B 2000 (Ovsyannikov) Plasma Diagnostics

Plasma diagnostics
Plasma
diagnostics
Ovsyannikov and Zhukov

Cambridge International Science Publishing Ltd.
7 Meadow Walk, Great Abington
Cambridge CB1 6AZ
United Kingdom Edited by A. A. Ovsyannikov and M. F. Zhukov
www.cisp-publishing.com
ISBN 1898326231
Cambridge International Science Publishing

PLASMA DIAGNOSTICS
i
ii
PLASMA DIAGNOSTICS
Editors
A.A. Ovsyannikov
Institute of Petrochemical Synthesis, Russian Academy of Sciences,
Moscow
M.F. Zhukov
Institute of Thermophysics, Suberian Division of Russian Academy of
Sciences, Novosibirsk
CAMBRIDGE INTERNATIONAL SCIENCE PUBLISHING
iii
Published by
Cambridge International Science Publishing
7 Meadow Walk, Great Abington, Cambridge CB1 6AZ, UK
http://www.cis-publishing.com
First published March 2000

Reprinted 2005
© Cambridge International Science Publishing
Conditions of sale
All rights reserved. No part of this publication may be reproduced or trans-
mitted in any form or by any means, electronic or mechanical, including pho-
tocopy, recording, or any information storage and retrieval system, without
permission in writing from the publisher
British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library
ISBN 1 898326231
Translated by Victor Riecansky

Production Irina Stupak
Printed and bound by Lightning Source UK Ltd
iv
Part I
SPECTRAL AND OPTICAL METHODS
Chapter 1 ....................................................................................................... 1
SPECTRAL DIAGNOSTICS METHODS ................................................ 1
1.1 Integral relationships. Optical measurement circuits .................................... 4
1.2 Spectral emission line ...................................................................................... 16
1.3 Width and shift of spectral lines ..................................................................... 22
1.4 Continuous spectrum ........................................................................................ 25
1.5 Kirchhoff law .................................................................................................... 27
1.6 Spectral absorption lines ................................................................................. 28
Chapter 2 ..................................................................................................... 32
KINETIC ENERGY, TRANSLATIONAL RELAXATION AND
DOPPLER BROADENING OF LINES IN NON-EQUILIBRIUM
PLASMA SPECTRA ............................................................................ 32
2.1 Introduction ...................................................................................................... 32
2.2 Determination of the type of velocity distribution of excited particles from
spectral measurements ..................................................................................... 34
2.3 Velocity distribution of atoms and molecules in excited electronic states .. 35
2.4 Measurement of gas temperature from Doppler broadening of spectral lines
............................................................................................................................ 37
2.5 A simple model of relaxation of the mean kinetic energy. Effect of gas
density ................................................................................................................ 41
2.6 Identification of excitation mechanisms ......................................................... 45
Chapter 3 ..................................................................................................... 50
OPTICAL ACTINOMETRY OF PLASMA ............................................ 50
Chapter 4 ..................................................................................................... 59
LASER METHODS OF PLASMA DIAGNOSTICS ............................. 59
4.1 Introduction ...................................................................................................... 59
4.2 Interference examination of plasma using lasers .......................................... 59
4.3 Holographic plasma diagnostic methods ....................................................... 74
4.4 Plasma diagnostics on the basis of scattering of laser radiation ................. 95
Chapter 5 ................................................................................................... 127

SPECTROSCOPY OF GROUND ELECTRONIC STATES OF
MOLECULES IN PLASMA USING TUNABLE LASERS ......... 127
5.1 Diode spectroscopy ......................................................................................... 127
5.2 CALS spectroscopy ........................................................................................ 136
i
Chapter 6 ................................................................................................... 147
DETERMINATION OF THE CONCENTRATION AND TEMPERA-
TURE OF HEAVY PARTICLES FROM THE SPECTRA OF
RAYLEIGH-SCATTERED LIGHT.................................................. 147
6.1 Introduction .................................................................................................... 147
6.2 Principles of the method ................................................................................ 147
6.3 Concentration and temperature of gas in flames and electric arc ............ 151
6.4 Plasma jet diagnostics .................................................................................... 155
6.5 Gas temperature in glow discharge .............................................................. 158
Chapter 7 ................................................................................................... 160

REFRACTOMETRIC PLASMA DIAGNOSTICS METHODS......... 160
7.1 Refractive index of plasma ............................................................................ 160
7.2 Plasma interferometry ................................................................................... 165
7.3 Shadow method ............................................................................................... 173
7.4 Diffraction interferometer based on IAB-451 shadow equipment for plasma
investigations ................................................................................................... 177
7.5 Examination of the nature of the plasma flow ............................................. 181
7.6 Temperature measurements by the interferometric method ..................... 189
Chapter 8 ................................................................................................... 196

DIAGNOSTICS OF PLASMA FLOWS WITH A DISPERSED PHASE ...... 196
8.1 Introduction .................................................................................................... 196
8.2 Methods of measuring particle size .............................................................. 199
8.3 Measuring the flow and particle velocities .................................................. 203
8.4 Particle temperature ...................................................................................... 206
Chapter 9 ................................................................................................... 207

MEASURING THE PLASMA FLOW VELOCITY BY THE TRACER
PARTICLE METHOD ....................................................................... 207
Part II
PROBE METHODS
Chapter 10 ................................................................................................. 219
ELECTRIC PROBES IN NON-EQUILIBRIUM PLASMA ............... 219
10.1 Introduction .................................................................................................. 219
10.2 Probe measurements of the energy distribution of electrons ................... 221
10.3 Determination of plasma parameters from probe measurements ........... 245
10.4 Selected problems of probe diagnostics ...................................................... 250
Chapter 11 ................................................................................................. 258

ELECTRIC PROBES IN CONTINUUM REGIME ............................ 258
11.1 Introduction ................................................................................................... 258
11.2 A system of equations and boundary conditions of the hydrodynamic
theory of electric probes ................................................................................. 259
ii
11.3 Volt–ampere characteristics of probes under hydrodynamic regimes .... 266
11.4 Diagnostic methods ....................................................................................... 275
11.5 Special features of experimental procedure ............................................... 292
11.6 Comparison of theoretical and experimental results ................................ 294
11.7 Conclusion ..................................................................................................... 297
Chapter 12 ................................................................................................. 299

PROBE METHODS OF DIAGNOSTICS OF CHEMICALLY REACT-
ING DENSE PLASMA ....................................................................... 299
12.1 Formulation of the problem and solution method ..................................... 300
12.2 Numerical solution results ........................................................................... 304
12.3 Experiments .................................................................................................. 305
12.4 Discussion of the results ............................................................................... 306
12.5 Electric probes in special conditions .......................................................... 312
12.6 Diagnostics of dense chemically reacting still plasma with electric probes
with a modulated potential ............................................................................ 326
Chapter 13 ................................................................................................. 338

ELECTRIC AND THERMAL PROBES IN THE PRESENCE OF
CHEMICAL REACTIONS IN NON-EQUILIBRIUM PLASMA . 338
13.1 Electric probe ............................................................................................... 338
13.2 Thermal probe .............................................................................................. 350
Chapter 14 ................................................................................................. 356

MEASUREMENTS OF THE ENTHALPY OF HIGH-TEMPERA-
TURE GAS FLOWS ........................................................................... 356
14.1 Introduction .................................................................................................. 356
14.2 Main methods of measuring the enthalpy of ............................................. 357
high-temperature gas ........................................................................................... 357
14.3 Enthalpy sensors and probes ....................................................................... 363
14.4 Sensitivity of the enthalpy sensor and measurement error ...................... 381
14.4 Conclusions ................................................................................................... 385
Chapter 15 ................................................................................................. 386

PROBE MEASUREMENTS OF HEAT FLOWS IN PLASMA JETS386
15.1 Measurement procedure .............................................................................. 386
15.2 Determination of plasma temperature ....................................................... 390
Part III
MACROSCOPIC DIAGNOSTICS
Chapter 16 ................................................................................................. 394
METHODS OF EXAMINING THE SPATIAL STRUCTURE OF
RADIO-FREQUENCY CAPACITANCE DISCHARGES ............. 394
16.1 Introduction .................................................................................................. 394
16.2 Specific features of producing, examining and using radio-frequency
iii
capacitance discharge plasma. Forms of existence of RFCD ..................... 395
16.3 Interpretation of volt–ampere characteristics of RFCD and their
relationship with the spatial structure of the discharge .............................. 402
16.4 Constant electric fields in RFCD ................................................................ 415
16.5 Using the battery effect in examining RFCD ............................................ 427
16.6 Effect of the frequency of RF field on the structure of the discharge and
plasma characteristics .................................................................................... 429
16.7 Active probing of RFCD using a constant electric current ...................... 434
16.8 The radial structure of the RF capacitance discharge ............................. 440
16.9 Optical methods of examining the spatial structure of the RF capacitance
discharge .......................................................................................................... 445
16.10 Conclusions ................................................................................................. 448
Chapter 17 ................................................................................................. 449

WAVE BREAKDOWN IN DISTRIBUTED SYSTEMS ...................... 449
17.1 Introduction .................................................................................................. 449
17.2 History ........................................................................................................... 450
17.3 Transition processes ..................................................................................... 454
17.4 Generalised block diagram of experiment ................................................ 457
17.5 Spatial–time structure of breakdown waves. Effect of preliminary
ionisation .......................................................................................................... 469
17.6 Electrodynamic processes in wave breakdown ......................................... 471
17.7 X-radiation during a wave breakdown ...................................................... 490
17.8 Initiation of laser radiation by the breakdown wave ............................... 494
17.9 Effect of the type of gas ............................................................................... 496
17.10 Current state of the theory of wave breakdown ...................................... 497
17.11 Wave breakdown - a method of examining processes in plasma ........... 501
17.12 Conclusion ................................................................................................... 505
Chapter 18 ................................................................................................. 506

PROBE MEASUREMENTS OF POTENTIAL DISTRIBUTION IN
DENSE PLASMA ................................................................................ 506
18.1 Special features of the method .................................................................... 506
18.2 Electric field of the high-current arc ......................................................... 510
18.3 Electric field in the laminar plasma jet of the electric arc plasma torch 515
18.4 The electric field of the plasma flow of a two-jet plasma torch...... 518
Part IV
METHODS OF PROCESSING RESULTS AND AUTOMATION OF
DIAGNOSTICS
Chapter 19 ................................................................................................. 522

REDUCTION TO A UNIFORM LAYER IN AXISYMMETRIC
OBJECTS ............................................................................................ 522
19.1 Smoothing experimental data ..................................................................... 522
iv
19.2 Numerical solution of Abel’s integral equation ......................................... 524
19.3 Numerical solution of the Volterra integral equation ............................... 527
19.4 Simplified reduction algorithms ................................................................. 532
Chapter 20 ................................................................................................. 538

RECONSTRUCTION OF VELOCITY DISTRIBUTION FUNCTIONS
OF EMITTING PARTICLES FROM THE SHAPE OF THE CON-
TOUR OF SPECTRAL LINES ......................................................... 538
Chapter 21 ................................................................................................. 545

AUTOMATION OF MEASUREMENTS IN PLASMA DIAGNOS-
TICS ...................................................................................................... 545
REFERENCES ......................................................................................... 549
INDEX ....................................................................................................... 577
v
vi
Spectral Diagnostics Methods
Part 1
SPECTRAL AND OPTICAL METHODS
Chapter 1
SPECTRAL DIAGNOSTICS METHODS

The spectral diagnostics methods are based on measuring the intensity
of the spectral lines of emission and absorption and the continuous spec-
trum, halfwidths and shifts of spectral lines.
The state of high-pressure plasma can be usually described using mod-
els of local thermodynamic (LTE) or partial local thermodynamic equi-
librium (PLTE). LTE plasma has the following characteristics. All
plasma particles, including free electrons, are distributed with respect
to the velocity of random motion in accordance with the Maxwell law:
F m I 3/ 2
F I
dN ( v ) = N 4 p
H 2pkT K GH
exp -
2 kTJK
mv 2 2
v dv. (1.1)
Here N is the concentration of the particles of the given type (the number
of particles in unit volume), N (v) is the concentration of particles with
velocity v; m is the particle mass; T is plasma temperature; k is the
Boltzmann constant.
The population of the energy levels is determined by the Boltzmann
law
gn E F I
Nn = N
S H
exp - n ,
kT K (1.2)
where N n is the concentration of particles of the energy level n with

energy E n and the statistical weight g n; S is the statistical sum with
respect to the states
1
Plasma Diagnostics
å gn exp FH - kTn IK ,
nmax
E
S= (1.3)
n=1
where n max is the maximum main quantum number of energy levels re-
alised in plasma. The latter is determined from the condition
E n max £ U – DU, where U is the ionisation energy, DU is the reduc-
tion of the ionisation energy of atoms (ions) in plasma,
F I 1/ 2
DUz =
(1 + z)e2
rD
, rD =
kT
8pe2 Ne
GH JK , (1.4)
r D is the radius of the Debye sphere; e is the electron charge; z is the

ion charge (for a neutral atom z = 0, for a singly-ionised atom z = 1,
etc.).
The relationship between the concentration of particles taking part
in the chemical reaction ab Û a + b, is determined in accordance with
the law of acting masses:
F
Sa Sb 2 pma mb kT I 3/ 2
FG IJ ,
GH JK
N a Nb Dab
Nab
=
Sab 2
h mab
exp -
H kT K
(1.5)
where N a, N b, N ab is the concentration of particles of components a,

b, ab of the chemical reaction, m a, m b, m ab is the mass of the corre-
sponding particles, D ab is the energetic threshold of the reaction, h is
the Planck constant. In particular, for the ionisation reaction
a « i + e we can write the Saha equation
Ne Ni 2 Si F 2pm kT I 3/ 2
F U - DU I
=
H h K exp -
H K
e
2
, (1.6)
Na Sa kT
where m e is the electron mass.

If it is required to calculate the composition of plasma, the equa-
tions (1.5) and (1.6) supplement the following relationships:
the equation of quasineutrality of plasma
2
p t
Ne = å å zNiz, x , (1.7)
x =1 z =1
where Ni,z x is the concentration of ions of component x with the charge

z;
the equation of state
p
P = kT å N x + kTN e , (1.8)
x =1
where summation is carried out over all neutral or ionised plasma com-
ponents;
The equation of conservation of the initial composition (the equa-
tion of elementary plasma balance)
t t
å N xz=1: å N xz=2 ... = c1: c2 ... . (1.9)
z =0 z =0
For example, for the plasma formed from steam, the initial composi-
tion NH: H O = 2:1; with an allowance made for all possible plasma com-
ponents H 2, H, H +, OH, O 2, O, O +, equation (1.9) assumes the form
(2 N H 2 + N H + N H + + N OH ):(2 N O2 + N O + N O + + N OH ) = 2 :1 .
PLTE plasma differs from LTE plasma by the fact the kinetic
energy of heavy particles (molecules, atoms, ions) is not equal to the
electron temperature since the excitation (1.2) and ionisation (1.6)
temperatures are equal to the temperature of free electrons.
The intensity of molecular bands, spectral lines and the continuous
spectrum is calculated in the LTE and PLTE plasma on the basis of
the equations of transfer of radiation using the probabilities of tran-
sitions and cross sections of the photoprocesses.
The diagnostics methods based on the emission spectra can be applied
directly to optically thin plasma objects where the absorption of radiation
is negligible. Verification of reabsorption can be carried out by measuring
the optical thickness t n = knl, where k n is the absorption coefficient at
frequency n; l is the effective length of the absorbing layer. For this
3
Plasma Diagnostics
purpose, the examined object is irradiated with an auxiliary source or

inherent radiation.
The optical thickness can also be evaluated by comparing the intensity
of examined radiation I n with the radiation of an absolute black body
B v at the same temperature. For a homogeneous emitter
FG InIJ
H
t n = k nl = - ln 1 -
Bn K
.
In order to ignore absorption, the following condition must be fulfilled

k nl << 1 (for example, ~0.01).
The emission spectra can be used for diagnostics and also in the case
of marked self-absorption (optically dense plasma). However, to deter-
mine the local emission coefficients, it is necessary to take absorption
into account.
A number of methods is based on measuring absorption, specially
the methods of determining the temperature using the Kirchhoff law and
the concentration of the atoms on the basis of the spectral absorption
lines.
The local values of the emission and absorption coefficients in
axisymmetric objects are found from the solution of corresponding in-
tegral inclusions. In operation with the contours of spectral lines, a
similar procedure is used for selecting the frequencies inside the con-
tour. This is followed by constructing the contour for a number of values
of the radial coordinate and by determining of the half with (shift) of
the spectral line as well as the function of the radial coordinate.
1.1 Integral relationships. Optical measurement circuits

We shall now present equations for a partial case of propagation of
monochromatic radiation along the chord of the cross section of a cy-
lindrical (axisymmetric) object. All local optical parameters are assumed
to be functions of only the radial coordinate. The following concepts
are used:
– the emission coefficient – the amount of energy emitted by the unit
volume per unit time into the unit solid angle in the unit frequency range;
– the radiation intensity – the amount of energy in the unit frequency
range passing through the unit area normal to the direction of propa-
gation of radiation, per unit time in the unit solid angle;
– the absorption coefficient – its numerical value is equal to the value
reciprocal to the distance along which the light flux is weakened
e @ 2.72 times;
4
– the transmittance coefficient – the fraction of radiant energy passing

through the object.
An optically thin source. We shall examine a cylindrical source
(Fig.1.1) whose emission coefficient e(r) depends only on the radial co-
ordinate (self absorption can be ignored). The intensity of radiation leav-
ing the source in plane XY along the chord [–y 0, y 0] at distance x from
the axis of symmetry is
z
y0
I( x) = e(r ) dy.
- y0
We set e (r ³ R) = 0 and, consequently, I(|x| ³ R) = 0. In the coordi-

nates (x, y), y = r2 - x2
z
R
e (r ) rdr
I( x) = 2 . (1.10)
| x| r2 - x2
The equation (1.10) is the integral Volterra equation of the first kind.
It is also referred to as the Abel integral transform. The following pro-
cedure can be used for inversion of the Abel transform. We shall trans-
form (1.10) using the operator: 1
z
R2
d dx 2
[...] . (1.11)
dz 2 2 x 2 - z2
z
The order of integration on the right hand side will be changed
F ig .1.1 Cross-section of an axisymmetric object.

ig.1.1
5
Plasma Diagnostics
F I
z z z
R2 R2 r2
d I( x)dx 2 d
= 2 e(r) GG dx 2
JJ dr .
2
dz 2
z2 x 2 - z 2 dz z2 H z2 r -x
2 2
x -z
2 2
K
The internal integral on the right hand side is reduced as a result of
substitution
x 2 = z 2 cos2 j + r 2 sin 2 j
to a tabulated integral and is equal to p. We shall write the result for

z = r
z
R
1 d I ( x) xdx
e(r) = - . (1.12a)
p rdr x2 - r2
r
Per parts integration of (1.12a) and subsequent differentiation with re-

spect to the parameter give
z
R
1 dI( x) / dx
e(r) = - dx. (1.12b)
p x 2
- r 2
r
The equation (1.12) are the inversions of the Abel transform (1.10).
An optically dense source. At an arbitrary value of coefficient k(r),
the intensity of radiation can be determined from the transport equa-
tion
dI
= e - kI . (1.13)
dy
This is a linear differential equation of the first order whose solution

has the form
e z
I ( y) = exp k (r ) dy j LMN z e z
e(r ) exp k (r ) dy dy + C j OP
Q
6
Transferring to integration in the ranges [–y 0, y 0] (Fig.1.1) and setting

k (r ³ R) = 0, e (r ³ R) = 0, I (–y 0) = I 0 (x) = C, I (y 0) = J (x), we
obtain
F I F I
J ( x ) = I ( x ) exp G - z k (r ) dyJ + z e(r ) exp G - z JJ
y0 y0 y0
0
GH - y0
JK - y0
GH y
k (r ) dy , dy
K
or
J ( x ) = I 0 ( x )w ( x ) + I ( x ). (1.14)
Here I (x) is the intensity of inherent radiation of the source; w (x) is

the transmittance coefficient, I 0 (x) and I 0 (x)w (x) is the intensity of
external emission prior to and after passage of the object. The equa-
tion for the transmittance coefficient
F I
w( x ) = exp G - z k(r ) dyJ
y0
GH - y0
JK (1.15)
can be written in the form of Abel's integral equation
z
R
k (r ) rdr
- ln w ( x ) º t( x ) = 2 , (1.16)
x r2 - x2
where t (x) is the optical thickness of the object. Equation (1.16) has
the inversion of the type (1.12).
The expression for the intensity of inherent radiation of the source
F I
z e(r ) exp G - z JJ
y0 y0
I( x) =
- y0
GH y
k (r )dy ¢ dy
K (1.17)
is transformed as follows
7
Plasma Diagnostics
F I LM F I F I OP
z z z z z
y0 0 y y0 y
GGH
I ( x ) = exp - k (r )dy JJK M e(r) expGH k(r)JK dy¢ + e(r ) exp GH JK PQ
k (r )dy ¢ dy =
0 N - y0 0 0 0
F I L F IO
z z z
e(r )MexpG - k (r )dy ¢J Pdy,
y0 y0 y
expG - k (r )dyJ
H 0 K 0
MN H K PQ
0
which takes into account the equality
F I F I
z z z z
0 y y0 y
e(r ) exp GH JK
k (r )dy ¢ = e(r ) exp - k (r )dy ¢ dy. GH JK
- y0 0 0 0
Using the notations of the hyperbolic cosine and the transmittance co-
efficient (1.15), we can write
F I
z z
y0 y
I ( x ) = 2 w( x ) w (r ) ch G k (r )dy ¢ dy JK
0 H 0
or in coordinates (x, r)
F I
z GG z JJ
R r
I( x) k (r ¢ )r ¢dr ¢ r dr
= 2 e(r ) ch
w( x )
x H x r¢ - x
2 2
K r - x2
2
which represents the integral Volterra equation of the first kind.
Measurement of radiation and transmittance. Figure 1.2 shows some

of the existing methods of measurement of radiation and transmittance.1
The image of the examined object O is projected by the lens L 1 on the
aperture of a spectral apparatus. When measuring the intensity I (x),
the translucent radiation is cut off by an obturator. In the circuits 1.2
a-c, the axis of the object should be normal to the direction of the
aperture. If necessary, the image of the object is rotated through 90°
using a prism of total internal reflection, a system of three or two mirrors
placed between the lens L1 and the object O. In photographic registration
we measure the distribution of blackening across the spectrogram and
8
Ma
A L2 O L1 S
Ma
M L2 O L1 S
L1 S
M
M
L1
S
O
F ig .1.2 Optical circuit of measurement of radiation and transmittance. A) auxilliary

ig.1.2
source; L 1 , L 2 (lenses); M) spherical mirror; Ma) mask; O) examined object; S) input
aperture of the spectral apparatus.
then transferred to the intensity distribution I (x) on the basis of the

characteristic curve of the photolayer. In photoelectric recording, the
intensity distribution I (x) is measured directly. Scanning in the direction
of the x axis is carried out by moving the diaphragm along the input
or output aperture of the device. In the diagram shown in Fig.1.2d the
object is scanned in the direction normal to the optical axis.
To measure the transmittance w (x), the examined object is illumi-
nated by the radiation of auxiliary source A (Fig.1.2a) or by natural
radiation reflected from the mirror M (Fig.1.2b-d). The total intensity
in illumination is
J ( x ) = I 0 ( x )w ( x ) + I ( x ),
9
Plasma Diagnostics
where I (x) is the intensity of radiation, I (x) is the natural radiation

of the source.
Therefore, knowing J (x), I 0 (x), I (x) we obtain
J ( x ) - I ( x)
w( x ) = .
I 0 ( x)
The diagram in Fig. 1.2 b gives the total combination of the image with
the object. The intensity of translucent radiation I 0 (x) = gI (x), where
g are radiation losses on the path from the source to the mirror and back.
The value of g is determined from the section of the spectrum where
absorption can be ignored, i.e. w (x) » 1 and g = (J (x) – I (x))/I (x).
If the mask Ma with non-transparent strips, parallel to the axis of the
source, is placed in front of the mirror M, in the focal plane of the device
we obtain alternating values of J (x) and I (x). The main disadvantage
of the setup shown in Fig. 1.2b is that the transmittance factor can-
not be measured outside the zone of radiation of the source. This dis-
advantage is not found in the setup in Figs 1.2 a,c,d. In the circuit in
Fig. 1.2 c, the image is projected by the mirror M in the plane of the
lens L 1. The examined object is illuminated with a wide, relatively uni-
form beam. The circuit showed in Fig.1.2 d is even more efficient in
this respect. The spherical mirror is rotated when scanning the object
O in such a manner that the object is illuminated for any values of x
by its brightest central part. In Fig.1.2 a–c, the value I 0 = const and
is determined from measurements outside the limits of the absorption
zone of the object O or in the setup in Fig.1.2 a in the absence of the
examined object. In photographic recording, the translucent radiation
is often modulated by setting the receiving apparatus to the modula-
tion frequency. This greatly increases the accuracy of measuring trans-
mittance and also enables auxiliary sources (Fig.1.2 a) of relatively small
brightness to be used.
The circuit shown in Fig.1.2 d is designed for examining stationary
sources. Nonstationary processes can be examined using the setup in
Fig.1.2 a-c with photographic registration. The time resolution is ob-
tained by placing an instantaneous gate (Kerr cell, a system of rotat-
ing mirrors) in front of the aperture or using other measures.
Spatial resolution power. The integral equations presented above were
obtained in the one-beam approximation. However, actual devices have
a finite angular aperture. Figure 1.3 shows the geometry of beams in
reflecting the object O on the aperture of the device by the lens L. The
normal incidence beam on the objective has the intensity I (x). The
oblique beam, propagating under angle j to the optical axis corresponds
10
F ig .1.3 Diagram of recording radiation with a finite appature of the device.

ig.1.3
to the chord whose distance from the centre of the object is x cos j
and intensity I (x cos j)cos j. The recorded intensity of radiation is
averaged out with respect to the angle 2a and is equal to
z
a
1
I( x) = I ( x cos j ) cos j dj. (1.21)
a
0
Replacing I (x) over the averaging region by a segment of the straight

line, for small angles we obtain a
F I FF II.
GH
I ( x) » 1 -
a2
6 JK GH GH
I x 1-
a2
6 JK JK (1.22)
The difference between the measured intensity I ( x) and true intensity

is determined to a first approximation by the scale factor
a2
b = 1- .
6
The value of b differs only slightly from unity, even for relatively high
angles a. Thus, b = 0.998 at a = 0.1. This shows that the reduction
of the spatial resolution power as a result of the finite angular aper-
ture of the device is insignificant in the majority of cases.
Taking transverse oscillations into account. Electric arcs, plasma
and flame jets are often show transverse oscillations over a wide fre-
quency range. The direct path of diagnostics of these objects is to obtain
11
Plasma Diagnostics
F ig .1.4 Coordinate system for examining an oscillating source.

ig.1.4
instantaneous lateral intensity distributions. This requires using

multichannel photoelectric registration or electron–optical converters with
a good time resolution. Experiment formulation is greatly simplified if
we use information on the transverse oscillations of the examined ob-
ject. 2,3
Let us assume that the properties of the cross section of the examined
object are constant and that the cross section wanders in the X'O'Y' plane
around some most probable position, Fig.1.4. The input aperture of the
spectral device is combined with the X' axis. It will be assumed that
small displacements of the source in the direction of the Y' axis do not
change the amount of light passing into the device. This is justified for
a small relative orifice of the projecting lens, as is usually the case.
Consequently, it is sufficient to take into account the displacement of
the source in the direction of the axis X'. ϕ(x') will denote the density
of the distribution of probability of deviation of the axis of the source
from the mean position x' = 0, coinciding with the centre of the inlet
aperture of the device. Correspondingly, the probability of some chord
x of the source being situated opposite the section x' ÷ x'+dx' of the
inlet aperture is
j( x ¢ - x )
j( x ¢ - x )dx = j( x ¢ - x)(dI / dx ) -1 dI = dI . (1.23)
I ¢( x )
The total probability α(I, x') dI of passage through the section of

the aperture x' ÷ x'+dx' with the intensity I ÷ I + dI is obtained by sum-
ming up equation (1.23) with respect to all chords x j of the source with
the same intensity I
j ( x ¢ - xi )
a ( I , x ¢ ) dI = å I ¢( xi )
dI . (1.24)
i
12
If the functions ϕ(x') and ϕ(I, x') are known, the determination of
the distribution of intensity I(x) is reduced to solving the problem of
non-linear algebraic equations, following from (1.24)
n j( x ¢j - xi )
å I ¢( xi )
= a ( I , x ¢j ), j = 1, m, m = 2n. (1.25)
i =1
At fixed values of I we exclude I'(x i) from the system (1.25) and

determine x i, which gives a discrete function I (x). In a partial case of
a symmetric and monotonically decreasing function I (x) at |x|>0, each
intensity I corresponds to two symmetric chords x and –x. Consequently
I (–x) = I (x), I'(–x) = I'(x) and the system (1.25) has the form
j( x1¢ - x ) + j( x1¢ + x )
= a ( I , x1¢ ),
I ¢( x )
j( x2¢ - x ) + j( x2¢ - x ) (1.25a)
= a ( I , x2¢ ).
I ¢( x )
Dividing the first equation (1.25a) by terms by the second equation, we

obtain
j( x1¢ - x ) + j( x1¢ + x ) a ( I , x1¢ )

= , (1.25b)
j( x2¢ - x ) + j( x2¢ + x ) a ( I , x2¢ )
We introduce the notations
j( x1¢ - x ) + j( x1¢ + x )
j( x1¢ , x2¢ , x ) = ,
j( x2¢ - x ) + j( x2¢ + x )
a ( I , x1¢ )
a ( I , x1¢ , x2¢ ) = .
a ( I , x2¢ )
b g
We specify x and calculate j x1 , x2 , x . For the value of α equal to j ,
from the known dependence of a on I, we determine the correspond-
ing I (x). This is the method of processing the results of measurements
of the radiation intensity of a vibrating object proposed in Ref. 2.
The function ϕ(x') is determined by experiments, and the examined
object is displayed on the aperture of the photographic recording
13
Plasma Diagnostics
device in such a manner that the axis of the source is perpendicular

to the direction of the aperture. On the photographic recording, with
sufficient time resolution, we calculate the mean position of the axis
of the source and the probability of its deviation to different distances
from the mean distance. This gives the function ϕ(x') after normalisation
with respect to unity.
To determine α(I, x') by the photoelectric method with a good reso-
lution, it is necessary to recorded the time dependence of the intensity
of the analytical spectral line or the section of continuum consecutively
for two or more co-ordinates x'. The function α(I, x') is calculated from
the frequency with which the intensity fits into the fixed ranges
I ÷ I + dI.
The authors of Ref. 3 proposed a different method of the spectroscopy
of vibrating objects. Like in Ref. 2, using the method proposed in Ref.
3, we measure the function ϕ(x') of vibration of the source but the spec-
trum is recorded without time resolution. The ratio between the distri-
bution of intensity q(x'), averaged with respect to time, and the true
lateral distribution of intensity I(x) of the source is determined as follows.
As mentioned previously, the probability of passage of intensity I(x),
corresponding to the chord x, through the section of the aperture x ÷
x' +dx' is ϕ(x'– x) dx. The radiation flux from the chord x through the
given section of the aperture, averaged of a relatively long period of
time, is
dq ( x ¢) = I ( x )j( x ¢ - x )dx. (1.26)
The total flux q(x') is obtained by integrating (1.26) throughout the entire
range of variation x ∈ [–R, R], in which I (x) differs from 0
z
R
q( x ¢) = I ( x ) j( x ¢ - x )dx. (1.27)
-R
The required distribution of intensity I (x) is the solution of the inte-

gral equation (1.27). The relationship (1.27) between the true I (x) and
the measured q (x') intensity distributions is also observed in record-
ing a stable source in a real device with the finite resolving power and
the apparatus function α(x'). The distortion of the intensity distribu-
tion in lateral examination as a result of the finite resolution of the re-
cording device may be considerable for sources with high intensity gra-
dients. If both reasons are valid, then ϕ(x') in equation (1.27) is the
convolution of the apparatus function of the device and the conditional
14
apparatus function, chatacterising the vibrations of the source. To

determine the apparatus function of the device, a uniformly illuminated
narrow aperture is made at the optical axis instead of the examined ob-
ject. The distribution of intensity in the focal plane of the device, meas-
ured for this auxiliary source, is also its apparatus function. Numeri-
cal realisation and examples of practical application of the method were
published in, for example, Ref. 10–13.
Brightness reference sources

The measurement of the absolute intensity of the spectral lines and
continuum and of the relative intensity of radiation in a wide spectral
range is carried out using brightness reference sources (standards of
radiation intensity) [4,5].
The primary intensity standard is the radiation of an absolutely black
body. This radiation is determined only by its temperature and does not
depend on the properties of the material from which it is made. The
brightness and spectral composition of its radiation are calculated using
Planck’s equation. Artificially, the properties of the absolutely black
body are reproduced by means of a small hole in the wall of a large
closed cavity. The material of the walls is tungsten or carbon. In most
cases, the device has the form of a pipe heated with electric current.
A large number of diaphragms is placed inside the pipe. These dia-
phragms prevent the direct exit of radiation from the cavity through the
examination hole. The pipe contains a small crucible with gold or plati-
num, and the intensity of radiation of the cavity is measured at the
melting point of these materials. The melting point of gold is 1337.58
K, that of platinum is 2044.9 K.
Secondary intensity standards are calibrated at specific wavelengths
by comparing the intensity of their radiation with the intensity of ra-
diation of the absolutely black body. The calibration error with respect
to intensity is +(5–10)%.
Tungsten reference sources are represented tungsten lamps with a
strip-shaped heating filament. The temperature of the filament is main-
tained at approximately 2700 K. The degree of blackness of tungsten
has been measured over a wide range of wavelength and filament tem-
perature, and its value is close to 0.45.
Another suitable secondary intensity standard is the crater of the anode
of a carbon arc (Fig. 1.5). In the carbon arc where the electrodes are
positioned under the almost right angle in relation to each other, at the
current intensity in the arc slightly lower than the value correspond-
ing to the start of hissing the temperature of the end surface of the anode
facing the examiner is equal to the sublimation temperature of graph-
ite (3995+20 K), at a degree of blackness of 0.7–0.75, depending on
15
Plasma Diagnostics
F ig .1.5 Distribution of electrodes in recording the crater of the carbon anode.

ig.1.5
the wavelength. The brightness of radiation of the crater of the carbon

anode is considerably higher in comparison with the tungsten strip heating
lamp. Consequently, it can be used as a brightness reference in the wave-
length range exceeding 2500 Å. The arc current is selected on the basis
of the dimensions of the anode crater. For example, for a carbon an-
ode with a right-angled cross-section of 5.5 × 3 mm, the arc current
is 9–9.5 A, which is slightly lower than the current of the hissing arc.
The carbon arc is not recommended for use in the range 3700–4200 Å
which contains the emission bands of cyanogen CN, and in the range
4500–4740 Å containing the bands of molecular carbon C 2.
In these regions of the spectrum heating is more efficient when carried
out using a high-pressure xenon lamp whose radiation consists of the
high-intensity continuous spectrum and contains only several lines at
the wavelengths below 6700 Å. The colour temperature in the range
3700–5300 Å is 5572+33K.
The electric arc, stabilised by the walls, is characterised by an even
higher brightnes in the group of the standards. The source can be used
for calibration in a wide range of the spectrum from the infrared to
vacuum ultraviolet range. The diagram of stabilisation of the arc is
shown in Fig. 1.6. At a relatively long arc column, the radiation in the
centre of the spectral lines corresponds to Planck’s radiation. This ra-
diation is taken through end windows and is used as reference radia-
tion. The temperature at the axis of the arc in relation to current and
the internal diameter of the pipe may vary from 10 000 to 30 000 K.
In order to eliminate the absorption of radiation in the cold regions,
the end zones of the pipe are blown with a different gas (argon, he-
lium).
1.2 Spectral emission line

The emission factor of the spectral line for plasma in a local thermal
16
Gas Gas
F ig .1.6 Diagram of the arc with stabilising walls. 1) water-cooled copper rings, hollow
ig.1.6
inside; 2) window; 3) insulating interlayers.
equilibrium is:
e nk =
h n nk Ank gn
4p S
N exp - E / kT ).a f (1.28)
Here N is the concentration of the atoms (ions) of the emitting element;

T is plasma temperature; n nk , A nk is the emission frequency for the
n ® k transition and the transition probability, respectively.
The functional dependence (1.28) is used as a basis of the method
of determining plasma temperature, particle concentration (atoms, ions)
and transition probabilities of spectral lines.
Absolute intensity of spectral lines. At the known transition prob-
abilities, initial composition and the pressure of LTE plasma, the tem-
perature dependence of the particle concentration can be calculated in
advance and the emission factor (1.28) is written in the form
e = e (T ). (1.29)
Consequently, by measuring the absolute intensity it is possible to de-

termine the plasma temperature. Figure 1.7 shows the dependence of
the intensity of the spectral line ArII 4806 Å on temperature for ar-
gon plasma at atmospheric pressure.
Varying (1.29) and taking into account (1.28), we obtain an estimate
of the error of the temperature measuring method
dT e ¶T LMFG de IJ FG de IJ dA OP
MNH e K H eK
= + +
PQ
. (1.30)
T T ¶e ran sys A
The first term in the square brackets is the random error of measure-
17
Plasma Diagnostics
ment reproducibility, the second term is the systematic error of cali-

bration of the reference brightness source, the third term is the error
e ¶T
of the transition probability. The factor is the coefficient of error
T ¶e
transfer (Fig.1.7).
In most cases (de/e)rand » 5%, (de/e)syst » 10%, dA/A » 10%, and for
an example shown in Fig.1.7, where (e/T)(¶T/¶e) < 0.1, dT/T » 3%.
If we know the transition probability and transfer temperature, equa-
tion (1.28) can be used to determine the atom (ion) concentration from
the measured intensity. The corresponding estimate of the error follows
from the variation of (1.28)
dN de dA E dT
= + + , (1.31)
N e A kT T
dN
which for the example shown in Fig.1.7 gives » 15 + 10 + 10 × 3 » 50% .
N
The ratio of the intensities of spectral lines. For the ratio of the
intensities of the spectral lines of some element of the same degree of
ionisation, equation (1.28) gives
e 2 ( Agn)2 F
E - E1 I
=
e1 ( Agn)1 H
exp - 2
kT
,
K (1.32)
and consequently
F ig .1.7 Dependence of the intensity of continuum λ = 4810 Å (I) (erg/(cm 3 sterad

ig.1.7
Å)) and spectral line Ar(II) 4806 Å (2) (erg/(cm 3 sterad) on temperature at P = 1
atm (3, 4 are the corresponding coefficients of transfer of the error). 6
18
( E 2 - E1 ) / k
T= .
( Agn) 2 e
ln - ln 2 (1.33)
( Agn)1 e1
Measuring the ratio of intensities, from equation (1.33) we determine

plasma temperature at a known ratio of the transition probabilities. The
variation of equation (1.33) gives the following estimate of the tempera-
ture determination error:
dT
=
kT
+
LM
d ( A2 / A1 ) d ( e 2 / e1 ) OP
N Q
. (1.34)
T E2 - E1 A2 / A1 e 2 / e1
In particular, equation (1.34) shows that to increase the accuracy it is

necessary to use spectral lines with a higher excitation energy differ-
ence. If, for example,
kT 1 d ( A2 / A1 ) d ( e 2 / e1 )
» , » 10%, » 5%,
E 2 - E1 3 A2 / A1 e 2 / e1
dT
then » 5%. .
T
The Ornstein method of measuring temperature on the basis of the
ratio of the intensities of two spectral lines can be extended to a case
of using a group of spectral lines belonging to the same type of atoms.
From equation (1.28), we obtain
FG e IJ Ei
ln
H Agn K i
= yi = a -
kT
, (1.35)
where a is the constant at fixed temperature T. Equation (1.35) gives

a straight line whose slope is tan j = 1/kT (Fig.1.8). To determine tem-
perature, we draw, through the experimental points y i, i = 1, n , a sec-
tion with a straight line by the methods of least squares. 7 We obtain
nå
F I
n
-Gå E J Ei2
n 2
kT =
H K
i =1 i =1
i
n n n (1.36)
n å yi Ei - å yi å Ei
i =1 i =1 i =1
19
Plasma Diagnostics
F ig .1.8 Graph y i = y ( E i) in the presence of Boltzmann distribution.

ig.1.8
and the error estimate
dT kT FG de + dAIJ ,
OP H e A K
=
T LMn
1F
n I 2 1/ 2
MNå E - n GH å E JK PQ
2 (1.37)
i i
i =1 i =1
where de/e is the error of measurement of intensity measurements,

dA/A is the inaccuracy in the transition probabilities. When using two
spectral lines, the equations (1.36) and (1.37) are usually replaced by
(1.33) and (13.4), respectively. The increase of the number of lines in-
creases the accuracy of temperature measurements and also gives an
answer to the question as to whether there is a Boltzmann’s distribu-
tion of atoms over the excited states. In this case, all points fit a sin-
gle straight line, within the measurement error range. If there is a large
deviation from Boltzmann’s distribution, the experimental points are
situated around some curve and not in the vicinity of a straight line.
For the ratio of the intensities of the atomic e a and ionic e i lines of
the same element, equation (1.28) gives
e a ( Agn) a Na E - Ei F I
=
e i ( Agn)i Ni
exp - a
kT H
.
K (1.38)
Consequently
Ni x ( Agn)a e i E - Ei F I
= =
Na 1 - x ( Agn)i e a
exp - a
kT
.
H K (1.39)
20
Measuring the ratio of the intensities of the atomic and ionic spectral
lines, from (1.39) we determine the degree of ionisation of the atom at
the known plasma temperature and the ratio of the transition probabilities.
Using (1.39) together with the Saha equation (1.6), we determine the
concentration of electrons in plasma
e a ( Agn)i F 2 pmkT I 3/ 2
LM (U - DU ) - ( Ea - Ei ) OP
Ne =
e i ( Agn) aH2
h2 K exp -
N kT
.
Q (1.40)
The error of determining N e is obtain by varying (1.40)
dNe d(e a / e i ) d( Aa / Ai )
= + + +
LM
3 (U - DU ) + Ea - Ei dT OP
N Q
. (1.41)
Ne e a / ei Aa / Ai 2 kT T
Assuming that E a » E i , U/kT » 10, DU « U, d (e a/e i )/(e a/e i ) » 5%,

d (A a/A i)/(A a/A i) » 10%, dT/T » 3%, we have dN e/N e » 50%.
Optimum temperature. With increasing temperature, the intensity
of spectral lines initially increases as a result of the processes of ex-
citation and ionisation of the atoms (ions), reaches a maximum value
and then decreases. Figure 1.9 shows the temperature dependence of the
intensity of a number of spectral lines. The temperature corresponding
to the intensity maximum is referred to as optimum temperature. In
stationary plasma with a temperature gradient, the maximum intensity
can be observed in a separate region thus recording the temperature of
this region. The spatial distribution of temperature is determined by
comparing the measured intensity distribution with the calculated tem-
perature dependence of intensity.
Optimum temperature depends of the ionisation energy of the atom
(ion), the excitation energy of the spectral line and the electron con-
centration in plasma. The method of determining the temperature from
the position of the maximum intensity of the line (Lorentz method) can
be applied if the maximum temperature in a heterogeneous examined
object exceeds the optimum temperature for the given line. The advan-
tages of the Lorentz method is that it does not require data on the tran-
sition probabilities and it is not necessary to measure the absolute in-
tensity. In addition, radiation readsorption has no significant effect in
measuring the temperature by this method on the basis of resonance
lines. 8
The maximum intensity of the spectral line occupies a relatively wide
21
Plasma Diagnostics
log ε rel a log ε rel b
F ig .1.9 Intensities of the line NaI 5890 Å in air plasma at the atmospheric pressure
ig.1.9
and at a sodium concentration of N = N a 0 + N i = 1×10 15 cm –3, 8 (a) and spectral lines
Hβ 4861 Å, OII 4651 Å and OIII 3447 Å (b) in air plasma at the atmospheric pressure.4
temperature range (Fig.1.9) and this also determines the error of the
method equalling 7–10%.
1.3 Width and shift of spectral lines

Doppler half width. As a result of random movement of the radiat-
ing atoms (ions), the spectral line broadens as a result of Doppler’s ef-
fect. The intensity distribution in the Doppler contour has the form
PD (w) =
1 LM
Dw2
exp - 2 .
OP
p Dw D N
Dw D Q (1.42)
w 0v
where Dw = w – w 0 ; Dw D = ; w 0 is the centre of the line;
c
v = (2 RT / m )1/ 2 is the most probable velocity of random movement of
the atom; R is the gas constant; c is the speed of light; µ is the atomic
weight of radiating particles. Dw D is the Doppler e-half width, i.e. at
the distance Dw = Dw D from the centre of the line w 0, the intensity is
e times less than the maximum intensity. Doppler half width, i.e. the
width of the line on the level of half of its intensity, expressed in wave
length, is
vl0 T
Dl D = 2 ln 2 = 7.16 × 10-7 l 0 . (1.43)
c m
Measuring the half width of the Doppler line, equation (1.43) can be
used to determine the kinetic temperature of the radiating atoms (ions).
The accuracy of temperature determination
22
dT d ( Dl D )
=2 (1.44)
T Dl D
is ~10%.
The Doppler contour (1.42) has the form of Gauss’ distribution in
contrast to the dispersion profiles determined by broadening at colli-
sions. Therefore, it is quite easy to determine whether the contour is
of the Doppler type. The Doppler half width can be separated if the
collision broadening is of the same order as the Doppler broadening.
Doppler broadening prevails in low-density plasma (small number of
collision) and at high temperature. The Doppler half width is greater
for the atoms with a low mass (T = 10000 K):
Element H O Fe Kr
Dl D, Å 0.36 0.09 0.05 0.04
For elements with unfilled internal shells, spectral lines with the purely
Doppler contour are observed even at high pressure. These lines are de-
termined by transitions between the levels not associated with the outer
electrons, and do not broaden under the effect of collisions. The use
of iron lines for determining the atom temperature from Doppler half
width in an electric arc at the atmospheric pressure is a well known
example. 4
In experiments it is important to use apparatus with a high resolu-
tion.
Stark half width. Broadening of the spectral lines under the effect
of charged particles (Stark broadening) is the main type of broaden-
ing in high-pressure plasma. The theory of Stark contours has been de-
veloped most extensively for hydrogen atoms. The error of theory is
estimated for half width at 7–10%. The corresponding contours can be
satisfactorily described by a dispersion (Lorentz) distribution
g / 2p
PL (w) =
+F I
2
.
g
(Dw)
H 2K
2 (1.45)
Here g is the Stark half width. When expressed in the wavelength, it

is equal to
Dl S = C -3/2 ( N e , T ) N e2 / 3 . (1.46)
23
Plasma Diagnostics
Figure 1.10 shows the dependences of the Stark half width of spectral
lines Ha and Hb on the electron concentration with broadening parameters
C (N e, T) from Ref.5. These graphs are used to determine the electron
concentration from the measured half widths of the spectral hydrogen
lines. The broadening parameter C (N e, T) for line H b is almost inde-
pendent of temperature. Since the contour Hb is not symmetric, it is rec-
ommended to use its short wave half (Fig.1.10). The method of deter-
mining the electron concentration from the Stark broadening of the
spectral lines does not require the presence of Maxwell’s velocity dis-
tribution of the electrons. The state of plasma can be evaluated by
measuring the electron concentration by this method and comparing it
with the results of measurements by another method assuming that
plasma is equilibrium.
The error of measuring the electron concentration is determined by
varying the equation (1.46):
dNe 3 LMFG dDl IJ FG dDl IJ OP.

MNH Dl K H Dl K
= +
Ne 2 ran sys PQ (1.47)
I rel
Fig .1.10 Dependence of the half width

ig.1.10
of spectral lines H α and H β (Å) on the
electron concentration (cm –3 ) at
different plasma temperatures. 5
24
Setting (dDl/Dl) sys» 7 ¸ 10%, (dDl/Dl) ran » 5%, we have (dN e/N e) »
15 ¸ 20%. Determination of the electron concentration from the Stark
half width of hydrogen lines is one of the most convenient and accu-
rate methods. It should be noted that hydrogen radiation is almost al-
ways present in plasma due to the presence of small amounts of hy-
drogen or water impurities in gases and other substances. The hydro-
gen lines are wide so that they are easy to record and process.
The Stark broadening of the spectral lines of other elements is smaller
and the values of half widths are determined with a larger error. There-
fore, it is not recommended to use these lines for diagnostics.
Doppler shift. In macroscopic movement of plasma in relation to
the observer the spectral lines carry out Doppler shift whose value is
v
Dl = l 0 cos j. (1.48)
c
Here v is the plasma velocity, j is the angle between the direction of

plasma movement and the observation beam, l 0 is the position of the
centre of the non-shifted line (observation normal to the direction of
plasma movement).Using equation (1.48), the plasma velocity can be
determined from the measured shift of the spectral line. The error of
the method
dv dDl
= (1.49)
v Dl
is 5 1 0%. The modern methods make it possible to measure the mini-

mum shift 4 Dl ~ 0.005 Å which at l 0 = 5000 Å and j = 0 gives the
lower restriction of the measured velocity v min = 300 m/s.
1.4 Continuous spectrum

The continuous spectrum of low-temperature plasma forms as a result
of photorecombination of the electron with the ion and deceleration of
the electron in the ion field. The emission factor of the continuum formed
during this process is written in the form 9
64p 3/ 2 e6 Ne2
en = x( n) , n £ ng ,
3 6 me3/ 2 c3 kT
g
en =
64p 3/ 2 e6
x af
n
Ne2 LM
( n - ng )h OP
, n ³ ng .
(1.50)
3 6 me3/ 2 kT
exp
N
kT Q
25
Plasma Diagnostics
Function x (v) and boundary frequency n g take into account the spe-
cific features of the energetic levels of the atom. For example, for argon
n g = 6.8·10 14 s –1 and the values of the function
ξν 1 0.73 0.88 1.15 1.47 1.7 1.75 1.9 2.1 2.05 1.85
ν · 1 0 –5, s–1 0 0.15 0.25 0.375 0.5 0.6 0.65 0.75 1.0 1.25 1.4
Equation (1.50) makes it possible to determine the electron concentration

from the measured absolute intensity of continuum and the known plasma
temperature. Varying the equation, we obtain the estimate of the error
FG
dN e 1 d e n d x 1 d T IJ
Ne
=
2 en
+
H
+
x 2 T
,
K (1.51)
which gives dN e/N e » 15% at de n/e n » 15%, dx/x » 10%, DT/T » 10%.
Relatively high accuracy is also obtained from the known temperature
approximation. This method does not require that LTE exists in plasma.
Plasma temperature can be determined from the decrease of continuum
beyond the boundary frequency. From equation (1.50) at n > ng we have
FG ( n - n g )h IJ
e nrel = exp -
H kT K
or
hn g hn
ln e nrel = - . (1.52)
kT kT
Plotting the dependence of ln e nrel on hn, we obtain a fragment of

the straight line whose slope gives the temperature. The accuracy of
temperature measurement by this method depends on the frequency range
n 1 ¸ n 2 in which the distribution continuum intensity is examined.
According to (1.52), we can write
e 2 hn2 - hn 2 Dhn
ln = = ,
e1 kT kT
26
and therefore
dT kT d (e 2 / e1 )
= . (1.53)
T Dhn e 2 / e1
For example, at Dhn/kT = 2 and de/e » 10%, the error dT/T » 5%. This
method gives the electron temperature for non-equilibrium plasma.
At the known composition of LET plasma en from (1.50) can be rep-
resented as a function of temperature
e n = e n (T ) (1.54)
and the plasma temperature can then be determined from the absolute
intensity of continuum. Figure 1.7 shows the dependence en(T) for argon
plasma at atmospheric pressure. It also shows the transfer coefficient
of the error (e n/T)(¶T/¶e n) from equation (1.55) for estimating the er-
ror of temperature determination. Varying (1.53) we obtain
LMF de I FG de IJ OP
MNGH e JK
dT e n ¶ T n n dx
T
=
T ¶e n n ran
+
He K
n sys
+
x PQ
. (1.55)
Accepting that de n / e nsys » 5% , de n / e n sys » 10%, dx / x » 10% we obtain

that at T < 130000 K dT/T » 3%, according to Fig.1.7.
1.5 Kirchhoff law

There is a relationship between the emission and absorption factors
(Kirchhoff’s law)
en 2 p hn 3 / c 2
= Bn , Bn = . (1.56)
kn exp(hn / kT ) - 1
Here B n is the radiation of an absolutely black body at frequency n and

at temperature T. Measuring e n and k n, we can determine the plasma
temperature from equation (1.56). If exp (hv/kT) >> 1 we have
en
= (2 phn 3 / c 2 ) exp( - hn / kT ), (1.57)
kn
27
Plasma Diagnostics
and therefore
hn / k
T= . (1.58)
ln(2 phn / c 2 ) - ln ( e n / k n )
3
The error estimate follows from the variation of (1.57)
dT kT de n dk nFG IJ
=
T hn e n
+
kn H
.
K (1.59)
At hn/kT = 4 (l » 3000 Å, T = 10 000 K), de n/e n » 15%, dk n/k n » 5%,

we have dT/T » 5%. In non-equilibrium plasma, this method gives the
electron temperature.
1.6 Spectral absorption lines

The atom concentration at the periphery of a plasma object outside the
radiation zone can be determined from the spectral absorption lines. For
this purpose, the object to be examined is irradiated with external ra-
diation with a continuous spectrum. Absorption lines are obtained in
a background of a continuous spectrum and are characterised by the
total absorption value
z FGH IJ
¥
I (w)
Aw =
0
1-
I0 (w)Kdw, (1.60)
where I 0 (w) is the intensity of the continuous spectrum outside the

absorption line, A w is the area defined by the contour of the absorp-
tion line (Fig.1.11). The value of Aw does not depend on broadening and
the deformation of the absorption line by the apparatus function of the
recording device. The absorption lines can be obtained on spectrographs
with a relatively low resolving power. Only the linear dispersion of the
device must be sufficiently high to ensure the required measurement
accuracy. The value of the total absorption Aw depends on the concen-
tration of the absorbing atoms N, the length of the absorbing layer l,
the power of oscillators of the spectral lines f and the parameters of
the contour of the absorption line. For spectral lines with Voight’s
contour, i.e. when their broadening is determined by the collision proc-
esses and the Doppler effect, the contour of the line is characterised
by the parameter
28
F ig .1.11 Contour of the absorption line.

ig.1.11
Dw L
a= , (1.61)
Dw D
where Dw D = (w 0 /c)(2RT)/µ) 1/2 is the Doppler e-halfwidth, Dw L =

4s2Nj(2pRT (1/µ + 1/µ j))1/2 Lorentz halfwidth, Nj, µj is the concentration
and atomic weight of the perturbing particles, s 2 is the effective cross
section of ‘optical’ collisions of the atoms of the examined element with
perturbing particles. Total absorption as a function of the number of
absorbing atoms can be calculated for different parameters a (the so-
called growth curves). 4 Figure 1.12 shows the growth curves for a ho-
mogeneous absorbing layer. These curves make it possible to determine
the number of atoms from the measured total absorption of the line,
if the gas temperature, parameter a and the power of oscillators f of
the corresponding transition are known.
The spectrograms,obtained by the method of transverse photography
in irradiating the heterogeneous object by radiation by a continuous spec-
trum (Fig.1.2a), record the absorption lines. The distribution of the total
absorption Aw along the line is determined. Photometric profiles are then
constructed for all sections of the absorption line adjacent to each other.
The characteristic curve of the photolayer is used for transition to relative
intensities, and total absorption is determined by measuring the area
defined by the photometric curve of the corresponding section of the
absorption line (Fig.1.11). The parameter a (1.61) and the Doppler
e-halfwidth are calculated from the known temperature and the concen-
tration of the absorbing atoms is determined from the measured total
absorption using the corresponding growth curve (Fig.1.12). A simplified
method of taking into account the non-uniformity of the asymmetric
object and determining the radial distribution of atom concentration in
the absorption line was described in Ref.8. The cross section of the arc
column is divided into annular zones of equal width and it is assumed
that the gas temperature and the atom concentration are constant within
each zone (Fig.1.13). The value (A w /2Dw D) 1 is determined from the
29
Plasma Diagnostics
F ig .1.12 Total absorption A a as a function of the number of absorbing ions N l .

ig.1.12
F ig .1.13 Calculation of the concentration of atoms at the absorption line.

ig.1.13
measured total absorption for the outer zone A w,1 and gas temperature
T 1 and the value (Nfl/Dw D) 1 is then taken from the growth curve cor-
responding to parameter a. This value is then used to determine the total
number of absorbing atoms (Nl) 1 and the concentration N 1 = (Nl 1)/l 11,
where l 11 is the length of the chord of the first zone (Fig.1.13). In de-
termining the number of absorbing atoms from A w,2 it is assumed that
the temperature along the absorbing beam is constantly equal to the mean
temperature D T2 = (T1 + T2 ) / 2 . The value (Nfl/Dw D) 2 is determined for
this temperature. This is followed by calculating the total number of
absorbing atoms (Nl) 2 and the concentration for the second zone N 2 =
(Nl) – 2N 1l 21 )/l 22. Thus, the atom concentration is gradually calculated
for all angular zones and each time it is assumed that the gas temperature
along the absorbing beam is constant and equal to the mean value with
respect to all intersecting zones.
30
The error of the method is evaluated using the equation on the right
hand part of the growth curve
F 4p e aDw
A =G
2 2 I
N f lJ
1/ 2
w
H mc e
D
K . (1.62)
The variation of the equation (1.62), taking into account that param-
eter a is inversely proportional to temperature, gives
dN dA 1 dT df da
=2 + + + . (1.63)
N A 2 T f a
Assuming in accordance with specific experiments 8 dT/T » 20%,

dA/A » 7%, df/f » 10%, da/a » 10%, we obtain dN/N » 50%.
Figure 1.40 shows the results of determining the concentration of so-
dium and copper atoms from the absorption lines in the cross section
of column of a carbon arc. 8 Plasma temperature was measured on the
basis of the ratio of the intensities of copper lines CuI 5106 Å and CuI
5218/20 Å (crosses in Fig.1.14) and by the Lorentz method from the
sodium line NaI 5890/96 Å (points in Fig.1.14), and the concentration
of the sodium and copper atoms was measured from the absolute
intensities of the lines NaI 4668 Å and CuI 5106 Å (crosses) and from
the absorption lines of NaI 3303 Å and CuI 3243 Å (points). It may
be seen that the absorption method greatly widens the spatial region of
measurements.
F ig .1.14 Distribution of temperature and concentration of sodium and copper atoms

ig.1.14
in the cross-section of the carbon arc. 8
31
Plasma Diagnostics
Chapter 2
KINETIC ENERGY, TRANSLATIONAL RELAXATION AND

DOPPLER BROADENING OF LINES IN NON-
EQUILIBRIUM PLASMA SPECTRA
2.1 Introduction
The nature of the translational motion of neutral particles in slightly
ionised low-temperature plasma influences directly or indirectly all proc-
esses taking place in it and in many cases this effect is very strong.
Therefore, any detailed examination of plasma objects includes the
determination of the energies of translational motion of the atoms and
molecules. In a large number of cases this is used to determine the tem-
perature of the gas component. This is carried out taking into account
that the translational relaxation is a fast process and leads to Maxwell’s
distribution with the temperature that is the same for all neutral com-
ponents.
There are a relatively large number of methods of measuring the gas
temperature. In the majority of cases, these methods are indirect in the
sense that they are not related directly to determining the velocity dis-
tribution of the particles and use only the features of manifestation of
different orders of the moments with respect to the distribution. They
include calorimetric, thermal probe, pressure gauge and interferometric
methods of measuring the particle density in different quantum states,
etc. 1,2
One of the widely used methods of measuring the gas temperature
is based on examining the form of the line contours with Doppler broad-
ening. It is well known that if the distribution of emitting particles with
respect to the velocity v z in the z observation direction is given by the
function (v z), then the contour of the radiation lines is 13
af c FG n - n0IJ
j n =
n0 H
Y c
n0 K
. (2.1)
Here n is the observed radiation frequency, c is the velocity of light,
32
Kinetic Energy, Translational Relaxation and Doppler Broadening...
n0 is the radiation frequency of the particle stationary in the examination

system. The applicability of the initial equation (2.1) is restricted by
the conditions of smallness of natural, impact and Stark broadening.
In addition, it is implicitly assumed that the radiation frequency is greatly
higher than the collision frequency, i.e. during radiation the particle does
not change its velocity or, otherwise, 2 l >> l, where l is the free path,
l is the wavelength. For the visible region of the spectrum
l ~ 10 –4¸10 –5 cm, and the last condition is justified up to the pressures
of the order of atmospheric pressure. We shall assume that these as-
sumptions are valid. The relationship between the form of the contour
of the Doppler-broadened line and the distribution function of the par-
ticles with respect to the velocity modulus P(v) is given in the general
case by the integral equation 4,5
za
¥
faf
K v, n P v dv = j n , af m=
n - n0
c
. (2.2)
m
The structure of the kernel K (v,n) is determined by the anisotropy of

the particle velocity. Several simplest cases were examined in Ref.5.
In the isotropic case, equation (2.2) assumes the form
P (v)
∞
∫ dv = ϕ(ν) (2.3)
µ
v
It is relatively simple to calculate (analytically in many cases) the

contour of the line for the derivative P(v). Usually, the Doppler con-
tour refers to the contour of the line corresponding to the Maxwell
distribution of the thermal velocities of radiating particles
 m 
3/ 2
 mv2 
PM (v ) =  
2
v exp  , (2.4)
 2πkT   2kT 
where m is the mass of the emitter, T is the gas temperature, k is Boltz-

mann constant. Integration of (2.3) with P(v) in equation (2.4) gives
the well known equation
c  m 
1/ 2  mc 2 (ν − ν )2 
ϕ (ν ) =  exp − 0
.
ν0  2πkT   2 kT ν 0 
(2.5)
33
Plasma Diagnostics
The width of the contour at half height Dn D
2 n0 2 ln 2 × kT
Dn D = , (2.6)
c m
Usually, (2.6) is also used to determine the gas temperature. How-

ever, the applicability of (2.6) both from the viewpoint of the Maxwell
nature of P(v) and with respect to the equality of the temperatures of
the emitter and the main mass of the non-excited gases has been
examined far less frequently. It should be noted that neither of these
aspects is ‘exotic’. In fact, as a result of the relatively low frequency
of intermolecular collisions in reduced-density plasma, the velocity dis-
tributions of the short-lived electronically-excited and non-excited par-
ticles may differ. Therefore, the specific features of the nature of move-
ment of the electronically excited particles are superimposed on the
profile of the Doppler broadened lines. These special features are de-
termined by both the rate of the relaxation processes of the special fea-
tures of the excitation mechanism of the electronic states.
2.2 Determination of the type of velocity distribution of

excited particles from spectral measurements
An inverse problem – determination of P(v) from the measured contours
of spectral lines, is in the mathematical sense an incorrect (according
to Adamar) problem. Its solution consists of two stages. In the first stage,
the experimentally recorded contour of the line f(v) is used to deter-
mine its true form j(v), i.e. apparatus effects are excluded. This is re-
duced to deconvolution of the given convolution with the difference kernel
∞
f (ν) = a (ν −ν′) ϕ(ν′) dν′ = f0 (ν) +ε (ν),
∫ (2.7)
−∞
where a(v) is the apparatus function, e(v) is the noise with the zero mean.
The second stage consists of determining P(v) from (2.2) or (2.3) which
requires differentiation of the experimentally determined j(v), i.e. the
procedure also leading to instability of the solutions in relation to the
experimental data errors. The progress in solving the problem exam-
ined in this section is determined to a large extent by the recently de-
veloped mathematical methods of solving such inverse problems using
various regularisation methods (for example, see Ref.6–8). These methods
and corresponding algorithms are also discussed to some extent in
Refs.2,5,9) and in Chapter 20 of this book.
34
The experimental method of examining the form of the contours of

the lines of emission spectra is relatively conventional and usually in-
cludes a high-resolution spectrometer linked with a Fabry–Perot inter-
ferometer. In photoelectric registration, the base of the interferometer
can be changed by an electrical method or by varying the pressure. 2
An important problem in these measurements is the reliable deter-
mination of the apparatus function a(v) – a classic problem for which
at least two new approaches have recently been proposed. Firstly, it is
the use of laser radiation with a stable neutral frequency in a narrow
line. Secondly, it is the application of the already mentioned mathematical
method of deconvolution of the convolution (2.7) in which the type of
j(v) is known with a high degree of reliability. 2,9 Of course, as
previously, the task of the experimentator is to ensure the narrowest
apparatus function so that the error in the final result can be reduced.
2.3 Velocity distribution of atoms and molecules in excited

electronic states
If the process of excitation of the radiation state takes place in a re-
action with the generation of a surplus energy
A + B ® C + D + DE , (2.8)
then the laws of conservation of the energy and the pulse show that the
resultant reaction product should acquire an additional kinetic energy.
If the radiating state has a short lifetime and the efficiency of colli-
sion processes is low, it should be expected that anomalous Doppler
broadening of the spectral lines will take place. An example of such
a phenomenon is shown in Fig.2.1. It shows the results of measurements
of the form of the contour of the transition line 2 in the oxygen atom
O (3 3P–3 3S) in a discharge when the following mechanism takes place
( ) ( ) ( )
Ar 3 p 5 4 s + O 2 3 P → O 3 3 P0,1,2 + Ar ( S ) + 0.6
1
0 eV. (2.9)
All curves are normalised with respect to the unit area. It may be
seen that the width of the contour of the Doppler-broadened lines de-
creases with increasing gas pressure, indicating that relaxation colli-
sion processes take place. For comparison, the contour of the line of
the same transition with normal Doppler broadening, corresponding to
the gas temperature of T = 330 K, is shown. It should be noted that
in most cases and under the same conditions the width of all three lines
of the transition 3 3P 0,1,2–3 3S 01 coincide.
35
Plasma Diagnostics
j,rel. units
P, rel. units
F ig .2.1. Contours of the spectral lines of radiation of atomic oxygen (3 3P 0,1,2 – 3 3 S 10 )

ig.2.1.
in a discharge in a mixture of O 2 –Ar (1:36). Discharge current i = 20 mA, pressure
of the mixture p = 0.2 torr (1); 0.5 (2); 3 (4); 10 (5); 15 (6); 7) Doppler contour
of the radiation line of the oxygen atom for the Maxwell distribution at a gas
temperature of T = 330 K; walls of the pipes with an inner diameter of 20 mm are
cooled with running water.
F ig .2.2. (right) Velocity distribution functions of excited oxygen atoms under the
ig.2.2.
conditions shown in Fig. 2.1. Pressure p = 0.2 torr (1), 3 (2), 5 (3), 10 (4) 15 (5);
6) Maxwell function at T = 330 K.
Figure 2.2 shows a set of velocity distribution functions of O( 3P)

atoms obtained in processing the contours shown in Fig.2.1. The Maxwell
function of the distribution of oxygen atoms at T = 330 K is also shown.
All curves are normalised for the unit area. The real distribution func-
tions of the O (3P) atoms are considerably wider than the Maxwell dis-
tribution. The width of the distribution gradually decreases with increas-
ing pressure. Breaks of the function in Fig.2.2 correspond to the val-
ues of intensities at the wings of the line contours (see Fig.2.1), where
the signal : noise ratio is ~1.
Thus, if the results of measurements of the Doppler width of the lines
in the radiation spectra excited in the non-equilibrium plasma in the proc-
esses with non-resonance interactions of heavy particles are used to
determine the neutral gas temperature, the results may be highly inac-
curate. In this example, the widths of the lines, recorded at low den-
sities, correspond to a temperature of ~2000 K, whereas the real gas
temperature is considerably lower.
36
Similar investigations of the anomalous Doppler broadening of the

lines in the spectra of molecular nitrogen and the examination of the
velocity distributions of N 2 molecules (C 3P) in a discharge were de-
scribed in Ref.9, 10.
2.4 Measurement of gas temperature from Doppler

broadening of spectral lines
A direct electron impact is one of the important channels of exciting
the electronic states of the molecules and atoms. A collision between
an electron and an atom is accompanied by a change of the velocities
of the colliding particles. This change may influence the form of the
spectral lines emitted by the short-lived excited particles for which the
collision of relaxation is ineffective. We shall estimate the nature of this
effect.
The pulse conservation law indicates that the velocity of the atom
after collision with an electron is
H H m H H
v ′ = v − e ( ve′ − v e ) , (2.10)
M
H
where v is the velocity of the atom prior to collision, M and m e are
H H
the masses of the atom and the electron, ve and ve¢ are the velocities
of the electrons prior to and after collision with the atom. The cases
with v e' << v e and v e' ≈ v e are most interesting for various applications.
The first case is observed in low-temperature gas-discharge plasma where
the mean electron energy is usually lower than the excitation thresh-
old of the radiating electronic state of the molecules and atoms. The
main role is played by the electrons with the energy slightly higher than
the threshold energy because in this region the electron concentration
rapidly decreases with increasing electron energy. The second case in
which v e' ≈ v e is realised in exciting the molecules with a beam of fast
electrons with the velocities v e' >> v oe (v oe is the velocity of atomic elec-
trons). These cases are encountered in, for example, probing the gas
flow with an electron beam.
In the first case, equation (2.10) shows that the relative variation
of the atom velocity ∆v/v in collision with an electron is
Dv me ve
= . (2.11)
v M v
In estimates, value v will be represented by the most probable ve-
37
Plasma Diagnostics
2 kT
locity of thermal motion v H = , and v e by the value of the elec-
M
tron velocity corresponding to the excitation threshold of the examined
states v the. In this case
Dv m v th
Dw = = e e (2.12)
vH 2 kTM
The following table gives the values of ∆w for some atoms and mol-
ecules at two gas temperatures T 1 = 100 K and T 2 = 300 K (v the cor-
responds to the excitation pressure of the state emitting the spectra) often
used for plasma diagnostics:
At om ,
Ex c i ta ti on DM ( %) a t
m ole cule s, L A t h × 1 0 8, c m s 1 DM ( %) a t 300 K
thr e shold, e V 100 K
e x cite d sta te
H ( n = 4) 12.75 2.12 90 52
H 2 (@ !Π) 13.87 2.21 66 38
D 2 (@ D )3
13.88 2.21 47 27
H e ( 3@ Π) 1
23.07 2.85 60 35
N 2( C Π)
!
11.03 1.97 16 9
C O( B Π)
1
10.77 1.88 15 9
It may be seen that, irrespective of the difference in the masses be-

tween the heavy particles and the electrons, in the process of excita-
tion of the atoms and molecules the velocity of translational motion can
greatly change. The value of ∆w increases with a decrease of the mass
of the atomic particles, with an increase of the excitation threshold of
the state and with a decrease of gas temperature. For example, for a
hydrogen atom at 100 K it is 90%, for a helium atom 60%, and for a
heavier nitrogen molecule at T = 300 K it is does not exceed 10%.
The excitation of electronic states of atomic particles by fast elec-
trons (this corresponds to the case with v e' ≈ v e has been studied quite
sufficiently. The problem permits Born’s approximation to be used, and
its solution has been analysed in detail in Ref.11. The main role is played
here by collisions causing scattering of the electrons through small angles
with a small momentum transfer. The maximum value of the transferred
momentum Pmax ≈ mev0 (v0 is the value of the order the velocity of atomic
electrons). This result is physically evident. The process of collision
with a transfer of momentum that is considerably greater than the mo-
38
mentum of atomic electrons, would lead simply to ionisation of the atom

or molecule. In fact, under the examined conditions where the veloc-
ity of the beam electrons ve is considerably higher than the velocity of
the atomic electrons v ae, the atomic electrons can be regarded as free,
and the collision with the atom as the elastic collision of the incident
electron with the initially stationary atomic electrons. In collision with
the transfer of a large momentum, both electrons (incident and atomic)
acquire velocities of comparable magnitude and this leads to ionisation.
Thus, to evaluate the relative variation of the velocity of the atoms
excited by the beam of fast electrons, we can use equation (2.12) since
v 0 ≈ v the.
We can now examine the behaviour of the Doppler contours of the
spectral lines. For this purpose, it is necessary to find the distribution
function of the excited atoms and molecules with respect to the com-
ponent of the velocity of translational motion along the observation axis
oz. We assume that the distribution prior to the interaction with the elec-
1   v  2 
trons is of the Maxwell type Ψ M (v z ) = exp  −  z   . . As a re-
π
  H  
vH v
sult of interaction with electrons, the velocity of the excited atom

changes. We also assume that the lifetime of the excited state of the
atom is shorter than the characteristic free path time, i.e. the excited
particles are not subjected to any collisions. It may easily be shown
that the distribution of the radiating atoms and the molecules is described
by the equation
∞
Ψ (v′z ) = ∫ Ψ ( v ) Ψ ( v , v′ ) dv ,
M z e z z z (2.13)
−∞
where the function Ψ e (v z, v'z) is the distribution of the excited atoms

with respect to the velocity component v'z under the condition that the
velocity of the atoms prior to collision with the electron was v z. Us-
ing (2.10) for the isotropic velocity distribution of electrons, we ob-
tain
 1
 , if v ′z − v z ≤ ∆ v ,
Ψ e (v z , v ′z ) =  2 ∆ v
(2.14)
 0, if v ′z − v z > ∆ v.

Substituting Ψ M (v z) and Ψ (v'z) into (2.13), we finally have
39
Plasma Diagnostics
v′z +∆v
1   v  
Ψ (v′z ) = ∫ exp −  z   dv z . (2.15)
2 π∆vvH v′z −∆v   vH  
As a result of comparing Ψ M (v z) and Ψ (v'z), we can conclude that

the velocity distribution of the excited atoms differs from the identi-
cal distribution of the non-excited atoms, and the distribution Ψ (v'z)
is non-Maxwellian.
Equation (2.15) shows that the contour of the spectral line is described
by the equation
1 v′
ϕ ( w ) dw = erf ( w + ∆w ) − erf ( w − ∆w ) dw, w = (2.16)
4∆w vH .
∞
Function (w) is normalised for the unit area, i.e. ∫ ϕ ( w) dw = 1 .
−∞
If
the projection of the velocity onto the observation axis does not change
during excitation, i.e. v'z = vz (excitation by the electron beam), the line
contour will be of the conventional Doppler type (Gaussian)
1 − w2
ϕ D ( w ) dw = e dw. (2.17)
π
Figure 2.3 shows the contours of spectral lines normalised for the
unit area. Curve 1 corresponds to the Gaussian contour (2.17), curves
2 and 3 to the contours (2.16). It may be seen that the contours de-
scribed by equation (2.16) are wider than the Gaussian contour and the
width of the contours increases with increasing ∆w. For example, for
∆w = 0.3 the difference from the Gaussian contour in respect of width
is ~3%, and at ∆w = 0.6 it is 13%.
Taking the actual measurement error into account, additional broad-
ening should be considered at w > 0.3. Figure 2.4 shows the actual gas
temperature dependence of the systematic error DT/T of determination
of temperature from the width of the spectral lines of some atoms and
molecules, associated with examining the above factor.
It can be seen that if measurements are carried out using the spec-
tral lines of relatively heavy molecules (N 2, CO), the systematic error
at the temperature found under the conditions of gas-discharge plasma
in practice is negligible. A different situation exists for light particles
at a low gas density. For example, DT/T > 10% for the H atom at
40
F ig .2.3 Contours of the spectral lines. 1) Gauss contour; 2,3) contours described
ig.2.3
by equation (2.16) at ∆ w = 0.3 and ∆ w = 0.6, respectively.
F ig .2.4 Dependence of the systematic error D T / T of determination of temperature

ig.2.4
on the basis of the width of the spectral lines H (1), H 2 (2), He (3), D 2 (4), N 2 and
CO (5) on gas temperature.
T < 550 K, for H 2 at T < 330 K, for He at T < 250 K, for D 2 at

T < 150 K. In Ref.12 these conclusions were confirmed by experiments.
2.5 A simple model of relaxation of the mean kinetic

energy. Effect of gas density
These considerations show that the processes of excitation of the elec-
tronic states of the atoms and molecules are accompanied by
perturbations of the velocity distributions of the particles. The real
distributions form under the effect of two factors. Firstly, excitation is
accompanied by the formation of particles with specific velocity dis-
tributions. Secondly, during the lifetime of the excited particles these
41
Plasma Diagnostics
initial distributions relax as a result of collisions with the particles and

transform to the form observed in experiments. In this case, especially
in the processes of excitation in interaction of heavy particles, the initial
deviations from Maxwellian can be large.
In Ref.13, the authors obtained general equations for the velocity dis-
tribution functions of the product of bimolecular reactions. The cor-
responding distributions for a number of modelling interaction potentials
can be determined numerically or analytically. Not discussing this aspect,
we shall only note that analysis carried out in, in particular, Ref.2, 14-
16 on the basis of the Boltzmann equation shows that for the major-
ity of cases interesting for practice the evolution of the mean energy
can be satisfactorily described using the approximation in which the
sequence of distributions in the relaxation process can be treated as a
sequence of Maxwell distributions and the relaxation interactions are
examined on the basis of the model of solid spheres.
Accepting these approximations, it is easy to construct a relatively
simple model of relaxation of the mean molecular energy in the medium
of a buffer gas at large deviations from equilibrium.
We examine the time dependence of the mean energy of translational
motion DE N of molecules N during their movement in a buffer gas
dE N = - n M v~N , M s N , M DE dt . (2.18)
Here nM is the concentration of the molecules of the buffer gas; v~N , M

is the mean velocity of the relative movement of N and M; DE N is the
mean energy lost by the molecule during a single collision; σ N,M is the
gas kinetic collision sections of N with M. We shall use the results of
Ref.17 in which the authors also examined the model of solid spheres
and it was assumed that both fast particles and the cold gas are de-
scribed by Maxwell velocity distributions:
1/ 2
m T 
vN ,M = vN  N M + 1 (2.19)
 mM TN 
(m N, m M are the mass of the particles N and M; T M, T N are the corre-

sponding temperatures, v N is the mean velocity of the molecules N),
 T  8 mN mM
∆E N = E N µ  1 − M , µ = . (2.20)
 TN  3 ( mN + mM )2
42
Taking into account (2.19) and (2.20), the relaxation equation is con-
verted to the form
1/ 2
−1/ 2  m 
dE
− N = 4nM σ N ,M µ (3πmN )  N EM + EN  ( EN − EM ). (2.21)
dt  mM 
It should be noted that if deviations from equilibrium are small, i.e.

E N » E M , equation (2.21) changes to the well-known relaxation equation
with the constant relaxation time τ rel (see, for example, Ref.11):
−
dEN
dt
=−
1
τrel
( E N − EM ) , (2.22)
1/ 2
1/ 2  
= 4nσ N ,M µ ( EM / 3πmN )
−1 mN
where τrel  + 1 .
 mM 
Solution of equation (2.21) at the initial conditions E N (t = 0 ) = E N0
is written in the following form:
{ }
2
 mN  c′ exp t τrel + 1
−1
m
EN (t ) = EM 1 +  − N EM ,
{ } (2.23)
2
 mM   c′ exp t τ −  m
rel − 1
1 M

where
 
1/ 2
mN  
1/ 2
mN 1/ 2 
′ 
c =  EN +
0
EM  + EM 1 +  ×
 mM   mM  
−1
 
1/ 2
mN  
1/ 2
(2.24)
mN 1/ 2 
×  E N +
0
EM  − EM 1 +   .
 mM   mM  
To ensure determinacy, we assume that N ≡ N 2, M ≡ Ar. Figure 2.5

shows the time dependence of EN 2 (t ) at EN0 2 (t = 0) = 2250 K,
EAr = 150 K . It can be seen that relaxation is very rapid: already af-
ter t = 0.5 τ rel the value of EN 2 decreases e times, and after t = 3τ rel
EN 2 ≅ EAr =.
43
Plasma Diagnostics
τ rel 2τ rel 3τ rel 4τ rel 5τ rel
F ig .2.5 Time dependence of the mean energy of translational movement EN 2 of the

ig.2.5
molecules of N 2 (C 3 Π) in argon.
In comparison with the experiments, carried out under the station-

ary conditions, it should be taken into account that the observed mol-
ecules formed, generally speaking, at different times in relation to the
observation moment (scintillation) and the period during which they
remain in the excited state differs. Therefore, when using equation (2.23)
for analysis of the experimental data, it is necessary to carry out av-
eraging on the basis of the lifetime of the molecules in the excited elec-
tronic state. The probability of an excited particle emitting a photon
during the time t to t + dt is given by the equation
dw (t ) = τ−p1e
−t / τ p
dt , (2.25)
where τ p is the radiation lifetime of the excited state. For N 2 (C 3Π)

τ p = (41 ± 2.9)·10 –9 s. Finally, averaging (2.23), we obtain
(
EN2 nAr , σ N2 , Ar ) = τ−p1 ∫ E (n
N2 Ar , )
σ N2 , Ar , t e
−t / τ p
dt . (2.26)
0
Integral (2.26) can be calculated numerically. Comparing the depend-

ence (2.26) with the experimental data 2,9 for discharges in the N 2–Ar
(1:9) and N 2–He (1:10) mixtures, we can determine the effective sec-
tions of relaxation collisions for excited molecules of N 2 (C 3Π) with
argon atoms s N2 ,Ar and helium atoms s N2 ,He . Figure 2.26 shows the
dependence of the most probable values of the energy for the N 2 mol-
ecules (C 3Π, v’ = 0, K’ = 26) (v’, K’ are the oscillatory and rotational
44
F ig .2.6 Dependence of the most probable values of energy for molecules of N 2 (C 3Π,
ig.2.6
v ’ = 0, K ’ = 26) on the concentration of He (a) and Ar (b). a) calculated from (2.26)
for N 2–He with sections of 10 15 cm 2) s N 2 ,He = 5.8 (1), 3 (2), 12 (3); b) calculated
from (2.26) for N 2Ar with sections (10 –15 cm 2 ) s N 2 ,Ar = 4.1 (1), 2 (2), 8 (3); the
points indicate the experimental results. 29
quantum numbers) on the He and Ar concentration. The best agreements

between the calculations and experimental values was obtained for
s N2 ,Ar =(4.1+1.5)×10 –15 cm 2, = (5.8+0.5)×10 –15 cm 2. It can be seen that
the calculated dependence is sensitive to selection of the cross section.
This makes it possible to determine the effective collision cross sec-
tion at a relatively small error. It should be noted that if in the case
of the N 2–Ar (1:9) mixture the value of s N2 ,Ar almost completely co-
incides with the gas kinetic collision section of the particles in the ground
electronic states ( s 0N2 ,Ar =4.3×10–15 cm2), then for the N 2–He (1:10) mix-
ture there is a difference ( s0N2 ,He =2.8×10 –15 cm 2): the cross section ob-
tained from the described comparison is approximately twice the
s 0N 2 value. It should be noted that the cross sections for the electroni-
cally excited molecule and the molecule in the ground state may or may
not coincide in collisions with another particle because the effective size
of the molecule in the excited electronic state, generally speaking, in-
creases.
2.6 Identification of excitation mechanisms

The proposed relaxation model can be used to determine the mean dis-
tribution energy at the known concentration of the buffer gas. However,
an inverse problem can also be formulated: from the measured mean
kinetic energy of the molecules we can determine the surplus of energy
45
Plasma Diagnostics
generated during excitation. This enables the process leading to molecular

excitation to be identified.
Initially, we examine the process of transfer of excitation from the
atom to the molecule. From the energy conservation law we obtain
Eea = Eemole
,v ,k - E k + E th
0
¢ - Eth0 , (2.27)
where Eae is the energy of the excited state of the atom, Eemol
,v,k is the en-
ergy of the examined electronic–oscillatory–rotational state of the mol-
ecule, E 0k is the rotation energy of the molecule prior to interaction, E0th
are the translational energies of the system of two particles prior to and
after interaction, respectively. The translational energy of the system
of two particles (with masses m 1 and m 2) can always be written in the
form of the sum of the energy of the centre of the masses E c.m and the
energy of the relative motion E µ : 18
E th = E c.m + Em . (2.28)
M S vc2.m m1m2
Here Ec.m = ; M S = m1 + m2 ; E m = mvrel . / 2; m =
2
.
2 m1 + m2 The ve-
H
locity of the centre of the masses vc.m and the velocity of the relative
motion are determined as follows
H 1 H H
vc . m = ( m1v1 + m2 v2 ), (2.29)
MΣ
H H H
vrel = v1 - v2 . (2.30)
H H
where v1 and v2 are the velocities of the first and second particles in
the laboratory reference system.
In interaction of the particles E c.m does not change
Eth¢ - E th0 = Em¢ - E m0 , (2.31)
where E µ0 = (1 / 2 ) v rel µ and E µ′ = (1 / 2 ) v rel

2
0
′ 2 µ are the energies of the rela-
tive motion of the particles prior and after interaction, respectively. From
(2.29) and (2.30) we obtain
H H H M
vrel = (v1 − vc.m ) Σ . (2.32)
m2
46
Substituting the corresponding value of the relative velocity into equa-

tion (2.31) we obtain
Eth′ − Eth0 =
m2 2 
(
m1 M Σ  2 02 H H
)
v1′ − v1 − 2vc.m q1  ,
 (2.33)
H H H
where q1 = v1¢ - v10 .
H
Averaging over all possible directions vc.m for the isotropic case leads
H H
to the disappearance of the term with vc.m q1 , and, consequently
M Σ  m1v1′2 m1v 0 
2
Eth′ − Eth0 =  − . (2.34)

m2  2 2 
 
Combining the equations (2.34) and (2.27) and setting

m1v ¢ 2
Ek0 = kT , (1/ 2 ) m1v10 = (3/ 2 ) kT and
2
= e1z we obtain that
2
 3 MΣ  MΣ z
Eeat = Eemole
,v ,k − kT  1 + + ε1 . (2.35)
 2 M 2  m2
Here ε z1 is the energy at the limit of zero gas concentration. To de-

termine this, energy it is necessary to plot the dependence of the most
probable energy on the concentration and find the asymptote of this de-
pendence at the zero concentration limit. In interaction of nitrogen with
argon for the N 2 (C3Π, v’ = 0, K’ = 26) molecules, we obtain εz1 = 2860
± 300 cm–1, and at T = 150 K equation (2.35) gives Eae = 93264 ± 3600
cm –1 which almost coincides with the energy of the state of Ar ( 3P 2)
equal to 93 144 cm –1. Using other spectral lines of the second positive
system of N 2 for measurements, in particular, those corresponding to
the transitions from high rotational levels (K' ~ 50), we can determine
the boundaries of the range of the states of argon from which excita-
tion is transferred to the nitrogen molecules.
We shall now examine the process of transfer of excitation from mol-
ecule to molecule. In this case, the energy conservation law gives
( )
Ee0,v = Ee′,v − ε 0k (1) − ε 0k ( 2 ) + Eth′ − Eth0 + ε′k (1) + ε′k ( 2 ) , (2.36)
47
Plasma Diagnostics
where E 0e,v and E'e,v are the values of the electronic–oscillatory energy
of the system prior to and after interaction respectively, ε 0k(1), ε'k (1),
ε 0k(2), ε'k (2) are the rotational energies of the first and second molecule
prior to and after interaction, E 0t and E't is the translational energy of
the system prior to and after interaction. Using equation (2.36) and set-
1 3 1
m 2 v1′ 2 = ε1z , ε0k (1) = ε0k ( 2 ) = kT , we obtain
2
ting m1v10 = kT ,
2 2 2
 3 MΣ  MΣ z
E e0,v = E e′ ,v − kT  2 +  + ε ′k (1) + ε ′k ( 2 ) + ε1 . (2.37)
 2 m 2  m2
In transferring excitation from molecule to molecule, in addition to

the translational energy whose value is easily determined from the con-
servation law, the second molecule can also transport the rotational
energy ε'k (2) after interaction. In this case, to determine E 0e,v we must
obtain additional information from the equation, in particular, we must
know the maximum possible value of the translational energy of the
second molecule max [ε'k (2)]. Equation (2.37) shows that the maximum
possible rotational energy of the second molecule is related to the mini-
MΣ z
mum value of the quantity ε1 + ε′k (1) for the first molecule, i.e.
m2
 3 MΣ   MΣ z 
Ee0,v = Ee′,v − kT  2 +  + min  ε1 + ε′k (1) +
 2 m2   m2  (2.38)
+ max ε′k ( 2 ) .
We shall examine a specific example of excitation of the second posi-

tive system of N 2 in a discharge in pure N 2 or in N 2–He mixture. In
the literature, there are at least two different viewpoints regarding the
formation mechanism of the ‘hot’ group N2 (C3Π) in similar discharges:
1) as a result of de-excitation of the molecules which are in the state
N 2(E 3Σ); 19 2) as a result of de-excitation of molecules present in the
states N 2(C' 3Π) or N 2(D 3Σ). 20
The products of these processes are the molecules N 2 (X 1Σ) and
N 2 (C 3Π). The moments of inertia of the molecules N 2 (X 1Σ) and N 2
(C 3Π) are very close and it is therefore natural to assume that on the
whole their rotational distributions coincide. The spectrum of the second
positive system N 2 associated with the transitions of the molecules of
‘hot’ group contains lines up to K'~50 and therefore max [ε'k(2)] = E k
48
(K' = 50). Investigations of the contours of the spectral lines showed

that min [2ε z1 + ε'k(1)] corresponds to the molecules N 2hot (C 3Π, v' = 0,
K' < 18). In particular, for K' = 18 we obtain 2ε z1 (K' = 18) + ε k
(K' = 18) = 2700 cm –1 . Finally, for the energy of the given state we
obtain E 0e,v = 95 800 ± 700 cm –1. This value is in good agreement with
the energy of the state E 3Σ +g equal to 95 772 cm –1 and even if we take
into account the measurement error, it is lower than the energy of the
state C' 3Π (E = 97 580 cm –1) and D 3Σ (E = 103 576 cm –1 ). Thus, the
results show that only the state E 3Σ g+ is responsible in this case for the
formation of the molecules of ‘hot’ group N 2 (C 3Π).
The agreement between the calculated and experimental results makes
it possible to recommend the use of a simple model for analysing the
processes of translational relaxation of the mean energy of the ‘hot’
particles. It should be noted that the examined cases are characterised
by a high non-uniformity: the initial particle energy is more than an order
of a magnitude higher than the equilibrium energy. As shown by com-
parison with the experiments, the simple analytical equations presented
here make it possible to describe satisfactorily the evolution of the mean
energy of ‘hot’ particles.
49
Plasma Diagnostics
Chapter 3
OPTICAL ACTINOMETRY OF PLASMA

A common disadvantage of the methods of determining the concentration
of molecules, atoms and radicals in plasma on the basis of the emis-
sion spectra is that these methods can be used to measure the parti-
cle concentration directly only in the excited state. The problem of de-
termining the particle concentration in ground states (which is often more
important) on the basis of the emission spectra is relatively complicated
because it requires quantitative information on the mechanisms of ex-
citation and deactivation of the particles of the examined type under
the conditions of the studied gas discharge.
The optical actinometry method makes it possible in a number of cases
to bypass these difficulties. The method is based on the following pro-
cedure.1–3 A small amount of the actinometer – a gas with a known spec-
trum and the excitation and deactivation constants – is added to the ex-
amined plasma object. If the emitting states of the actinometer and the
examined gas are excited by a direct electron impact from the ground
state and have similar excitation potentials, the same group of electrons
will take part in populating these states. Consequently, the rate con-
stants of excitation of the emitting states should depend in the same
manner on the plasma parameters.
In cases in which the radiation channel is the main deactivation chan-
nel of these states, the equations for the intensities of the emissions lines
of the actinometer and the emitting gas have the same form:
I ik = N i Aik hnik = N 0 ne sv 0i
Aik t i hnik , (3.1)
where Iik, Aik, νik are the intensities of the lines, the probability of spon-
taneous emission and the frequencies of radiation transition i → k, re-
spectively; N i, N 0 are the populations of the excited and ground elec-
tronic states; n e is the electron concentration; <σv> 0i are the rate con-
stants of excitation of the emitting states by the electron impact from
the ground electronic states; τ i is the radiation lifetime of the emitting
state.
50
Optical Actinometry of Plasma
In accordance with (3.1), the following equation can be written for

the ratio of the intensities of the emission lines of the actinometer (the
values with index 1) and the examined gas (values with index 2):
I1 N01 σv A1τ1ν1
= 01
=
I 2 N02 σv0 Aτ ν
02 2 2 2
∞
∫ ϕ (ε ) f (ε ) ε dε
1/ 2
01
N01 σ01
m
∆E1 A1τ1ν1
= m ∞
. (3.2)
N02 σ02 A2τ2ν 2
∫ ϕ (ε ) f (ε ) ε dε
1/ 2
02
∆E2
In (3.2), the rate constants of excitation are written in the explicit

form 4
∞
σv =σ0mi ∫ ϕ ( ε) f ( ε ) ε dε,
1/ 2
0i 0i (3.3)
∆Ei
where σ m0i is the size of the excitation cross section at the maximum;
∆E i is the excitation threshold of the emitting state; f(ε) is the energy
distribution function of electrons in the plasma; ϕ 0i (ε) is the normal-
ised function describing the form of the dependence of the excitation
cross sections of the emitting states on the electron energy.
The exact solution of equation (3.2) encounters principal difficul-
ties associated mainly with insufficient knowledge of the energy dis-
tribution function of the electrons f(ε). However, these difficulties can
be overcome if the conditions ϕ 01(ε)¾ ϕ 02(ε) and ∆E 1 ¾ ∆E 2 are ful-
filled. As shown in, for example, Refs.4 and 5, the first of these con-
ditions is approximately fulfilled for a large number of atomic and mo-
lecular states. The second condition can be satisfied by selecting the
corresponding excited emitting states of the actinometer and the examined
gas. If both conditions are fulfilled, the following equation can be written
I1 N
= k 01 , (3.4)
I2 N 02
where k is a constant that depends only on the properties of specific

atoms
sm
01 A1t1n1
k= . (3.5)
sm
02 A2 t 2 n 2
51
Plasma Diagnostics
Thus, the absolute concentration of the particles of the examined gas

in the ground electronic state is determined only by the ratio of the
intensities of the corresponding spectral lines and by the absolute con-
centration of the particles of the gas–actinometer
I2
N 02 = k N . (3.6)
I1 01
When examining the molecular states, the equation (3.2) should

slightly be modified owing to the fact that it is necessary to take into
account the excitation of the molecular emitting states by electron impact
from the different vibration levels of the ground electronic states of the
molecules. Taking into account the experimental data presented in Refs.6
and 7 according to which the relative cross sections of excitation of
the electronic–vibration states by an electron impact are proportional
to the Frank–Condon factors of the corresponding transitions, and taking
into account the actual populations of the vibrational levels of the ground
electronic states, we can write a relationship for the relative intensities
of two molecular bands
∞
∞
∫ ϕ ( ε ) f (ε ) ε dε ∑ q ( ) Ψ( ) ( v )
1/ 2 1 1 0
01
v0
I1 N01 σ01
m
∆E1 v0 =0 A1τ1ν1
= m ∞ ∞
.
N02 σ02 A2τ2ν2
∑ q ( ) Ψ( ) ( v )
I2 (3.7)
∫ ϕ (ε ) f (ε ) ε
2 2 0
02
1/ 2
dε v0
∆E2 v0 =0
Equation (3.7) differs from the examined relationship between the

intensities of two atomic spectral lines (3.2) by the fact that it contains
∑ q( )Ψ( ) (v ) , where qI vbi g0 are

∞
i i 0
additional factors containing the sums v0 0
0
v =0
the Frank–Condon factors linking the vibrational levels of the ground

electronic state v 0 with the zero vibrational level of the electronically
excited state from which the emission transition starts; Ψ (i) (v 0) is the
distribution function of populations of the vibrational levels of the ground
electronic state of the examined molecules.
For the relative intensities of the atomic spectral line and the mo-
lecular band, we must write a slightly different relationship
52
∫ ϕ (ε ) f (ε ) ε dε
1/ 2
01
I1 N01 σ01
m
∆E1 1 A1τ1ν1
= .
N02 σ02 ∞
m ∞
A2τ2ν2
I2
∫ ϕ (ε ) f (ε ) ε
02
1/ 2
dε ∑q v0 0
Ψ v0 ( ) (3.8)
∆E2 v0 =0
After appropriate transformations, the equations (3.7) and (3.8) can

be reduced to the relationship (3.6) in which, however, constant k has
a slightly different meaning. For example, for the case described by (3.7)
∞
∞
σ01
m ∫
∆E1
ϕ01 (ε ) f (ε ) ε1/ 2dε ∑ q ( ) Ψ( ) ( v )
1
v0 0
1 0
A1τ1ν1
v0 =0
k= ∞ ∞
.
σ02
m
A2τ2ν2
∫ ϕ (ε ) f (ε ) ε
02
1/ 2
dε ∑ q( ) Ψ( ) (v )
v0 0
2 2 0 (3.9)
∆E2 v0 =0
Evidently, the discussed method was proposed for the first time and
applied in Ref.1 and 8 where the absolute concentrations of oxygen atoms
were determined under the condition of glow discharge plasma in carbon
dioxide.
The authors of Refs.1 and 8 compared the relative intensities of oxy-
gen lines O (3p 5P → 3s 5S) and the bands of the Angström system of
carbon oxide CO (B 1Σ, v’ = 0 → A 1Π, v''). The upper levels of these
emission transitions have similar excitation potentials (energy differ-
ence between them is ∆E ¾ 320 cm –1 and, in addition to this, the ex-
citation cross sections of these states are well known from the litera-
ture). In Refs.1, 8, the given concentration of carbon oxide molecules
were used to determine the concentration of oxygen atoms in relation
to the discharge conditions. The advantage of the proposed method is
that the actinometric gas was represented by CO present naturally in
the discharge.
Optical actinometry was reborn in studies of the diagnostic zone of
non-equilibrium chemically active plasma generated in low-pressure high-
frequency discharges used for etching silicon and silicon dioxide by active
halogen atoms (fluorine, chlorine, bromine). 9–18 It was shown that the
etching rate is proportional to the absolute concentration of the halo-
gen atoms.
53
Plasma Diagnostics
The authors of Ref.2 proposed to add gaseous argon to the plasma

as an actinometer. The intensity of the argon line Ar λ = 7504 Å
(excitation potential ¾13.5 eV) was compared with the emission intensity
of the line of the fluorine atom F λ = 7037 Å (excitation potential ¾14.5
eV). The results show that the concentrations of the fluorine atoms,
determined by the actinometer, in the high-frequency gas discharge
(f = 13.56 MHz) with a power of W = 100 W in a mixture CF 4/O 2 at
a pressure of p ¾ 0.04 torr are in good agreement with the data ob-
tained by other methods.
The authors of Ref.3 carried out detailed examination of the
actinometer properties of argon (λ = 7504 Å) and molecular nitrogen
((0–2) band of the second positive system N 2 (C 3Π→ B 3Π)) for meas-
uring the absolute concentration of F and O atoms and CO and CO 2
molecules in the high-frequency discharge (f = 27 MHz) with a power
of W = 50 W in a mixture of CF 4/O 2 at a pressure of p ¾ 1 torr. In
particular, it was shown that although the excitation energies of the ex-
amined states vary over a wide range (8–20 eV), the concentrations of
F, O, CO and CO 2 can nevertheless be calculated using equation (3.6),
where Ar and N 2 are used as an an actinometer.
The accuracy of the optical actinometer method has been verified
many times by comparison with the independent method of determin-
ing the particle concentration in the ground electronic states. For ex-
ample, in addition to optical actinometry, the concentration of fluorine
atoms in Refs.9 and 10 was determined independently by the titration
method in which CO 2 was added to the initial gas mixture CF 4/O 2 and
as a result of rapid substitution reaction F + Cl 2 → FCl + Cl the au-
thors recorded recombination chemiluminescence of chlorine atoms in
the range 600–700 nm. The glow intensity was used to determine the
concentration of the fluorine atoms. Results show that the actinometry
and titration methods give identical results.
In Ref.11, optical actinometry was used to determine the concentration
of bromine atoms in etching GaAs crystals with bromine. The absolute
concentration of the Br atoms was measured both by actinometry
(argon was used as the actinometer) and by the linear absorption method.
The results show that the ratio of the intensities of the bromine and argon
lines is proportional to the absolute concentration of the Br atoms in
a wide range of rearrangement of the discharge excitation frequency
(f = 0.1 – 13 MHz). In accordance with (3.6) this shows that actinometry
can be used to examine discharges of this type.
The accuracy of the optical actinometry method was verified in Ref.12
by laser-induced fluorescence enabling independent measurement of the
concentration of CCl radicals in CCl 4 plasma (molecular nitrogen was
used as an actinometer. In this case, actinometry could not be used be-
54
cause the excited CCl* radicals formed in the discharge not during direct
electronic excitation but during dissociation of CCl 2, CCl 3 and CCl 4. 12
The general conclusion relating to the results obtained in Refs.9–
12 can be described as follows: under the conditions where it is pos-
sible to use the linear absorption methods or laser-induced fluorescence
it may not be necessary to use optical actinometry. However, as a re-
sult of its availability and simple experimental set up, actinometry is
often more attractive.
An important moment which forms the basis of the method of op-
tical actinometry is the assumption on the excitation of emitting states
by a direct electron impact. In Ref.13, this circumstance was verified
by investigating the form of the contours of the spectral lines of fluorine,
chlorine and argon atoms in the plasma of ac discharge (f = 20–50 kHz)
in CF4/O2/Ar and Cl2/Ar mixtures. It is well known that in a low-pressure
discharge the form of the contours of the spectral lines, excited by direct
electron impact, should have the form of the Gauss function with the
gas temperature used as the parameter. 19 In Ref.13 it was shown that
the contours of the emission lines of the fluorine and argon atoms for
the CF4/O 2/Ar mixture indeed have the Gaussian form with the tempera-
ture T = 360 ± 70 K which corresponds to the gas temperature. At the
same time, the Cl 2/Ar mixture was characterised by large differences
in the contours of the emission lines of the chlorine and argon atoms
in comparison with the Gaussian form at the gas temperature. It was
concluded that the chlorine and argon atoms in the discharge in the Cl2/
Ar mixture are excited during dissociation in the processes of non-reso-
nance collision of heavy particles but not in excitation with the elec-
trons, i.e. the optical actinometry method cannot be used in this case.
Analysis of the strength of the possible effect of the gas–actinometer
on the properties of the examined plasma was carried out in Ref.14 where
the effect of the additions of argon, helium and molecular nitrogen on
the properties of the plasma and the SF6/O2 mixture was examined theo-
retically and by experiments. The results show the effect of argon, up
to 10% content in the working mixture, has almost no influence on the
energy distribution function of the electrons, whereas the additions of
molecular and nitrogen have a strong effect on the properties of SF 6/
O 2 plasma as a result of effective dissipation of the electron energy in
excitation of the vibrational levels of N 2.
Optical actinometry can also be used to identify the excitation chan-
nels of the quantum states of the molecules of chemically reacting gases.
For example, the authors of Ref.15 examined the mechanism of exci-
tation of the N 2 (C 3Π) state (argon was used as the actinometer) and
of the N +2 (B 2Σ) state (actinometer – neon). The results show that the
relationship (3.6) is fulfilled in a wide range of the variation of the dis-
55
Plasma Diagnostics
charge conditions for the corresponding states. Since the emitting states
of argon and neon were excited with the direct electron impact under
the given experimental conditions, it was concluded that the N 2 (C 3Π)
and N +2 (B 2Π) states are also populated by the direct electron impact.
In Ref.20, the absolute concentration of the nitrogen atoms was meas-
ured under the low-pressure arc discharge conditions with a consum-
able cathode made of titanium. Argon was used as the actinometer.
Taking into account the fact that the excitation potentials of the lines
of nitrogen atoms NI λ = 4099.94 Å (E 1 = 13.7 eV), NI λ = 4109.98
Å (E 1 = 13.7 eV) and argon atoms ArI λ = 4158.96 Å (E 2 = 14.55 eV)
are quite close, the nitrogen atoms concentrations can be measured in
principle using equation (3.6). All assumptions made in deriving equation
(3.6) were confirmed for the experimental conditions used in Ref.20.
The experimental results of the measurement of concentration in rela-
tion to the conditions in the vacuum arc discharge plasma were explained
in Ref.20 on the basis of a kinetic model which takes into account the
formation of nitrogen atoms in dissociation of N 2 by a direct electron
impact, dissociated recombination of N +2 with electrons in the recom-
bination of N on the walls of the discharge chamber.
Comparison of the experimental results obtained for the dependences
of the absolute concentration of nitrogen atoms and the growth rate of
the nitride–titanium condensate on the substrate of gas pressure made
it possible to propose an important channel of formation of titanium
nitride in the vacuum arc discharge through recombination of the ni-
trogen and titanium atoms on the surface.
The excitation potentials of the line of the oxygen atom OI λ = 7771.9
Å, of the Angström system of the CO molecule and the second posi-
tive system of N 2 are relatively similar, and to measure the absolute
concentrations of the oxygen atoms and the carbon oxide molecules in
the ground electronic state it is convenient to use the optical actinometry
method. The authors of Refs.22 and 23 carried out detailed measure-
ments of the concentration of these components in chemically active
plasma. Molecular nitrogen was used as the actinometer in all cases.
Naturally, the initial stage was the confirmation of the assumptions of
the method under the experimental conditions.22,23 The experimental object
in Ref.22 was a capillary glow discharge in the CO 2/N 2/He (1:1:8)
mixture at medium pressure (p = 20÷100 Torr) used for pumping
waveguide CO 2 lasers, and in Ref.23 it was the superhigh frequency
(microwave) medium pressure discharge (p = 70–100 Torr) with trans-
verse blowing of CO2 used to dissociate the carbon dioxide. In the latter
case, molecular nitrogen was added to the discharge in small amounts
for diagnostic purposes.
56
The results for the spatial distribution of the chemically active mol-
ecules of CO and O atoms, presented in Refs.22 and 23, provided im-
portant information of the special features of plasma chemical trans-
formations in the examined discharges so that it will be possible to
optimise these systems.
When adding the actinometric gas M to the examined gas X, the proc-
esses of direct electron excitation may be accompanied by reactions in
which heavy particles take part. For example, the reaction of quasi-
resonant energy transfer in collisions should take place:
k1
X + M * ® X * + M ± DE . (3.10)
The rate of the process (3.10) is often very high because ∆E is small.
As already mentioned, a low value of ∆E is one of the main prereq-
uisites of the actinometry method.
Taking into account the fact that in addition to (3.10) the reactions
of excitation of the emitting states by the direct electron are also ef-
fective:
k2
M + e ® M * + e, (3.11)
k3
X + e ® X* + e, (3.12)
the equations of the balance of excitation and deactivation of the

actinometer and the examined gas assume the form
M*  ( A + ν ) =[M] nek2 , (3.13)

  M
 X*  ( A + ν ) = [X ] ne k3 +
[M ]ne k2 X k ,
  [ ]1
X
(A + ν)
M
(3.14)
where [ ] is the concentration of the corresponding particles; A i and v i

are the probability of spontaneous emission and the quenching rate of
the excited states M* and X* respectively, k 1, k 2, k 3 are the rate con-
stants of the reactions (3.10), (3.11) and (3.12); ne is the electron con-
centration.
Since the equations for the intensities of the lines (or bands) have
the form
57
Plasma Diagnostics
I M = M * AM ,
(3.15)
I X = X* AX .
equations (3.13)–(3.15) show that
−1
[X ] = I X (1 + ν / A )X  k 3 [M ] 
 + k1  .
[M ] I M (1 + ν / A )M  k 2 ( A + ν )M 
(3.16)
It can be seen that in justifying the actinometry method it is important

to take into account the processes of interaction of the heavy particles,
especially the processes of quenching the emitting states and excitation
transfer processes. This is especially important (as indicated by equation
(3.16)) when examining the plasma of discharges of higher pressures
and also the plasma with a higher concentration of the particles of the
actinometric gas.
58
Laser Methods of Plasma Diagnostics
Chapter 4
LASER METHODS OF PLASMA DIAGNOSTICS

4.1 Introduction
The invention of the laser has offered investigators a light source which
greatly expands the possibilities of the optical methods of plasma di-
agnostics. As a result of the unique properties of laser radiation –
extremely high brightness, high coherence and the possibility of pro-
ducing ultrashort laser pulses – lasers are not only used widely as light
sources in classic diagnostic methods such as shadow, interference and
schlieren methods, but have also been used to develop completely new
diagnostic methods such as non-linear, resonance and laser interferometry,
holographic diagnostics, scattering and fluorescence methods. In this
chapter, we examine briefly the plasma diagnostic methods in which the
use of lasers as light sources of principal importance.
More detailed information (of course, restricted by the publication
dates) on this problem can be found in a number of monographs 1-3 and
review articles. 5,6
4.2 Interference examination of plasma using lasers

The advantages of lasers as light sources in classic two-beam
interferometry with visualisation of field are obvious. It is the higher
time and spatial coherence of radiation which reduces the requirements
on straightening of the optical paths in two branches of the interfer-
ometer and on the accuracy of aligning the corresponding beams on the
interferogram. The high intensity and monochromatic nature of laser
radiation make it possible to avoid illuminating the interferogram with
natural radiation of plasma when using corresponding optical filters and
this greatly widens the range of plasma objects available for interfer-
ence investigations. The possibility of generating ultrashort laser pulses
greatly increases the time resolution.
Because of these obvious advantages, the lasers are used as light
sources in the majority of interference methods, especially in investi-
gations on large plasma systems and in diagnostics of dense pulsed
plasma.
59
Plasma Diagnostics
In this paragraph, we shall examine only the interference methods

that can be realised only when using laser radiation. These methods
include non-linear and resonance interferometry and also laser
interferometry with photoelectric recording.
Non-linear interferometry
The development of laser technology has been accompanied by rapid
advances in non-linear optics. In turn, the non-linear conversion of the
frequencies of laser radiation (by doubling or displacing frequencies)
has been used widely in interference plasma investigations. In particular,
there have been a large number of studies in which plasma probing is
carried out by simultaneous radiation of the basic frequency and the
second harmonics of a ruby or neodymium laser. At the output of the
interferometer these radiations are divided using light-dividing mirrors
and selective filters and two interferograms, corresponding to two
wavelengths, are recorded. As a result of combined treatment of these
interferograms, it is possible to separate the contributions to the refraction
of the plasma of electrons and atoms.
In all these studies the laser radiation frequency is converted to pas-
sage through the interferometer and the examined object. However, there
is a whole group of non-linear interferometers7 in which non-linear trans-
formation of the waves takes place after the passage through the ex-
amined object. This offers a number of new possibilities to interference
plasma diagnostics.
They include the possibility of probing plasma with the radiation of
a single frequency situated in the spectral range and suitable for re-
cording (for example, in the infrared range), followed by transforma-
tion of the radiation frequency to the visible range and recording the
interferogram using conventional photographic material. To realise this
possibility in plasma diagnostics, it is necessary to greatly increase the
sensitivity of measuring the electron concentration. In fact, it is well
known that the contributions of electrons and heavy particles (atoms
and ions away from their absorption lines) to plasma refraction are de-
scribed by the relationships
Ce N e
ne - 1 = - = -4.49 × 10-14 l2 N e , (4.1)
w2
na - 1 = Ca N a , (4.2)
2 pe 2
where Ce = ; e and m are the charge and electron, respectively;
m
60
ω and λ is the frequency and wavelength of probing radiation, respec-

1  B 
tively; C a =  A + 2  is the refraction of the atoms of ions per single
NL  λ 
3
particle, cm ; N e and N a is the concentration of electrons and atoms;
N L is the Loschmidt number. The coefficients A and B are presented in
tables in Ref.8.
In the visible and infrared ranges for the majority of atoms
B/λ2<<A and, consequently, C a is almost completely independent of ra-
diation frequency. In accordance with (4.1) and (4.2), the phase shift,
introduced into the wave by the electronic component
ω ( ne − 1)l Ce N e l
ϕe = =− (4.3)
c cω
(l is the thickness of the irradiated layer, c is the speed of light), in-

creases in transition to the IR range. At the same time, the phase shift
determined by the atoms
ω ( na − 1) l ωC a N a l
ϕa = = , (4.4)
c c
increases in the IR region. Thus, the transition from the visible to the
IR range is an efficient method of increasing the sensitivity of meas-
uring the electron concentration in the plasma.
If plasma is probed with IR radiation with frequency ω 1 which is
then mixed with the flat pumping wave with frequency ω2 (Fig.4.1), the
phase relief ϕ 3 of the wave of the total (or difference) frequency ω 3 =
ω 1 ± ω 2 in the first approximation is equal to the phase relief ϕ 1 of
the wave with frequency ω1. Thus, the sensitivity of measuring the elec-
tron concentration corresponds to frequency ω1 of IR radiation and the
F ig .4.1 Diagram of non-linear transformation of the waves. The wave U 1 (ω 1 ) passes

ig.4.1
through the object O and is mixed with the flat wave U 2 (ω 2 ) in a non-linear crystal
K; wave U 3 with a frequency ω 3 = ω 1 + ω 2 is separated by the filter F.
61
Plasma Diagnostics
interferogram can be recorded at frequency ω 3 in the spectrum range

suitable for recording.
Another new possibility of non-linear interferometry is the direct com-
parison of the shape of the fronts of waves of different frequency.
The difference in the shape of the wave fronts of waves of differ-
ent frequency, passing through a transparent object, is determined by
differences in the refractive index of the object for radiation with these
frequencies, i.e. by the dispersion of the object. The problem of direct
determination of the difference of the refractive indices, i.e. compari-
son of the form of the wave fronts of waves of different frequency
passing through the object, cannot be solved by classic interferometry
methods because waves of different frequency do not form a station-
ary interference pattern. A non-linear interferometry makes it possible
to solve this problem.
The principal diagram of a non-linear dispersion interferometer is
shown in Fig.4.2. The object O (plasma) is situated between two non-
linear elements K 1 and K 2. Two waves with frequency ω 1 and ω 2 are
directed to the input of the interferometer. These waves are partially
transformed to the wave with total frequency ω 3 in the first non-
linear element. The object is then irradiated with three waves U 1 (ω 1),
U 2 (ω 2) and U 3 (ω 3), and the waves U'1 (ω 1), U'2 (ω2) and U'3 (ω 3) leave
the object. The phases of these waves are respectively
l
ϕ1′ = ω1  n ( ω1 ) − n0 (ω1 ) ; (4.5)
c
l
ϕ′2 = ω2  n (ω2 ) − n0 (ω 2 ) ; (4.6)
c
l
ϕ′3 = ω3  n ( ω3 ) − n0 (ω 2 ) . (4.7)
c
where n and n 0 are the refractive indices of the object and the envi-
ronment.
In the second non-linear element, the frequencies of the waves U' 1
(ω 1), U'2 (ω 2) are mixed and this leads to the formation of the wave
U''3 (ω 3) whose phase is ϕ''3 = ϕ'1 + ϕ'2. Interference of two waves U'3
F ig .4.2 Diagram of a non-linear dispersion interferometer.

ig.4.2
62
and U''3, separated by the filter F, takes place at the outlet of the inter-
ferometer. The frequency of these waves is ω 3 and their distinguishing
feature is that one of these waves was obtained as a result of mixing
of the waves with the frequencies ω 1 and ω 2 prior to their passage
through the object, and the second – after passage through the object.
The phase difference of these waves
Dj D = j 3¢¢ - j ¢3 = j 1¢ + j ¢2 - j ¢3 , (4.8)
and the shape of their wave fronts differs only as a result of differ-
ences in the refractive indices of the object and the environment for ra-
diation with the frequencies ω 1, ω 2 and ω 3. If the dispersion of the en-
vironment can be ignored, the shifts of the fringes on the interferogram
will be determined only by the dispersion properties of the object. Sub-
stituting the values ϕ'1, ϕ'2 and ϕ'3 from (4.5), (4.6) and (4.7) into (4.8)
and setting n 0 (ω 1) = n 0 (ω 2) = n 0 (ω 3), we obtain
{ } l
∆ϕ D = ω1  n (ω1 ) − n (ω3 ) + ω2  n (ω2 ) − n ( ω3 ) .
c
(4.9)
A partial case of the dispersion interferometer is the non-linear inter-

ferometer based on frequency doubling proposed by the authors of this
book 9 and almost simultaneously in Ref.10. In this case, the frequen-
cies ω 1 and ω 2 are equal, ω 3 = 2ω 1 and
2ω1l
∆ϕ D =  n ( ω1 ) − n (ω3 ) , (4.10)
c 
and the shifts of the fringes on the interferogram are determined by the
differences of the refractory indices of the object for radiations of the
basic frequency and the second harmonics.
Representing plasma refraction as the sum of refractions of the elec-
trons and the atoms (see equations (4.1) and (4.2)) gives
Ce N e
n -1= - + Ca N a . (4.11)
w2
This shows that the dispersion of the refraction index of plasma is

determined mainly by the electron concentration. Substituting n from
(4.11) into (4.9) gives an equation for the shifts of interference fringes
on the dispersion interferogram
63
Plasma Diagnostics
∆ϕ D C N l 1 1 1 
kD = = e e  + − . (4.12)
2π 2 π c  ω1 ω 2 ω 3 
For the case of the interferometer with frequency doubling ω 1 =

ω 2 = ω 3/2 and
3Ce N e l
kD = . (4.13)
4 pcw1
Therefore, the dispersion interferometer is not sensitive to the atomic

component of plasma and makes it possible to take selective measure-
ments of the electron concentration on the same dispersion interferogram.
In Ref.11, the non-linear dispersion interferometer based on frequency
doubling was used to determine the electron concentration in the plasma
of a laser spark in air. One of the interferograms, obtained in Ref.11,
is shown in Fig.4.3a. For comparison, the photograph also shows an
interferogram in the light of the basic frequency ω 1 (Fig.4.3) of the ra-
diation of a ruby laser. The shifts of the fringes on the latter
interferogram are considerable and determined mainly by the redistri-
a
F ig .4.3 Interferograms of a laser spark. a) dispersion, b) in the light of the main

ig.4.3
frequency of radiation of a ruby laser.
64
bution of heavy particles (displacement from internal regions into the

shock wave). Small shifts, determined by the electronic component,
observed on the dispersion interferogram, are found only in the central
highly ionised region, and in the region in the immediate vicinity of the
front of the shock wave the fringes remain practically straight thus
indicating that the interferometer is not sensitive to heterogeneities with
no dispersion and associated with the distribution of heavy particles.
The accuracy of measuring Ne on the dispersion interferogram is con-
siderably higher than when determining N e using two interferograms ob-
tained in the light with ω 1 and ω 3 because in the latter case the small
shift, due to the electronic component ∆k 1,3, is determined as a differ-
ence of large values (∆k 1,3 = 2k 1 – k 3). Here k 1 and k 3 are the shifts of
the fringes on the interferograms corresponding to ω 1 and ω 3.
Of special interest is the case in which one of the mixed frequen-
cies, for example, ω1, is situated in the long-range or medium IR range,
and the second one, ω 2, in the visible range. Therefore, ω1<<ω 2, ω 3 and
Ce N e l
kD » .
2 pcw 1 The sensitivity of measuring the electronic component of
plasma depends on frequency ω 1 and is ω 3/ω 1 higher than when prob-
ing plasma with the radiation with frequency ω 3 whose light is used to
record the interferogram.
Resonance interferometry
The resonance interferometry methods are based on obtaining an
interferogram in the light of radiation close to the absorption lines of
one of the components of the examined plasma. Refraction (n–1) and
the absorption factor χ (Fig.4.4) in the vicinity of the line with a dis-
persion contour are described by the relationships
F ig .4.4 Dependence of refraction and the coefficient of absorption in the vicinity

ig.4.4
of the spectral line on the wavelength.
65
Plasma Diagnostics
n − 1 = C λ 30 N a f
(λ − λ 0 )
, (4.14)
(λ − λ 0 )2 + (∆λ / 2 )2
∆λ
κ = 2πC λ 02 N a f 2
.
 ∆λ 
(λ − λ 0 )
2 (4.15)
+ 
 2 
Here λ0 is the length of the wave corresponding to the maximum of the

absorption line, ∆λ is the width of the line measured at half height, λ
is the wavelength of radiation, f is the force of the line oscillator, N a
e2
is the concentration of the atoms on the absorbing level, C = .
4pmc 2
On approach to the absorption line, the refraction of the corresponding
atoms rapidly increase and can exceed by several orders of magnitude
the refraction of the same atom away from the absorption line. Thus,
using the radiation with the wavelength close to the absorption line for
obtaining interferograms, it is possible to increase greatly the sensitivity
of measuring the concentration of appropriate atoms.
A proposal to produce the interference pattern in the light of radiation
close to the absorption line of one of the components of the examined
plasma was made prior to inventing lasers. 12 However, the invention of
lasers facilitated the introduction of this method to the practice of plasma
investigations. A radiation source with the lines close to the absorp-
tion lines of the atoms to be determined was specially selected in the
first studies of resonance interferometry. For example, the authors of
Ref.13 presented a table of laser radiation lines situated close to the
absorption lines of different atoms.
Of special interest is the possibility of using tuneable dye lasers in
resonance interferometry. This not only greatly widens the range of ob-
jects that can be studied by this method but also makes it possible to
vary over a wide range the sensitivity of interference measurements dis-
placing the position of the generation line in relation to the wavelength
of the absorption line.
The sensitivity and ranges of application of resonance interferometry
were calculated in Ref.14, 15. In Ref.14, calculations were carried out
assuming monochromatic probing radiation for the case of the absorption
line whose contour is determined by the combined effect of dispersion
and Doppler broadening. In Ref.15, calculations were carried out for
the absorption line with a dispersion contour, taking into account the
finite width of the probing radiation line.
66
The relationship between the shift of the interference fringes on the

resonance interferogram and the concentration of the absorbing atoms
can be found using equation (4.14) for plasma refraction
( n − 1)l ≈ C λ 02 N a fl (λ − λ 0 )
k= . (4.16)
λ (λ − λ 0 ) + ( ∆λ / 2 )2
2
Therefore, setting k min = 0.1, it is easy to obtain the concentration de-

termined by this method as a function of the distance between the wave-
length of the line of probing radiation λ and the absorption line λ0. Equa-
tion (4.16) shows that the minimum detectable concentration is obtained
at a distance from the centre of the absorption line equal to its half width,
Dl
i.e. at l = l 0 ± :
2
k min Dl
N a,min » . (4.17)
Cl20 fl
Thus, the minimum detectable concentration of the atoms is directly

proportional to the width of the absorption line.
The upper concentration limit available for investigations by laser
interferometry is determined by the absorption of probing radiation by
plasma and a corresponding reduction of the contrast of the interfer-
ence fringes. Figure 4.5 shows the boundaries of applicability of the
methods calculated in Ref.15 using these assumptions. The curves, shown
in this graph, correspond to the probing radiation line whose width is
considerably smaller than the width of the absorption line. For the case
in which the width of the probing radiation line is comparable with that
of the absorption line, the shifts of the fringes, corresponding to dif-
ferent parts of the contour of the probing line, differ and this reduces
the contrast of the interference pattern. As shown by the calculations
in Ref.15, when resonance interferograms are produced using a radiation
with the line width exceeding the width of the absorption line, the sen-
sitivity of the method greatly decreases and the range of the measured
concentrations is smaller.
Laser interferometry with a dye laser as a light source was used to
examine the plasma of the laser jet on a target made of an alloy of
lithium with lead. The equipment circuit (Fig. 4.6) consisted of two lasers
whose Q-factor was modulated by the same rotating prism P. Radia-
tion of neodymium laser A 1 was focused onto the surface of the solid
67
Plasma Diagnostics
F ig .4.5 Boundaries of applicability of the method of resonance interferometry at

ig.4.5
different widths of the absorption line and the infinitely narrow line of the probing
radiation.
L2 F
L3
FP
L1
MZ
F ig .4.6 Diagram of equipment for interference examination of the laser jet using a
ig.4.6
dye laser.
target inside a vacuum chamber K placed in one of the arms of the

Mach-Zender interferometer (MZ). The dye was pumped into the cu-
vette by radiation of the second harmonics of the ruby laser L 2. The
cavity of the dye laser L 3 consisted of a semitransparent mirror and a
diffraction grating which when rotated resulted in smooth readjusting
of the length of the generation line in the vicinity of the resonance doublet
of lithium (λ0 = 6707.8 Å). A Fabry–Perot (FP) interferometer with the
distance between the mirrors of 0.01 cm was placed inside the cavity
of the dye laser to reduce the width of the generation line. The gen-
eration spectrum of the laser had the form of a single line 0.5 Å wide.
68
a b c d e
F ig .4.7 Interferograms of the laser jet on a target made of a lithium alloy with lead
ig.4.7
for different moments of time. t = 0.3 (a), 0.8 (b), 1.3 (c), 1.8 (d), 2.3 5s (e); λ-
λ 0 = 30 (a,b), 8 (c,d), 4 Å (e).
The radiation of the ruby dye laser was directed onto the interfer-
ometer using glass prisms and sheets. Two interferograms were pho-
tographed at the outlet of the interferometer in the light of radiation
of the ruby laser λ 1 and the dye laser λ 2.
Figure 4.7 shows interferograms obtained in the light of a dye la-
ser and corresponding to different stages of jet acceleration (t = 0.3-
2.3 µs). Optimum sensitivity was selected for each specific moment by
varying the value λ–λ 0. For times of 0.3 and 0.8 µs when determin-
ing the concentration of lithium atoms, it was necessary to process
interferograms obtained simultaneously in the light of radiation of the
dye laser and the ruby laser. Starting from 1.3 µs, there are no shifts
of the fringes of the interferograms obtained in the light of the ruby
laser. Therefore, the shifts of the fringes on interferograms, corresponding
to radiation of the dye laser, can be fully attributed to the refraction
of lithium atoms.
The spatial distributions of the concentration of lithium atoms, shown
in Fig.4.8, were obtained as a result of processing the interferograms
for different stages of jet acceleration. As a result of modifying the wave-
length of the radiation of the ruby laser, it was possible to take reli-
able measurements of the concentration of lithium atoms with the con-
centration varying in the range of approximately four orders of mag-
nitude.
In Ref.17, resonance interferometry was used to determine the con-
centration of excited hydrogen atoms in the plasma of a laser spark.
The laser spark was produced in a chamber filled with hydrogen with
a pressure 2 to 10 atm. Interferograms were obtained using the radiation
of a dye laser with a wavelength close to the line H α. The measured
69
Plasma Diagnostics
F ig .4.8 Distribution of lithium atoms along the laser beam axis for the moments
ig.4.8
of acceleration of the laser jet of 0.3 (1), 0.8 (2), 1.3 (3), 1.8 (4), 2.3 µs (5).
parameter in this case was the concentration of the atoms on the sec-
ond energy level.
Laser interferometry
The term ‘laser interferometry’ usually refers to different interference
methods in which the laser plays simultaneously the role of a light source
and an interferometer, and the phase shifts are recorded by photoelectric
methods.
If a diaphragm separating part of the interference fringe is placed
at the output of an interferometer, and a photoelectric receiver is placed
behind this diaphragm, then the variation of the photoflux can be used
to examine the variation of the optical difference of the path with time.
However, in this type of recording, it is not possible to obtain infor-
mation on the spatial distribution of the refractive index that can be
obtained in systems with visualisation of the field. The spatial resolution
is determined by the diameter of the probing beam.
An advantage of the photoelectric recording method is the possibility
of increasing the sensitivity of determining the electron concentration
by using long wave radiation because a relatively large range of pho-
toelectric receivers is available for operation in the IR range. In ad-
dition, photoelectric recording opens wide possibilities for using dif-
ferent radioelectronic systems employed in interference diagnostics of
plasma in the microwave range (see Ref.18). Consequently, the accu-
70
racy of measuring the phase shift can be greatly increased.

The diagram of the simplest laser interferometer 19 is shown in
Fig.4.9a. A flat mirror M 3 is added to the cavity of the helium–neon
laser consisting of spherical M 1 and flat M 2 mirrors. The mirrors M 2
and M3 form an additional Fabry–Perot cavity linked with the main laser
cavity. The intensity of laser radiation depends strongly on the phase
of the signal reflected from mirror M3. The variation of the optical length
of the M 2–M 3 cavity as a result of changes of the parameters of the
plasma placed inside it caused periodic changes of the phase of the
reflected signal and the modulation of laser radiation leaving through
the mirror M 1. The signal modulated with respect to amplitude is re-
corded by the photoreceiver and oscillographically processed. Evidently,
the variation of the optical length of the cavity corresponding to a single
modulation period is λ/2.
It is well known that the helium–neon laser can emit simultaneously
a number of lines in the visible and infrared spectrum ranges, includ-
ing lines λ = 0.63; 3.39 µm. Both these lines have a common upper
level and, therefore, the modulation of the intensity of laser radiation
at a wavelength λ = 3.39 µm is also accompanied by the modulation
of visible radiation λ = 0.63 µm. Placing a silicon filter between mirror
M 2 and plasma it is possible to probe the plasma by IR radiation char-
acterised by higher sensitivity to determination of the electronic den-
sity, and the modulation of the intensity of laser radiation in the vis-
ible range can be examined using a conventional photomultiplier.
Laser interferometry methods are used mainly for examining pulsed
M1 a M2 M3
PL
L
P
M3
M1 M2 PL
b
L
P
M3
M2 PL M2
c
L
P
F ig .4.9 Diagrams of three-mirror laser interferometers. a) internal resonator of the

ig.4.9
Fabry–Perot type; b) resonator with a spherical mirror; c) plasma is situated inside
the laser resonator; L - helium-neon laser; PL - plasma; P - photoreceiver.
71
Plasma Diagnostics
plasma. If the number of maxima on the oscillogram from the moment

of formation of plasma to the given moment of time t is k (t) and it
is known that refraction is caused mainly by the electronic component
of the plasma and changes monotonically during this period of time,
then taking into account the double passage of radiation through the
plasma, we have
1.12 ⋅1013
N e (t ) = k (t ). (4.18)
λl
To determine the parameters of stationary plasma, it is necessary to

disconnect rapidly the energy source supplying power to the plasma and
count the modulation maxima from the moment of disconnection to com-
plete collapse of plasma.
When the concentration changes non-monotonically, the form of the
oscillogram does not make it always possible to determine whether the
shift per fringe corresponds to the increase or decrease of Ne. Additional
phase modulation is often used to remove ambiguity in interpreting the
fringes. In the absence of plasma, laser radiation is modulated at a
constant frequency. The introduction of plasma into the cavity disrupts
the periodicity and the direction of variation of the refractive index of
plasma can be determined by examining whether the distance between
maxima becomes smaller or greater.
The accuracy and sensitivity of measuring N el depend on the wave-
length of probing radiation and the accuracy of measuring the shift of
the fringes ∆k:
1.12 ⋅ 1013
∆ ( N el ) = ∆k . (4.19)
λ
As indicated by (4.19), when using IR radiation of a helium–neon

laser with λ = 3.39 µm a shift per fringe corresponds to ∆(Nel) = 3·1016
cm 2. In direct count of the maxima on the oscillogram, the accuracy
of measurement is, according to the data of different authors, 1 to 0.25
cycles. The method of phase modulation using an acousto-optical cell
introduced into the M 2–M 3 resonator, or vibrations of mirror M 3, make
it possible to measure the phase difference with an accuracy of approxi-
mately 10 –3 fringe. 20,21 Substituting λ = 3.39 µm and ∆k = 10 –3 into
(4.19), the resultant measurement error is ∆(N el) = 3·10 13 cm 2.
Sensitivity can be increased using an interferometer in which the flat
mirror in the outer resonator is replaced by a spherical one22 (Fig.4.9b).
72
Not only longitudinal but also transverse modes are excited in this reso-
nator so that when measuring the length of the resonator, the intensity
of laser radiation is modulated with a frequency S times higher than
the laser interferometer with a flat mirror M 3. As shown in Ref.23, the
quantity S has a simple geometrical meaning – it is the number of double
passages of the light beam between the mirrors of the cavity after which
the beam returns to the initial (entry) point. This geometrical interpre-
tation is especially convincing when the laser beam is not directed along
the axis of a reference resonator (Fig.4.9b). The increase of sensitiv-
ity in this case can be regarded as a result of multiple (S times) pas-
sage of radiation through plasma. In successive passages, the beam is
displaced thus impairing the spatial resolution. This can be avoided by
placing the plasma inside a laser resonator 24,25 (Fig.4.9c).
As already mentioned, the sensitivity of determining the electron con-
centration can be increased by increasing the wavelength of laser ra-
diation. For example, when using the radiation of a CO2 laser (λ = 10.6
µm) λ(N el) ≈ 10 16 cm –2 corresponds to the shift per fringe, and when
using a HCN laser (λ = 337 µm) it is ∆(N el) ≈ 3·10 14 cm –2 .
When measuring the atom concentration, the sensitivity can be in-
creased selectively using the radiation close to the absorption lines of
these atoms (see equation (4.14)). For example, in Ref.26 the concen-
tration of metastable helium atoms on the level 23S1 was determined using
the radiation of a helium–neon laser (λ = 1.0798 µm) whose distance
from the absorption line λ = 1.0830 µm was 32 Å. A similar method
was used in Ref.27 to determine the concentration of metastable neon
atoms on the level 1s 5. Plasma probing was carried out by the radia-
tion of a helium–neon laser (λ = 0.6401 µm) 1 Å from the line formed
in transition to the level 1s 5 (λ 0 = 0.6402 µm). In this case, the con-
centration of the metastable neon atoms corresponding to the shift per
fringe was ∆N a ≈ 1.1·10 11 cm –3 (column length l = 42.5 cm).
Laser heterodyne systems are characterised by a very high sensitivity
to the changes of plasma refraction. The addition of plasma inside a
laser resonator shifts the laser radiation frequency
Dw lDn Dw lDn
= , = , (4.20)
w dr w dr
where l is the thickness of the plasma layer, d r is the laser resonator

length. Frequency shift can be measured by optical heterodyning. Ac-
cording to the estimates in Ref.28, the minimum frequency shift measured
by this method is around 10 Hz which corresponds to the ∆n =
2·10–13 and an electron concentration of 3·108 cm–3 (in calculation it was
73
Plasma Diagnostics
assumed that l = 20 cm, d r = 100 cm and λ = 1.15 µm).

A laser heterodyne system was used for plasma diagnostics in Ref.29.
The radiation of two helium–neon lasers with plasma added to the reso-
nator of one of these lasers and the second laser was a reference one,
was directed onto a photoreceiver. The recorded signal showed pulsa-
tions at frequency Ω equal to the difference of the generation frequency
of two lasers (Ω = ω 1– ω2). The introduction of pulsed plasma into the
laser resonator caused changes in the pulsation frequency. The mini-
mum measured value of the electron concentration, obtained by the
authors of Ref.29, was 10 11 cm –3 (at a layer thickness of 24 cm).
A superheterodyne laser interferometer was described in Ref.30, 31.
In this case, the radiation of a laser was divided into two channels in
a Mach–Zender interferometer. An ultrasound cell was placed in one
of the arms of the interferometer. The light signal at the output was
recorded using a photomultiplier. Ultrasound frequency Ω was present
in the frequency spectrum of the photoflux. The addition of plasma to
the second arm of the interferometer changed the phase of the oscil-
lations and this frequency situated in the radio range. The phase variation
was recorded using an electronic phase meter. The sensitivity of deter-
mining the phase shift was 1°, 31 which corresponded to (N el) min ≈ 10 15
cm –2 at λ = 0.63 µm.
Different circuits of laser interferometers with heterodyne and
homodyne frequency conversion were described in Ref.32.
4.3 Holographic plasma diagnostic methods

Possibilities and special features of holographic plasma diagnostic
methods
Holography is a method of recording and subsequent restoration of waves
based on recording the interference pattern formed as a result of the
interference of the subject wave and the reference wave coherent with
it. The interference structure recorded on the light-sensitive material is
referred to as the hologram. The diffraction of the reference beam on
the hologram leads to restoration of the subject wave.
In examining the plasma by classic shadow, schlieren and interfer-
ence methods, the phase distortions of the light wave which has passed
through the plasma are visualised in the form of a shadow pattern or
interferogram directly at the moment of passage of radiation through
plasma. Holography makes it possible to record the wave passed through
the plasma, restore it and examine it by various methods. As in the direct
application of these methods, the measured quantities are the refrac-
tive index or its derivatives, and in measuring the amplitude distortions
– the absorption factor.
Data on the plasma parameters are obtained through their link with
74
the refractive index and the absorption factor, i.e. the holographic method
offers nothing new in comparison with the conventional optical method.
Nevertheless, in many cases the holographic methods have considerable
advantages in comparison with the conventional optical methods and
sometimes make it possible to solve problems which could not be solved
by these conventional methods. The new possibilities of the holographic
method include:
– recording and subsequent restoration of the wave passed through
the pulsed plasma, diagnostics of the latter under the standard condi-
tions by different optical methods;
– restoration by means of a single hologram of light waves passed
through the plasma in different directions within the limits of the solid
angle defined by the hologram makes it possible to obtain the spatial
distribution of the parameters of the three-directional object in the ab-
sence of axial symmetry;
– interference examination of the plasma enclosed in a vessel with
optically non-uniform windows;
– low requirements on the quality of the optical components of the
interferometer and, consequently, the possibility of interferometric ex-
amination of plasma of almost unlimited dimensions;
– simultaneous or consecutive recording of holograms in the light
of radiation passed through the plasma with several wavelengths and
examination of the interference pattern formed by the waves restored
by such a hologram;
– increase of the sensitivity of interference measurements;
– recording of the amplitude distortions of the transmitted wave and
measurement of the absorption factor of the plasma without interfer-
ence from the side of its natural radiation;
– the possibility of using various movie holographic systems for ex-
amining plasma dynamics.
The holographic methods were used for the first time for examin-
ing plasma in 1966.33 At present, they are widely in plasma experiments.
We shall now present the results of a number of investigations in which
these special features were most evident together with the possibilities
of holographic methods of plasma diagnostics. More detailed information
can be found in a number of books and review articles. 2-5
Examination of pulsed plasma under the stationary conditions

The light wave passed through the pulsed plasma is recorded on the holo-
gram and then restored and examined by different methods which sup-
plement each other.
The schlieren circuit was used in the restoration stage already in the
initial investigations of holographic plasma diagnostics.33,34 It was found
75
Plasma Diagnostics
F ig .4.10 Schlieren-system for examining wave fronts restored by a hologram. 1)

ig.4.10
laser, 2) spot diaphragm, 3) bent mirror, 4) hologram, 5) blade, 6) scanning
photoreceiver.
that the plasma of the laser spark is equivalent to the negative astig-
matic lens, and the focusing distance of this lens was used to deter-
mine the refractive index of the plasma and estimate the electron con-
centration in it.
A schlieren circuit, shown schematically in Fig.4.10, was used to
record a wave restored by the hologram in Ref.35 where a laser spark
in argon induced by the radiation of a CO 2 laser was examined. Infor-
mation on the spatial distribution of the gradient of the refractive in-
dex of the plasma was obtained using a scanning photodetector.
The authors of Ref.36 describes a holographic interferometer which
enables the frequency and orientation of the fringes on the interferograms
to be changed in the restoration stage. The circuit of the interferom-
eter is shown in Fig.4.11. The hologram 1 was produced by the dou-
ble exposure method, and the angle of incidence of the subject beam
on the hologram was varied between exposures. The waves restored by
this hologram and corresponding to each exposure propagate under an
angle to each other and can be separated in the focus of the objective
2. Reflected from different faces of the prism 3, the waves propagate
in two different arms of the interferometer and form an interferogram
in plane 4 where the frequency and orientation of the fringes are meas-
F ig .4.11 Diagram of a holographic interferometer.

ig.4.11
76
ured by tilting one of the mirrors 5 of the interferometer. Overlapping

one of the interference beams, the shadow images of the examined object
are obtained. This interferometer was used by the authors of Ref.36 to
examine Z- and θ-pinch plasma.
Examination of the spatial distribution of the plasma parameters using

holographic interferometry data
As in the majority of optical methods (except the methods of laser scat-
tering and resonance fluorescence, Fig.3), the holographic method makes
it possible to determine only the plasma parameters integral with re-
spect to the direction of the beam. In the case of objects with axial
symmetry, the transition from the measured quantities to the radial dis-
tribution is usually carried out by solving the integral Abel equation.
In the absence of axial symmetry of plasma the information carried
by the light wave, passed through the plasma in one direction, is in-
sufficient to determine the local plasma parameters. If in producing a
hologram the plasma was illuminated over a wide solid angle range, for
example, using a diffusion scatterer, then using a single hologram it is
possible to restore the light waves passed through the plasma under
different angles so that the spatial distribution of the refractive index
in the plasma can be determined.
In principle, some information can be obtained by classic
interferometry. In this case, it is necessary to examine the plasma si-
multaneously using several interferometers or carry out consecutive
examination of the interferograms of plasma with the variation of the
mutual position of plasma equipment and the interferometer. From the
technical viewpoint, both methods are highly cumbersome and difficult
to apply and, in addition, when using the second method it is neces-
sary to ensure exceptionally high stability or reproducibility of the
examined object because random errors, associated with the non-repro-
ducibility of the plasma, can lead to an unacceptable reduction of the
accuracy of calculations of the local values of its parameters. There-
fore, it is not thought that classic interferometry can determine t the
local parameters of non-axis symmetric plasma with sufficient efficiency.
At the same time, the holographic methods can be used to solve these
problems.
The problems of examining the spatial distributions of the refrac-
tive index of three-dimensional objects were examined in Ref.37-40. The
most detailed and consecutive examination of these methods was car-
ried out in Ref.41-43. Without discussing the mathematical aspect of
the problem, we shall examine only the main procedures of obtaining
holographic interferograms corresponding to irradiation of the phase
object under different angles.
77
Plasma Diagnostics
O
D
a
H
b P
L S
H
F ig .4.12 Diagram of recording the hologram of a transparent object with a scatterer

ig.4.12
(a) and restored waves corresponding to the specific direction of illumination (b).
The simplest method is to irradiate the object O using the scatterer

D, for example, matt glass (Fig.4.12a). In the restoration stage, the spot
diaphragm P is placed in the focal plane of the lens L (b) which projects
the apparent image of the object on to the screen S; the diaphragm sepa-
rates beams restored by the hologram H and the parallel beam corre-
sponding to the specific irradiation direction. Changing the position of
the diaphragm, it is possible to obtain a series of interferograms cor-
responding to different examination directions using a single double-
exposure hologram.
It should be noted that when producing holograms by equipment with
a scatterer, various parts of the initial subject would interfere with the
reference beam. This imposes stringent requirements on the spatial co-
herence of the light used to produce the hologram. Therefore, the systems
with the scatterer usually include single-mode lasers. An exception is
a system proposed in Ref.44 in which the scatterer is projected on to
a hologram so that it is possible to combine the mode structure of subject
and reference beams and use a multimode laser as a radiation source.
The second disadvantage of the system with the scatterer is the spotty
structure typical of images of diffusion objects illuminated with coherent
radiation.
In certain cases, the matt scatterer can be efficiently replaced by a
phase diffraction grating which enables the object to be irradiated under
several discrete angles.45,46 For example, in Ref.45 the holographic grating
with 50 lines/mm generated bright beams of three diffraction orders on
each side of the normal so that it was possible to obtain 7 interferograms
with a vision angle of 11° (in 1.8° steps) from a single hologram. The
restored interferograms do not have the spotted structure because in this
case there are no diffusion scattering elements. The equipment with the
78
diffraction grating was used in Ref.40 to obtain holographic

interferograms corresponding to different irradiation directions. These
interferograms were used to restore the field of gas temperatures above
the heated wire.
Insensitivity of holographic diagnostics to the quality of optical

elements
In a conventional two-beam interferometer, two light beams propagating
along its two arms interact. Defects of mirrors and other optical ele-
ments lead to different distortions of the beams and this has a direct
effect on the form of the interference pattern. However, in holographic
interferometry (see, for example, Ref.3, 43) the waves passing along
the same path interact but at different moments of time. This is one of
the most important special features of the holographic method – its
differentiality.
Recording on a hologram two waves corresponding to two states of
the examined object and restoring them at the same time, it is possi-
ble to examine an interference pattern whose form is determined only
by changes which took place in the object during the period between
the two exposures. Thus, the comparison wave forms automatically and
repeats in every detail the waves scattered by the object in the initial
state. Consequently, the requirements on the quality of optical components
of equipment becomes less stringent. Distortions of the wave fronts, as-
sociated with the imperfection of the optical elements of the holographic
system are the same for two interference waves and have almost no effect
on the form of the interference pattern.
Because of the insensitivity of holographic interferometry to imper-
fections of the optical elements it is possible to examine plasma in large
plasma systems, for example, Z- and θ-pinch. 47,49 It is difficult to carry
out these investigations by classic interferometry because devices with
large mirrors are extremely difficult to produce and are very expensive.
Because of this advantage of holographic interferometry it is also possible
to diagnose the plasma enclosed in a vessel with non-uniform walls,
especially plasma of pulsed lamps. 50-52
If distortions in interference comparison of two waves distorted to
the same extent by optical elements are excluded, then in examining the
waved restored by the hologram using shadow or schlieren methods the
distortions of the wave front are recorded to the same degree both as
a result of the passage through the examined phase object and as a result
of the imperfection of the optical systems, especially due to aberration
of the lenses and spherical mirrors. However, holography makes it
possible to exclude distortions of the wave front caused by the imper-
fection of the components of the optical system, 53,54 by recording two
79
Plasma Diagnostics
F ig .4.13 Repair of aberration of wave fronts. a) recording of the holograms H 1 and

ig.4.13
H 2 ; b) restoration of non-aberration wave front V'2 .
holograms on the same holographic equipment: H1 in the absence of the

object and H2 in its presence. In the restoration stage the conjugate wave
front 1, restored by the first hologram, with the phase relief inverse in
relation to the wave V1, is used to illuminate the second hologram. This
results in the restoration of the aberration-free wave front V'' 2 whose
phase relief is determined only by the passage of the wave through the
examined object (Fig.4.13).
Sensitivity of holographic plasma diagnostic methods and methods

of measuring it
The sensitivity of double-exposure holographic interferometry is almost
identical with that of the conventional method. The minimum observed
variation of the optical thickness ∆(nl) min is associated with the mini-
mum measured shift of the fringes on the interferogram k min by the re-
lationship
∆ ( nl )min = k min λ, (4.21)
where λ is the wavelength, l is the thickness of the plasma layer.

The error of measuring the shifts on the fringes on the holographic
interferogram of medium quality, as in visual processing of conventional
interferograms, is assumed to be usually equal to 0.1 fringe.
Minimum value of the electron and atom concentration can be de-
termined substituting into (4.21) the values of the refraction of elec-
trons (4.1) and atoms (4.2):
kmin
( Nel )min = 2.2 ⋅1013 , (4.22)
λ
80
kmin
( N al )min = . (4.23)
Ca λ
As in conventional interferometry, the transition to the IR range is

the selective method of increasing the sensitivity of measuring the electron
concentration in accordance with (4.22). At the same time, the selec-
tive increase of the sensitivity of measuring the concentration of a specific
sort of atoms or ions can be achieved by displacing the wavelength of
probing radiation to their absorption lines (see (4.14)), i.e. by resonance
holography. 55-58
In addition, there are a number of methods of increasing the sen-
sitivity of interferometry. Some of these methods are used in both con-
ventional and holographic interferometry, others are holographic only.
Sensitivity of both conventional and holographic interferometry can
be increased by systems with multiple passage over the subject beam
through plasma. The plasma is situated between two semitransparent
mirrors A and B (Fig.4.14a). To separate the light beam, correspond-
ing to the given number of passes, one of the mirrors is tilted in re-
lation to the other one by small angle . The consecutive beams are in-
clined in relation to each other by angle 2α and can easily be separated
using a spatial filter (a lens with a diaphragm positioned in its focal
plane). A similar variant of the multipass method was used to include
the sensitivity in examining shock waves in ballistic trajectories at re-
duced pressure. 59 Interferograms corresponding to 11 passages of ra-
diation through the object were obtained.
RB
Fig .4.14 Holographic diagram with multiple passage of the probing light beam through
ig.4.14
the object (a) and the diagram of increasing sensitivity as a result of multiple passage
of the light beam through the hologram in the restoration stage (b).
81
Plasma Diagnostics
A considerable disadvantage of this variant of the multipass method

is the displacement of the light beam in consecutive passages through
the object thus reducing the spatial resolution of the method. This short-
coming can be eliminated by another variant of the multipass holographic
method proposed in Refs.60, 61 in which the mirrors A and B between
which the object O is situated are positioned strictly parallel in rela-
tion to each other and normal to the probing beam (Fig.4.14a). If the
distance between the mirrors is greater than the coherence length of
radiation of the lasers, then from the set of the beams with different
numbers of passes through the plasma only the beam whose width differs
from that of the reference beam by less than the coherence length of
laser radiation will take place in the formation of the hologram. Moving
the mirror C, it is possible to change the length of the optical path trav-
elled by the reference beam RB and produce holograms corresponding
to the given number of radiation passes through the plasma.
The sensitivity of holographic interferometry can be increased not
only by the multiple passage of radiation through the object in the stage
of producing the hologram but also as a result of multiple passage of
radiation through the hologram in the restoration stage. 62,63 In this case
(Fig.4.14b), one of the semitransparent mirrors is perpendicular to the
beam restored by the hologram and directs the beam onto the hologram
as a restoring beam. Since the thickness of the hologram is consider-
ably smaller than the longitudinal dimensions of the examined plasma,
the transverse displacement of the light beam during its multiple pas-
sage to the hologram is considerably smaller than in passage through
the plasma. A disadvantage of this method is the faster (than in direct
passage through the object) reduction of the intensity of consecutive
beams determined not only by the transmission factor of the mirrors
but also by the diffraction efficiency of the hologram.
The second specific holographic method of increasing the sensitiv-
ity is the use of non-linear effects and the possibility of producing an
interference pattern in superimposition of conjugate waves (corresponding
to the apparent and real images) with inverse phase reliefs.
If the hologram is recorded on a non-linear photomaterial, in addi-
tion to the zeroth order it also restores ±1-st order of the wave of higher
orders (Fig.4.15) whose phase relief is k times (k is the number of the
order) higher than the waves of the first order. A similar method of in-
creasing the sensitivity of holographic interferometry was proposed in
Ref.54, 65.
Illuminating a hologram obtained by non-linear recording with two
light beams results in the propagation of two waves of different orders
in the given direction (Fig.4.16). In particular, if the restoring beams
are directed in such a manner that the waves of +k-th and –k-th order
82
a b
F ig .4.15 Diagrams of recording the hologram (a) and restoration of the wave fronts
ig.4.15
of a non-linearly recorded hologram (b).
F ig .4.16 Diagram explaining the increase of the sensitivity of holographic

ig.4.16
interferometry by a factor of 2 k in interference of the waves of the + k -th and k -th
orders.
are restored in the direction normal to the hologram, then the gain in
sensitivity as a result of their interference is 2k.
However, the use of non-linear effects for increasing the sensitiv-
ity intensifies not only the distortions of the wave front caused by the
examined object but also any distortion caused by the imperfection of
the optical elements of holographic equipment. Therefore, either holo-
graphic systems designed for increasing sensitivity are produced on the
basis of interferometers with high quality mirrors or it is necessary to
compensate the distortions introduced by the optics in the restoration
stage. The methods of this compensation of aberrations using waves of
high orders were examined in Ref.66, 67.
Infrared and holographic interferometry

As already mentioned, the transition to the IR range leads to a selec-
83
Plasma Diagnostics
tive increase of the sensitivity of determining the electron concentra-

tion. For example, in accordance with equation (4.22), the use of a CO2
laser (λ = 10.6 µm) instead of a ruby laser (λ = 0.69 µm) leads to a
15 fold increase of the sensitivity of determining N e, with other con-
ditions being equal.
One of the main difficulties of transition to the IR range is that no
high sensitivity and high-resolution media for recording IR holograms
are available at the moment.
The upper limit of the spectral sensitivity or silver-halide
photomaterials corresponds to a wavelength of the order of 1 µm. To
record holograms and interferograms in the range of longer waves, it
is necessary to use different materials (wax, gelatine, graphite,
thermochromic materials, polymer films, metal films deposited on a glass
substrate, etc.) whose optical properties undergo reversible or irreversible
changes as a result of heating with laser radiation.
The common property of the majority of these materials associated
with the thermal recording mechanism is the relatively low sensitivity
in the range from 0.1 to 1 J/cm2, whereas the sensitivity of silver halide
materials for recording holograms in the visible range is 10 –3 –10 –7
J/cm 2.
A second special feature of the materials for recording IR holograms
is the dependence of their resolution power on the exposure time de-
termined by ‘erosion’ of the temperature relief as a result of heat con-
ductivity. Therefore, continuous lasers can be used only for recording,
with these materials, interferograms with a frequency of several lines
per millimetre, whereas when using powerful pulse lasers with the same
materials it is possible to record holograms with a frequency to 100
lines/mm.
A new promising material for recording holograms is a phase inter-
ference transformation reversed light reflector (PITRLR) 68 based on the
semiconductor–metal phase transition in vanadium oxides. The sensi-
tivity of this material to the radiation of a ruby laser is 5·10 –4 J/cm 2
but at longer wavelengths the sensitivity decreases as a result of a de-
crease of the absorption factor of vanadium oxides in the IR range. In
recording all PITRLR holograms in the light of radiation of a CO 2 laser
(λ = 10.6 µm) the layer of vanadium oxides is thicker as a result of
thermal conduct with the substrate absorbing radiation with the absorbing
layer of silicon oxide deposited on the surface of PITRLR. According
to the data in Ref.69 the sensitivity in this case is 5·10 –3 J/cm 2.
High sensitivity is also typical of the method of recording IR holo-
grams proposed in Ref.70–72 which utilises the temperature depend-
ence of the sensitivity of conventional photomaterials. Infrared radia-
tion generates a temperature relief on the surface of photographic material
84
which is also illuminated with the actinic radiation of a pulsed lamp

synchronised with the radiation of the IR laser. The sensitivity of the
laser is 10 –2 J/cm 2, resolution power 50 lines/min.
A more detailed analysis of the properties of different media and the
methods of recording IR holograms can be found in Refs.73 and 74.
Infrared holographic interferometry was used to examine electric dis-
charges, 75,76 exploding wires, 75 and plasma generated in focusing laser
radiation in gases and on solid targets. 67,77,78
The diagram of holographic equipment for diagnostics of a spark
breakdown differs from conventional holographic systems in the vis-
ible range only by the presence of germanium optics (Fig.4.17). Holo-
grams were recorded using a pulsed CO2 laser on bismuth films deposited
on a glass subject.
Holography in the light of a pulsed CO 2 laser was used in Refs.77,
78 for examining late stages of development of the laser spark in air
and of the laser jet on a solid target. The plasma was induced by a
second powerful CO 2 laser synchronised with the first one. Holograms
were recorded on triacetate cellulose films.
The same material was used in holographic examination of an arc
discharge. 76 The holographic equipment (Fig.4.18) included the germa-
nium light-dividing wedge 2, and the electrodes of the arc 6, situated
in the vicinity of the wedge, were projected by means of the spherical
mirror 3 with ×2 reduction on the surface of triacetate cellulose film
5. The threshold properties of the material and the relief-phase nature
of recording resulted in highly non-linear recording and the appearance
of intensive waves of high orders in the restoration stage. Interferograms
with improved sensitivity (Fig.4.19) were obtained as a result of inter-
ference of the restored waves of high orders (up to ± 7 th order). The
sensitivity of measuring the electron concentration was (N e l) min =
CO 2 laser
F ig .4.17 Equipment for examining the spark breakdown using IR holography. W is

ig.4.17
the germanium wedge, L 1 and L 2 are germanium lenses, H is the hologram, O are
the electrodes of the spark.
85
Plasma Diagnostics
F ig .4.18 Diagram of equipment for producing IR holograms of arc discharge. 1)

ig.4.18
CO2 laser, 2) germanium wedge, 3,4) spherical mirrors, 5) hologram, 6) arc electrodes.
a b
c d
e f
g h
F ig .4.19 Holographic interferograms of the arc discharge plasma produced at

ig.4.19
interference of the waves of the orders of 0 and +1 (a), +1 (b), +2 (c), +3 (d), +4
(e), +5 (f), +6 (g) and +7 (h).
86
2·1014 cm–2. This high sensitivity makes the described method promising
for diagnostics of low-density high-temperature plasma in Tokamak-type
systems. The equipment designed for these investigations was described
in Ref.79; it was also used for analysis of late stages of acceleration
of the laser jet on a carbon target. 67
Two-wavelength, resonance and dispersion holography

To separate contributions to the refraction of electrons and atoms in
both conventional and holographic interferometry, it is necessary to record
at least two interferograms at two wavelengths. If the object and ref-
erence beams in the stage of producing the hologram consist of radiation
with two wavelengths λ 1 and λ 2, the resultant hologram represents a
superimposition of two interference (holographic) structures with dif-
ferent spatial frequencies ν 1 and ν 2 which depend on the wavelength
2 sin a / 2
( n1,2 = where α is the angle between the subject and refer-
l 1,2
ence beams).
In illuminating such a hologram with monochromatic radiation, the
restored waves corresponding to each hologram, propagate under dif-
b
F ig .4.20 Holographic interferograms of the laser spark in helium at a pressure of
ig.4.20
6 atm, corresponding to the radiations of the basic frequency (a) and second harmonics
(b) of a ruby laser.
87
Plasma Diagnostics
F ig .4.21 Variable distributions of the electronic concentration N e and gas density

ig.4.21
ρ, obtained at combined processing of the interferograms shown in Fig.4.20; r f is
the radius of the front of the shockwave, ρ 0 is the initial density of the gas.
ferent angles. If the hologram is produced by the double exposure

method, it can be used to restore two interferograms separated in space
and corresponding to each wavelength. Two such interferograms of a
laser spark in helium, corresponding to the radiation of the main fre-
quency in the second harmonics of a ruby laser, are shown in Fig.4.20.
Combined processing of these interferograms 80 yielded the spatial dis-
tribution of the concentration of electrons and gas density in the laser
spark (Fig.4.21).
If in recording the hologram one of the wavelengths present in the
radiation was close to the absorption line of the specific type of par-
ticles, the sensitivity of measuring their concentration increases in ac-
cordance with (4.14), i.e. we have a holographic variance of a reso-
nance interferometry in this case.
Resonance holographic interferometry was used for the first time in
Refs.55 and 56 for examining the laser jet on a potassium target. Plasma
probing was carried out by radiation with three wavelengths: a ruby
laser (λ 1 = 6943 Å), its second harmonics (λ 2 = 3472 Å) and forced
combination scattering of the radiation of a ruby laser in nitrobenzol
(λ3 = 7658 Å). Wavelength λ3 differs by only 7 Å from the wavelength
of one of the components of the resonance doublet of potassium. As
a result of processing three simultaneously produced interferograms
corresponding to the three wavelengths, it was possible to separate the
contributions to the refraction of electrons, air molecules and potassium
atoms described by the equations (4.1), (4.2) and (4.14), respectively.
The method of resonance holographic interferometry with a dye la-
ser used as a light source was used for examining the laser jet on a
88
barium target 57 and cesium plasma of a pulsed arc discharge. 58

In all the studies described previously interferograms corresponding
to different wavelengths and separated in space were produced in the
restoration state. However, holography makes it possible to examine
directly the pattern of interference of the restored waves which are
holographic copies of waves of different frequency.
As regards plasma diagnostics, of special interest is the method of
dispersion holographic interferometry 81 in which the hologram on the
phase object is recorded by means of the main frequency (λ 1) and the
second harmonics (λ 2 = λ 1 /2) of laser radiation on a non-linear
photomaterial. In the process of restoration of waves using such a holo-
gram H in relation to the waves of the ±1-st order waves of high or-
ders were restored (Fig.4.22). The wave of the second order, correspond-
ing to λ1, propagates in the same direction as the wave of the first order
corresponding to λ 2. Interference of these restored waves results in the
formation of an interferogram where the shifts of the fringes are pro-
portional to the dispersion of the object, as in the case as when using
a non-linear dispersion interferometer (see section 4.2). In turn, the dis-
persion of the object is determined by the electron concentration in the
plasma (see equations (4.10)–(4.11)).
F ig .4.22 Diagram of producing a dispersion holographic interferogram.

ig.4.22
Dispersion holographic interferometry was used in Ref.82, 83 to ex-

amine the laser spark in air. The resultant interferograms were used to
determine the electron concentration in the laser spark without making
any corrections for the redistribution of the concentration of air mol-
ecules associated with the formation of a shock wave.
Kinoholography
Holography offers considerable possibilities for constructing
kinoholographic systems, i.e. equipment producing a series of holograms
89
Plasma Diagnostics
F ig .4.23 Kinoholographic equipment for examining the laser spark. a) top view, b)
ig.4.23
lateral view, 1) pulsed ruby laser, 2,3) co-focal lenses, 4,5) mirrors of the optical
delay line, 6) wedges, 7) holograms, 8) light-dividing wedge, 9) wedge used to produce
finite-width fringes.
of a high-speed process corresponding to successive stages of its de-

velopment. These systems differ in the number of frames and time periods
between them, the method of formation of light pulses displaced with
time and the methods of holographic recording.
Figure 4.23 shows the circuit of one of the first kinoholographic sys-
tems used for examining the laser spark in air.84,85 The optical delay line
consists of opaque and semitransparent mirrors and is used to shape a
series of light pulses shifted in space and with time. The distance be-
tween the mirrors is 6 m which corresponds to the time period between
the frames of 40 ns. The light beams, passed through the semitransparent
mirror, were directed onto a laser spark using glass wedges. Each of
the beams was divided into subject and reference and then mode structures
were combined in the plane of the hologram. This gave five holographic
interferograms corresponding to five consecutive moments of develop-
ment of the laser spark (Fig.4.24). A similar five-frame system with the
period between the frames of 0.1 ns was described in Ref.86.
In the previously examined system, the object beams passed through
plasma under different angles. However, in certain cases, for example,
in examining Z- and θ-pinches and current layers, it is important to
ensure that all subject beams illuminate the plasma strictly in one di-
rection (along the discharge axis). The following holographic systems
can be used in this case (Fig.4.25). In examining holograms using the
system shown in Fig.4.25a, the subject beams, correspond to different
times are combined in space whereas the reference beams illuminate the
hologram under different angles. Thus, the same section of the photo-
sensitive layer records several holograms with different frequency and
orientation of the holographic structure. In the restoration stage, such
90
42 ns
84 ns
Time
126 ns
168 ns
210 ns
F ig .4.24 Five consecutive holographic interferograms of a laser spark.

ig.4.24
a hologram makes it possible to examine the images of the objects sepa-

rated in space and corresponding to different moments of time.
A shortcoming of this system is the low diffraction efficiency D 1
which is inversely proportional to the square of the number n of holo-
grams recorded in the same area, and for the case of amplitude holo-
grams is described by the equation
91
Plasma Diagnostics
F ig .4.25 Diagrams of producing kinoholograms in spatial combination of the beams

ig.4.25
( V 1 -V 3 ), corresponding to various moments of time. a) three holograms are recorded
on the same section of the photolayer, b) holograms corresponding to different
moments of time are recorded in different sections of the photolayer, U 1 , U 2 , U 3
are reference beams.
F ig .4.26 Diagram of equipment for kinoholographical examination of the θ-pinch.

ig.4.26
G is the diffraction grating, H 1 – H 3 are holograms, L 1 , L 2 , L 3 are lasers, P is plasma.
1
D1 £ . (4.24)
16n 2
In the system shown in Fig.4.25b, the subject beams combined in

space are divided into several parts using diffraction grating G and are
directed onto different areas of the photosensitive material. Each of these
sections is illuminated with one reference beam. This system was used
in Ref.48 to examine the θ-pinch. Three pulsed ruby lasers were used
to produce light passes separated in time (Fig.4.26).
As shown in Ref.87, the diffraction efficiency of the hologram is de-
scribed by the relationship
92
1
D2 ≤
( )
2
4 n +1 (4.25)
and decreases at a considerably lower rate when the number of holo-

grams increases. For example, at n = 5 D 2 is almost an order of mag-
nitude higher than D 1.
The system with the diffraction grating was also used to examine
the plasma of a current layer. 87–89 In this case, beams delayed in time
were produced using an optical delay line (Fig.4.27). As an example,
Fig.4.28 shows five holographic interferograms produced during a single
pulse of plasma equipment and corresponding to the processes of failure
of the current layer. As indicated by the photographs, in the initial
uniform distribution of the electronic concentration the layer (frame 1)
contains a region with reduced concentration in the centre of the layer
(frames 2, 3) and this is followed by a decrease of the concentration
in the entire layer (frames 4, 5). Figure 4.28b shows curves of equal
electron concentration calculated as a result of processing a series of
interferograms (see Fig.4.28a) and illustrating the process of failure of
the layer.
The real time method can be used for kinoholographic plasma
diagnostics. The circuit of one of these systems, designed for examining
Laser
M2
M1
F ig .4.27 Diagram of kinoholographic equipment for examining current layers. G

ig.4.27
is the diffraction grating, M 1 , M 2 are the mirrors of the optical delay line, P is the
plasma chamber, H is the hologram.
93
Plasma Diagnostics
1.40 µs
1.46 µs 1.52 µs
1.58 µs 1.64 µs
Fig4.28 (continued on page 95).
94
b
1.40 µs
1.46 µs
1.52 µs
1.58 µs
1.64 µs
F ig .4.28 Five consecutive holographic interferograms of the current layer (a) and
ig.4.28
the lines of equal electronic density constructed from them (b).
the plasma jet, 90,91 is shown in Fig.4.29. Radiation was generated by

ruby laser 1 operating in the free generation regime. Initially, the holo-
gram 2 was exposed in the absence of plasma. The hologram was de-
veloped and processed on site using a special cuvette. The wave, re-
stored by this hologram, then interfered with the wave passing directly
through plasma 3. The interferograms were photographed using the
superhigh speed camera 4. The exposure time of the individual frames
was determined by the duration of the laser radiation beams and did
not exceed 1 µs.
Identical equipment was used in Ref.92 to examine flames. Radia-
tion was generated by a 50 mW continuous He–Ne laser. This restricted
the time resolution of the method to 10 –3 s.
4.4 Plasma diagnostics on the basis of scattering of laser

radiation
In all previously described optical methods of plasma diagnostics, the
95
Plasma Diagnostics
F ig .4.29 Equipment for kinoholographic examination of the plasma jet by the real
ig.4.29
time method.
information obtained directly from the experiment is integral with re-

spect to the examination direction. The transition from integral to lo-
cal values takes place by means of relatively complicated calculations
and results in a large reduction of accuracy. Therefore, it is of special
interest to examine the diagnostics method enabling direct measurements
of the local plasma parameters. The method of scattering of laser ra-
diation is one of these methods.
Scattered radiation is observed from a small volume formed in in-
tersecting the light beam, probing the plasma, with a cone within which
the radiation scattered by the plasma is recorded.
If the wavelength of the probing radiation is situated at a relatively
large distance from the absorption line of the atoms and ions, the main
contribution to light scattering by the plasma is provided by the elec-
trons. The total, the so-called Thomson scattering cross section of ra-
diation by an electron is equal to 6.65·10–25 cm2 which results in a weak
signal of the scattered radiation measured in the experiments. Because
of this weak signal, the scattering method could not be used for labo-
ratory plasma diagnostics until lasers were invented.
The proposal to use the scattering of laser radiation for determin-
ing the plasma parameters was made almost simultaneously in Ref.93
and 94 in 1962. The theoretical fundamentals of the scattering method
and the possibilities resulting from this method in plasma diagnostics
were examined in a number of review articles 95–98 and also in mono-
graphs. 1,2,32,99-101
The spectrum of the radiation scattered by plasma and its interpre-
tation depend strongly on the plasma parameters. Scattering at free elec-
trons (Thomson scattering) takes place in low-density plasma, whereas
collective scattering takes place at high values of N e.
Scattering on a free electron

H H
Under the effect of an incident light wave E = E 0eiw 0t , propagating along
the x axis, the electron starts to vibrate and becomes at the same time
96
F ig .4.30 Scattering of the light wave by a free electron.

ig.4.30
an emitting centre. Figure 4.30 shows the wave vectors of the incident
H H
k 0 and scattered k S waves whose absolute values are equal to
H 2p w 0 H 2p w s
k0 = = and kS = =
l0 c ls c , and the directions coincide with the
directions of propagation of the incident wave and the examination di-
rection.
The amplitude of the radiation scattered by an electron at distance
r from it is
r0
Es = E sin j, (4.26)
r 0
e2
where r0 = 2
= 2.8 × 10 -13 cm is the classic electron radius, ϕ is the
mc
H
angle between the vector of the strength of the incident wave E0 and
H
the wave vector of scattered radiation k s .
The effective scattering cross section, representing the ratio of the
light flux, scattered in the direction ϕ within the limits of the unit solid
angle, to the intensity of incident radiation is
E s2 r 2
se = = r02 sin 2 j. (4.27)
E02
Integrating (4.27) in all directions gives the total Thomson section
97
Plasma Diagnostics
8p 2
s Th = r = 6.65 × 10 -25 cm2 . (4.28)
3 0
If the velocity of the electron on which scattering takes place is not

equal to zero, in accordance with the Doppler effect the frequency of
scattered radiation is displaced in relation to the incident radiation fre-
quency by the value
H H H H H
e i
w = w s - w 0 = v × ks - k0 = v × k , (4.29)
H
where v is the vector of the electron velocity. Thus, the frequency shift
in scattering is proportional to the projection of
H the velocity
H of the elec-
tron on the difference of the wave vectors k s and k 0 and not on the
examination direction as in the case in studying the radiation of a moving
source, forH example,
H an atom.
Since k s » k 0 , then
H w q 4p q
k » 2 0 sin » sin (4.30)
c 2 l 2
w 0vk q
w»2 sin , (4.31)
c 2
where θ is the angle between the direction of propagation of the inci-

dent wave and the examination direction, v k = v cos ϕ is the projec-
H
tion of the electron velocity on the vector k .
Scattering of radiation by plasma

The complex amplitude of the radiation scattered by plasma is deter-
mined by summing up the vectors of the complex amplitudes scattered
by the individual electrons.
Thin layers within which all particles are scattered in the same phase
are defined in the plasma. (In the case in which both the incident and
scattered waves are flat, these layers are parallel to the bisectrix of the
angle θ between directions of propagation of these waves (Fig.4.31).)
At distances l, 2l, 3l and so on from the first layer there are layers giving
scattered waves whose phases differ from the phase of the waves scat-
tered by the first layer by 2π, 4π, 6π, etc. From geometrical consid-
erations
98
Incident
wave
d
re
wa atte
ve
Sc
F ig .4.31 Scattering of a light wave by plasma.

ig.4.31
l 2p
l= = H .
2 sin q / 2 k (4.32)
In the gap between these layers there are layers of the same thick-
ness scattered in the antiphase in relation to the first layers. Conse-
quently, if the number of electrons of each layer were the strictly the
same, the scattered wave would be completely quenched. In reality, be-
cause of fluctuations of the electron density in the plasma, the com-
plete quenching of the waves, scattered by the individual layers, does
not take place.
The amplitude of the waves scattered by the k-th layer is Ak = Esnkeik,
where nk = ne + dnk is the number of the electrons in the k-th layer dif-
fering from the mean value ne by the value n k, and ϕ k is the phase
of the wave scattered by the given layer. The total amplitude of radiation,
scattered by all layers, is
A= å Ak = Es å bne + dnk geij k

= Es å dnk eij ,
k
(4.33)
k k k
å n e ij k = 0 in accordance with the previous considerations.

k
The mean value of the intensity of the radiation scattered in the given
direction is
c c 2 ibj - j g
å
*
Is = AA = Es dnk dn j e k .
8p 8p (4.34)
k, j
99
Plasma Diagnostics
Thus, the intensity of the radiation scattered by the plasma is

directly linked with the correlations of the fluctuations of the electron
density in plasma. The case of Thomson scattering (or scattering on free
electrons) corresponds to the complete absence of correlation. The so-
called collective scattering takes place in the presence of correlations.
Thomson scattering of radiation by plasma

In low-density plasma where the energy of charge interactions is neg-
ligible in comparison with the thermal energy of the particles, it may
be assumed that individual electrons are separated in space randomly
and the fluctuations of the number of particles within the limits of the
elementary layers are independent. Statistical averaging leads in this
case to the following expression for the intensity of scattered radiation:
c 2
Is = E
8p s å dnk2 , (4.35)
k
and since for the random distribution dnk2 = nk , then
c 2 c c
Is = E
8p s å nk = 8p E s2 ne = 8p E s2 N e DV . (4.36)
k
Here ne = å nk is the total number of electrons in the scattering vol-

k
ne
ume V; N e = is the electron concentration in plasma. It is evident
DV
that this equation can also be obtained by direct summation of the
intensities of the radiation scattered by all electrons within limits of the
volume V.
Using equation (4.27), equation (4.36) can be easily transformed to
the form
c E2 I
Is = s e 20 N e DV = s e 02 N e DV , (4.37)
8p r r
cE 02 W0
where I 0 = = is the intensity of incident radiation, W 0 and S
8p DS
is the total power and area of the cross section of the light beam falling
onto the plasma.
100
The total light flux of the scattered radiation, recorded by the de-
tector within the limits of the solid angle ∆Ω, is determined by the ex-
pression
F s = I s r 2 DW = s e N e DVDWI 0 . (4.38)
Since each electron in this case scatters radiation independently, the

scattered radiation spectrum forms as a result of superimposition of the
radiations of the individual electrons. The shift of the frequency of ra-
diation scattered by an individual electron is related with its velocity
by equation (4.29). If the velocity distribution of electrons is described
by the function f (v x), the probability of the electron having the com-
ponent of the velocity v x in the range from v x + dv x is
b g
dw = f vx dvx , (4.39)
The scattered radiation spectrum S (ω) is determined substituting the

value of v x from (4.31) into f (v x) (in this case it is assumed that the
H
x axis coincides with the direction of the vector k )
F I
b
S ws - w0 = g s0
f HGG
ws - w0
. JJ
f0 k H K (4.40)
Thus, examining the spectrum of the radiation scattered by the plasma

it is possible to determine directly the velocity distribution function of
the electrons. For Maxwell’s distribution
mvx2
b g
-
m 2 kTe
f vx = e (4.41)
2 kTe p
and the contour of the line of scattered radiation has the form of a Gauss
curve
LM m bw g OP = S expLM-FG w - w IJ OP,2
Sbw - w g = S exp -
2
- w0
MM 2kT PP MN H w K PQ
s s 0
s 0 0 H2 0
N Q
(4.42)
e k e
101
Plasma Diagnostics
2lTe H 2
where w e º k . The width of the scattering line, measured at
m
half height, is
LM kT kH OP
2 1/ 2
Dw = 2w aln 2f = 2 2aln 2f
MM PP
1/ 2 e
e . (4.43)
N m
Q
H
Transferring to the wavelength and substituting k from (4.30) gives
l0 F I F 2kT ln 2I
q
1/ 2
Dl = 4
H KH m K
e
sin . (4.44)
c 2
i.e. the electronic temperature of plasma can be determined from the

width of the scattering line.
The electron concentration can be determined from the absolute in-
tensity of scattered radiation (integral over the spectrum) in accordance
with equation (4.38).
We shall now determine the plasma parameters at which the acts of
radiation scattering by individual electrons can be regarded as independ-
ent.
As a result of the interaction of the charged particles in plasma the
disruption of quasi-neutrality of plasma takes place only within the limits
of the volume whose linear dimensions are considerably smaller than
the so-called Debye radius
F kT I
D= G
1/ 2
H 4pN e JK
e
2
. (4.45)
e
In the volume with the linear dimensions considerably smaller than

D the fluctuations of the number of electrons can be regarded as in-
dependent. Thus, the criterion of applicability of the Thomson scattering
model is the relationship between the Debye radius and the distance l
(see equation (4.32)) between the plasma layers for which the scattered
radiation phase changes by 2p. According to Salpet 102 such a criterion
is the parameter
102
1 1 l 4 pN e e 2
a= H = = .
k D 2 pD 4 p sin q kTe (4.46)
2
The case a << 1 correspond to scattering on three electrons, and the

case a >> 1 to collective scattering.
Collective scattering
At an arbitrary value of parameter α the intensity and spectrum of the
radiation scattered by the plasma are determined by the time and spatial
spectra of the fluctuations of the electronic density. The spectral flow
of the radiation scattered by the plasma can be calculated from the equa-
tion
H H
dFS
dw
e
= se s ks - k0 , w s - w 0 DVDWI0 , i (4.47)
H
e i
where s k , w is the spatial–time Fourier transforma from the function
of the paired correlations of the electronic density.100 Integrating (4.47)
with respect to frequencies gives
H H
e i
Fs = S ks - k0 se Ne DVDWI0 , (4.48)
ze
¥
H H H
e
where S ks - k0 = i 1
Ne
i
s k , w dw characterises the difference of the
-¥
collective scattering section from the scattering section on free electrons.
When l >> D (a >> 1) the thickness of the layers with coherent ra-
diation is greater than the dimension of the regions within which the
quasineutrality can be disrupted. The motion of each electron is strongly
affected by the electrostatic field of ions and other electrons. The motion
of the individual electrons can no longer be regarded as independent
and the fluctuations of electronic density are random. When the elec-
tronic and ion temperatures are equal (Te » Ti), the velocity of the elec-
trons is considerably higher than that of the ions. Consequently, any
displacement of the ion is accompanied by the collective displacement
of the entire electron cloud screening the ion charge. On the other hand,
the ions do not manage to follow the movement of the electrons and
this leads to the disruption of quasineutrality for the period of the order
103
Plasma Diagnostics
F I 1/ 2
of t e »
D
ve
= GH
m
4 pN e e 2
JK . The value reciprocal to this time is
F
4 pN e e 2 I 1/ 2
wp =
1
te
= GH
m JK , (4.49)
and represents the so-called electronic plasma frequency.

According to these considerations, in the centre of the fluctuations
of the electronic density we can separate the high-frequency (electronic)
component determined by the movement of the free electrons, and the
low-frequency (ion) component characterising the collective displace-
ment of the electron cloud associated with the movement of the ions.
H
Consequently, as s( k , w ) can be written with the form
H H H
e i e i e i
s k , w = se k , w + si k , w . (4.50)
A similar equality can also be written for the quantities integral over
the spectrum
H H H
ei ei ei
S k = Se k + Si k . (4.51)
The spectrum of electronic density fluctuations was calculated by

many authors. 102-106 Here we present (Fig.4.32a) the results of calcu-
lation carried out by Salpeter for the case of Maxwell’s velocity dis-
tribution of electrons assuming that N eD >> 1, i.e. in a sphere whose
radius is equal to Debye radius there are a large number of electrons.
The value a = 0 is related with the Gauss contour whose width is linked
with T e by the relationships (4.43) and (4.44). As the parameter a is
increased the electronic component of the spectrum breaks up into two
components whose distance from the probing radiation line is ±Dw and
which satisfies the equation
3kTe H 2 3 F I
Dw 2 = w 2p +
m
k = w 2p 1 + 2 .
a H K (4.52)
At a >> 1Dw tends to w p.
104
a b
F ig .4.32 Scattering spectra at different plasma parameters. a) electronic component

ig.4.32
of the spectrum at different values of parameter α, 102 b) ionic component of the
spectrum at α > 1 and different ratios of the electronic and ionic temperatures; 103
the abscissa gives the distances from the laser line ω = ω S – ω 0 related to ω e =
(2 kT e / m ) 1/2 and ω i = (2 kT i/ M ) 1/2, where M is the mass of the ion.
The integral (with respect to frequency) contribution over the elec-

tronic component to the intensity of scattered radiation is determined
by the equation
H
ei
Se k =
1
1 + a2
(4.53)
and rapidly decreases with increasing a, whereas the contribution of

the ion component to the scattering radiation intensity increases with
increasing a:
H
ei
Si k =
F
za 4
I .
d1 + a iGH1 + a x T a JK
2 2T e
i
2 (4.54)
The width of the ion component of the scattering spectrum is deter-

mined by the velocity of the ions and is considerably smaller (approxi-
mately M / m times) than the width of the electronic component. The
shape of the ion line depends on the ratio of the electron and ion tem-
peratures and on the ion charge. At T e << T i it consists of two com-
ponents shifted in relation to w 0 by the frequency of ion acoustic vi-
brations, and at T e << T i the ion line has the Gaussian contour
(Fig.4.32b).
105
Plasma Diagnostics
Thus, at the values a >> 1 the scattered radiation spectrum consists

of a narrow central peak whose width is determined by the thermal ve-
locities of the ions, and two weak electronic components displaced in
relation to the centre by the value of the electronic plasma frequency
w p. Consequently, at a >> 1 the width of the ion peak is used to cal-
culate the ion temperature and the electron concentration is calculated
from its intensity integral with respect to the spectrum. In addition, the
electron concentration can be found from the distance from the central
maximum to the side maxima, in accordance with equation (4.49).
When a » 1, the plasma parameters are estimated by comparing the
observed contour of the scattering line with the calculated theoretical
contours corresponding to different temperatures and electron concen-
trations in the plasma.
As indicated by equation (4.46), the value a depends not only on
the plasma parameters but also the examination angle and the wave-
length of scattered radiation. For example, for the same plasma in scat-
tering under large angles where sin (q/2) @ 1, the value of α can be
lower than 1 and the scattering close to Thomson scattering. In scat-
tering under very small angles where sin (q/2) << 1, there may be cases
in which the value of parameter a exceeds the unity and collective effects
appear in the scattered radiation spectrum. The appearance of the col-
lective effect is also supported by the increase of the wavelength. The
appearance of the collective effect is also supported by the increase of
wavelength. For example, for the same examination angle, the transi-
tion from the radiation of a ruby laser to that of a CO 2 laser increases
the value of a up to 15 times.
Experimental procedure and apparatus for examining plasma-scat-

tered radiation
Experimental equipment for examining the plasma by the scattering
method should ensure the introduction of laser radiation into a plasma
chamber, collection of radiation, scattered within the limits of the given
angle, and recording the spectrum of scattered radiation.
Figure 4.43 shows the circuits of two typical systems for examin-
ing scattering. In the first system, (Fig.4.33a) scattered radiation is
observed under a high angle q to the direction of incident radiation (in
the majority of cases q = 90°). Figure 4.33b shows equipment for ex-
amining radiation under small angles. Both systems consist of lasers
1, focusing optics 2, plasma chamber 3, optical elements 4, collecting
the scattered radiation and directing on to the spectral device 5 whose
output contains the radiation detector 6, and the device for recording
the scattering signal 7.
106
As already mentioned, the main problem in measuring the intensity

of plasma-scattered radiation is the weak used signal. The light flux,
leaving the spectral device (monochromator), is determined by the re-
lationship 110 within the limits of the spectral range
a f
F reg = bl L dl ,
2
(4.55)
where bl is the spectral brightness of the light source, L = tbSlj is the

light force of the monochromator, t is the transmission factor which
takes into account the losses of light in the device, b is the angular height
of the input slit, S is the effective area of the dispersing element, D j
is angular dispersion.
In turn, the spectral brightness is represented by the light flux emitted
by the unit visible surface of the source in the unit solid angle and the
unit spectral range
DS
bl = , (4.56)
ldDWDl
where l is the length of the scattering volume, d is its diameter. Sub-

stituting F S for (4.38) into (4.56) and taking into account that
DV = lDS, and I 0 = W 0/DS, and gives
L1 L2 D
1 D2
Lc
DFS-12 oscilloscope
D1 PEM 2 Oscilloscope
L
PD D2
PEM 1
L1 L2
θ-pinch
Oscilloscope
F ig .4.33 Typical diagrams of equipment for plasma diagnostics by the scattering

ig.4.33
method. a) equipment for examining scattering under an angle of θ = 90°, 107,108 b)
the same, under a small angle θ (in the forward direction). 109
107
Plasma Diagnostics
s e N eW0
bl = . (4.57)
dDl
To construct the contour of the scattered radiation line, the width

of the separated spectral range dl must be p times smaller than the width
of the spectrum of scattered radiation Dl (p is the number of points
used to construct the scattered radiation spectrum), i.e. p = Dl/dl. Con-
sequently, substituting b from (4.57) into (4.55) we obtain
s e N eW0 LDl
F reg = . (4.58)
p2d
The useful signal recorded by the photodetector can be determined

as the number of electrons ne knocked out by the recorded radiation from
its photocathode
hF reg Dt hs e N e LlDlE 0
ne = = . (4.59)
hv hcp 2 d
c
Here h is the quantum yield of the photocathode, hn = h is the quan-
l
tum energy, Dt is the laser pulse time, E0 = W0Dt is the total laser pulse
energy.
In accordance with (4.59), the strength of the useful signal is de-
termined directly by the laser pulse energy, the light force of the spectral
device and the sensitivity of the radiation detector. Therefore, general
requirements of the elements of equipment for examining scattering are
reduced to using lasers that emit a sufficiently large amount of energy
in the pulse and to the use of apparatus with a high light force and the
most sensitive radiation detectors (with a high quantum yield).
At the same time, the useful scattering signal is recorded on the back-
ground of various types of interference associated with the intrinsic glow
of plasma and also the parasitic scattering of laser radiation at opti-
cal elements and components of plasma equipment. Therefore, to increase
the signal/noise ratio, it is necessary, on the one hand, to increase the
laser radiation power and, on the other hand, take measures to reduce
the level of parasitic scattering.
The requirements of apparatus used for examining scattering were
examined in detail in Ref.2. Here, we shall confine ourselves only to
more detailed examination of two typical experimental systems described
previously.
108
Equipment shown in Fig.4.33a was used for examining the scattering

of the light by the plasma of dc arcs in a magnetic field. 107,108 The
radiation source was a ruby laser in the free generation regime (E 0 =
25 J). Laser radiation was focused in the centre of the plasma column
using a system consisting of two lenses L 1 and L 2, the first of which
focuses radiation on diaphragm B and the second constructs the image
of the diaphragm in the centre of the chamber. The diameter d 1 of dia-
phragm B is determined from the condition d 1= xf 1, where x is the di-
vergence of laser radiation, f 1 is the focusing distance of length L 1. A
large part of laser radiation passes through the diaphragm and, at the
same time, the cone of the beams directed onto the plasma is restricted.
This is very important for attempts to reduce the intensity of parasitically
scattered radiation.
Two nozzles with glass windows are used for input and output of
radiation from the chamber. The output is positioned under the Brewster
angle to the incident plane-polarised laser beam thus preventing the
propagation of the radiation reflected from it into the chamber. Inside
the nozzles there is a system of diaphragms D 1 and D 2 used to reduce
the radiation scattered by the windows and walls of the chambers.
The radiation scattered by the plasma under angle q = 90° is col-
lected by the condenser lens L c from the volume 7 mm long and 0.6
mm in diameter in the solid angle DW = 1/32 mean and after passing
through the prism P, which rotates the image by 90°, is directed onto
the slit of the DSS-12 double monochromator whose parameters, de-
termining its light force, are t = 0.1, S = 180 cm 2 , D = 2.5·10 4
cm –1 , b = 0.05.
Radiation is recorded using the FEU-38 photodetector and an oscil-
loscope. The signal corresponding to scattering in plasma is determined
from equation
IS = I - I pl - I par (4.60)
where I is the signal recorded in the presence of the laser pulse and
in the presence of plasma, I pl is the signal determined by the natural
glow of plasma in the absence of the laser pulse, I par is the parasitic
signal observed in the presence of the laser pulse and in the absence
of plasma. The contour of the scattering line is recorded from individual
points. Regardless of the fact that the parasitic signal I par at the wave-
length of laser radiation exceeded 30 times the useful signal I S, the
‘wings’ of the scattering line were recorded with sufficient reliability.
Absolute calibration of the entire measuring system was carried out
on the basis of Rayleigh scattering in helium with which the plasma
109
Plasma Diagnostics
chamber was filled. This made it possible to eliminate errors associ-

ated with measuring quantities such as the absolute values of the la-
ser energy, the light force of the spectral device, the sensitivity of
photodetector, etc.
The measured width of the scattered radiation contour Dl = 43 Å
was used to determine the electron temperature T e = 1.8 eV, and the
electron concentration N e = 2.5·10 13 cm –3 was determined from the ab-
solute intensity of scattered radiation. The parameter calculated for these
values of Te and N e was a = 0.04, i.e. scattering in this case took place
in free electrons.
The equipment shown in Fig.4.33b is typical for examination of scat-
tering and small angles.109 Scattering took place in the plasma of q-pinch
in the ‘forward’ direction. The direct laser beam, passed through the
plasma, was absorbed by special traps 8. The radiation, scattered under
small angles, was outputted using a conical length (axicone) A 1 with
a orifice in the central pass through which the direct laser beam passed.
The axicone deflects all beams falling on it under the same angle
j » (n – 1)b (b is the angle at the base of the cone). Thus, all the beams,
leaving the scattering volume under the angle q = j to the axis, propagate
parallel to the axis of the system after passing through the axicone and
are separated by the diaphragm D1 positioned in the focus of the lengths
L. Moving the diaphragm along the optical axis it is possible to ob-
serve the radiation scattered by the plasma under different angles.
The spectrum of scattered radiation was examined by a Fabry-Perot
(FP) interferometer and the second axicone A2 directing the beams, cor-
responding to one of the interference rings, to diaphragm D2 positioned
in the focus of the lens L 2. The diameter of the diaphragm determined
in this case the width of the separated spectral range and the spectrum
was scanned by the moving axicone along the optical axis.
In initial studies in this area 107-109,111,112 the scattering signal was re-
corded using a photodetector and an oscilloscope and the spectrum was
obtained from the individual lens. Subsequently, to record the entire spec-
trum of scattered radiation in a single pulse it was necessary to develop
different multichannel systems 103,104 and also electron–optical convert-
ers (EOC). For example, in Ref.115,116 EOC was used to examine the
spectrum of collective scattering in examining a laser spark. The re-
gion of the spectrum 10 mm long with the centre on the line of the ruby
laser and separated by the spectral device was projected on to the cathode
of the EOC. To avoid superimposition of the natural glow of the plasma
on the scattered radiation, the time sweep of the spectrum was carried
out. Figure 4.34 shows microphotographs of the spectrum photographed
from the screen of the EOC and corresponding to moments of time 3.6
110
(a), 10.4 (b) and 21 µs (c) after formation on the laser spark. The spec-
trum shows clearly the central ion maximum and satellites at the plasma
frequency.
Further advances in experimental techniques used in the laser scat-
tering method were associated with using this method of diagnostics of
the high-temperature plasma in large toroidal systems of the Tokamak
type. 6 To increase the information content of the method, investigators
used multichannel recording systems based on polychromators of dif-
ferent types. The focal plane of a spectral device is separated into a
number of spectral regions using a package of light guides so that the
entire spectrum of scattered radiation can be recorded in a single plasma
pulse.
The radial distribution of the plasma parameters is determined us-
ing a set of packets of light guides where each light guide shows an
image of a specific region of the slit of the spectral device in the di-
rection of height, or a set of modules (polychromators) of the same type
is used where each records the radiation spectrum scattered from some
point of the plasma cord.
Simultaneous recording of scattering spectra in a large number of
spatial points in TFTR Tokamak equipment 117 was carried out, using
as a detector a two-coordinate television system matched with a
polychromator. Measurements were taken in 76 points in space and 25
channels with respect to spectrum.
F ig .4.34 Microphotographs of spectra of collective scattering photographs from the

ig.4.34
screen of the EOC. 116
111
Plasma Diagnostics
The information possibilities of the scattering method are indicated

most convincingly by the diagnostic complex used in ASDEX Tokamak
equipment. 118,119 It includes a laser (l = 1.06 µm) operating in the fre-
quency regime (the energy in each pulse is 0.8 J), silicon avalanche
detectors and a set of 16 modules of three-channel polychromators. The
system makes it possible to measure, on the basis of the scattering spec-
trum, to measure the temperature and electron concentration simulta-
neously at 16 points on radius. The frequency laser generation regime
makes it possible to find the radial distributions Te (r) and N e (r) every
16 µs. Up to 400 temperature and electronic density profiles can be re-
corded in a single discharge.
Determination of plasma parameters from scattered radiation spectra

Since it is not possible to review the large number of investigations in
which the diagnostics of various plasma objects was carried out using
the scattering method, we shall demonstrate here the possibilities of
evaluating various plasma parameters on the basis of scattered radia-
tion spectra using several typical studies.
As already mentioned, the interpretation of the scattering spectra de-
pends on the value of parameter α. In scattering of radiation on low-
density plasma (Da << 1) with the Maxwell distribution of electron
velocity, the scattering line has the Gaussian contour (see (4.42)), and
the width of this contour can be used to determine the electronic tem-
perature of plasma T e using equation (4.44). As an example, Fig.4.35
shows the spectra of scattered radiation recorded in Tuman-2 toroidal
equipment. 120 The Gaussian contours were constructed on the basis of
experimental points using a computer. The spectra shown in Fig.4.35
were obtained prior to (a) and (b) adiabatic compression of the plasma
by the magnetic field and indicate a large increase of the electron tem-
perature in the compression stage.
If the velocity distribution of the electrons differs from Maxwellian,
the contour of the scattering line can be used to draw conclusions on
the nature of these differences. In particular, the authors of Ref.121 found
an anisotropy in the velocity distribution of the electrons in the direction
along and across the magnetic field, and the form of the spectrum ex-
amined in Ref.122 indicated the presence in plasma of two electron
groups each of which had Maxwellian velocity distribution corresponding
to two different temperatures.
In the presence of directional movement of the electrons the scat-
tering line is shifted in relation to the wavelength of laser radiation.
For example, in scattering on an electron beam with an energy of 2 kV
(see Ref.123), the Doppler shift of the scattering line examination under
an angle of 65° in relation to the direction of electron motion reached
112
I s , rel. units
I s , rel. units
F ig .4.35 Spectra of scattered radiation obtained in equipment Tuman-2. 120 Points

ig.4.35
are experimental data, the curves are the calculated Gauss contours.
259 Å. The Doppler shift of the scattering line was also observed in
investigations of the laser spark 124-126 so that it was possible to meas-
ure the travel velocity of the plasma boundary in the direction against
the laser radiation which equalled around 100 km/s.
In accordance with (4.38), measuring the scattered light flux F S, we
can determine the electron concentration Ne in plasma. However, the ab-
solute energy measurements are always associated with large experi-
mental errors. It is far easier and more accurate to calibrate appara-
tus on the basis of Rayleigh scattering. For this purpose, the plasma
chamber is filled with a gas with the known scattering section and the
Rayleigh scattering signal is measured at the same intensity of the laser
pulse, the geometry of optical equipment and the sensitivity of the meas-
uring system. Consequently, all absolute measurements are eliminated
and only the ratio of two signals – scattering on plasma and Rayleigh
scattering – is measured:
F S s e N e s Th N e
= = . (4.61)
F R sj N a s R N a
Here sj is the Rayleigh scattering section in the direction j in the unit

solid angle calculated for a single atom in the unit volume, and s R is
the Rayleigh scattering section integral with respect to angles. This
section can be calculated from Rayleigh’s equation 127
113
Plasma Diagnostics
F I
8p w 4
sR =
H K
3 c
a 2a , (4.62)
where a a is the polarisability of the atoms.

Equation (4.61) can be used to determine N e when a << 1. In the
H
opposite case, s Th in equation (4.61) should be replaced by s Th S( k )
H H H
where S( k ) = S e( k ) + S e( k ). At a >> 1, the main contribution to the
intensity of scattered radiation, integral with respect to spectrum, is
provided by the central ion component and, in accordance with (4.54),
at z = 1 and T e = T i S i @ 1/2, i.e. the scattering section is half the
Thomson section. The intensity of the ion peak was used to examine
the spatial distribution of the spectrum concentration in a laser spark
plasma in air. 112 Figure 4.36 shows the distribution of N e in the laser
spark obtained in Ref.112.
When a » 1, the main contribution to the intensity Hof the scatter-
ing spectrum is provided by the electronic component Se( k ). The plasma
parameters are determined comparing the contour of the scattering line
with the calculated contours corresponding to different a. The observed
and theoretical contours can be matched only by selecting both the ac-
curate value of a and a specific scale on the frequency axis. Since the
theoretical curves (Fig.4.32b) are formed by plotting w/w e on the ab-
F 2kT I 1/ 2
scissa, then we determine not only a but also w e = k HmK
e
and this
F ig .4.36 Spatial distribution of the concentration

ig.4.36
of electrons in the plasma of the laser spark in
the stage t = 3 5s. 112
114
a b
I, rel. units
F ig .4.37 Spectra of radiation scattered by the plasma of q-pinch. 128 Points are the
ig.4.37
experimental values, solid lines are the experimental contours of the scattering line:
a) T e = 90 000 K, N e = 2·10 16 cm –3 , a = 0.53; b) T e = 55 000 K, N e = 4·10 16 cm 3 ,
a = 0.97; broken lines are the positions of the laser lines.
is used to determine Te. Substituting Te into (4.46), N e is calculated from

the known value of a.
As an example, Fig.4.37 shows the scattering spectra on the plasma
of q-pinch obtained in Ref.128.
When a >> 1, the scattering spectrum consists of a narrow central
ion peak and weak satellites whose distance from the ion peak is ap-
proximately equal to the distance of the electronic plasma frequency w p
so that equation (4.49) can be used to calculate N e. If the width of the
range defined by the spectral device is considerably greater than the
width of the ion and electron peaks, the ration of their intensities is
H H
proportional to 2S i( k )/S e( k ) and in accordance with (4.53) and (4.54)
for z = 1 and T e = T i is equal to
H
2 Si ( k ) 2a 4
H =
ei
,
Se k 1 + 2a 2 (4.63)
so that the value of a can be determined. Consequently, substituting

the values of a and N e into (4.46) we can determine T e.
Of special interest for experiments is the plasma for which a > 1
in examining scattering at large angles, and in examination at small
angles a > 1. An example of such a plasma is the q-pinch character-
ised by the values N e » 10 15 ¸ 10 16 cm –3 . Figure 4.38 shows the scat-
tering spectra on the plasma of q-pinch obtained in the postglow stage
in examination under angles q = 13.5° (a) and 90° (b). 129 The values
a = 3, N e = 2.4·10 15 cm –3 and T e = 1.1 eV were determined from the
ratios of the intensities of the central and lateral maxima and from the
distance between them (Fig.38a). Similar values, N e = 2.4·1015 cm–3 and
Te =1.0 eV, were determined from the intensity integral with respect to
the spectrum and from the width of the contour of the scattering line
115
Plasma Diagnostics
(Fig.4.38b). The value of a in this case was » 0.5.

The information on the plasma parameter is also obtained by exam-
ining the shape of the ion line and the plasma satellite. 130 In particu-
lar, the width of the ion peak is used to determine the ion temperature
T i and the ratio T e/T i is determined from the shape of this peak (see
Fig.4.32b).
It is also important to note the possibility of measuring the strength
of the magnetic field in the plasma from the scattering spectrum. In the
presence of a magnetic field, the electrons move in the plasma along
eB
a spiral around force lines with the rotation frequency w B = (elec-
mc
tronic cyclotron frequency). Consequently, the contour of the scatter-
ing line is modulated by the frequency w B. The modulation depth is
H
maximum if the angle g between vector k and the vector of the strength
H
of magnetic field B is equal to 90°. If g differs from 90°, the modu-
lation depth rapidly decreases and this restricts the value of the solid
angle DW within which the radiation scattered by the plasma is collected.
Regardless of considerable experimental difficulties, in a number of
studies 131,132 it was possible to record the modulation of the scattering
spectrum with respect to frequency w B (Fig.4.39). The strength of the
magnetic field (125 kHz) was determined from the distance between the
adjacent components (» 5.7 Å).
Plasma diagnostics based on radiation scattering on atoms

As mentioned at the beginning of this section, if the wavelength of laser
a
I, rel. units
F ig .4.38 Spectra of radiation scattered by θ -pinch plasma in the forward direction

ig.4.38
(a) and under a right angle to the laser beam (b). 129
116
I, rel. units
F ig .4.39 Spectra of radiation scattered by the plasma in the magnetic field. 131
ig.4.39
B = 125 kilogauss (1), 0 (2); 3) theoretical curve.
radiation is at a sufficiently large distance from the absorption lines

of the atoms, the main contribution to the scattering signal comes from
the electronic component of the plasma. Rayleigh scattering on atoms
or molecules is used (because of its small section (σ R << σ Th)) mainly
for calibrating the measuring system. To ensure that the ratio of the
scattering signals with respect to the electronic component F S to the
Rayleigh scattering signal F R is of the order of unity (see equation
(4.61)), the atom concentration N a should be several orders of magni-
tude higher than the measured electron concentration N e.
However, when approaching the absorption line, the scattering section
on the atoms rapidly increases so that this phenomenon can be used for
plasma diagnostics. The scattering process consists of two acts: absorp-
tion of the photon and subsequent radiation. The scattering section on
the atoms s a can be determined from equation (4.15) transferring from
wavelength to frequencies
sa =
k
= r0 cfki
a Dn / 2 f
b g a f
2 2
, (4.64)
na n ki - n + Dn / 2
where r 0 is the electron radius, f ki , ν ki is the force of the oscillator and

the frequency corresponding to the transition between the levels k and
i, Dn is the width of the absorption line which in the case of transi-
tion between the ground and excited levels is determined by the radiation
lifetime of the upper level.
When approaching the absorption line, both Rayleigh scattering with-
out frequency change, i.e. at frequency n, and fluorescence at frequency
n ki, induced by collisions (Fig.4.40a), can take place simultaneously.
The laser scattering lines are narrow and the width is determined by
the temperature of the atoms and ions, whereas the lines of induced fluo-
117
Plasma Diagnostics
rescence broaden as a result of collisions and their contents provide in-

formation on the processes of inelastic collisions of excited atoms. The
ratio of the intensities of Rayleigh scattering and induced fluorescence
depends on frequency difference nki–n, the laser radiation intensity and
the parameters of the examined plasma, especially its optical thickness.133
Resonance fluorescence appears when the frequency of the scatter-
ing radiation coincides with the natural frequencies of the atoms. This
name was used due to tradition because in the initial studies, carried
out prior to invention of the lasers, investigators used resonance radiation
corresponding to transition between the ground and first excited lev-
els of the atom. However, in principle, radiation can coincide with any
sufficiently intense absorption line, including the one which starts on
the excited level. As a result of the absorption of the light quanta with
frequency n ik corresponding to the transition between the levels k and
i (Fig.4.40c), part of the atoms is transferred from the lower state k
to the upper state i. Spontaneous transitions of the atoms from the level
i to k results in radiation with the initial frequency n ik. If transitions
from state i are possible not only to level k but also other underlying
levels (n, m, etc.), the fluorescence spectrum also contains frequencies
n in and n im , etc.
The fluorescence method has greater historical merit in comparison
with spectroscopy because it was one of the main methods of exam-
ining the distribution of energy levels of the atoms. Recently, as a result
of extensive use of tuneable dye lasers, this method has been used on
an increasing scale in fluorescence analysis 134 and also plasma diag-
nostics. 6,135
The fluorescence intensity is determined by the number of absorb-
ing atoms and its measurement can be used to determine their concen-
tration N k. The lifetimes of the atoms of a given level are measured
examining the time dependence of the fluorescence signal. The form of
the fluorescence lines is determined by the same processes as the form
a b c
F ig .4.40 Scattering of radiation on atoms. a) Rayleigh scattering, b) induced, c)

ig.4.40
resonance fluorescence.
118
of the radiation lines of plasma particles so that the contour of the fluo-
rescence line can be used to determine the electron concentration if the
Stark effect in the electric fields of charged particles is the main broad-
ening mechanism, or temperature if the line broadens as a result of the
Doppler effect.
An advantage of the resonance fluorescence method in comparison
with the conventional emission diagnostic method is its localisation. As
in the laser scattering method, the examined radiation is observed from
a small plasma volume in which the laser beam and the cone of the
beams recorded by the detector intersect and, consequently, provides
information on the plasma parameters within the limits of this volume.
Strictly speaking, this holds for optically thin plasma. In examination
by the fluorescence method of the plasma with a large optical thick-
ness it is important to take into account the absorption of resonance
radiation in the vicinity of the scattering region where the effective
volume of this region may increase and the glow brightness may change.
The strength of the recorded fluorescence signal is
F fl = jik DVDW, (4.65)
where j ik is the light flux emitted by the unit volume in the unit solid
angle, DV is the plasma volume from which fluorescence is observed,
DW is the solid angle within which the radiation is recorded. In turn,
1
jik = A hv DN , (4.66)
4p ik ik i
where A ik is the probability of a radiation transition, DN i = N i – N i0

is the change of the population of the upper level as a result of absorp-
tion of probing radiation. Substituting (4.66) into (4.65) gives
1
F fl = A hn DN DVDW. (4.67)
4p ik ik i
The population of level N 0i in the absence of laser radiation is de-

termined by the combined effect of different processes (collisions with
electrons and atoms, spontaneous and forced radiation, absorption,
cascade transitions from higher levels, etc.). Under LTE conditions, the
combined effect of these processes leads to the Boltzmann distribution
119
Plasma Diagnostics
hvik
N i0 gi - kTe
= e , (4.68)
N k0 g k
where g i and g k are the statistical weights of the levels.

In passage of a laser pulse through the plasma the equilibrium dis-
tribution of the population is disrupted. For accurate calculation of the
populations of the level, it is important to take into account all the proc-
esses leading to an increase or decrease of the population, and it is also
essential to know the probabilities of the corresponding impact and ra-
diation transitions. However, in a number of partial cases, the prob-
lem can be greatly simplified.
As an example of such a simplification, we shall examine a two-level
system in an electromagnetic field. It is assumed that the initial popu-
lation of the upper level prior to applying the field is close to zero
(N 0i » 0 and DNi » Ni). This corresponds to the case of low temperatures
(kT e << hv ik ). Further population of the level takes place as a result of
absorption of radiation, and its depletion – as a result of forced and
spontaneous radiation transitions and also quenching collisions. Con-
sequently, the differential equation, describing the pumping process, has
the form 136
D
dNi
dt
af af af af
= Nki rv t N k t - Bik rv t + A Ni t , (4.69)
where A = Aik + A q; Bki, A ik are the Einstein coefficients; A q is the prob-

ability of quenching collisions; rn (t) is the spectral density of the pump-
ing radiation flux. Assuming that the total number of the atoms on both
levels N 0 = N k + N i does not change, and the laser pulse is right-an-
gled, i.e. r n (t) = r n at 0 £ t £ t and r n (t) = 0 outside this range,
equation (4.69) is transformed to the form
dNi
dt
b g
= Bki r n N0 - Bki + Bik r n + A Ni . (4.70)
If a laser pulse is sufficiently long, a stationary state is established

dN i
in which = 0 . Therefore, from (4.70) we determine the stationary
dt
value
120
Nor n Bki
Nist =
b g
Bki + Bik rn + A
. (4.71)
We shall now examine two extreme cases – high and low values of
A
r n. If r n << B + B , then
ki ik
N 0 N ki r n
N ist » , (4.72)
A
i.e. the population of the upper level in the stationary state is directly
proportional to the radiation density. Therefore, the fluorescence sig-
nal also increases linearly with increasing rn. Substituting Nist from (4.72)
into (4.67) instead of DN i and replacing A by A ik (this corresponds to
the absence of quenching collisions) we obtain
1
F fl = hn DVDWN 0r n . (4.73)
4 p ik
A
In the second extreme case, where r n >> B + B , as indicated by
ki ik
(4.71)
N 0 Bki N 0 gi
N ist » = (4.74)
Bki + Bik gi + g k
(it is taken into account here that B ki g k = B ikg i and does not depend on
r n). Correspondingly, the fluorescence signal is also independent of r n
and in this case is equal to
1 N 0 gi
F fl = A hn DVDW. (4.75)
4 p ik ik gi + g k
This case corresponds to absorption saturation and fluorescence.

Strictly speaking, saturation starts at infinitely high values of rn (rn®¥),
but high non-linearity of the dependence of Fsl on rn is already observed
at
121
Plasma Diagnostics
A
r n = r sat
n º . (4.76)
Bki + Bik
This value is usually used to estimate the saturating flux density.

In the absence of quenching collisions, taking into account that
8 ph g k c
A = Aik = Bki , and transferring to r l = r n 2 , we obtain
l 3 g
i l
8phcgk
l =
r sat
b
l gi + gk
5
.
g (4.77)
A characteristic feature is the strong dependence of rsatl on wavelength.

This leads to a rapid increase of the values of r lsat in the ultraviolet
region of the spectrum. The table of calculated values of the saturat-
ing flux density for different atomic transitions was presented in Ref.134.
From the equations (4.73) and (4.75) for the light fluorescence signal
we can transfer to the total number of the photoelectrons recorded by
the detector during the laser pulse time Dt
ne = F fl ThDt , (4.78)
where T is the transmission of the optical system, h is the quantum yield

of the photocathode. Consequently, after specifying the minimum meas-
ured value of n e, this makes it possible to estimate the minimum atom
concentration N 0 detected by the examined method.
Sensitivity of the fluorescence method is so high that individual atoms
can be recorded in some cases. 137 This is possible due to multiple ab-
sorption and repeated radiation of the photons in an ideal two-level sys-
tem. When using a three-level system, sensitivity is lower. For exam-
ple, the sensitivity of detecting Pb atoms in Ref.138 was 250 at/cm 3.
The equations (4.73) and (4.75) were derived ignoring the popula-
tion of the upper level in the absence of laser radiation. At sufficiently
high temperatures when the initial population of the upper level can-
not be ignored, N 0 in equations (4.73) and (4.75) should be replaced
FG gk N i IJ
H
by N k 1 -
gi N k K
. Here the expression in the brackets takes into ac-
count the depletion of the upper level as a result of induced emission.

The fluorescent signal is observed in the background of natural plasma
radiation.
122
The natural radiation flux of the plasma, recorded by the detector,

originates from the volume V and is determined by the relationship
1
F pl = A hn N 0VDW. (4.79)
4 p ik ik i
From (4.67) and (4.79) we obtain that
F fl DNi DV F DV I
F pl
= 0
Ni V
= GH
Ni
0
Ni
-1
V
.JK (4.80)
The ratio of the volumes DV/V in the first approximation is equal

to the ratio of the diameter of the focusing spot of the laser beam to
the linear dimension of the plasma and is usually far smaller than unity.
Substituting in (4.80) the value N 0i, corresponding to LTE conditions
(see (4.68)), and N i which corresponds to the saturation conditions, and
assuming that N 0I + N 0k = N i + N k = N 0, we obtain
LM F hn I - 1OP
MN GH kT JK PQ DV .
ik
gk exp
F fl e (4.81)
=
F pl gi + gk V
Equation (4.81) shows that at the values kT e > hn ik , the ratio

F fl
D << 1 . This reduces the sensitivity of measuring the atom concen-
F pl
tration.
Thus, natural glow is in fact the main interference when examining
the fluorescence signal. The interference from the side of parasitic scat-
tering of the optical elements of the system and the walls of the plasma
chamber in this case play a considerably less significant role than in
the laser scattering method, owing to the fact that the fluorescence cross
section is considerably greater than the Thomson scattering cross section.
In addition, the corresponding signals can be separated on the basis of
their different time dependence (parasitic scattering takes place with-
out inertia, and the duration of the fluorescence signal is determined
by the lifetime of the upper level). In cases in which the observed signal
does not correspond to the transition whose emission causes pumping
of the upper level, the wavelengths of the fluorescence and parasitic
123
Plasma Diagnostics
scattering signals differ and they are quite easy to separate. This also
relates to examining the induced fluorescence signal displaced with
respect to frequency in relation to the laser radiation frequency by the
value n ki – n.
In initial studies concerned with the application of the resonance fluo-
rescence method for plasma diagnostics, 139,140 investigations were car-
ried out using barium plasma and the light source was a high-pressure
arc. This was followed by using this method in a series of investiga-
tions into the diagnostics of the plasma containing alkali metals, us-
ing tuneable lasers as a radiation source. In Ref.141, the concentration
of barium atoms in the flame was determined by the fluorescence of
the resonance barium line excited using a dye laser.
In Ref.142, the fluorescence method was used for diagnostics of po-
tassium plasma. Fluorescence was excited by radiation of a ruby la-
ser whose wavelength coincided with the potassium line l = 6939 Å
(4P 3/2 ® 6S 1/2) as a result of cooling the ruby crystal to a temperature
of –45°C. The fluorescence signal was observed under an angle of 90°
to the incident laser beam and was recorded using a monochromator,
a photomultiplier and an oscilloscope. When the wavelength of laser
radiation was displaced by 0.06 Å from the absorption line, the fluo-
rescence signal disappeared. Fluorescence was observed at two lines
l = 6939, 6911 Å. The decay time of fluorescence in measuring the
plasma parameters was examined. In this case, the decay time depended
on the electron concentration so that these measurements could be used
to determine Ne. The measurements of the ratio of the fluorescence signal
to the intensity of the emission line enabled the authors to determine
T e (see equation (4.81)). According to the estimates of the authors, the
method is suitable for potassium plasma diagnostics with the param-
eters N e = 10 11 ¸ 10 11 cm –3 and T e = 2000 ¸ 3000 K.
In Ref.143, the resonance fluorescence method was used to meas-
ure the concentration of sodium atoms in the octupole plasma and in
Ref.144 – for determining the spatial distribution of the atoms and ions
of barium in an arc discharge in a magnetic field. The spatial resolu-
tion power obtained in Ref.144 was 0.2 mm 3.
In a number of studies, the fluorescence method was used to examine
the distribution of impurities in the plasma evaporating from the walls
of the plasma chamber and separated from metallic electrodes. The con-
centration of Fe, Cr, Ni, Mo, Ti, W in both cold 145,146 and hot 147,151
plasma was determined.
The fluorescence method was used to measure the concentration of
hydrogen, 152,153 helium, 154 argon 155 and other gases.
In Ref.154, a dye laser set to a wavelength of l = 4471 Å was used
for selected excitation of the 4d 3D level of a helium atom. The fluo-
124
rescence signal was detected at both the line l = 4471 Å, corresponding

to the transition 4d 3D ® 2p 3P and in a number of other helium lev-
els, including those originating from levels to which radiation transi-
tions from the level 4d 3D are forbidden. The population of these lev-
els is determined only by collisions with atoms and electrons. Exam-
ining the time dependence of the fluorescence signal for the corresponding
lines, the authors of Ref.154 estimated the probabilities of radiation and
impact transitions.
In low-temperature hydrogen plasma, the fluorescence method in the
visible range was used to measure the concentration on the second 152
(in excitation of fluorescence of the line H a) and on higher levels. 153
Fluorescence on the line H a in Ref.156 and H b in Ref.156 was used
for determining the concentration of excited hydrogen atoms in high-
temperature plasma in Tokamak-type equipment. The atom concentration
was measured directly on levels with the main quantum number n = 2;
3, and the concentration of normal atoms (n = 1) was calculated on the
basis of balance of populations of the excited state in accordance with
the physical model of the plasma. For Tokamak plasma, the physical
model can be represented by the coronal model. Figure 4.41 shows the
spatial distribution of the concentration of neutral atoms in the plasma
of ST-1 Tokamak in two regimes differing in the electron concentra-
tions. 156
The direct measurement of the concentration of neutral hydrogen at-
oms from the fluorescence on the resonance line is very interesting.
However, the resonance line of hydrogen L a (l = 1216 Å), unlike the
resonance lines of the majority of gases, is situated in the vacuum ul-
traviolet range. As already noted, in movement into the ultraviolet region
of the spectrum the spectral density of the flux at which fluorescence
is saturated (r sat
l
~l –5 ) greatly increases. For example, if in examining
fluorescence on H a, r satl is several kW/(cm 2×nm), in the vicinity of L
represents several thousands of kW/(cm2×nm). In addition, tuneable lasers
are not available for this region of the spectrum.
To vary the wavelength of radiation in vacuum ultraviolet, it is nec-
essary to use the methods of non-linear optics, especially third harmonic
generation. The initial radiation is the radiation of ruby, neodymium,
excimer lasers and dye lasers and the non-linear medium is a mixture
of inert gases with metal vapours. For example, radiation at a wave-
length of La with a record power of hundreds of watts was produced.158,159
Regardless of the fact that this power can be considerably lower than
that required for fluorescent saturation, this radiation source was used
in ASDEX Tokamak for examining the interaction of plasma with the
chamber wall in ‘cleaning’ discharges. 158
125
Plasma Diagnostics
F ig .4.41 Distribution of the concentration of hydrogen atoms along the diameter

ig.4.41
of the plasma column in a FT-1 Tokamak produced by the fluorescence method on
the line H a . 156
With the existing technical possibilities, the minimum concentration

of neutral hydrogen, measurable by the examined method, is 10 9 ¸ 10 10
cm –3 . Thus, the sensitivity of measuring N 1 from the fluorescence L a
is approximately an order of magnitude lower than when measuring on
the basis of the Ha line with subsequent conversion in accordance with
the accepted model. 156,157
Another possibility of direct examination of the concentration of non-
excited atoms is to utilise the two-photon excitation. The fluorescence
signal in two-photon excitation is proportional to the square of inten-
sity, so that at higher flux densities two-photon excitation may prove
to be more promising than a single-photon one. In addition, the wave-
length of pumping in this case is twice the wavelength of the resonance
line. In particular, for the two-photon excitation of the fluorescence of
the line L a it is necessary to apply radiation with the wavelength l =
2434 Å which is far easier than in the region l = 1216 Å. The first
experimental results were obtained in Ref.160 and 161.
As shown by the results, the resonance fluorescence method is used
at present mainly for determining the local values of the concentration
of normal and excited atoms and ions in the plasma. In individual studies,
the temperatures of the atoms and ions were determined from the contours
of the fluorescence line. 143,162
The resonance fluorescence method can also be used to examine the
dynamics of processes leading to population and depletion of the lev-
els. Therefore, in addition to the previously mentioned study in Ref.157
in which the kinetics of population of the levels in the helium plasma
was investigated, it is also important to note the Ref.134, 164 concerned
with the analysis of the processes of population of the hydrogen lev-
els.
There is another possibility of the resonance fluorescence method -
examination of the diffusion processes in plasma. 165,167
126
Spectroscopy of Ground Electronic States of Molecules in Plasma
Chapter 5
SPECTROSCOPY OF GROUND ELECTRONIC STATES

OF MOLECULES IN PLASMA USING TUNABLE LASERS
The advantages of using tunable lasers in comparison with conventional
light sources in quantitative plasma spectroscopy are often caused by
the following parameters: high monochromatism, high intensity, coherence
and directionality of radiation. 1 If these properties are efficiently uti-
lised, it is possible to change qualitatively many conventional
spectroscopy methods. The possibilities of absorption spectroscopy, 2
interferometry,3 the spectroscopy of Rayleigh,4 Thomson,5,6 Raman7 light
scattering have been greatly expanded. These possibilities are used
efficiently in spectral and optical plasma diagnostics. Methods based
on recording changes of the plasma properties in resonance absorption
of laser radiation by plasma particles are being developed. 8,10 In this
book, we shall discuss two methods of spectroscopy of molecular plasma
using the frequency-tunable lasers: diode spectroscopy and CALS (co-
herent anti-Stokes spectroscopy of Raman light scattering).
These methods are attractive because of the following reason. The
molecular plasma conditions are characterised by a low (Ù10–6) degree
of ionisation and the fraction of electronically excited emitting parti-
cles (Ù 10 –5 –10 –7 ). Therefore, in the majority of cases special inter-
est is attracted by the processes in which vibrationally-rotationally-excited
molecules in ground electronic states take part. These methods make
it possible to determine directly the behaviour of the molecules under
these conditions. The second reason is that these methods often suit-
ably supplement each other.
5.1 Diode spectroscopy

Laser diodes
The name of this method was derived from the fact that the light source
in adsorption measurements is represented by laser diodes made of
semiconductor materials generating radiation in passage of current
through them in the region of the (p–n)-junction. The characteristic di-
mension of the diode in each measurement are less than 1 mm thick
127
Plasma Diagnostics
Current
µm
0 Polished
25 surface
Cleavage
300 µm p- surface
ty
pe
n-
ty
pe
50
0
µm Laser
beam
F ig .5.1 Laser diodes.

ig.5.1
(Fig.5.1). The cleaning surfaces are used as the reflecting elements of

the resonator.
The generation frequency can be tuned by many methods. The most
useful methods are the variation of the pumping current and the specimen
temperature. The typical region of frequency tuning for the diode of the
given composition is ~100–300 cm–1. Pulsed and continuous generation
regimes are possible. In the pulsed regime, frequency tuning forms in
non-stationary heating of the (p–n)-junction by a current pulse. The
tuning characteristics are piecewise-continuous characteristics with zones
of smooth tuning Of several centimetres to the minus first degree. This
is associated with the mechanisms of formation of laser modes. In the
continuous regime, the width of these bands is usually 1–2 cm–1, in the
pulsed regime to 10 cm –1 and more. 11 The spectrum range is selected
by selecting the diode materials and can be varied from 0.7 to 50 µm.12
Figure 5.2 shows schematically the data on the types of laser used and
the characteristic frequencies in the spectra of several relatively sim-
ple molecules.
Strain vibrations
Valency vibrations
Obertons, combined vibrations
Electron transitions Rotational transitions
µm
F ig .5.2 Characteristic types of laser diodes used in various areas of the spectrum
ig.5.2
and the frequency in spectra of certain molecules.
128
L2 L1 L
L L
F ig .5.3 Circuit of the pulsed–periodic diode spectrometer.

ig.5.3
Spectrometer
The diagram of a diode laser spectrometer 11 operating in the pulsed–
periodic regime, is shown in Fig.5.3. The radiation of the diode laser
1, shaped by the lenses L1 and L2 into the beam with a transverse
dimension of ~1 mm, takes place through the examined object 2 (dis-
charge tube with a high voltage current source 16). The monochromator
3 is used for rough determination of the frequency of generation and
selection of the laser modes. The relative frequency calibration is speci-
fied by the Fabry-Perot interferometer 4. The signal from the photodiode
5 passes through the amplifier 6 to the oscilloscope 7 and the
stroboscopic integrator 8 followed by recording on the automatic re-
cording device 9. PbSe diodes, used in Ref.11, with the generation in
the range 4–6 µm, are excited by rectangular current pulses with a fre-
quency of ~1 A, duration 200 µs and a repetition frequency of 10–100
Hz. The generation power is 10–100 µW. The pulses are shared by the
generator 10 and magnified with the magnifier 11. The generation fre-
quency is tuned by passing a current pulse through the laser. The po-
sition of the tuning zone is varied by varying the temperature (in the
range 30–80 K) of the cooling line on which the laser is secured. The
sensitivity of the generation frequency to the variations of temperature
for the laser diode is ~1 cm –1 K –1. Therefore, to obtain a high spec-
tral resolution, it is sufficient carry out accurate regulation of tempera-
ture. The long-term stability of temperature is not lower than 10–2 K
and is achieved using the electronic system 12 as a feedback. The mo-
ment of the start of recording the spectrum with a stroboscopic inte-
129
Plasma Diagnostics
grator is set by the shaper of the time delay 13. The temperature fluc-
tuations in the period between two consecutive pulses of the pumping
current are compensated by the optical triggering system 14 with the
delay shaper 15.
This system makes it possible to ensure a resolution power of
~10 –4 cm –1, the width of the zone of continuous tuning to 11 cm –1, fre-
quency tuning rate 10 4–10 6 cm –1 ×s –1, operating speed of the recording
system ~10 –8 s. At these powers and widths of the radiation spectrum
the effective brightness temperature of the emitter can be estimated at
~105 K which enables quantitative absorption measurements to be carried
out even at very high excitation levels of the examined system. At the
same time, the relatively low power of the semiconductor lasers ena-
bles measurements to be taken in the linear absorption regime.
Special features associated with higher resolution under the conditions

of piecewise-continuous tuning characteristics
The presence of breaks in the frequency-tuned characteristic of the laser
creates certain difficulties in investigations in practice. However, the
latter is partially compensated by a higher spectral resolution. This shall
be explained on the following example of examination of distribution
of CO 2 molecules on the vibrational–rotational levels in the active
medium of a gas discharge CO 2 laser. 11,13,14
The absorption bands of CO 2 corresponding to its four normal vi-
brations n 1, n 2 (twice degenerate), n 3, are situated in the spectrum re-
gions 8, 15 and 4 µm. In conventional approaches utilising the meth-
ods of spectroscopy with moderate resolution, to examine the overall
pattern of distribution of vibrational energy investigation must be carried
out especially in these regions, and n 2 and n 3 are reflected in the ab-
solute spectra and the symmetric vibration n 1 in the Raman scattering
spectrum. The application of the high resolution technique, restricted
by the width of the line like, for example, in diode spectrometry, per-
mits a different approach. It is sufficient to work in the region of one
of the frequencies, for example, n3, and use the superimposed rotational
structure of sequential transitions v 1v 2¢v 3 ® v 1v¢ 2(v 3 + 1), where v and
l are the quantum numbers in the generally accepted notations.
Figure 5.4 shows a fragment of the absorption spectrum of CO 2 in
a band approximately 1 cm –1 wide in the region 4.3 µm in a discharge
in the CO 2–N 2–He mixture and in the absence of discharge. In the dis-
charge the spectrum is enriched. This corresponds to absorption from
excited vibrational levels. For the example discussed here, it is important
to note that certain vibrational states of the molecules are found already
in such a narrow (in the conventional sense of the word) spectral range,
including the states belonging to isotopically substituted molecules. In
130
F ig .5.4 A fragment of absorption spectra of CO 2: a) in the CO 2 –H 2–He mixture (1:1:8)

ig.5.4
in the absence of discharge, pressure 36 torr; b) in the discharge, the mixture and
the pressure are the same, current density 1.5 mA/cm 2 ; c) transmittance of the Fabry-
Perot resonator.
identifying the transition shown in Fig.5.4, we note a branch and also

rotational and vibrational states being lower states for the transitions.
The asterix indicate the transitions in the C 12 O 182. Other transitions are
in the normal isotope C 12 O 162.
Using the intensities of absorption and the literature data on the values
of the matrix elements of the dipole moment for the vibrational–
rotational transitions, it is possible to determine the distributions of
populations of the rotational and vibrational levels.11,14 Figure 5.5 shows
the distribution of CO 2 molecules on rotational levels for several vi-
brational levels, and Fig.5.6 is the distribution on the vibrational lev-
els. It may be seen that the rotational distributions are of the Boltz-
mann type, and since they exist in long-life ground electronic states,
the rotational temperature corresponds to the kinetic temperature of the
neutral particle. 15 The vibrational distributions also demonstrate on the
whole the Boltzmann-type distributions within the limits of each vibra-
tional mode. Only the block with anti-symmetric vibrations n 3 shows
a weak tendency for Trinor-type deviations.16 The vibrational temperature
of the anti-symmetric mode T 3 = 2650 ± 30 K, and of the symmetric
and strain modes T 1 = T 2 = 560 ± 15 K and does not coincide with the
gas kinetic temperature T g = 470 ± 10 K.
We shall discuss here physical processes that determine the forma-
tion of corresponding distributions because we are interested in this case
in the procedural aspect of the matter. It is only important to note that,
firstly, the relative error of temperature determination is sufficiently
131
Plasma Diagnostics
F ig .5.5 Distribution of CO 2 molecules on the basis of the rotational levels for several
ig.5.5
vibrational levels. Discharge in a capilliary 2 mm in diameter, 50 mm long, current
i = 9 mA, gas mixture CO 2 –N 2 –He (1:1:8:), pressure 30 torr.
F ig .5.6 Distribution of CO 2 molecules on vibrational levels. Symbols are the same

ig.5.6
as in Fig.5.5.
small, being ∆T/TÙ 2%. This is due to the fact that a large number
(~1000) of vibrational–rotational transitions was used in the measure-
ments. Secondly, the information on the vibrational distributions in
Fig.5.6 was obtained in the energy range of vibrational states ~15000
cm –1 , but within the limits of a very narrow spectral range ~75 cm –1
since the differences in the frequencies of vibrational transitions are as-
sociated only with anharmonism. Naturally, this approach can be used
only when using high spectral resolution. This is also possible in di-
ode spectroscopy.
132
Recording low-concentration particles

Due to high spectral resolution, diode spectroscopy has high sensitiv-
ity. When used for plasma diagnostics, this makes it possible to examine
excited states and measure low particle concentrations.
In the simplest case of direct linear absorption, the reliability of
measurements on the level of recording the changes of intensity of laser
radiation DI/I ~ 10 –2 is sufficiently reliable. For the example of exami-
nation of the energy distribution of CO 2 molecules, described in the
previous section, this corresponds to a molecule concentration of ~10 8
cm –3 on one vibration-rotational level and an optical path of 50 mm.
The use of the methods of two-beam spectroscopy enables measurements
to be taken at DI/I~10 –5. Approximately the same possibilities are of-
fered by the use of autocorrelation methods in recording complicated
spectra. 17 Modulation methods can be used to increase sensitivity: the
frequency modulation of laser radiation, Stark and Zeeman modulation.
In recording the ions in plasma in discharges excited by an alter-
nating field or DC discharges with a specially applied alternating field,
it is also possible to carry out the modulation of the speed of motion
leading to the modulation of the Doppler shift, combined the method
of phase-sensitive detection of radiation. 18 Figure 5.7 shows a fragment
of the modulation spectrum of absorption of the three-atom negative ion
NCS – in ac discharge (25 kHz) in a gas mixture of 1.7 torr NH 3 +0.5
torr CS2. Figure 5.8 show the radial distribution of density of ArH + ions
in DC discharge in an H 2 + Ar mixture. 20
The radicals, ions and electronically-excited particles in the plasma
recorded up to the year 1988 by diode spectroscopy, are given below.
In many cases, investigations were not restricted to concentration meas-
F ig .5.7 Fragment of the absorption spectrum of NCS - . 119 a) band n 1 , b) n 1 + n 2 –

ig.5.7
n 2 , c) n 1 + n 3 – n 3.
133
Plasma Diagnostics
F ig .5.8 Radial profile of ArH + ions in a discharge tube. 20

ig.5.8
urements but also included determination of the distribution of parti-

cles on the vibrational–rotational states, analysis of the shape of the
contours of spectral lines and obtaining information on molecular con-
stants:
Radicals, unstable particles

AsH, BF, CCl, CF, CN, CS, FO, NCl, NF, NS, OH, OD, PH, PO, PCl, PN, PS, PF,
SF, SCl, SiH, SO, SiN, BO 2 , C 2 D, CD 2 , CH 2 , CF 2, C 2 O, C 2H, FCO, FO 2 , HCO, HO 2,
DO 2 , NF 2 , CD 3, CF 3 , CH 3 , CH 2 F, NO 3 , ClBO, FBO, HBO, S 2 O, HBNH
Positive ions
Ar + , ArH + , CF + , ClC + , HCl + , NO + , NeH +, OH + , CO 2 + , ClH 2 + , D 3 +, DCO + , DN 2 + , H 3 ,
HBF + , H 2 D +, HD 2 +, HCO + , HCS + , HN 2 +, H 2 O +, HCNH + , H 3 O + , SH 3+ , HeH + , SH +
Negative ions
C 2 – , OH – , SH – , FHF – , ClHCl – , FDF – , N 3 – , NCO – , NCS –
Electronically-excited particles
I, Kr, D 2 , N 2
Non-stationary processes
It has already been noted that the pulsed-periodic regime of operation
of the laser diode is characterised by a very high frequency tuning rate.
It is thus possible to record the spectra of non-stationary objects. For
example, Fig.5.9 shows the evolution of amplification in a pulsed pe-
riodic CO2 laser at atmospheric pressure on the P(20) line of the 00°1–
10°0 generation transition. In contrast to the conventional methods of
measuring the amplification factor in the centre of the line, it is pos-
sible to examine the change of the entire profile of amplification with
high (less than 1 µs) time resolution. 18a
134
F ig .5.9 Evolution of amplification in a CO 2 laser.

ig.5.9
It should be noted that as a result of obtaining extremely high rates

of frequency scanning in diode spectroscopy combined with the high-
speed radiation recording system, the rate of recording the absorption
spectra can be limited no longer by technical but purely physical fac-
tors. If the width of the absorption line is Dw, and the duration of re-
cording the contour of the line Dt, it is evident that if the indetermi-
nacy relation DwDtÚ 1 is not fulfilled, the contour is recorded in the
distorted form. In fact, the coherent radiation passing through it po-
larises this contour. The phases of relaxation processes during periods
longer than t rel ~ Dw –1 leads to dephasing of dipoles and induced po-
larisation disappears. In scanning the contour during the periods shorter
than or comparable with trel, after the effect of laser radiation (i.e. when
w(t) > w ik + Dw/2, where w(t) is the tuned frequency, w ik is the fre-
quency corresponding to the centre of the contour) the medium retains
the molecules with the induced dipole moment. The molecular dipoles
continue to vibrate at frequency w ik, they are phased, and this leads to
emission of coherent radiation by the medium in the direction of inci-
dent light. This radiation is combined with probing radiation, and de-
pending on the ratio of the phases of these fields the resultant signal
may both increase or decrease.
This situation was demonstrated by experiments in Ref.21 in recording
the contour of the line from the vibrational–rotational spectrum of CO2
(l » 4.2 µm) with a diode spectrometer with a frequency scanning speed
of up to 10 6 cm –1 ×s –1 .
Figure 5.10 shows the oscillograms of the radiation passed through
a gas. These oscillograms were recorded at tuning rates of µ = 105; 106
cm –1×s –1 and the width of the line Dw = 4.4·10 –3 cm –1 . Case a corre-
sponds to ‘slow’ recording Dw 2/µ = 36.6, and the oscillogram reflects
135
Plasma Diagnostics
100 ns
10 ns b
F ig .5.10 Oscillograms of intensity in recording the contour of the absorption line

ig.5.10
of CO 2 at scanning speeds of m = 10 5 (a), 10 6 cm –1 ×s –1 (b).
the actual line contour. In the case b Dw 2/µ = 36.6 and the effects de-
scribed previously are observed.
Restrictions in the application of diode spectroscopy are associated
mainly with two circumstances: 1) absence in a number of cases of
adsorption vibration IR spectra (in dipole approximation), for exam-
ple homonuclear diatomic molecules; 2) no localised measurements in
the direction of propagation along the laser beam.
CALS spectroscopy has no such restriction.
5.2 CALS spectroscopy

Introduction
Coherent anti-Stokes spectroscopy of Raman light scattering (CALS)
is a non-linear analogue of the classic Raman scattering method. 22 The
effect of two power coherent sources with frequencies w 1 and w 2 on the
radiation medium generates new coherent radiation at frequency w3. From
the microscopic viewpoint, CALS is a four-photon process in which the
molecule absorbs two photons with frequency w 1 and emits two pho-
tons with frequencies w 2 and w 3 = 2w 1 – w 2, and the quantum state of
the molecule does not change (Fig.5.11). The radiation at frequency w 2
is forced in the field of incident radiation with frequency w 2. The ra-
diation with w 3 is coherent and characterised by the wave vector
H H H
k 3 = 2 k1 - k 2 . In principle, this process takes place in any medium, but
the radiation intensity at frequency w 3 rapidly increases if the differ-
ence w 1 – w 2 is close to the frequency of the Raman-active transition.
Varying the difference w1 – w2 we obtain a spectrum containing the same
information as the spectrum of spontaneous Raman scattering (SRS).
The two main advantages of CALS in comparison with SRS are the rapid
increase of the signal amplitude in the directional light beam (this is
important when examining objects with a low density), and the possi-
136
F ig .5.11 Generation of CALS signal with a frequency of w 3 = 2w 1 - w 2 under the

ig.5.11
effect on the molecule of two laser beams with frequencies of w 1 and w 2 ( i , f are
the quantum states of the molecules).
bility of localising the region of generation of the CALS signal, due

to the non-linear nature of non-interaction of the light with the medium.
A similar process also takes place in the Stokes region where w¢ =
2w 2 – w 1 although the anti-Stokes region is more suitable for practice
because of the absence of undesirable radiations in it: luminescence of
optics, scattered laser radiation, etc.
To determine the intensity of CALS spectra and their relationship
with the densities of the quantum states of the molecules, it is impor-
tant to use the conclusions of quantitative theory, similar to that de-
scribed in a number of reviews and monographs (see, for example,
Ref.22).
H Under the effect of the field of the light wave the medium is polarised
P . This polarisation can be described by a series with respect to the
exponents of the total strength E of the electric field
P = c a1f E + c a 2 f E 2 + c a3f E 3 +...,

H H H H
(5.1)
where c (i) is the dielectric susceptibility tensor. The first term with c (1)
describes refraction and light reflection phenomena. Susceptibility c (1)
is linked with permittivity by the relationship e = 1 + 4pc (1). The term
with c (2) describes the phenomena of doubling the frequency and op-
tical detection. For isotropic media, such as gases relevant to this work,
this and all subsequent even terms in (5.1) are equal to zero because
the tensor c is symmetric. Thus, the term with c (3) is a term of a lower
order describing the non-linear effect in isotropic media, including all
coherent defects of Raman scattering, including CALS.
Calculations within the framework of conventional electromagnetic
theory which take into account the symmetry properties c (3) for a me-
137
Plasma Diagnostics
dium with magnetic permittivity µ = 1, give the following expression

for the intensity of the CALS signal:
25p 4 w 32
c a 3f - w 3 , w 1 , w 1 , - w 2
b g 2 F Dkl I .
I3 =
b g b g d ic
n w3 n w2 n w12 4
I12 I2 l 2 sin c
H 2 K (5.2)
Here I j is the intensity of the wave with frequency w j; l is the length
of interaction; k = k 3 – 2k 1 + k 2 ; c is the speed of light; n is the
refraction index. In accurate phase matching (Dk = 0) I 3 ~ l 2 . At
k ¹ 0, I3 is the periodic function of l which reaches the first maximum
at a length lc = p/Dk referred to as the coherence length. Generally speak-
ing, the condition w 3 = 2w 1 – w 2 does not ensure that Dk = 0 is ful-
filled, but in gases due to low dispersion (weak dependence n (w j)) in
the case of colinear propagation of the beams the phase synchronism
is fulfilled over a large length. For the majority of gases l c ~ 1 m, and
in the majority of real systems l < l c so that it can be assumed that
I 3 ~ l 2.
Examination of the susceptibility tensor c (3) on the microscopic level
and determination of the relationship with the density of particles and
molecular constants shows that its component, corresponding to reso-
nance with the Raman-active transition, can be expressed as follows:
c ap3f =
dN i
- Ni n1c 4 F ds I dw i -1
H sW K
f
if - w1 + w 2 - iGif . (5.3)
2 Dn2 w 24 if
Here N i and N f are the densities of the molecules in the states i and
f; s is the cross section of spontaneous Raman scattering; W is the solid
angle; G if is the half-width at half height of the line of spontaneous
ds
Raman scattering. Cross sections SRS s and are well known for
dW
a large number of molecules. 23
Thus, in a relatively narrow spectral range
d
I3 ~ N f - Ni I12 I2 . i2
The direct relationship of c (3) with SRS cross sections shows that
the rules of selection of CALS are identical with those for SRS. In par-
ticular, for diatomic molecules they have the form
138
R| 2 is S - branch
Dv = 1; D J = S 0 is Q - branch
|T-2 is O - branch (5.4)
Dv = 0; D J = ±2 is S -, O - branch,
where v, J are the vibrational and rotational quantum numbers. At the

same time, since CALS is a parametric process in which the amplitude
of the scattered wave represents the sum of contributions from all
molecules in the interaction region, in calculating I 3 it is necessary to
carry out averaging with respect to molecular orientation. Consequently,
differences in the matrix elements of the polarisability tensor form in
the microscopic sense in the SRS and CALS cases. However, this does
not affect the selection rules (5.4), 24 and the matrix elements of
polarisability a if will have the following dependence on v and J (in
models of a harmonic oscillator and a rigid rotator):
2 D LM 4 OPa f
Q - branch: v, J a if v + 1, J =
2 Mw v N 45 Q
a ¢ 2 + bJ , J g ¢ 2 v + 1 ,,
2 D 1LM O (5.5)
g ¢ Pa v + 1f.
N Q
2
O-, S - branches: v, J a if v + 1, J ± 2 = bJ , J ± 2
2 Mw v 15
Here M is the reduced molecular mass, w v is the frequency of vibra-

tions, b is the Plachek–Teller coefficient
a f
J J +1 3 J a J - 1f
bJ , J =
a fa f
2J - 1 2J + 3
, b =
2a2 J - 1fa2 J + 1f
J , J -2 ,
3a J + 1fa J + 2 f F c IJ F ds I ,
, a =G
4
bJ , J + 2 =
2a2 J + 1fa2 J + 3f H w K H dW K
if
2 if
a¢ and g¢ describe the mean polarisability and its anisotropy.

For the majority of molecules, the anisotropy is relatively small. 25
To record the Q-branch with high intensity (in comparison with O- and
S-branches) in the CALS spectrum, the correction for the intensities of
the vibrational–rotational lines is small (this correction is the largest
in the region of small J).
For the H 2 and N 2 molecules it is given below, %:
139
Plasma Diagnostics
J 0 1 2 3
H2 –8 5 1 0.3
N2 –6.5 4 0.8 0.3
This holds if the resonance component of the tensor c (3) is consid-

erably higher than the non-resonant one. In practice, this can be ful-
filled in cases in which the examined molecules represent a small fraction
of the total number of particles in the gas. In these cases, the inter-
pretation of intensities in CALS spectra is greatly complicated and, most
cases, steps are taken to exclude the non-resonance component in for-
mulating an experiment. In particular, the authors of Ref.26 proposed
an efficient method of suppressing it based on the fact that when the
polarisation rates of pumping rates do not coincide with the polarisations
of the resonant and non-resonant signals of the CALS do not coincide.
Spectrometer. Figure 5.12 shows the typical circuit of a CALS
spectrometer. 27,28 Since the particle concentration in the gas and plasma
is usually relatively low, to use non-linear-optical methods it is nec-
essary to use radiation of a relatively high power. The master oscil-
lator in the described circuit is the pulsed-periodic aluminium–garnet
laser 1 (Nd-YAG). Radiation of with a wavelength of l = 1.06 µm, a
repetition frequency of 20 Hz and a pulsed time of ~10 ns is converted
to the radiation with a doubled frequency (second harmonics, l = 532
nm) in DKDP crystal 2 (the efficiency of energy conversion is »30%)
which is used as a beam with reference frequency w1. The radiation pulse
energy in the second harmonics is ~30–50 mJ, the spectral line with
~0.1–0.2 cm –1 . Residual radiation with l = 1.06 µm, not converted in
the doubler 2, is doubled with respect to frequency in the second DKDP
L1 L2
M1 DM 3
DM 2
DM 1
M2
F ig .5.12 Diagram of CALS spectrometer.

ig.5.12
140
crystal 3 (~20 mJ with l = 532 nm) and is used to pump the dye la-
ser 4. To separate the radiation beams of the first and second harmonics
of the (Nd-YAG) laser, the authors used the DM 1 and DM 2 dichroic
mirrors. The width of the dye laser radiation line was 0.1-0.2 cm –1, the
pulsed energy 1–5 mJ. The frequency of the laser 4 is tuned by the mi-
croprocessor 5. Telescope T is used for optimum matching the diam-
eters d 1 and d 2 of the beams w 1 and w 2. After passage through the tel-
escope, radiation w 2 is co-linearly combined with the beam w 1 using
the DM 3 mirror. Both beams are focused in the examined object 6 by
the power unit 7 using the lens L 1. The CALS signal is separated from
laser radiation using the symmetric four-prism monochromator 8,29 which
enables operation in a relatively wide spectrum range at the constant
position of the prisms. The radiation detector was the photodetector 9
with the power unit 10. The signal from the photoelectric multiplier trav-
els through the cable delay line 11 to the stroboscopic integrator 12 op-
erating in the synchronous detector regime. The integrator is activated
from the avalanche photodiode 13 illuminated with laser radiation 1.
The delay line is selected in such a manner as to ensure the transfer
of the signal from the photoelectric multiplier in the working range of
the time delays of the integrator. The beam passage path ω 1 and ω 2 are
selected such as to ensure the simultaneous arrival of the light pulses
with the given frequencies to the examined object. This is controlled
by the avalanche photodiode 14 whose signal travels to the oscilloscope
15 through the delay line 16.
Typical parameters of this spectrometer are: spectral resolution 0.1-
0.2 cm –1 ; the range of combination vibrational frequency 1000−4200
cm –1 , sensitivity with respect to concentration N f – N i~10 10–10 14 cm –3
for H2, N2, CO molecules…; localisation of the region of measurements
across the beams ~10–100 µm, along the beams 1–20 mm.
This spectrometer was constructed on the basis of the so-called
‘narrow-band’ ‘co-linear’ systems. Other variants can also be used, de-
pending on the requirements imposed by the specific features of the ex-
amined object. For example, if necessary, the localisation of the meas-
urements along the probing beams is improved using a non-co-linear
system in which the conditions of phase synchronism of the beams are
satisfied at the large angles from their convergence selected by specific
procedure.30 Additional advantages of this system are due to simple re-
cording of CALS spectra with small combination shifts, for example,
purely rotational spectra.
The ‘narrow-band’ system in which both lasers operate with a high
monochromaticity is preferred when examining objects operating under
stationary or pulsed-periodic conditions. In the case of objects with a
monopulse regime, it is recommended to use the ‘wide-band’ variant
141
Plasma Diagnostics
of the spectrometer. 31 in which a dye laser generates a wide spectrum

and the CALS spectrum is recorded in a single laser pulse under the
condition of photography, microchannel or matrix (instead of photoelectric
multiplier) detection. However, as shown by practice, the sensitivity in
this case is approximately an order of magnitude lower than in the ‘nar-
row-band’ system.
In recent years, the method of using CALS spectrometers has been
successfully developed; we shall mention the use of lasers with the
picosecond pulse duration, 32 systems with increased spectral resolution
power which include, as the master laser, a high-stability gas laser with
subsequent light amplification in the (Nd-YAG) medium,33 and a number
of other interesting proposals.
Examples of using CALS for investigations of vibrational–rotational

distribution of molecules in gas discharge plasma
The majority of applications of CALS in low-temperature plasma di-
agnostics are associated with examination of vibrational–rotational dis-
tributions of simple diatomic molecules.
The authors of Ref.34 examined the distribution of nitrogen molecules
on vibrational–rotational levels of the ground X1S state in a low-pressure
discharge (p = 2¸4 torr, i = 80 mA). The sensitivity of the method
enabled molecules to be recorded on vibrational levels with v £ 14. The
measured rotational temperatures in the vibrational states v = 0 ¸ 10
at p = 2 torr were within the error range equal to T rot = 530 ± 30 K.
The vibrational distribution was of the non-Boltzmann type with the
vibrational temperature describing the relative population of lower levels
with v = 0 and v = 1 equal to T 10 = 5300 ± 350 K. However, the
vibrational distribution at a pressure of p = 2 torr was not described
within the framework of the well-known Trinor model (Fig.5.13). This
F ig .5.13 Populations of the vibrational levels of N 2 in the discharge. Pressure p =

ig.5.13
2 torr, discharge current i = 80 mA; 34 solid line - experiments, broken line are the
calculated data.
142
F ig .5.14 Distribution of the concentration of N 2 molecules on the vibrational states

ig.5.14
in a pulsed (duration 40 ns) discharge (current 1.4 A) after a pulse with different
time delay. t = 50 (1), 100 (2), 6 (3) and 20 ms (4).
is attributed to the effect of vibrational relaxation on the walls. An

additional argument, according to the authors, was that at high pres-
sure (p = 4 torr) the agreement between theory and experiment improved.
The authors of Ref.35 examined the distribution of N 2 molecules in
the vibrational–rotational states in the condition of a pulsed (t = 40
ns) high-current (i = 1.4 A) discharge. The discharge zone was probed
both during the current pulse and after its completion over a period of
20 µm. Results are shown in Fig.5.14. The non-Boltzmann type of
distribution is again found. Comparison with the theory of vibrational
relaxation enabled the authors to determine the constants of the (V–V)
exchange. Similar measurements were taken in Ref.36 at higher energy
inputs (t = 200 ns, i = 260 A), at delay times after the start of the
current pulses of 400 ns–1 ms. In this case, to explain the experimental
results, the authors had to assume that in addition to the direct elec-
tron impact and (V–V) processes, there are additional channels of for-
mation of vibration distributions, especially the population of the vi-
brational levels of the ground electronic state in the processes of quench-
ing of metastable electronic levels.
Detailed investigations of the distribution of hydrogen molecules on
the rotational levels in a gas discharge were carried out in Refs.27 and
28. The higher value of the rotational quantum of the H 2 molecule and
the presence of modifications with different nuclear spin caused that
at high gas temperatures the rotational distribution was non-Boltzmann,
even under the stationary conditions. Figure 5.15 shows an example of
such a non-equilibrium distribution in H 2 (X 1S, v = 0) under the con-
ditions of a gas discharge in a H2–He (1:3) mixture at a pressure of
0.5 torr and a current of 30 mA in a discharge tube 14 mm in diam-
143
Plasma Diagnostics
Fig .5.15 Distribution of the concentration of H 2 molecules ( X1S; v = 0) on the rotational

ig.5.15
levels at the discharge axis. The straight line is the Boltzmann distribution at a
gas temperature of T g = 145 K; 1) vapour - H 2 ; 2) ortho-H 2 .
eter with the walls cooled with liquid nitrogen. The population of the
rotational levels is related to the unit statistical weight taking into account
the nuclear statistical weight so that we can examine a single dependence
of the number k of the levels for ortho- and paramodifications of the
molecules. The slope of the straight line corresponds to the gas tem-
perature measured by independent methods (thermocouple, the rotational
structure of the spectrum of small nitrogen impurities, the width of
Doppler contours) which in this case is T g = 145 K at the discharge
axis. To provide further information, the upper part of the figure shows,
on the linear scale, the ratio of the measured population N k of the level
~
to the calculated population N k , corresponding to the Boltzmann dis-
tribution at the rotational temperature equal to the gas temperature. The
quantitative interpretation of the deviations was provided by the authors
of Ref.28 on the basis of examining the balance of the excitation rates
of the rotational levels by the electronic impact and rotational relaxation.
Some other possibilities of CALS spectroscopy as a local measurement
method are indicated by Figs.5.16 and 5.17 which show the spatial dis-
tributions of the H2 molecules (X1S, v = 0) on individual rotational levels
under the conditions of a discharge in hydrogen with standing layers
and cooled with liquid nitrogen.
Figure 5.16 shows the radial distributions in the relative units of the
concentration of H 2 molecules (X 1S, v = 0, k = 2; 3) in a discharge.
For comparison, the graph also shows the graph of the Bessel function
describing the radial distribution of a concentration under the diffusion
discharge regime.
Figure 5.17 shows the distribution of the H 2 molecules (X 1S, v = 0,
k = 2; 3) along the stationary layer of the discharge under the experiment
144
N k , rel. units
F ig .5.16 Radial distribution of the concentration of H 2 molecule ( X 1 S; v = 0; k = 2

ig.5.16
(1), 3 (2)) in a H 2 discharge in a tube cooled with liquid nitrogen. Pressure P = 0.5
F I
torr, discharge current i = 40 mA, solid curves is the Bessel function J0 2.4
GH JK
L3 .
R
N k , rel. units
F ig .5.17 Distribution of the concentration of H 2 molecule ( X 1S; v = 0; k = 2 (1), 3

ig.5.17
(2)) in a discharge. Calculations were carried out disregarding (I) and taking diffusion
into account (II).
conditions corresponding to Fig.5.16. The directions of propagation of

the laser beams coincide with the axis of the discharge tube (Fig.5.12).
The spatial resolution along the beams is 1 mm, the layer length
L = 1 cm. As a result of a large change of the parameters of the elec-
tronic component along the layer one could also expect changes in the
concentration of the molecule in different rotational states. However,
measurements show that no such changes take place. The result is ex-
145
Plasma Diagnostics
plained by the authors of Ref.28 by the fact that during the rotational
relaxation time which is relatively short for the hydrogen molecules the
axial profiles are smoothed out by diffusion. The results of the corre-
sponding calculations are shown by solid curves in Fig.5.17.
Concluding this section, it may be noted that the described methods
of diode and CALS spectroscopy enable us to transfer to a relatively
new level of investigation of the processes in a non-equilibrium sys-
tems in which vibrationally–rotationally excited molecules take part in
comparison with conventional methods of classic spectroscopy. This,
however, does not mean that the classic spectral methods of diagnos-
tics have lost their importance. On the contrary, the combination of
various approaches opens new wider possibilities of detailed investi-
gations of these systems.
146
Determination of the Concentration and Temperature of Heavy Particles
Chapter 6
DETERMINATION OF THE CONCENTRATION AND

TEMPERATURE OF HEAVY PARTICLES FROM THE
SPECTRA OF RAYLEIGH-SCATTERED LIGHT
6.1 Introduction
The method of diagnostics of low-temperature plasma, based on exam-
ining the spectra of Rayleigh-scattered radiation probing the plasma,
has been attracting special attention recently. This is associated not only
with the fact that the method is relatively simple but mainly with the
possibility of direct measurement of the concentration of heavy parti-
cles without using complicated and expensive methods of measuring light
absorption in the vacuum ultraviolet range or the methods of multiphoton
absorption of laser radiation which are still difficult to apply. In a number
of cases, the Rayleigh scattering method can be used to measure in-
dependently also the translational temperature of the particles. This is
of special interest in investigations of non-equilibrium plasma.
As is well known, Rayleigh scattering occurs when the size of the
scatterers is considerably smaller than the wavelength of probing ra-
diation. Light scattering by gas molecules can be regarded as a con-
sequence of fluctuations of the concentration of molecules causing fluc-
tuations of permittivity. The latter is directly included in the equation
that determines the intensity of Rayleigh-scattered light.1 The spectrum
of this radiation (the structure of the Rayleigh line) depends on the nature
of damping with time of concentration fluctuations which in turn is
associated with the nature of the physical processes determining the
propagation of concentration fluctuations in the gas. 2
6.2 Principles of the method

Depending on whether the Rayleigh scattering is a microscopic or mac-
roscopic process, there are two limiting scattering conditions: kinetic
and hydrodynamic whose boundaries are determined by the value of the
scattering factor
147
Plasma Diagnostics
a = 1/Kl, (6.1)
where l is the mean H free

H path
H of gas molecules,
H H K is the modulus of
the wave vectors K = Ks - K0 , where K0 , Ks are the wave vectors of
the probing and scattered waves (K 0 = 2p/l 0; K s = 2p/ls, l0, l s are the
wave lengths of the probing and scattered light). Since Ks » K0, the value
K = 2K0·sin (q/2), where q is the angle between the vector of the probing
and scattered waves. Setting K = 2p/L, we obtain
L = l 0/2·sin (q/2) (6.2)
It may be concluded that the scattering in the examination direction

forming the angle θ with the direction of the probing beam is deter-
mined by concentration fluctuations with a wavelength of L/2p. For the
scattering parameter we obtain the following equation, taking into account
(6.2)
a = L/2pl = (l 0/l) (1/4p) sin (q/2) (6.3)
which shows that a can be treated as the reciprocal value of the Knudsen
criterion for the scattering process.
The inequality a >> 1 determines the boundary of the hydrodynamic
scattering regime in which the scattering spectrum consists of the central
line l 0 and two Brillouin components l B symmetric in relation to the
central line. These components are caused by thermal fluctuations of
the concentration propagating as a sound wave through the gas and are
therefore shifted with respect to frequency in accordance with the Doppler
effect. 3
At a ~ 1, the wavelength of fluctuations is close to the mean free
path of the molecules. In this case, the scattering regime is kinetic and
the scattering spectrum can be determined only by solving the system
of kinetic equations;4,5 The Brillouin components in the spectrum merge
almost completely with essential line.
The inequality a << 1 corresponds to the collisionless scattering re-
gime. Concentration fluctuations are non-correlated, the scattering acts
in every molecule of the probed volume become independent, and the
frequency of the light scattered by the individual molecules is shifted
in accordance with the Doppler effect by the value which depends on
the velocity of the molecule and the direction of its movement in re-
lation to the examination direction. In other words, under conditions
in which the molecule velocities have some distribution (for example,
Maxwell distribution), the line of probing radiation simply assumes
Doppler broadening in the scattering spectrum.
148
In calculating the scattering parameter for the specific plasma con-

ditions it is important to know the mean free path of the molecules which
can be evaluated using relationships obtained in Ref.6 assuming that
the model of collisions of solid spheres and the ideal gas flaw are valid:
F I
8m 2 kT 1/ 2
l=
H K
5 p pm
, (6.4)
where m, p, T is the viscosity, pressure and temperature of the gas, re-

spectively, k is Boltzmann’s constant, m is the mass of the gas mol-
ecule. The accuracy of estimating l from equation (6.4) is fully accept-
able for the majority of cases relevant for practice, whereas the data
from Ref.7 can be used for more accurate estimates.
For the large majority of plasma and plasma chemical systems, with
the exception of cases with very small scattering angles of very high
pressures, the parameter a < 1, i.e., the scattering regime is close to
collisionless. The scattering spectrum consists of the components de-
termined by the Doppler shift of the line of probing radiation as a result
of kinetic movement of the light-scattering molecules. If the distribu-
tion of the molecules of every component i of the plasma with respect
to velocities v is of the Maxwell type
af
fi v = Bi1/ 2 exp - Bi , b g (6.5)
where B i = m i/2pkT; m i is the mass of the molecule of component i;

T is temperature, then the scattering spectrum is
F L bn - n gc O I 2
af
Si n = Bi1/ 2 exp G - pB M
GH N 2n sin q / 2 PQ JJK ,
i
0
(6.6)
0
where n is the frequency of scattered light; n0 = c/l0, c is light velocity;

q is the scattering angle. Thus, measuring the width of the scattering
line, we determine the translational temperature of scattering molecules.
The concentration of scatterers can be determined in principle by
measuring the integral intensity I (n) of the scattering line on the ba-
sis of the equation
af å xi s R Si a nf,
z
I n = AI0 N i (6.7)
i =1
149
Plasma Diagnostics
where A is some constant, I 0 is the intensity of probing radiation, N

is the total gas concentration, x i is the molar fraction of the particles
of the i-th thought, s Ri is the Rayleigh scattering cross section, z is the
number of components. Equation (6.7) shows that for single component
plasma it is quite easy to measure the molecule concentration. Obvi-
ously, in the case of multicomponent plasma it is necessary to know
the plasma composition. In some cases, this difficulty can be bypassed
by taking relative measurements of the intensity of scattered light, i.e.
comparing the value measured in the plasma with that in the gas un-
der known conditions and the same composition as the plasma. Evidently,
in the presence of chemical reactions leading to a large change of the
initial composition of the plasma gas, correct measurements of concen-
tration are relatively difficult. Preliminary calculations of the compo-
sition of the reacting gas make it possible to bypass this problem in
a number of cases, at least, estimate the error of concentration deter-
mination. For example, according to the data of the authors of Ref.9
who use the calculation model, the error of determination of concen-
tration by the Rayleigh scattering method does not exceed 35% even
for gas mixtures of some flames of rather complicated composition.
To estimate the contributions of each plasma component to the scat-
tering signal, it is essential to know the Rayleigh scattering section sRi
of the molecules of this component; the section is usually computed from
the equation 9
FG
4 p 2 ni - 1 IJ
s Ri =
l4 H
NL
sin 2 q,
K (6.8)
Ga s H O C N Ar He Ne
(n i – 1 ) × 1 0 4 (2 7 3 K , 1 0 5 P a ) 0.72 1.36 1.78 1.50 2.82 0.34 0.67
σ Ri × 1 0 28, c m2 (θ = 9 0 °, λ = 4 8 8 nm) 0.47 1.78 3.05 2.16 76.7 0 . 11 0.41
Ga s Cl H2 O2 N2 OH CO NO
(n i – 1 ) × 1 0 4 (2 7 3 K , 1 0 5 P a ) 3.81 1.44 2.73 3.00 2.06 3.40 2.97
σ Ri × 1 0 28, c m2 (θ = 9 0 °, λ = 4 8 8 nm) 13.9 1.89 7.18 8.66 4.08 11 . 1 2 8.50
HC l Air H2 O HO 2 CO2 N 2O H2
Ga s (witho ut
C O 2)
(n i – 1 ) × 1 0 4 (2 7 3 K , 1 0 5 P a ) 4.47 2.93 2.55 3.43 4.50 5.08 6.40
σ Ri × 1 0 28, c m2 (θ = 9 0 °, λ = 4 8 8 nm) 19.2 8.16 6.26 11 . 3 19.5 24.8 39.4
Ga s O2 N H3 C 2 H2 C H4 C 2 H4 C 2 H6 C 3 H8
(n i – 1 ) × 1 0 4 (2 7 3 K , 1 0 5 P a ) 6.61 3.76 5.80 4.44 6.36 7.76 10.94
σ Ri × 1 0 28, c m2 (θ = 9 0 °, λ = 4 8 8 nm) 42.1 13.6 32.4 18.96 38.9 53.0 104.1
150
where ni is the refraction index of the component i; NL is the Lochsmidt

number. The data on the refractive indices and the Rayleigh scatter-
ing sections for some gases are given below:
6.3 Concentration and temperature of gas in flames and

electric arc
Possibilities, advantages and shortcomings of the Rayleigh scattering
method were demonstrated most extensively in studies of the well-known
investigators of turbulent flames L. Talbot and his colleagues. 6,9-13 The
physical–chemical conditions in turbulent flames are relatively similar
to those in plasma chemical objects (with the exception of evidently the
temperature level); therefore, the results of these studies (or at least,
the procedure) can be efficiently used in the diagnostics of plasma
objects.
Figure 6.1 shows the diagram of experimental equipment 6 for exam-
ining the Rayleigh scattering spectra of laser radiation in flames. The
source of probing radiation was an argon ion laser operating in the
single-mode regime with frequency stabilisation (using an intraresonator
reference) at a radiation power of approximately 1 W with at a wave-
length of 488 mm. The laser beam was focused in the volume of the
examined flame (constriction diameter around 0.03 mm). Diaphragms
and traps were used to reduce the intensity of parasitically scattered
light. The scattered light spectrum was produced using a piezoelectrically
tuned Fabry–Perot interferometer (free spectrum range 0.41 cm –1); the
tuning range and speed were set using a computer. Radiation intensity
at a hundred points uniformly distributed throughout the spectrum was
measured using a photon counter with a corresponding discriminator
at the input. The measured data were transferred to the same compu-
F ig .6.1 Diagram of equipment for examining Rayleigh scattering of laser radiation

ig.6.1
in a flame. 6 1) laser, 2) mirror with 5% reflection, 3) scattering lens, 4) black screen,
5) collecting lens, 6) optical trap, 7) hydrogen flame in air, 8) screen, 9) Fabry-
Perot interferometer, 10) photoelectric multiplier, 11) computer, 12) oscilloscope.
151
Plasma Diagnostics
ter where they were processed using an algorithm enabling determination

of the maximum intensity of the scattering line and the half-width of
the profile of this line. In addition to the scattering line, the position
and shape of the initial radiation line were inspected. This radiation was
directed on to the input of the interferometer using the screen controlled
by the same computer. An interference light filter was used to weaken
the natural emission of the flame and increase the signal:noise ratio.
For the flame conditions6 the scattering parameter did not exceed 0.1
so that the authors of Ref.6 could determine the gas temperature in the
flame (~2 kK) from the half-width of the scattering line. With the
measurements repeated ten times, the error of temperature measurement
did not exceed 3%. In addition to the errors determined by a low
signal:noise ratio, the total error was determined to a large extent by
the instability of the spatial position of the flame. For the conditions
outside the flame the scattering parameter was around 0.7 (kinetic scat-
tering regime) so that it was necessary to calculate the contributions
of Brillouin components to the scattering spectrum.
The intensity of scattered radiation was measured by numerical in-
tegration of the resultant spectra. Since the error of these measurements
proved to be 6 relatively high (~20%), the authors of Ref.6 took only
relative measurements of the density distribution in the flame. Evidently,
the use of a spectral device with a smaller resolution (a narrow-band
filter in the limit) made it possible to measure also the absolute val-
ues of the gas density. In fact, the majority of investigators took this
approach in subsequent investigations; for example, in Ref.9–13 it was
possible to measure the spatial distribution in the flame of not only the
absolute values of gas density but also of its turbulent pulsations.
Two-dimensional temperature and concentration distributions of the
components in turbulent flames were obtained in Ref.14 on the basis
of measurements of the intensity of Rayleigh and Raman scattering of
laser radiation (a dye pulsed laser with a radiation energy of 200 mJ
in a pulse with a duration of 1.8 µs at a wavelength of 440 nm). The
authors of Ref.15 measured the two-dimensional distribution of tem-
perature in the diffusion turbulent flame of a mixture of methane with
hydrogen where this distribution was obtained during a single pulse of
laser radiation (duration 10 ns) using a special optical system of illu-
mination of the flame by the laser radiation – the second harmonics of
a neodymium laser (the energy in the pulsed 200 mJ). To select the scat-
tered radiation they used an interference filter and the effective increase
of the signal:noise ratio was achieved by using a vidicon with pulse
control as a registering device. The spatial resolution of the optical
system was 0.12×0.12×0.3 mm 3, the error of temperature measurement
in the range 1800–2300 K was around 7%.
152
Without discussing the experiments concerned with Rayleigh scat-

tering in ‘cold’ gas flows (see, for example, Refs.16 and 17) in which
measurements were taken of both the molecule concentration and its
pulsations, attention will be given to the Ref.18 where the Rayleigh
scattering method was used to examine the plasma of a dc arc freely
burning in an argon atmosphere (purity 99.999%) at a current of 100
A (arc length 5 mm, argon flow rate 10 l/min). The plane-polarised ra-
diation of the argon ion laser (wavelength 514.5 nm), modulated with
respect to the intensity by a rotating perforated disc with a frequency
of 1 kHz, was focused in the arc plasma. The laser beam leaving the
arc chamber was absorbed in a trap. Scattered light was examined under
an angle of 90° to the direction of the probing beam in the solid an-
gle of 0.01 steradian. To separate the line of scattered radiation at the
background of arc plasma radiation, the authors used gradually the
Fabry–Perot resonator (free spectral range 0.25 nm) and a
monochromator (focusing distance 1 m). The spatial resolution was
0.05×0.005 mm2. To reduce the intensity of parasitically scattered light,
they diaphragmed the laser beam and the laser itself was placed in a
light impermeable jacket. The level of parasitically scattered light was
determined by measuring the intensity of light scattered in helium and
argon which filled the chamber (the ratio of the Rayleigh scattering
sections of the gases was 66). 18 The results show that this level equals
~3% of the intensity of the light scattered by argon under the normal
conditions so that sufficiently reliable measurements could be taken in
plasma up to a plasma temperature of approximately 9 kK where the
signal:noise ratio decreased to unity.
Measurements were taken of the radial distributions of the argon con-
centration in the arc column in different sections of the column; these
data were then used to compute the distribution of plasma temperature
using the equation of state of the ideal gas. Plasma temperature was
also measured by an independent method on the basis of the absolute
F ig .6.2 Radial distributions of temperature in

ig.6.2
the arc column in the cross-section 2 mm from
the cathode. Total length of the arc column 5
mm, current intensity 100 A, argon; 1) results
of measurements by the Rayleigh scattering
method, 2) results of spectroscopic measure-
ments, 3) calculations in the LTE approximation.
153
Plasma Diagnostics
F ig .6.3 Diagram of equipment for investigating Rayleigh scattering or laser radiation

ig.6.3
in a plasma jet. 1) laser, 2) system for pumping and filling the laser tube, 3) light
flux stabiliser, 4) sensor of laser radiation intensity, 5) collecting lens, 6) a set of
diaphragms, 7) camera, 8) optical trap, 9) interference light filter, 10) polariser,
11) monochromator, 12) photoelectric multiplier, 13) unit for amplifying and recording
signals and controlling equipment, 14) automatic recording device, 15) drive of the
modulator, 16) light diode, 17) photoresistance, 18) plasma torch, 19) unit for cooling
the photoelectric multiplier, 20) distribution of the plasma torch in the chamber
(axial section), 21) feed of purified argon, 22) plasma jet.
intensity of ArI line at 696.5 nm. Figure 6.2 show some of the results.18
It may be seen that the temperatures measured spectroscopically and
by the scattering method greatly differ in the peripheral regions of the
arc column. At the same time, the results of measurements by the scat-
tering method are in good agreement with the results of calculations 19
carried out assuming the validity of the LTE model for the arc plasma.
The calculated results are also in good agreement with spectroscopic
data obtained for the near-axial regions of the arc. Thus, the periph-
eral regions of the arc were characterised by large deviations of the
population of the excited levels from those corresponding to the LTE
model. These deviations are explained by the excitation of argon atoms
at the arc periphery by resonant radiation of the argon plasma of the
high-temperature (~15 kK) core of the arc, as confirmed by the results
of calculations published in Ref.20 in which the process of excitation
and deactivation of argon atoms in the peripheral regions of the arc
column was taken into account.
Thus, experiments with the method of Rayleigh scattering of light
made it possible to determine the reasons for the deviation of the state
of plasma from that described by the LTE model generally accepted for
the examined object.
154
6.4 Plasma jet diagnostics

In our experiments, the method of Rayleigh scattering of light was used
for the diagnostics of the argon plasma jet. The diagram of equipment
is shown in Fig.6.3. We used a LG-106M argon ion laser produced in
Russia, with a radiation power of 0.5–0.7 W at a wavelength of 488
nm. Since the operating life of lasers of this type is very short, the laser
tube was connected to the argon pumping and inlet system so that in
addition to replacing the working gas it was possible to maximise the
radiation power by varying the argon pressure in the tube. Instead of
the standard laser power unit, we used a special voltage stabiliser con-
trolled by the sensor of the laser radiation intensity; it was thus pos-
sible to reduce the fluctuations of radiation intensity from 20 to 1.5%.
The radiation of the laser output was modulated by a rotating perfo-
rated disc at a frequency of 729 Hz. Using a lens (focusing distance
207 mm), the laser beam was focused inside a chamber at its axial line
which coincided with the axial line of the plasma jet travelling into the
chamber from the plasma torch nozzle (diameter 3 mm); the plasma torch
operated at frequencies of 90–120 A, arc voltage 22–24 V, argon flow
rate 0.15–0.25 g/s. At the exit from the chamber, the laser beam was
absorbed in an optical trap. Scattered radiation was examined under an
angle of 90° to the axial line of the beam in the solid angle of ~0.03
steradian using a lens with a focusing distance of 195 mm. An inter-
ference filter was placed in the quasi-parallel light beam behind this
lens (the position of the maximum of transmittance and the width of
the band were 486.9 and 18 mm respectively). The second lens of the
receiving optical system produced the image of the cross section of the
probing beam at the input slit of the monochromator in the focus con-
striction area (the constriction diameter was around 60 µm); the slit of
the spectrometer cut out from the image a section approximately 15 µm
wide. These dimensions characterised the spatial resolution of equip-
ment. A FEU-79 multiplier was positioned at the output of the
monochromator. Its cathode was cooled using a thermoelectric cooler
to a temperature of around –15°C. The signal from the output of the
photoelectric multiplier was transferred to a selective amplifier with a
synchronised detector at the output. The signal was recorded in an
automatic recording device.
The main task in measuring the intensity of Rayleigh-scattered light
was to ensure an acceptable signal:noise ratio. Noise sources were the
parasitically-scattered light, natural plasma radiation, the noise of the
photoelectric detector and the appearance of macroscopic particles in
the probed volume of the plasma flow. To reduce the intensity of the
parasitically scattered light, the internal walls of the chamber were
ground and bevelled in relation to the axial line under an angle of 8°;
155
Plasma Diagnostics
the conicity of the chamber walls also made it possible to prevent re-
turn flows in the chamber bringing macroscopic particles from the outside
into the examined jet. The windows positioned initially under the
Brewster angle in the chamber walls for transmitting the laser beam in
examining scattering were dismantled so that the level of parasitically
scattered light was greatly reduced. The beam was transmitted through
the hole in the wall with a diameter of around 3 mm; a series of dia-
phragms with a gradually decreasing aperture were placed along the path
of the beam from the length to inlet into the chamber. Optical systems
of illumination and examination were placed in duralumin thin-walled
tubes with black walls. The optical traps for absorbing probing radiation
at the exit from the chamber and for absorbing parasitically scattered
light (opposite the aperture of the receiving optical system) were made
of glass with the black external surface. The natural radiation of plasma
was suppressed quite efficiently using an interference filter, a
monochromator and a polariser. Cooling of the cathode of the photo-
electric multiplier made it possible to reduce the level of natural noise
of the multiplier to a level at which the contribution of this noise to
the noise signal was insignificant. To reduce the frequency of appearance
of macroscopic particles in the probing volume, the argon fed into the
plasma torch and for circular blowing of the plasma jet in the cham-
ber was cleaned to remove dust in three consecutive filters FV-1.6; argon
was dried prior to passing through the filters. All these measures and
also careful adjustment of the optical system caused that the signal:noise
ratio approached unity in measurements of the intensity of scattered ra-
diation in the plasma with a temperature of around 9700 K.
In addition to determination of the intensity of Rayleigh-scattered light,
we also measured the pulsations of the intensity (in the peripheral re-
gions of the jet); the laser beam was not modulated in this case.
It is also important to note that the appearance of macroscopic par-
ticles in the probed volume could not be completely prevented; the fre-
quency of appearance of the particles was on average 12 min –1. Since
the intensity of the light scattered by the particles exceeds the inten-
sity of Rayleigh scattered light by 6–7 orders of magnitude, the par-
ticles greatly complicated the measurements. To suppress the noise signals
(including those from the particles) whose amplitude was higher than
the amplitude of the measured signal of Rayleigh scattering, we developed
a special electronic circuit including a photoelectric multiplier for the
duration of the effect of a strong noise; the level of signal discrimination
was selected during measurements. More efficient suppression of the
noise may be ensured by using the method of digital filtration of the
signal. 21
This equipment was also used to measure the temperature distribution
156
in the plasma jet on the basis of the absolute intensity of lines ArI 430
and 696.5 nm. The radiation standard was SI-10/300 tungsten strip lamp.
In the measurements were used the same monochromator and the am-
plifying-recording part of equipment.
Figure 6.4 shows the radial distribution of temperature in the plasma
jet in the vicinity of the outlet of the plasma torch nozzle for two
different mean mass enthalpies of the plasma. Temperature T was
determined from the results of measurements of the outlet concentra-
tion by the scattering method using the equation of state of the ideal
gas p = NkT, where p is pressure (here 10 5 Pa), N is the partial con-
centration, k is Boltzmann’s constant. The same graph shows the
results of spectroscopic measurement. In contrast to the experiment
discussed previously,18 there were no large differences between the results
of individual measurements. It is possible that this is caused by a
considerably lower level of temperatures in the jet core in comparison
with that in the arc, 18 if we take into account the results of calcula-
tions carried out in Ref.20. Figure 6.5 shows the radial distribution of
mean quadratic pulsations of the intensity of scattered light, i.e.
pulsations of the argon atom concentration. It may be seen that the pul-
sation intensity at the periphery of the jet where rapid turbulent mix-
ing of the plasma jet with the surrounding cold air takes place, the
pulsation intensity reaches a high value. The presence of these pulsa-
tions leads to a large increase of the measurement error of the mean
concentration and, consequently, the temperature in the corresponding
regions of the jet.
F ig .6.4 Radial distributions of the temperature in the cross-section of the plasma

ig.6.4
jet of argon situated 1.2 mm from the outlet from the plasma torch nozzle. 1,3)
results of spectroscopic measurements, 2,4) results of measurements by the method
of Rayleigh scattering; specific enthalpy of plasma 2.4 (1,2) and 1.4 MJ/kg (3,4).
157
Plasma Diagnostics
F ig .6.5 Radial distribution of the relative rms pulsations I ′ of intensity I of Rayleigh

ig.6.5
scattered light in the plasma jet in the section situated at a distance of 1.2 mm
from the outlet of the plasma torch nozzle (specific enthalpy of the plasma 2.4 MJ/
kg).
6.5 Gas temperature in glow discharge

This parameter is one of the important parameters that determine the
course of many physical–chemical processes in the discharge plasma
and on substrates placed in the plasma. Gas temperature T g is meas-
ured on the basis of the rotational structure of the spectrum of diatomic
molecules assuming that T g coincides with the rotational temperature
measured directly in this case, or by different contact methods on the
basis of assumptions whose validity cannot always be verified. Inter-
ferometric methods of measuring the gas concentration in the discharge
are complicated and the corresponding optical systems are very
difficult to use. It was therefore interesting to measure the gas concen-
tration in the discharge and, consequently, determine Tg by the Rayleigh
scattering of light method.
Under the glow discharge conditions at p ~ 3·10 2 Pa the gas
concentration is 10 16–1017 cm –3, and in comparison with the previously
examined objects at the atmospheric pressure, the intensity of the scat-
tering signal (with other conditions being equal) should decrease by
almost three orders of magnitude. To increase the signal:noise ratio, the
equipment whose circuit is shown in Fig.6.3, was modified. Instead of
an argon laser, an LT-401 laser fitted with a radiation frequency con-
verter was used; radiation power at a wavelength of 543 nm was 1.2-
1.5 W. The conventional design of the glow discharge tube (internal
diameter 48 mm) was modified in such a manner as to reduce the in-
tensity of parasitically scattered light. The recording system used the
Fabry-Perot interferometer, a polariser and a NDR-6 monochromator;
signals were recorded in the memory of a D3-28 computer. The main
interference in this case was the natural radiation of the discharge plasma.
Since the signal:noise ratio at working pressures of the gas in the dis-
charge of 0.1–1 kPa was 0.6–3, the scattering line was recorded many
158
T g, K
P, kPa
P , kPa
Fig .6.6 Results of measurements of the argon temperature in glow discharge at different
ig.6.6
pressures. The axial line of the cylindrical discharge tube with a diameter of 48
mm, discharge 50 mA.
times (usually 900–2000 times) in the presence and absence of plasma;

the natural radiation of the plasma was recorded with the overlapped
laser beam. The recording cycles were synchronised using a computer
with the cycle of movement of the screen overlapping the beam. The
algorithm of subsequent data processing made it possible to determine
the average value of the intensity of the radiation scattered in the plasma
normalised to that measured in the absence of plasma. The rms error
of measurements was computed for each measurement cycle and its value
varied with respect to temperature from ~17% at lower pressures to
~10% at higher pressures in the pressure range given previously.
Figure 6.6 shows the data on the gas temperature at different pressures
in the discharge 1 obtained by the scattering method. The temperatures
measured by Yu.A. Ivanov and V.A. Timakin by the thermal probe
method 2, described in Chapter 13, are also presented here. Regard-
less of the relatively good agreement of the results of the individual meas-
urements, it should be noted that the Rayleigh scattering method in this
form is characterised by a very high error. On the basis of the experi-
mental material, it appears that in Rayleigh diagnostics of stationary
plasma it is necessary to use pulsed lasers with the pulsed time in the
range from one to several microseconds with radiation energy in the pulse
of several microJoules permitting retuning with respect to the wavelength
(to eliminate the interfering lines of plasma radiation). It is thus pos-
sible to increase greatly the accuracy of measuring the gas concentration
at low gas pressures in the discharge. It should only be remembered
that for specific discharge conditions it is important to evaluate the degree
of perturbation of the plasma by laser radiation at high laser radiation
energy levels.
159
Plasma Diagnostics
Chapter 7
REFRACTOMETRIC PLASMA DIAGNOSTICS METHODS

Refractometric examination of plasma is based on measuring its refractive
index. Heterogeneous media are examined by interferometric and shadow
methods. In the interferometric method, the refractive index is computed
from the experimentally measured values of the optical difference of
the path. In the shadow method, the refractive index is determined from
the angle of deviation of the light beam from its initial direction as a
result of deformation of the light wave during passage through the
examined object. In the shadow method, we examine the shadow on the
screen of some non-transparent barrier along the path of light propa-
gation, and this is also reflected in the name of the method. Subsequently,
using the relationship of the refractive index with the internal plasma
parameters (temperature, electron and atom concentration), the values
of these parameters can be determined. Refractometric devices together
with the measurement of plasma temperature and the concentration of
atoms and electrons can be used to examine the nature of plasma flows.1
In the laminar gas discharge flow (plasma) there are no local density
fluctuations. The interference pattern has fringes stable with time. Large-
scale density pulsations leads to erosion of the interference pattern when
the period of large-scale plasma flow fluctuations is considerably smaller
than the exposure time. The large-scale pulsations are not visible on
the interferograms with a short holding time. Local heterogeneities with
high-density gradients, characteristic of turbulent flows, can be seen on
interferograms as dark and light spots.
7.1 Refractive index of plasma

The group and phase velocities are defined in propagation of light in
a medium. The group velocity is associated with energy transport and
is always lower than the light velocity in vacuum. Phase velocity c p is
associated with the displacement of the phase of the light wave and can
be both lower and higher than the light velocity in vacuum. The refractive
index is determined by the difference of the phase velocity from the light
velocity in vacuum c and is equal to the ratio n = c/c p and the values
160
Refractometric Plasma Diagnostics Methods
n > 1 and n < 1 are possible. The value of the phase velocity is de-
termined by the interaction of an electromagnetic wave with the elec-
trons of the medium, both bonded and free.
The refractive index n and the absorption factor of the medium k
are linked by the Kramers–Kronig relationship 2
z a f
¥
af
n w -1 =
c
4p
k w ¢ dw ¢
a f
w¢ w¢ - w
. (7.1)
-¥
The refractive index at frequency w depends on the absorption factor

in the entire spectrum range.
The dependence (7.1) holds for any processes leading to light ab-
sorption and can be used to calculate the refractive index from the avail-
able spectral absorption factor. For the spectral lines, the absorption
factor is:
af
kw =
pe 2
me c
NfPw, af (7.2)
where N is the concentration of absorbing atoms, f is the force of the

oscillators in absorption.
Substituting (7.2) into (7.1) and computing the integral, we obtain
for the Lorentz contour P L(w)
af
n w -1= -
pe 2 Nf w - w0
b
me w 0 w - w 0 + g / 2 g b f 2
, (7.3)
where w 0 is the centre of the line, g is its half width.

Figure 7.1 shows the course of the function n (w)–1 in the vicin-
ity of the centre of the spectral line. The range of rapid variation of
the refractive index in the vicinity of the spectral line is referred to as
the anomalous dispersion zone.
Measurements of the refractive index in this zone are usually taken
by the Rozhdestvenskii ‘hook’ method which enables, together with equa-
tion (7.3) to determine either the concentration of absorbing atoms or
the force of the oscillators. The method is highly accurate in operation
with homogeneous objects (for example, plasma in a shock tube). In
the case of heterogeneous plasma, the ‘hooks’ are eroded and the method
is difficult to use in this application.
161
Plasma Diagnostics
F ig .7.1 Refractive index for the Lorentz contour of the spectral line. 1) n ω, 2) P L(ω).
ig.7.1
Away from the centre of the spectral line at frequency w, satisfy-

ing the conditions w<<w 0, w 0 – w >>g/2, equation (7.3) assumes the
form
af
n w -1=
2 pe 2
me w 20
Nf . (7.4)
It may be seen that in this case the refractive index does not depend
on the probing radiation frequency w. Equation (7.4) holds for any
contour of the spectral line (with the exception of the purely Doppler
contour) because they have the same asymptotics in the distribution of
the absorption factor. In calculating the refractive index of plasma at
frequencies away from the spectral lines, it is sufficient to take into
account the contribution of ground states of the atoms (ions) because
their value Nf is considerably higher than for the excited states. The
condition w << w 0 is fulfilled in examining gases (nitrogen, oxygen,
argon, etc.) in the visible region of the spectrum because their reso-
nance lines are situated in the vacuum ultraviolet region of the spec-
trum.
The contribution of the free plasma electrons to the refractive in-
dex can be explained by the interaction of the probing electromagnetic
wave with a wave in plasma at the plasma frequency. The latter is de-
termined by the superimposition of the ordering effect of Coulomb in-
teractions in the Debye sphere on the chaotic motion of the electrons
and is equal to
162
F e Ne I
= 2p G
2 1/ 2
wp
H pm JK
e
.
Under laboratory plasma conditions, the plasma frequency is usu-

ally low and situated in the infrared range of radiowave frequency range.
At the probing frequency w >> w p the refractive index, determined by
the free electrons, is lower than unity: 2
af
n w -1 e = -
2 pe 2 N e
me w 2
. (7.5)
When calculating the refractive index at a given frequency, it is im-

portant to take into account the contribution of the spectral lines and
free electrons. For example, for argon plasma:
(n – 1) = (n – 1) ArI, (n – 1) Ar*I + (n – 1) ArII + (n – 1) e.
The components of this sum at λ = 5000 Å: 3
(n – 1) ArI = 1.04·10 –23N ArI, (n –1) Ar*I = 13.2·10 –23N Ar*I ,

(n – 1) ArII = 0.715·10 –23 N ArII, (n – 1) e = –13.4·10 –23N e.
Here ArI, Ar*I, ArII denote the neutral and excited argon atoms and
argon ions, respectively. Since N Ar*I << NArI, then (n – 1)Ar*I << (n – 1)ArI
and the contribution of the excited states to the refractive index can be
ignored, as mentioned previously. The contribution of argon ions is also
negligible: it is considerably smaller than the contribution of free elec-
trons (N ArII » N e). Thus, it is sufficient to take into account the con-
tribution of the argon atoms in the ground state and the contribution
of the free plasma electrons. As an example, Fig.7.2 shows the depend-
ence of the refractive index of the argon plasma at atmospheric pres-
sure at a wavelength of l = 6328 Å. The refractive index of the plasma
decreases with increasing temperature as a result of a decrease of the
atom concentration due to thermal expansion of the gas at constant
pressure and atom ionisation. The free electrons provide a negative
contribution to the value n – 1 and determine mainly the refractive index
at an argon ionisation degree higher than 10%. For most gases, the re-
fractive index can be calculated quite accurately because it is deter-
mined by the spectral lines of the main series (the forces of oscillators
of these lines are available with a high degree of reliability) and by free
163
Plasma Diagnostics
F ig .7.2 Refractive index of argon plasma at a wavelength of λ = 6328 Å at P = 1

ig.7.2
atm (1) and the coefficient of transfer of the error (2).
electrons whose contribution is computed with high accuracy from

equation (7.5).
In the case of LTE plasma, the dependence n (T), identical with that
shown in Fig.7.2, can be used to determine the plasma temperature from
the measured refractive index. If the plasma is in the partial thermo-
dynamic equilibrium state, the measurements of the refractive index at
two relatively different frequencies w 1 and w 2 make it possible to de-
termine the concentration of atoms and ions. In fact, the refractive index,
determined by the atoms, does not depend on frequency (see equation
(7.4)), whereas the refractive index, determined by the free electrons,
is inversely proportional to the frequency (see equation (7.5)). We have
two equations
n (w 1) – 1 = (2 e 2/m ew 20) fN a – (2pe 2/m ew 21) N e,

n (w 2) – 1 = (2 e 2/m ew 20) fN a – (2pe 2/m ew 2e) N e
for determining the concentration of atoms Na and the electrons Ne. Using
Sah’s equation, it is possible to calculate the electron temperature T e
and compare it with the temperature determined by a different method
assuming LTE and evaluate at the same time the deviation of the state
of plasma from local thermodynamic equilibrium.
The common feature of any measurement system of the refractive
index is that the value z = n – n R is determined directly; here n R is the
refractive index of the medium outside the examined object. In calcu-
lating temperature n = z + nR is used. It is therefore necessary to know
the refractive index n R of the surrounding environment. Since nR is usu-
ally considerably higher than the refractive index of plasma, this pa-
rameter can be determined with high accuracy. The value n R can be
measured directly by the interferometry method using, as a reference,
164
a cuvette with a thoroughly scrubbed gas whose refractive index is known

accurately (hydrogen). n R can also be determined indirectly, as is the
case in the combined spectroscopic–refractometric method of temperature
detemination. 1 From the temperature T 0 at the axis of the plasma ob-
ject determined by the spectroscopic method it is possible to find, us-
ing the dependence n(T) (Fig.7.2), the refractive index at the axis n 0
from which n R = n 0 – z 0 , where z 0 is the value measured by the
refractometric method at the axis of the object.
The relationships between the errors of measurement of the refrac-
tive index and temperature in the case of LTE plasma will now deter-
mined. Varying n = n (T), gives dn = (dn/dT)dT, and consequently dT/
T = (dT/Tdn)dn.
In turn, dn = dz + dn R, since dT/T = (dT/Tdn)(dz + dn R). The last
equation will be written in the form dT/T = (dT/Td(n –1)) [dz + d
(nR – 1) since in practice it is more convenient to operate with the value
n – 1 because of the fact that n differs only slightly from unity. Ap-
proximately, it can be written that z = n R – n » n R – 1 and dz/(n R –
1) » dz/z and, consequently,
dT n R - 1 ¶T LM dz + dbn - 1g OP.
a f N z n -1 Q
R
» (7.6)
T T ¶ n -1 R
The dependence of the transfer factor of the error [(n R – 1)/T] [¶T/
¶(n– 1)] on the argon plasma temperature is shown in Fig.7.2.
The accuracy of the refractometric method is considerably lower than
that of the spectroscopic method. The refractometric method gives reliable
results only for a highly stable object at a sufficiently high frequency
of the measurements and detailed processing of the results. As shown
in Ref.1, it is possible to measure the temperature with an error of 5-
10%. An important advantage of the refractometric method of the meas-
urement is that is enables the latter to determined at the periphery of
the plasma object outside the radiation zone where the spectrometric
method cannot be used. The combined application of the spectroscopic
and refractometric methods enables a complete radial temperature dis-
tribution to be plotted.
7.2 Plasma interferometry

Optical systems4,5 for determining the refractive index in interferometric
experiments are oriented at measurements of the optical difference of
the path of two beams, one of which passes through the examined object
with thickness l with a refractive index n, the other one passes outside
165
Plasma Diagnostics
Beam 1
L3 SC
Beam 2
S
SL
L1 L2
F ig .7.3 Mach–Zender interferometer.

ig.7.3
the object where the refractive index is equal to nR. The measured optical
path difference is
µ = l (n – n R) (7.7)
The most efficient device for measuring the optical path difference
of gas or plasma objects is the Mach–Zender interferometer. However,
this equipment is of complicated design and is expensive. An interfer-
ometer based on a modification of a standard IAB-451 shadow equipment
is used more widely. It is also less sensitive to vibrations and is sim-
ple to set up.
Mach–Zender interferometer
The circuit of the Mach–Zender interferometer is shown in Fig.7.3. It
is a two-beam interferometer consisting of two mirrors M1 and M2 and
two plain-parallel semi-transparent sheets P 1 and P 2. The distances
M 1P 1 = M 2P 2 and M 1P 2 = P 1M 2 are equal. The mirrors M 1, M 2 and the
sheets P 1, P 2 are parallel and ranged in pairs. The light source S is re-
flected by the lens L 1 on the slit SL of the device situated in the fo-
cus of the collimator L2. The beam light from the source of illuminating
radiation is divided by the sheet P1 into two beams which after reflection
from M 1 and M 2 and passage through P 2 are projected onto the screen
SC. The examined object in placed in the beam bundle 1. The lens L 3
forms the image of the object O on the screen. The beam bundles, the
object 1 and reference 2, are coherent in relation to each other and
generate an interference pattern on the screen. If all the mirrors and
sheets are parallel, then in the absence of the object the width of the
interference fringes, formed on the screen, is infinite (the interference
field is uniformly illuminated). The introduction of an object leads to
the appearance of fringes whose shape corresponds to the curves of equal
166
values of the optical path difference. The interference fringes are lo-
calised in the planes of the object and its image on the screen so that
it is possible to determine the correspondence between the position of
the fringe on the screen and the line equal to the optical path differ-
ence in the objects. If mirrors M and sheets P are positioned under a
small angle in relation to each other, an additional path difference forms
between the beams 1 and 2 and finite-width fringes appear on the screen.
The fringes are lines of the equal path difference which differ for the
adjacent lines by the light wavelength. The fringe width depends on the
angle between M and P and increases with its decrease. In the pres-
ence of an optically heterogeneous medium (examined object) the in-
terference fringes on the screen change their shape in accordance with
the distribution of the optical path difference in the object. The distance
between the branches of the Mach–Zender interferometer is sufficiently
large so that very large objects can be examined. Interferograms can
be recorded using a camera placed instead of the screen. Lasers are
usually used as the source of probing radiation in plasma diagnostics.
As a result of the high brightness of laser radiation and the use of a
narrow-band light filter, it is possible to prevent illumination of the
interferogram by plasma radiation.
Figures 7.4 and 7.5 show interferograms of the flame of a spirit torch
in setting the interferometer in relation to the infinite-width and finite-
width fringe. The fringes of the equal optical path difference (Fig.7.4)
correspond to the additional path difference µ = kl, k = 1, 2, …, with
counting carried out from the periphery to the centre of the object. In
processing the interferogram, the position of the fringes on the coor-
dinate axis X with the start of counting in the centre of the object is
measured in the given cross section of the object (axisymmetric in this
case) and the dependence µ=µ (x) is plotted. When processing the
interferogram of finite-width fringes (Fig.7.5a), the additional path dif-
F ig .7.4 Interferogram of the flame of a spirit torch. The fringes of equal optical
ig.7.4
path difference.
167
Plasma Diagnostics
F ig .7.5 Interferogram of the flame of a spirit torch. Finite-width fringes (in the
ig.7.5
absence of the object the interference fringes are horizontal over the entire plane).
ference introduced by the examined object, is determined using the

following procedure (Fig.7.5b). In the given cross section measurements
are taken, for different values of the co-ordinate x, of the shift of the
fringe in relation to the initial position in fractions of the distance
between the pair of unperturbed fringes whose optical path difference
is λ. The relationship between the measured additional path difference
µ (x) and the required distribution of the value z(r) = n(r) – n R(n(r) is
the refractive index at the circumference of the radius r, n R is the re-
fractive index outside the object) is given by the Able integral equa-
tion
af z af
R
n r - n R rdr
m x =2 , (7.8)
x r2 - x2
which is converted to
af af z af
R
1 dm x dx
z r = n r - nR = - (7.9)
p dx x - r2
2
r
168
Equations (7.8) and (7.9) are valid for small angles of deflection of
the light beam in passage through the object. If the deflection angle is
large, the interference fringe and its processing are greatly complicated.
In this case, it is convenient to transfer to shadow examination meth-
ods.
The ‘slit-filament in the focus’ interferometer

The main simplified circuit of this interferometer is shown in Fig.7.6.
The light source S is reflected by the lens L 1 onto the slit SL of the
device which is situated in the focus of the collimator objective L 2. The
parallel light beam, leaving the objective L 2, is focused by the camera
objective L3 in the focal spot F and then projected on to the screen SC.
The non-transparent filament H is placed in the focus of the camera
objective and overlaps the image of the slit. The thickness of the filament
is slightly larger than the image width of the slit (filament thickness
~0.1 mm, the width of the slit image ~0.08 mm). The light carrying
out diffraction on the filament uniformly illuminates the screen in the
absence of the heterogeneous object. The examined object O is placed
in the light beam between the objectives L2 and L 3 and reflected by the
objective L 3 on to the screen. Because of the presence of the gradient
of the refractive index in this object, the light beams are deflected from
the initial direction and pass away from the filament onto the screen.
The deflected light beam is displayed on the screen at the same point
as the unperturbed beam because the planes of the object and the screen
are connected by the lens L3. The rays of the light of the parallel beam,
passing outside the object, are overlapped by the filament, but part of
the light falls on to the screen as a result of diffraction on the filament.
The light rays formed in this manner and passing through the object
and outside it interfere and the screen shows the fringes of equal op-
tical path difference, similar to those shown in Fig.7.4. At specific ratios
between the size of the object and the width of the slit and the fila-
ment the interference fringe in the described interferometer is identi-
cal with that in the Mach–Zender interferometer with setting to the
infinite-width fringe.
S
SL
L1 L2 L3 SC
F ig .7.6 Diagram of the “slit-filament in focus” interferometer.

ig.7.6
169
Plasma Diagnostics
D G
Beam 1
S SL Beam 2
L1 L3
SC
L2
F ig .7.7 Diagram of an interferometer with a diffraction grating.
ig.7.7
An interferometer with diffraction grating

A simplified circuit of the interferometer with a diffraction grating is
shown in Fig.7.7. The parallel light beam is divided into two beams
using diaphragm D. The examined object O is projected by the objective
L 3 on the screen SC and is placed in one of these beams. The trans-
parent diffraction grating DG (50–100 lines/mm) is positioned in the
vicinity of the focus F of the camera object L 3. Each of the probing
beams is split by the diffraction grating into a number of beams. The
rays from the first and second bundles that overlap on the screen in-
terfere together. In the absence of the object, the screen shows inter-
ference fringes of finite width with the spacing
mb f
D=
b g
k1 - k2 s
, (7.10)
where m is the scale of reflection of the object on the screen; b is the

grating constant; s is the distance of the grating from the focus F; f
is the focusing distance of the objective L 3; k 1, k 2 is the order of in-
terference diffraction maxima of the first and second beams. When the
examined object is placed in the beam 1 the interference fringes are
distorted. This gives an interferogram identical with that shown in
Fig.7.5a for the Mach–Zender interferometer. If the diffraction grating
is placed in the focus F then in accordance with the equation (7.10)
s = 0 and D ® ¥ and the screen shows the fringe of infinite width.
For this setting of the interferometer and placing the examined object
in the beam 1 the screen shows fringes of the equal optical path dif-
ference of the type shown in Fig.7.4. Thus, the interferometer with a
diffraction grating makes it possible to realise two setting schemes: finite
width fringes at s > 0 and an infinite width fringe at s = 0.
Shearing interferometer
The diagram of the shearing interferometer differs from those described
previously by the fact that a light-dividing device (a system of mirrors,
170
F ig .7.8 Displaced images of the object on the screen of a shearing interferometer.

ig.7.8
Wollaston prism) is positioned in the vicinity of the focal plane of the

camera objective and splits the light beam into two beams in such a
manner that the screen shows two images of the examined object shifted
in relation to each other (Fig.7.8). In this method, the light beams which
pass through the object along the chords are separated by the shift dis-
tance x interfere. The corresponding optical path difference
z zb
y0¢
af a f
y0
m x = n r , x - n R dy - g
n r, x + x - n R dy, (7.11)
- y0 - y0¢
where the first integral corresponds to the optical path difference for
the chord whose distance from the centre is x (initial image), the sec-
ond integral for the chord (x + x) (shifted image). For the most inter-
esting case – small shift – we can write
a f b
n r, x - n r, x + x » g a f
¶n r , x
dx =
¶n r , x
x
a f
¶x ¶x
and then
af z ar, x f - n br, x + xgdy = x z a f

y0 y0
¶n r , x
m x = dy. (7.12)
¶x
- y0 - y0
Taking into account the rotational symmetry r 2 = x 2 + y 2
¶n ¶n ¶r dr x rdr
= = , = , dy = .
¶x ¶r ¶x dx r r - x2
2
171
Plasma Diagnostics
Consequently, equation (7.12) takes the form

R
dn dr
µ ( x ) = 2ξx ∫ dr
x r 2 − x2
or
µ( x)
R
dn rdr
ξx
=2
x
∫
rdr r 2 − x2 (7.13)
Equation (7.13) describes the relationship of the optical path difference

with the gradient of the refractive index in the shearing interferometry
method at a small image shift. This is an integral Able equation that
is transformed to
µ ( x ) dx
R
dn 1 d
rdr
=−
ξπ rdr ∫
r x2 − r 2
. (7.14)
Multiplying both parts of (7.14) by rdr and integrating over the region
from r to R gives
1 µ ( x ) dx
R R
− dn ≡ n ( r ) − nR =
∫ ∫ . (7.15)
r
ξπ r x2 − r 2
For comparison, the equation of the normal interference method (7.9)

has the form
1 dµ ( x ) dx
R
n ( r ) − nR =− ∫ .
π r dx x2 − r 2
It may be seen that equation (7.15) does not contain the operation
of differentiation of the experimental function m(x). This must increase
the accuracy of determining n (r) from the shearing interferograms in
comparison with normal ones.
Figure 7.9 gives the shearing interferogram for a cylindrical object.
The shearing interferograms are processed by the same procedure as
normal ones. The shift of the fringe (7.9b) in relation to the initial
172
F ig .7.9 View of a shear interferogram (a), b) processing of shearing interferograms,

ig.7.9
L ( x ) is the geometrical shift of the fringe, ∆ is the distance between the maxima
of the non-perturbed fringes.
position is measured for difference values of co-ordinate x in fractions

of the distance between the pair of unperturbed fringes whose optical
path difference is λ. This gives the dependence µ (x).
7.3 Shadow method

This method is based on widening the angle of deflection of the light
beam from its initial direction in passage through an optically hetero-
geneous object. In experiments, the shadow from some barrier in the
path of the light is observed on the screen. The displacement of this
shadow from the initial position is measured. The more suitable method
is the shadow method of the defocused inclined filament6 (Fig.7.10). The
following symbols are used in Fig.7.10: a is the distance from the object
to the lens L 3 with a focusing distance f; s is the distance of the bar-
rier (filament) to focus F (defocusing); A is the barrier (inclined fila-
ment); b) is the distance between the screen S and the barrier. The
examined object is illuminated with a parallel light beam, B is one of
the points of the object. Length L 3 reflects the object onto the screen,
B¢ is the image of the point B.
Two beams, one of which (beam 1) passes through the object (point
Beam 2 L3
Beam 1 S
F ig .7.10 Diagram of a shadow device (chamber path) of a defocused inclined filament.

ig.7.10
173
Plasma Diagnostics
B), and the other one (beam 2), outside the object will now be exam-
ined. In the absence of the object, all beams pass through the focus F
and the shadow from the barrier A by the beam 2 is projected to point
A¢. In the presence of the beam 1 is deflected by the angle θ, passes
away from the focus and forms a shadow from barrier A at point A²
whose distance from A¢ is d. The beam 1 deflected by the object is dis-
played on the screen at the same point as the unperturbed one because
the planes of the object and the screen are connected by the lens L 3.
The relationship of the measured value of displacement of the shadow
d with the required deflection angle q will be determined. In accord-
ance with Fig.7.10, c = a tan q » aq for small angles. From the similarity
of the triangles with common tips at the points A and F we have
d/b = (c + d)/(f + s), d/f = d(b + s). Using the equation of the lens
1/f = 1/a + 1(f + b + s) and the derived relationships finally gives
θ = ( a − f ) sδ / f  f 2 − s ( a − f ) ,
Usually s << f and a = 2f (the object is displayed on the screen in

the natural size) giving
s
q= d. (7.16)
f2
At the deflection angle q the displacement of the shadow on the screen

d increases with decreasing defocusing of the filament s. However, the
slit width also increases proportionately (Fig.7.11). In the beam optics
approximation h¢/h = (b + s)/s. From Fig.7.10 at a = 2f we obtain
b + s = f and, using equation (7.16)
d f
=q . (7.17)
h¢ h
F ig .7.11 Projection of a filament onto a screen.

ig.7.11
174
F ig .7.12 A shadow pattern of the method of the inclined defocused filament for a
ig.7.12
cylindrical non-uniform object (a) and the function δ (x) (for processing of the shadow
pattern) (b).
It can be seen that the fraction of the displacement of the shadow

in relation to its width d/h¢, which also determines the accuracy and
sensitivity of the method, does not depend on the value s. The width
of the shadow can be decreased using, as a barrier/filament, a phase
sheet at half the wavelength.
The form of the shadow fringe in the method using an inclined de-
focused filament shown in Fig.7.12. To simplify considerations, it is
assumed that the heterogeneous object is cylindrically symmetric with
the gradient of the refractive index only in the radial direction. The initial
shadow of the filament – the direct inclined line – is transformed when
the examined object is placed in the path of the beams. The displace-
ment of the shadow for several characteristic points is clearly visible.
The lower part of the figure shows the dependence d(x) obtained on the
basis of the shadow pattern. Measuring d(x) equation (7.16) can be used
to determine q(x).
The relationship of the deflection angle with the gradient of the re-
fractive index and the radial distribution of the refractive index is de-
termined using the following procedure. For a small deflection angle
of the light beam (Fig.7.13)
¶ ln n
dq = dy. (7.18)
¶x
175
Plasma Diagnostics
F ig .7.13 Diagram of passage of the beam through a non-uniform cylindrical object.

ig.7.13
Integrating (7.18) along the chord [–y 0, y0], whose distance from the
axis of the object is x, gives
y0
∂ ln n
θ( x) = ∫ dy. (7.19)
− y0
∂x
Taking into account the rotational symmetry r 2 = x 2 + y 2
¶ ln n ¶ ln n dr dr x rdr
= , = , dy =
¶x ¶r dx dx r r 2 - x2
and equation (7.19) takes on the form
θ( x )
R
dln n dr
x
=2
x
dr ∫r 2
− x 2
. (7.20)
The integral Able equation (7.20) gives the relationship of the an-
gle of deflection of the light beam with the radial gradient of the re-
fractive index. To determine the radial distribution of the refractive index,
equation (7.20) is transformed to
θ ( x ) dx
R
d ln n 1 d
rdr
=−
π rdr ∫r x2 − r 2
. (7.21)
Multiplying (7.21) by rdr and integrating it from r to R gives
176
1 θ ( x ) dx
R R
n
− dln n ≡ − ln R = n ( r ) − nR = −
∫ ∫ .
r
n(r ) π r x2 − r 2 (7.22)
Equation (7.22) is used to calculate the radial distribution of the re-

fractive index from the measured deflection angle of the light beam. It
should be noted that the equation (7.22) for the shadow method is iden-
tical with (7.15) for the shearing interferometry method and does not
contain the derivative of the measured value. The latter determines the
high stability of the calculations of n (r) in relation to random meas-
urement errors. It should be noted that the values q(x) in the shadow
method and µ(x)/x in the shearing interferometry method are identical,
although they were obtained using different optical systems.
As mentioned, the equations (7.15) and (7.22) are valid for small
angles of deflection of the probing beam light when the integral (7.19)
is taken along the chord [–y 0, y 0]. Integration along the beam with an
arbitrary trajectory leads to the expression 7
n ( r ( s )) = nR exp −Φ ( r ( s )) . (7.23)
Here
1 θ ( x ) dx
R
Φ (s ) = ∫ , (7.24)
π s x2 − s2
r ≈ s ⋅ exp Φ( s ) . (7.25)
From the measured deflection angle q(x) we calculate the value F(s)
from (7.24) and then determine n (r(s)) from (7.23) where the relationship
of r with s is described by equation (7.25). The small angle approxi-
mation is present in these equations as a limiting case: F(s) << 1. Since
F(s) is determined from the experimental data on q(x), we can also verify
the degree of the reliability of this approximation. At F(s) << 1, we
have r » s and n(r) » n R(1 – F (r)) » n R – F(r), which is in agree-
ment with equation (7.22).
7.4 Diffraction interferometer based on IAB-451 shadow

equipment for plasma investigations
The circuit of IAB-451 equipment was proposed and developed by D.D.
Maksutov. An advantage of this equipment is the possibility of easy sepa-
177
Plasma Diagnostics
ration of the collimator and camera parts from the main objectives so
that it is possible to use different optical and mechanical attachments
designed for operation with a specific method. The universality of this
equipment in carrying out shadow and interference investigations was
described by S.A. Abrukol and L.A. Vasil’ev.5,8 The results5,9 were used
efficiently for the rapid development and construction of a diffraction
interferometer on the basis of IAB-451 shadow equipment. IAB-541 was
converted to a diffraction interferometer by introducing transparent
optical gratings into the optical system of the equipment.
The main circuit of IAB-451 equipment used to examine the nature
of flows is shown in Fig.7.14. Equipment consists of two identical sys-
tems: the collimator system, generating a parallel light beam for illu-
minating the examined heterogeneity, and the chamber system used for
examination and photography. The collimator part consists of a
mirror–meniscus objective with a diameter of 230 mm and a focusing
distance of 1917 mm and a flat mirror for changing the direction of
the beams. The condenser produces an intermediate image of the light
source in the focal plane in the mirror–meniscus objective. The cam-
era part has the same structure. Instead of the slit and the blade, trans-
parent diffraction gratings P 1 and P 2 are placed in the focal planes of
the objectives O 1 and O 2. To produce two beams (working and refer-
ence), the authors of Ref.9 proposed to place a field diaphragm with
two windows in the plane of the object P. The grating P 1 is used only
for increasing the intensity of illumination of the interference field from
the source with a continuous or linear radiation spectrum. The appli-
cation of the laser as a light source together with the grating P 1 does
not require field diaphragm to be positioned in the object plane of the
device because the required two light beams can be produced directly
behind the grating P 1. The light beam from a helium-neon laser, fall-
ing through the objective O on the transparent diffraction grating P 1,
is divided by the grating to a number of beams. Each beam represents
Ob
LG-75
F ig .7.14 Diagram of IAB-451 device, P is the photorecording device.

ig.7.14
178
a specific order of diffracted light, two of which – zero and first

orders – pass through the objective of the collimator O1 and are shaped
into beams with flat wave fronts. One of the beams passes through the
examined plasma, the other one is a reference beam. The remaining
beams leave the grating P 1 under diffraction angles which are such that
the beams are cut off by the mounting of the objective of the collimator
and do not arrive in the reception part. The diameter of the beams a
is given by the focusing distance F of the objective O placed between
the laser and the diffraction grating P 1
f +F f
a = a¢ » a¢ ,
F F
where f = 1917 mm is the focusing distance of the collimator objec-

tive, a¢ = 3.5 mm is the diameter of the beam at the exit from the la-
ser.
In the reception part of the device, these two beams are directed on
to the diffraction grating P2. The transparent grating P2, which is a com-
pulsory element of the circuit of the diffraction interferometer, ensures
the formation of an interference field which coincides with the image
plane. The distance between the centres of the light beams b satisfies
the condition of producing an interference pattern: bd = lf, since the
collimator and reception sections of IAB-451 device are identical (l is
the wavelength of probing radiation, d is the spacing of the grating P 2).
The diffraction grating represents the object plane in the form of the
maxima of zeroth, ±1st and so on orders which are differently readjusted9
in relation to the constant of the grating P2 forming the interference field.
Interference is of the two-beam nature because only two beams from
two superimposed orders meet at each point of the interference field.
If the spacing of the grating P 1 is equal to that of P 2, the interference
pattern forms in superimposition of the zeroth and +1 st or zeroth and
–1 st orders of the diffracted light of the reference and working beams.
The even orders of light diffraction (±2 nd, ±4 th) do not appear because
the ratio of the width of the transparent groove to the non-transparent
one in the diffraction gratings is equal to unity. The spacing of the
gratings P 1 and P 2 is selected in accordance with the recommendations
in Ref.5. The diffraction gratings P 1 and P 2 have the spacing d = 0.01
mm. The source of radiation in the device is a continuous gas laser LG-
75, l = 6328 Å. A pattern formed by the interference of the zero and
±1 st orders of the working beam is used for the investigations. The
grating grooves were parallel to each other. The infinite width fringes
are used when the diffraction grating P 2 is situated exactly at the
179
Plasma Diagnostics
focal point of the main objective of the reception part of the device.
In this case, the light waves from different points of the working field
(and the reference field) arrive in the image plane in the same phase,
and the intensity of illumination of the interference pattern is the same
for all points of the image and depends on the difference of the phases
between the waves propagating from the working section and their
comparison section.
The working and the reference beam are positioned in space alongside
each other in the horizontal plane or one above the other, depending
on the system used.
The objects characterised by strong intrinsic glow are usually dif-
ficult to record in interferometric diagnostics because additional measures
must be taken to weaken this glow. Although the addition of narrow-
band interference filters to the optical system of the device leads to the
required weakening of the radiation of the object, it nevertheless greatly
impairs the quality of interferograms: the monochromatic radiation of
a laser generates additional interference fringes on the interferogram after
passing through the light filter and these fringes influence the quality
of the main interference pattern. The use of conventional absorbing light
filters leads to smaller distortions but is not always suitable because
such filters are characterised by a wide transmission band.
The required coefficient of weakening of radiation of the plasma jet
is obtained in this case using several simultaneous methods. Firstly, at-
tenuation is carried out using a KS-14 light filter placed in the reception
part of the device in front of the photorecording device. Secondly, plasma
radiation is weakened by a diaphragm positioned on the mirror objective
of the reception pattern device. The dimensions and configuration of
the holes in the diaphragm correspond to those of probing beams. The
light from the plasma jet is weakened by the value determined from the
ratios of the entire area of the mirror objective to the beam area. For
example, at an objective diameter of 230 mm and diameters of two
identical orifices in the diaphragm of approximately 60 mm, radiation
is weakened by approximately an order of magnitude. The third fac-
tor weakening the parasitic plasma radiation is the diffraction of this
radiation on the grating P 2. The grating divides ‘wide’ radiation of the
plasma jet into a spectrum, the image of the jet is stretched in the di-
rection normal to the groove of the grating. The degree of attenuation
increases with increasing frequency of the grating. This property of the
diffraction interferometer is an advantage in comparison with other types
of interferometers. As a result of combined use of this method it is pos-
sible to produce an interference pattern that is not ‘burdened’ by the
presence of natural plasma radiation.
180
The interferograms are recorded by a mirror-type photographic camera

highly suitable for visual examination. The gain factor of the system
in recording the interferograms is selected to ensure the photograph of
the standard size shows the examined region. The high monochroma-
ticity of probing radiation and overexposure enables interferograms of
good quality to be obtained.
Both interfering beams are produced by the same components of the
device and, consequently, this device is far less sensitive to vibrations
than Mach–Zender-type interferometers. Special measures are taken to
protect IAB-451 device against vibrations duration operation: the de-
vice is positioned on a foundation and there is a rigid link between the
collimator and camera parts. The two-beam diffraction interferometer
results in the error of measurement of the fringe shift of ~0.05l.
7.5 Examination of the nature of the plasma flow

Two-jet plasma torch
The interferometric method described above was used to examine jets
and the main plasma flow of a two-jet dc plasma torch. 10 Plasma jets
were discharged into air. The main operating regime of the plasma torch:
arc current 105 A, the consumption of the working gas (argon) 0.12-
0.36 g/s, supplied power 16 W, efficiency up to 0.9. IAB-451 device
was constructed for infinite width fringes. Exposure time was 2 ms. Typi-
cal interferograms of the plasma flow for gas flow rates of 0.12; 0.24;
0.36 g/s are presented in Fig.7.15a-c. The initial angle of convergence
of the jets was 60°. For intermediate gas flow rates, the shadow pat-
tern of the plasma flow does not differ qualitatively from those pre-
sented here and enables regions of the plasma where the flow is still
laminar to be defined. The presence of stable interference fringes, de-
tected visually, indicates the absence of large scale pulsations of den-
sity. Examination under different angles shows a direct relationship
between the angle of convergence of the jets and the flow regime. In-
creasing initial angle between the plasma jets decreases the flow rate
of the plasma forming gas at which turbulent pulsations appear.
Analysis of the interferograms shows that at argon flow rates less
than 0.15 g/s and the angles of convergence of the jets of less than 60°
the plasma flow is laminar over a large length. The length of the sta-
ble plasma flow is approximately 50 cm. At a gas flow rate of 0.24
g/s, the main plasma flow is laminar at a distance of up to 30 mm from
the area of merger of the jets. This is followed by the appearance of
pulsations. At a gas flow rate higher than 0.36 g/s these pulsations
‘spread’ the interference fringes downwards along the plasma flow up
to the area of convergence of the electrode jets. A further increase of
the gas flow rate results in unstable operating regime of the plasma torch,
181
Plasma Diagnostics
a b
c d
e f
F ig .7.15 Interferograms of a plasma flow ( I = 105 A) in the regime of infinite width

ig.7.15
fringes. At a flowrate the plasma forming gas of 0.12 (a), 0.24 (b), 0.36 g/s (c); d-
f) at a flowrate of the plasma forming gas of 0.12 g/s and introducing a substance
into the zone of convergence of the jet; argon with a flowrate of 0.03 g/s (d), aerosol
(e) and suspended mixture (f) with a flowrate of the transport gas of 0.03 g/s.
182
formation of breaks in the jets and arc extinction. Arc extinction can
be explained by the insufficient reserve of the voltage of the power
sources; if the voltage is increased, a higher gas flow rate is required
to extinguish the arc. However, the electrode jets of the plasma remain
laminar even under these limiting conditions. The passage of electric
current prevent stabilisation of the current-conducting jets.
Analysis of the experimental data on the nature of plasma flow in
arc channels show that at Reynolds numbers Rein ~3000–5000, estimated
on the basis of the parameters of the cold gas at entry into the chan-
nel, the flow is still laminar. When converted to plasma temperatures
(~10 000 K), the Reynolds number is ~1000. The estimates, obtained
on the basis of the parameters of the plasma jets at an exit from the
nozzle of the two-jet plasma torches in the equation
Re in= 4G/pdm
at G = 0.4 g/s, d = 5 mm, gave Re = 300 ¸ 400, i.e. lower than critical
(here d is the nozzle diameter, µ is plasma viscosity). During the passage
of electric current, the temperature in plasma is maintained sufficiently
high as a result of Joule heat generation and the Reynolds number is
lower than critical. In the case of currentless plasma flows, rapid cooling
of these flows is the result of heat exchange with the environment re-
sults in a rapid increase of the Reynolds number due to a rapid reduction
of viscosity. This creates suitable conditions for the transition from the
laminar to turbulent flow.
Analysis of the interference patterns shows that no reversed flows
were observed at initial angles of convergence of the jets of less than
60° and a gas flow rate lower than 0.15 g/s.
It should be noted that the information interferograms, shown in
Fig.7.15, corresponding to the laminar plasma flow, can also be used
to determine the temperature and density field in examination from
different directions. Figure 7.16 shows interferograms in finite width
fringes.
For spectral analysis, plasma chemistry, the technology of
spheroidisation of microparticles and coating, it is essential to know
the special features of the processes of entry of the solutions and powders
into the plasma and their effect on the flow regime. Qualitative exami-
nation of these processes was carried out by the interferometric method
in IAB-451 equipment.
Analysis of the literature data shows that the laminar regime is suit-
able for a number of technological plasma processes. Therefore, the effect
of added substances was investigated for the laminar flow (Fig.7.15a).
This gave interferograms of the plasma flow when adding argon, aerosol
183
Plasma Diagnostics
F ig .7.16 Interferograms of plasma jets (a) and the view of interference fringes in
ig.7.16
the absence of plasma (b) in the regime of finite-width fringes.
and aerosol suspension into the area of convergence of the plasma jets.
The results are presented in Fig.7.15d-f. Figure 7.15d shows that blowing
cold gas (0.03 g/s of argon) greatly changes the nature of the flow. The
gas velocity in blowing through a pipe 0.08 cm in diameter and a gas
flow rate of 0.03 g/s is on average 35–40 m/s. Disruption of the laminar
nature of the flow takes place as a result of rapid cooling of the main
plasma flow by the cold gas. This is due to an increase of the Reynolds
number above the critical level due to a reduction of viscosity. The
184
addition of aerosols no longer leads to such large changes in the na-

ture of the flow (Fig.7.15e). In this case, pulsations evidently form as
a result of explosion-like evaporation of the droplets of the solution in
the plasma. When adding the powders (powder flow rate 0.3 g/s) with
the flow of the transport gas (0.03 g/s of argon) even a strongly pul-
sating plasma flow becomes laminar (Fig.7.15f). The Reynolds number
for powder particles with a diameter of 100 µm, estimated taking into
account the temperature of the surrounding plasma, is ~1. At this
Reynolds number, a laminar boundary layer forms at the surface of each
particle. The presence of a large number of such particles stabilises the
general flow. Evidently, this effect of laminarisation of the plasma flow
by adding finely dispersed materials is common to all plasma jet de-
vices and indicates the directions of controlling the flow regime not only
in open discharges but also in plasma reactors where the turbulence
greatly determines the losses on the walls.
‘Laminar’ plasma torch

The flow regime of a plasma jet of an ‘laminar’ plasma torch1 was in-
vestigated in IAB-451 shadow equipment by the slit and filament method.
The slit and the filament were placed parallel to the axis of the plasma
jet. The slit was 0.07 mm wide, with the filament 0.08 mm thick. The
shadow pattern of the plasma jet was recorded from a SVDSh1000
mercury lamp through a ZhS-18 light filter on RF-3 photofilm, expo-
sure time 0.002 s. Typical photographs for different gas flow rates are
shown in Fig.7.17. Each photograph shows two sections of the jet greatly
a b c d
F ig .7.17 Shadow patterns of the plasma jet at a gas flowrate of 0.09 (a), 0.18 (b),
ig.7.17
0.36 (c) and 0.72 g/s (d).
185
Plasma Diagnostics
differing in the type of flow. Interference fringes appear in the section

at the base of the jet. These fringes form as a result of diffraction, at
the filament, of the light passed through the plasma jet. The presence
of stable interference fringes, clearly observed by visual examination,
indicates that the flow is laminar. Consequently, at an argon flow rate
of 0.09 g/s, the plasma jet is laminar almost along its entire length, and
at 0.18 g/s the laminar flow is clearly visible at a distance of 14 mm,
i.e. along the entire bright part of the jet. At a flow rate of 0.36 g/s
there is no laminar section, although an essential part of the plasma
jet can again be laminar as a result of high viscosity.
Welding arc
Investigations were carried out into a welding arc in Ar and CO 2 for
currents I = 50÷90 A, a flow rate of the plasma gas of 0.1–0.6 g/s,
at different geometry and configuration of the nozzles and different dis-
tances from the outlet of the nozzle to the component (anode). Typi-
cal interferograms are presented in Fig.7.18. Similar interferograms were
a
F ig .7.18 Interferograms of a welding arc in argon at a current of 60 A, distance

ig.7.18
from the outlet of the nozzle to the component 6 mm, flowrate of the plasma forming
gas 0.1 (a), 0.25 g/s (b).
186
G, mg/s G, mg/s
a b
F ig .7.19 Effect of the flowrate of the plasma forming gas, current intensity and
ig.7.19
distance H from the component to the nozzle on the boundary of transition of the
flow to the turbulent regime at a nozzle diameter of 3 mm. a) I = 60 A, b) H = 5
mm, I is the turbulent, II the laminar region; 1) argon, 2) CO 2 .
obtained at all examined conditions. The photographs show clearly the

laminar and turbulent flows. These investigations were used to deter-
mine the boundary of transition from the laminar to turbulent flow for
different arc parameters (Fig.7.19). The results of these investigations
were used as a basis for recommendations for selecting the optimum
parameters of the welding conditions.
Alternating current arc

Figure 7.20 shows interferograms of an ac arc (current 10 A, carbon
electrode) in infinite width fringes. The dynamic interferograms were
recorded using an SKS-1 camera. At the same time, time marks from
the generator of rectangular pulses were recorded on the film. The signal
from the generator was also fed to the input of N-115 loop oscilloscope
where arc current oscillograms were also recorded. This method makes
it possible to identify each frame of the interferogram with the arc
burning phase. Current oscillograms show that the arc burns for 5 ms
and the remaining 15 ms is used up by the current break. This is con-
firmed by visual examination of the interferograms. The presence of axial
symmetry enables the interferograms to be processed by a corresponding
procedure (see section 7.1) and obtain quantitative data on the plasma
arc temperature. At the initial moment of arc ignition, the frame 1 shows
the region of strong turbulence between the electrodes. The tempera-
ture at this moment increases rapidly. In the following frame 2 (τ ~ 0.75
ms), the fringes start to assume the form of concentric rings, their number
increases and the temperature at the axis is close to 1000 K. In the fol-
lowing frame 3 (after 0.75 ms), the number of interference fringes in-
creases and the temperature in the arc increases to 4000 K. The size
of the arc cloud (frame 4) reaches 20 mm with the temperature at the
axis being 6500 K. This is followed by a current break and the tem-
perature of the axis slowly decreases to 3000–1000 K and the arc cloud
187
Plasma Diagnostics
188
F ig .7.20 Interferograms of ac arc with half-wave current.

ig.7.20
starts to travel upwards (frames 5 and 6). Examination of the

interferograms shows that the radial velocity of the front of the ther-
mal wave in arcing reaches ~1.8 m/s, the rate of convection in the current
break is ~0.2–0.3 m/s.
7.6 Temperature measurements by the interferometric

method
A two-beam diffraction interferometer, constructed on the basis of the
shadow equipment, was used to measure the temperature distribution
of the argon plasma in the jet of the ‘laminar’ plasma jet. A typical
interferogram of the plasma jet, obtained during the experiments, is
shown in Fig.7.21. The procedure of temperature measurements in the
plasma jet by the interferometric method is reduced to determining the
radial distribution of the refractive index n (r), followed by the tran-
sition to temperature T (r) using the well-known dependence n (T).
Processing the interferograms

Plasma jets are examined by lateral examination with interference fringes
of finite width oriented normal to the axis of symmetry of the jet. In
a general case, the shifts of the fringes ϕ(x) are proportional to the
variation of the optical path in the plasma in relation to the surrounding
medium. For the axisymmetric plasma with non-uniformly distributed
parameters, the local values of the refractive index n (r) were deter-
mined by the integral Able equation written on the basis of the results
of lateral examination of the shift of the fringes at a wavelength of
F ig .7.21 Interferogram of a jet.

ig.7.21
189
Plasma Diagnostics
λ = 6328 Å:
af af z af
R
m x 2 rdr
j x = = n r - nR ,
l l r - x2
2
x
where R is the radius of the plasma jet, n R (r) is the refractive index
of the medium surrounding the jet. To determine n (r), we used the trans-
formation of the Able integral (7.9)
af z af
R
l dj x dx
n r = nR - . (7.26)
p dx x2 - r2
r
The accuracy of determination of the refractive index n (r) from the

data on the shift of the fringes j(x) depends on observation in the ex-
periment of the expected axial symmetry and also the accuracy of
measuring the dimensions of the zone of perturbation of the jet R and
the refractive index n R of the unperturbed region.
The deflection of the fringes j can be measured from the coordi-
nates of the shift for the whole or half numbers of the fringes (posi-
tion of the maxima and minima of the interference field). For this pur-
pose, in the examined section AA (Fig.7.22) we draw a line and measure
the co-ordinates of the points of intersection of the line with the inter-
ference fringes. In this method, the experimentator obtains limited in-
formation on j(x) because the number of experimental points is small
and the error in determining the position of at least one fringe leads
to large errors in n (r).
The shift of the fringes was determined in the form j(x) = L(x)/D,
where L (x) and D is respectively the linear deflection of the fringe from
the initial position and the distance between the light (or dark) inter-
ference fringes. This method of construction of j(x) makes it possible
F ig .7.22 Processing of interferograms.

ig.7.22
190
to obtain a considerably larger number of experimental points (their

number depends only on the resolution power of the interferometer) and,
consequently, use the available information more efficiently. However,
it should be remembered that for real objects (arcs, jets, flames) process-
ing must be carried out taking into account the axial variation of the
parameters of the object. In addition, the accuracy of the measurement
of j(x) depends on the non-ideal nature of the specific circuit of the
interferometer causing apparatus distortions.
Taking apparatus distortions into account

In an ideal device, the interference fringes should be strictly straight
and mutually parallel, but production defects result in distortion of the
interference pattern in an actual device. Apparatus distortions were taken
into account by counting L (x) from the position of the corresponding
initial interference fringes and not from the horizontal straight line normal
to the jet axis. For this purpose, we measured the co-ordinates of the
initial interference fringes in relation to the basic straight line on the
interference patterns obtained without plasma. Measurements of the
interferograms were taken on an x-y comparator with an accuracy of
0.01 mm.
Figure 7.23 shows the effect of apparatus distortions on the deter-
mination of the deflection of the interference fringe in the section
z = 25 mm. In this case, ignorance of the apparatus distortions results
in systematically too high values of L (x) for the plasma axis jet (by
~10%). If necessary, a correction can be made not only for the deviation
of L but also for the co-ordinate x (taking appropriate measures). When
distortions are taken into account, the accuracy and reliability of meas-
urements increases.
F ig .7.23 Diagram of taking into account apparatus distortions. 1) measured shift

ig.7.23
of the fringes, 2) shift of the fringes with apparatus distortions taken into account
3) magnitude of distortions.
191
Plasma Diagnostics
Taking axial gradients into account

The correction for the axial gradients was introduced owing to the fact
that in measuring L (x) we determined not the shift of the phase of the
wave front of the given fixed section but of the section positioned at
a distance of L (x) from it. For a flow with a cylindrical symmetry, these
corrections are not required because the interference fringes, representing
a boundary of the same delay of the front of the light wave, are identical
in any section. For the real plasma jet, with steep axial power of density
gradients, the temperature error may be 500–1500 K if appropriate cor-
rections are not made.
To take axial gradients into account, it is necessary to reduce the
deviation of the interference fringes to the fixed section. The follow-
ing procedure is used. In the region of the plasma jet containing the
relevant section the deviations of 3–4 adjacent interference fringes are
checked. Subsequently, the graphs of deviation L (x) of fringes is con-
structed for the fixed values of the x co-ordinate as a function of the
distance z from the outlet of the plasma nozzle. The corresponding points
for the adjacent fringes are connected by a straight line from which we
can determine the deviation at a given fixed value x in the examined
cross section of the jet (for a cylindrically symmetric source this curve
degenerates into a horizontal straight line). Figure 7.24 shows the di-
rectly measured distribution of the deviation of the interference fringe
at the deviation reduced on the basis of the described procedure to a
single section. The graph reflects indirectly the processes of cooling the
central parts of the jet and of heating the peripheral parts of the jet
and the associated corrections.
Efficient processing and consideration of even small axial gradients
are essential because the resultant curves represent integral quantities
with respect to the examination beam, and the accuracy of subsequent
F ig .7.24 Deviation of the interference fringe taking into account (1) and disregarding
ig.7.24
(2) axial gradients.
192
determination of the local data is especially sensitive to the deforma-

tion of the initial integral contour.
The distribution L (x) was constructed from 80–100 points which
corresponds to a measurement range of »0.3 mm and is consistent with
the resolving power of the shadow devices of ~10 lines/mm. Subse-
quently, equation (7.26) was used to determine z(r) in the form
af af z af
R
l dL x dx
z r = n r - nR = - (7.27)
pD dx x - r2
2
r
Numerical integration in (7.27) was carried out by dividing the ex-

perimental curve L (x) into 40 zones. Statistical processing of the meas-
urements was carried out by averaging the data for z(r). To obtain re-
liable results and evaluate random errors of the method, it was necessary
to process 20 interferograms. The resultant values of z (x) were used
to determine the distribution of the refractive index in different sections
of the jet.
Refractive index of the environment

The value of n R was determined by means of mutually supplementing
methods: analytical and experimental.
The analytical method is based on the functional dependence of re-
fraction of air on pressure, temperature and moisture content
an - 1f = an - 1f
R 0
273 P
T 760
-
4.1 × 10 -5 m
760
where P, T is the partial pressure and temperature of air, n is the partial

pressure of water vapour, (n – 1) 0 is the refraction of air under nor-
mal conditions. The error of calculations of nR by this method is » 0.5%,
but the method is applicable only in cases in which the nature and
magnitude of deviation of the composition of the atmosphere from the
normal conditions are known.
In the atmosphere with unknown impurities, the value of n R can be
determined with high accuracy by the experimental method based on
comparing the values with the refractive index of the reference gas
(helium, hydrogen). Experiments were carried out on a two-beam dif-
fraction interferometer where a gas cuvette with transparent plane-parallel
glass sheets was placed in one of the arms of the interferometer. The
required refractive index n R = n e + kl/l, where n e is the refractive in-
193
Plasma Diagnostics
dex of the reference gas at the experiment temperature and pressure,

l is the length of the cuvette, i.e. the path travelled by the light beam
in the reference gas, k is the number of fringes by which the interfer-
ence pattern was displaced when letting the reference gas into the cuvette,
λ is the wavelength of probing radiation. According to the estimates,
the error or measurement is 1.5% at a cuvette length of 5 cm and 0.9%
and l = 20 cm.
Similar measurements of nR were taken after each experiment series
with a plasma jet. The resultant absolute values of the refraction of the
atmosphere were in good agreement with the analytical determination
data. It may therefore be concluded that anomalous impurities were not
present during the experiment. The analytical method is more convenient,
but in single measurements or if only insufficient statistical data are
available, it can result in large errors of measuring the plasma tempera-
ture.
From the resultant average distributions of the refractive index of
argon plasma in the jet of a laminar plasma torch, we determine the
distribution of plasma temperature T(r, z) using the standard depend-
ence of plasma refraction on plasma temperature. The refractive index
is an additive quantity and for a mixture of particles it is expressed
as the sum of components forming this mixture:
an - 1f = å dn - 1i = å K N ,
j j j
j j
where K j and N j is the specific refraction and concentration of type j

particles, respectively (atoms, ions, electrons). We used the following
values of specific refraction: K a = 1.042×10 –23 , K i = 0.715×10 –23 ,
K e = –1.798×10 –22. The plasma composition was computed using the
standard procedure for thermally equilibrium argon plasma at the char-
acteristic atmospheric pressure in the experiment of P = 0.9 atm. The
calculated values of n (T) were used to plot the distribution T (r, z)
and estimate the experiment error.
Figure 7.25 shows the radial temperature distribution in the section
z = 5 mm (from the outlet of the plasma torch). Vertical sections
indicate the measurement errors. The rms error of measuring the tem-
perature from a series of 20 measurements is ~5% for the entire tem-
perature distribution. However, the error of a single measurement reaches
20%. Figure 7.25 also shows the distribution T (r) obtained by the
spectroscopic method.
Since plasma refraction at high temperatures is caused in most cases
194
T, 10 3 K
interferometer
spectrometer
r, mm
F ig .7.25 Temperature distribution in a 5 mm cross-section of the plasma jet of argon.

ig.7.25
by the contribution of the electrons, in the temperature range 13000-

15000 deg these two methods determine almost the same quantity –
electron temperature. Therefore, in principle, the spectral data can be
used to determine n R by the normalisation method:
af
nR = n T = T0 -z 0 .
Combined with the spectral data (or independently), interferometric

measurements make it possible to construct the temperature distribu-
tion in the plasma jet from the maximum temperature at the axis to room
temperature at the periphery.
195
Plasma Diagnostics
Chapter 8
DIAGNOSTICS OF PLASMA FLOWS WITH A

DISPERSED PHASE
8.1 Introduction
Heterophase plasma flows (in particular, those with a dispersed phase)
are found most widely in plasma and plasma chemical technologies. The
examination of the characteristics of these flows is important in process-
ing powders of metals and non-metallic materials, processing coal and
other carbon-containing solid substances, waste from various produc-
tions, in plasma chemical production of commercial carbon (soot), pig-
ment titania, ultradispersed powders of a large number of metals and
inorganic materials, etc. 1
The two-phase plasma system examined up to now can be classified
in two groups. The first group includes the processes in which the second
phase is the initial one and is to be processed to the target product under
the plasma flow conditions. The second group includes systems (ini-
tially homophase) in which a phase transition takes place under the
plasma flow conditions. From the viewpoint of diagnostics this divi-
sion is principally of secondary importance although the examination
methods used can slightly differ.
It is well known that tasks of the diagnostics of homophase plasma
flows include the measurement of the concentration of atoms and mol-
ecules, determination of the form of the distribution of energy between
these particles (temperature in quasi-equilibrium plasma) and measure-
ment of the flow velocity. Plasma flows are usually systems open in
the thermodynamic sense and they contain gradients (often steep) of the
concentration and temperature of the particles and the flow velocity.
In the majority of cases interesting for practice the examined flows are
turbulent. Because of these special features it is necessary to measure
local values of these parameters and also ensure a sufficiently high reso-
lution power in measurements of turbulent pulsations of the concentration,
temperature and velocity.
196
Plasma of Plasma Flows with a Dispersed Phase
When examining two-phase plasma flows, in addition to these tasks

it is also necessary to solve a number of other problems associated with
the appearance (or consumption) of the second phase in the flow. They
include the measurement of the size of the second phase particles, their
concentration, velocity and temperature. Until recently, the main method
of examining the two-phase flows was the method of taking samples
using probes of various type. For many low-temperature systems, this
method gives fully satisfactory results even at present. However, when
examining high-temperature (2.5–10 kK) plasma two-phase flows us-
ing probes, the results of probe measurements of the parameter of the
second phase may be unreliable. This is associated mainly with the fact
that the processes of condensation from the gas phase taking place in
the channel of the probe in cooling the sample may greatly distort the
initial size distribution of the particles (SDP). Distortion of SDP also
forms as a result of the losses of particles in transport lines of the probe
(these losses depend on the particle size). Finally, under normal steep
radial gradients of the parameter of the plasma flow it is difficult to
ensure the conditions of isokinetic sampling of second phase particles
from the flow using finite size probes.2 This also distorts the initial SDP.
Various optical methods are at present more promising for examining
two-phase plasma flows. These methods are of two types. The first type
includes the methods based on obtaining images of the individual particles
or of their ensembles. The methods of the second type are based on meas-
uring the characteristics of the light scattered by an individual parti-
cle or an ensemble of particles.
Measurement of the size of single particles in a flow without counting
the number of particles of different sizes (i.e. without size classifica-
tion of the particles) is of no significance for practice because it does
not make it possible to determine the most often required mean tem-
perature of the examined high-temperature aerosol. Therefore, one of
the main types of diagnostics of two-phase plasma flows is the deter-
mination of SDP. In most cases, when examining these flows it is re-
quired to determine the kinetic characteristics of the system, especially
the SDP kinetics. It is necessary to measure SDP at different times or
in individual points of the flow which appears to rotate in space the
corresponding time processes of formation of SDP. To relate the spa-
tial co-ordinate of the flow and time it is necessary to know the ve-
locity of the particles. It is well known that particles not larger than
1–2 µm are dragged away almost completely by the plasma flow, whereas
as the particle size increases ‘slipping’ becomes more extensive. It is
important to measure the velocity of distribution of the particles (VDP)
correlated with the size distribution of the particles.
197
Plasma Diagnostics
The currently available optical methods combine laser Doppler ve-

locity measurement devices (LDVMD) and devices for analysis of the
amplitude and duration of pulses of the light scattered by particles (see,
for example, Ref.3 and 4). These methods are unsuitable for examin-
ing the kinetics of SDP under the conditions of the initially homophase
plasma flows with a phase transition (at least, at the most interesting
initial stage of nucleation of a new phase). This is due to the fact that
the dimensions of nuclei of the phase are distributed in this case in the
so-called Rayleigh region restricted by the condition pd/l Ù 0.3, where
d is the size of the particles scattering the light, l is the wavelength
of probing radiation.
The restriction in this and of a number of other methods of meas-
uring the SDP as regards the range of possible dimensions is not the
only one. Almost all methods of light scattering are efficient under the
conditions in which the concentration of scattering particles does not
exceed some threshold value starting from which it is necessary to take
into account secondary scattering of probing radiation. Optical meth-
ods require relatively complicated equipment and equipment itself is
difficult to adjust. These methods do not make it possible to determine
simultaneously the chemical composition of the reacting plasma flow.
All these factors indicate that it is necessary to combine efficiently the
optical and probe methods of diagnostics of two-phase plasma flows.
The extent of perturbations caused by the probes in the flow (these
perturbations are often the main obstacle in the application of the probes)
is quite easy to control using corresponding optical methods. At the same
time, taking samples with probes makes it possible to control the re-
sults of measurement by the optical method using the methods of mass
concentration of the second phase particles and enables information to
be obtained on the chemical composition of the flow.
Analysis of the problem of mutually correlated measurement of the
size and velocity distribution of the particles transported by the plasma
flow and, as shown later, also of the temperature distribution shows that
in many cases the solution of this problem can be made easier by the
use of the methods of mathematical modelling for describing the move-
ment and heat exchange of the particles and also the kinetics of chemical
reactions in the particle–plasma flow system. The simplified models of
movement and heating of the particles in the plasma flow without chemi-
cal reactions were examined in Ref.5–7. Even this very simplified ap-
proach yields conclusions that may prove to be useful from the view-
point of diagnostics. For example, as a result of losses through radiation
the particle in plasma cannot, under certain conditions, be heated above
a specific temperature which depends on the temperature and compo-
198
sition of the plasma. 7 Radiation effects are important for the relatively
large particles whose material is characterised by a high density (black-
ness) coefficient and evaporation (sublimation) temperature, and also
under the conditions with a small difference of the temperature of the
plasma and particle surface. The mathematical model makes it possi-
ble to describe, on the basis of a relatively small volume of the experi-
mental data, the kinetics of heating particles as a whole (and not only
of its surface) and physical–chemical processes taking place in the plasma
volume and on the particle surface. 8 It is evident that this information
cannot be obtained using only contactless diagnostics methods.
8.2 Methods of measuring particle size

The main requirements on the method of measuring the size of dispersed
particles in the plasma flows are usually reduced to the following:
contactless nature, localisation, real scale time, the range of measured
dimensions 1–10 µm, only weak dependence of the measurement results
on the particle form and their relative refractive index.
We shall examine the currently available laser–optical methods of
measuring the particle size and evaluate their suitability for the diag-
nostics of plasma jets with a dispersed phase in the context of the re-
quirements described previously. Firstly, it is a group of methods that
uses different parameters of the output signal of the LDVMD: visibility,
phase, amplitude.
Visibility of the LDVMD signal (depth of modulation of the
photoflux) depends on the particle diameter and the period of the in-
terference field in the measured volume. 9 However, in order to use this
dependence to determine the particle size it is necessary to satisfy a
number of requirements 10 which in practice is possible only in a small
number of cases. In addition, the method has shortcomings which com-
plicate its use for the diagnostics of plasma flows: dependence on the
measured results on the refractive index of the particles (in the case
of extra-axial reception of the scattered radiation); sensitivity of the
method to mutual coherence, polarisation and relative intensity of the
laser probing beams; insufficient dynamic range (d max/d min < 10, where
d is the particle size); very stringent requirements on the quality of setting
the optical system.
The phase method of determining the particle size is based on meas-
uring the phase difference between LDVMD signals from two or more
photodetectors.11 The application of this method for a number of tested
objects and non-plasma flows gave the following results. The method
is characterised by a linear relationship between the measured phase
shift and the particle size; its dynamic range d max/d min ¾ 40 without re-
199
Plasma Diagnostics
adjusting the optical system; the range of the measured parameters 1–

10 µm; the size and velocity of particles can be measured simultane-
ously; only slight sensitivity to optical perturbations, weakening of the
probing beams and detuning of the optical system; relatively high spatial
resolution. The LDVMD manufactured by Dantec company 12 is fitted
with a two-channel processor of the counting type which enables the
Doppler period and phase shift to be measured; the range of the measured
parameter is 1–10 µm; the error of determination of the dimension ±3%;
the range of the measured Doppler frequency is 2–8 MHz; the maxi-
mum measured velocity 250 m/s; the error of velocity determination ±2%;
maximum measured particle concentration 103 mm –3; the error of meas-
uring particle concentration ±30%.
Experience with the application of this diagnostic apparatus to plasma
objects is not available but the following can be noted: the device has
an insufficiently high upper limit of measured velocity; the LDVMD
signal in the diagnostics of plasma flows is usually measured on the
background of intensive noise, and the possibility of reliable analysis
of the signal using a processor of the counting type is problematic (analy-
sis of LDVMD signals in the diagnostics of plasma flows is usually
carried out in spectrum analysers and photon correlators). 10,13
The absolute intensity of radiation scattered on a spherical particle
is proportional to the square of its diameter and can be calculated from
the equations of the Mi series. 14 Consequently, the amplitude values of
the base component of the LDVMD signals 15 can be used to determine
the particle size by calibrating the measuring system in advance using
particles with the known optical properties and the dimensions.16–24 The
velocity of the particles is measured at the same time using standard
methods of laser Doppler anemometry.
The same principle of determination of the particle size from the in-
tensity of radiation scattered on the particles is used in the so-called
laser particle counters. 20,21 However, the Gauss distribution of the in-
tensity and the indeterminacy of the particle path in the measuring volume
cause that the measurements are ambiguous: a small particle in the centre
of the measured volume can give a signal of the same amplitude as a
large particle at the periphery of the volume. There are methods of
eliminating this ambiguity that can be classified in two groups.
They include methods 17–21 in which the size distribution of particles
is determined from the distribution of the amplitudes of the signals of
a photodetector using a corresponding mathematical procedure for cal-
culating the probability of intersection by the particle of specific points
of the measuring volume for the known form of distribution of the in-
tensity of the probing beam in this volume. In this case, it is not possible
200
to take direct and simultaneous measurements of the particle size and

velocity. The SDP can be determined only from a large number of
recorded signals by computer processing. In addition, it is necessary
to make several assumptions: the form of the SDP should be specified,
the velocity of the particles in the measuring volume should be the same
irrespective of their dimensions, etc. The error of measurements depends
to a large extent on the accuracy of statistical processing and the ac-
curacy of computer calculations. The range of the measured dimensions
varies from fractions of a micron to 100 µm at d max/d min » 10 ¸15 and
an error of ±10–15%. A shortcoming of the methods in the examined
group is the complicated form of the algorithms of mathematical process-
ing and the corresponding programmes. 22
In the methods of the second group, a measuring volume with the
known and homogeneous illumination is formed within the limits of the
examined object. This is carried out by stopping down the probing beams,
using additional detectors with matching circuits and other measures
for separating the given region. The method proposed in Ref.23 and 24
is the simplest and most reliable. In the given optical circuit two beams
of identical dimensions intersect in the centre of a large laser beam of
a different colour or polarisation, forming a Doppler measuring vol-
ume which also represents the region with homogeneous illumination
in the larger beam. The signal of the LDVMP from ‘narrow’ beams is
used to measure the particle velocity, and the size of the particles is
determined from the amplitude of the signal from ‘large’ beam. Theo-
retical investigation of the possibilities of the method and its experi-
mental verification on a number of modelling objects have shown that
it can be used efficiently for simultaneous measurements of the parti-
cle size and velocity, and the error of determination of dimensions is
±10%, with a dynamic range of d max/d min » 30.
Disadvantages of the methods of measuring the particle size on the
basis of the intensity of scattered radiation include the dependence of
the accuracy of measurements on the level of background radiation of
heated gas and particles, fluctuations of the relative refractive index
of the particles, absorption of radiation in components of the optical
system, absorption and multiple scattering of radiation and the parti-
cles outside the measuring volume in operation in dense flows, and also
from the level of the noise of the laser, photodetector and radioelectronic
equipment.
Regardless of these shortcomings, the method of determining the par-
ticle size from the intensity of the radiation scattered by them is used
widely for measuring the SDP in the flows of heated gases of various
types. 18,20,21,23,24 In Ref.19, the amplitude of the base component of the
201
Plasma Diagnostics
LDVMP signal was used to determine the size distribution of Al 2O 3

particles in the plasma jet conditions.
The small-angle scattering method25–31 makes it possible to plot the
SDP under the conditions of two-phase flows in the particle size range
3–150 µm. Two approaches can be used here. In the first appr-
oach, 25–28 the SDP is determined from the indicatrix of scattering of
probing light at particles in the small-angle range solving the incorrect
problem of light scattering. There are various methods of solving the
inverse problem but they all lead to a large error (±10–15%), especially
at relatively narrow SDPs.
The second approach 29,30 is based on the method of measuring the
mean (with respect to the beam of sight) particle size in an ensemble
on the basis of the integral characteristics of scattered light without solv-
ing the incorrect inverse problem. An interesting variant of the method
of estimating the size of individual particles was proposed in Ref.31
where the integral characteristics of the radiation, scattered by each
particle flying through a flat focused beam, was measured under small
angles. Scattered radiation is recorded simultaneously through slit and
sector diaphragms. The ratio of the amplitude of the corresponding
signals at the output of the photodetectors is linked unambiguously with
the particle size. Knowing the parameters of the receiving optical system
and the dimension of the diaphragm, a calibrating curve can be plot-
ted from the corresponding functional dependence. The range of the
measured dimensions is 3–350 µm. In testing with lycopodium parti-
cles, the error of the method was ±2%.
An important advantage of the small angle method is its insensitivity
to the refractive index of the particle material and slight sensitivity to
the shape of these particles. It is promising to examine the possibil-
ity of combining the local small angle method 31 with a time-of-flight
anemometer. It is also important to note the original and relatively simple
method of measuring the dimensions of glowing particles 31 (see below
the examination of the problem of measuring the surface temperature
of particles). Here the image of the glowing particle intersects consecu-
tively (during displacement of the particle itself) two slit diaphragms.
The width of the first diaphragm is selected smaller than and that of
the second one larger than the particle size. The ratio of the intensities
of the light passed through these diaphragms is linked unambiguously
with the particle size.
Optical methods of diagnostics of two-phase plasma flows in the
Rayleigh region. The recently developed photon correlation methods33,34
make it possible to measure quite reliably the mean particle size in a
laminar flow without taking additional measurements of the light ab-
sorption factor. Until recently, these additional measurements until re-
202
cently the only methods (combined with measurements of the intensity

of scattered light) of estimating the mean size of Rayleigh particles.35,36
However, the photon correlation methods have not as yet been used in
the diagnostics of two-phase plasma flows. Evidently, these methods are
promising for the diagnostics of laminar plasma flows with ultrafine
particles.
When using certain methods of diagnostics of two-phase systems based
on measuring the light scattering, it is usually necessary to know the
particle shape and the refractive index of their material. For small and,
in particular, ultrafine particles, the refractive index is a relatively
indeterminate value. In the formation of particles as a result of plasma
chemical reactions, it is often necessary to face the situation in which
precursors of the solid phase particles are represented by ultrafine drop-
lets of an almost unknown chemical composition. The problem of reliable
determination of the shape and refractive index of the particles has not
as yet been completely solved. It is possible that the concepts proposed
in Ref.37,38 will help the development of appropriate methods. Spherical
particles were examined in Ref.37. It is well known that when illumi-
nating these particles with laser radiation, scattered light components
are not found in the plane normal to the vector of polarisation of the
probing beam. If these components are present, their magnitude is used
to evaluate the degree of deviation of the shape of the particles from
spherical.
8.3 Measuring the flow and particle velocities

The plasma flow velocity is measured by the tracking particle method
whose velocity is determined either by means of LDVMP 9,15 or by the
time-of-flight method. Experience with the use of LDVMP for the di-
agnostics of heated gas flows, including plasma flows 9,15,39–41 shows that
the use of conventional methods of photodisplacement of direct photo
heterodyning for detecting the Doppler frequency shift in measurements
in the plasma flows is associated with considerable difficulties due to
a low signal:noise ratio caused by intense phonon radiation and also
the disruption of mutual coherence and fluctuation of the phase of laser
probing beams passing through the highly heterogeneous flow plasma.
In addition, at small dimensions and high velocities of the light-scat-
tering particles the energy of the LDVMP signal becomes so low that
the signal can be analysed only using the methods of photon correla-
tion spectroscopy. However, at velocities higher than 300 m/s and under
the conditions of highly turbulent flows, the correlation methods are
difficult to use because of difficulties in interpreting the correlation func-
tion and the need to use rapidly acting correlators (the upper limit of
203
Plasma Diagnostics
the analysed frequencies for the majority of correlators does not exceed
50 MHz).
Thus, the conventional methods of separating the Doppler frequency
shift, based on photodisplacement or direct photodetection, are either
unsuitable for examining turbulent two-phase plasma flows or are
characterised by errors that are so large that the measurement results
can be interpreted using the modelling method only. 10,41,42 These diffi-
culties can be overcome using LDVMP with direct spectral analysis 42,46
which do not require the use of very complicated electronic devices for
signal processing. In this case, the upper limit of the measured Dop-
pler frequency shifts is not restricted, i.e. there is no upper limit of meas-
ured velocities. The spectral method can also be used to determine the
direction of the velocity vector and is not sensitive to phase noise and
the polarisation of scattered light. This is especially important in
examining non-isothermal turbulent flows. LDVMP with direct spec-
tral analysis can operate in a wide dynamic range of the intensity of
light signals, including at a level of these signals where traditional
LDVMP operate only with photon correlators. Examples of using the
LDVMP with direct spectral analysis investigations of high-speed heated
gas flows were described in Ref.45–49. In Ref.50, the LDVMP of this
type was used to measure the particle velocity in a plasma jet.
When measuring the velocity of tracer particles (3 ± 1.5 µm in size)
in the plasma flow good results are obtained using the time-of-flight
method 10,13,50 when the intensity of the laser radiation scattered at such
particles is no longer sufficient for reliable operation of conventional
LDVMP. The main shortcoming of the time-of-flight anemometer,
especially in measurements of turbulent flows, is the low rate of
collection of the data limited by the small number of particles inter-
secting two measuring volumes (laser beam focused to 10–50 µm at a
distance of 400–500 µm from each other). Experience with the use of
high-speed plasma flows shows that the time-of-flight anemometer is
preferred under these conditions to the conventional LDVMP. The method
enables measurements of the velocity of up to 1000 m/s and is almost
insensitive to gradients of the flow parameters and their pulsations. 10,50
The amplitude of the signal of the time-of-flight anemometer is 2–3
orders of magnitude higher than that of the LDVMP signal.
Objective comparison of the efficiencies of the time-of-flight anemom-
eter and LDVMP with direct spectral analysis (in scanning or track-
ing regime) is not possible because there is almost no experience with
the use of the latter in the diagnostics of plasma flows.
When using anemometers of various type based on the tracer par-
ticle method for measuring the velocity in plasma jets, arc discharges
and other plasma objects characterised by high absolute values of tem-
204
perature and velocity and also steep gradients of these parameters, there
are large number of problems, with the main ones being: firstly, to meas-
ure the velocity, it is necessary to have particles that enter into heat-
and mass-exchange processes with the plasma and influence its state:
i.e., in this case, the method can no longer be regarded as a true
contactless method. 39,40 Secondly, at relatively high temperatures the
particles evaporate and the rate of evaporation increases with decreasing
particle size. Therefore, on the one hand, the particle should be rela-
tively large in order to avoid evaporation up to the moment when its
measuring volume is reached and, on the other hand, it should be as
small as possible in order to follow all pulsations of the flow veloc-
ity. In real situations where the diameter of the particle is d » 2 ¸ 5
µm, the ratio of the densities of the particle material in the surround-
ing plasma is r/r p = 10 3 ¸ 10 4 , with high gas velocities and high
accelerations, the particle velocity can differ from that of the carrier
phase (‘sliding’ effect). 10,39–41 To obtain the true profile of the average
flow velocity it is necessary to introduce corresponding corrections based
on modelling the laws of movement of the particles and results of
measurement of their dimensions. Thirdly, the non-uniform distribution
of the concentration of the particles in the flow results in additional
errors of velocity measurements. It is therefore useful to examine the
methods of measuring the velocity of the plasma flow based on the
Doppler shift of emission spectral lines 51,52 and resonance Doppler
anemometry. 53,54
The measurement of the velocity of the dispersed phase particles in
plasma flows does not differ greatly from the measurement of the ve-
locity of tracer particles and, consequently, the previously examined
methods of laser Doppler and time-of-flight anemometry can also be
used in this case. When examining the two-phase plasma flows special
attention is given to the problems associated with the processes of
interphase heat and mass transport, the non-stationary interaction of the
particles in the gas – the sliding effect of the particles and its dependence
on various factors (material density, particle size and their concentration,
etc.); interaction of the particles with each other leading to their coa-
lescence or fragmentation, i.e. variation of the initial SDP; the effect
of the dispersed phase and the structure of the carrier flow, etc. The
use of the LDVMP for measuring the velocity of the particles in such
flows requires overcoming not only technical but also system difficul-
ties. 55–58 In previous sections we have stressed the promising nature of
the LDVMP method for detailed measurements of the velocity and size
distribution of the particles in the region of Mi scattering (i.e. at pd/
l >> 1).4,10,59 It is interesting to note that for initially homophase plasma
flows in which a phase transition takes place measurements of the SDP
205
Plasma Diagnostics
by this method gives information on the form of the SDP also in the
region of small particles which form an auto-modelling size distribu-
tion, for example, soot particles, quite rapidly from the moment of their
formation. 60,61
Thus, the currently available optical methods based on measuring
the characteristics of the light scattered by the particles enable rapid
advances to be made in solving the problem of measuring correlated
size and velocity distributions of the particles. The situation with the
measurement of particle temperature in the plasma flow is less satis-
factory.
8.4 Particle temperature

The currently available methods can be used to measure the brightness
surface temperature of particles with a relatively low accuracy. The error
of measuring the true temperature of the particles is associated to a large
extent with the inaccurate information on the emisivity of the particle
material which also depends on the state of the particle surface and
surface temperature.62 The measurement error is reduced using the method
of multicolour pyrometry. 63–66 In this case it is possible to carry out
simultaneous measurements of the surface temperature, size and velocity
of the particles. 32,67–69 The total error of the method of two–colour
pyrometry is estimated to be ±10–12%. 63,67,68,70 In variants of this
method, 13,32,67–69 measurements were taken of the diameters of particles
larger than 10 µm at a surface temperature higher than 1500 K
(restrictions are caused only by the sensitivity of measuring apparatus)
with the particle concentration in the flow not greater than 104 cm–3.32,67,68
Further advances in the methods of measuring the particle surface
temperature may be associated with using wide-aperture holographic
gratings 13 to record the radiation of particles simultaneously in several
areas of the spectrum. This will enable the methods of multicolour
pyrometry to be used 62 and increase the overall sensitivity of appara-
tus and measurement accuracy.
206
Measuring the Plasma Flow Velocity by the Tracer Particle Method
Chapter 9
MEASURING THE PLASMA FLOW VELOCITY BY THE

TRACER PARTICLE METHOD
In a number of studies, the plasma flow velocity was determined by
measuring the reaction of a probe body F introduced into the flow 1-3
F = CD Sru 2 / 2, (9.1)
where r, u are the density and velocity of plasma, C D is the drag co-
efficient that depends on the form and state of the body surface, S is
the middle section of the probe body.
Sheets, discs and wires are used as probe bodies. The disadvantages
of this method include the relatively large size and inertia of such sensors.
The method was developed further by using free solid particles as probe
bodies. Since such particles are thermally and electrically insulated, a
number of sources of errors is eliminated. Particles accelerated in a
centrifuge are thrown through the examined section of the plasma flows.4
After passing through the flow the particles stick to the specially prepared
surface. If the powder is sufficiently uniform, the particles reproduce
the profile of the dynamic pressure of the plasma flow. Subsequently,
the Able integral equation is used to determine the local flow veloc-
ity. In Refs.5 and 6, steel balls 1–5 mm in diameter were thrown through
the plasma jet, and in Ref. 7 borosilicate balls 0.3–0.5 mm in diam-
eter were used. Measuring the deflection of the ball after passing through
the flow and using the Able equation, the authors calculated the local
values of the deflecting force, and the equation (9.1) was used to cal-
culate the profile of the plasma velocity with other parameters known.
The method of measuring the plasma velocity using tracer particles
was proposed in Ref.8. It is based on measuring the acceleration of
probed spherical particles introduced along the diameter of the cross
section of an axisymmetric vertical plasma flow with the known prop-
erties
207
Plasma Diagnostics
EGDG
P
D
SPR
LG-75
C
F ig .9.1 Diagram of photographic recording of particles moving in a free plasma
ig.9.1
jet. D is a screen with a point orifice, LG-75 is laser, C is the plasma jet (cross-
section), SPR is the speed recording device, P is the particle.
af
m a x -g =
pd 2
8
afaf af af
CD x r x U x - V x , (9.2)
where m, a, d is the mass, acceleration and the diameter of the sam-

ple particle; g is the free fall acceleration; v is the particle velocity in
the direction of plasma movement; r, U is the density and axial velocity
of the flow; C D is the aerodynamic drag coefficient of the particle; x
is the transverse co-ordinate (in relation to the movement of the flow).
Two optical methods were developed to apply this method, for open jets
and jets in channels, respectively. Figure 9.1 shows the diagram of pho-
tographic recording of the movement of particles in free plasma flows,
Fig.9.2 – for plasma flows in the channels.
The particle was introduced into the flow using an electrogasdynamic
gun (EGDG) (Fig.9.3) whose barrel was in the form of a medical needle
with an internal diameter of (0.1 ¸ 0.35)·10 –3 m. A voltage of ~3 kV
is applied to the gun electrodes. The pulsed charge (duration
t~2·10–6 s) between the electrodes expands the gas in the capillary tube
LG-38 laser
Photofilm
Arc
M1 D 1 L2 SFR camera
L 3P EMG D 2 EMG
F ig .9.2 Diagrams of photographic recording of a particle moving in a flow in closed
ig.9.2
channel. D 1 , D 2 are diaphragms, M 1 , M 2 are mirrors, L 1 , L 3 are lenses, O 1 , O 2 are
objectives, P is the picture frame, EMG is the electromagnetic gate.
208
F ig .9.3 Schematic section of EDGD with a device for charging particles (4). 1)
ig.9.3
discharge chamber, 2) gun barrel, 3) spherical particle, 5) a pusher for charging
particles into the barrel.
and pushes the particle from the barrel. The trajectory of the particle
in the plasma flow is recorded frame by frame using an SFR-1 cam-
era. The frames of the film record a shadow from the ball on the back-
ground of laser radiation. The pulsed discharges initiated synchronously
with operation of the SFR chamber using standard systems of synchro-
nisation and initiation of a high-voltage discharge. A single particle is
shot each time through the flow and it is then trapped and used in
repeated experiments. The velocity of rejection of the particle is selected
from the condition that it is necessary to ensure that it does not melt
during its stay in plasma. Inspection is carried out by examining the
surface of the ball under a microscope after ejection and by weighing
on an analytical balance. The particles are produced by pulsed elec-
tric melting aluminium wire in argon. Particles are carefully selected
on the basis of surface roughness, spherical form and the diameter
d = (0.15–0.3)· 10 –3 m on a type IZA-5 comparator (IZA-7) with an
error of ±1·10 –6 m. The results of filming for each particle ejection are
used to determine the trajectory of its movement Z (x) and the horizontal
velocity W (x). Acceleration is determined as the second derivative of
the particle trajectory in the direction of movement of the plasma flow:
af
a x =
d2 Z x af =
af
d 2 Z x W02
, (9.3)
2
dt dx 2
since W (x) = W 0 with an error of ±1%, W 0 is the initial velocity of

the particle.
209
Plasma Diagnostics
Particle acceleration can be computed on the basis of approximat-

ing the experimental function W 20Z (x) by exponential polynomials,
Chebyshef polynomials, modelling functions, etc. In the first case, we
used the polynomial with degree N which describes the experimental
function with an error
The unknown coefficients a k are determined by the method of least
squares from the system of algebraic equations of the N-th degree
t W Z ( x ) ≈YN ( x ) =
2
∑a x .
k =0
k
k
where n is the number of experimental points. The number of terms of

the series is determined from the minimum of the error of closure
NF n Ia = x N
axf,
åGH å x j -1 k -1
i xi JK å k
j -1 2
i W Z j = 1, N
k =1 i =1 i =1
Acceleration in accordance with equation (9.3) is determined as fol-

lows
2
1 n
N 
DN =
n−N
∑∑  ak xi − W Z ( xi ) .
i =1  k =0
k 2
A shortcoming of this method is the need to solve a system of al-

gebraic equations; this increases the error at high N.
The use of Chebyshef polynomials greatly simplifies the calculation
procedure. In this method, the experimental function W2Z (x) is expanded
into a series
N
a ( x ) ≈ YN′′ ( x ) = ∑k (k −1)a x
k =2
k
k −2
.
The polynomials P k (x) are computed from the known recurrent re-
lationships
210
Hk
Pk +1 ( x ) = ( x + βk +1 ) Pk ( x ) − Pk −1 ( x ),
H k +1
n
P0 ( x ) = 1, P1 ( x ) = x − x, Hk = ∑ P ( x ),
i =1
k
2
i
n n
1 1
x=
n
∑
i =1
xi , βk +1 =
Hk
∑ x P ( x ).
i =1
i k
2
i
Coefficients a k are determined from the equation
n n
ak = ∑W Z ( x ) P ( x ) / ∑P ( x ).
i=1
2
i k i
i=1
k
2
i
The number of the terms of the series is determined from the minimum
of the error of closure
n 2
1
Dn =
n− N
∑ Y ( x ) −W Z ( x ) .
i=1
N i
2
i
Acceleration of the particle is determined in accordance with expres-

sion
N
a ( x ) ≈ YN′′ ( x ) = ∑a P′′( x).
k =2
k k
For the polynomials P''k (x), we obtain the following recurrent rela-
tionships
Hk
Pk′′+1 ( x ) = 2 Pk′ ( x ) + ( X + β k +1 ) Pk′′( x ) − ( x ),
H k −1
Hk
Pk′+1 ( x ) = Pk ( x ) + ( x + β k +1 ) Pk′ ( x ) − Pk′−1 ( x ) , ,
H k −1
P0′ ( x ) = 0, P1′ = 1, P1′′( x ) = 0, P2′′( x ) = 2.
In addition to the described methods, a method of determining sec-
211
Plasma Diagnostics
ond derivatives using modelling functions was developed. In this case,

approximation is carried out in the form
x x′
W Z ( x) ≈ Y ( x) =
2
∫∫ a ( x)dxdx′, (9.4)
00
where a (x) is the particle acceleration to be determined. The task is

to select the distribution a (x). Taking into account the axis symme-
try of the plasma jet, we can use the three-parameter function
N
 r  r
a ( r ) = a0 1 −  1 + N  ,
 δ  δ
where d is the radius of the plasma jet. The optimum set of the param-
eters a 0, d, N and the position of the axis jet symmetry are determined
from the minimum of the error of closure
2
1 n
D= ∑ W 2Z ( xi ) − Y ( xi ) .
n i=1  
These methods were applied in a computer. To verify the process-

ing methods, numerical calculations were carried out using a test function
selected from apriori considerations in the form
 r2   r2 
a ( r ) = a0 1 − 2  exp  −5 2  , (9.5)
 δ   δ 
Random errors with a relative value of d = 3; 10% (Fig.9.4) were

superimposed on the discrete values of Y(xi) computed from (9.4) with
substitution from (9.5). It may be seen that the method of the model-
ling function gives a solution close to the accurate one even at d = 10%
whereas the Chebyshef polynomials do not ensure sufficient accuracy
at the peripheral parts of the flow, especially at high experiment er-
rors. Statistical tests using the method of the modelling function at a
random error of 3% show that the error of measuring acceleration in
the near-axial zone of the jet does not exceed 5%, and in the periph-
eral section it is ~10%.
212
a
b
F ig .9.4 Restoration of the probe function (1) by Chebyshef polynomials (a) and
ig.9.4
modelling functions (b) at s = 3 (2), 10% (3).
The resultant values of particle acceleration are then used to plot

the plasma velocity profile, using equation (9.2), at other plasma pa-
rameters (temperature, density, viscosity) and the aerodynamic drag
coefficient of the particle.
A special feature of this method is the direct determination of the
local acceleration of the particle. Consequently, it is not necessary to
use Abel's transformation and use the method for examining the non-
axisymmetric flows. Advantages of the method include the possibility
of determining the complete velocity profile in single channeling of the
particle, a sufficiently high spatial resolution and small perturbations
even in diagnostics of plasma jets with respect to a small transverse
dimension. Examination of the plasma flows using tracing by spheri-
cal particles and the relationships (9.1) was carried out in Ref.6 for
determining density, in Ref.5 to determine viscosity, in Ref.6 plasma
velocity and in Ref.9 the aerodynamic drag coefficient of the particles.
The error of the method is estimated from the equation
δ u 1  δρ δ a δ C D 
=  + + .
u 2 ρ a CD 
The error of determining plasma density is determined by the error

of temperature measurements and it can therefore be accepted that
dr/r ~5%. The error of measuring particle acceleration, determined by
statistical tests, is ~5% at the axis and ~10% at the periphery of the
plasma flow. The maximum area of estimating the head drag coefficient
is close to 20–25%. Thus, the error of the velocity measurement method
is 15–20%.
213
Plasma Diagnostics
The aerodynamic drag coefficient. Using this method, measurements

were taken of the coefficient C D for a sphere in argon plasma. Experi-
ments were carried out in the channel of a dc electric arc stabilised with
cold walls with an internal diameter of 1.5 and 3 cm, arc current of
70–190 A, the flow rate of the plasma forming argon 0.2–2.75 g/s, and
the diameter of probed particles 0.15–0.3 mm. The measured local
average values are presented in Fig.9.5 in relation to the local Reynolds
number calculated from the parameters of the non-perturbed incident
flow. The maximum error of determining C D and Re for random non-
correlated measurement errors of initial values is 20–25%. The results
of measurements of C D at different values of arc current I, gas flow
rate G and the diameter of the channel 2R and particles d were used
to plot (using the method of least squares) the approximation dependence
C D = f (Re) in a wide range of the values of the Reynolds number situ-
ated below the standard drag curve.
Figure 9.5 shows the standard drag curve for a sphere under the con-
ditions of isothermal flow of a non-viscous incompressible gas (curve
1), the results of measurements in plasma10–12 and also the expected de-
pendence C D = 16.6 Re–0.75 + 0.2 (curve 2) which approximates the ex-
perimental results obtained by the authors of this book and the data
obtained in Ref.13 for Re = 0.4÷220. The same graph shows the cal-
culated data from Ref. 14 where the authors analysed by numerical meth-
ods, the movement of aluminium particles with a diameter of (5–50)·
10 –6 m in the laminar flow of argon plasma at the atmospheric pres-
F ig .9.5 Dependence of the coefficient of aerodynamic resistance of the sphere on

ig.9.5
the Reynolds number in argon plasma. 1) standard drag curve, 2) approximating
curve, 3) data from Ref.12, 4) data from Ref.13, 5) data from Ref.14, 6) this work.
214
sure taking into account the change of the plasma properties in the
boundary layer and also showed a large decrease of the values of C D
in relation to the flow and particle temperature.
It is important to note that there are no experimental data for C D
for the plasma conditions in the range Re = 10–50. Taking into account
the monotonic form of the dependence C D = f (Re), it may be assumed
that in this range there will be no large deviations of the values of C D
from the values calculated from the equation approximating the experi-
mental data for the Reynolds numbers Re = 0.4–220.
Measuring the velocity of the laminar plasma jet at the outlet from
the plasma torch
Investigations were carried out in the laminar jet of argon plasma gen-
erated by a plasma torch 15 with the conditions: arc current I = 80 A,
arc voltage U = 80 V, the flow rate of the plasma forming argon 2.5·
10 –2 kg/s, the diameter of the outlet nozzle 3.3·10 –3 m. The plasma jet
was discharged into air. To prevent mixing of the jet with the surrounding
air, it was necessary to apply co-axial blowing of the jet with an ar-
gon flow. The tracer particles were spherical and made of aluminium,
copper and tungsten. The trajectory of the particles was recorded us-
ing the set up shown in Fig.9.1.
In calculating the plasma velocity using equation (9.2), the effect of
gravity forces g and the axial component of the particle velocity v~10
m/s was ignored (Fig.9.6). The error of determining the plasma velocity
was obtained by varying the equation (9.2) taking into account the
dependence C D = f (Re, M, γ) and equalled ~15% for the axis of the
plasma jet and ~20% for the periphery of the latter.
The error of measuring acceleration is determined by the local scatter
of the experimental data with respect to the extent of deflection of the
particles, and by the method determining the second derivative. In ad-
dition, systematic errors are also important. In this case, they are as-
sociated with the variation of the dynamic pressure of the jet ru2/2 within
the range of the sphere size.
In movement across the plasma flow, the particle is subjected to the
effect of the vertical dynamic pressure of the jet with a radial gradi-
ent. The presence of this gradient leads to lateral (in the direction of
the gradient ru 2/2) displacement of the particle (the so-called profile
effect9). The ratio of the transverse (horizontal) Fh and vertical Fv forces,
acting on the particles, determined by the authors of this book within
the limits of Newton’s approximation 16 of the interaction of a spheri-
cal particle with the gas flow, is determined by the equation
215
Plasma Diagnostics
w 2 z, m 3 /s 2 a α , km/s 2 b u, km/s c
a b c
F ig .9.6 Experimental function (a), particle acceleration (b), and plasma velocity
ig.9.6
(c), obtained in shooting particles through the diameter and three chords of the cross-
section y . y = 0(1), 0.6×10 –3 (2); 0.9×10 –3 (3); 1.2×10 –3 m (4).
Fh ¶ ln ru 2
= 013
. . (9.6)
Fv ¶r
Under the given experiment conditions this ratio is on average equal

to ~0.05. The influence of the profile effect increases in tracing along
some chords of the cross section of the jet. The presence of the gra-
dient of dynamic pressure causes ejection of the particle to the flow
periphery. The particle is deflected from the initial direction and in turn
affects the region of the flow where the dynamic pressure is 5–8% lower
than the expected pressure (determined by the direction of introduction
of the particles). To take into account the profile effect in tracing along
the diameter it is sufficient to use the local velocity of the particle
W (x) in plotting the experimental function W 2Z (x). Under the given
experimental conditions, the error dW/W, determined by the profile effect,
is, according to estimates, ~1.5%.
To determine more accurately the strength of the influence of the pro-
file effect, particles were introduced in the same section of the jet
Z = 5·10 –3 m along the chords whose distance from the diameter was
(0.6, 0.9, 12)·10 –3 m. The data were processed by the method of mod-
elling functions. Figure 9.6b, c shows the results of measurements of
acceleration and the corresponding plasma velocity. It may be seen that
the distributions obtained in introduction along the chords and the di-
ameter differ. The systematically growing displacement of the distribu-
tions, obtained in introducing the particles along different chords, cor-
responds to increase of local values ¶ln ru 2/¶r.
In the flows with high radial gradients of the parameter the scatter
216
in the direction of introduction of the particles in tracing along the chords

increases the random error of the experimental data (Fig.9.6a). The in-
fluence of the profile effect in the experiments with the tracer parti-
cles will be strongest if processing is carried out using only the finely
resultant deflection of the particle after intersection of the flow by it,
and not its total trajectory. 4
The undesirable consequences of the influence of the profile effect
can also be observed in technological processes of plasma treatment of
dispersed materials. In moving along an extended non-uniform flow, the
particles are displaced under the effect of the radial ejection force to
the periphery of the flow and stick to the walls of the reactor or the
plasma torch channel. These consequences of the influence of the profile
effect are evidently weakened by the superimposition of the displacing
force acting in the opposite direction, for example, by means of annular
blowing of an accompanying flow of cold gas into the plasma.
Measuring the velocity in plasma jets of a two-jet plasma torch

In the experiments, the anode and cathode jets at a distance Z = 5 mm
from the outlet of the nozzle of the plasma torch were traced along the
diameter with aluminium particles 150 µm in diameter. Plasma jets were
discharged into the atmosphere. The operating regime of the two-jet
plasma torch was characterised by an arc current of 105 A and the flow
rate of the working gas (argon) of 0.075 g/s per head at an initial jet
diameter 5 mm. These experiments were carried out using the meas-
ured spectroscopic radial temperature profile extrapolated at the periphery
to the temperature of the environment. Local values of the velocity and
U, m/s U, m/s
a b
F ig .9.7 Profile of the plasma velocity in electrode jets of the two-jet plasma torch.
ig.9.7
a,b) cathode and anode jet, respectively; 1,2) calculations at z = 2 and 10 mm; points
are the experimental data at z = 5 mm.
217
Plasma Diagnostics
acceleration of the particles were computed from the method described

previously. To evaluate the error of the results of measuring the velocity
comparison was made with the calculated characteristics of the flow
in individual plasma jets obtained on the basis of numerical solution
of a system of MHD equations using the two-temperature model in the
boundary layer approximation. The experimentally measured and cal-
culated profiles of the plasma flow velocity are in good agreement
(Fig.9.7).
218
Electric Probes in Non-Equilibrium Plasma
Part II
PROBE METHODS
Chapter 10
ELECTRIC PROBES IN NON-EQUILIBRIUM PLASMA
10.1 Introduction
The electric probe is a small metallic electrode placed in plasma and
used to determine its characteristics. It usually measures the volt–ampere
(probe) characteristics of the system, which includes a measuring probe,
a reference electrode and a voltage source. The reference electrode can
be represented either by one of the electrodes of the gas discharge system
or by a specially introduced reference probe. An example of such a
characteristic and typical potentials of the probe are shown in Fig.10.1.
If the system satisfies the relevant requirements, it can be used to
determine the concentration of charged plasma particles, the energy
ip
F ig .10.1 Volt–ampere characteristic of the probe i p ( U ). U is the potential difference

ig.10.1
between the reference and measuring probes; U sp is the plasma potential (space);
U f is the floating potential; U p = U – U sp is the potential of the measuring probe in
relation to plasma; I) U p > 0 is the electronic saturation current; II) U p < 0 is the
electronic current on the probe; III) U p < 0, | U p | > kT e / e is ionic saturation current.
219
Plasma Diagnostics
(velocity) distribution of the electrons, the plasma potential in the

vicinity of the measuring probe, potential pulsations and the fluxes of
charged particles. After appropriate modification, the method can also
be used to examine chemical processes in plasma, such as plasma chemi-
cal polymerisation and etching. In contrast to the majority of other di-
agnostic methods, the probe method gives local values of the plasma
characteristics.
Unfortunately, the use of probes for plasma diagnostics is faced with
a number of difficulties which restrict the advantages of the method as-
sociated with its apparent simplicity. The probe and the plasma must
satisfy a number of relatively stringent requirements, and only in this
case the results of simple electrical measurements can be linked with
the plasma parameters. These are also essential restrictions of the di-
mensions of the probes associated with plasma perturbation and ful-
filment of the requirements of probe theory, and the need to take into
account the variation of the plasma potentials, and a number of oth-
ers. In addition, the probes are introduced into the plasma using various
devices and it is therefore necessary to take into account their perturbing
effect on the plasma. The use of a method must be justified in each
specific case, otherwise both incorrect measurements and erroneous
interpretation of these measurements may take place.
The studies of the theory of electric probes and apparatus used for
the probe methods have progressed in the direction of expanding the
number of plasma objects examined using probes. They are now used
to examine plasma of dc discharges, radiowave and microwave plasma
at pressures from 10 –5 mm Hg to atmospheric pressure. The number of
studies in this direction increases and this section cannot provide a
complete pattern of the current state of investigations. We shall present
only several reports on the probes associated with the problems of their
practical application in non-equilibrium plasma and, in particular, to
the problem of measuring the electron energy distribution function
(EEDF) in the plasma. The EEDF is the most important characteris-
tic of the electronic component of equilibrium plasma which determines
the kinetics of all processes associated with the interaction of electrons
with a heavy plasma component. Details of the probe method can be
found in reviews and monographs1–8 where extensive literature lists are
also provided. In this book, the list is supplemented by the latest pub-
lications which, we believe, have made it possible to widen greatly the
potential of the probe method. Special attention is given to the experi-
mental aspects of probe diagnostics.
220
10.2 Probe measurements of the energy distribution of

electrons
The theory of current on to a metallic body (probe), placed in plasma,
which makes it possible to determine the electron temperature in the
case of a Maxwell distribution of the velocity of charged particles, was
developed by Langmuir and Mott–Smith. 9 At the arbitrary form of the
F I
z
¥
energy distribution of the electrons f a e f G f a e f = n f a eef, f a e fde = 1J in
H e 0
K 0
0
the region of the negative potentials of the probe the electron current
on the probe i e (U p) is linked with f (e) by the relationship 1
∞
2πn e
( )
ie U p = 2e
m eU ∫ (
f0 (ε) ε − eU p dε, ) (10.1)
z
where e, m is the charge and mass of the electron, e is the electron en-
ergy, U p is the potential of the probe in relation to plasma, n e is the
electron concentration. Equation (10.1) holds for isotropic EEDF and
probes with a convex surface.
An important step in the development of probe diagnostics of plasma
was the solution, by Druyvesteyn, of the problem of determination of
the EEDF in the plasma on the basis of the second derivative of the
electronic probe current i e using the probe potential U p (U p< 0) 10
ne f0 (ε ) = 2
3/ 2 1/ 2 −3/ 2
m e S p−1U 1/p 2
( ) = const U
d 2ie U p 1/ 2 ( ),
d 2ie U p
p (10.2)
dU p2 dU p2
where S p is the probe area. This equation holds for isotropic EEDF and
does not depend on the probe geometry if its surface is convex. It is
known as the Druyvesteyn equation and represents the basis of the ab-
solute majorities of studies concerned with the measurement of EEDF
in non-equilibrium plasma.
Geometrical criteria of the theory of Langmuir probes. Theory 9
assumes that in measuring current the electrons move in the near-probe
double electric layer without collisions in the potential field generated
by an external source. The probes operating under such a regime are
referred to as Langmuir probes.
The dimensions of the probe should be such that as to fulfil the as-
221
Plasma Diagnostics
sumption of the theory that enable the measured quantities to be linked

with the plasma characteristics, and that the perturbations caused by
the probe should be low. The requirement for a collisionless near-probe
layer gives a relationship between the free path of the electrons l and
the Debye screening length l D. This also defines the lower boundary
of the electron concentration in plasma at which probe measurements
can be still be taken
( ),
2
l ne >> 5 ⋅105Te N σ ( ε) (10.3)
where Te is the electron temperature, eV; N is the concentration of heavy

particles, cm –3 ; ás(e)ñ is the mean value of the collision section of the
electrons with heavy particles.
Placing the probe in plasma leads to screening, by the probe, of some
areas of the plasma in relation to others. The geometrical criterion of
the smallness of perturbations of the plasma by the probe has the form
a p << l, (10.4)
where a p is the characteristic probe dimension. Thus, the condition

l >> a p + l D should be fulfilled for Langmuir probes.
In additional to geometrical perturbations, two other causes of er-
rors are associated with the probe dimensions: the sink of electrons to
the probe, the effect of the finite resistance of plasma.
Sink of the electrons to the probe

The properties of plasma in the vicinity of the probe can change ow-
ing to the fact that the diffusion of electrons from the unperturbed plasma
does not manage to compensate their losses associated with departure
to the probe. 11,12 The VAC of the probe is distorted, and the extent of
these distortions increases with the decrease of the difference between
the potential of the probe and the plasma potential and with increas-
ing sink parameter d. The sink parameter depends on the ratio of the
characteristic dimensions of the probe and the free path and also on
probe geometry. For spherical and cylindrical probes the values of d
were determined in Ref.11 and 12, respectively
2
3  rp  3 rp lp
4 λ 
(
δspher =   / 1 + rp / λ , ) δ cylin =
4 λ
ln
2 rp
, (10.5)
222
where r p and l p are the radius and the length of the probe. Since the
free path l depends on electron energy e, then also d = d(e).
The sink of the electrons to the probe decreases the electron con-
centration calculated from the electronic saturation current, increases
the electron temperature determined from the VAC, displaces the floating
potential of the probe in relation to the plasma potential and distorts
the second derivative of the probe current with respect to the probe
potential. In particular, EEDF is not proportional to the second derivative
of the probed current with respect to the probe potential, as at
d = 0.
The effect of sink to the probe on the measurement results can be
corrected by calculations. Thus, the distorted and true EEDFs at
d << 1 are linked by the relationship 12
f m (ε ) = f (ε ) 1 − θ ( f (ε )) ,
δf (ε′) dε′
∞
θ ( f (ε )) = 2 ∫ ε′f (ε) 1 + δ (1 − ε / ε′) 3
, (10.6)
ε  
and the concentration of the mean energies of the electrons by the re-
lationship
ne = (1 + 4δ / 3 ) ne m , ε = εm (1 − δ / 2 ). (10.7)
Analysis of the work of the probe at d >> 1 show that also in this
case the probe characteristic gives EEDF but it is proportional to the
first (not second) derivative of electron current on to the probe with
respect to the probe potential (see further sections).
The effect of end resistance in the probe circuit

When determining the probe characteristic (Fig.10.1), it was assumed
that the entire voltage of the external source U is applied to the probe-
plasma transition layer. Differences in the experimental conditions of
this model may lead to an error in interpreting the probe measurements.
This discrepancy appears if the probe circuit contains an element con-
nected in series with the probe and the voltage drop on this element
decreases the voltage drop in the near-probe layer during current passage
(at a constant voltage of the external source). 7 This is indicated by ex-
amining equivalent circuits of the probe (Fig.10.2). Such elements are
either the resistance of plasma between the measuring and reference
223
Plasma Diagnostics
a b
Ep
Rm Rp
R pl
∆ U pl
R cp
E cp
c
d
a Zp
b
F Ep
Rm Cp Z pl
Cp
Cp ∆ U pl
Z cp
E cp
d c
R ~p
Zv b
a
Ep
Cp
F ig .10.2 Equivalent diagram of the probe circuit for direct ( I ) and alternating ( II )
ig.10.2
current and the equivalent circuit of the probe ( III ). R p, R cp and Z p, Z cp is the resistance
and impedances of double electrical layers at the probe and the counter probe,
respectively; E p and E cp is the emf of the layers at the probe and the counterprobe,
to their floating potentials, determined by the plasma parameters in the areas in
which they are situated; R pl and Z pl is the resistance and impedance of the plasma;
∆ U pl is the difference of the plasma potentials in the area of distribution of the probe
and the counterprobe; C p are the parasitic copacitances; R m is the measuring resistance;
U is the source of the probe voltage; Z v is the impedance of the probe as the vibrators;
Cp and RNp is the capacitance and resistance of the near-probe with respect to alternating
current; Φ is the high-impendance element (filter).
probes Rpl or the resistance of conducting lines and resistances connected

to the external part of the probe circuit. Consequently, the electron cur-
rent to the probe in the circuit with a resistor is lower than without the
resistor at the same U. The nature of the effect of this factor on the
results of probe measurements is identical with the effect of the sink
of the electrons to the probe.
Since the area of the counterprobe is usually considerably greater
than the probe area (see later), neglecting the voltage drop in the
counterprobe circuit, we can write the following equation for the probe
circuit
224
( ) ( )
ie U p =ie* U U p  ,
 (10.8)
U =Up + i (U ) R + ∆U ,
e p s (10.9)
where the asterisk denotes the values of quantities measured in the circuit
with a resistor Rs connected in series; DU is the difference of the plasma
potentials at the points of positioning the probe and the counterprobe.
Double differentiation of equation (10.8) with respect to U p with an
allowance made for (10.9) gives a relationship between the experimentally
determined second derivative of the probe current with respect to the
potential (di 2e*/dU 2) with the quantity determined by EEDF (di 2e/dU 2p):
−3
d 2ie d 2ie*  Rs 
= 2 
1 − ~*  , (10.10)
dU p dU  R p 
2
where Rp~* = (di*e / dU)–1 is the differential resistance of the probe–plasma

layer calculated from the measured probe characteristic, Rp~* = Rp~* (U).
The right-hand part of equation (10.10) contains the quantities deter-
mined from the measured VAC of the probe.
The error in determining the second derivative of the probe current
with respect to the probe potential decreases with decreasing ratio and
R s/R p~*. As a rule, the components of the external electrical circuit are
selected in such a manner as to cause only small distortions in the probe
measurements, and the main component of the Rs is the plasma resistance
R pl. The value of R pl depends on both the plasma parameters and the
geometry of the probe system. Calculations carried out for different struc-
tures of the probe show that at r p << l the error of the probe meas-
urement is proportional to the ratio rp/l, i.e. increasing rp increases both
the sink of electrons to the probe and the effect of end resistance of
the probe circuit on the VAC of the probe. 7,14 Distortions of the VAC
become greater with approach of the probe potential to the space po-
tential because R p~* is minimum at U p = 0. In Ref.13, the authors
examined the combined effect of electron sink and the resistance of the
probe circuit on the VAC of the probe in a wide range of the sink pa-
rameter and R s/R p. The nature of distortions of the VAC of the probe
is clear from Fig.10.3. In particular, the author of Ref.13 proposed a
method of computing the temperature of plasma electrons (assuming
Maxwell’s energy distribution of the electrons) in which additional re-
sistors are added specially to the probe circuit. Measurements are taken
225
Plasma Diagnostics
i''
p
, rel.units
Fig .10.3 Qualitative pattern of the effect of electron sink on the probe of the resistance
ig.10.3
connected in the probe circuit R p , on the second derivative of the electronic current
to probe with respect to the probe potential (calculations) 13 (δ = 0.1). 1) non-distorted
derivative; R p i e (0)/ kT e ×(1+δ) 2 = 0(2); 0.5(3); 1.0(4).
of the potentials Um of the maxima of the second derivative of the probe

current with respect to the probe potential at two values of Rs and, con-
sequently, T e is described by the equation
e
(U m′ − U m′′ )[ Rs′ / Rs′′] .
−1
Te = − (10.11)
k
The effect of R pl can be compensated by experiments if in plotting

the VAC the potential is counted in relation to a special tracking probe
situated in the vicinity of the measuring probe and subjected to the float-
ing potential (the current of the tracking probe is equal to zero). 15
Effect of the surface area of the counterprobe

In probe measurements in electrodeless discharges, such as radio- and
microwave discharges, it is necessary to use two-probe circuits. To obtain
a single-probe characteristic, the area of the reference probe (counter-
probe) should greatly exceed the area of the measuring probe. This is
indicated by the fact that the ionic saturation current of the counterprobe
should be considerably higher than the saturation current of the probe.
Estimates of the required ratio of the areas of the probe and the
counterprobe from the equation from Ref.1 a = S cp/S p >> (Mi/m) 1/2 are
too low owing to the fact that under the real conditions the dimensions
of the counterprobe are greater than the free path of the charged par-
ticles and the random current to the counterprobe from the plasma is
226

i e /i es
eU p /kT e
F ig .10.4 Effect of the ratio of the areas of the counterprobe and the probe on the
ig.10.4
VAC of the probe (calculated). 15 R p < λ e , λ i ; λ D/ R cp = 0.01; T i/ T e = 0; Σ α = 100(1),
200 (2), 400 (3), 10 3 (4), 10 4 (5), ∞ (b); VAC normalised the electronic saturation
current i es .
lower than that calculated disregarding this fact (Mi is the mass of ions).
Analysis of the problem carried out in Ref.16 for probes of different
forms showed that the condition applied in most cases, a > 104, should
be replaced by a new condition Sa > 10 4, where S a is obtained by mul-
tiplying the value of a by the coefficient C a ¾ 10 –2, which takes into
account the decrease of the random current to the probe and depends
on its shape.
Thus, the new condition is considerably more rigid. The effect of
Sa on the VAC of the probe and the nature of the distortion of the VAC
due to the small area of the counterprobe are clear from Fig.10.4.
The influence of all these factors leads to ‘stretching’ of the VAC
of the probe along the voltage axis, to smoothing of the VAC in the
region of the plasma potential, and becomes weaker with decreasing ratio
r p/l. The contribution of these factors to the total error of the probe
measurement depend on the probe dimensions and the plasma charac-
teristics. For example, in measurements in microwave plasma in atomic
and molecular gases with a probe with r p = 5 µm at pressures of
1–5 mm Hg, the errors associated with the sink of electrons to the probe
and with the end resistance of the plasma were similar.
The dependence of the effect of all factors on the value of r p makes
it possible to correct the results of probe measurements if experiments
with the probes of different thickness can be conducted. Extrapolating
the results of measurements to a zero thickness probe, it is possible to
reduce the error of measurements. 7
Effect of alternating fields on VAC

A large procedural error may form in probe measurements in the plasma
227
Plasma Diagnostics
generated using electromagnetic fields (rf and microwave discharges)

and also in the plasma with instabilities of different type leading to a
periodic variation of the plasma potential with time. This error is
associated with the fact that the probe–plasma layer (this layer yields
the measured VAC of the probe) is a non-linear element and the vari-
able signal on this layer which is rectified changes the measured VAC
averaged out with respect to time. 1,7,17-20
If a voltage of type
U p = U p0 + u0 cos wt , (10.12)
is applied to the near-probe layer, where Up0 and U0 is the constant volt-
age and amplitude of the alternating voltage in the near-probe layer,
then (assuming Maxwell’s distribution of the electrons), the mean elec-
tronic current on the probe is expressed by the equation
 eU p 0   eu0 
ie* = ies exp  −  I0  , (10.13)
 kTe   kTe 
where i es is the electronic saturation current, I 0 ( ) is the modified

Bessell’s function of the zero order.
Equation (10.13) shows that identical values of the electronic cur-
rent on the probe in the distorted characteristic (i*e) are obtained at higher
negative values of U p0 than in the case of the non-distorted VAC
(u 0 = 0), and the shift becomes greater with increasing U 0 (Fig.10.5).
Correspondingly, the derivatives of the probe current with respect to
the probe potential are also distorted (Fig.10.6 and 10.7). The nature
of distortions is the same under the effect on the probe of hf and
microwave signals, and the calculated results and experimental data are
in qualitative agreement.
i e /i es
eU p /kT e
F ig .10.5 Effect of alternating voltage on the near-probe layer on the VAC of the
ig.10.5
probe (calculated assuming the Maxwell EEDF). 17 eu 0 / kT e = 0(1); 5(2); 10(3).
228
eU p /kT e
F ig .10.6 Effect of alternating voltage on the near-probe layer on the second derivative
ig.10.6
of probe current averaged with respect to period (calculated). 20 eu 0 / kT e = 0(1), 2(2),
4(3), 7(4).
i''
p
, i''
e
i''
p
,
eU p /kT e
F ig .10.7 Effect of alternating field on the VAC of the probe (a) and the second
ig.10.7
derivative of the probe current with respect to probe potential (b) in microwave
discharge, in N 2 under a pressure of 1.5 mm Hg (probe radius 5 µm). 7 1) non-distorted,
2) distorted curves.
The methods of reducing the error of probe measurements, determined

by the effect of alternating fields, are clear from the equivalent diagram
of the probe circuit using alternating current (Fig.10.2). They are all
d i
associated with the decrease u0 U p = DU ~ Z p U p / d i å Z , where DU
i
~
is
i
the alternating voltage acting in the probe–reference electrode system;
Z p (U p) is the potential-dependent impedance of the probe; Z i are the
in-series connected elements of the equivalent diagram of the probe circuit
(Fig.10.2). It is important to note that U 0 depends on U p and, conse-
quently, the VAC is not just simply displaced from the voltage axis but
is also deformed.
229
Plasma Diagnostics
One of the methods of decreasing U 0 is to position the probe and

the counterprobe at equipotential (with respect to the alternating sig-
nal) points of the plasma. It should be noted that the counterprobe is
larger and if it is placed in the vicinity of the probe it may alter the
plasma parameters.
The second method is to increase the probe surface. This is accom-
panied by a reduction of its impedance
4kTe  eU 
Rp~ = 2
exp−1  − p  , (10.14)
1.5neve S p e  kTe 
2πl p ε0
Cp = ,
(
ln x / rp ) (10.15)
where ve is the mean electron velocity, Cp is the capacitance of the near-

probe layer at a floating probe potential, e0 is the permittivity of vacuum,
lp, rp are the length and radius of the cylindrical probe, x is the thickness
of the near-probe layer (x ¾ (5÷7)l D). 7 It is obvious that, with other
conditions being equal (all other components of the equivalent diagram
of the probe circuit are constant), the alternating voltage acting on the
layer decreases. It should be noted that when using this approach all
errors examined in the previous sections become larger because to reduce
them it would be necessary to reduce the probe dimensions.
The same effect can be achieved not by changing the impedance of
the probe but by increasing the impedance of the external parts of the
probe circuit. For this purpose, additional elements are added to the
circuit (in the rf range it is filters with concentrated elements, in the
microwave range – with distributed elements 7).
There are active methods of reducing U0 in which an alternating signal
with the field frequency generating plasma is introduced into the probe
circuit. The amplitude and phase of the signal are such that they com-
pensate the alternating voltage on the probe generated by plasma (see,
for example, Ref.20–22). For this purpose, the circuit shaping the re-
quired signal (and this signal is part of the voltage supplied to the dis-
charge) should contain an attenuator and a phase-shifting device.
In the last and penultimate cases, it must be taken into account that
the alternating voltage at the near-probe layer has the frequency of not
only the external energy source but also multiple frequencies. This is
associated with the fact that the discharge is a non-linear element in
which frequency transformation takes place. There are cases in which
230
the probe circuit must contain filters for both the first and second har-
monics of the external voltage.
The criterion for evaluating the efficiency of the methods of decreasing
u0 is the sensitivity of the floating potential of the probe Ufl to the value
of u 0. Equation (10.13) shows that the change of the probe potential
(self-displacement of the probe) and, correspondingly, Ufl is linked with
u 0 by the relationship
kTe  eu 
∆ U fl = − ln I 0  0  . (10.16)
e  kTe 
Thus, reaching the minimum shift of the floating potential, it is pos-

sible to minimise the effect of u0 on the results of probe measurements.
In equation (10.16), the value U fl is counted from the plasma poten-
tial (see Fig.10.1) which is not known prior to starting the probe meas-
urements. In experiments, all potentials are usually determined in re-
lation to the potentials of reference electrodes or electrodes of the gas
discharge system. Therefore, the condition (10.16) can be reformulated
for a specific type of discharge and a specific measurement system. For
example, in Ref.22 a criterion in a capacitance rf discharge was the
establishment of the maximum positive value of the contant potential
of the probe in the absence of any current in relation to an earthed elec-
trode.
Effect of ion current on EEDF

EEDF is determined from probe measurements using equations (10.1)
and (10.2) if the dependence of electron current on the probe i e on U p
at U p £ 0 is known. In the experiments, we measure not i e but the to-
tal current to the probe i p = i e + i i , where i i is the ion current to the
probe, and the contribution of the ion current to the measurement of
EEDF becomes significant at eU p ≥ (3 ÷ 4) ε . 1,7,23
The electronic component can be separated from i p by extrapolat-
ing the ion saturation current to the region of lower probe potentials.
However, this method has a number of significant shortcomings asso-
ciated with the fact that the accurate dependence of the ion saturation
current on the probe potential is not known and the extrapolation law
is not available. This may cause a large uncontrollable error in the value
of i e.
In most cases, investigations were carried out using the derivatives
of the probe potential and not current. Work in the region of high po-
tential is possible due to the fact that the dependence of ii on U p is far
231
Plasma Diagnostics
i p , rel.units
a
i''p , rel.units
b
U p, V
F ig .10.8 Restoration of the second derivative of electronic current on the probe with
ig.10.8
respect to the probe potential ( i e ") from the modelling second derivative of the probe
current ( ip") at non-monotonic EEDF. Numbers at the curves correspond to the numbers
of iterations. 23
weaker than the dependence i e (U p). Consequently, the effect of i i on

the curve of the second derivative of i p with respect to U p should be
detected at values U p higher than on the VAC of the probe.
The authors of Ref.23 described an iteration method of separating
d 2i e/dU 2p from the measured d 2i p/dU 2p using the expressions for the ion
current on the probe in limitation movement of the ions in the case of
an arbitrary form of EEDF. In the first approximation and assuming
that the measured curve d 2i p/dU 2p= d 2i e/dU 2p, we determine the values
that control the ion current, and also determine i i and its second de-
rivative. Subtracting this derivative from the measured curve, we carry
out the first correction of the measurement results. The procedure is
then repeated. Since the dependence of i i on EEDF is weak, the pro-
cedure rapidly converges. An example of application of the method of
a non-monotonic EEDF is shown in Fig.10.8.
Experimental methods of correcting the VAC of the probe for ex-
cluding i i are available. The VAC can be produced using probes of
232
i p'', rel.units
a b
U p, V
F ig .10.9 Application of two probes for widening the range of measurement of i p" in
ig.10.9
the range of high energies in cesium vapours. 24 a) 2.8×10 –3 mm Hg, 200 mA; b)
1.8×10 –2 mm Hg, 100 mA; r p = 0.025 (1), 0.25 mm (2).
different thickness and since the relative contribution of the second

derivative of the ion current with respect to the probe potential decreases
with an increase of its diameter (ion current of the flat probe does not
depend on U p), 2 measurements of d 2i p/dU 2p by the probes of different
thickness make it possible to widen the range of measurement of EEDF
towards higher electron energies (Fig.10.9). In this method, as in the
previously described method of correcting the low-energy part of the
EEDF using probes of different thickness, there are certain difficulties
in combining the curves obtained by different probes along the voltage
axis. This is caused by distortion of the VAC of the probe. The situ-
ation is made easier if the EEDF contains characteristic points, for
example, inflections at specific energies.
The second method is based on measuring d 2i p /dU 2p to the probe
potentials where this quantity completely determined by the ion current.
The VAC of the probe is obtained at the same time and is approximated
in the region of the ion current by some dependence on Up (for example,
i i ~ U np, where n is determined from the experimental curves; for ex-
ample, in the case of a thick near-probe layer (where r p/(5¸7)l D<< 1),
n ¾ 0.5). 7 The accuracy of approximation is verified by comparing the
second derivative with respect to U p of the resultant dependence for the
ion current with the measured dependence. Subsequent subtraction of
d2ii/dU2p from d2ip/dU2p makes it possible to separate d2ie/dU2p. This method
enables the EEDF to be measured in microwave plasma to electron
energies of e ¾ 10 e (Fig.10.10).
The ratio of the second derivatives of the ion and electron currents
to the probe and, consequently, the electron energy to which the EEDF
can be measured depends on the type of gas, the conditions in plasma
(which determine the selection of ion current theory) and of the EEDF
itself.
233
Plasma Diagnostics
i p'' , i ''e,f
ε, eV
F ig .10.10 Restoration of the second derivative electronic current on the probe in

ig.10.10
respect to probe potential; 1) from the measured second derivative of probe current;
2) by subtracting the ionic components. 7 Microwave discharge plasma in CO 2 at a
pressure of 1 mm Hg, ε = 1.9 eV, r p = 5 µm; 3) EEDF, 4) Maxwell's distribution
with the same mean energy.
The methods of determining EEDF from probe measurements

To determine the EEDF from probe measurements it is necessary to use
the relationships (10.1) and (10.2). The Druyvesteyn method (equation
(10.2)) is used most widely. In this method, the EEDF is linked with
the second derivative of the electrode current on the probe with respect
to the probe potential d 2i e/dU 2p = i²e, and i²e is determined using the fol-
lowing methods:
– double electrical differentiation of the VAC (modulation of VAC
by special form signals, natural noise);
– double numerical differentiation of the VAC;
– double analogue differentiation;
– double graphical differentiation.
The modulation methods of differentiating the VAC of the probe are
based on the non-linear properties of the near-probe layer and on the
associated capacity to change the form, carry out displacement, and de-
modulation of the weak alternating signal u(t) applied to it. The ex-
pansion of probe current ip[Up + u(t)] into a Taylor series in these cases
shows that different expansion harmonics contain the terms proportional
to i''p. The methods used for these cases are based on analysing vari-
ous variable components of the probe current because they are less sen-
234
sitive to the instability of the discharge than the constant component.1,4,7,25

We shall present several examples of u(t):
u1 (t ) = a1 cos ωt;
u2 (t ) = a2 (1 + b cos ω1t ) cos ω2t;
u3 (t ) = a3 (b cos ω1t + c cos ω2t );
u4 (t ) = a4 (1 + b sgn (cos ω1t ))cos ω2t; (10.17)
u5 (t ) = a5 (1 + b sgn (cos ω1t ))sgn (cos ω2t ),
where sgn (x) = {–1, x < 0;0, x = 0; 1, x > 0} is the signature func-
tion.
In modulating the signal of the type u 1(t) the second derivative of
the probe current is obtained using the second harmonic of frequency
(the second harmonic method); for the signal of type u 2(t) – the am-
plitude of the current with frequency ω1 (demodulation method); for the
signal of type u3(t) – the amplitude of the current of total or difference
frequency w 1 ± w 2 (‘wobbling’ method); at signals of type u4(t) and u5(t)
– the amplitude of current with frequency w 1 (modulation with a right-
angled signal).
The error of the modulation method is linked with the fact that the
amplitude of the measured harmonics contains not only i''p but also higher
derivatives, for example, even derivatives of probe current for the second
harmonics method. The contribution of higher derivatives becomes
smaller with increasing amplitude of the modulating signal.
The second reason for the error is the distortion of the VAC of the
probe by an alternating signal fed into the probe circuit. This has al-
ready been examined. The amplitude of the modulating signal should
be small ( eai < e ).
The error of the measured values i''p is also caused by the end re-
sistance of plasma elements of the probe circuit because this leads to
a difference between u(t) which depends on the probe potential, and to
the corresponding voltage in the near-probe layer.
In Ref.26, comparative analysis was carried out on the basis of two
parameters for the first out of four modulation signals described pre-
viously: the intensity of the signal S, proportional to i''p and the ratio
D of the terms proportional to i''p'' and i''p in the measured signal. In method
2, the maximum value of S is obtained at b = 1, and in method 3 at
b = c. Comparison was carried out at the same amplitudes (from maxi-
mum to maximum) of the variable signal Du. It was concluded that the
the methods 1 and 4 have advantages. In the first of these methods, the
235
Plasma Diagnostics
maximum value is S, in the fourth the minimum value D. An ampli-

fier makes it possible to compensate a small value of S, and the method
4 it is preferred.
The distorting effect of various methods of differentiation of the VAC
(difference between the measured i''p* and the true value) is taken into
account by utilising the formalism of the apparatus functions (by analogy
with optics) for each method. 26–28 The measured signal (for example,
the voltage at the measuring resistance) is a convolution of the real sec-
ondary derivative i''p and the normalised apparatus function of the cor-
responding method A i
( )
i′′p* = C i′′p ⋅ Ai , (10.18)
where C includes the values of measuring resistance, the gain factor,

the amplitude of the modulating signal, etc. Examples of Ai for the sig-
nals of the type (10.17) are shown in Fig.10.11. The true i''p is restored
as a result of solving the Fredholm equation of the first kind (10.18).
Advantages of this approach are the possibility of describing by the same
procedure all methods of obtaining i''p (see later), an increase of the
signal:noise ratio by increasing the amplitude of the modulating signal
and taking into account the resultant distortions of the measured results
in the apparatus function, and also the possibility of taking into account
various factors affecting the result of probe measurements (for exam-
ple, reflection of electrons from the probe surface was taken into ac-
count in Ref.29).
Without using the previously described procedure for decreasing the
distortions it is necessary to decrease u(t). This impairs the signal:noise
F ig .10.11 Normalised apparatus functions for modulated voltages of the type (10.17).
ig.10.11
The numbers of the curves correspond to the numbers of expressions in (10.17); 25
variable z links V p and V 'p .
236
ratio. In Ref.30, it was proposed to determine EEDF by the Druyvesteyn

method using the natural and minimum (for specific plasma) modulating
signal – spontaneous fluctuations of the plasma potential (noise) n(t).
The probe–plasma system is regarded as an inertialess element carry-
ing out non-linear transformation of the stationary random process
n (t) ® i p [U p + n(t)], and the Druyvesteyn equation is converted to
the form giving the relationship of the EEDF with the statistical char-
acteristic of the spontaneous functions of plasma measured in the ex-
periments. The results obtained by this method are in satisfactory agree-
ment with the data of double numerical differentiation of the VAC of
the probe.
The numerical methods of differentiating the VAC are used with suc-
cess for obtaining i''p, and ip is often approximated by polynomials. Com-
parison of the results of numerical and double electric differentiation
of the VAC showed that the agreement between the two methods is
obtained when using the polynomials with the factor n ³ 11. However,
at eU p > (2.5 ¸ 3) e the error i''p and numerical differentiation remain
large and non-monotonic features can also appear. This is associated
with the error in measuring the VAC in the region of the ionic satu-
ration current. 7
Diagnostics of non-stationary plasma and plasma with a high elec-
tron concentration is often carried out using the pulse methods of ob-
taining the VAC and differentiation using analogue devices. 7 These
measurements are based on applying a saw-like voltage to the probe
and using in-series connected differentiation terms. This also results in
substitution of differentiation with respect to U p by differentiation with
respect to time t:
arbitrary units
F ig .10.12 Oscillograms of VAC of the probe (1), i'p (2), i"p (3) in microwave discharge
ig.10.12
in argon obtained by the pulse method. 7
237
Plasma Diagnostics
d 2i p d 2i p  dU p
 di p d U p
2
= + . (10.19)
dt 2 dU p  dt
 dU p d t
2
Equation (10.19) shows that the error of this differentiation method

is due to the linear form of the saw-like voltage used (d 2U p/dt 2 ¹ 0).
In addition, wide-band electronic circuits must be used to transfer the
pulsed signal. Therefore, a shortcoming of the method is the small energy
range of measurement of EEDF ( e < 3e ) caused by the wide-band na-
ture of the differentiating amplifiers and plasma noise. Examples of the
VAC of the probe and of its first and second derivatives, produced by
the pulsed methods, are shown in Fig.10.12.
The graphical methods of differentiation are used only very seldom
because their error is very high.
The previously described formalism of the apparatus functions can
be used to compare different methods of differentiation of the VAC. In
Ref.31, this comparison was made for the modulation methods (equa-
tion (10.17) at b = c = 1), the double numerical differentiation (this
will be denoted as method 6) and the double differentiation using dif-
ferentiating amplifiers (method 7) at the same sensitivity of the meth-
ods (i.e. with selection of the corresponding values of A i in equation
(10.17)). According to the resolution power (the minimum energy in-
terval in which two monoenergetic electron distributions are permitted),
the introduced distortions and the quality of the restored EEDF (res-
toration of modelling EEDF from equation (10.8) using the regularis-
ing algorithm proposed by A.N. Tikhonov32), the examined method can
be distributed in the following sequence with decreasing efficiency: 6,
5, 4, 2, 1, 3. The method 7 is not included in this sequence because
it was the best method as regards the resolution power and the resto-
ration of the EEDF and the worst method as regards the distortion caused
by it.
When selecting the differentiation method, it is important to take into
account other factors that are not linked or not taken into account by
the apparatus functions. Some of them have already been mentioned,
for example, difficulties in obtaining high-energy parts of the EEDF by
the methods 6 and 7. In comparing the modulation methods, it is im-
portant to note that method 1 enables the first, second and higher de-
rivative of the VAC (the latter is very important for examining the form
of the EEDF) to be obtained by simple rearrangement of the frequency
of the selective recording circuit. However, stringent requirements are
imposed on the shape of the modulation signal (absence of higher har-
monics). The methods 2–5 are less sensitive to the distortion of the signal
form.
238
The EEDF is often determined using equation (10.1) by transferring

from the Volterra equation (10.1) to the Fredholm equation of the first
kind
( ) ∫ ( )
ie U p = C K ε, eU p f (ε) dε, (10.20)
0
where K (e, eU p) is the kernel of the integral equation,
(
 K ε , eU p , ε > eU p )
( )
K ε , eU p = 
e ≤ eU p
(10.21)
 0,
As in the previous equation (10.18), the problem of determining the

integrand function is incorrectly formulated. Regularisation methods are
used to solve it. 32 The regularisation method was used for the first time
for determining the EEDF using equation (10.20) in Ref.33 and it was
developed further in Ref.23 and 34. The agreement between the restored
and true EEDF depends on the method used to solve the incorrectly for-
mulated problem. For example, the regularisation method based on the
assumption of the smoothness of the initial solution gives the best agree-
ment between the restored and true EEDF in the high-energy range. 34
The search for a solution on a large number of limited monotonic func-
tions gives the best results in the low-energy range. 23
The kernel of the integral equation can take into account different
factors leading to the distortion of ip (U p) and, at the same time, makes
it possible to compensate them when restoring the EEDF. For exam-
ple, taking into account the sink of electrons to the probe 23
 ε − eU p
 , ε > eU p
(
K ε , eU p ) ( )
=  1 + δ ( ε ) 1 − eU p / ε
(10.22)

 0, ε ≤ eU p
where d (e) is the parameter of the sink and is determined by the equa-
tions of the type (10.5).
Plasma potential
To calculate EEDF and interpret probe characteristics, it is important
to know the probe potential in relation to U p, whereas the experiments
239
Plasma Diagnostics
yield the probe potential in relation to some reference electrode U and

U p= U – U sp, U sp is the plasma potential (see Fig.10.1). In accordance
with the conventional assumptions regarding the VAC of the probe, the
plasma potential is determined as the potential of the inflection point
of the VAC in transition from U p < 0 to U p > 0. 2
In the actual VAC of the probe there is no distinctive inflection point.
This is associated with the effect of the variation of the plasma potential,
the effect of a weak alternating signal for measuring i''p superimposed
on the constant displacement of the probe, contamination of the probe
surface, etc. Characteristic points on the derivative of the probe cur-
rent with respect to the probe potential are used for determination. There
are two approaches to determining U sp: U sp corresponds either to the
probe potential at which i''p is maximum, or to the potential at which
i''p passes through zero (i p is maximum). This problem was analysed in,
for example, Ref.1, 7, 13. Although views differ, in the majority of
studies U sp is determined by the condition i''p (U sp) = 0.
Probe measurement of EEDF at medium pressures

These measurements are taken using conventional Langmuir probes, i.e.
probes under the conditions where l >> rp + lD. This condition restricts
the upper value of pressures at which EEDF can be calculated from
the probe characteristics, at a pressure of 5–6 mm Hg and the EEDF
is associated with electronic current to the probe or with its second-
ary derivative with respect to the probe potential using the equations
(10.1) and (10.2).
However, it is possible to determine the EEDF from the probe meas-
urements also at higher pressures 35,36 for which
λ ε >> rp + λ D >> λ, λ p = 4 De τeff , De = vλ / 3,

(10.23)
−1
τeff = ν e + ( 2m / M ) ν el + (8 B / T ) ν r + ( hω ) / εν v + ν* = κν eff ,
where l e is the characteristic length of the electron energy loss, t eff is

the effective relaxation time of the electron energy, D e is the coefficient
of electron diffusion, n e, n el, n r, n n, n* are the characteristic frequen-
cies of collisions of electrons with energy exchange between themselves,
with energy transfer to translational, rotational, vibrational and elec-
tronic degrees of freedom of heavy particles, respectively; B is the
rotational constant, hw is the vibrational quantum, k is the mean fraction
of energy losses by the electron in a single collision, n eff is the effec-
tive frequency of electron collisions. Since k ¾ 10 –5 ¸ 10 –2 then Ök and
this determines the expansion of the range towards higher pressures or,
240
at the same pressure, the possibility of taking measurements using large-

diameter probes.
The authors of Ref.35 derived a relationship of the EEDF and elec-
tronic current on the probe at l e >> r p + l D which gives in the limit-
ing case r p + l D << l relationship for the Langmuir probe, and at
r p + l D << l e the relationships of diffusion theory.36 The equations are
identical with those obtained in Ref.11, 13 but were derived on the basis
of more general considerations. The equation for the current has the
form of (10.20) with the kernel of the type (10.22).
The main difference of the probe theory at high pressures is the re-
lationship of EEDF not with the second derivative of the electronic cur-
rent with respect to the probe potential i''p but with the first derivative
i''p
3mγ 0 δ die
(
ne f0 eU p = − ) 8πe2 U 1/p 2 dU p
, (10.24)
where 4/3 ³ g 0 > 0.71 is a parameter which depends on r p/l (g 0 = 4/3

at r p << l and g 0 = 0.71 at r p >> l), eU p = e:
rpC
δ (ε ) = ,
γ 0λ (ε ) (10.25)
where C = 1 for a spherical probe and C = ln (l/r p) for a cylindrical

probe (see equation (10.5)).
Analysis carried out in Ref.36 shows that EEDF can be computed
from i''p at d £ 1, and at d ³ 10 from i¢p.
At higher pressures or higher values or r p, there is no relationship
between the electron current to the probe and the non-perturbed
EEDF.The values obtained in the probe measurements can be used to
determine only the mean electron energy and the concentration of charged
particles (for example, using the equations in Ref.6).
Measurement of EEDF with time resolution. Examination of non-
stationary discharges faces the need to measure instantaneous EEDF,
i.e. EEDF obtained during periods shorter than the characteristic pe-
riod of variation of the plasma parameters.
For these purposes, it is possible to use pulsed methods of obtain-
ing the VAC and its derivatives where saw-like or triangular voltage
is supplied to the probe. This method has principal restrictions at the
bottom on the duration of voltage pulse t v. This is due to the fact that
241
Plasma Diagnostics
the layer of the probe–plasma spatial charge manages to rearrange

itself when the applied voltage changes. In the opposite case, the VAC
is distorted by the transition processes.7 The layer formation time is de-
termined by the flight time t i of the slowest charge carriers (ions) to
the layer with the characteristic dimension x ¾ (5 ¸ 7)l D
τi ≅ x / vi ≅ 1011 ( M i ne ) ,
1/ 2
(10.26)
which takes into account the acceleration of the ions in the prelayer;
n i = (kT e/M i) 1/2 , M i is the ion mass, kg; electron concentration, cm –3 .
For example, for Ar (M i = 6.68·10 –26 kg) at n e > 10 10 cm –3 t i < 5·
10 –6 s. At these characteristic pulse times the results of the measure-
ments are strongly influenced by the parasitic capacitance of the probe
circuit (see, for example, Fig.10.2). Time t v is restricted at the top by
the characteristic time of variation of the plasma characteristics T, i.e.
t i < t v < T. The curves obtained in Fig.10.12 were obtained during a
period of ~50 µs.
Measurements in plasma with periodically changing parameters can
be taken using methods in which the probe receives pulses of variable
amplitude under a negative potential so that the VAC can be recorded
or the probe is connected to the measuring circuit for a short period
of time. The pulses acting on the probe are synchronised by some method
with the periodic process. The time resolution is determined by the du-
ration of the pulses or switching time, with the minimum value of the
latter, as previously, restricted to τ v > τ i. To obtain i''p, we can use dif-
ferent methods. The authors of Ref.37 used the method of modulation
of the signal of type (10.17) (second equation), and in Ref.38 meas-
urements were taken by differentiation with analogue devices (τ i =
4 ÷ 40 µs).
Effect of the state of the probe surface and processes taking place
on it on probe measurements
During probe measurements, various processes can take place on the
probe surface: gas adsorption, deposition of conducting, semiconduc-
tor and dielectric films, removal of deposited coatings under the effect
of fluxes of charged and excited particles from plasma, also due to heat-
ing of the probe, etc. The sources of contamination are either chemi-
cal compounds present in the gas medium, or vapours of organic com-
pounds appearing in the discharge when using oil pumps for pumping,
and also substances dissolved from the walls of the discharge cham-
ber and from objects placed in the plasma. Consequently, the proper-
ties of the probe surface in plasma differ from those of the probe material
242
and vary with measurement time. The effect of contaminants on the VAC
of the probe is associated with the fact that the layers formed have an
impedance which should be connected in series with a probe in the
equivalent circuit in Fig.10.2, and also with the fact that the work func-
tion of the probe surface changes. 1,3,6,7
In this case, the equation for the voltage of the near-probe layer in
relation to plasma U p differs from that mentioned previously
( ) (
U p = U − i p Rl −U sp − ϕ p − ϕc. p , ) (10.27)
where R l is the resistance of the layer on the probe surface, j p, j c.p is

the work function of the surfaces of the probe and the counterprobe
taking the layers on them into account. All the considerations expressed
previously regarding the effect of a resistance connected in the probe
circuit apply to the resistance formed by the surface layers.
The variation of the properties of the probe surface shifts the VAC
along the voltage axis, causes deformation of the VAC (as a result of
a change in the properties of coatings during the measurement time) and
smoothes the inflection point of the VAC at U sp, and results in a hys-
teresis.
The sensitivity of the probe to contamination depends on the probe
material. Typical examples of the effect of contaminants on the VAC
are shown in Figs.10.13 and 10.14. The characteristic shown in Fig.10.14
ip
i p , mA
0.25 A/div
Up
5 V/div
U p, V
F ig .10.13 Effect of contaminants on the VAC of the probe in measurements in dc

ig.10.13
glow discharge in argon at a pressure of 0.167 mm Hg. Time from the moment of
cleaning the probe: 0 (1), 3 (2), 7 (3), 20 min (4). 39
F ig .10.14 Variation of the VAC of a tungsten probe when changing the properties
ig.10.14
of its surface. 14
243
Plasma Diagnostics
contains a region of a rapid variation of the slope. One of the possi-

ble explanations is the variation of the work function of the probe surface
on approaching U sp as a result of its heating and, consequently, the
variation of U p (equation (10.27)). This is also confirmed by the fact
that with a rapid increase of U c the VAC is described by curve 1 and
when U p rapidly decreases – by curve 2. It is evident that the deriva-
tives of the distorted probe current should be non-monotonic in such
cases.
In Ref.41, non-monotonic i''p was observed in investigations with a
heated probe in nitrogen plasma with cesium vapours at low probe tem-
peratures. The authors attributed this to the formation of cesium spots
on the tungsten surface of the probe and by representing the VAC by
a superimposition of probe characteristics displaced by Dj due to dif-
ferent work functions of clean areas and areas coated with cesium. In
Ref.20, it is reported that U sp can be shifted as a result of the forma-
tion of oxide films on the surface of the tungsten probe and the alu-
minium counterprobe and the associated variation of the work function.
The nature and strength of the effect of contaminants on the results
of probe measurements must be analysed in every case. For example,
in Ref.2 it was concluded that the formation of a film of amorphous
hydrogenised silicon a = Si:H at temperatures higher than 200° C has
no effect on the results of probe measurements.
One of the important problems of probe diagnostics is the problem
of criteria of contamination of the probe because on the basis of the
form of the VAC and its derivatives it is difficult to draw any conclusions
regarding the presence of contamination. Thus, hysteresis may not form
if the rate of contamination is relatively high.
The low-frequency impedance of the probe is an objective param-
eter enabling the presence of contaminants to be evaluated and inspect
their size. 43 Cleaning of the probe increases the active and decreases
the capacitance component of probe impedance.
In discharge system where the probe surface can be contaminated,
advantages are often offered by pulsed measurements. They are less sen-
sitive to contamination due to a higher capacitance component of the
impedance; in addition, the probe surface does not manage to change
during short periods of time. The inflection at Usp on the VAC produced
by the pulsed method is more distinctive than on the static character-
istics.
In probe measurements in plasma with spattering of the metal or with
the formation of conducting coatings, the VAC may be distorted as a
result of an increase of the probe area in depositing conducting films
on the elements of probe insulators that are in contact with the probe.
This can be avoided by moving the area of contact of the probe with
244
the insulator away from the plasma region.

The main methods of cleaning the probes is ion bombardment at high
negative potentials of the probe and heating using an external source.
Usually the probe is cleaned for a long period of time and is then con-
nected to the measuring system for a short period of time.
The appearance of coatings on the probe and the variation of its sur-
face is regarded as a phenomenon that complicates the measurements.
However, the sensitivity of the probe to the surface condition makes
it an efficient tool for examining the processes of plasma chemical po-
lymerisation and etching (see Ref.7 and also Chapter 13 of this book).
Another reason for the errors should be mentioned. It is associated
with the emission of electrons from the probe under the effect of ions
at high negative probe potentials. 3,6 This distorts the potential distri-
bution in the near-probe layer and increases the apparent ion current
and may cause an error in the ion concentration determined from the
ion current. Since it is very complicated to consider and detect this phe-
nomenon, it is necessary to use materials with a low secondary emis-
sion coefficient and work with low negative values of U p.
10.3 Determination of plasma parameters from probe

measurements
Mean electron energy e is the first moment of the EEDF and is de-
scribed by the relationship
∞ ∞ ∞
ε = εf (ε )dε / f (ε ) dε, ε = εf0 (ε )dε,
∫ ∫ ∫ (10.28)
0 0 0
where f (e) = n e f 0(e) is linked with the probe current by the relation-
ships (10.12) or (10.24) (dimension f(e) – cm –3 ×eV –1). To calculate e
with an error lower than 5%, it is sufficient to measure EEDF to en-
ergies lower than (3 ÷ 4) ε .
Assuming Maxwell’s EEDF, we can calculate the electron temperature
Te
−1
kTe  d ( ln ie ) 
= −
 dU p 
. (10.29)
e  
Since the real EEDF differs from Maxwell’s EEDF, Te calculated from
equation (10.21) can be both too high and too low in comparison with
245
Plasma Diagnostics
Teff = 2 e / 3, and this difference depends on the probe potential at which

Te.7,23 When Te is determined at a floating potential Te > Teff. It was shown
in Ref. 27 that the best agreement between T e and Teff is obtained when
using the VAC in the vicinity of the plasma potential U sp. It should be
taken into account that this region of probe potentials is characterised
by the maximum possible distortions of the VAC as a result of the sink
of the electrons to the probe, the finite resistance of the plasma and
elements of the probe circuit.
Electron concentration ne is computed by integrating equations (10.2)
and (10.24) with respect to energy
∞
ne = f (ε )dε,
∫ (10.30)
0
For the Langmuir probe (equation 10.2)
∫
ne = const U 1/p 2i′′p dU p .
0
(10.31)
If f(e) is not determined in the absolute measure, the equations (10.30)

and (10.31) give the relative value of n e.
The electron concentration can be determined from the random current
on the probe and the plasma potential
di
ne = 4ie U sp / evS p , (10.32)
where v = 5.9 ⋅105 ( ε )

1/ 2
is the mean electron velocity in m/s; e in eV,
all other quantities are determined in the SI system.
Ion concentration n i is determined from the VAC in the region of
the ion saturation current. The problem of the relationship of the plasma
parameters with the ion current on the probe is one of the most com-
plicated in the probe diagnostics and is not examined here. Several ap-
proximate solutions of this problem have been described in Ref.1-6.
Equations of the following type
1/ 2 n
 kTe   eU p 
ii = S p eni     , (10.33)
 2πM i   kTe 
246
j i*
j i*
r p/ λ D
eU p /kT e
F ig .10.15 Dependence of the normalised ion current on the cylindrical probe j i* on

ig.10.15
probe potential. r p /λ D = 0 (1), 3 (2), 4 (3), 5 (4), 10 (5), 20 (6), 50 (7), 100 (8);
T i/ T e = 0. 44,45
F ig .10.16 Dependence of the normalised ion current on the cylindrical probe on
ig.10.16
r p /λ D. eU p / kT e = –25; T i / T e = 0 (1), 0.5 (2), 1.0 (3). 44,45
are used quite often. Here n is determined by experiments, M i is the

ion mass. For a thin probe and a collisionless layer n = 0.5.3,4,7 Equation
(10.33) takes into account ion acceleration in the preliminary layer, i.e.
the fact that the ions arrive at the layer boundary with the probe at ve-
locities corresponding to T e and, consequently, the information on ion
temperature T i recorded in this manner in the ion current is lost.
It is preferred to use numerical solutions of the equations describ-
ing the current on the probe from the surrounding plasma. The most
detailed examination of this problem has been carried out by
Laframboise. 44,45 In the case of a collision-less layer in plasma, con-
sisting of neutral particles, positive ions and electrons with Maxwell
distributions, the ion current on the probe is determined by the equa-
tion
1/ 2
 kTe 
ii = eni S p   ji* , (10.34)
 2 πM i 
where j*i = j*i (r p/l, r p/l D, T i/T e) is the normalised current density on
the probe which also takes into account the dependence of ion current
to the probe on the probe potential. The values of j*i, calculated for
T i/T e = 0 for a cylindrical probe, are shown in Fig.10.15 and the ef-
fect of T i is evident from Fig.10.16.
To calculate n i from the VAC of the probe in the region of the ion
saturation current, independent measurements must be taken to deter-
mine the ion composition of plasma and ion temperature T i. In addi-
247
Plasma Diagnostics
tion, the expression for the ion current includes T e and U p which can
be determined if the VAC is measured in the region of not only the ion
current but also electron current up to the plasma potential (U p is
measured in relation to U sp).
In quasi-neutral plasma in the absence of negative ions n i = n e, and
the ion parts of the VAC can be used to determine ne. It should be noted
that since the value lD, required for calculations, includes the unknown
value n e, an iteration procedure must be used to determine this value.
Iterations can be avoided if we use the method proposed in Ref.46. In
this procedure, the known value of ion current i i at the selected probe
potential (sufficiently high to eliminate the effect of electron current)
and T e are used to calculate j*i (r p/l D) 2 which in contrast to j*i depends
only on i i and T e
ii rp
( ) −
( M i / 2 πe )1/ 2 ( kTe / e ) 3 / 2 ,
2
ji* rp / λ D = (10.35)
l pε0
where e 0 is the base of natural logarithms. From the graph showing the
dependence of j*i on j*i (rp/lD)2 at the corresponding probe potential we
determine j*i and then from equation (10.36) the value of n i
−1
ni = ii  ji*erpl p ( 2πkTe / M i )  .
1/ 2
(10.36)
 
Many attempts have been made to verify the ion current theories,
including Laframboise theory. They were based on comparing the values
of n i, determined from ion current, with n e measured by independent
methods (microwave or from electronic saturation current). Some of the
results from this comparison are presented in Ref.6. Good agreement
was obtained using the Laframboise theory. However, there are also data
indicating differences in the values of n i and n e, with n i being higher
than n e obtained from the VAC of the plasma potential. For example,
in Ref.47 the values of n i were systematically three times higher than
ne. This may prove to be important in determining the concentration of
charged particles in plasma containing negative ions because their con-
centration is assumed to be equal to n i – n e.
The results of calculations of n i from the ion current may be influ-
enced by the difference between the EEDF and the Maxwell EEDF that
is usually used in calculating i i . The authors of Ref.23 examined the
effect of the EEDF on the Debye screening length and i i on an exam-
ple of a wide range of modelling functions. It is shown that although
248
l D depends on the EEDF, the role of high-energy part of the EEDF is

small. An equation was derived determining the dimensionless multi-
plier in the equation of type (10.33) for the current taking into account
the form of the EEDF. Calculation of these multipliers from the EEDF
measured in hf discharge in inert gases shows that they differ by up
to 30% from the values obtained assuming the Maxwell EEDF.
Collisions of the ions together and with neutral particles change the
VAC of the probe in the ion current region. 3,6 For example, the former
leads to an increase of ion current, the latter to its decrease, with the
exception of the case of a single collision in the layer where the cur-
rent may increase as a result of disruption of orbital motion.
It should again be noted that the apparent ion current can be increased
by electron emission from the probe under the effect of different fac-
tors and this may become significant at high negative probe potentials
in ion bombardment of the probe surface.
In electrodeless discharges, there are considerable difficulties in po-
sitioning the counterprobe with a large surface to obtain a single-probe
characteristic. Double probes are used extensively under these condi-
tions.48 They consist of a system of two probes of the same area placed
in plasma at a distance at which there is no mutual screening. If the
plasma properties in the areas where these probes are positioned do not
differ, the volt–ampere characteristic of such a system has the form
shown in Fig.10.17. The abscissa gives the voltage between the probes
U = U p1 – U p2 + DU, where DU equals the difference of the plasma
potentials in the areas in which the probes are positioned and the dif-
ference of the contact potentials of the probes. At DU ¹ 0, the char-
acteristic is displaced along the voltage axis by DU. The maximum
current in this system is determined by the ion saturation current on
the probe under high negative potentials. The second probe is also at
a negative potential which is such that the electron current to this probe
i p (U)
F ig .10.17 VAC of a symmetric double probe.

ig.10.17
249
Plasma Diagnostics
is equal to ion current of the first probe. When measuring the polar-
ity U, the probes change their roles.
The double probe has a number of advantages in comparison with
a single probe. Since the probes are usually quite close, in measure-
ments in hf plasma the alternating voltage acting between them is greatly
reduced thus reducing the degree of distortion of the VAC. In addition,
the effect of alternating voltage on the ion part of the VAC is consid-
erably smaller than that on the electron part. It must be considered that
parasitic capacitances (Fig.10.2) lead to the appearance of high-frequency
currents and distort the VAC.
In addition, the collisions in the near-probe layer have a smaller effect
on the double probe in comparison with a single probe so that the double
probe can be used to determine the plasma parameters under the con-
ditions in which the single probe gives inaccurate results. 6
To obtain the plasma characteristics from the VAC of the double probe
we can use the assumption of the Maxwell’s distribution of electrons
in plasma. There are several methods of processing the VAC (see, for
example, Refs.4 and 5). One of them is similar to the procedure used
in the method of the semi-logarithmic graph for single probes
e d i +i 
= ln  i1 i 2 − 1  ,
kTe dU  ie1 

(10.37)
where i i1, i i2, i e1 are the ion electron currents on the probe. The prin-
ciple of determination of these currents is clear from Fig.10.17 (lin-
ear approximation of ion current was used).
The ion concentration is calculated by the previously described
methods of processing the ion parts of the VAC.
10.4 Selected problems of probe diagnostics

It is not possible to examine here all types of application of probes for
investigating the plasma. We shall therefore mention only some of them
and quote the literature where further information can be found, and
we shall examine in greater detail only the applications which have not
been described sufficiently in the review of literature. Many plasma
devices are based on plasma in a magnetic field. The magnetic field
greatly complicates the probe diagnostics. The velocity of the particles
along and across the field differs and there are a large number of prob-
lems associated with plasma anisotropy (the measurement results de-
pend on the probe of orientation). In the presence of a magnetic field
the electron current at the plasma potential decreases, the transition from
250
the region U p < 0 to the region U p > 0 is smoothed out, the section of
the VAC at U p < 0 is distorted (in this section the current on the probe
is still determined by the electrons). The effect of the magnetic field
on the ion part of the VAC is considerably weaker than that on the elec-
tron part. At present, there is no complete theory of probes in the mag-
netic field, and some of the attempts made to develop such a theory have
been described in Ref.1, 3, 4, 6, 13.
Specific problems appear in probe diagnostics of plasma flows under
the conditions of a collisionless layer, the transition regime and the con-
tinuum regime. 2-6
Probe measurements in plasma containing negative ions

Regardless of the fact that the non-equilibrium plasma of electronic active
gases is used widely and there are large number of investigations in which
the results of probe measurements have been published, this region of
application of probes is one of the least examined. The presence of
negative ions leads to a number of effects.
1. The difference between electron and ion saturation currents and
between the floating potential and plasma potential decreases. 1,6,49-52
Consequently, the effect of the layer of current on the VAC of the probe
is observed at almost all probe potentials, and in calculating the pa-
rameters of the electronic component it is necessary to deduct the ion
current. The role of this factor becomes important with increasing
n –/n e, where n – is the concentration of negative ions. A decrease of the
difference of the ion electron currents of saturation simplifies the re-
quirements imposed on the ratio of the areas of the probe and the
counterprobe examined in (10.2).
2. In determining n e from the electronic saturation current (equation
(10.32)) it is important to take into account that i p (U sp) can be dis-
torted by the current of negative ions i –. The ratio of these currents is
determined by the equation
(
i − / ie = n − / n e ) (T
i
−
)
m / Te M i− , (10.38)
where T –i and M–i is the temperature and mass of the negative ions. At
m/M–i < 10–4, T –i /T e < 10 –1 the effect of ions with i p(U sp) can be ignored
if n –/n e < 10 –3 .
3. The ion part of the VAC of the probe in the regime of the
collisionless layer at the electron temperature considerably higher than
the ion temperature, and at n–/ne < 2 does not change and ion ‘collection’
takes place in the same manner as in the plasma without negative ions.
When determining n i from the ion saturation current, we can use the
251
Plasma Diagnostics
well-known equations in which the positive ions travelled to the boundary

of the near-probe layer with the velocity of ion sound ((kT i/M i) 1/2). At
n –/n e > 2 the structure of the electric field in the layer is distorted and
the velocity of the positive ions at the boundary of the layer is
(kT i /M i) 1/2. 51
4. The second derivative of the probe current with respect to the probe
potential i''p at probe potentials close to U sp can contain information on
the energy distribution of the negative ions 1
( )
U 1/p 2 i′′p U p ≅ const  ( e / m ) ( ) ( ) ( )
n − f 0− eU p  ,
1/ 2
ne f 0e eU p + e / M i−
1/ 2
 
(10.39)
where f e0 and f 0– are the energy distributions of the ions and electrons,
respectively. Examples of the distribution of the negatively charged par-
ticles (obtained from the second derivative of the distributions) in the
stratified discharge of direct current and in oxygen and in a plasma-
beam discharge in SF 0 are shown in Fig.10.18 (i''p was determined by
modulation methods using signals of the type (10.17)) (the fourth equa-
tion) (O 2) and (10.17) (the first equation) (SF 6)). Peaks in the energy
range ~0.1 eV are attributed to negative ions. Integration of the resultant
distributions with an allowance made for m and M – i gives absolute
concentrations n – and n e. 50,52
5. The presence of negative ions has no influence on the position of
the potential of the inflection point of the VAC in transition from the
region with U p < 0 to U p > 0 and the inflection at U sp is more marked
than without the negative ions. 49
It should be noted that the presence of the peak of negative ions on
U p 1/2 i p|| U p 1/2 i p||

arbitrary units arbitrary units
a b
eU, eV
F ig .10.18 Energy distributions of negatively charged particles in the plasma of a

ig.10.18
stratified discharge in oxygen at n – / n e ¾ 20(a) 51 and in plasma-beam discharge in
SF 6 at n – / n e = 150 (b). 50
252
i''p was not observed in most investigations (see, for example, for plasma
of the same gases but in rf discharge 20,53 ). The authors of Ref.49
described their method of probe investigation of the plasma of electronic
active gases with the energy distribution of electrons slightly differing
from Maxwell’s distribution.
Probe measurements in anisotropic plasma

We previously assumed that the distribution of the electrons in the non-
perturbed plasma region is isotropic. Under the conditions of real gas
discharge plasma there are gradients of the strength of the electric field,
the beams of electrons in the near-electrode regions of dc discharges
and rf discharges, the plasma can be sustained in magnetic fields, the
plasma often moves in relation to the probe, etc. All this leads to plasma
anisotropy and, consequently, the previously described methods of probe
diagnostics cannot be used. The results of probe measurements must
depend on the probe orientation. 1
There are several approaches to solving the problem of probe meas-
urements in anistropic plasma.
In axially-symmetric plasma, the electron velocity distribution function
(EVDF) in the spherical co-ordinate system with the polar axis coin-
ciding with the axis of symmetry of the plasma, can be represented by
an expansion with respect to Legendre polynomials
∞
f (v, θ) = ∑ f (v)P (cos θ),
j=0
j j (10.40)
where v, q is the velocity modulus and the polar angle. Coefficient f 0

determines the velocity modulus distribution of electrons, f 1 is linked
with the convective velocity of the electrons. The representation of
f (v, q) only through f 0 and f 1 is known as the Lorentz approximation.
Equation (10.40) shows that determination of the EVDF is reduced to
determining the coefficients in the expansion f (v, q).
In Ref.54, this problem was examined with special reference to various
types of probes. It was shown that for axial symmetry of plasma, the
determination of the finite number of the expansion coefficients
f (v, q) is possible when taking probe measurements for fixed probe ori-
entations. Measurements taken using flat two-sided and cylindrical probes
enable even expansion coefficients to be determined. Flat one-sided and
flat double probes can be used to determine even and odd expansion
coefficients. The spherical probe is not sensitive to plasma anisotropy
and gives f 0.
253
Plasma Diagnostics
The possibilities of this approach were illustrated in Ref.55 on an

example of investigating a plasma-beam low-voltage arc in He using
cylindrical probes, with some of the probes oriented along the axis of
symmetry of the discharge gap and others normal to this axis.
At U p < 0 and with all the requirements of probe diagnostics ful-
filled, with the exception of isotropic distribution, the density of electronic
current on the probe is determined by the equation
2π ∞ (
arccos eU p / ε )
2πn
( )
j eU p = 2e dϕ εdε
m 0 eU ∫ ∫ ∫ f (ε, θ1, ϕ)sin θ1 cos θ1dθ1. (10.41)
p 0
where q 1, j are the polar and azimuthal angles in the spherical co-or-
dinate system whose axis is directed along the external normal to the
probe surface; f (e, q 1, j) is the distribution function normalised with
respect to 1. After integrating equation (10.41) and taking into account
(10.40), with double differentiation with respect to U p, we have
2πe2ne S p ∞
( )
ie′′|| eU p =
m2
∑ F (eU ) P (0),,
2J p 2j (10.42)
j =0
π
2e2 ne S p
( )
ie′′⊥ eU p =
m2
∑ P (cos θ)dθ,
1j (10.43)
j =0
∞
∂
( ) (
F2 j eU p = ef2 j eU p − ) ∫ f2 j (ε )
∂U p
P2 j ( )
eU p dε, (10.44)
eU p
where the signs || and ^ indicate the axial and perpendicular position
of the probe.
Equations (10.42)–(10.44) show that i²e in anisotropic plasma depends
on probe orientation. In addition, the signs of the terms of the series
in the equations (10.42)–(10.44) change and, consequently, i²e may assume
negative values. These facts are used as a basis for experimental veri-
fication of the anisotropy of EVDF. It should also be noted that the cur-
rent on the cylindrical probe does not depend on odd harmonics, i.e.
it is independent of convective current. Examples of i²e for two probes
at different points of the arc are shown in Fig.10.19. Anisotropy of
plasma of the cathode and isotropic transformation of the EVDF at a
distance from it are visible. It was assumed in Ref.55 that f 2 >> f 4, and
f 0 and f 2 were also determined.
This method was developed further in Ref.56–58. The characteris-
254
i p||
arbitrary
units
eU , eV
F ig .10.19 Second derivatives of probe current i"p at different distances z from the
ig.10.19
cathode in a low-voltage arc in helium at a pressure of 1 mm Hg and a discharge
current of 0.4 A. Spacing of the electrodes 6 mm, U = 0 is the cathode potential,
the scale of the high-energy part is magnified 10 times; 1) perpendicular probe; 2)
parallel to the discharge axis. 55
tics of the cylindrical and flat probes in anisotropic plasma were also
studied in Ref.59.
The authors of Ref.60 and 61described the method of measuring the
directional part of electron distribution f 1 (v), if the distribution is rep-
resented in the Lorentz approximation
H
f (v ) = f0 (v ) + f1 (v ) cos θ. (10.45)
Measurements should be taken using two flat probes oriented in the

opposite directions, or using a double flat probe. Measuring the cur-
rent difference of two probes at q = 0, we can determine f 1 (v)
 ( )  ,
d ∆i p
f1 ( )
2eU p / m =
m2
4πnee3S p
U 1/ 2 d
p

dU p 
U −1/ 2
p
dU p 
(10.46)

or, after differentiation
255
Plasma Diagnostics
v f0.1 /[v f ], m
v, V
F ig .10.20 Functions vf 0 ( v )/[ vf (v )] m (1,2) and vf 1( v )/[ vf ( v )] m (3,4) in discharge in helium

ig.10.20
at a pressure of 0.11 mm Hg. 61 [ vf ( v )] m is the maximum value of the Maxwell function
at a current of 30 mA; 1,3) 40 mA; 2,4) 30 mA.
( ) ( )  .
m2  d ∆i p d ∆i p
2
f1 ( 2eU p / m = )  2
8πe3S p  dU p
−
dU p 
(10.47)

The equations (10.46) and (10.47) were obtained assuming the iden-
tical plasma potentials at the areas where the probes were situated, i.e.
DU sp = 0. If DU sp ¹ 0, the expressions for f 1 are greatly complicated
and an addition appears. The value of this addition depends on U p and
is maximum at U p ~ DU sp. The equations which enable corrections for
DU sp ¹ 0 to be made are presented in Ref.61 and they were analysed
and determined more accurately in Ref.13. The application of the double
flat probe is illustrated in Fig.10.20.
If the EVDF in a system with cylindrical symmetry is represented
in the form f (v) = f (v z, v ^), where v z and v ^ are the velocities of the
electron in the direction of the z axis and the perpendicular plane, and
the longitudinal f 1 (v z) and transverse f 2 (v ^) EVDF, expressed by the
equations
∞
f1 (vz ) = 2 π f (vz , v⊥ )v⊥ dv⊥ ,
∫
0
∞
(10.48)
f 2 (v⊥ ) = ∫ f (v , v )dv ,
z ⊥ z
−∞
are introduced, the EVDF can be determined using the method proposed
in Ref.13 and 62. Measurements with a flat one-sided probe, oriented
in the direction normal to the z axis, and the cylindrical probe oriented
along the z axis can be used to determine f 1 (v z) and f 2 (v ^)
256
mdi p
f1 ( vz ) = , (10.49)
nee 2 S p dU p
∞ −1/ 2
m3/ 2 d 2i p
f 2 (ε ⊥ ) = 3/ 2 2
2 nee S p ∫ (eU
ε⊥
p − ε⊥ ) dU p2
, (10.50)
where e ^ = mv 2 /2.
The VACs of the flat and cylindrical probes make it possible to cal-
culate the ‘longitudinal’ and ‘transverse’ temperatures using equation
(10.28). Orienting the normal to the surface of the flat probe along and
across the z axis, the drift electron velocity can be determined from the
current difference at the space potential
vdr = Di p / ne S p . (10.51)
In Ref.1 it was concluded that as a result of screening the plasma

with a probe, the value v dr, computed from (10.51) may be too low.
257
Plasma Diagnostics
Chapter 11
ELECTRIC PROBES IN CONTINUUM REGIME
11.1 Introduction
Electric probes as a means of diagnostics of low temperature plasma
have a number of advantages, the main ones being simple experimental
procedures and the possibility of determining local parameters. Unfor-
tunately, there is no general theory of electric probes suitable for in-
terpreting the results of probe measurements over a wide range of the
conditions.
The parameters characterising the operating regime of the probe (i.e.
the regime of passage of current in the near-probe region) are numer-
ous. Usually, the regimes are subdivided in relation to the ratio between
the free path of the ions and electrons under the conditions of elas-
tic collisions l i, l e and the linear dimension of the near-probe region
of perturbation. To simplify considerations, it will be assumed that the
thickness of the near-probe layer of the volume charge is not greater
(as regards the order of magnitude) than the characteristic damage of
the probe a (this assumption holds if the Debye radius is comparable
with the probe size or is smaller, and the probe potential is not too high).
Consequently, the probe size a can be regarded as the linear scale of
the near-probe region of perturbation.
When l i, l e >> a, the collisions in the near-probe region are not
significant. These operating conditions of the probe are referred to as
collisionless or molecular. The theory of these regimes was developed
in the twenties 1 and described in the well-known studies 2–8 (we should
also mention Ref.9 which contains, in particular, an extensive bibliog-
raphy; please also refer to Chapter 10 of this book). It should be noted
that, in this case, the volt-ampere characteristic of the probe contains
also information on the non-perturbed function of electron distribution,
in addition to information on the parameters such as, for example, the
electron concentration in non-perturbed plasma or plasma potential.
At l i, l e Ù a, important kinetic scales are the length of the en-
ergy relaxation of the electrons as a result of collisions with neutral
258
Electric Probes in Continuum Regime
particles l u = l e/Öd and the length of Maxwellisation as a result of

interelectron collisions l m = l e(n en /n ee) 1/2 where d is a parameter char-
acterising energy transfer from the electrons to neutral particles (for
elastic collisions, this parameter is equal to the value of the double ratio
of the electron mass to the mass of the neutral particle, for non-elastic
collisions it is equal to the same value multiplied by the coefficient of
inelastic losses), n en, nee are the local values of the frequency of elastic
collisions of electrons with neutrals and the frequency of interelectron
collisions (it is assumed that the plasma is weakly ionised and n en >>
n ee ). It should be noted that usually d << 1 and, consequently,
l u>> l e; in the examined case of weakly ionised plasma we also have
l m >> l e .
In the case with l i, l e Ù a << l ul m, collisions in the near-region
are significant but the change of electron energy as a result of colli-
sions is small. The theory of electron current on the probes under these
regimes was developed in Ref.10 and 11. It should be noted that in
this case the VAC retains information on the energy distribution of
electrons on non-perturbed plasma.
In the opposite limiting case where at least one of the lengths l u,
l m is considerably smaller than the probe size, l i , l e << (l u- 1 +
l m–1) –1 << a, collisions in the near-probe region are the more signifi-
cant process. Evidently, there is no relationship between the VAC of
the probe and the type of non-perturbed electron distribution function
in this case; on the other hand, the non-perturbed distribution function
under these conditions should differ only slightly from the Maxwell’s
function. To describe the distribution of both ions and electrons in a
large part of the near-probe region (outside the layer with a thickness
of the order of (l u–1 + l m–1) –1 adjacent to the surface of the probe) we
can use the hydrodynamic approximation in this case. These regimes
of operation of the probe are referred to as the continuum regimes or
hydrodynamic regimes. Hydrodynamic equations were used for analysis
of the characteristics of electric probes in Ref.12; in Ref.13 and 14,
the hydrodynamic equations were used in the entire near-probe region
of perturbation without dividing this region in advance into a quasi-
neutral region and a volume charge layer. Reviews of the hydrodynamic
theory of electric probes were published in Ref.6, 8, 15-17.
In this chapter, we examine probes in the hydrodynamic regimes;
its aim is to add to the reviews mentioned previously and describe the
current state of probe diagnostics procedures.
11.2 A system of equations and boundary conditions of the

hydrodynamic theory of electric probes
We shall examine stationary moving or still slightly ionised plasma con-
259
Plasma Diagnostics
taining neutral particles (generally speaking, of several different types),

ions (in a general case also of several types) and electrons. The plasma
contains a conducting solid (electric probe). The effect of charged par-
ticles on the flow field of neutral components and plasma as a whole
will be ignored. It will be assumed that the degree of dissociation and
ionisation of the main neutral components of the plasma (i.e. the com-
ponents whose contribution to collisions with the charged particles is
considerable) are low and the molar fractions x q (x q = n q/n, where n q
is the concentration of the q-th component, n is the total concentra-
tion of plasma particles) changes only slightly in the vicinity of the probe.
It will be assumed that the frequencies of elastic collisions of the
charged particles with the neutral particles greatly exceed the frequen-
cies of collisions of charged particles.
The system of equations which determine the distributions of the
concentrations of charged particles n m, the densities of their fluxes J m
and electrostatic potential ϕ, includes the equations of conservation of
the charged particles and Poisson's equation (see, for example, Ref.6
and 8):
nm
nvÑ + ÑJ m = wm , m = i , e, (11.1)
n
Dj = -4pe å zmnm , (11.2)
m = i ,e
and also the hydrodynamic equations of transfer which can be writ-

ten in the form (see, for example, Ref.18)
 P 
J m = − µ m  ∇ m + zm nm∇ϕ  , m = i, e (11.3)
 e 
Here v is the mean mass velocity of the plasma, w m is the rate of

variation of the concentration of the m-th component and the result of
volume reactions, µ m, z m, P m is the mobility, charge number and the
partial pressure of the m-th component, index i assumes the values
corresponding to different types of ions, the index e is given to the
electrons. The quantities n, v as well as the temperature of neutral
particles T, can be determined by solving the corresponding gas-dynamic
problem of flow around the probe and by neglecting the effect of
charged particles. It is assumed that they are given functions of the
coordinates. The first term in the left-hand part of the equations of
conservation of the charged particles (11.1) describes convective trans-
260
fer, the second term – diffusion and drift transfer, the term in the right-
hand part - the formation or disappearance of the charged particles
of the given type as a result of volume reactions. Poisson’s equation
(11.2) links the volume charge of plasma with the induced electric field.
First term in the brackets in the right-hand part of the hydrodynamic
transfer equation (11.3) describes the concentration diffusion, the second
term - the drift in the electric field. It should be noted that when writing
the last equations the terms describing thermal diffusion were omit-
ted together with some terms taking into account baric diffusion; usually,
these effects are small and they can be taken into account only in special
cases. 6,19-21
The hydrodynamic equations of transport (11.3) for the ions
(m = i) are applicable at l i << L, where L is the local macroscopic
scale, i.e. the linear dimension of the examined hydrodynamic zone.
In writing the diffusion term we can accept that P i = n ikT, where k
is the Boltzman’s constant.
In the case of a weak field (eEl i << kT, where E is the strength
of the electric field) these equations can be regarded as the partial case
of the Stefan–Maxwell equations for multicomponent diffusion 22-24 (see
also Ref.19 and 25). If only one component is the main in the group
of neutral plasma components, then the quantity µ in is the function of
only temperature and, to a first approximation, is equal to
3e/(16m iqW iq(1,1) ), where W iq(1,1) is the collision integral, 23,26 m iq is the
reduced mass; to determine this quantity we can use either the experi-
mental data or the results of calculating the collision integral (these
calculations can be carried out for the known transport scattering cross
section or the ion–neutral particle interaction potential). If the number
of the main neutral component of the mixture exceeds unity, quantity
µ in also depends on the molar fractions x q; to a first approximation,
these dependences are described by Blanc’s law 19.,27 and the mobil-
ity is expressed through mobility in pure gases
−1
 1  3e
µi = 
 ∑  ,
µiq 
µiq =
16nq miqΩiq( )
1.1
.
 q
In the case of a strong field (eEl i Ú kT), the diffusion term of the
transfer equations has the order l i/L in relation to the drift term and
under the conditions of applicability of the hydrodynamic approxima-
tion is low. It should be noted that this circumstance determines in the
majority of examined problems the use of the Einstein equation in writing
261
Plasma Diagnostics
the diffusion term of the equation (11.3) at the temperature of the neu-
tral particles. The validity of this equation in strong fields is violated.
The quantity µ in in the case of the strong field is the function of also
E/n, and Blanc’s law, generally speaking, is no longer valid. To deter-
mine this quantity, it is necessary to determine the experimental data
of the results of solving the kinetic equation.
An extensive review of the data on the mobility of ions in pure gases
was published in Ref.28–30.
The hydrodynamic transfer equation (11.3) for the electrons
(m = e) is valid if at least one of the following two conditions is ful-
filled:
l u << L , l m << L. (11.4)
The first of the conditions is discussed in Ref.31, the second fol-

lows from Ref.32–34. The mobility of electrons at the known electron-
neutral particle transport collision sections and the known electron dis-
tribution function or, more accurately, its main isotropic part f 0 is de-
scribed by the Lorentz equation for slightly ionised plasma. 18,31,35,36
If the first condition (11.4) is fulfilled, the function f 0 (and, con-
sequently, the quantity µ en) is given by the local values of the param-
eters T, E/n, xq, xe; when writing the diffusion term of the transfer equa-
tion we can accept P e = n ekT. When E << kT (el u) f 0 is the Maxwell
function with the temperature of the neutral particles and the dependence
of µ en on E/n, x e disappears. When E Ú kT (el u ) at n ee /n en £ df 0
is not, generally speaking, Maxwell’s function; 18,31,35,36 at n ee/n em >> d,
f 0 is the Maxwell function with temperature T e determined by the lo-
cal balance of Joule heat and energy exchange with the neutral par-
ticles. It is important to stress that when E Ú kT (el u) the diffusion
term of the electron transport equation has the order l u/L in relation
to the drift term so that the Einstein equation can be used in the majority
of examined problems at the temperature of neutral particles.
If only the second condition (11.4) is fulfilled (this is possible if
n ee /n em >> d), then f 0 is the Maxwell function with the temperature
T e. To determine this temperature, the system (11.1)–(11.3) should be
supplemented by a differential (containing spatial derivatives) equation
of electron energy. 6,18 When writing the diffusion term of the trans-
port equation, it should be accepted that P e = n ekT e.
A large number of data on the mobility and transport sections of
electron scattering was published in Ref.37–41.
The boundary conditions with the concentration of the charged par-
ticles and the potential on a probe surface (which is assumed to be
262
ideally absorbing and non-emitting) and away from the probe have the
form: 6
On the probe surface
n m = 0, m = i, e; ϕ = ϕ w, (11.5)
away from the probe
nm = nm¥ , m = i , e; j = 0, (11.6)
where jw, nm∞ is the potential of the probe surface in relation to plasma
and the concentration of the charged particles in the non-perturbed
plasma (given values).
We shall examine briefly the justification of the first boundary con-
ditions (11.5), initially for the ions. Evidently, the most consistent con-
clusion is the one based on the asymptotic analysis of the Boltzmann
equation. 42 In this approach, the entire region of the plasma adjacent
to the surface of the probe is subdivided into the Knudsen layer, i.e.
the wall region of the plasma with the thickness of the order l i, and
the adjacent hydrodynamic region with the characteristic linear size
L g >> l i. In each of these zones, we construct its own asymptotic ex-
pansion of the distribution function with respect to the small param-
eter li/L g. Combining the expansions gives, in particular, the condition
that must be satisfied by the first term of the asymptotic expansion
which holds in the hydrodynamic region. This condition is also the re-
quired macroscopic boundary condition.
For the case in which the field in the near-wall region is weak, the
first condition (11.5) was justified in Ref.42 and can be formulated as
follows. Let it be that n ki, n gi are the characteristic values of the ion
concentration in the Knudsen layer and the hydrodynamic region. The
flux of the ions to the surface in the Knudsen layer can be evaluated
(taking into account the anisotropy of the distribution function) with
regard to the order of magnitude as n kiC i, and in the hydrodynamic re-
gion as D in gi /L g, where C i, D i are the characteristic thermal velocity
and the diffusion coefficient of the ions. Since the flux across the
Knudsen layer is maintained (with the accuracy to the ratio of the fre-
quencies of ionisation, recombination, adhesion, etc. to the frequency
of elastic collisions), these estimates are compared and we obtain
n ki/n gi ~ l i/L g << 1. To enable combining, the first term of the asymp-
totic expansion of the distribution function in the hydrodynamic region
can convert to zero on the surface of the probe. This leads to equa-
tion (11.5).
263
Plasma Diagnostics
In the case of a strong field, the diffusion term of the equations of

ion transfer in the hydrodynamic region is, as shown previously, small
in comparison with the drift term and can be omitted. The order of
the degenerate system of the hydrodynamic equations, compiled in this
manner and, consequently, the number of boundary conditions required
to solve it decrease. Evidently, in this formulation of the problem for
the ions drifting to the probe, the boundary condition on their surface
is superfluous; for the ions drifting from the probe the first condition
(11.5) applies. On the other hand, the solution obtained within the frame-
work of this formulation, coincides, with the accuracy l i/L g, with the
solution described by the initial formulation that takes into account the
diffusion term and uses the first condition (11.5). An exception is only
the distribution of a concentration of the ions drifting to the probe in
the near-wall region with the thickness of the order of l i; if the con-
centration in the framework of the degenerate problem is constant to
a first approximation, then in the framework of the problem with dif-
fusion it decreases to zero (in fact, it should be noted that both the
first and second solutions are physically incorrect in the given region
in the general case and to determine the true distribution of the con-
centration it is necessary to find the first term of the asymptotic ex-
pansion of the distribution function in the Knudsen layer 43 ). In order
to provide standard and accurate description (this is important, for ex-
ample, when carrying out numerical calculations), equation (11.3) with
the diffusion term and the first boundary condition (11.5) can also be
used in the presence of a strong field; however, it should be remem-
bered that the solution obtained in this approach is not reliable in the
region of diffusion-induced decrease of the concentration of the ions
drifting to the surface.
For the electrons, the structure of the near-wall non-hydrodynamic
region is more complicated; 32–34 in addition to the Knudsen layer with
a thickness of the order of l e in which the distribution function is iso-
tropic, there is another kinetic layer in which the distribution function
is isotropic to a first approximation but is not local and is not of the
Maxwell type. To determine this function, it is necessary to solve the
equation with spatial derivatives.18,31,35 The thickness of this kinetic layer
is considerably greater than le and is determined by the smaller of the
scales l u , l mkin = l e (n en /n eekin ) 1/2 , where n eekin is the frequency of elec-
tron collisions evaluated on the basis of the characteristic concentration
of electrons in the kinetic layer n ekin.
If dn en > n eekin , the thickness of the kinetic layer is of the order of
lu. In the adjacent hydrodynamic region, the first condition (11.4) must
be fulfilled. When E << kT (el u), estimating the fluxes of electrons
264
in the kinetic layer and the hydrodynamic region as n ekinC eÖd and D en ge/
L g (n ge, C e, D e is the scale of the electron concentration in the hydro-
dynamic region, thermal velocity and diffusion coefficient of the elec-
trons) and equating these estimates (with the accuracy to the ratio of
the ionisation frequencies, and so on, to dn en ), we obtain n ekin /n ge ~ l u/
L g << 1, and the boundary condition (11.5) is valid. As in the case of
the strong field for the ions, this condition can also be used at E Ú
kT/(el u ).
If dn en << n kee
in
, the thickness of the kinetic layer is of the order of
l m . Only the second condition (11.4) is fulfilled in the adjacent hy-
kin
drodynamic region because the quantity l u in this case is itself the

macroscopic scale – at the distance of this order the transport of elec-
tron energy by thermal conductivity is comparable with the energy ex-
change with the neutral particles. Estimating the fluxes in kinetic layer
and the hydrodynamic region as n ekin C e (n kin ee
/n en ) 1/2 and D e n ge/L g, we
obtain n e /n e ~ l m /L << 1, and (11.5) is again valid.
kin g kin g
Thus, the boundary condition (11.5) is also substantiated for the elec-
trons. It should be noted that when dn en << n eekin, to close the hydro-
dynamic formulation of the problem it is also necessary to formulate
the boundary condition on the probe surface for the electron energy
equation. Taking into account that this equation has a singularity 6 on
the probe surface, we find the asymptotics of its solution close to the
surface. The asymptotics includes two terms, one of which is restricted
and the other one algebraically increasing so that the restriction of T e
on the probe surface should be used as the required boundary condition.
Writing the differential equation of the first order (which is satisfied
by the first of the terms mentioned previously) for the surface, we
obtain the equivalent form of the accepted condition:
−1
 K he 5  ∂T 5 ewe  ∂ ne 
 + k e =   − eE y ,
 khe De 2  ∂ y 2 µ e  ∂ y 
where K he is the coefficient of electronic heat conductivity, and the

y axis is directed along the normal from the probe surface; this equation
can be regarded as a generalisation of the conditions derived by several
other methods in Ref.44 and 45.
In the general case, the conditions of applicability of the solution,
determined within the framework of the described hydrodynamic for-
mulation of the problem, are as follows. The characteristic size of the
probe a must satisfy one of the conditions (11.4). When the resultant
solution describes hydrodynamic zones with different linear scales, each
scale must satisfy one of the conditions (11.4).
265
Plasma Diagnostics
11.3 Volt–ampere characteristics of probes under

hydrodynamic regimes
We shall analyse the form of probe VACs, predicted by hydrodynamic
theory, on a frequently encountered example of a probe in a plasma
flow where the gas-dynamic regime of flow of plasma around the prob
is the regime of a viscous boundary layer (i.e. Re = v¥a/n ¥ >> 1, where
Re is the Reynolds number, n is the coefficient of kinematic viscos-
ity of plasma, the indices ¥ and w here and below are given to the
values of the corresponding quantities in non-perturbed plasma and on
the probe surface, respectively), and the Debye radius h, estimated from
the parameters of non-perturbed plasma, is considerably smaller than
the scale of the thickness of the viscous boundary layer in the vicin-
ity of the probe surface D = a/ÖRe. It will be assumed that the flow-
around is chemically non-equilibrium, i.e. the duration of reactions in
which the charged particles of each type form and disappear is com-
parable with the hydrodynamic (time of flight) time a/v ¥. It should also
be noted that the first condition (11.4) is fulfilled in each of the hy-
drodynamic zones. It will be assumed that the Schmidt numbers of the
ions n/D i under normal conditions have the order of unity.
The asymptotic solutions of the problem of this type were exam-
ined in Ref.46–48. We shall present several asymptotic estimates re-
sulting from these solutions.
The entire near-probe region can be divided into the region of non-
viscous flow, the quasi-neutral region of the near-probe viscous boundary
layer and the near-wall layer of volume charge (Debye layer, or DL)
(Fig.1.11).
In the quasi-neutral part of the boundary layer and in the DL the
electric field and fluxes of the charged particles are directed to a first
approximation along the normal to the probe surface (i.e. along the y
axis). In the region of non-viscous flow the distribution of electric field
and fluxes are three-dimensional in the general case.
Fig.11.1 Asymptotic structure of the

near-probe region. 1) region of non-
viscous flow; 2) quasi-neutral part of
the near-probe viscous boundary layer;
3) DL (Debye layer).
266
In the region of non-viscous flow and the quasi-neutral part of the

bounadry n m ~ n m¥ . The scale of the flux density in all three zones
is the same and is determined by the order of the diffusion term of
the equations (11.3) in the quasi-neutral part of the boundary layer:
Jmy ~ Dmnm¥/D. The order of magnitude of the electric field in the region
of non-viscous flow and in the quasi-neutral region of the boundary
layer is equal to kT/(eD). In the region of non-viscous flow, the con-
vective transport of the charged particles is considerably greater than
the drift transport which, in turn, is far more extensive than the dif-
fusion transport. The convective, drift and diffusion transports in the
quasi-neutral part of the boundary layer are comparable.
When estimating the orders of magnitude of the quantities in the
DL, we shall into account the results of asymptotic analysis which has
been carried out many times in the literature for collision layers of the
volume charge, starting with Ref.13 and 14; in particular, the authors
of Ref.49 and 50 obtained a solution by the method of combined as-
ymptotic expansions with respect to the small parameter (when men-
tioning Ref.49 it should be noted 50,51 that the more exact definitions
made in Ref.50 and cited in Ref.6 regarding the conditions of appli-
cability of the solution [14] and the presence of two (not one as in Ref.
13) transition regions between the main part of the DL and the
quasineutral region have not been substantiated). The convective trans-
port of the charged particles in the DL does not play any significant
role. When the DL is homogeneous (i.e. has no internal structure) 14,50
the diffusion and drift transport in the DL are comparable and have
the order D mn m¥/D, which gives the equations for the orders of mag-
nitude of n m, E y in the layer through the scale of its thickness y D :
nm ~ nm¥ yD/D, Ey ~ kT/(ey). Substituting these estimates into the Poisson
equation, we determine y D ~ (h 2D) 1/3 << D.
When the thickness of the DL is considerably greater than
(h 2D) 1/3, but its order of magnitude is not higher than D, the layer is
heterogeneous13,49 and includes the main part of and the transition region
and the zone of the diffusion layer of the concentration of the parti-
cles drifting to the probe separating it from the quasi-neutral region
and the probe surface. The estimates made previously, i.e.
n m~n m¥ (h/D) 2/3 , E y ~ kT/[e(h 2D) 1/3], hold in the transition region, and
the scale of the thickness of this region is (h 2D) 1/3 and considerably
smaller than y D. In the main part of the layer the concentrations of
the particles repulsed by the probe field are low. The controlling transport
mechanism for the attracting particles is a drift mechanism and the
concentration of the particles is n m¥h (y DD) –1/2; E y ~ (y D/D) 1/2 kT/(eh).
The thickness of the main part is, because of the small thickness of
267
Plasma Diagnostics
the adjacent regions, close to the thickness of the DL in the whole y D.

In the zone of the diffusion decrease, the concentrations of particles
are the same as regards the orders of magnitude as in the main part,
the drift transport of the attracting particles is compared with the dif-
fusion transport, the strength of the electric field is constant, the scale
of the thickness of this zone is equal to h(D/y D) 1/2 and considerably
smaller than y D as well as the scale of the thickness of the transition
region. Evidently, with decreasing parameter yD and with this parameter
becoming comparable with (h 2D) 1/3, the structure of the layer degen-
erates and the estimates transfer to corresponding estimates for the
case of a homogeneous DL.
These considerations show that the condition of fulfilling the first
inequality of (11.4) in all hydrodynamic zones is equivalent to the con-
dition lu << h(D/y D) 1/2 which means that in all zones f 0 is the Maxwell
function with temperature T.
In the case of a thin DL (y D << D), the intensity of generation of
the particles in the DL is low due to the assumption that the charac-
teristic duration of the volume reactions is comparable with the time
of flight. The fluxes of the charged particles on the probe and, con-
sequently, the probe current are determined by the fluxes from the
quasi-neutral region to the external boundary of the DL. We shall examine
the problems that describe the distribution of the concentration of the
charged particles in the region of non-viscous flow and the quasi-neutral
part of the viscous boundary layer. In the region of the non-viscous
flow, the distributions of the concentration of the charged particles are
described by the problem
nm
nvÑ = wm , (11.7)
n
and away from the probe
nm = nm¥ . (11.8)
It is important to stress that the electric field does not influence in

the first approximation the distribution of the concentration of the charged
particles in the region of non-viscous flow.
The system of equations described in the distribution of the con-
centration of the charged particles, the density of their fluxes and also
the strength of the electric field in the quasi-neutral part of the vis-
cous boundary layer have the form
268
nm ¶J my
nvÑ
n
+
¶y
= wm ; å zmnm = 0, (11.9)
m= i ,e
 kT ∂ nm 
J my = µ m  − n + z m nm E y  . (11.10)
 e ∂y n 
We shall formulate the boundary conditions for the given system of

equations. The concentration of the charged particles at the outer bound-
ary of the viscous boundary layer tends to the values resulting from
the solution of the problem (11.7), (11.8) on the probe surface. The
boundary conditions for the concentration of the charged particles on
the probe surface should be determined from combination with the as-
ymptotic expansion of the concentrations which holds in the transition
region (or in the DL as a whole, if it is homogeneous). As indicated
by the above considerations, expansion starts with the terms of the order
of n m¥ (h/D) 2/3 << n m¥ , and the quasi-neutral concentrations on the
probe surface can assumed to be equal to zero. The insufficient
boundary condition for the system of equations (11.9) and (11.10) can
be written in the form 52
K - q kTw 1
y ® 0: Ey = +...,
K +1 e y
−1
   
where K =  ∑
 m=i , e
J my / µm 

∑
 m=i
J my / µm  (the first and second sums

 − w  + w
are taken with respect to the components with the negative or posi-
tive charge, respectively), and all ions are assumed to be having the
same charge. Evidently, the quantity K determined in this manner can
assume only non-negative values.
It is important to stress that in the problem formulated here which
specifies the distribution of the concentration of the charged particles,
the densities of their fluxes and the strength of the electric field in the
quasi-neutral part of the viscous boundary layer, the effect of the probe
potential is manifested only through parameter K. At K ® 0 (K®¥),
i.e. when suppressing the flux of the particles with the negative (positive)
charge to the probe, this effect disappears and, consequently, the probe
current tends to some constant value which will be denoted I + (I –).
Thus, when the DL is thin, the probe current is restricted by the
limiting values I t, I –. If the probe current is assumed to be positive in
cases in which it is directed from the probe into plasma, then I + < 0,
269
Plasma Diagnostics
I– > 0. Evidently, |I+| ~ eD in e¥a 2/D, I – ~ eD e·n e¥a2/D. It should be noted

that |I + |/I –~ b << 1, where b is a parameter whose order of magni-
tude is equal to the ratio of the mobility of the ions to the mobility of
the electrons (this parameter usually does not exceed 10 –2).
Initial considerations enable the VAC of the probe to be analysed.
In the current range I + < I < I – the DL is homogeneous and the main
contribution to the total difference of the probe–plasma potential comes
from the region of non-viscous flow. Since the distribution of the con-
centration of charged particles in this region does not depend on the
probe potential, and the current transport is of the drift type, the VAC
of the probe is linear in this range.
In the range I > I – we have y D ~ D, and the contributions to the
probe–plasma potentials come from the region of non-viscous flow in
the DL. As regards the order of magnitude, the contributions are equal
to respectively (kT/e) a/D and (kT/e) D/h. The ratio between them
depends on the parameter ah/D 2 (it should be noted that the last pa-
rameter, as regards the order of magnitude, is equal to Re h/a or, which
is the same, Mh/li, where M is the Mach number of the incident flow).
In the case with ah/D 2 >> 1, the controlling contribution is the con-
tribution of the region of non-viscous flow, and the VAC remains linear.
When ah/D 2 >> 1 the contribution of the DL is dominant, and the char-
acteristic value of dI/dj w, which gives the slope of the VAC in rela-
tion to the potential axis, is s ¥a2h/D 2 (s is the conductivity of plasma),
which is considerably lower than the characteristic value of this quantity
in the range I + < I < I –, equal to s ¥a. In other words, when the probe
current reaches the value I – the slope of the VAC rapidly decreases.
Naturally, this variation of the slope is referred to as saturation and
the value of the current I – is the saturation current. When ah/D 2 ~ 1
the contributions of the region of non-viscous flow and the DL to the
total potential difference between the probe and the plasma in the range
of currents I > I – are comparable and the VAC has an intermediate
shape.
In the range I < I +, we have y D ~ D, and the contributions of the
region of non-viscous flow on the DL with respect to the order of mag-
nitude are b (kT/e) a/D, (kT/e)·D/h. Depending on the order of mag-
nitude of bah/D 2 the VAC is linear, although there is saturation at
I = I + or the VAC has the intermediate shape.
It should be noted that as a result of the non-uniformity of the prob-
lem the densities of the limiting currents in different areas of the probe
surface differ and are obtained at different values of the probe po-
tential. Because of the limited space, this effect is not discussed here,
see Ref.48.
270
These estimates show that the floating potential with the error of
the order kT/e coincides with the plasma potential. More accurate
analysis gives the following equation for the value of the floating po-
tential of the probe in relation to plasma (to a first approximation, i.e.
with the error to terms of the order of kT/e):
kTw  h  
2/3
ϕf = ln   β  . (11.11)
e  ∆  
Thus, the characteristic special features of the VAC in the exam-

ined case are: the presence of a linear section; the difference of the
saturation criteria in the regions of positive and negative potentials of
the probe, with the saturation criterion at positive potentials, i.e. on the
electronic part of the VAC, which has the form h << D 2/a 1, being con-
siderably more ‘rigid’ than the saturation criterion at negative potentials
(h << D 2/(ba), D); the fact that the floating potential of the probe is
close to the plasma potential. It should be stressed that, when discussing
the linear section, we are concerned with the relatively long part of
the VAC and not, for example, the small section of the VAC in the
vicinity of the inflection point. Saturation here refers not to the absence
of the dependence of the probe current on its potential but to the change
of the curvature of the VAC which is caused by a rapid increase of
the resistance of the near-probe region under the conditions in which
the capacity of diffusion as regards supply of charged particles to the
probe surface is exhausted and a further increase of current can take
place only as a result of increasing the thickness of the near-probe layer
of the volume charge.
Previously, it was assumed that the condition l u << h (D/y 0) 1/2 can
evidently be ‘weakened’: it is sufficient if l u << D. It is evident that
under the last condition, the hydrodynamic approximation remains valid
in the region of non-viscous flow, the quasi-neutral part of the viscous
boundary layer and (if y D ~ D) and the main part of the DL, i.e. in
these regions which determine the main elements of the VAC. Evidently,
if the given condition is fulfilled, the function f 0 in the region of non-
viscous flow and the quasi-neutral part of the viscous boundary layer
is a Maxwell function with temperature T.
It has been assumed that the electron concentration in non-perturbed
plasma is comparable with the concentration of positive ions and elec-
trons remain the main carriers of the negative charge. All these con-
siderations, with exception of equation (11.11), remain also valid in the
case of the plasma of the electronegative gas when the contribution
271
Plasma Diagnostics
of electrons to the transfer of the negative charge is not controlling.

In this case, parameter b should be set equal to unity, and the cur-
rents I + and I – have the same order of magnitude as the saturation cri-
teria in the region of positive and negative potentials (evidently, the
presence of the ions should decrease the asymmetry of the VAC; for
example, if there are no electrons, and the mobility of the negative and
positive ions are identical, then the function I (j w) is odd, i.e. there
is no asymmetry in the VAC).
Previously, we examined the case in which the plasma flows around
the probe under the regime of the chemically reacting non-equilibrium
viscous boundary layer. It may easily be seen, on the other hand, that
the proposed theory is also applicable for the case in which the flow-
around takes place under the regime of the viscous boundary layer but
is chemically ‘frozen’ (evidently, in this case, the source terms in equa-
tions (11.7) and (11.9) should be omitted).
The theory can also be applied to the case of stationary or mov-
ing plasma with rapid reactions where the reaction times are short in
comparison with both the time of flight a/v ¥ and the diffusion time of
the ions (or, which is the same as regards the order of magnitude, the
ambipolar diffusion term) in the near-probe a 2/D i. In this case, ana-
logues of the region of non-viscous flow and the near-probe viscous
boundary layer are the region of ionisation equilibrium and the near-
probe layer of non-equilibrium ionisation. The convective terms in the
equations (11.7) and (11.9) should be omitted. Instead of the param-
eter D, the estimates will include a linear scale characterising the re-
actions, for example, the recombination length d calculated from the
parameters of non-perturbed plasma (evidently, in this case, this scale
is considerably smaller than the probe diameter and, if the Reynolds
flow-around number is high, the thickness of the near-probe viscous
boundary layer). It is only important that the Debye radius remains small
in comparison with the previously mentioned scale, and the equilibrium
concentration of the charged particles in the vicinity of the probe is
not too small in comparison with that in the non-perturbed plasma.
In particular, the theory described previously is applicable to the case
of a spherical probe in stationary plasma containing neutral particles,
ions of the same kind and the electrons, with volume ionisation and
recombination under the condition that h << d << a (in this case a
is the probe radius). Figures 11.2 and 11.3 53 show the corresponding
electron and ion parts of the VAC, obtained by numerical solution of
the initial problem (11.1)–(11.3), (11.5), (11.6) (in this case, the problem
is, because of spherical symmetry, unidimensional and its numerical so-
lution is not difficult). In the calculations it was assumed that the ioni-
sation of the neutral particles is carried out by an external ioniser or
272
Fig.11.2 Electron branch of the VAC of a spherical probe in stationary plasma with
fast reactions (numerical calculations).
Fig.11.3 Ion branch of the VAC of a spherical probe in stationary plasma with fast
reactions (numerical calculations).
in collisions with neutral molecules; recombination – in collisions with

the neutral molecules or dissociative recombination; the plasma tem-
perature, the rate of volume ionisation, the coefficient of recombina-
tion and the mobility of the ions and electrons are constant in the near-
probe region; h/a = 10 –5 , b = 0.01. Both VAC branches are normal-
ised with respect to the values of the saturation currents computed from
the corresponding asymptotic equation. The normalised multipliers for
the probe potential j ± = I ± (4pas ¥) are selected taking into account
the asymptotic results in such a manner that the angle of inclination
of the linear section, predicted by the asymptotic theory, is 45°. Each
curve in Fig.11.3 is displaced along the potential axis by the value com-
puted from the asymptotic equation (11.11). Taking this normalisation
into account, it may be seen that the numerical calculations are in
complete agreement with the asymptotic results. The normalised values
of the floating potential for the probe of the given case, obtained by
the numerical solution, are given below.53 It may be seen that they differ
only slightly from the values determined from equation (11.11):
a /d 10 1 0 3 /2 102 1 0 5 /2 103 1 0 7 /2 104
ej f /k T
(nume ric a l - 10.8 - 10.0 - 9.20 - 8.43 - 7.66 - 6.86 - 6.03
so lutio n)
(11 . 11 ) - 10.7 - 9.98 - 9.21 - 8.44 - 7.68 - 6.91 - 6.14
273
Plasma Diagnostics
It is evident that the relationship between the Debye radius, the scales
∆ and d and the size of the probe at which the previously described
qualitative analysis is valid, can be written in the form h << l << a,
where l is the quantity which coincides with the smaller of the scales
D, d. In the case of stationary or slowly moving plasma with slow or
frozen reactions in the near-probe region, when the order of magni-
tude of the Reynolds number does not exceed unity, and the duration
of the reactions is not small in comparison with the diffusion time of
the ions in the near-probe region, the right-hand inequality is violated.
In this case, the VAC does not contain the distinctive linear section.
The saturation conditions in the region of positive and negative potential
coincide and have the form h << a. The VAC is situated at a large
distance from the origin of the co-ordinates; for example, for the probe
in stationary plasma with frozen reactions and constant temperature,
we have I (0) = (I + + I –)/2 (I (0) is the value of current on the probe
at the potential equal to the plasma potential).
Finally, in the case in which the concentration of charged particles
in the plasma is not high and the Debye radius is comparable with the
smallest of the scales D, d, a, there is no justification for expecting
the existence of a distinctive region of saturation at positive or negative
probe potentials. The conditions of existence of the linear section of
the VAC has the form l << a in this case.
These considerations show that the effect of movement of the
plasma on the VAC of the probe under the examined conditions is char-
acterised by parameter Re. Since, as already shown, the Schmidt
number of the ions under the normal conditions has the order of unity,
this number is naturally interpreted in the probe theory as the ratio of
the velocity of the incident flow to the diffusion rate of the ions in the
near-probe region. On the other hand, the order of magnitude of this
parameter is equal to the ratio of the Mach number of the incident
plasma flow to the Knudsen number l i/a. Since the Knudsen number
under the examined conditions is small, the influence of the convec-
tive effects under the hydrodynamic conditions becomes more significant
at considerably lower values of the velocity of the incident flow than
under collisionless conditions. For example, for a probe with a char-
acteristic size of 1 cm in the plasma of slightly ionised air at 3000 K
and at the atmospheric pressure, the movement of plasma must be taken
into account already from the velocities of the order of units of cen-
timetres per second (under these conditions, the Reynolds number be-
comes equal to unity at v ¥ » 8 cm/s).
The influence of the convective effects on the VAC of the probes
under the hydrodynamic conditions is, as indicated previously, of the
dual type. On the one hand, the movement of plasma results in a general
274
increase of the current on the probe caused by an increase of the dif-

fusion rate of the charged particles to the probe surface resulting from
a decrease of the thickness of the region in which this diffusion is lo-
calised. On the other hand, at relatively fast movements (Re>>1) the
form of the VAC changes: a linear section appears, and so on. Evi-
dently, this change is caused by the transformation of the structure of
the near-probe region of perturbation: the quasi-neutral region becomes
of the two-scale type, i.e. includes two regions with greatly differing
scales (the region of non-viscous flow and the quasi-neutral region of
the viscous boundary layer). The volume reactions in the near-probe
region have the same dual effect.
11.4 Diagnostic methods

As shown by the results of qualitative analysis described in the pre-
vious section, the physics of the passage of current in the near-probe
region of high-pressure slightly ionised plasma is relatively complicated.
It is not surprising that there are no universal solutions of the prob-
lem (11.1)–(11.3), (11.5), (11.6) that would enable the diagnostic methods
to be developed for the entire range of hydrodynamic conditions.
The problem of numerical solution has been examined with special
reference to unidimensional problems (see Ref.8 and the studies quoted
in there, and also Ref.20, 53–63 and Chapter 10 of this book). At
present, effective numerical methods have been developed for these
problems and enable the influence of different physical effects, such
as thermal diffusion,20 high-intensity volume ionisation and recombina-
tion, 53,58 the emission of electrons from the probe surface, 8,59,63 the
presence in the plasma of negative ions in addition to neutrals, posi-
tive ions, electrons 8,61,62 to be considered.
There are only a few studies concerned with the accurate numerical
solution of the problems of the hydrodynamic theory of electrical probes
in non-unidimensional formulations. In addition to the studies quoted
in Ref.6 and 8, we should mention the studies in Ref.64 and 65 which
present the results of two-dimensional numerical calculations for the
case in which the concentration of charged particles in the plasma is
not high and the Debye radius is comparable with the probe size.
Evidently, in this situation, the methods of accurate numerical so-
lution can be used efficiently as the main means of analysis of the VAC
in the undimensional problems, for example, the problems of spheri-
cal, double cylindrical (in the form of two coaxial cylinders) and double
flat probes in stationary plasma. It should be noted that when construct-
ing the numerical algorithms, it should be taken into account that at
low values of the Debye radius the Poisson’s equation is not suitable
for determining the potential 58,66 because the only term of the equa-
275
Plasma Diagnostics
tion which contains the potential is small in the main part of the plasma
volume. In non-unidimensional problems, including the problems of
probes in moving plasma, it is more efficient to use the approximate
approaches which enable specific regions of the VAC to be calculated.
On the other hand, the degree of accuracy and the region of appli-
cability of these approaches can be efficiently evaluated using the
numerical solutions of simulation unidimensional-problems.
Qualitative analysis, described in the previous paragraph, leads to
three relatively simple approaches within which we can calculate ap-
proximately either the linear section of the VAC or the sections of
currents of positive or negative particles I < I + , I > I – (in this case,
we take into account the voltage drop only in the main part of the DL)
or the saturation current of the positive or negative particles I+, I–. Fig-
ure 11.4 shows schematically the VAC of the probe (solid line) and
the linear section (line 1), the sections of the currents of positive or
negative particles (lines 2, 3, respectively) and the values of the satu-
ration currents, calculated using these approaches. These approaches
will now be discussed in greater detail.
The first approach. To calculate the linear section of the VAC,
it is initially necessary to find the distribution of the concentration of
the charged particles in the region of non-viscous flow (or, in the case
of plasma with fast reactions, regions of ionisation equilibriurm) de-
scribed by the problem (11.7), (11.8). Subsequently, in this region we
determined the distribution of the conductivity of plasma and solve the
problem for the distribution of the potential, neglecting the voltage drop
in the near-probe viscous boundary layer (the non-equilibrium ionisa-
tion layer)
∇ ⋅ (σ∇ϕ ) = 0. (11.12)
Fig.11.4 VAC of the probe and its elements, calculated using approximate approaches.
276
On the surface of the probe
j = jw , (11.13a)
j = 0. (11.13b)
In a number of cases, changes of the pressure and temperature of

plasma in the examined region are small. For example, for stationary
or slowly moving plasma, this situation is realised in the case of a non-
cooled probe. If the Reynolds flow-around number is high, such as in
the case in which the Mach number of the flow is moderate (for ex-
ample, for a diatomic gas at M £ 0.6 the changes of pressure and tem-
perature in the region of non-viscous flow do not exceed approximately
28 and 7%, respectively). For trans- and supersonic flows, this situ-
ation occurs in particular in the case of a thin probe, for example, in
the form of a disc or flat sheet. If the flow is hypersonic, M >> 1,
and the probe is ‘blunted’, the extent of withdrawal of the head shock
wave, formed in front of the probe, is often not large and the controlling
contribution to the resistance of the region of non-viscous flow comes
from the non-perturbed region in front of the shock wave.
In these cases, the changes of the concentration of the charged par-
ticles and, consequently, plasma conductivity in the examined region
are not large. Cases can also occur in which the variations of the pres-
sure or temperature in the given region are large but they do not cause
any change of conductivity. In particular, in cases in which the flow-
around the probe is chemically ‘frozen’, the molar fractions of the
charged particles in the region of non-viscous flow are constant; if the
value µ en in the examined temperature range is constant (this approxi-
mation is satisfactory for, for example, the plasma of combustion prod-
ucts), the conductivity of plasma in the region of non-viscous flow is
constant regardless of the dependence on M.
Since the conductivity of plasma in the given region is constant, the
proposed approach used in these cases is greatly simplified: equation
(11.12) becomes the Laplace equation, and the expression describing
the linear section of the VAC has the form
I
= 4 pCs ¥ , (11.14)
jw
277
Plasma Diagnostics
where C is the electrical capacitance of the probe. This quantity de-

pends only on the form of the probe and its value is well known for
the probes of the simplest configurations. In particular, for a spheri-
cal probe with radius a we have C = a; for a probe in the form of
a thin cylinder with height H and radius R (R << H) we have C =
H/[2 ln (H/R)]; for a disc-shaped probe with radius a it is C = 2a/
p; the equations for the probe in the form of an ellipsoid of revolu-
tion are also known (see, for example, Ref.67, 68); for a wall probe
in the form of a circle with radius a, positioned flush on the non-con-
ducting plane, C = a/p; for a hemispherical probe with radius a po-
sitioned on a non-conducting plane, C = a/2, and so on. It should be
noted that using the results published in Ref.69, the equation (11.14)
is generalised to the case of a probe in a magnetic field.
As indicated by the results obtained in the previous paragraph, the
discussed approach can be used at h Ù l << a; l u << l. It should also
be noted that when l u is comparable with the scale l and the last
inequality is violated, function f 0 in the examined region will, gener-
ally speaking, differ from the Maxwell function and will depend on the
local value E/n and, consequently, on the probe potential. Therefore,
in the given case the analysed section of the VAC is no longer lin-
ear. On the other hand, we can attempt to calculate this section on
the basis of the combined solution of the problems (11.7), (11.8) and
(11.12), (11.13).
The second approach. In the majority of studies carried out within
the framework of the second and third approach, it is assumed that
the plasma contains charged particles of two types – electrons and posi-
tive ions of the same type, and we recalculate the region of the ion
current of the VAC or ion saturation current. The condition of appli-
cability of these approaches is written in the form of inequalities h,
l u << l, a; bha << l 2.
To calculate the region of the section of ion current, it is neces-
sary to find initially the distribution of the concentration of the charged
particle in the quasi-neutral region and determine the value of the ion
flux on the outer boundary of the DL as the function of the thickness
of the DL. This is followed by solving the problem in the DL, and the
diffusion transport of the charged particles is ignored in comparison
with the drift transport and we determine the voltage drop in the DL
which is regarded as identical with the total difference of the probe-
plasma potentials. It should be noted that the function I (j w), obtained
within the framework of this approach, which describes the region of
ion current of the VAC, assumes the value I + if j w = 0 (Fig.11.4).
When solving the problems in the DL in the case in which the in-
equality l i << h is not fulfilled, it may be necessary to take into ac-
278
count the dependence of the mobility of the ions on the local value of
E/n. The electron concentration in the DL is considerably lower than
the ion concentration. However, in certain situations it is essential to
take into account the presence of the electrons in the DL (in particular,
as a result of the high energy of the electrons an important role is played
by ionisation by the electron impact, for example, Ref.70); it should
therefore be noted that the dependence of function f 0 in the DL on
the local value of E/n becomes significant at l u Ú h.
This approach will be illustrated on an example of a spherical probe
(discussed in the previous paragraph) in still plasma with volume ioni-
sation and the recombination and constant properties in cases in which
h, l u << d << a, and within the framework of the model of the ki-
netics described previously when examining Figs. 11.2 and 11.3. In the
region of ionisation equilibrium, the concentration of the ions and the
electrons is constant and equal to that in the non-perturbed plasma:
n i = n e = n i¥ = n e¥ (here and when examining the model of the plasma
with the charged particles of two types – positive ions and electrons
- the ion or electron concentration in non-perturbed plasma will be de-
noted by n c¥: n c¥ = n i¥ = n e¥). In the quasi-neutral part of the near-
probe layer of non-equilibrium ionisation the distribution of the concen-
tration of charged particles n c (n c = n i = n e) is determined by the prob-
lem
d 2 ns
Da = k r nc2 - S , (11.15)
dy
y = y D : nc = 0; y ® ¥: nc ® nc¥ , (11.16)
where D a = 2D i D e/(D i + D e) » 2D i is the coefficient of ambipolar dif-

fusion, S is the volume ionisation rate, k r is the recombination coef-
ficient (evidently, k r n 2c¥ = S). In accordance with the assumption
d << a the curvature of the probed surface is not taken into account
here.
Equation (11.15) will be multiplied by dn c/dy and integrated. Tak-
ing into account the second boundary condition (11.16) gives
FG IJ
Da dnc
2
kr 3 2
2 dy H K =
3
nc - Snc + Snc¥ .
3
(11.17)
Dividing the equation (11.13) for the ions and electrons by D i, D e,
279
Plasma Diagnostics
respectively, and combining the results, in the quasi-neutral approxi-

mation we obtain
Ji J 2
+ e = − ∇ ( nc T ) . (11.18)
Di De T
Consequently, for the examined model, using (11.17) and taking into
account the first boundary condition (11.16), the determined relation-
ship between the fluxes of the ions and electrons on the outer boundary
of the DL:
Ji ( y D ) J e ( y D ) n
= - 2 c¥ ,
FG 3D IJ 1/ 2
+ d=
H 4k n K
a
. (11.19)
Di De d r c¥
The problem in the DL has the form (ignoring the effect of the cur-
vature – this is justified when the absolute value of the probe poten-
tial is not too high and y D << a):
dJ i dJ e dE
= S, = S, = 4pe(ni - ne ), (11.20)
dy dy dy
J i = m i ni E , J e = - m e ne E , (11.21)
y = 0: J e = 0; y = y D : E = 0. (11.22)
When writing the equations of conservation of the charged parti-

cles, recombination was ignored because the concentration of these par-
ticles in the DL was low. The first boundary conditions indicates that
the probe is under the high negative potential and the electron flux on
its surface is suppressed, the second condition corresponds to neglecting
the electric field at the outer boundary of the DL in comparison with
the values inside the layer. The equation (11.19) is the insufficient
boundary condition.
If the dependence of the mobility of ions and electrons on the strength
of the electric field is ignored, the problem (11.19)–(11.22) is easily
solved. In particular, the solution shows that n e /n i ~ b << 1, and, con-
sequently, the presence of the electrons in the DL can be ignored with
the error of the order of β. In other words, the second term in the
left-hand part of equation (11.19) and the second term in the brack-
280
ets in the right-hand part of the third equation (11.20) are insignificant
and can be omitted. Consequently, the problem (11.19)–(11.22) is split
up: the functions n i, J i, E are determined independently of the elec-
tron distribution, and this is followed by determining n e, J e; the solu-
tion obtained using this procedure coincides with the error b with the
solution of the complete problem (11.19)–(11.22).
The equations for the dependence I (y D) and also the equation for
the dependence j w (y D), obtained by integrating the solution for the
electric field in the range from 0 to y D, have the form
b
I = -4pea 2 SyD + 2 Di g nd ,
c¥
(11.23)
hdFF
16 6 p 3y I
jw = -
H 8d K
D
enc¥ (11.24)
9
with the function F = F (u) is given by the equation
af a fd i
1/ 2 LM d i
1/ 2 OP
F = F u = 2 2u + 1 u 2 + u
N
- ln 2 u 2 + u
Q
+ 2u + 1 .
These equations represent the parametric description of the section

of ion current of the VAC. The terms in the brackets in the right-hand
of (11.23) are naturally interpreted as the contributions to the probe
current of the generation of charged particles in the DL and of the
ion diffusion from the quasi-neutral region. Evidently, the absence of
the dependence of the flux from the quasi-neutral region on the probe
potential is associated with neglecting the effect of the curvature and
by the fact that the plasma properties in the near-probe region are
constant within the framework of the examined model. The influence
of the curvature effects under the conditions in which the generation
of charged particles in the DL is ‘frozen’ was analysed in Ref.71–73
(the authors of Ref.72 also took into account the dependence of the
mobility of ions on the local strength of the electric field). Under the
conditions in which the generation in the DL is the controlling proc-
ess, the influence of the curvature effects was examined in Ref.74.
The influence of the variability of the plasma properties has been dis-
cussed in Ref.75 and 76.
The authors of Ref.77 and 78 calculated the regions of the ion cur-
rent on the VAC of wall probes on sheets and also conical probes, in
plasma flows with frozen ionisation and recombination at Re>> 1. In
addition to these studies where the authors examined the case where
281
Plasma Diagnostics
yD >> D (the so-called sheath–convection regime); the appropriate stud-

ies have been reviewed in Ref.15 (see also Ref.79–83). A distinguishing
feature of the last case is the simple calculation of the ion flux on to
the outer boundary of the DL: in this case, the flux is equal to n i¥v ^
(v ^ is the projection of the mean mass velocity on the normal to the
upper boundary of the DL). To explain this special feature, we shall
examine briefly the buffer zone dividing the region of the non-viscous
flow and the main part of the DL. It will be assumed that h << D <<
yD Ù a, and we shall take into account the results published in Ref.84.
At hy D /D 2 << 1, the zone is identical with that in the case where
yD ~ D and includes the quasi-neutral diffusion region in which the con-
vective, drift and diffusion transports of the charged particles are com-
parable, and the transition region in which the non-quasi-neutrality is
significant and the intensity of convective transport is low. Taking into
account that v^ ~ v¥ yD /a, we shall determine the scale of the thickness
of the regions: D 2/y D, (h 2D 2/y D) 1/.3. At hy D /D 2 ~ 1, the scales coincide:
the buffer zone is homogeneous and all the previously mentioned ef-
fects are simultaneously significant in this zone. At hy D /D 2 >> 1, the
buffer zone again becomes inhomogeneous and includes the layer of
decrease of the electron concentration in which the electrons are repulsed
by the probe fields, the concentration of the ions remains constant, and
the convective-drift layer in which the ion concentration decreases from
n i¥ to the values corresponding to the main part of the DL (it is as-
sumed that by D h/D 2 << 1). The scales of the thickness of the layers
are equal to h, h 2 y D/D 2.
Using these estimates, it can be easily be shown that the flux of
the ions across the buffer zone is considerably stronger than their con-
vective transport in the longitudinal direction. Therefore, the density
of the ion flux across the buffer zone changes only slightly and is de-
termined by the convective and drift fluxes at distances from the zone
considerably greater than its thickness. Finding the electric field from
the condition of suppression of the electron flux, we obtain that the
density of the ion fluxes n i¥ v (1 + m i/m e), and since µ i/ µ e is low we
obtain the expression described previously.
The voltage drop in the DL at DÙ y D Ù a has the order
(kT/e) y 2 D /(hD) (for comparison, we shall show that at
(h 2D) 1/3 Ù y D Ù D the drop is of the order of (kT/e) y D3/2 (hD 1/2)). To
neglect the voltage drop in the region of non-viscous flow, it must be
that bah/(Dy D) << 1. If y D << a, the DL is treated as locally uniform;
if y D ~ a, non-unidimensional effects must be taken into account.
The third approach. The saturation current is determined by solving
the problem of the distribution of the concentration of charged parti-
cles in the quasi-neutral region. It is important to stress that the specific
282
features, associated with the presence of the electric field in the plasma,
are not very important in this problem and the formulation of this prob-
lem is similar to the formulation of the well-known problem of the flow
of a mixture of neutral gases around the solids. To solve these prob-
lems, the gas dynamics and heat and mass exchange theory offer suf-
ficiently developed methods, both analytical and numerical.
Initially, we examine the model of plasma with two types of charged
particles (electrons and positive ions). Assuming that the electron flux
on the probe surface (or, more accurately, on the outer boundary of
the DL) is equal to zero, the equation (11.18) gives the relationship
between the density of ion saturation current and the quasi-neutral con-
centration of the charged particles
FG ¶nc IJ
j+ = -2e Di
H ¶y K w
. (11.25)
From equation (11.18) we exclude the density of the electron flux

using the equation J e = J i – j/e (j is the density of electric current).
Expressing J i from the resultant equation and substituting this result
into (11.1) for the ions, we obtain the equation for n c
nc D LM b gOPQ + ej Ñ D D+ D
nvÑ
N
- Ñ a Ñ nc T = wc ,
i
(11.26)
n T i e
where w c is the rate of variation of the concentration of the ions or

electrons as a result of volume ionisation or recombination.
The corresponding boundary conditions have the form:
on the probe surface
nc = 0, (11.27a)
nc = nc¥ . (11.27b)
It may be assumed that under the ion saturation regime the den-
sity of the electric current in the entire near-probe region of pertur-
bation is equal to the right-hand part of (11.25) (as regards the order
of magnitude). Since the third term in the left hand part (11.26) has
the order b in relation to the second part, the third term can be ig-
283
Plasma Diagnostics
nored. Consequently, the problem (11.26) and (11.27) becomes closed:

solving this problem and using equation (11.25), we determine the density
of saturation current. It is interesting to note that in the case in which
the coefficients of the diffusion of ions and electrons depend on plasma
temperature in the same manner, the distribution of the concentration
of charged particles in the quasi-neutral region is independent of the
probe potential not only in the saturation regime but also in the entire
potential range in which the DL remains thin (the third term in the left-
hand part of the (11.26) is actually equal to zero).
When volume ionisation and recombination are ‘frozen’, equation
(11.26) (here and in further references to this equation it is assumed
that the third term in its left part is omitted) is linear and the satura-
tion current is proportional to the concentration of ions or electrons
in non-perturbed plasma n c¥. Analysis of the results of a large number
of experiments shows that the effect of changes in the surface tem-
perature of the probe of the ion saturation current on the probe in still
plasma or plasma moving at a moderate Mach number is quite weak.
This effect has been observed many times in experiments. It is caused
by the fact that the changes of saturation current, caused by the change
of the field of gas-dynamic parameters (primarily gas density) and by
changes of the diffusion coefficient of ions, compensate each other to
a first approximation. In this situation the ion saturation current is
calculated to a first approximation ignoring the changes of the trans-
port properties of plasma in the near-probe region, and the overall ex-
pression for the current on the probe of the given geometry is writ-
ten in the form
I + = -Sh enc¥ Di¥ a. (11.28)
The dimensionless coefficient Sh (Sherwood number) introduced here,

is a function of Schmidt ambipolar number Sc = n ¥/D a¥ and of the dif-
fusion Peclet number Pe = Re Sc = v ¥a/D a¥ . The coefficient is de-
termined using either the experimental results or by solving the approxi-
mate hydrodynamic problem. In particular, at Pe << 1, if the charac-
teristic size of the probe is represented by its electrostatic capacitance,
the equality 68 is fulfilled with the accuracy to terms of the order
Pe 2 ln (Pe –1 )
Sh = 4p (2 + Pe). (11.29)
It is interesting to note that this equation makes it possible to de-

termine the concentration of charged particles in non-perturbed plasma
284
using the measured dependence of the ion saturation current on the

speed of the probe in relation to plasma without using information on
the values of the diffusion coefficient under the given conditions:
b g
d - I+
= 2peC 2 nc¥ .
dv¥ v¥ = 0
At Pe >> 1 when plasma flows around the probe in the regime of

the viscous boundary layer, Sh » FÖPe, and equation (11.28) can be
written in the form
I+ = -
F
2
d
enc¥ v¥ Di¥ a 3 i
1/ 2
, (11.30)
where F = F (Sc) is the dimensionless coefficient which depends on

Schmidt’s number. The studies concerned with applying equation (11.30)
to probes of different geometry are reviewed in Ref.17. In particu-
lar, the distribution of the density of ion saturation current along the
‘windward’ surfaces of a sphere and a thin cylinder can be calculated
using the equations 85
a F
2 a I -1/ 2
j+ (a ) = j+ (0) cos 2
2 H
1 - sin 2
3 2 K ,
1.3 F
v D I 1/ 2 (11.31)
j+ (0) = -
Sc 1/ 6 H
enc¥ ¥ i¥
a K ,
a 11
. v D F I 1/ 2
i+ (a ) = j+ (0) cos ,
2
j+ (0) = - 1/ 6 enc¥ ¥ i¥
Sc R H K , (11.32)
where a is the polar angle counted from the direction opposite to the
vector of the speed of the non-perturbed incident flow (i.e. from the
front critical point).
These equations can be used to calculate the integral ion satura-
tion current on spherical and cylindrical probes whose collecting surfaces
are represented by any segments of the ‘windward’ surface. In par-
ticular, when the collecting surface is the entire ‘windward’ surface
(i.e. the front hemisphere or front half cylinder respectively), the in-
tegral saturation currents are described by the equation
285
Plasma Diagnostics
I+ = -
6.7
Sc1/ 6
d
enc¥ v¥ Di¥ a 3 i
1/ 2
, I+ = -
3.2
Sc1/ 6
b g
enc¥ v¥ Di¥ R
1/ 2
H. (11.33)
When the entire surface of the sphere or cylinder is collecting, theo-

retical determination of the contribution of the ‘leeward’ surface is as-
sociated with difficulties due to taking into account the separation of
the boundary layer and formation of dead zones. In this situation this
contribution can be evaluated on the basis of an analogy between mass
and heat exchange and using the experimental data for the Nusselt
number. Using the semi-empirical equations for the Nusselt number,
presented for the sphere and cylinder in Ref.86, 87, the integral points
on the total surface of the probe are described by the relationships similar
to (11.33) where the first factors in the right hand part are replaced
by the factors 6.8 and (3.3/Sc 0.13 ). Using semi-empirical equations
published in Ref.88 gives equations that are also identical (11.33) but
the numerical coefficients 6.7 and 3.2 are replaced by 8.3 and 3.3. Tak-
ing into account that the above equations are similar to (11.33), it can
be concluded that the contribution of the rear hemisphere or rear half-
cylinder to the integral saturation current is not large. It should be noted
that this conclusion is confirmed by the results of measuring the dis-
tribution of the density of ion saturation current around the circumference
of the cylindrical probe described in Ref.89, 90, see also Ref.91. It
should be noted that the contribution of the ends to the integral saturation
current on the cylinder is also insignificant under these conditions.
The distribution of the density of ion saturation current along the
surfaces of a flat sheet or cone is calculated using equations 6
0.47 v DF I 0.81 FG
v D IJ 1/ 2
Sc xH
j+ ( x ) = - 1/ 6 enc¥ ¥ i¥ ,
K Sc xH
j+ ( x ) = - 1/ 6 enc¥ xd i¥
K , (11.34)
where x is the distance from the edge of the sheet or tip of the cone,
v xd is the speed at the external boundary of the boundary.
The density of saturation current at the critical point of a ‘blunted’
solid is determined from the equation 6
2 e × 0.66 LM
dv xd OP
1/ 2
j+ ( 0) = -
Sc 1/ 6
enc¥
N
dx
(0 ) Di¥
Q , (11.35)
where dv xd /dx (0) is the value at the critical point of the derivative
of the speed at the outer boundary of the boundary layer, e = 0 or
286
Fig.11.5 Normalised ionic saturation current on the spherical probe in a subsonic flow.
solid line - calculations; points - experiments in Ref.122(1), 151(2), 153(3).
Fig.11.6 Normalised ionic saturation current, a cylindrical probe in a subsonic field .

solid line – calculation; points – experiments. Ref.52 (1), Ref.90 (2), Ref.113 (3), Ref.118
(4), Ref.148 (5), Ref.149 (6), Ref.150 (7), Ref.151 (8), Ref.152 (9), Ref.153 (10),
Ref.154 (11), Ref.155 (12)
e = 0.5 respectively, in cases in which the flow-around the probe is

two-dimensional or axisymmetric. It should be noted that the value of
dvxd/dx (0) for spherical and cylindrical probes is equal to 3vx¥/(2a) and
2v ¥/R, respectively, and equatioan (11.35) for these probes is similar
to the second equation of (11.31), (11.32) (the numerical coefficients
287
Plasma Diagnostics
1.3 and 1.1 are replaced by the coefficients 1.2 and 0.94).
To calculate the saturation current at Pe ~ 1 it is necessary, strictly
speaking, to find (numerical) solution of the elliptical problem (11.26),
(11.27). On the other hand, the approximate analytical expression for
Sherwood’s number for this case is obtained 92 by algebraic interpo-
lation starting with the asymptotics (11.29), (11.30):
4 pFPe
Sh = 8p + . (11.36)
F + 4 p Pe
Note that in the case of a spherical probe (for which in accord-

ance with the first equation (11.33) it can be assumed that F =
9.5Sc –1/6) this equation agrees with the accuracy to 20% with a similar
equation derived on the basis of the correlation equation for Nusselt’s
number published in Ref.86.
In the case of a probe in the form of a cylinder the condition of
validity of equation (11.29) has the form v ¥H/D a¥ = Pe h << 1 (if the
probe is characterised by several (not 1) greatly differing dimensions,
the criterion of validity of equation (11.29) is the smallness of the Peclet
number estimated from the largest of these dimensions). On the other
hand, the validity of the second equation (11.33) (which is the vari-
ant of equation (11.30) for a cylindrical probe) is limited by the range
of high values of the Peclet number estimated from the radius v ¥R/
D a¥ = Pe R >> 1. In this situation it is difficult to hope for a rational
accuracy of the interpolation (11.36). Therefore, for approximate cal-
culations of the integral ion saturation current on the thin cylindrical
probe whose entire surface is collecting, the following procedure is rec-
ommended at Pe R of the order of unity. 99 The second equation (11.33)
j + , µA/cm 2
Ref.11.7 Density of saturation current at a

critical point of a cylindrical probe in the flow
of plasma of combustion products with an
addition of potassium in relation to the
concentration of potassium atoms in non-
perturbed plasma. Lines - calculations: T w =
800(1,3), 400 K(2,4); points - experiments.
288
is used, in accordance with the data in Ref.87, at Pe R > 20. For the
range 0.5 < Pe R < 20, taking into account the correlation relationship
for the Nusselt number 87 the following equation holds
6.3
I+ = - 0.03
PeR0.4 enc¥ Di¥ H . (11.37)
Sc
Finally, for the range R/H << Pe R £ 0.5
13enc∞ Di∞ H
I + =− .
( )
ln Pe−R1 + 4.2 + 0.8 (11.38)
This formula is in agreement with the two-term asymptotic expansion

of the Sherwood number with respect to parameter 1/ln (Pe R–1 ) and
agrees at the point Pe R = 0.5 with equation (11.37).
It should be stressed that in the equations (11.29), (11.31)-(11.35),
(11.37), (11.38) the dependence on the ambipolar Schmidt number Sc
is either not present or is very weak. It should be assumed that it also
remains weak in the general case, i.e. the Sherwood number in (11.28)
depends on approximately only on Peclet number. In other words, the
quantity I+/(enc¥aD i¥ ) for the probe of the given geometry is a universal
function of the Peclet number. These functions for the spherical and
cylindrical probes calculated using equations (11.36) and the second
equation (11.33) and equations (11.37), (11.38) are presented by solid
lines in Fig.11.5 and 11.6. 93
The validity of the first equation (11.34) is not limited by the as-
sumption on the moderate values of the Mach number of the incident
flow. If the index ¥ in the second equation (11.34) and in equation
(11.35) is replaced by the index d (given to the values of the corre-
sponding functions at the external boundary of the boundary layer), the
validity of the equations is also not limited by this assumption; of course,
in the case trans- or supersonic flows these equations link the satu-
ration current with the parameters not in the perturbed plasma but at
the outer boundary of the boundary layer. The dependence of n c¥ of
the ion saturation current of the spherical probe in the flow with high
M and moderate Re, when plasma flows around the probe in the re-
gime of the viscous impact layer, were obtained in Ref.94, 95; the
authors of Ref.96 presented the results of numerical calculations of
the distribution of the concentration of charged particles and the po-
tential at the axis of symmetry of the viscous impact layer.
289
Plasma Diagnostics
We have examined the situation in which the recombination length

is considerably greater than the smaller of the scales a, D and the volume
ionisation and recombination in the near-probe region are ‘frozen’. If
this equality is not fulfilled, the ionisation recombination must be taken
into account. Equation (11.26) becomes non-linear (due to the non-linear
dependence of the recombination rate on the concentration of charged
particles) and the current is no longer proportional to the concentra-
tion of ions or electrons in non-perturbed plasma n c¥.
At the limit of the high rate of the reactions (d << a; D) the equation
(11.26) is solved analytically. 97,98 However, the solution links the satu-
ration current not with n c¥ but with some value of the equilibrium con-
centration in the perturbed region so that the saturation current can
be used to determine n c¥ only under conditions in which the equilib-
rium concentration in the vicinity of the probe is constant (as an ex-
ample of the corresponding diagnostic equation we shall mention the
equation (11.23) at y D = 0).
Calculations of the saturation current on the probes with plasma flow-
ing around them under the regime of the viscous boundary layer with
non-equilibrium volume ionisation and recombination (d ~ D) were cal-
culated in Ref.52, 99–103. In all these studies, with the exception of
Ref.101, the authors used numerical methods; in Ref.101, calculations
were carried out using the approximately linear (with respect to the
concentration of charged particles) equation for the rate of recombi-
nation. As an example, Fig.11.7 52 shows the results of calculations of
the density of ion saturation current at a critical point of a cooled
cylindrical probe in the flow of the plasma of combustion products with
a potassium addition: the broken line was calculated using the second
equation (11.32), lines 1 and 2 - numerical calculations taking into account
ionisation and recombination in the volume of the boundary layer (these
calculations, the only type of ions present in the plasma were the ions
of atoms of the addition; in particular, the quantity n c¥ in the second
equation (11.32) for each value of the concentration of free potassium
atoms in non-perturbed plasma n k¥ was computed using Saha's equa-
tion). Evidently, taking into account volume reactions reduces the rate
of increase of saturation current with increasing concentration of the
charged particles in non-perturbed plasma n c¥ (this concentration is
proportional to nk¥ ). The reasons for this effect is clear: in the case
of a cooled probe in plasma with thermal ionisation the ionisation rate
in the vicinity of the probe is low and the current is influenced to some
extent by recombination. Therefore, with increasing n c¥ the recombi-
nation time decreases and the current increases slower than n c¥. The
recombination time increases with increasing temperature and this effect
290
weakens. This also explains the appearance of the dependence on the

surface temperature of the probe T w.
It should be noted that for a non-cooled probe an increase of n c¥
decreases the spatial scale of the measurement of the concentration
of charged particles at a probe surface proportional to the recombi-
nation length. Consequently, the current increases more rapidly than
n c¥ (for example, within the framework of the model corresponding to
the equation (11.23) the saturation current is proportional to n 3/2c¥).
We have discussed the effect of volume reactions on the value of
the ion saturation current under the conditions in which the distribu-
tion of plasma temperature in the vicinity of the probe is monotonic.
The equation for the saturation current on to a wall probe on a sharp
cone under the conditions in which the local Mach number is high and
the temperature distribution in the boundary layer contains a maximum
to which ionisation takes place was presented in Ref.6 and derived more
accurately in Ref.104.
Until now, we examine the model of plasma with two types of
charged particles. In cases in which there are large numbers of charged
particles of more than two types, to determine the saturation current
of the particles with a positive (negative) charge it is necessary to solve
numerically a system of equations (11.1), (11.3) with an additional con-
dition of quasineutrality, with the first boundary conditions (11.5), (11.6)
and the boundary condition
kTw 1  kT 1 
y → 0 : Ey = − + ...  E y = w + ...  .
e y  e y 
As an example illustrating the effect of the presence of negative

ions on the level of the saturation current of the positive ions, the lines
3 and 4 in Fig.11.752 show the results of numerical calculations in which
a number of negative ions, especially HCO –3 ions, were considered in
addition to the K + ions and electrons. It can be seen that the effect
of the negative ions under these conditions is relatively strong, in par-
ticular: the range of concentration of the addition atoms in which the
saturation current is proportional to nc¥ is considerably wider; saturation
current generally increases; the dependence of saturation current on
the surface temperature of the probe is greatly weakened. In Ref.5,
on the basis of the analysis of the calculation results it was established
that an important role in the mechanism of this effect is played by both
the presence of the negative ions in the non-perturbed incident flow
and by the ‘sticking’ of electrons to the neutral molecules in the volume
of the boundary layer. It is interesting to know that the difference of
291
Plasma Diagnostics
the dependence j + (n k + ¥), determined within the framework of this

model, from a similar dependence calculated using the second equa-
tion (11.32) (with n c¥) replaced by n k + ¥) is relatively small – around
3040% at n k¥ Ù 10 14 cm –3 .
The authors of Ref.105 calculated numerically the ion saturation cur-
rent onto a flat probe in a still reacting plasma within the framework
of a model which takes into account the presence in plasma of elec-
trons and two types of positive ions.
Previously, it was assumed that the degree of ionisation of plasma
is relatively small and the frequencies of elastic collisions of charged
particles with the neutral particles greatly exceed the frequency of
collisions of charged particles together. However, it may be expected
that the validity of a number of the previously examined diagnostic results
is not limited by this assumption. In particular, taking into account ion-
electron collisions, the authors of Ref.48 substantiated the first of the
previously examined approaches, including (11.14). Similarly, the equation
(11.18)104 considering the ion–electron collisions is also valid (the terms
of the equations of ion and electron transfer, taking into account ion–
electron collisions, are shortened when deriving (11.18). Consequently,
the method of calculating the ion saturation currents was in the frame-
work of the model of plasma with two types of charged particles, based
on the equations (11.25)–(11.27) is valid, together with subsequently
derived diagnostic equations. It should be noted that, in some cases,
the equations for the ion saturation current, obtained within framework
of the previously examined model in which the function of the energy
distribution of the electrons in the quasi-neutral region – Maxwell’s func-
tion with the temperature of heavy particles, are approximately accurate
also in the framework of the model with the ‘frozen’ electron tempera-
ture. 6
11.5 Special features of experimental procedure

The procedure of the probe experiment has been sufficiently devel-
oped with respect to both measuring systems and probe design. 2-5,7,8,16
In particular, there are systems in which a computer is used for con-
trolling the experiment, and collecting, processing and storing the re-
sults (see, for example, Ref.106, 108).
We shall examine briefly some of the problems often arising in car-
rying out probe experiments in high-pressure plasma. In many situa-
tions strong heat flows travel to the probe from plasma and the probe
surface has a tendency to heat up rapidly during the experiment. This
heating often causes a number of undesirable consequences. Firstly,
it results in intensive thermal electronic emission from the probe surface
which influences the ion part of the VAC (the effect of electroemission
292
on the electronic part of the VAC is usually less marked). In inter-

preting the measurements it is usually difficult to take this emission
into account because the work function under the conditions typical
of real technical devices is not known. Secondly, insulation is less effect
and this results in both an increase of the effective collecting surface
of the probe and in the appearance of strong leakage currents. Finally,
at relatively high probe temperature the probe fails. Therefore, if ex-
periments are not pulsed, then in these situations it is necessary to use
either cooled probes or reduce the time during which the probe stays
in plasma (for example, the time-of-flight probe can be used). It should
be noted that since the density of thermoemission current depends quite
strongly on the surface temperature of the probe, the strong sensitivity
of the ion part of the experimental VAC to the change of the surface
temperature under the conditions in which thermal ionisation in the vi-
cinity of the probe surface is insignificant, indicates that emission exerts
an effect.
It is important to examine the problem of a reciprocal electrode
through which the probe current is closed. Within the framework of
the theory of a single probe examined in the previous sections, the
distance between the probe and the return electrode is considerably
greater than the size of the probe, and the probe-plasma potential
difference greatly exceeds the plasma-return electrode potential dif-
ference. Under these conditions, the probe potential should be deducted
from the return electrode. Evidently, the second condition can be fulfilled
more efficiently if the surface area of the return electrode and the
temperature of the surface is increased. It should be noted that the
return electrode is represented by an electric arc, either already present
in the given equipment (for example, the working arc of the plasma
torch) or a specially produced arc. If diagnostics is carried out by ana-
lysing the ion part of the VAC, it is then quite easy to fulfil the sec-
ond condition. In cases in which the first of the approach is discussed
in the previous section is used for diagnostic, and it is difficult to fulfil
the second condition, an additional (reference) electrode under a float-
ing potential can be introduced into the plasma and the probe poten-
tial can then be deducted from the potential of this electrode. The dis-
tance from the probe to the reference electrode should be greater than
the probe size.
Under the conditions typical of real technical devices the probe sur-
face contains various contaminants. Deposition of non-conducting ma-
terials decreases the effective collecting surface of the probe and that
of conducting materials increases this area and also causes leakage
currents to appear. The presence of contaminants is also reflected in
a change of the VAC in relation to the holding time of the probe in
293
Plasma Diagnostics
the plasma. It is also rational to examine the dependence of the VAC

of the collecting surface of the probe. If the probe operates in the
electronic part of the VAC and the saturation criterion of the VAC in
the region of the negative potentials is satisfied, then the presence of
strong leakage currents is indicated by the absence of any change of
the slope of the VAC in transition to the ion part. It should be noted
that in some cases the effect of contamination can be minimised by
ion bombardment of the probe surface. 6
Organisation of probe measurements in high-speed and high-energy
flows of chemically active plasma in large-scale technical devices, such
as MHD generators of electric energy, 109-112 has its specific features.
11.6 Comparison of theoretical and experimental results

Both the qualitative and quantitative comparison of the results of the
hydrodynamic theory of electric probes with the experimental data has
been carried out quite often in the literature (see, in particular, reviews
in Ref.6, 8, 15, 16. We shall discuss briefly comparison of the theo-
retical results examined in section 11.4 with the experimental results.
The diagnostic method, based on equation (11.14) was optimised in
Ref.48, 53, 113, 114). The parameters of plasma in which these ex-
periments were carried out are given in the table on p.295 of this book.
The shape of the VAC recorded in experiments, is in complete agree-
ment with the shape predicted for these conditions by theory in sec-
tion 11.3 (in particular, the floating potential proved to be close to the
plasma potential, is also confirmed by the measurements 115-118 ). The
slope of the linear section of the experimental VAC in each case is
consistent with the values estimated from equation (11.14).
There are a relatively large number of studies in which sections of
the current of positive ions were measured and compared with the ex-
periments, see, in particular, Ref.15, 74, 78-83, 116, 119-125.
The number of studies in which the saturation current of positive
ions was measured was even larger. When analysing these measure-
ments it is natural to use, as the basic model, the model of plasma with
two types of charged particles and ‘frozen’ volume ionisation and re-
combination. As shown in section 11.4, in this model, the normalised
ion saturation current on a probe of a given configuration in a subsonic
flow can be regarded approximately as a universal function of the Peclet
number whose form does not depend on the type of gas, parameters
of non-perturbed plasma, etc. Therefore, within the framework of this
model it is natural to examine, from the same position, all experimental
data available in the literature on the saturation currents of positive
ions on the probes of the most frequently used configurations – spherical
and cylindrical. The results of this interpretation 93 are presented in
294
C o mb ustio n p ro d uc ts(C P ) n e∞, c m–3 T ∞, 1 0 3K v ∞, m/s S o urc e
Air + K 1 0 13– 1 0 14 2.5 300 [4 8 ]
C P C 3H8 in a ir + N a , K 1 0 10– 1 0 12 2.2 6 [5 2 ]

12
C P C 3H5O H in a ir + K 3 × 10 – 6 2.8 400 [5 3 ]
×103
Ar 1 0 17 13 230 [9 0 ]
13 15
Air 10 – 10 5–7 100 [11 3 ]
N2 + O2 8 × 107 – 2 × 0.35 0 [11 4 ]

109
Ar 2 × 1 0 17 15 130 [11 8 ]
10 13
C P C 3H18 in a ir + N a , C s 3 × 10 – 10 2.3 5 [1 2 2 ]
Ar 5 × 1 0 15 – 2 × 9 – 15 80 [1 4 8 ]
1 0 17
C P C 3H8 in a ir + Li, N a , K , 1 0 10 – 2 × 1 0 12 2.2 4 [1 4 9 ]

Rb , C s
Ar + K (1 ÷ 4 )1 0 13 2.3 25 [1 5 0 ]
C P C O in O 2 + N 2,, Ar + N a , 9 × 109 – 2 × 2.4 2 [1 5 1 ]

K, Cs 1 0 12
C P H2 in a ir + C s 5 × 1 0 10 – 8 × 2.5 40 [1 5 2 ]
1 0 11
C P C 3H8 in a ir + K 7 × 1 0 10 – 4 × 2.1 8 – 20 [1 5 3 ]
1 0 11
C P C 2H2 in a ir + N a (2 ÷ 4 ) 1 0 10 2.4 4 [1 5 4 ]
11 12
C P C H4 in O 2 + N 2 + K 2 × 10 – 10 2.2 10 [1 5 5 ]
C o mme nt. 1 . Exp e rime nts (with the e xc e p tio n o f tho se in Re f. 11 4 ) we re c a rrie d o ut in
the rma l p la sma a t a tmo sp he ric p re ssure , the e xp e rime nts in Re f. 11 4 we re c a rrie d o ut in
the a c tive zo ne o f a nuc le a r re a c to r a t a p re ssure fro m 0 . 5 to 2 a tm. 2 . C P – c o mb ustio n
p ro d uc ts.
Fig.11.5, 11.6. The sources of experimental data in the plasma param-

eters at which all these experiments are carried out are given in the
Table. The concentration of charged particles n c used to normalise the
saturation current, were either taken from the sited or calculated using
Saha's equation from the normal values of the plasma temperature and
the concentration of ionising atoms. The values of the densities of
saturation currents are the critical points of spherical and cylindrical
probes, presented in Ref.153, 52, were converted to the saturation
current to the front hemisphere and the front half cylinder using the
first equations (11.31) and (11.32).
It can be seen that the simple analytical equations in section 11.4
describe a large part of the experimental data with the accuracy to
the factor 3 (the boundaries of the corresponding corridor are shown
by dashed lines in Fig.11.5, 11.6). It should be stressed that all experi-
mental points distributed above this corridor were obtained in the ex-
periments with gas torches at low concentrations of charged particles
295
Plasma Diagnostics
(to 4·1010 cm–3) were, as shown by estimates, it is not possible to expect

ionisation equilibrium in the plasma incident on the probe and the evalu-
ation of the concentration of the charged particles using Saha's equation,
used in normalising the saturation current, may prove to be too low,
as is usually the case under these conditions. On the other hand, the
points situated below this corridor were obtained at higher concentration
values where the ratio of the recombination length to the radius of the
probe or (at Pe > 1) to the radius divided by Pe are less than unity,
and the saturation current is strongly influenced by the recombination
in the volume of the boundary layer.
As an example of interpreting the measurements of saturation current
using a more complicated model we can mention Fig.11.7. 52 It must
be stressed that in the probe surface temperature range 400-800 K which
also includes the experimental data shown on this Figure, 52 the ion
currents are independent (within the experiment range) of the surface
temperature and material. Evidently, the results of numerical calculations
within the framework of the model which takes into account the ef-
fect of negative ions are consistent with the experimental data.
It should be stressed that the presence of negative ions as a probable
reason for the special features of the ion saturation currents in the
plasma of combustion products was mentioned in Ref.126, whereas
HCO –3 ions were linked with the effect in Ref.52 after publishing ther-
modynamic data. 127 Although examination of the role of negative ions
in the plasma of combustion products has a very long history, suffi-
ciently reliable experimental data on the concentration of charged par-
ticles in the temperature range in the order of 2200 K and lower which
would enable an unambiguous confirmation or rejection of this conclusion,
have appeared only recently – see Ref.128 in which a specially de-
veloped high-sensitivity laser interferometer was used to measure the
electron concentration in the plasma of combustion products with an
addition of potassium in the temperature range 2000–2500 K. At tem-
peratures lower than approximately 2300 K this concentration was (in
the case of a stoichiometric mixture) considerably lower than that
calculated using Saha's equation from the measured values of the con-
centration of the free potassium atoms and temperature. Since the de-
viation from equilibrium is, according to estimates, small under these
conditions, the result evidently shows the presence of a large number
of negative ions, as confirmed by the conclusions made in Ref.126 and
52. Thus, electric probes in high-pressure plasma can be used not only
for controlling the operation and experimental equipment but also for
examining the processes in plasma.
296
11.7 Conclusion
Of the three diagnostic methods discussed previously and based on the
linear section of the VAC, the section of the saturation current of posi-
tive ions, respectively, the simplest method is often the first one. In-
deed, equation (11.14) can be used to calculate directly the value of
s ¥ from the slope of the linear section of the VAC measured in the
experiments, and it is not necessary to know accurate values of the
transport and kinetic plasma coefficients. Of course, if it is required
to determine n e¥ after determining s ¥, it is necessary to know µ e. This
method is especially efficient in inspecting the conductivity of plasma
directly during the experiments. 48,53
Within the framework of the plasma model which takes into account
the types of charged particles (positive ions and electrons), the sec-
ond and third method can also often be used in the form of explicit
equations which link the expression for the section of the ion current
or the value of the ion saturation current with n c¥. In particular, such
equations are available for a number of situations in which ionisation
recombination in the near-probe region are frozen. It should be noted
that the values of the diffusion coefficients of the ions included in these
equations are often known or can be estimated with the acceptable
accuracy (with the error not exceeding 10%). 28 Within the framework
of the model of the multicomponent chemically active plasma the cal-
culation method is greatly complicated and requires taking into account
the kinetics of reactions in the near-probe region. Information of the
kinetics is often insufficient so that the possibilities of using this method
for determining the concentration of charged particles in non-perturbed
plasma are limited. On the other hand, the method can be used to
examine the discussed kinetics.
Numerical methods are an efficient means of calculating and ana-
lysing the probe VAC for unidimensional problems.
It should be noted that in developing a method of probe diagnos-
tics of high-pressure plasma for new conditions the formulation experi-
ment should be as complex as possible: it is rational to carry out ex-
periments with probes of different configuration, using different diag-
nostic methods, at different surface temperatures and holding time of
the probe in the plasma, etc.; it is also efficient to combine the probe
method with other diagnostic methods. On the one hand, this increases
the reliability of interpreting the results and, on the other hand, ena-
bles additional information on the plasma parameters and properties to
be obtained.
In conclusion, it is convenient to mention some studies concerned
with the problems of probe diagnostics of high-pressure plasma which
297
Plasma Diagnostics
are outside the framework of this book (see also books in Ref.6, 8):
in Refs.57, 60, 63-65, 129–133 the authors examined non-stationary probe
measurements in Ref.134 measurements were taken in plasma with an
applied electric field, in Ref.135–139 measurements were taken in a
magnetic field, in Ref. 16, 140 in turbulent plasma, in Ref.145, 146 in
a plasma with a higher degree of ionisation, and in Ref. 145, 146 meas-
urements were taken in flames. As regards the problem of double probes,
it is necessary to mention Ref.147.
298
Probe Methods of Diagnostics of Chemically Reacting Dense Plasma
Chapter 12
PROBE METHODS OF DIAGNOSTICS OF

CHEMICALLY REACTING DENSE PLASMA
The probe method of plasma diagnostics is a simple and efficient means

of obtaining information on the local plasma parameters. A large amount
of information has been accumulated on the operation of electric probes
in greatly differing situations. This information has been systematised
in Ref.1, 4 and 12. The probe theory in the case of collisionless plasma
where the mean free path of the electrons and ions in the plasma is
considerably longer than the characteristic size of the probe has been
sufficiently developed and is a reliable means of obtaining information
on the properties of such plasma. The situation is complicated in tran-
sition to collisional plasma where the mean free path of charged particles
is of the same order or smaller than the characteristic probe size.
In this case, the electric probe starts to disturb significantly the ex-
amined plasma in its vicinity and greatly changes the concentration of
charged particles in this region in comparison with non-perturbed
quasineutral plasma. This perturbation can no longer be regarded as
small and, in addition, it generates the probe current measured in the
experiments which carries information on the plasma parameters. The
task of theory in this case is to learn how to calculate, using the given
parameters of the plasma and electric probe, electric layers around the
probe, electron and ion currents, the total current on the probe, and
also develop approaches to solving an inverse problem, i.e. determine
the plasma parameters from the probe current.
It should be noted that in the case of dense plasma attention is usu-
ally given to situations with frozen or non-frozen chemical reactions,
in the presence or absence of negative ions or ions of different types,
and moving or still plasma. 1
Each specific case of dense plasma requires separate accurate theo-
retical development because the relationship of various plasma param-
eters with probe currents is non-linear. We shall pay special attention
to the case of an electric probe placed in dense stationary plasma
299
Plasma Diagnostics
formed as a result of chemical ionisation processes in molecular gas

mixtures.
12.1 Formulation of the problem and solution method

We shall examine low-temperature (T = 2000 ¸ 3500 K) dense still
plasma (with the pressure equal to approximately atmospheric pres-
sure) in which charged particles form as a result of reactions of chemi-
cal (associative) ionisation and they annihilate during dissociative re-
combination processes. These processes are the most efficient proc-
esses of formation and annihilation of charged particles in low-tem-
perature plasma without an external ionisation source. 3,15 A cylindri-
cal probe with radius R p and length L is placed in plasma. The probe
is under potential j p in relation to the earthed walls of the chamber
in which the examined plasma is generated.
We shall now list assumptions which must be made when formu-
lating the problem.
1. It is assumed that the probe problem has cylindrical symmetry.
The asymptotic boundary conditions at r*®¥ are replaced by conditions
at some finite point r* = r*m (r*m >> 1). The boundary conditions at
r* = r*m correspond to the assumption according to which the plasma
sufficiently far from the probe is not perturbed and there are no ex-
ternal electric fields and currents.
2. The effect of the cold probe surface on the thermophysical prop-
erties of plasma is ignored. The solution of the problem of the cooled
probe in stationary plasma, obtained in Ref.2, taking into account the
change of the transport properties, shows that in the region of nega-
tive values of the probe potential in plasma the effect of temperature
perturbation is associated mainly with the thermodiffusion effect and
can decrease the ion current on the probe.
3. The probe surface is assumed to be ideally catalytic, i.e the re-
actions of recombination of the charged particles on the surface oc-
cur at an infinite rate and surface ionisation or ionisation by an electron
impact at the boundary of the Knudsen layer in the vicinity of the
surface are regarded as insignificantly small. The numerical solution
of the problem shows that defining the quantity n*i (l, t*) =
n*iw (t*) ¹ 0 on the surface has only a slight effect on the current at
jp* < 0. As shown by calculations, the probe is not heated during short
examination times of the order of 100–300 µs and there is no
thermoelectronic emission. The intensity of autoelectronic emission is
low because of the relatively low strength of electric fields on the probe
surface.
4. It is assumed that the equilibrium distribution of the velocity of
all particles is established in the plasma and the temperatures of the
300
electrons, ions and neutral plasma or neutral particles are equal.

5. Plasma is regarded as still, and the convective transport of charged
particles during the measurement period is ignored.
The equations that describe, within the framework of the electro-
dynamics of solid media, the distribution of the electric potential ϕ and
the concentration of singly-charged positive ions n i and free electrons
n e in plasma around a cylindrical probe, will be written in the
dimensionless form to facilitate numerical solution of the problem 27
¶ns* d *s ¶(r * js* )

+ = W * (t * ) - a *ni*ne* , (12.1)
¶t * r* ¶r *
ε* ∂  * ∂ϕ* 
 = ne − ni ,
* *
r (12.2)
r * ∂r *  ∂r * 
¶ns* * ¶j
*
fs* = - * - sign (es )ns * , s = i, e,
¶r ¶r
d *
e ¶j
* 2 *
(12.3)
j * = ji* - *c je* - * * * .
di d i ¶r ¶t
The dimensionless variables, denoted by the asterix, are introduced

in the following form
t * = t (W0 α 0 )
0.5
r* = r / Rp , , W * = W / W0 , α* = α / α 0 ,
0.5
eϕ α 
ϕ =
*
, ns* = ns  0  ,
kT  W0 
(12.4)
kTb W 0.5
js* = js / jsd , jsd = 0.5s 0 , j = j / jid , s = i , e ,
*
α Rp
here W is the effective ionisation rate that changes with time in non-
stationary plasma; a is the effective recombination coefficient; W 0, a 0
are the characteristic (normalisation) values of these quantities; R p is
the probe area; T is the equilibrium plasma temperature; k is the Boltz-
mann constant; b s is the coefficient of mobility of the ions and elec-
trons; j is the density of total current recorded by the probe; j s is the
density of currents of charged components; e s is the charge of the
301
Plasma Diagnostics
particles of the s-th kind. In this definition of the dimensionless pa-

rameters, three dimensionless coefficients are introduced into the system
of equations (12.1)–(12.3) and
α0.5
0 kT kTbs
ε =
*
, δ*s = 2 , s = i, e.
eRp (W0α0 )
4πe W0 R p
2 0.5 2 0.5 (12.5)
Parameters d*s in the continuity equation of the components (12.1)

have the meaning of the ratio of the characteristic time of develop-
ment of the chemical process (t ch = (W 0 a 0) –0.5 ) to the characteristic
diffusion time of the charged particles (t sd = R 2p/D s, where D s = kTb s/
e is the diffusion coefficient of component s).
Parameter e* in Poisson’s equation (12.2) represents the ratio of
the square of the characteristic Debye length R d to the square of the
probe radius (R 2d = kT/4 e 2N, N = W 00.5/a 00.5 ). It should be noted that
in the examined non-stationary problem the true length of Debye’s
screening, determined from the instantaneous value of the concentration
of charged particles non-perturbed by the probe, changes over a very
wide range, because the concentration of the ions and electrons in the
gas at the start of the ionisation process is low. The introduced
dimensionless variables (12.4) are suitable for the numerical solution
of the problem:
Initial conditions
( )
t * = 0, ns* r* , 0 = 0, s = i, e; (12.6)
boundary conditions
( ) ( )
r * = 1, ns* 1, t * = 0, ϕ* 1, t * = ϕ*p = const,
ϕ (r , t ) = 0, ∂n / ∂r
(12.7)
r * = rm* , * *
m
* *
s
*
= 0.
From the solution of the problem (12.1)–(12.7) it is necessary to

find the time dependence of the total current on the probe I p = Sj idj*
at the given values of the quantities W* (t*) and a*. The geometri-
cal dimensions and the probe potential are assumed to be given.
The problem was solved numerically on the basis of the method de-
veloped in Ref.2. The difference system implicit with respect to time
was useful. The boundary problem for the system for the resultant
302
difference equations was solved by the method of matrix run with it-
erations and variable with respect to the spatial co-ordinate with the
integration step. This was carried out using a special method of de-
fining the previous iteration based on analysis of solution at points with
the highest gradients of the parameters. The accuracy of the solution
was inspected using the integral of the system of equations (12.1)–
(12.3) r*j* = f*j* (t*), where f* (t*) is some function that depends
only on time and does not depend on the spatial co-ordinate.
The programme makes it possible to find the accurate solution of
the problem for arbitrary functions W*(t*) and a*. If W* = const,
a* = const is considered, the solution at relatively high t* reaches the
stationary level.
As a typical example, we examine the results of solving the prob-
lem for a methane–oxygen mixture
0.5% CH 4 + 2% O 2 + 97.5% Ar. (12.8)
We examine the conditions behind a reflected shock wave at tem-

peratures of 2000–3000 K and the atmospheric pressure.
Under these conditions, the main ionisation process is the reaction11
CH + O → CHO + + e (12.9)
The appearance of the primary ion CHO + causes a whole series of

subsequent ion–molecular reactions, especially overcharging reactions.
Several types of positive ions form in the mixture: CHO +, H 3O, NO +
and, possibly, some others.3 The quantities n*i and W*(t*) in the equations
(12.1)–(12.3) correspond to the total concentration and the total rate
of formation of ions of all types. It is assumed that the properties of
the transport of ions of different types are similar. 5–7
For the examined conditions, it can be assumed that the values of
the function W*(t*) are determined by the rate of the process of as-
sociative ionisation (12.9). The corresponding values of the rate of
formation of CHO + positive ions or free electrons at T = 2750 K and
the atmospheric pressure are presented in Fig.12.1 (curve 1). They were
obtained as a result of solving a direct kinetic problem describing the
process of chemical ionisation in this system. 3 The initial approxima-
tion for the function W (0) (t) in the probe problem was the rate of ioni-
sation from the direct kinetic problem W(t).
The low degree of ionisation under the conditions examined here
enables us to ignore the interaction between charged particles when
calculating the transport properties.
We examined the results of the calculations for the studied meth-
303
Plasma Diagnostics
W ×10–16, cm–3 s–1
t, µs
Fig.12.1 Dependence of the rate of formation of charged particles on time for a mixture
of 0.5% CH 4 + 2% O2 with argon at 2750 K and atmospheric pressure behind a reflective
shockwave. 1) W (0)(t) from the kinetic calculation of ionisation processes, 3 2) W (1)(t) from
dependence (12,11), 3) W (t)(2) is the corrected velocity.
Fig.12.2 (right) Space–time distribution of the concentration of positive ions (I) and
free electron (II) around the probe. t = 10 (1), 14 (2), 20 (3), 23 (4), 39 µs (5).
ane-oxygen mixture at T = 2750 K and atmospheric pressure. The

following data were used in the calculations: b i = 33.6 cm 2/(V·s); 5,6
b e = 1.07·10 4 cm 2 /(V·s); 6,8 a* = 1, a 0 = 6·10 –7 cm 3 /s, W 0 = 10 16
cm –3 ·s –1 ; j p = –9 V, R p = 0.015 cm, S = 1.51 cm 2.
In calculating the coefficient of mobility of the ions b i it was as-
sumed that the main contribution to the interaction of the ion with the
neutral particle at T = 2000 ¸ 3000 K is provided by the polarisation
interaction, and the section of elastic collisions was estimated using
the procedure described in Ref.5. The coefficient of mobility of the
electrons b e was calculated assuming that the elastic collisions of the
electrons with the argon atoms are dominant. The decrease of the
section in the temperature range 2000–3000 K was considered. 8 As
shown by the calculations, at the negative probe potentials the vari-
ation of quantity b e within the limits of several orders of magnitude
influences only slightly the intensity of total current on the probe.
12.2 Numerical solution results

Figure 12.2 shows the values of the concentration of positive ions and
electrons in the vicinity of the probe at different moments of time. From
the qualitative point, the distribution resembles the situation in stationary
plasma. 1 Figure 12.2 can be used to examine the dynamics of the
304
change of the layer of the free charge around the probe. The region
of perturbation of the plasma by the probe equalling (5 ¸ 6) Rp is clearly
visible.
The distribution of the strength of the electric field is shown in
Fig.12.3. The maximum strength of the field at the probe surface for
the given parameters is of the order of 10 3 V/cm. The electric field
changes to the maximum extent inside the charged layer, although it
does penetrate into the region of quasi-neutral plasma.
Figure 12.4 shows the distribution of the electric potential ϕ for dif-
ferent moments of time, and the calculated total current to the probe
is shown in Fig.12.5 (curve 3).
12.3 Experiments
To compare the calculated and experimental values of probe currents
experiments were carried in a shock pipe designed in Ref.3. At the
same time, the non-stationary concentration of the electrons was
recorded using a microwave interferometer with a two-lead line as a
probe system 3 and the current on the probe was determined using the
methods described in Ref.13. This was achieved by combining the two-
lead line with the electric probe: the conductor is made of stainless
steel and parts of the waveguides to which they were connected were
electrically insulated from the remainder of the microwave circuit using
thin teflon gaskets which have no effect on the parameters of the mi-
crowave wave. Both wires received the same electric potential (constant
with time) in relation to the earthed walls of the shock pipe. Conse-
quently, the electric probe consisted of two wires 0.3 mm in diameter
E, V/cm ϕ, V
Fig.12.3 Distribution of the strength of electric field in the vicinity of the probe surface.
t = 1.44 (1), 5.76 (2), 10 (3), 20 µs (4).
Fig.12.4 (right) Distribution of the electric potential in the vicinity of the probe surface.
t = 1.22 (1), 5.76 (2), 10 (3), 20 µs (4).
305
Plasma Diagnostics
t, µs
Fig.12.5 Experimental and calculated dependences of probe current and electron

concentration for a mixture 0.5% CH4+2% O2 with argon at T = 2750 K and the atmospheric
pressure behind the reflected shockwave. Experiments: 1) probe current I(t), 2) electron
concentration ne(t); calculations: 3) Ip(0) (t) using the initial approximation of the function
W 0(t), 4) Ip(2) (t) using the second approximation of the function W (2)(t), 5) concentration
of electrons n e(t) with functions W (2)(t).
and 80 mm long separated by a distance of 4.5 mm. The wires were

stretched in the direction normal to the direction of the speed of propa-
gation of the shockwave at a distance of 10 mm from the end of the
shock pipe.
The results of one of the typical experiments are presented in
Fig.12.5. It can be seen that from the moment of arrival of a reflected
shockwave to the probe current (curve 1) increases initially with time
and then, passing through the maximum (I max = 1.76·10 –4 A at t = 21
µs) starts to decrease. The electron concentration (curve 2) reaches
the maximum value slightly later, at t = 30 µs, and the decrease of the
electron concentration due to the recombination process is slower than
the decrease of probe current as a result of a rapid decrease of the
ionisation rate.
12.4 Discussion of the results

Figure 12.5 shows that the time to establishment of the maxima of the
calculated and experimentally measured currents coincides but the cal-
culated values of the total current I p (t) in the region of the maximum,
obtained using the initial approximation of the function W (0) (t) is ap-
proximately half the experimental values (curves 1, 3).
306
To understand the effect of W (t) and a on the probe current, we

carried out calculations with different values of these parameters. The
results show that the current depends mainly on W (t), and the vari-
ation of a over a relatively wide range has only a slight effect on
current. The maximum current with respect to time almost completely
coincides with the maxima of the ionisation rate. These special fea-
tures of the effect of W (t) and on the current enables the ionisa-
tion rate to be corrected with respect to the experimental current using
the following procedure. From the numerical solution of the problem
with the initial approximation W (0) (t) we can plot the dependence of
the calculated current I p (t) on W (t) (Fig.12.6). To consider a wider
range of the variation of W (t), in calculations, the function W (0) (t)
was multiplied by the constant coefficient 20. It can be seen that there
is a peculiar hysterisis loop in the dependence of I p (t) on W (t), if
the values of current prior to and after the maximum are plotted. This
indicates that the current is also affected by the diffusion and mobil-
ity processes which change the structure of the layer of the volume
charge with time. However, to a first approximation, this effect can
be ignored and the dependence of current on the ionisation rate can
be regarded as an unambiguous function. If we examine the situation
in which the ionisation rate changes from experiment to experiment while
other plasma parameters (the coefficient of diffusion and mobility, tem-
perature, pressure, etc.) remain almost constant, the dependence Ip (W)
can be approximated by the relationship
W ×10–16, cm–3 s–1
Fig.12.6 Dependence of calculated current on probe I p(t) on the actual value of the rate
of formation of ions W(t) for different values of probe potential. ϕ p = –9 (1), –48 (2),
–72V (3); the arrow on the curves indicates the direction of the passage of time.
307
Plasma Diagnostics
0.6
. × 10 -17 S j p
0.75
I p (t ) = 172 W (t ) , (12.10)
where I p is in A, j p is the probe potential and V;W (t) is in cm –3 ·s –1 ;

S is the surface area in cm 2. The dependence of current on the nega-
tive potential is determined by processing the results of calculations
carried out for several different values of the probe potential.
The relationship (12.10), which is valid at all times, with the exception
of the very late stages of the processes of disappearance of charged
particles, represents an approximate volt–ampere characteristic of the
cylindrical probe for non-stationary plasma with chemical reactions.
Because of the dependence of the function W (t) on time, the non-
stationary volt–ampere characteristic (12.10) is a surface in space
(I p, j p, t). The dependence of total current on the probe on plasma
temperature and the transport coefficients will be determined in
examining stationary plasma with chemical reactions where dynamic
equilibrium is established between the process of formation and an-
nihilation of the charged particles.
Calculations show that in the equation (12.10) it is necessary to cor-
rect the value of the constant numerical coefficient in order to obtain
the best agreement with the calculated values of current. This depends
on the gas mixture examined and the specific experiment conditions.
It is also possible to obtain agreement between the calculated values
of current and the values calculated from equation (12.10) in the range
20%.
Converting the resultant dependence and using the experimentally
measured probe current I (t) instead of I p (t), we can determine the
dependence of the ionisation rate in the system in the first approxi-
mation
-0.8
W (1) (t ) = 2.25 × 1022 S -4 / 3 j p
4/3
I (t ) . (12.11)
The values of W (1) (t), obtained from the equation (12.11) for the
experimental conditions shown in Fig.12.5, are presented in Fig.12.1
(curve 2). The calculations show that a small correction of the first
approximation W (1) (t) in the region of the maximum current is suffi-
cient to obtain agreement between I p (t) and I (t). Curve 3 in Fig.12.1
has the corrected profile W(2) (t). For this function there is a good agree-
ment between I p (t) and I (t) (curves 1 and 4 in Fig.12.5) for almost
all times, with the inspection of the late stages of the process.
For the non-perturbed region of plasma away from the probe, the
308
equations of continuity for the probes and electrons (12.1) give
¶ni ,e / ¶t = W (t ) - ani ne , ni = ne . (12.12)
Substituting here the values of W (t) = W [I (t)] we can select the

value of the effective recombination coefficient α which is such that
the profile of the electron concentration n ep (t), determined by solv-
ing this equation, is in satisfactory agreement with the experimental
profile from microwave measurements (curve 2 in Fig.12.5) within the
experiment error range (±20%). The profile obtained in this manner
is shown in the same graph (curve 5). However, the recombination
coefficient is several times higher than the literature values for the
dissociative recombination coefficient. 14–16
The intensity of the chemical ionisation processes depends strongly
on temperature. 3 We carried out a series of experiments at different
temperatures behind a reflected shockwave with the same mixture and
using the same procedure with simultaneous measurement of the probe
current and the concentration of free electrons at a constant probe
potential, j p = –9 V. Using the relationship (12.11), the experimental
values of the probe currents were used to determine the ionisation rate
values W (1) (t) shown in Fig.12.7. Dashed curves correspond to the
corrected values of W(2) (t) at which there is good agreement between
I p (t) and I (t). Solution of the equation (12.12) with the functions
W (2) (t) and selection of the corresponding values of α to obtain good
agreement between the calculated and experimental electron concen-
trations (Fig.12.8) leads to the following dependence of the effective
recombination coefficient a temperature:
a = 4·10 –2 T –1 – 9.8·10 –6 ,
where a is in cm –3 ×s –1 .
As previously, the value of a is too high. When the temperature in-
creases from 2250 to 2900 K, the values of a change from 8·10 –6 to
4·10 –6 cm 3 ×s –1 .
This method can be used to determine the values of the effective
ionisation rate W (t) and the effective recombination coefficient from
the probe current and electron concentration measured in the experi-
ments. The values of a are higher than the literature data, and if the
literature data for α and the ionisation rate at which the currents are
in good agreement are considered, there is a difference between the
calculated and measured electron concentrations. The calculated con-
centration is approximately twice as high as the concentration meas-
309
Plasma Diagnostics
W ×10–16, cm–3 s–1
t, µs t, µs
Fig.12.7 Dependences of the rate of formation of charged particles W(t) on time for a
mixture of 0.5% CH 4 +2% O 2 with argon at different temperatures and atmospheric
pressure behind the reflected shockwave. I) W (1)(t), obtained from (12.11); II) W (2)(t)
is the corrected velocity; T = 2250 (1), 2440 (2), 2600 (3), 2750 (4), 2800 (5), 2900
K (6).
Fig.12.8 (right) Dependence of the concentration of free electrons on time for a mixture
of 0.5% CH 4 +2% O 2 with argon at different temperatures and atmospheric pressure.
Lines – experiments for temperatures of: 2250 (1), 2400 (2), 2600 (3), 2600 (4), 2800
K (5); points – calculated electron concentrations.
ured in experiments in the region of the maximum. Combining these

curves by decreasing the ionisation rate leads to calculated current values
which are several times lower than the experimental data. 29
Possible reasons for the differences in these values were examined
theoretically: the presence of a large number of negative ions in plasma,
the difference between the electron temperature and the temperature
of heavy particles, the presence of convective transport of the charged
particles by the gas flow, the chemical reactions of ionisation under
probe surface in the diffusion layer around the surface, possibly with
electronically excited particles taking part. 35
We believe that the most likely reason for this difference are the
ionisation reactions of neutral particles (possibly electronically excited)
on the probe surface or in its vicinity. Further analysis is required and
the model of the work of the probe in non-stationary chemically re-
acting plasma must be improved.
However, on the basis of the results it is already possible to pro-
pose a semi-empirical method of determining the electron concentra-
tion and the effective ionisation rate from results of probe measure-
310
ments. The correcting factor must be introduced into the W (1) (t) cal-
culated from equation (12.11). This factor can be evaluated by direct
comparison of the chemical ionisation rates obtained in experiments
simultaneously by probe and microwave measurements. Such a com-
parison for a methane–oxygen mixture (12.8) and the same mixture
with helium in the temperature range 2150–2900 K at the atmospheric
pressure behind the reflected shockwave leads to the relationship
W micro = kW (1) (t), where k = 0.13. The quantities W micro and W (1) are
found for the same moment of time corresponding to the maximum of
probe current: the first – from the first time derivative of the region
of the time dependence of the free electron concentration in the re-
gion of the maximum derivative, the second – from equation (12.11).
Using the constant correction factor k = 0.13 for the entire time dis-
tribution of the probe current, we obtain an approximate profile of the
effective ionisation rate. The resultant solution of equation (12.12) with
this profile is the calculated profile of the electron concentration n e
(t) which is in good agreement with the experimental profile n emicro (t)
for the literature values of the recombination coefficient.
As already mentioned, for constant ionisation rates and recombination
coefficient a dynamic equilibrium is established in plasma after some
time between the rates of formation and annihilation of the charged
particles. Their concentrations become stationary and the probe cur-
rent also ceases to depend on time.
In this case, the solution of the problem (12.1)–(12.5) is determined
completely by the dimensionless parameters d i *, d e *, e* and the
dimensionless probe potential (j*p ). Calculations were carried out in
which these parameters were measured in a relatively wide range and
the results were used to determine the dependence of total current on
the probe on these parameters. The dependence can be described ap-
proximately as an exponential function of these parameters
( ) (ε )
0.66 0.12 0.4
I = 8 ⋅10−10 eW0 Rp S δ*i *
ϕ*p , (12.13)
where I is in A; S in cm 2; R p is cm; W 0 in cm –3 ×s –1 .
As indicated by (12.13), the total current at the negative probe
potentials is almost independent of parameter d*e which is not present
in (12.13). The resultant relationship is an analytical approximation of
the general self-modelling solution of the problem of the probe in chemi-
cal reacting equilibrium plasma and is highly universal for any case of
ionisation processes, with the exception of electron impact ionisation.
Since the ionisation rate becomes a complex self-consistent (with other
311
Plasma Diagnostics
parameters of the problem) function of the distance from the probe

surface, this case requires special examination.
If the expressions for the dimension parameters d* i, e*, |j* p| are
substituted into equation (12.13), we obtain the dependence of total cur-
rent on the main parameters of the examined plasma
0.4 0.66 0.61 -0.27 0.38

I = 3.367 × 10 -19 ( R p ) 0.44 L j p bi W a T , (12.14)
where I is in A; b i in cm 2 /(W s); W in cm –3×s –1 ; j p in W; T in K; a

in cm 3 s –1 ; R p and L are in cm.
If it is taken into account that the equation (12.12) in the case of
stationary plasma gives a simpler relationship between the ionisation
rate, the recombination coefficient in the concentration of heavy particles
in the plasma non-perturbed by the probe, the total current on the probe
can be expressed by the equilibrium concentration of charged parti-
cles
0.4 0.66 1.22 0.34 0.38

I = 3.367 × 10-19 ( R p ) 0.44 L j p bi ni , e a T , (12.15)
where n ie is in cm –3 , other parameters as in (12.14).

If we use an independent method to determine the electron concen-
tration in the equilibrium plasma (with a microwave interferometer or
some other method), the measurements of the probe current at the
known plasma temperature enable us to determine the coefficient of
mobility of the ions b i from equation (12.15).
12.5 Electric probes in special conditions

In the literature dealing with the probe methods of plasma diagnostics
it has become traditional to examine the group of the following problems
in the section under this name: the electric probe in the transition regime,
the probe in multicomponent plasma with different types of positive ions
of different masses, with positive and the negative ions, the effect of
electronic emission on the probe characteristics, the probe under the
conditions with the strong effect of convection in the molecular regime
and in the continuum regime, the probe in a magnetic field.
In the transition regime, the mean free path of the charged parti-
cles coincides, as regards the order of magnitude, with the characteristic
size of the probe. In the general formulation, the problem of the probe
in this case includes a complete system of Boltzmann equations and
Maxwell equations. This problem has not as yet been solved. Usually,
312
experimental data are processed using different interpolation equations.

A comprehensive list of the studies concerned with the probes in the
transition regimes can be found in the reviews in Ref.1, 4 and 12.
Investigations of the behaviour of the probe in multicomponent plasma
were carried out because of a number of circumstances. Firstly, if the
multicomponent nature of the plasma is ignored, the physico-chemical
processes, used to model the real properties of the plasma during the
experiment, can be greatly distorted. Secondly, with the exception of
probe methods, at present there are no reliable means of determining
the concentration of negative ions under real plasma conditions. The
attempts to use mass spectrometric methods for this purpose encountered
principal difficulties: very rapid ‘sticking’ of the free electrons to the
electronegative particles in the zone of the cooled sampler of the device.
In most cases, the multicomponent plasma forms in electric discharge
and chemical lasers, in plasma chemical reactors in different etching
processes with halides taking part, in upper layers of the Earth’s at-
mosphere, in the products of combustion of chemical fuel, in the wake
of a shockwave formed during movement of solids in the atmosphere
at a high velocity. The method of producing chemically active media
using the energy of free electrons is used on an increasing scale. The
energy of the external electric field through the free electrons can be
used for selective excitation of different quantum states of the atoms
of molecules thus stimulating atomic–molecular transformations. By
varying the plasma parameters, especially the concentration and electron
temperature, the chemical process can be directed towards the for-
mation of useful products. The transition from the electron to the nega-
tive ion is capable of changing the course of recombination and will
lead to a loss of energy which had been previously used to form the
free electrons. It is therefore important to examine the problem of
ensuring sufficiently active measurements of the concentration of elec-
trons and negative ions in the plasma where a large number of reactions
of both the formation and annihilation of the electrons and different
ions take place. 15,23,24
The effect of the multicomponent nature of the plasma on the probe
characteristics in the molecular regime has been examined in Ref.18,
19,12. The appearance of negative ions leads to changes in the structure
of the layer of the volume charge in the vicinity of the probe, the dis-
tribution of the electric field changes and the ion and negative probe
potentials increases by 30–50%. 12
In the continuum regime for plasma with frozen chemical reactions,
the problem of the probe in the mutlicomponent plasma has been ex-
amined in Ref.1, 12, 17. The main result of these investigations was
the conclusion according to which the increase of the ‘sticking’ electrons
313
Plasma Diagnostics
slightly widens the zone perturbed by the probe, the current density
of positive ions slightly increases, and the electronic current decreases
in accordance with the decrease of the concentration of free electrons.
We shall examine in more detail the method of determining the con-
centration of negative ions in a chemically reacting high-pressure low-
temperature plasma, in the continuum regime, on the basis of the volt-
ampere characteristic of the electric probe. We shall examine the con-
ditions under which the main processes of formation of the primary
charged particles, i.e positive ions and free electrons, are the proc-
esses of chemical associative ionisation A + B ® AB + + e. The nega-
tive ions then form as a result of sticking of the electrons in
C + e + M ® C – + M ternary collisions. Reversed reactions repre-
sent the processes of rapid dissociative electron-ion recombination and
separation of the electrons from the negative ions. Within the framework
of the examined model we can also investigate the thermal ionisation
and dissipative sticking of the electrons. At high concentrations of the
negative ions it is necessary to consider the ion–ion recombination under
the actual plasma conditions, depending on its composition, tempera-
ture and parameters, situations can arise in which the parameter
l = n 0– /n 0e, characterising the ratio of the equilibrium concentrations
of the negative ions and the electrons, changes in a very wide range
– from zero when there are almost no negative ions, to very high values
where almost all free electrons formed transform to negative ions.
In this system of reactions the rates of variation of the concentration
of charged particles in plasma w s can be described using five effec-
tive parameters:
w e = W + gn- - an+ ne - kne ,

w i + = W - bn+ n- - an+ ne ,
(12.16)
w i - = kne - gn- - bn+ n- .
Here W is the effective rate of the formation of the electrons and

positive ions in the associative ionisation processes; a is the electron-
ion recombination coefficient, k is the effective frequency of sticking
of the electrons to the electronegative components, g is the effective
frequency of separation of the electrons, b is the ion–ion recombination
coefficient, n e, n i+, n i– is the concentration of the electrons, positive
and negative ions. The values W, g, k depend on the concentration n s
of the negative particles.
W = k1na nb , k = k2 nc n M , g = k3 n M . (12.17)
314
Here k 1, k 2, k 3 are the constants of the rates of associative ioni-

sation, three-particle sticking and separation of the electrons. It is
assumed that in (12.17) the concentrations of the neutral particles are
constant in the vicinity of the probe. The effective parameters W and
k can also describe thermal ionisation in the collisions of neutral particles
and dissociative sticking of the electrons CD + e ® C – + D. k 1 is the
constant of the rate of thermal ionisation, and parameter k is written
in the form k = k* 2n CD , where k* 2 is the constant of the rate of dis-
sociative sticking.
The equations describing the distribution of the densities of charged
particles and the electron potential j* in the vicinity of the cylindri-
cal probe in the approximation of the electrodynamics of the continuum
in the dimensionless variable 27 have the form
∂ns* δ*s ∂
+ * * (r * js* ) = ωs , s = e, i +, i −,
∂t *
r ∂r
ε ∂  * ∂ϕ*  * *
*
 = ne + ni+ − ni − ,
*
r
r * ∂r *  ∂r * 
∂ns* * ∂ϕ
*
js* = − − sign ( e )
s sn , (12.18)
∂r * ∂r *
δ*i − * δ*c *
j = ji + − * ji − − * je .
* *
δi + δi +
Here j*s, j* are the densities of the currents of the components and
the total current. When writing the equations of continuity of the com-
ponents the convective transport of current to the probe is ignored. The
left-hand parts of the equation (12.18) include for dimensionless pa-
rameters.
e * = ( Rd / R p ) 2 , d *s = t / t ds , s = e, i + , i - ,
(12.19)
Rd2 = kT / (4 pe 2 n0 ), t ds = R 2p / Ds .
Here R d is the characteristic Debye length, t ds is the characteris-

tic diffusion time of component s. The right-hand parts of the equa-
tions of continuity of the components in the variables (12.4) have the
form
315
Plasma Diagnostics
ω*e = 1 + γτni*− − ne* ( ni*+ + κτ),

(
ω*i + = 1 − ni*+ ne* + β ni*− / α , ) (12.20)
ω*i − = τκne* − ni*− ( γτ + β ni*+ / α ).
The boundary conditions for the equation (12.18):
r * = 1, ns* = 0, j *p = j * , s = e, i + , i - ,
(12.21)
r * = rm* , j * = 0, ¶ns* / ¶r * = 0, rm* >> 1.
The initial conditions for the equation (12.18):
t * = 0, ns* (r * , 0) = 0, s = e, i + , i -. (12.22)
The problem (12.18)–(12.22) was solved numerically on the basis

of the method proposed in Ref.2. The main assumptions, made in for-
mulating the problem and also the special features of solving the boundary
Fig.12.9 Distribution of the concentration of charged particles and electric potential

for ϕ p = –9 (I), + 9 V (II). 1) n*i+ , 2) n*e , 3) n*i–, 4) |ϕ*|.
316
problem for the system of the resultant different equations, were dis-
cussed previously in this book and described in Ref.2 and 22.
The distribution of the concentrations of the charged particles and
the electric potential in the vicinity of the probe for one of the vari-
ance of calculations are presented in Fig.12.9. Calculations were carried
out for the following variants of the parameters: e* = 2.7·10 –2 , d i+ =
d i– = 6.7, d e = 6.7·10 3 , where W = 10 13 cm –3 ×s –1 , a = 10 –7 cm 3 ×s –1 ,
b = 5·10 –8 cm 3×s –1 , g = k = 1.5 × 10 3 s –1, the ratio of the equilibrium
concentration of the negative ions and electrons l = 0.73. These data
simulate the situation in air with an addition of CCl 4 vapours at a tem-
perature of T = 2200 K and the atmospheric pressure.
For these values of the parameters, the size of the region of electric
perturbation of the plasma by the probe under the positive potential ex-
tends over a large distance, over more than 100 R p. In the case of the
negative potential of the probe, the main decrease of the electric po-
tential takes place on the length » 5 R p. Under the examined condi-
tions, the dimensions of the layer of the volume charge equal (2–4)R p,
and the size of the entire region where the concentration of the charged
particles greatly differs from the non-perturbed values is 10Rp. It should
be noted that the analytical and numerical solutions of the probe problem,
available in the literature, were obtained mainly for the limiting cases
where the size of the layer of the volume charge is either small in
comparison with the radius of the probe or, on the other hand, is con-
siderably greater.
The problem (12.18)–(12.22) contains five parameters character-
ising the ionisation processes, and three transport coefficients – the
effective coefficients of mobility bi+, bi–, be. For the conditions examined
below, b e >> b i–, b i and the conditions b i+, b i– are of the same order
of magnitude. It is also assumed that b i+, b i– = b.
The numerical solution of the problem (12.18)–(12.22) enables the
distribution of the quantities n e, n i+, n i–, j in the region perturbed by
the plasma to be determined at the given probe potential and calcu-
late the current on the probe for each set of the given parameters, i.e
solve a direct problem. However, our main task is to solve the inverse
problem, i.e to develop a method of determining the plasma parameters
on the basis of experimental data for the current on the probe. We shall
examine the possibilities of fulfilling this task, as previously, by con-
structing analytical approximation dependences of the probe current on
the main plasma parameters.
Under the chemical equilibrium conditions for each specific initial
gas mixture the definition of temperature and pressure determines the
values of the kinetic parameters W, a, b, g and k, and also the val-
ues of the coefficients of mobility of the charged particles. The kinetic
317
Plasma Diagnostics
parameters are linked by the chemical equilibrium conditions which

include the equilibrium concentrations of the neutral components formed
as a result of a large number of chemical transformations. In the ex-
periments, the concentrations of the neutral components, taking part
in the formation of charged particles in accordance with (12.17), can
vary over a relatively wide range by varying the initial composition of
the mixture and the experimental condition as regards temperature and
pressure. Analysis of the literature data makes it possible to indicate
the ranges of the values of the kinetic parameters and the transport
coefficients which in principle can occur under the experiment con-
ditions.
The values of the effective coefficients, describing the annihilation
of the charged particles (the coefficient of electron–ion recombination
α and ion–ion recombination b) when the temperature varies from 1000
to 3500 K are in the range 15,24,25
10 -8 £ a £ 10 -6 ,
(12.23)
5 × 10 -9 £ b £ 5 × 10 -7 ,
where a and b are in cm 3×s –1 .

The parameter W, which describes the processes of chemical and
thermal ionisation, has the values in the range 15,26
. × 101 £ k £ 15
15 . × 108 ,
(12.24)
. £ g £ 15
15 . × 104 ,
where W is in cm –3 ×s –1 .
The parameters k and g which characterise the rates of ‘sticking’
and separation of the electrons depend on the concentration of the elec-
tronically active particles present in the plasma. To examine the ef-
fect of the negative ions on the volt–ampere characteristic of the probe,
we examine the following ranges of the values of these parameters
. × 101 £ k £ 15
15 . × 108 ,
(12.25)
. £ g £ 15
15 . × 104 ,
where k and g is in s –1 .
The corresponding change of the parameter l is:
318
2 × 10 -3 £ l £ 2 × 104 ,
(12.26)
l = ni0- / ne0 .
The values of the coefficients of mobility of the ions and electrons

for the gas mixtures at the pressure of the order of the atmospheric
pressure are in the range 7
1 £ b £ 100,
(12.27)
103 £ bc £ 105 ,
where b and b e are in cm 2/(V s).

At the negative probe potentials where |j* p| >> 1, the change of
the coefficients of mobility of the electrons has almost no effect on
the probe current.
These parameters correspond to a relatively wide spectrum of pos-
sible experiment conditions and, consequently, the equations derived
below can be used for the diagnostics of a large number of partially
ionised media which are of interest for practice.
We shall examine the solution of the problem (12.18)–(12.22) at nega-
tive probe potentials. A large number of numerical calculations were
carried out for different combinations of the parameters including the
problem, which correspond to possible experiment conditions. We shall
use the results obtained in kinetic modelling the processes of chemi-
cal ionisation in complex molecular systems, such as mixtures of oxygen
with hydrocarbons methane and acetylene,3 air with different additions,27,
etc.
Analysis shows that the results of numerical solution of the prob-
lem of a probe with the accuracy of 20% in relation to accurate nu-
merical calculations can be approximated by the correlation equation:
0.55
. × 10 -19 C × R 0p.44 LW 0.68b 0.6T 0.34 j p
I - = 55 ,
(12.28)
C = a -0.07 k 0.018 g 0.004b -0.02be0.02 .
where I – is the total current on the probe at the negative potential, A;

R p, L is the radius and length of the probe; cm; T is the plasma tem-
perature, K; j p is the probe potential, V. 20
Under the examined conditions, the layer of the volume charge in
the vicinity of the probe is quite long. Chemical ionisation reactions
319
Plasma Diagnostics
take place inside this layer; positive ions and electrons form from neutral
components in these reactions. The electric field displaces the elec-
trons from the layer thus preventing the reactions of electron–ion
recombination and electron sticking. These circumstances result in a
relatively strong dependence of I – on the ionisation rate and a very
weak dependence on the parameters a, b, g, k, which control the for-
mation of negative ions.
At the negative probe potential in the plasma with no negative ions,
the probe current is considerably stronger than the current of the same
negative potential. 1 However, if the plasma does not contain free elec-
trons, and the transport properties of the positive and negative ions are
similar, the volt–ampere characteristic of the probe is almost completely
symmetric. Therefore, it may be expected that the maximum sensitivity
of the probe current to the presence of the negative ions must be in
the region of the positive probe potentials.
Analysis of the results of numerical calculations in this region shows
that the dependence of the total current I + at j*p >>1 on the kinetic
parameters W, a, b, and temperature T is almost the same as in equa-
tion (12.28). However, the relationship of I + with the mobility coeffi-
cients b, b e and with the effective rate of electron sticking is more
complicated. The dependence on these parameters changes with the
change of the ratio of equilibrium values of the concentration of negative
ions and electrons l. Therefore, the parameter l = n 0i–/n 0e– will be used
as the controlling parameter. In our case of equilibrium plasma, l is
related with the kinetic parameters, included in (12.16), by algebraic
relationships.
The results showing that there is a small difference in the depend-
ences of the currents I +, I – on W, a, b, T indicate that it is rational
to use the dimensionless quantity Y = I+(|j p|)/I–(–|j p|) which is almost
independent of these parameters. On the basis of numerical calcula-
tions for W we can write the following equation
Y = f 1 (l ) h f 2 ( l ) , h = be / b. (12.29)
Functions f 1, f 2 are shown in Fig.12.10. It can be seen that higher

sensitivity of the ratio of the currents to the concentration of the negative
ions is at l ³ 1. In the region where l >> 1, the functions f 1® 1,
f ® 0 so that the ratio of the currents Y ® 1, and the volt–ampere
characteristic of the probe is close to symmetric. When l<<1, the ratio
of the currents becomes
Y = be / b. (12.30)
320
Fig.12.10 Dependences f 1 = f 1(log λ) (1), f 2 = f 2 (log λ) (2).
Consequently, in reacting plasma with a small number of negative

ions the ratio of the currents measured at the positive and the same
negative potential (e|j p |/kT >> 1), can be used to determine the ra-
tio of the mobility coefficient b e/b.
Equation (12.30) was derived assuming that at j p > 0 the probe in-
duces slight electric perturbations in the plasma, i.e there is a coun-
ter probe of a relatively large area at some distance from the probe.
At the known value of b e/b the relationship (12.30) makes it possible
to inspect the accuracy of probe measurements at the positive probe
potential.
We shall now return to equation (12.29). It gives the dependence
of the ratio of the probe currents at the same absolute probe poten-
tial in relation to the plasma on two parameters: the ratio of the equi-
Fig.12.11 Sections of the surface log Ψ

(log λ, log η) by planes η = const. η =
10 4 (1), 10 3 (2), η = 10 2 (3).
321
Plasma Diagnostics
librium concentrations of the negative ions and electrons l and the ratio
of the coefficients of mobility of electrons and ions h. Figures 12.11
and 12.12 shows the cross section of the surface log Y (log l, log
h) by planes l = const, h = const. Using the functions f 1, f 2, shown
in Fig.12.10, it is easy to construct a graph of the function log Y (log
l) also for the intermediate values h. We can propose the following
method of processing the probe measurements in the plasma with nega-
tive ions.
We assume that the ratio of the mobility coefficients b e/b is known.
Using the dependence of log Y or log l as shown in Fig.12.11, from
the ratio of the currents Y 1 = I +/I – measured in the experiments we
determine the value log l which is the abscissa of the point of inter-
section of the log Y = log Y 1 with a curve log Y (log l) which cor-
responds to the value h = h 1. The determined value of l gives the
ratio of the concentrations of the negative ions and electrons in the
examined plasma.
Dashed lines in Fig.12.12 show the values of a, corresponding to
the cases in which the mobility coefficients are determined by the col-
lision sections of the charged particles with the He and Ar atoms. It
can be seen that depending on the type of gas (diluent) log (Y) changes
relatively appreciably.
When adding a relatively small amount of an electronic active ad-
dition to the plasma, the effective ionisation rate W and the recom-
bination coefficient a remain almost unchanged. Equation (12.28) shows
that the probe current for the negative potentials is almost constant,
although the equilibrium value of the concentration of positive ions slowly
Fig.12.12 Sections of the surface log Ψ (log λ, log η) by the planes λ = const., λ =
10 –27 (1), 1 (2), 10 (3), 10 2 (4), 10 3 (5), 10 4 (6), 7) η for helium, 8) η for argon.
322
Fig.12.13 Dependence of log K on log λ 1. 1) η = 10 2 (θ =10), 2) η = 10 3(θ = 10), 3)

η = 10 4 (θ =10), 4) η = 10 2(θ = 10 2), 5) η = 10 3 (θ =10), 6) η =10 4 (θ = 10 2 ), 7) η =
10 4 (θ =10 2), 8) η = 10 4 (θ =10 3 ), 9) η = 10 4 (θ =10 4).
increases as a result of the shift of the chemical equilibrium in the for-

mation of negative ions. Calculations carried out using the examined
reaction system show that the variation of the equilibrium concentration
of the positive ions does not exceed 50% of the entire range (12.25)
of the values of k.
It is evident that the formation of negative ions leads to a change
in the electron concentration. We introduce parameters q = n e1 /n e2 ,
K = Y 1/Y 2, where n ej is the equilibrium electron concentration, Y j is
the ratio of the probe currents at the same absolute value of the probe
potential in the initial plasma (j = 1) and in the plasma with the electron
negative addition (j = 2). Since at j p < 0 the current remains almost
unchanged, parameter K is equal to the ratio of the currents of the
positive potential
K = I1 ( + j p ) / I2 ( + j p ). (12.31)
Figure 12.13 shows the dependences of log K on log l1 for different

values of the parameters q and h, where l 1 is the ratio of equilibrium
concentration of the negative ions and electrons in the initial plasma.
Figure 12.13 shows that the dependence of log K on the parameter
h = b e/b becomes stronger when the magnitude of the change of the
electron concentration increases, i.e. with increasing parameter q. At
323
Plasma Diagnostics
Fig.12.14 Dependence of log K on

log θ. 1) η = 10 2 (λ 1 ), 2) η =
10 2 (λ 1=10 –2), 3) η = 10 3 (λ 1 =1),
4) η = 10 3 (λ 1 = 10 –2), 5) η = 10 4
(λ 1 = 1), 6) η =10 4 (λ 1 =
10 –2), 7) experiments.
q = 10 the dependence of log K on h up to the value log l 1 = 0 is

very weak and it can be assumed that there is a linear relationship
between K and q, as clearly indicated by Fig.12.14. The values of log
K change only slightly with the variation of l 1 in the range log l 1 <
0.5. This is due to a very significant contribution of the electrons to
the current at a negative probe potential. Consequently, in the region
where l 1 < 0.3, the probe measurement at the positive potential can
be used to determine the relative variation of the equilibrium concen-
tration of the electrons after adding electron-absorbing components to
the plasma. When log l 1 increases the dependence of K on l 1 shows
a non-linearity which is determined by the increase of the contribu-
tion of the negative ions to the current formation process. At higher
values of l 1, i.e high initial concentration of the negative ions in the
plasma, the contribution of the negative ions to the current becomes
controlling and the value of K tends to unity. As indicated by Fig.12.13,
at log l 1 ³ 1 the dependence of log K on q and h is relatively com-
plicated and to determine θ from the value of log K, it is necessary
to know the value h. At log l 1 ³ 2.5 log K tends to zero and the sen-
sitivity of K to the parameters q and h rapidly decreases. Therefore,
at very high initial concentrations of negative ions in plasma it is not
possible to determine the change of the concentration of electrons using
probe measurements at a constant probe potential.
Figure 12.14 shows the dependence of log K on log q for differ-
ent values of the parameters h and log l 1. The value of K is directly
proportional to the parameter q up to the values q = 10 and only slightly
324
Fig.12.15 Experiment to determine

dependence of the concentration of
electrons n ep and probe current on time.
t, µs 1,2) n ep (t) for mixtures I and II,
respectively; 3,4) (t) mixutres I and II.
sensitive to the change of h. Consequently, probe measurements at

q < 10 can be used to determine the variation of the electron concen-
tration when the electronegative component is added, even if there is
no information on parameter h.
Experimental verification of the calculated dependences was car-
ried out on a shock pipe in low-temperature plasma formed behind the
front of the reflected shock wave. Simultaneously, probe currents on
the concentration of free electrons were measured in the same cross
section of the shock pipe using a microwave interferometer with a high
spatial resolution. Experimental equipment and procedure were described
in Ref.22. An acetylene–oxygen mixture 0.27% C 2 H 2 + 1.4% O 2 +
98.3% Ar were used. In this mixture, the process of chemical ioni-
sation, ensuring a relatively high electron concentration (of the order
of 10 10 –10 11 cm –3 ) occurred at the relatively low temperatures of the
order of 2000–2500 K and pressures of the order of the atmospheric
pressure. The electromagnetic addition was carbon tetrachloride CCl 4.
Figure 12.15 shows the dependences of the electron concentration
and probe currents, measured with the microwave interferometer, at
j p = 9 V, on time in the mixture without (I) and with the addition of
CCl 4 (II). Temperature T = 2500 K, pressure − atmospheric.
The results of processing a series of experiments at temperature
325
Plasma Diagnostics
of 2200–2600 K and the atmospheric pressure using the method de-

scribed previously in the mixtures I and II are presented in Fig.12.14.
The maximum deviation of the experimental points from the calculated
dependence is around 30%. 21
12.6 Diagnostics of dense chemically reacting still plasma

with electric probes with a modulated potential
A single measurement in plasma diagnostics using an electric probe
with a constant potential can provide information on the amplitude char-
acteristics of the probe signal associated with the plasma parameters.
In modulation of the probe potential by specific pulses, in addition to
the amplitude characteristics of the signal we obtain the phase char-
acteristics, especially the phase shift between the current signal and
the probe potential. The function of dependences of the probe method
widen in this case.
The method of pulsed probe characteristics is used widely for ex-
amining different non-stationary plasma formations. At a high formation
content, at high operating speed and simple calibration, this method is
characterised by the appearance of surges of probe current at pulsed
probe excitation. These surges greatly exceed the stationary value. The
effort of experimentators is directed mainly to developing measuring
systems in which information on the current to the probe during the
transition process was excluded in order to eliminate the effect of non-
stationary current surges. For example, in Ref.28 each consecutive
change of the pulsed voltage on the probe was carried out only af-
ter completing the transient process caused by a previous change. How-
ever, this resulted in a loss of valuable information on the plasma pa-
rameters and required relatively complicated measuring systems. We
shall now examine how to utilise the ‘interfering’ surge of probe current
in pulsed probing for determining the plasma parameters. 34
We shall examine the case of dense still plasma in which chemi-
cal reactions of ionisation and recombination take place. The law of
variation of the probe potential is shown in Fig.12.16. The dynamic
properties of the probe–plasma system will be examined in the stationary
section of the probe current when the total current to the probe be-
comes constant and the concentration of charged particles in the plasma
region not perturbed by the probe is also constant under the conditions
of establishment of equilibrium of the ionisation and recombination rates.
The equations determining the distribution of electric potential and
the concentration of singly-charged positive ions and electrons in the
vicinity of the cylindrical probe can be written, on the basis of the elec-
trodynamics of solids, in the dimensionless form similar to the equa-
326
t, µs
ϕ, V
Fig.12.16 Calculated dependence of the probe current on time for the pulsed variation
of probe potential. W = 10 16 cm –3 s –1 , n 1,e = 4.25 × 10 11 cm –3 .
tions (12.1)–(12.5). The dimensionless variables are introduced as in

the case of equation (12.4). The initial conditions remain in the form
(12.6). The boundary conditions will be written in the form:
r * = 1, ns* (1, t * ) = 0, s = i , e,
j * (1, t * ) = j *p = const , t * < t0* ,
j * (1, t * ) = j *p + Dj * , t 0* £ t * < t1* ,
j(1, t * ) = j *p , t * ³ t1* ,
(12.32)
r * = rm* , j * (rm* , t * ) = 0, ¶ns* / ¶r * = 0, s = i , e.
The moments of time t*0 and t*1 correspond to the time of supplying
a pulse to the probe which changes the probe potential by the value
±Dj*. Assumptions made in formulating the problem, and details of this
327
Plasma Diagnostics
numerical solution are identical with those discussed in the problem

(12.1)–(12.7).
We shall examine the results of the numerical solution of the problem
for plasma formed behind a reflected shock wave in air at T = 3200
K and the atmospheric pressure. The main ionisation process under
these conditions is the reaction
N + O ® NO + + e –. (12.33)
The primary ion NO + is dominant.30 Therefore, when calculating the

coefficient of mobility of the ions we use the sections of elastic col-
lisions of the NO + ion with N 2, O 2 molecules calculated taking into
account the polarisation interaction of the particles by the method
described in Ref.5. The coefficient of mobility of the electrons is de-
termined using the data on the sections of elastic collisions of the elec-
trons with nitrogen and oxygen molecules published in Ref.32. In the
numerical solution of the problem of the probe, the value of b i was
varied to explain the nature of the dependence of the current on the
probe. The following data were used in the calculations: b i = 38.6 cm 2/
(V×s); b e = 2·10 4 cm 2/(V×s); a* = 1; a 0 = 5·10 –8 cm 3/s; 31 W 0 = 10 16
cm –3 ×s –1 ; R p = 0.015 cm; S =1.51 cm 2; j 0p = –9 V; j p = –5 V. The
variants with j p = –11; –12.8 V were also considered.
Figure 12.16 shows the results of numerical calculations of the time
dependence of probe current for a stepped change of the negative po-
tential of the probe in relation to the earthed walls of the shock pipe.
For quantitative analysis of the calculation results of transient cur-
rents it was convenient to transfer to a new dimensionless variable:
I1c - I1 (t ) I 22 - I 2 (t ) t ch 1
Y1 = , Y2 = , x= = , (12.34)
eW0 R p S eW0 R p S t - timp Dt (W0a 0 )1/ 2
where I 1 (t) and I 2 (t) are the instantaneous values of current; t imp are
the moments of variation of the potential (t 0 or t 1, respectively); Dt =
t – t imp.
Figure 12.17 shows, in the form of dependences Y 1(x) and Y 2(x),
the results of calculations of the transient currents I 1 (t) and I 2 (t) on
the probe for different values of the equilibrium concentrations of the
charged particles of the plasma under the stationary regime conditions.
The concentration values varied by changing the ionisation rate at a
constant recombination coefficient.
The numerical solution of the problem (12.1)–(12.32) enables us to
328
Fig.12.17 Dependence of Ψ 1 and Ψ 2 on ξ.
determine the dependence of the current of the probe on the param-

eters which control the state of the plasma. For this purpose, we carried
out calculations in which the values of W, a, d i , |Dj| were varied at
different values of the initial probe potential. In all variants the value
|Dj| was taken as lower than |Dj 0p |.
The non-stationary current in the transient sections t 0 £ t < t 1 and
t1 £ t < t2, excluding a short period of time of the order of 0.5 µs where
bias currents play a significant role, depends mainly on the concen-
tration of the charged particles, non-perturbed by the probe in the quasi-
neutral region of the plasma and on the mobility coefficient of the ions.
Separating the dependences on the varied parameters, we obtain the
approximation relationships
0,35
Y1 = 0,025d i0.5n* Dj* F (x), (12.35)
0.35
Y2 = 0.0375d i0.5n* Dj * f (x), (12.36)
The functions F (x) and f (x) are presented in Fig.12.18. They

depend only slightly on the parameters mentioned previously, and can
be regarded as constant for a set of dependences 1−4 and 1¢–4¢, shown
in Fig.12.17 and corresponding to a wide range of the variation of the
parameters n* and d i.
329
Plasma Diagnostics
Fig.12.18 Dependence of F and f on ξ.
The relationships (12.35) and (12.36) include the main plasma pa-
rameters: W, a, b i. These relationships can be used as diagnostic equa-
tions for determining any plasma parameters both directly and together
with relationships for stationary probe currents. In this case, we al-
ready have two independent equations and can determine two unknown
plasma parameters.
The function F (x) and f (x) can be approximated by analytical re-
lationships with the accuracy sufficient for measurements in practice.
Consequently, we obtained the following dimensionless dependences on
the plasma parameters for transition currents:
0.35  1 
∆I1 = 1.11 ⋅10 −20 R p LW 0.75α −0.25bi0.5T 0.15 ∆ϕ  − 1 ,
 (W α) (t − t1 )
0.25 0.5

(12.37)
0.35  1 
∆I 2 = 1.39 ⋅10−20 R p LW 0.75α −0.25bi0.5T 0.15 ∆ϕ  − 2 .
 (W α) (t − t2 )
0.25 0.5

(12.38)
Here R p, L are the radius and length of the cylindrical probe, cm;
b i is the coefficient of mobility of the ions, cm 2/(V s); T is the equi-
librium plasma temperature, K; Dj is the amplitude of variation of the
probe potential, V; W is the ionisation rate, cm 3×s –1 ; a is the recom-
330
bination factor, cm –3 ×s –1. The method of determining DI i and t – t i is

shown in Fig.12.19.
The plasma temperature and the coefficient of mobility of the ions
will be regarded as known. Consequently, from the equations (12.37)
and (12.38) we can derive equations for determining the ionisation rate
and the effective recombination coefficient
∆Ii0 Ai
W= , α= ,
(i ) Ai (∆ti0 )0.5 (i ) 3
∆Ii (∆ti0 )1.5
0 (12.39)
A i = B iR pLb i0.5 T 0.15 |Dj| 0.35 , B 1 = 1.11·10 –20, B 2= 2.78·10 –20, i = 1.2.
The method of determining the values DI 0i and Dt 0i is shown in
Fig.12.19. It can be seen that in the co-ordinates |I i –I i (t)| 1/(t –t i ) 0.5
the time dependences of the transient currents are straightened out.
The points of intersection of these straight lines with the ordinates and
abscissa give the values of DI 0i and Dt 0i, respectively. Index 1 relates
to the case in which the pulse potential reduces the absolute value of
the probe potential, and the index 2 corresponds to the opposite case.
These cases are completely equivalent when determining W and a and
either of them can be used.
Experimental verification of the resultant relationships was carried
Fig.12.19 Dependence of transition currents on time for determining ∆I 0i and ∆t0i . I,

II relate to the currents in Fig.12.16.
331
Plasma Diagnostics
out on a shock pipe in a series of experiments with air in the tempera-

ture range T = 2700 ¸ 3200 K and at the atmospheric pressure be-
hind the reflected shock wave. The gas parameters were calculated
from the measured velocity of the incident shock wave taking into
account the chemical reactions of dissociation and ionisation using the
method proposed in Ref.33. the following conclusions can be drawn
as a result of comparing the calculated and experimental results.
1. The concentrations of charged particles, obtained on the basis
of the transient currents with the pulsed change of the probe poten-
tial, are in satisfactory agreement with the concentrations determined
from the results of probe measurements at a constant probe potential
in the section of the stationary concentrations of charged particles in
the plasma.
2. Determination of the recombination factor of the charged par-
ticles using the two methods gives values close to the literature data.
3. The pulsed probe can be used to determine the coefficient of mo-
bility of the positive ions if the recombination factor of the charged
particle is available.
4. The concentration of charged particles, obtained from the probe
measurements, is 1.5–2 times higher than the concentration of elec-
trons measured using the microwave interferometer.
We shall examine the method of diagnostics of chemically react-
ing plasma by using an electric probe working in the regime of potential
modulation by high-frequency sinusoidal voltage.
The methods available at present are based on using approximate
analytical solutions of the problem of probe impedance.36–39 This problem
is complicated by the need to calculate the propagation of weak
perturbations caused by the potential modulation, through the highly
heterogeneous state of the medium in the vicinity of the probe. Therefore,
the procedure of determining the analytical solution is accompanied by
a large number of simplifying assumptions whose validity for specific
experimental conditions cannot always be checked. The analytical
solution derived as a result which link the impudence characteristics
with the plasma parameters are often complicated and cumbersome and
cannot in many cases be resolved in the explicit form in relation to the
required plasma parameters. It is possible to solve only a direct problem
− obtain agreement between the theoretical and experimental data on
impedance by substituting, into analytical relationships, appropriately
selected values of the parameters of the medium examined in the ex-
periment. 39
The solution of the inverse problem, i.e determination of the plasma
parameters from the impedance measured in the experiments, requires
332
further considerable simplifications which can greatly restrict the range

of applicability of the final relationships.
We shall examine plasma consisting of one type of positive ions and
free electrons. The rate of formation of the charged particles w is de-
scribed in this model by two effective parameters w = W – a n i n e ,
where W is the effective rate of chemical ionisation, a is the coeffi-
cient of electron–ion recombination. The values of these parameters
give the equilibrium quasineutral density of the charged particles.
The system of equations which determines the distribution of the
densities of charged particles and electric potential j in the approxi-
mation of the solid medium, is formulated in the region of the cylin-
drical probe in the same manner as the equations (12.2)–(12.5).
The boundary conditions are written in the form
r * = 1, ni* (1, t * ) = ne* (1, t * ) = 0,

j* (1, t * ) = j*p + Dj* sin (2pf *t * ),
j*p , Dj* = const, Dj* > 0, (12.40)
r * = rm* (rm* >> 1), ¶ns* / ¶r * = 0, s = i, e, j* (rm* , t * ) = 0.
Here Dj* is the amplitude of the variable component of probe

potentials; j*p is the constant component of the potential; f* = ft ch , f
is the modulation frequency of the potential. We shall analyse the case
of negative potentials ϕ*p < 0, the absence of convective transport of
charged particles, and thermal diffusion. Inertia terms in the relationships
for the current density of charged components are not investigated.
The latter circumstance restricts the modulation frequency of the probe
potential: f << n i, where n i is the frequency of collisions of ions with
the neutral particles in plasma. The ionisation rate and the recombi-
nation factor are constant with time.
The problem was solved numerically using the difference method
implicit with respect to time. The solution of the non-linear system of
different equations in each time layer was determined using iterations
in relation to the modulation frequency of the potential. Usually, the
ratio of the time step to the period of potential change was not greater
than 5·10 –3. It was thus possible to determine the phase shift of cur-
rent on the probe with the error not greater ±2°.
The numerical solution gives the distribution of the quantities n*i, n*e,
j* in the vicinity of the probe at every moment of time. These dis-
tributions were used to calculate the value j* and the time depend-
333
Plasma Diagnostics
ence of the total dimension current on the probe I = 2pR pLj idJ*. The
calculations show that under conditions in which |Dj*/j* p| < 0.2, the
function j* (t*) can be approximated with sufficient accuracy using
their relationship
j * (t * ) = j0* + Dj * sin (2 pf *t * + Y ), Dj * > 0, (12.41)
where j*0 is the stationary value of the dimensionless density of the

current on the probe at a constant probe potential j*p, Y is the phase
shift between the current and the potential; Dj* is the amplitude of the
variable component of current density.
We shall examine the results of the numerical solution of the problem
obtained in the values of the parameters: b i = 38.6 cm 2/(V s); b e =
2·10 4 cm 2(V s); a = 1.5·10 –8 cm 3 s –1; W = 10 16 cm –3 s –1; n i,e = 4.47·10 11
cm –3 ; R p = 1.5·10 –2 cm; j = –9 V; Dj = 1 V; f = 10 5 Hz. These pa-
rameters correspond to air plasma at T = 3000 K and atmospheric
pressure. It should be noted that for these conditions the simplifying
conditions at which the authors of Ref.37–39 obtained the approximate
solutions of the impedance problem are not fulfilled.
In accordance with equation (12.3), the total current density recorded
by the high-frequency probe consists of the conduction current (the
first two terms in the right-hand part of (12.3)) and the bias current.
In the numerical solution it is possible to find separately each com-
ponent and analyse its frequency dependence. With increasing modulation
frequency the phase difference between current and potential tends
to 90° because the layer of the volume charge resulting in the appearance
of an effective capacitance in the probe circuit does not manage to
react to the potential change. The amplitude of the oscillations of bias
current increases and that of the conduction current decreases.
The aim of solving the impedance problem is to obtain relationships
for determining the plasma parameters from the current amplitudes
(measured at different frequencies) on the probe DI and the phase shift
Y. The equations (12.1)−(12.4), (12.40) show that the values of Y and
Dj* (amplitude Dj* is proportional to the measured amplitude DI) are
determined completely by the values of the dimensionless parameters
d* i, d* e, e*, j*p, Dj*, f*.
The numerical solutions enables us to find Dj* and Y for each set
of the values for these parameters. The values of j*p, Dj* at the known
plasma temperature are given, other quantities depend on the plasma
properties and the modulation frequency of the potential f. It is pos-
sible to estimate the range of variation of these parameters by examining
the permissible range of the values of W, a, b i, b e for different gas
334
mixtures at the atmospheric pressure and a temperature of 2000–3500

K. For the negative probe potential the variation of d e in the range
5·10 2 – 1·10 4 has only a very slight effect on the values of j*0, Dj*,
and therefore, calculation of d e from the data on the value of Dj* is
incorrect. The range of variation of other parameters will be restricted
on the basis of the estimates by the following inequalities:
0.1 £ W / W0 £ 50; 1 £ a / a 0 £ 100; 0.1 £ bi / b0 £ 10; 2 £ T / T0 £ 4;

W0 = 1016 cm -3 × s -1 ; a 0 = 5 × 10 -8 cm 3 × s -1 ;
(12.42)
b0 = 40 cm 2 / (W × s); T0 = 10 3 K .
These inequalities determine the relatively wide range of the val-

ues of the effective parameters W, a, b i and enable the following
relationship to be used for a large number of gas mixture compositions
which are interesting for practice.
The given values of j p, Dj, f were varied in the following ranges:
–45 V £ j p £ –1 V; 0.5 V £ Dj £ 5 V; 10 5 Hz £ f £ 10 7 Hz. In
calculations, the value of ∆ϕ for every value of j p was selected in
such a manner as to fulfil the relationship |Dj/j p | < 0.2. The
dimensionless quantities e*, d*i varied in the range 10 –4 < e < 2·10 –2 ;
0.1 < d i < 30.
Analysis of the results of numerical calculations, the amplitude of
the variable component of the current density on the probe and the phase
shift between the current and the potential can be written in the form
1/ 2
Dj * = ( Dj *p ) 2 + ( Djc* ) 2 ,
(12.43)
Y = arctan ( Djc* / Dj *p ),
where Dj*c, Dj*p are the amplitudes of the density of bias current and
a variable component of the conduction current density, respectively.
In the examined range of the variation of the parameters, the results
of the numerical solution can be approximated by the following rela-
tionships
0.9
Dj *p = 0.967(d *i ) -0.2 ( e * ) -0.06 j *p ( f * ) -0.1 ,
-0.4 (12.44)
Djc* = 8.12(d *i ) -1.04 ( e * ) 0.74 j *p Dj * ( f * ).
335
Plasma Diagnostics
The resultant equations (12.43) and (12.44) can be used to determine

the effective rate of ionisation W and the coefficient of mobility of the
ions b i on the basis of the results of measuring the amplitude of the
variable component of current on the probe and the phase shift be-
tween the current and the potential.
We shall write the relationships (12.44) in a dimensionsless form
-0.1
DI p = 2 pRjid Dj *p = A p f ,
DI c = 2 pR p Ljid Djc* = Ac f ,
-0.6
A p = 8.74 × 104 R 0p.52 LDj 0.9bi0.8W 0.68 T 0.44 j p a -0.38 , (12.45)
-0.4
Ac = 2.615 × 10 -6 R p LDjbi-0.04W 0.15T 0.1 j p a -0.11 .
Here DI p, DI c are the amplitudes of the total conduction current and

of the bias current on the probe.
Solving the equations (12.45) in relation to the quantities b i and W,
we obtain
bi = Cb DI p DI c-4.62 ,
-1.24
W = CW DI p0.27 R p2.24 L3.6 Dj 3.7 f 4.72
jp a -0.12 ,
(12.46)
2.33
CW = 9.5 × 10 28
R p-3.4 L-5.7 f -5.4
jp a 0.7
T -0.66
Dj -5.67
.
The values of DIp, DIc are determined from the amplitude of the vari-
able component of total current on the probe (measured at a given fre-
quency) DI and the phase shift between the current and the potential
Y using the relationships
DI p = DI cos Y; DI c = DI sin Y. (12.47)
When the measurement of the phase shift at frequency f is diffi-

cult, the values of DI p, DI c can be obtained from the values of DI 1 and
DI 2 – the amplitudes of the total current measured at two different fre-
quencies f 1 and f 2. Consequently, to determine A i and A c we obtain
336
A p = D( DI12 × f 22 - DI 22 × f 12 )1/ 2 , f1 < f 2 ,

Ac = D( DI 22 × f1-0.2 - DI12 × f 2-0.2 )1/ 2 ,
(12.48)
D = ( f 22 × f 1-0.2 - f 12 × f 2-0.2 ) -1/ 2 .
The values of DI p and DI c are calculated from the resultant values

of A p and A c for the corresponding frequency. The equations (12.45)
- (12.48) can be used to estimate the error in determining the values
of b i, W at a known error of measuring the current amplitudes at each
of the frequencies. For example, if the error of measuring the current
is 10%, the error of determining b i and W is approximately 100% of
the frequency of f = 10 6 Hz. 40
Thus, equations (12.46) at a known temperature and the recombi-
nation factor can be used to calculate the coefficient of mobility of
the ions and the effective rate of ionisation in equilibrium plasma. For
this purpose, it is necessary to measure, at the selected frequency, the
amplitude of the variable component of the current on the probe and
the phase shift between the current and the probe potential or measure
the current amplitude at two different frequencies. The unperturbed
equilibrium concentration of the electrons is then determined from the
relationship n 0 = (W/a) 1/2 assuming the mechanism of chemical ioni-
sation and dissociative recombination.
337
Plasma Diagnostics
Chapter 13
ELECTRIC AND THERMAL PROBES IN THE

PRESENCE OF CHEMICAL REACTIONS IN
NON-EQUILIBRIUM PLASMA
13.1 Electric probe

Recently, non-equilibrium electric discharges at reduced pressures have
been used on an increasing scale in the technology of depositing and
removing (etching) thin layers on the surface (from the surface) of
solids. 1–3 Layers can be dielectric, semiconducting or conducting, and
have optical, diffusion and other properties interesting from the applied
viewpoint.
It was therefore necessary to carry out diagnostics of the electronic
and ionic components of the plasma under the conditions when dielectric
or other coatings grow on the probe surface or when the dimensions
of the probe decrease as a result of etching its surface layers. Simi-
lar changes can also occur in structural elements of the probe device
(current conducting lines, insulation, etc.). Consequently, deviations occur
from the normal operating regime of the probe, i.e disruption of some
important postulates of the theory of the probe method. In order to
resolve these contradictions, it was proposed to use the electric probe
also for the diagnostics of the process of growth of the dielectric polymer
layers 4,5 and then for diagnostics of the kinetics of growth and removal
(etching) of layers of different nature. 6–10,25–27 At present, the elec-
tric probe in the film-forming or etching plasma is used for the diag-
nostics of electronic and ionic components of the plasma and also for
examining the kinetics of the layers on the surface of the probe/sub-
strate.
In the first case, theoretical fundamentals of the method, developed
for plasma diagnostics without chemical reactions, remain unchanged.
However, during measurements it is necessary to take measures to clean
the surface to remove dielectric deposits, and take into account the
338
Electric and Thermal Probes in the Presence of Chemical Reactions
Fig.13.1 The probe-wire device. 1,2) connected conductors; 3,4) insulating capillaries;
5) conical plug; 6) conical ground joint; 7) discharge tube; 8) working part of the probe.
Fig.13.2 (right) Block diagram of the probe circuit for measuring U p , r pf and VAC of
the probe. 8.10 1) power source for the discharge; 2,4) electrodes; 3) discharge tube; 5)
probe; 6) bias voltage source of the probe; 7) unit for setting and measuring probe
temperature; 8) current stabiliser; 9) unit for matching the stabiliser with the automatic
recording device and setting I p ; 10) automatic recording device.
change of the dimensions of the probe as a result of deposition on its

surface of conducting or semiconducting layers or as a result of etching
the probe material.
In discharges in hydrocarbons, the probe can be efficiently cleaned
to remove the polymer film grown on it during rapid heating of the probe
to a temperature of T p ¾ 900 ¸ 1000 K by a current from a special
current source. The probe suitable for this operation has two lead-outs
(Fig.13.1). The diagram of the electric circuit of the probe, used in dc
glow discharge, 8,10 is shown in Fig.13.2. After rapid heating, the volt-
ampere characteristic or its derivatives with respect to voltage are
measured in the pulsed regime during the period in which the probe
surface remains clean. Measurements with a heated probe can also
be taken. The temperature of the probe is regulated by the filament
current and should be selected on the level at which no deposits form
on the probe surface. For discharging hydrocarbons, the temperature
should be around 800–900 K.
The discharges in the vapours of fluorine- and chlorine-containing
compounds are used to etch metallic and semiconductor materials.3 Un-
der these conditions, it is possible to etch metals, including noble ones,
which are often used to produce electric probes. The etching rate of
noble metals is usually r etch Ù 0.1 nm/s, and the radius of the probe-
wire r p ¾ 10 µm.
The variation of the dimensions of the probe made of gold or platinum
remains unchanged over an operating period of 1 h. This estimate re-
quires verification by experiments.
339
Plasma Diagnostics
Thus, the occurrence of the processes of deposition of the layers

on the probe or etching of the probe material complicates the appli-
cation of the probe method of diagnostics of charged particles in prac-
tice. After all, we have mentioned here only some partial cases of the
processes of etching and depositing the layers. Their number rapidly
increases. 1,3,9
The kinetics of growth and etching of the layers on the surface of
the solids under the conditions of electric discharges at reduced pressure
has been examined using the methods of optical interferometry, micro-
scopic, weighing methods, etc. Each method has its advantages and
disadvantages.11 The electric probe method 8,10 is suitable for examining
the level of charged plasma particles in heterogeneous stages of the
process. 8–10,25 It can also be used to examine some electrophysical prop-
erties of the dielectric layer during their growth. 6,8,10,12,25–27
We shall examine the principle of this method on an example of a
dc discharge in hydrocarbons, where the polymer film with the dielectric
properties grows on the surface of the probe/substrate, and the probe
has the form of a thin wire with a diameter of 2r p ¾ 10 ¸ 20 µm and
a length of l p ¾ 1 cm (Fig.13.1). In this case, the method has the fol-
lowing characteristics mentioned in Ref. 7,8,10,12–15,25.
1) high sensitivity, the capacity to record changes of film thickness
of several nanometers;
2) reliable recording of the first monolayers of the film on the probe-
substrate;
3) deposition of a film at the given densities of electron fluxes W e
and ion fluxes W i of the plasma on its surface, where the range of
specified values of W e and W i is more than 10 orders of magnitude;
4) localisation of measurements determined by the probe diameter;
5) negligible (at the given probe dimensions) thermal and geometrical
perturbations of the plasma in the vicinity of the probe;
6) definition of the temperature of the probe/substrate T p in a wide
range T melt Ú T p Ú T 0p, where T m is the melting point of the probe
material, T 0p is the probe temperature in the plasma with preheating
current switched off (T 0p can differ from the gas temperature of
plasma); 8,10
7) it is convenient to combine in the same structural device of the
electric probe for investigating the kinetics of film growth, the Langmuir–
Mottsmith probe for diagnostics of charged plasma particles and a
microcalorimeter for examining the thermal effects on the surface of
the probe/substrate (in the latter case, the probe temperature is meas-
ured on the basis of its electrical resistance using, for example, the
device shown in Fig.13.3);
340
8) modification of the method for high-frequency discharge, 9,25–27

for diagnostics, by this method, of the etching processes, 9 for using, as
the probe, flat substrates 9,25–27 non-conducting substrates, etc. 8
Figure 13.4 shows a set of the VAC of the probe/wire in a dc glow
discharge in a mixture of neon with methane at different moments of
growth of the polymer film (PF) on its surface. It is assumed that the
thickness and all other properties of the PF are the same over the entire
working surface of the probe (the probe–plasma contact surface), that
the plasma is homogeneous in the vicinity of probe and that the probe
is placed in the equipotential plane of the electric field in plasma. We
shall make a section through the set of the VAC by the straight line
I p = const, where I p is the probe current. The same current also flows
through the PF, thus I p = I PF . The transition from one VAC to another
at I p = const is possible at an appropriate change of the voltage U of
the external source (see Fig.13.2) |DU| = I pR PF , where R PF is electric
resistance of the PF at a given moment of time. In other words, with
the growth of the PF the densities of the currents W e and W i of the
charged particle from the plasma on the external (facing the plasma)
surface of the PF remain unchanged
I p = W e + W i = f (j pl – j PF ) = const, (13.1)
is the difference of the potentials of this surface j PF and the plasma

j pl is constant. It is evident that
(j PF – j pl) = const at |DU| = I pR PF . (13.2)
Of course, all other the quantities characterising the probe, the plasma
and the film are constant with time. The distribution function of the
plasma electrons with respect to energy is arbitrary, but not necessarily
D2
D1
K153UD2 K153UD2
Fig.13.3 Principal electrical diagram of a device for measuring the temperature of the
probe–wire. RX =30÷100 ohm is the probe resistance, R 1=10 kohm, R 2=40 ohm, R3=100
ohm, R 4 =R 5=R 6 =12 kohm, R 7 =1 Mohm, R 8 =10 kohm; D 1 ,D 2 are amplifiers.
341
Plasma Diagnostics
I p , µA
∆U p
I p = const
Fig.13.4 Time dependence of

U p, V the VAC of the probe with
Um growing dielectric film. Growth
time of the film t 5 >t 4>t 2 >t 1 .
Maxwellian. The condition required for applying the theory of the

Langmuir–Mottsmith probe and, in particular, the ratio of the Debye
length l D and the mean length l e of the free path of the electrons in
the plasma l e >> l D, may not be fulfilled. The film thickness d PF, sat-
isfy the condition
d PF << R p (13.3)
The size of the working surface of the probe-wire, coated with the
film, does not depend on the film thickness (with the accuracy of ful-
filling the condition (13.3)). If the conditions (13.1) and (13.3) are si-
multaneously fulfilled and the plasma parameters are constant, the
condition (13.2) is also fulfilled. For a flat probe with any thickness
of the PF the area of the outer surface of the film is constant with
time. We do not examine the growth of the first monolayers of the film
when the properties of the probe such as the work function of the elec-
tron changes. This disrupts the condition (13.2). Evidently, the large
changes of the useful signal DU, observed in the initial period of the
growth of the PF 7,26 are caused mainly by this effect.
The VAC of polymer films, measured during its growth 8,10,12,25–27
proved to be non-linear and non-symmetric in relation to the origin of
the co-ordinates (Fig.13.5). The value of RPF, included in equation (13.2),
does not have the meaning of the ‘ohmic’ resistance. In Ref.7, 8, 10,
13, 25, 26, it was shown that at the given discharge conditions of the
current probe of I p the value DU is directly proportional to the film
thickness of the probe d PF at d PF Ù 100 nm. Consequently, it can be
assumed that
R PF = rd PF at I p = const, (13.4)
342
where r has the meaning of ‘specific’ resistance of the film. Under

these conditions, the value r does not depend on the thickness and,
consequently, the mechanism of conductivity through the film remains
unchanged.
The problem of the mechanism of conductivity through the thin
(d PF Ù 100 nm) polymer films grown in glow discharge remains un-
changed. Until now, all investigations into this problem were carried
out after extracting the films from the reactor and depositing the up-
per electrodes on them by some method. Experimental data shown in
Fig.13.5 are unique in that they were obtained during film growth where
one electrode of the measuring cell was the probe/wire, the second
electrode with the plasma around the probe. The method of produc-
ing the VAC during film growth will be examined below.
The mechanism of passage of the electric current through dielec-
tric polymer films up to 250 nm thick, produced by plasma chemical
polymerisation of organic substances, was investigated in Ref.16,17,
18,19 and 20. A detailed analysis of the experimental and theoretical
data on the passage of current through the thin directive films was
published by Simmons. 21 It was shown here that the electrical char-
acteristics of the electrode–thin dielectric film–electrode cell are de-
termined not only by the special features of the dielectric material of
the film. They also depend on the nature of the electrode–dielectric
contact.
At a low strength of the field in the film, the passage of current
can be governed by the Ohm law but the current intensity in this case
is very low. 21 A current of high intensity appears at a field strength
of the film of E PF = 10 4 ¸ 10 5 V×cm –1. The VAC shown in Fig.13.5
was measured at E PF Ú 10 5 cm –1 . In such high fields, the current
I *p , µA
Fig.13.5 VAC of polymer films

produced in Xe+1.7% vol. CH 4 1
min after start of growth. Current
Ip at which the polymer film grows:
1) 10 µA, 2,4) 0, 3) +10 µA;
"measuring current" I*p: 2) I*p = |Ie|–
|I j |>0, 4) I*p = |I e |–|I j |<0; P = 240
∆U, V
Pa, I p = 50 mA.
343
Plasma Diagnostics
through a thin polymer film is transported by the electrons (accord-

ing to the view of a large number of investigators) and ‘the bottleneck’
of the transport process is the generation of current carriers. This gen-
eration takes place as a result of the emission of Schottky electrons
from the electrode into the film and/or the melt (ionisation) of acceptor-
donor centres in the film in a strong homogeneous electric field by the
Poole–Frenkel mechanism. 20,21 The transport mechanism in which the
current is restricted by the spatial charge (CRTS) does not correspond
to the experimental data, as shown in Ref.20 and 21.
For the conductivity mechanism by the CRTS, the dependence of
current through the film on the voltage of the cell U PF and the film
thickness d PF is described by the equation 20
l +1 2 l +1
I PF ~ U PF / d PF , l Ú 1, (13.5)
or
ln I PF ~ (l + 1) ln U PF at d PF = const.
When restricting the current by the generation of current carriers

by the Schottky and Poole-Frenkel mechanisms
/ 2 -1/ 2
ln I PF = (b / kT )U 1PF d PF + C, (13.6)
where k is the Boltzmann constant; T is the cell temperature; C is the

constant; b = (e 3 /apee 0) 1/2 , a = 4 for the Schottky mechanism and
a = 1 for the Poole–Frenkel mechanism; e is electron charge; e is the
dielectric constant of the film; e 0 is the permittivity of the vacuum.
Thus, b PF = 2b SCH .
The majority of experimental data are in agreement with the de-
pendence lnI PF ~ U 1/2 , i.e they confirm that the current is restricted
by the generation of current carriers by the Schottky and/or Poole-
Frenkel mechanism. 20 For example, in Ref.22, for films obtained in
plasma chemical polymerisation of the styrol better agreement was
obtained regarding the value of b with the hypothesis on the conductivity
by the Poole–Frenkel mechanism, and, in addition to this, the VAC of
the film did not depend on polarity of the electrodes. This is typical
of the Schottky emission. The electrical conductivity of films of styrol
was examined in greater detail in Ref.23. The VAC measured in Ref.10,
12, 25 and 26 (Fig.13.5) was non-symmetric in relation to the varia-
tion of the current direction. However, the specific feature of the ‘meas-
344
uring cell’ where plasma can be used as one of the electrodes, does
not make it possible to make an unambiguous conclusion on the oc-
currence of the Schottky conductivity mechanism in this case.
Equation (13.6) shows that a constant current I PF, the voltage on
the film U PF and its thickness d PF are proportional in relation to each
other: U PF = Bd PF , where B is a constant typical of the given cell.
Consequently, the value r, included in equation (13.4) is related with
the value V by the ratio r = B/I PF . It is evident that B = E PF is the
mean strength of the field in the film with respect to its thickness. It
is not known in advance how the value of B changes with the
variation of the conditions of plasma chemical polymerisation: pressure,
chemical composition of the flow rate of the plasma forming gas,
specific energy contribution to the plasma, the density of the flows of
charged particles of the plasma on the surface of the growing film,
etc. In the experiments with discharge the mixtures of xenon on neon
with methane 10,13 and argon with perfluorobutane 25,27 the authors com-
pared the data on the growth rate of the film, obtained by the probe
and optical methods. They show that under the operating conditions 10,13
a variation of the pressure of the mixture from 0.3 to 4 torr and a dis-
charge current of 20 to 75 mA and a film thickness of d PF = 1 ¸ 100
nm the value of B remain constant with an error not greater than
15–25%. For other discharge conditions, the constancy of the value
B must be verified by experiments. If B = const is fulfilled, the probe
method can be used to examine the kinetics of growth of the film and
some electrophysical properties of the latter.
For the conditions of discharge in the mixture of xenon or neon with
methane the electrical resistance of the film growing on the probe
changes by only 1–2%, and the variation of the temperature of the
probe–substrate by DT p = 50 K and T p ¿ 350 K. Consequently, the
probe method can be used to examine the dependence of the kinet-
ics of film growth on substrate temperature.8,10,15 Since the thermal inertia
of the probe/wire is not high – around 0.1s, on the basis of the vari-
ation of DU PF at the moment of variation of the probe temperature we
can estimate the activation energy of the electrical conductivity of the
film during its growth. In Ref.6,10,12–15 it was of the order of
E a ~ 0.1 eV.
As already mentioned, the main advantage of the probe method is
that it enables the film to be grown on the probe at different densi-
ties of the electron and plasma ion fluxes on its surface and, at the
same time, reveals the role of the charged particles in heterogeneous
stages of plasma chemical polymerisation. The following procedure is
used: the film is grown at a given discharge conditions at different
values of I p, i.e at different points of the VAC of the probe. However,
345
Plasma Diagnostics
I p = I PF, i.e a displacement of the probe along the VAC indicates si-
multaneous displacement on the VAC of the film. Since the VAC of
the film is non-linear, direct comparison of the growth rates at different
values of Ip on the basis of the rate of variation DU PF is incorrect. This
problem is solved by two methods.
The film is grown at unchanged current I p during a relatively long
period of time. The probe/wire is then extracted from the reactor and
its thickness is measured using an optical microscope. The procedure
is repeated for a number of other values of I p. From the VAC of the
heated probe we determine the correspondence of I p to the densities
of the electron fluxes W e and ion fluxes W i so that we can determine
the dependence of the growth rate of the film r PF on W e and W i. This
method can also be used to examine the kinetics of growth of the
conducting and semiconducting films.
It is also suitable for dielectric substrates, for example, in the form
of glass filaments. 8 In the latter case, measurements are simple only
at Ip = 0, i.e at the point of ‘the floating’ potential, when We = Wi. Re-
placing the glass filament by the capillary with an inserted wire-heater
enabled the dependence of the growth rate on substrate temperature
to be examined. The use of the thin capillary with the diameter smaller
than the mean length of free path of the molecules enables extensive
thermal disruption of the plasma to be avoided. This method can be
used to examine efficiently the distribution of the growth rate over the
volume of the reactor.
Fig.13.6 Electrical block diagram of the probe circuit. 1) reactor; 2,8) electrode; 3) the
probe coated with the dielectric film; 4) heating battery of the probe; 5) resistance for
regulating the heating current; 6) source of the bias voltage of the probe; 7) stabilising
of probe current I p .
Fig.13.7 (right) Principal electric diagram of current stabiliser. 1-3) transistors with an
isolated gate; 4) regulated voltage source; 5) microammeter, R 1, R 2 is the matching
voltage divider.
346
The second method is more complicated but provides a large amount

of information. A current stabiliser (Fig.13.6, 13.7) is introduced into
the probe circuit. The decrease of the voltage at the stabiliser is identical
with the amount of increase of the voltage drop on the film. To compare
the growth rates of the films at different I p (different densities of the
flows W e and W i), it is necessary to measure the rate of variation of
their resistances at the same point of the VAC of the film, i.e. at the
same current I*p. (It is assumed that the ‘specific resistance’ at a given
I*p is the same for the films grown at different Ip. This assumption must
be verified by experiments.) The following procedure can be used. Two
kinetic curves DU PF (t) for currents I p and I*p are recorded in an au-
tomatic recording device (Fig.13.8). The value of current Ip is selected
in such a manner as to ensure that the surface of the growing film
receives the required electron W e and ion W i fluxes. The current is se-
lected using the VAC of the ‘pure’ probe measured in advance (see
above). The regime of switching over the current stabiliser from I p to
I*p is specified in such a manner as to ensure that the film grows at
current I p for 90–95% of the entire growth period. Switching to the
‘measuring current’ I*p is carried out over short periods of time, equalling
5–10% of the total duration of the growth process. The procedure is
repeated for another value of I p, and I*p remains unchanged. Conse-
quently, this gave the dependences of the growth rate on the density
of the electron ion flux in the discharge in mixtures of neon or xenon
with methane 6,8,10,12 and argon with perfluorobutane. 25-27 If different
values of the ‘measuring current’ I*p are applied quite rapidly for short
periods of time whilst retaining unchanged film growth conditions, i.e
−∆U p , V
−∆U p , V
I *
p1 I p
I *
p2
I *
p3
I *
p4
t, s
t, s
Fig.13.8 Kinetic voltage curves on the film. U PF = –U p ; 1) U PF for I p ; 2) U*PF for I*p .
Fig.13.9 (right) Construction of the VAC characteristic of growing film on the basis
of kinetic curves obtained for different measuring currents I*p (k) (polymer film is
grown at probe current I p ).
347
Plasma Diagnostics
I p, then the resultant data can be used to measure the VAC of the film
(see Fig.13.5) during its growth at different moments of time (Fig.13.9).
The absolute values of the growth rates of the film are determined by
calibrating the probe method by the first method (see above) or by some
other independent method. To facilitate measurements of both parts of
the VAC of the film we can use the ‘bridge’ circuit of the current sta-
biliser.
Cleaning the probe to remove the polymer film is carried out by rapid
heating the probe with a current from a special source (see Fig.13.2).
To set and measure the temperature of the probe-wire produced in the
form of a loop (see Fig.13.2) we can use a device whose main electrical
circuit is shown in Fig.13.3. The circuit of the device was proposed
in Ref.24. A cartridge with resistances can be used as a variable resistor
R 3. The values of R 2 and R 3 for the probe-wire made of platinum wire
10–20 µm in diameter and approximately 1 cm long are around 20–
50 W. Regulating R 1 we set a current through the probe I heat and its
temperature T p . The cartridge of the resistances R 3 is used to cali-
brate the output device in the values of the resistance of the probe
at each current I heat. The transition from the values of the probe re-
sistance to its temperature is evident:
R 3 (T p) = R 3 (T 0) (1 + aDT p) (13.7)
where a is the temperature coefficient of resistance of the wire.

Measurements of the dependence of the growth rate of the film on
Tp at given Ip and the discharge regime (Fig.13.10) enable the effective
activation energy of heterogeneous stages of the film growth to be de-
termined.
At a film thickness of the order of d PF ¾ 100 ¸ 200 nm the au-
thors of Ref.7 observed an electrical breakdown of the film at a given
−∆U p , V
Fig.13.10 Measurement of the

dependence of the rate of growth on
dielectric film on substrate tempe-
ratures of the given I p . T 3 >T 2 >T 1
is the temperatures of the probe -
substrate, ∆U p is the voltage jump
on the film with the variation of
t, s temperature.
348
I p. Evidently, the conductivity mechanism changed here, other ava-

lanche-like processes developed and caused heating and destruction of
the film (Fig.13.8).
Until now, no experiments have been carried out to examine the prob-
lem of the process of polarisation of the dielectric film during its growth
and the effect of polarisation on the error of measurements of the
growth rate by the method described previously. When switching from
current I*p to I p the ratio (13.2) is disrupted if an electric polarisation
field forms in the film. This also relates to the case in which the probe
method is used to examine the etching kinetics of dielectric films.
Evidently, to examine the film polarisation it is convenient to use the
probe method based on alternating current. 8 It can also be used to
examine the kinetics of growth (etching) the film and its VAC.
The method is based on the following procedure. A source of
sinusoidal voltage of low amplitude u 0 sin wt, where eu 0 << e , e is
introduced into the probe circuit (see Fig.13.2) in series with the dc
source; here e is the mean energy of plasma electrons expressed in
electron-volts; e is the electron charge. A condenser is connected in
parallel to the current stabiliser. The value of the capacitance and the
frequency w of the weak signal are selected such that the stabiliser
‘does not smooth’ the current components formed in the probe circuit
with a frequency equal to or multiplicable by ω. The amplitude of the
first harmonics of the variable component of the current is proportional
to dI p/du p. Here u p is the potential of the probe-substrate at a point
of the VAC of the probe coated with a film corresponding to constant
current I p. The value of the current is set by the current stabiliser. The
amplitude of the component of the current with a frequency of 2ω is
proportional to d 2I p/du 2p. 8 It is evident that the resistances of the probe
(i.e dual electric layer in the plasma at its vicinity) of the film are con-
nected in series. If at the initial moment of time the probe is clean,
the variation of the variable signals with a frequency w, 2w, etc. with
time is proportional to the variation of the first, second, and other de-
rivatives of the VAC of the film with respect to time. Recording of
these signals makes it possible to examine the kinetics of film growth
and fine effects of the variation of its VAC during growth (etching).
The amplitude and procedure for measurements using alternating cur-
rent were described in the section concerned with the probe measure-
ments of the function of the energy distribution of the electrons and
also in Ref.8.
Frequency w can easily be measured over a wide range which in
principle enables polarisation effects of the film to be detected. The
main difficulty in measurements using alternating current is associated
349
Plasma Diagnostics
with a sensitivity to the ‘noise’ of the plasma and instability of the

discharge.
13.2 Thermal probe

In the conditions of non-equilibrium electric discharges at reduced pres-
sure, the thermal probe method is used to determine the gas temperature
of the plasma and thermal effects of the recombination and deactivation
reactions on the surface. 8 Two design variants of the thermal probe
are used – a thermocouple coated or not coated with a dielectric
screen, and thin wire with two lead-outs. In the first case, the tem-
perature of the thermal probe is determined from the measured value
of the emf of the thermocouple, in the second case it is determined
from the change of the electric resistance of the wire. The first variant
is far simpler as regards practical application. The second variant has
a considerably higher information content as a result of the possibil-
ity of regulating its temperature by means of an external current source
during the process of measuring the heat flows on the probe. It is also
convenient that the same device can be used as a microcalorimeter,
the electric Langmuir–Mottsmith probe and the probe for investigat-
ing the kinetics of growth (etching) of the layers. Both variants use
in the explicit form models of the mechanisms of the processes of heating
plasma and energy exchange between the probe, plasma and the re-
actor. This is the most important feature of the microcalorimeter (thermal
probe) method when used for problems of non-equilibrium plasma di-
agnostics. We shall examine it on an example of the thermal probe in
the form of heated wire immersed into the plasma of dc glow discharge,
under the deposition regime of a polymer film.
The thermal balance of the thermal probe-wire is described by equa-
tion
Q h.t + Q e + Q rad + Q he + Q Ip + Q eq + Q ch + Q PF + Q c + Q dea = 0

(13.8)
Here Q h.t is the heat flow through the side surface of the thermal
probe as a result the change of the kinetic energy of the neutral plasma
particles in collision with the surface of the probe, Q e is the heat flow
through the ends of the probe-wire on the holders, Q rad is the heat
exchange of the probe with the environment through radiation, Q he is
the heating of the probe-wire by current by the external source, in-
cluding in the operation of measuring the wire temperature from its
electrical resistance (see Figs.13.2, 13.3), Q eq is the heating of the ther-
mal probe as a result of passage of probe current I p through it when
the thermal probe like the Langmuir–Mottsmith electric probe is not
350
Ie
Ip
I he
C1 C2
Fig.13.11 Simplified principal electrical diagram of the microcalorimeter. 1,3) leadouts

of the microcalorimeter; 2) microcalorimeter – wire; 4,5) holders; 6) electrode–anode
of the discharge tube; 7) device for setting and measuring heating current I he of
microcalorimeter measuring its temperature; 8) circuit for setting probe current I p and
measuring the VAC of the probe–mircocalorimeter; A1 , A2 are the current measurement
device, B 1 , B 2 are the voltage measurement devices, C 1 and C 2 are power sources, R 1
is the regulated resistor, I e is the equalisation current.
|I p|, µA; ∆T w , K
U p, V
Fig.13.12 Dependence of probe temperature and fluxes on the probe of ions I j and
electrons Ie on probe potential Up in direct current discharge in Ar + 10% vol H2. Pressure
P = 0.3 torr, discharge current I D = 25 mA, diameter of the discharge tube 5 cm; 1) I e,
2) T pc = T p–T w, T w is the wall temperature of the tube; 3) I j (measurements by V.I.
Timakin).
at the ‘floating’ potential (Figs.13.11, 13.12), Q eq is the heating of the

thermal probe by the ‘equalisation’ currents caused by differences in
the potentials of the plasma along the probe, Q ch is the heat flow
through the side surface of the thermal probe as a result of chemi-
cal reactions on the surface (recombination of radicals, hydrogen at-
oms, etc.), Q PF is the heat generation in the polymer film growing on
351
Plasma Diagnostics
the surface of the thermal probe, in passage of equalisation current

I eq through it and the probe current I p, Q c is the heat generation in
recombination of the charged particles and accommodation of the ki-
netic energy on the surface of the thermal probe, Q dea is the heat gen-
eration in deactivation of the electronically and vibrationally excited
plasma particles on the surface of the thermal probe. It will be assumed
that d p << l, where d p is the diameter of the thermal probe, l is the
mean path of the neutral particles in the plasma. This simplifies cal-
culations of the fluxes of the particles from the plasma on the probe
surface.
The equation (13.8) can be used in practice only if a number of sim-
plifying assumptions are made.
The term Q h.t cannot be measured directly in the plasma or calcu-
lated with sufficient accuracy. In the gas of the same chemical com-
position and pressure, Q h.t can be easily measured as the function
Q h.t = f (T p – Tg), where T g is the gas temperature, T p is the temperature
of the probe/wire, and T p > T g. The quantity T p can be specified and
measured using a device shown in Fig.13.3. It can assumed that at small
differences of the temperatures T p and T g the modulus Q h.t does not
depend on the sign of the difference T p – T g and in transition from the
gas to the plasma with the same gas temperature and the chemical
composition Q h.t does not change. It is assumed that the characteris-
tics of the probe surface in both cases are the same.
The total value (Q e + Q rad) = j (T p, T w) is measured by heating the
probe-wire with current I he at the pressure in the reactor without dis-
charge at which the term Q h.t can be ignored. According to the ex-
perimental data published in Ref.10 it is sufficient to set the pressure
below (2 ¸ 3)10 –2 torr. The temperature of the reactor walls T w should
be set in the range of its values under the experimental discharge
conditions. It should be noted that a given temperature T w the tempera-
ture of the walls and the contacting leads under the discharge con-
dition is slightly higher than T w. The dimensions of the thermal probe
should be selected such as to ensure that the end losses are minimum
(large length and small probe diameter).
The measured quantities Q he = I 2he R p , where R p is the resistance
of the probe-wire at a given current I he , represent part of the proce-
dure of measuring the probe temperature from the probe resistance
using a device whose diagram is shown in Fig.13.3. It was assumed
that the resistance of the conducting leads is far less than R p.
The upper limit of quantity QI p can be easily estimated assuming
that the probe current I p flows along the entire length of the thermal
probe. Usually, the value of QI p can be ignored if the probe is not cov-
352
ered by the dielectric film. This also applies to Q eq because the equali-
sation current is of the same order of magnitude as I p.
If the thermal probe is covered with a dielectric film, the situation
does not change. For example, for a probe with a length of 1 cm and
a diameter of 10 µm in a discharge in the mixtures of argon with meth-
ane at a pressure of 1 torr Q h.t ¾ 10 –3 ¸ 10 –2 W. For the probe current
I p ¾ 10 –5 A and the thickness of the dielectric film d PF ≈ 100 nm the
voltage drop in the film reaches Vfi » 100 V, i.e. Q PF ¾ 10–3 W is com-
parable with Q h.t. According to the estimate, the extent of heating the
film with the equalisation current can also be significantly high.
The term Q ch describes heat generation in the probe during chemical
reactions on its surface. It can be reliably measured in discharges and
mixtures of inert gases with hydrogen. Here it is interpreted as the ther-
mal effect of recombination of the hydrogen atoms under probe sur-
face: Q ch = g recW (H) e rec, where g rec is the thermal coefficient of re-
combination; W (H) is the flux of hydrogen atoms onto the probe sur-
face; g rec is the binding energy of the atoms in the H 2 molecules. The
values of e rec are available for some materials, for example platinum. 29
If d p << l (see previously), then W ( H ) = 1 / 4[ H ]vt , where [H] is the
atom concentration in the plasma; vt is the mean thermal velocity of
the atoms. Evidently, interpretation of Q ch does not cause any diffi-
culties in other cases as well.
Q p can be evaluated by experiments by measuring the temperature
of the probe-wire Tp at different points of the volt–ampere characteristic
of the probe, i.e at different values of the probe current I p (see
Fig.13.11). The minimum T p is observed in the vicinity of the ‘float-
ing’ potential of the probe where the electronic and ionic components
I p are equal as regards their modulus. The thermal effects of the ions
and electrons can be separated only away from the ‘floating’ poten-
tial. The mechanism of heating the probe as a result of the kinetic and
potential energy of the charged particle remains insufficiently exam-
ined because only approximate polarisation from remote parts of the
VAC to the ‘floating’ potential of the contributions of the ions and
electrons is possible. For reliable measurements of other terms (usu-
ally Q h.t and/or Q ch ) it is important to verify by experiment that Q p is
small.
When interpreting the meaning of the term Q da it is important to re-
member that the following condition is fulfilled in the plasma of non-
equilibrium electric discharges at reduced pressure:
T g » T rot Ù T vib < T el
353
Plasma Diagnostics
where T g, T rot, T vib , T el is the gas (translational), rotational, vibrational

and electronic ‘temperatures’ of the plasma particles, respectively (the
concept of the temperature of the internal degrees of freedom of the
neutral particles in the non-equilibrium plasma is not always correct,1,8,10,13
as indicated by placing the term in inverted commas). It is usually
assumed that T g @ T rot . In discharges in nitrogen, hydrogen and mix-
tures of nitrogen with CO2, Tvib was considerably higher than Tg.30 Under
these conditions, the heating of the thermal probe in the deactivation
process of vibrationally-excited molecules on its surface can be con-
siderable. 31 In non-equilibrium plasma of hydrocarbons, as a result of
rapid vibrational–translational relaxation, according to the estimates,
T vib @ T g at a pressure of P Ú 0.1 torr and a specific energy con-
tribution of W Ù 1 W×cm –3 . 10 The calorific effect of the deactivation
of electronically excited particles should be taken into account in, for
example, discharges with a low energy contribution W Ù 10 –2 W×
cm–3 in inert gases with small (Ù1 mol.%) additions of molecular gases.
The concentrations of the metastable excited atoms in such a plasma
reach A M ¾ 10 10 ¸ 10 11 cm –3 . 10,32
Figure 13.12 illustrates the meaning of the terms Q p, Q eq, Q he, Q PF .
Thus, the relationship between the quantities included in (13.8) de-
pends on the external parameters of the discharge: pressure, chemi-
cal composition of the plasma forming gas, specific energy contribu-
tion, the presence of the dielectric coating on the surface of the thermal
probe, etc. Preliminary evaluation of the thermal effects can be car-
ried out on the basis of the order of magnitude or even more approxi-
mately. Therefore, the use of the method is associated with carrying
out a number of procedure experiments: measuring Q s.t in the gas with-
out discharge, Q e + Q rad in vacuum, Q e in plasma at different points
of the VAC of the thermal probe in the electric probe regime. It may
be useful to compare the measurements in the given reactor in the dis-
charge in the inert gas (where there are no heterogeneous chemical
reactions on the probe surface) and in the ‘working’ gas, etc.
Another useful the method is used mainly for examining the ther-
mal effects of chemical reactions on the surface under the condition
that the gas temperature of the plasma must be obtained by other ex-
perimental or calculation methods. This comment applies to both types
of the thermal probe: thermocouple and the heated wire. The use of
the same design of the device enables three diagnostic methods to be
used (see previously) widens the possibility of each method under the
conditions of investigating the extremely complicated object –
non-equilibrium plasma – in the presence of chemical reactions.
354
The method can be used efficiently for measuring the gas tempera-
ture of the plasma Tg in cases in which in equation (13.8) we can ignore
the terms Q da , Q p, Q PF , Q ch , Q eq , Q Ip . Most difficulties are associated
with the preliminary evaluation of the value of the term Q ch . Assum-
ing that the thermal probe is thin
F1v I F1v I
Qch / Qht @
H4 t K H4
R g rec e rec / t K
N ak (Tg - Tp ) , (13.9)
where a and g rec have the thermal effects of accommodation and re-
combination on the surface, respectively; vt is the thermal velocity of
the heavy particles of the plasma, e rec is the binding energy (recom-
bination energy) or the radicals, k is the Boltzmann constant, [N] are
the concentration of radicals and the total concentration of the heavy
plasma particles. For example, for a thermal probe made of platinum
in a discharge in a mixture of argon with hydrogen at a pressure
P ¾ 1 torr we can accept that: a ¾ 1; g rec ¾ 10 –2 ; e rec ¾ 4 eV; T g
– T p ¾ 10 ¸ 100 K; [H] ¾ 10 14 –10 15 cm –3 ; [N] ¾ 10 16 cm –3 .
Substituting into (13.9) gives Q ch/Q h.t ¾ 1·10 –2 ¸ 1, i.e. Q ch and Q h.t
are the same order of magnitude. If the platinum thermal probe-wire
is coated with a dielectric film (for example, by polymerisation in a
discharge), then e rec should greatly decrease. 33 Consequently,
Q c/Q h.t << 1 so that Q h.t. and T g can be measured.
In high-frequency discharge the thermal probe is heated by the elec-
tromagnetic field. Heating can be easily evaluated by experiments by
switching the hf field at low pressure in the reactor when there is no
gas discharge. Special features of using the electric probe in rf and
microwave discharges have been examined in Chapter 10.
355
Plasma Diagnostics
Chapter 14
MEASUREMENTS OF THE ENTHALPY OF

HIGH-TEMPERATURE GAS FLOWS
14.1 Introduction
Enthalpy is one of the most important characteristics of high-temperature
media, especially moving ones, in the presence of ionisation. Enthalpy
data are especially important in applying various technologies.
Determination of enthalpy using different devices started at the end
of the 50s when work was carried out in many countries to develop
high-temperature processes and equipment. The largest number of de-
velopments of the measuring devices was reported in the 60s and
70s. 1–60 Measurement methods were proposed and substantiated, ap-
propriate sensors were developed and subjected to successful tests in
measuring the enthalpy of different high-temperature media: inert, hot,
plasma. At present, the methods and devices for measuring the enthalpy
are usually developed along the path of improving and automating the
measurements.
In the group of the methods of measuring enthalpy there are usu-
ally three most important groups: 8,21,37 1) sound flow, 2) direct deter-
mination of the enthalpy by taking gas samples, 3) heat flow – deter-
mination of the enthalpy from the heat exchange of the solid with the
gas flow. The first of these methods, which used special thermal probes,
was not used widely because of complicated design of the probes and
limited possibilities in comparison with the other two methods 8 which
are used extensively. The main type of device are enthalpy probes (or
sensors) based on gas samples. The enthalpy probe is a device intro-
duced into the gas flow in order to measure the flow enthalpy. The
enthalpy sensor is an enthalpy probe in a complex with a sensitive
element ensuring reception of an electric signal.
No detailed analysis of the design of probes and methods of meas-
urement have been carried out from the moment of development of
the first devices for enthalpy measurements. Some studies in this di-
rection are of limited nature and reflect only partial aspects of measuring
356
Measurements of the Enthalpy of High-Temperature Gas Flows
the enthalpy of high-temperature media. It should be noted that the

enthalpy of the low-temperature flows is measured by conventional
methods, for example, using thermocouples, thermal resistors and other
heat-sensitive elements.
14.2 Main methods of measuring the enthalpy of

high-temperature gas
The currently used methods of measuring the enthalpy of high-tempera-
ture gas flows are based on the second and third method.
The second method has been used widely; it makes it possible to
obtain high accuracy of measurements and develop various devices for
enthalpy measurements.
Determination of enthalpy by calorimetric measurements. The cur-

rent available methods of measuring the enthalpy by calorimetric meas-
urements of the gas sample in the application range relate to two different
groups – stationary and non-stationary. Historically, the stationary
method was the first one to be developed it was then used on a ba-
sis for developing various devices – enthalpy probes, sensors, etc. The
founder of the method is seen to be Grey 2,3,6,8 who founded its own
company in the USA (Grey-rad) producing enthalpy sensors included
in recording devices. Subsequently, another USA (Calprobe) started
to produce sensors on the basis of Grey’s licence.
It should be noted that the devices for determining the character-
istics of the flow by taking gas samples were constructed a long time
ago. For example, Fig.14.1 shows a probe 1 for taking gas samples and
examining their chemical composition. This probe is very similar to
enthalpy probes. Grey 8 also reported that prototypes of enthalpy probes
existed prior to 1962 when a method of measuring the enthalpy using
probes was substantiated for the first time.
Fig.14.1 Microprobe for examining flames. 1) attachment, 2) calorimeter, 3) inlet and

outlet of water, 4) discharge of gas.
357
Plasma Diagnostics
L meas
T 2, G B
T g , C pg , G g
T1
Water
Fig.14.2 Stationary enthalpy probe. 1) measuring, 2) outer tube, 3) screen, 4,5) inlet
and outlet of water, 5) discharge of gas.
Stationary method. The device using the stationary method of

enthalpy measurement is shown in Fig.14.2. The main elements are the
internal tube 1 for sucking in (taking sample) of the gas and the outer
tube 2 for protecting the sensor against failure in the high-tempera-
ture flow and for ensuring that the shape of the sensor is such that
it minimises the flow perturbation. The tail part of the sensor usually
contains tubes for introducing and discharging the cooling liquid, sucking
away the gas and the screen 3 for protecting auxiliary devices against
the effect of the flow. Enthalpy determination is based on two meas-
urements. The first measurement makes it possible to find the heat flow
Q 1 on the probe without sucking the gas through the internal tube of
the probe, i.e. the amount of heat received by the probe per unit time
only as a result of an exchange of the flow with the outer surface.
The result of the second measurement – heat flow Q 2 on the probe
with sucking away the gas with flow rate G g when the probe is in con-
tact with the high-temperature flow on both the outer and inner sur-
faces. Taking into account the results of these measurements, the
enthalpy of the flow is calculated from
Q2 - Q1
h0 = + C pg Tg , (14.1)
Gg
where Q 2 = G lC plDT 2; Q 1 = G l1 G pl DT 1; DT 2 = T 2 – T 0, DT 1 = T 1 – T 0 ;
G l, C pl, T is the flow rate, specific heat capacity and the temperature
of the liquid cooling the probe; C pg , T g is the specific heat capacity
and temperature of the gas probe at the exit from the probe.
Usually, the flow rate of the cooling liquid is maintained constant,
i.e. G l1 = G l2 = G l, and
358
C pl Gl
h0 = (T2 − T1 ) + C pg Tg , (14.2)
Gg
It should be noted that h0 is the total enthalpy including the enthalpy

of the frozen flow and its kinetic energy
h0 = h + v 2 / 2, (14.3)
where h, v are the enthalpy and speed of the flow.

Ignoring the kinetic energy of the flow increases the enthalpy of
the frozen flow.
The speed of the incident flow can be determined using the same
sensor. During measurement of the heat flow on the probe without
sucking away the gas, a device for measuring the pressure is connected
to the measuring tube, i.e. the enthalpy probe operates as a pressure
sensor p:
p = p0 + Cd rv 2 / 2, (14.4)
where p 0 is the static pressure in the region of the inlet orifice of the
probe. The value p 0 is measured by other sensors; for jet flows with
subsonic speed at the atmospheric pressure p 0 close to atmospheric
pressure and, consequently,
p = Cd rv 2 / 2. (14.5)
Coefficient C d which takes into account ‘sticking of the gas’ at low

Reynolds numbers, as determined from the dependence C d = f (Re). 37
In a number of cases Re > 10 C d ¾ 1, and with satisfactory accu-
racy r = rv 2/2, i.e. v 2 = 2p/r, where r ¾ A/h;37 A is a constant. Thus,
h is determined from the transcendental equation
h0 = h + 2 p / ρ, ρ = f ( h ), (14.6)
where h 0 and p are measured by the same enthalpy probe.

The problems of design of the probes, sensitivity and error of meas-
urements will be examined in detail below and recommendations will
be made for using other probes.
359
Plasma Diagnostics
Fig.14.3 Nonstationary enthalpy probe. 1) specimen measuring tube, 2) shielding shell-

screen, 3) high-speed valve, 4) evacuated cavity, 5) connection of vacuum pump, 6)
shielding screen, 7) thermal insulation, 8) thermocouple.
The non-stationary method. The design of the device illustrating the

non stationary method of determining gas enthalpy 19 is shown in
Fig.14.3. The main element here is the standard measuring tube 1 with
the wall thickness d and the base length L (the section in which its
temperature is measured using, for example, a thermocouple). On the
outside there is a shielding shell whose shape defines, as in the sta-
tionary method, the nature of heat exchange of the probe with the flow
and ensures the minimum heat flow. At the outlet of the probe there
is a high-speed valve, an evacuated container and a system for sucking
away the gas.
The measuring tube is used in such a manner that the entire en-
ergy of the gas flow, sucked away by the pump, is absorbed in the tube.
In this case, the temperature of the tube continuously increases when
taking gas samples. It is important to prevent the penetration of the
heat flow through the outer tube, i.e. the dimensions of the air or some
other space between the tubes 1 and 2 should be such as to ensure
that the heat flow to the tube 1 during the measurement arrives only
from the gas sample. The equation, describing the variation of the
enthalpy of the gas, passing through the measuring tube, has the form
v 2 C phρhVh ∂ 1
L
Q
h − hout + = Th ( x )dx ± loss ,
∫ (14.7)
2 Gg ∂t L 0 Gg
where h out is the enthalpy of the gas at the outlet of the probe, C p,
r, V, T is the heat capacity, density, volume, gas temperature at the
inlet into the probe, index t relates to the measuring tube, g to the gas.
Q loss /G g characterises the error of determining the enthalpy as a re-
sult of losses or additional supply of heat. The value h out is in the
majority of cases close to zero because the probe is designed in such
360
a manner that the entire energy of the sucked away gas is absorbed
by the measuring tube. The derivative determines the variation of the
mean temperature of the tube, i.e.
L
∂ 1 ∂T
∫T ( x ) dx = m . (14.8)
∂t L 0 ∂t
Consequently
v 2 C ph rhVh dTm Qloss

h - hout = - = ± . (14.9)
2 Gg dt Gg
Producing an enthalpy probe using a specific procedure it is easy

to ensure that h out » 0 and the speed v can be measured with the same
probe in the regime without sucking away the gases; the measuring
tube is connected to a high-speed pressure recording device, for ex-
ample, a condenser microphone (at low pressures) or to a pressure
gauge. The value Q loss /G g with other conditions being equal should be
insignificantly small. Under these conditions
v 2 C ph rhVh dTm
h+ = (14.10)
2 Gg dt
The flow rate of the gas to the measuring tube is calculated measuring
the rate of variation of pressure in constant volume 4 (see Fig.14.3)
from the equation
Gg = Adp / dt , (14.11)
where dp/dt is the rate of variation of the pressure in the filled vol-
ume, and A = const for the given volume and the given gas. Thus, we
obtain
v 2 C ph rhVh dTm / dt dT
h+ = =B m (14.12)
2 A dp / dt dp
The value dT m/dp is determined by experiments; if the flow speed
361
Plasma Diagnostics
is known, the enthalpy of the gas in the examined flow is calculated.

This approach was used as a basis for developing a large number
of devices. Their design, special features of application and applica-
tion areas will be described below in a separate section.
Determination of enthalpy from heat exchange. The method of
determining the enthalpy of the gas on the basis of heat exchange with
the solid, introduced into the flow, is less accurate than the method
examined previously but can be applied more easily in a number of cases
because it is simpler: when disruptions of the high-temperature flow
should be as low as possible or when the flow dimensions are very
small and when the stationary method of measuring enthalpy described
above is not applicable.
This method has been known for some time. 4,5,10,21,22,26,33 The solid
used when examining heat exchange is represented either by a sim-
ple solid, for example, a cylinder with longitudinal or transverse flow
around, 26,37,44 or a flat solid or solid enabling modelling of some
technological process, the conditions of interaction of the high-tempera-
ture medium with the component, etc.
The flow enthalpy is measured using the following procedure. For
example, we examine a cylinder with a transverse flow around. 37 In
this case, the density of the heat flow is defined by the equation
Dh
q=a , (14.13)
Cp
where a is the heat transfer coefficient; Dh is the enthalpy pressure

(Dh = h f – h w); h f , h w is the enthalpy of the flow at the temperature
of the flow and the wall of the sensor, respectively; C p is the heat
capacity of the gas. The heat transfer coefficient is generally deter-
mined from the criterial dependence
l
a= Nu, (14.14)
d
where the Nusselt criterion Nu is determined from the equation
0.4  ρ f µ f 
 h∂ 
Nu = 0.5 Re 0.5
f Pr f  (
 1 + Le f − 1
0.52
) , (14.15)
 ρ w µ w   h f 
362
where l is heat conductivity, d is the cylinder diameter, ρ, µ is the den-

sity and viscosity of the gas, Re, Pr, Le are the Reynolds, Prandtl and
Lewis criteria, the indices "f" and "w" relate the thermophysical prop-
erties of the gas to the flow temperature or wall temperature.
If the right hand part of the (14.14) is divided by Öp, where p is
the dynamic pressure, taking above considerations into account, we
obtain
q λ Nu ∆h
= = f (T ) , (14.16)
p d p Cp
where
0.2
λ f ρ 0.25  ρfµf   h  ∆h
f (T ) = 0.6
µ 0.5
f
0.5
Pr 0.4

 ρw µ w

 ( 0.52
)
1 + Le f − 1 ∂  , (14.17)
f df    h f  C pw
and the value f (T) can be easily calculated if we know the proper-
ties of the gas.
Thus, measuring experimentally the density of the heat flow q and
the dynamic pressure p, we determined the enthalpy of the high-tem-
perature gas flow from the dependence f (T).
This method has been tested many times37 in flows of various gases:
in argon, 37 oxygen 24 ad others, and can be recommended for the tem-
perature region characterised by dissociation, ionisation and other re-
actions associated with energy release. In this case, the measurement
error decreases because the dependence f (T) is very steep. Disad-
vantages of the method include the complicated and sometimes inac-
curate dependence of q / 4 p on the thermophysical properties of the
gas.
14.3 Enthalpy sensors and probes

The development and improvement of the methods of measuring the
enthalpy of high-temperature gases, which started in the 60s, was ac-
companied by development of various devices for enthalpy measure-
ments – sensors and probes, with the most intense activity observed
up to the middle of the 70s. In later stages, the development of the
measurement procedures continued along the path of optimisation and
improvement of the existing devices57,58 and development of automated
systems. 59.60
363
Plasma Diagnostics
Detailed analysis of the design solutions and procedure special fea-

tures of determining the enthalpy makes it possible to classify the
enthalpy probes and design two large groups, stationary and non sta-
tionary enthalpy probes (in accordance with the method of measuring
enthalpy by calorimetric measurements of gas samples). The device,
designed for measuring the enthalpy on the basis of the heat flow on
the probe, fully corresponds to the heat sensors used for evaluating
the heat flows are not examined here.
Stationary enthalpy probes. According to the application, these
probes are divided into probes for subsonic and supersonic flows. They
have many similar features, especially in the procedural part, but also
some design differences.
There are various designs of these probes. According to the posi-
tion of the measuring part of the probe in the flow, there are probes
with longitudinal and transverse flow-around (they will be referred to
as longitudinal and transverse enthalpy probes).
Longitudinal enthalpy probe. As mentioned previously, the first
probe of this type was developed by Grey, et al., 2,3,6,7 and is therefore
referred to as Grey’s probe (Fig.14.4). It consists of three coaxial tubes;
two of these tubes, the inner and outer, are connected together at the
inlet part and form a hemispherical head with a measuring orifice at
the inlet. The intermediate tube is positioned between the outer and
the measuring tubes and is used to divide the cooling flow and directs
it into the inlet part of the probe. In the outlet part of the probe the
tubes are connected in such a manner that they ensure the inlet and
outlet of the cooling liquid, positioning of temperature gauges in the
inlet and outlet areas, for example, thermocouples, and connection of
Fig.14.4 Longitudinal enthalpy

probe. 1,2) inlet and outlet of water,
3) discharge of gas.
364
Fig.14.5 Outer view of longitudinal enthalpy probes with an outer diameter of d = 1.5
(a), 5 mm (b).
the sensor to the systems for gas sampling

and pressure measurements.
Figure 14.5 shows the external appearance
of longitudinal enthalpy probes whose design
corresponds to Fig.14.4. Only the methods of
supply and discharge of cooling water differ.
In Fig.14.5, the tube 1 in the probe is used
to supply water, the tube 2 is used for dis-
charging water and the tube 3 for taking a
gas sample.
In some cases, the enthalpy probes can be
produced in the form of compact replaceable
modules connected to the measuring unit us-
ing a sleeve with holes for supplying and dis-
charging the cooling liquid and gas sampling
(Fig.14.6). The measuring orifice is situated
in the nose part of the probe, and the gas is
sucked away from the rear side of the probe
into the end part of the sleeve. The cooling
water is supplied and discharged through ori-
fices made in the side surfaces of the sleeve.
This probe is easy to use, is placed in the lon-
gitudinal direction in the high-temperature
flow 13–15 (like the probe in Fig.14.2), and a
Fig.14.6 A replaceable shielding screen is placed in its tail part.
longitudinal enthalpy The two-tube enthalpy probe is similar to
probe. Grey’s probe as regards its design.32 The only
365
Plasma Diagnostics
Fig.14.7 Two-tube enthalpy probe. 1,2) inlet and outlet of water, 3) discharge of gas,
4) direction of the heat flow.
difference is the cooling system (Fig.14.7) Here the design of the in-
ternal measuring tube together with the outer tube ensures the required
cooling rate and defines the direction of movement of the cooling liquid.
Therefore, the intermediate tube is now required. Although the dimen-
sions of these probes can be very small, they are not yet used widely
because they do not make it possible to measure the enthalpy with the
required accuracy. At the circumference of the probe in the cross
section (A–A) there are two sections in which the internal measur-
ing tube is poorly cooled by the liquid. Consequently, the internal tube
receives the heat flow q that is difficult to control. For example, in
the first measurement (without sucking away the gas) calorimetric
measurements are taken to determine the heat flow to the probe Q 1
which travels through all the surfaces, including the flow q through the
regions of contact of the tubes. In the second measurement (with suck-
ing away the gas), the probe receives the heat flow Q 2 which includes
the heat flow brought in by the sucked-away gas. In this case, the
sucked-away gas transfers, in addition to its energy, also part of the
heat flow q (∆q), i.e.
Q 2 – Q 1 – Dq = G g Dh (14.18)
If Dq is comparable with the measurement enthalpy h (Dh = h –

h out), the error is large and this restricts the application of sensors of
this type, regardless of their simple design.
The enthalpy probe with internal heat insulation (Fig.14.8). This
probe is very similar to Grey’s probe and, as mentioned by Grey himself,8
had been known previously and was referred to as the calorimetric
probe. The presence of heat insulation reduces the intensity of heat
366
Fig.14.8 Enthalpy probe with a heat-insulated intermediate tube. 1,2) inlet and outlet
of water, 3) outlet of gas, 4) insulator.
exchange between the counterflows of the cooling liquid, increases probe

sensitivity and reduces the measurement error. However, these probes
are not used in practice because of production problems and problems
in developing the methods of decreasing the intensity of this heat ex-
change.
The enthalpy probe with the divided flow of the cooling liquid.
The diagram of the probe (Fig.14.9) shows that an additional interme-
diate tube is used here. Three outer tubes form the cooling systems. 9,17
There is one inlet for the cooling liquid and two outlets. 17 The cool-
ing liquid is supplied between two intermediate tubes and, consequently,
is directed to the area with highest thermal stresses, i.e. the inlet part
of the probe. Here the liquid is divided into two parts: one part cools
the probe from the outside, protecting it against the effect of the high-
temperature medium, and the other part of the liquid passes in the gap
between the intermediate and measuring tubes. The last part of the
liquid receives mainly the heat flow from the measuring tube. To measure
enthalpy it is sufficient to know the flow rate of this liquid (‘meas-
uring’ liquid) and its temperature difference (DT = T 2 – T 1) in the pres-
ence and absence of sucked away gas.
Thus, the slightly more complicated design of the probe and its larger
Fig.14.9 Enthalpy probe with the distributed flow of cooling liquid. 1,2) inlet and outlet
of water, 3) discharge of gas.
367
Plasma Diagnostics
dimensions are compensated by improved measurement sensitivity.

The enthalpy probe with a screen. There are two types of these
probes – with a cooled and uncooled screen.
The enthalpy probe with an uncooled screen slightly differs from
the probe without a screen. Usually, the screen is positioned on the
outer side of the probe, is made of a heat-insulating material, for ex-
ample, graphite 12,25 and repeats the shape of the probe. The screen
not only protects the probe against overheating but also increases the
sensitivity of calorimetric measurements of the gas sample if these meas-
urements are carried out on the background of a smaller total heat flow
on the probe. These enthalpy probes are characterised by several work-
ing conditions depending on the start of measurements in relation to
the moment of introduction of the probe into the high-temperature flow.
The first regime: the heat-insulating screen only started to be heated,
the temperature of its internal surface slightly differs from the initial
temperature, and the calorimetric measurements of the gas sample make
it possible to obtain the maximum sensitivity of the probe and the mini-
mum measurement error. Operation in this regime requires the use of
high-speed equipment. In the second regime, the heat flow to the outer
cooled tube of the probe increases almost linearly with time and this
must be taken into account in enthalpy measurements. However, the
error of measurement will be the highest of the three conditions ex-
amined. The measurements in the third regime are carried out in the
steady heat flow on the probe using the conventional procedure. In this
case, as a result of decreasing the heat flow the sensitivity and ac-
curacy of measurements increase.
The enthalpy probes with the uncooled screens are used on a limited
scale because the presence of the screen increases the dimensions and
amplifies the perturbing effect of the probe. They can be used effi-
ciently for the diagnostics of long flows.
The enthalpy probe with a cooled screen (Fig.14.10). This probe
Fig.14.10 Enthalpy probe with a cooled screen. For symbols see Fig.14.9.
368
is sometimes referred to as the ‘six-barrel probe’ according to the in-

cluded in probe design. 42,43,45 In this case, a screen of three coaxially
positioned tubes is placed on the normal longitudinal enthalpy probes
(Grey probe). A heat insulator, for example, air, is placed between the
measuring probe and the screen. 20 The cooling liquid is supplied into
the probe in such a manner that the internal tube of the screen and
the outer tube of the measuring part of the probe are cooled by the
liquid with the same temperature. The cooling liquid is fed in parallel
directions to these parts of the probe thus ensuring equal temperatures
of the walls of the measuring part of the screen in the absence of heat
overflow. Therefore, in random contact of the walls of tubes (as a result
of disruption of heat insulation or due to some other reason) the op-
erating regime of the probe is not disrupted.
The sensitivity of the probe of the accuracy of measurements are
sufficiently high because the parasitic heat flow to the measuring part
of the probe rapidly decreases. For example, according to the estimates
in Ref.27 and 42, the error of measuring the enthalpy decreases from
6 to 4%, although the perturbation which can be caused by the probe
with larger dimensions (the diameter of the probe with a screen 6.2
mm, without a screen 3 mm) 3 is not taken into account. Evidently, the
examined probes can be used efficiently for the diagnostics of flows
with large transverse dimensions which exceed at least by an order
of magnitude the outer diameter of the probe. 34
The enthalpy probe for measurements in supersonic flows
(Fig.14.11). These probes are usually of the six-barrel type, i.e. fit-
ted with screens. 18,23,26 The nature of gas flow in the top part of the
probe in the regimes with and without gas sampling can differ as a
result of changes in the position of the shock wave so that the speed
of gas sampling should be minimum. To ensure that the amount of gas
Fig.14.11 Enthalpy probe in a supersonic flow of high-gas. 1,2) inlet and outlet of
water, 3) discharge of gas, 4) shockwave.
369
Plasma Diagnostics
is sufficient from the viewpoint of ensuring required sensitivity, the size

of the measuring orifice must be increased. These considerations are
taken into account in the design of the probe in Fig.14.11. The screen
is separated from the calorimetric part of the probe by an air gap. In
the leading front of the probe all tubes are brazed together because
the generating lines of the outer and inner cones are positioned un-
der an angle of 30°. 42 The sensitivity of this probe is sufficiently high,
and enthalpy measurements does not require large amounts of this sam-
pled gas.
In Ref.42, the enthalpy probe was produced from 0.1C18Cr9NiTi
stainless steel: the outer tube with a diameter of 6.2 mm, d 0 = 4 mm,
d i = 1.8 mm (Fig.14.11). The large size of the orifice d 0 enables the
consumption of the sampled gas and the perturbing effect of the probe
to be reduced.
The ‘angular’ enthalpy probe. This probe was designed as a result
of improvement of longitudinal enthalpy sensors. Grey 8,11,16 also tried
to minimise the part of the probe introduced into the flow. Usually, the
probe was bent 38,40 in such a manner that its ‘body’ did not cause any
significant perturbations in the flow in the region of the measuring ori-
fice. As a result of a compromised variant, the ‘angular’ enthalpy probe
was developed.
This probe is shown schematically in Fig.14.12 and is external ap-
pearance is in Fig.14.13. Here, the measuring internal tube in the in-
Fig.14.12 "Angular" enthalpy probe.
370
Fig.14.13 External view of the angular

enthalpy probe.
let part is bent so that it is possible to place the initial part of this tube
along the flow, and the external pipe under an angle to the flow. The
intermediate tube is placed in the same position as in the longitudinal
probes and is used as a guide for the cooling liquid.
The ‘angular’ enthalpy probe is efficient in operation and retains
the advantages of the longitudinal enthalpy sensor. The angle under
which the probe is positioned in relation to the movement of the flow
is usually around 45°, although other variants are also possible. Po-
sitioning the probe under an angle of 45° ensures only slight perturbations
of the flow with sufficiently simple design of the top part of the probe.
The transverse enthalpy probe (Fig.14.14). In contrast to the lon-
gitudinal probes, the measuring part of the transverse probe is normal
to the flow and the relatively short initial section of the measuring tube
of the probe is positioned in the direction along the flow. The probe
consists only of two tubes: internal–measuring, and the external used
for transport of the cooling liquid; therefore, it is possible to reduce
the dimensions of the probe. For example, the authors of this book 22
produced a probe with an external diameter of 1 mm (external tube
1.0 × 0.85 mm, inner tube 0.5 × 0.4 mm) (Fig.14.15). A decrease of
the probe dimensions leads not only to positive results. In similar small
probes it is difficult to ensure high flow rates of the cooling liquid at
pressures usually encountered under the laboratory conditions; therefore,
decreasing the diameter of the external tube slightly reduces the pos-
sibilities of measuring high heat flows. The most suitable probes are
those with an external diameter of 2–3 mm which also have higher
371
Plasma Diagnostics
Fig.14.14 Transverse enthalpy probe, for symbols

see Fig.14.9.
Fig.14.15 External view of transverse enthalpy probes, diameter 2 (1) and 1 mm (2).
mechanical strength in comparison with probes with a smaller diam-

eter.
Disadvantages of the transverse enthalpy probe include, firstly a high
level of perturbations of the high-temperature flow determined by the
asymmetric flow-around of the flow around the body of the probe. Sec-
ondly, as a result of the small length of the initial section of the measuring
tube, positioned along the flow, the probability of appearance of dis-
tortions in the results of the measurements as a result of the distor-
tion of the gas flow lines in front of the probe decreases.
Disadvantages of the transverse enthalpy probe include, firstly, the
possibility of decreasing its transverse dimensions and the perturbations
caused by the probe. Secondly, the inlet and outlet parts of the probe
372
are separated thus reducing the error of measurements as a result of

preventing overflow of the heat. Thirdly, these probes are suitable for
measuring the flow characteristics in closed channels.
The transverse enthalpy probe was described for the first time in
Ref.22; it was developed on the basis of previously used 23 cylindrical
heat sensors with transverse flow-round. In subsequent stages, the
enthalpy probes of this type were used as a basis for constructing a
measuring system. 57,58 They were gradually improved and, as in the
case of longitudinal enthalpy probes, a number of their characteristics
were improved. 54,56 For example, sensitivity, service lifetime, etc. 30
A schematic of the transverse sensor is shown in Fig.14.16. Here
the enthalpy probe 1 is placed in the measuring block 2 made of or-
ganic glass. The probe is secured by bolts 6 with flanges 3. The gasket
4 and half rings 5 restrict the movement of the probe. The rubber seal
7 prevents penetration of the cooling agent into the gas circuit of the
probe. A differential battery of the thermocouples was installed to record
temperature. Only one half of this battery is shown in Fig.14.16. In
the upper part, the probe is secured through auxiliary straps with screws
to increase mechanical strength. Prior to measurements the straps are
removed.
Figure 14.17 shows a sketch of the design of the transverse enthalpy
probe with symmetric (in relation to the flow direction) sampling of gas,
and its external appearance is in Fig.14.18. The outlet ends of the meas-
uring tube are connected to both end pieces of the enthalpy probe so
Fig.14.16 Transverse enthalpy sensor.
373
Plasma Diagnostics
Fig.14.17 Transverse enthalpy probe with symmetric sampling of gas. 1) tip, 2) insert
3) inlet section of measuring tube, 4) insert, 5,6) auxiliary straps, 7) external, 8) measuring
tube, 9) securing screws.
Fig.14.18 External view of a transverse enthalpy probe with symmetric sampling of

gas.
that gas sampling, required under the measurement conditions, is carried

out in each half of the measuring tube with a lower velocity. This in-
creases the efficiency of heat transfer from the gas sample, and in
the majority of cases, it is not necessary to measure the gas temperature
at the outlet from the probe.
The transverse enthalpy probe with separate supply of the cool-
ing liquid. These probes are shown schematically in Fig.14.19 where
two main types are given: asymmetric and symmetric. A special feature
of these probes is the supply of the cooling liquid directly to the area
of position of the inlet hole of the inlet tube in the body of the probe.41
This liquid is then branched out (separated) and directed into the in-
let and outlet parts of the probe. At the same time, the total heat flow
in the measuring part of the flow rapidly decreases. Gas enthalpy is
determined from the following equation:
374
a Gg, T g b G g2, T g2
G l2 G l2
T2 T2
G l0
T0
CHANGE
G l0 G l1
T0 T1
G l1, T 1 G g 1 ,T g1
Fig.14.19 Asymmetric (a) and symmetric (b) transverse enthalpy probes with the distributed
supply of cooling liquid.
C pl Gl 2 ( 2 )
h0 = (T2 - T2(1) ) + C pg Tg , (14.19)
Gg
where T 2(2) and T 2(1) is the temperature of the cooling liquid flowing
through the measuring part of the probe, in sampling and without sampling
the gas, respectively. The values C l0, T 0, G l1, T 1 are used to measure
the total heat flow on the probe.
The examined enthalpy probes are characterised by high sensitiv-
ity and low measurement error. In addition, they operate with higher
flow rates of the cooling liquid and enable the range of measurement
of heat flows to be expanded.
The enthalpy probe with a symmetric supply of the cooling agent
is of slightly more complicated design but enables a different method
of measuring the enthalpy of a high-temperature gas flow to be ap-
plied, i.e. differential method.
375
Plasma Diagnostics
1/2 G l
Gl Gg
T0 Tg
1/2 G l
Fig.14.20 Differential enthalpy sensor.
The differential enthalpy sensor (Fig.14.20). This sensor includes

a transverse enthalpy probe with symmetric supply of the cooling agent
1, the distributor of the sucked-away gas 2 and the circuit for meas-
uring the temperature T 0, T 1, T 2 and the flow rate of the cooling liq-
uid G l . 52
The differential enthalpy operates as follows. The gas is sucked
away using a distributor of gas alternatively through the first or sec-
ond half of the probe. Thus, the enthalpy of the sucked away gas h e
is added to the heat flows arriving to the first (lower) or second (upper)
halves of the probe, Q 1 and Q 2, respectively. To increase the efficiency
of measurements, it is convenient to equalise the flow rates of the gas
G g in the flow rates of the cooling liquid through both halves of the
probe. In this case, in sucking away the gas through the first part of
the probe we have
Q 1 + h gG g = C plG l (T 1 – T 0) 1 (14.20)
Q 2 = C plG l (T 2 – T 0) 1 (14.21)
Consequently Q 1 + h g G g – C 2 = C plG g (T 1 – T 2) = C pl G l DT 1.
In sucking away the gas through the second part of the probe we
have
Q 1 = (Q 2 + h gG g ) = C pl G g (T 1 – T 2) 2 = C pl G lDT 2 (14.22)
Thus, the results of the first and second measurements give
h gG g = 1/2C pl G l (DT 1 – DT 2) (14.23)
376
so that we can determine the enthalpy from the equation
1 G
hg = C pl l ( ∆ T1 − ∆T2 ). (14.24)
2 Gr
Measuring G l, G g, DT 1 and DT 2, we can compute the enthalpy of
the gas.
On the whole, the procedure for measuring the enthalpy becomes
more complicated because instead of three parameters (G l , G g, DT)
it is necessary to measure four parameters (temperature difference is
measured twice). However, the sensitivity of measurements increases
because the total gas enthalpy is not measured on the background of
the total heat flow (Q 1 + Q 2) but on the background of the difference
(Q 1 – Q 2 ).
The planar enthalpy probe (Fig.14.21). The principle of the method
of enthalpy measurement is almost the same as in the previous case.
This probe has much in common with the enthalpy probes with a screen,
with the exception of the fact that the screen in the working part is
made in the form of a plane in order to enable measurements to be
carried out at higher heat loads and in modelling the flow-around of
Fig.14.21 Planar enthalpy probe. 1) flat screen, 2) enthalpy probe, 3) heat insulator,
4,5) inlet and outlet of water, 6) discharge of gas.
377
Plasma Diagnostics
the high-temperature flow around the plane, and the enthalpy probe in
its working part is made flush with the plane.
In measuring the enthalpy with these sensors, during taking gas up-
wards it is necessary to consider the nature of flow around of the liquid.
Of considerable importance here is the mutual position of the meas-
ured orifice and of the region where the high-region temperature flow
changes its direction. At a high flow rate of the gas during taking
samples of the gas and in the regime without sampling the positions
of the this region may differ thus increasing the measurement error.
In the majority of cases these probes are suitable for determining
the gas enthalpy in the near-axial region of the axis symmetric tem-
perature flows.
Nonstationary enthalpy probes. These probes are used far less
frequently than their stationary analogues, evidently due to their more
complicated design and the need to use high-speed apparatus. In ad-
dition, it is difficult to analyse the measurement error in these probes
and, as a whole, these probes and sensors have been described only
seldom in the literature.
The non stationary probes can be classified in two groups. The first
is similar to stationary probes where the enthalpy is determined by ca-
lorimetric measurements, and the second – the enthalpy sensor with
a compensator.
Fig.14.22 High-speed enthalpy probe. 51 1) shielding screen, 2) insulator, 3) threaded

joint, 4) securing of measuring pipe, 5) inlet of the second cooling circuit, 6) rubber
seal, 7) supply of cooling water, 8) discharge of cooling water.
378
High-speed enthalpy probe (Fig.14.22). 51 This probe consists of

three coaxial tubes with external diameters of respectively 1.0, 2.3 and
3.3 mm. The measuring channel of the probe into which the high-tem-
perature gas is introduced during measurement, has a diameter of 0.4
mm and is 100 mm. To exclude heat losses, the central tube is separated
from the measuring tube by an air gap of 0.35 mm. The cooling wa-
ter is supplied (7) to the front part of the probe, and flows through
the external shell (Z) between the outer and central tubes and leaves
outside in the vicinity of the inlet region, radial to the axis through eight
orifices with a diameter of 0.4 mm. Penetration of cooling water into
other parts of the probe is prevented by the seal (6). The body of the
probe is cooled using the second cooling circuit. Here the water is sup-
plied through the tube (5), flows through the outer part of the probe
body and is discharged through the second side outlet (8).
The enthalpy of the gas sample is determined by measuring the elec-
trical resistance of the central tube. This tube of the probe should be
made of nickel because of the almost linear dependence of its resistance
on temperature (usually to 340°C).
In accordance with the non-stationary method of measuring the
enthalpy it is necessary to know nature of the time dependence of the
tube temperature and pressure at the inlet into the enthalpy sensor (see
Section 14.1). The temperature of the measuring tube is determined
from its electrical resistance
L
ρ
R= ∫ S [1 + α
0
0 (Tx − T0 )]dx, (14.25a)
where T x is the temperature of the measuring tube along its length; r

is the specific resistance of the tube with length L; S is the effective
section of the tube equal to p(d 2out – d 2inn /4); a 0 is the temperature co-
efficient of resistance. At a 0 = const from equation (14.25a) we ob-
tain
L
R − R0 1
=
α0 R0 L 0 ∫
(Tx − T0 )dx. (14.25b)
Consequently DR/a 0R 0 = DT m.
Taking into account (14.12), we obtain an expression for the increase
of the enthalpy of the gas along the length of the measuring tube L
379
Plasma Diagnostics
dR dp dR
Dh = a / =a , (14.26)
dt dt dp
rt C ptVt
where a = ; rtVt is the mass of the tube; C pt is the heat ca-
a 0 R0
pacity of the tube material.
Consequently, enthalpy determination is reduced to simultaneous meas-
urement of resistance and pressure as a function of time. Constant a
is determined at relatively low temperatures at which the enthalpy can
be measured by another method, for example, using some thermal el-
ement.
In the experiments, the enthalpy probe is inserted into the high-tem-
perature flow for a short period of time. The water jacket protects it
against overheating. The probes with a water jacket 19 are used so that
the measurement time can be shortened. Such a probe was developed
in Ref.19. Its external diameter is 1.3 mm, with a diameter of the
measuring orifice being d i = 0.4 mm. The measurement time in a ni-
trogen flow with an enthalpy of h = 4.5 ¸ 12.8 kJ/g is 20–50 ms.
5 µs is sufficient for an enthalpy measurement cycle.
When using sensors of this type, the sampled gas flows through the
measuring pipe at a constant rate so that the pipe is usually connected
to the evacuated vessel. The enthalpy measurement procedure may be
described as follows: the probe is introduced into the plasma flow, the
resistance of the internal pipe of the probe is measured as a function
of time at a constant flow through the pipe. The quantity dp/dt is de-
termined by recording the pressure in the evacuated vessel as a function
of time.
Thus, the previously examined high-speed enthalpy probes enable
the enthalpy of the high-temperature gas flow to be determined within
the limits of milliseconds; this is especially important when long-term
holding of the probe in the flow is now possible or the duration of the
effect of the flow is limited.
Nonstationary enthalpy sensor with a compensator (Fig.14.23). Here,
in contrast to the previous enthalpy sensor, the sensor contains two
identical tubes. 48 One of them, the basic tube 2, is used for sampling
the gas, the other one is compensating. Thus, any body, placed in the
main tube, with the exception of the gas sample, for example, from
the external side of the probe, has no effect on the measurement results.
The main and compensating tubes (resistances R meas and R comp, respec-
tively) form a bridge circuit together with R1 and R2 which is connected,
through the external pipe of the probe, to a stabilised power source.
380
Gas samples
View C
SS
Fig.14.23 Nonstationary enthalpy sensor with a compensator. 1) enthalpy probe, 2)

measuring, 3) compensating tube, 4) current recording device, 5) stabilised source.
This circuit is characterised by high sensitivity and minimum circuitry

errors.
Advantages of non-stationary enthalpy probes are quite clear: 1) high
operating speed (measurement time is several milliseconds); 2) small
size (an enthalpy probe with a diameter of 1 mm can be produced);
3) enthalpy is measured within a single cycle – during taking gas sam-
ples, thus increasing the accuracy of measurements.
Disadvantages include higher requirements on the quality of assem-
bling the enthalpy probe – high accuracy of centring pipes is essen-
tial.
The nonstationary enthalpy probes are not yet used widely because
of complicated production and insufficient stability in the high-tempera-
ture flow. Erosion, cavitation, surface damage due to electric currents,
etc., greatly shorten the service life of the probes and have an adverse
effect of the extent of their application. Nevertheless, the given ad-
vantages of these devices often compensate these shortcomings.
14.4 Sensitivity of the enthalpy sensor and measurement

error
The sensitivity of the sensor σ is determined as the ratio of the heat
flow, received by the sensor from the gas sample in the measuring tube
to the total heat flow to the sensor as a whole from the examined flow:
Q2 - Q1 DT2 - DT1
s= = , (14.27)
Q1 DT1
where ∆T 2, ∆T 1 is the increase of temperature of the cooling water

with and without taking gas samples, respectively. Thus, the sensitivity
381
Plasma Diagnostics
of the enthalpy sensor is the fraction of the heat flow resulting from
taking gas samples on the general ‘background’ of the heat flow re-
ceived by the probe.
When measuring the temperature of cooling water with an error of
+0.1°C, the sensitivity of the sensor is close to 0.05. 22 In the general
case, the latter also depends on the following factors: the accuracy of
determination of the flow rate of cooling water and sucked-away gas,
the stability of parameters, determining the intensity of the heat flow
to the sensor.
Evaluation of sensitivity according to Ref.8

The flow of the high-temperature gas is laminar and the flow of the
gas in the measuring tube is governed by the Stokes law. Sensitivity
is evaluated from (14.27). Using the ratios 53 between the heat flow,
arriving to the measuring tube and gas sampling, at the heat flow on
the probe as a whole which takes into account the effect of the length
of the measuring tube and the gas sampling conditions.
The maximum sensitivity of the enthalpy sensor corresponds to
the case in which the sampled gas is cooled down in the measuring
tube to the temperature of its walls. In this case, the conditions of heat
exchange between the gas and the tube walls and the nature of the
gas are determined only by the dimensions of the measuring channel.
The flow rate of the gas to the sensor should not exceed the local flow
rate of the flow in the sampling area 50 otherwise the structure of the
flow will be greatly disrupted. This is a very important restriction which
will be discussed later.
The maximum sensitivity of the enthalpy sensor is calculated from
the equation
hrv f Si
sm = , (14.28)
Qf
where h is the enthalpy of the gas in the examined flow; rv f is the

mass flow rate of the gas in the flow; S is the cross section area of
the orifice of the probe for gas sampling (S i = pd i2/4); Q f is the to-
tal heat flow to the sensor.
The value of s m for every specific sensor can be calculated if we
know its geometrical dimensions and the parameters of the high-tem-
perature gas. As an example, Fig.14.24 shows the calculated depend-
ences of the maximum sensitivity of the transverse enthalpy sensor on
temperature for plasma flows of argon and oxygen. Calculations were
carried out for a number of fixed values of the linear flow speed of
382
Fig.14.24 Dependence of maximum sensitivity of the enthalpy sensor on gas temperature

in the flow. Argon, outer diameter of the sensor 5 mm, gas flow speed 20 (1), 100 (2),
1000 m/s (3), 4) v f = v 0 (T f/T 0 ), T 0 = 1000 K, v 0 = 10 m/s.
the flow (10; 100 and 1000 m/s). The total heat flow to the sensor was
calculated from the criterial dependence
Q f = Q + hrv f Si ,
LL O
MMN b g OPPQ 1 + (Le
0 .2
l (rm ) f Dh (14.29)
Q = 0.5 Re 0.5 Pr 0.4 0 .52
- 1) pdl,
d rm ) c Cp
where Q f is the total heat flow received by the sensor under the con-
ditions with the gas sampling; l, r, µ is the thermal conductivity, density
and viscosity of the plasma, respectively; d is the outer diameter of
the sensor sheath, and l is the length of the sensor counted from the
measuring orifice. Calculations were carried out per unit length of the
sensor l = 1 cm for a diameter of 0.5 cm. The diameter of the measuring
orifice was assumed to be equal to 1 mm.
As indicated by Fig.14.24, the sensitivity of the sensor decreases
with increasing temperature, whereas it increases with increasing flow
speed. Curve 4 was calculated under conditions in which the increase
of the flow speed is directly proportional to the increase of the plasma
temperature. Similar dependences have been recorded for oxygen, 22
nitrogen and air. In the temperature range above 8000 K the value of
sm is close to or lower than 10%, which is already unsuitable from the
practical viewpoint or requires using high-accuracy measuring appa-
ratus.
383
Plasma Diagnostics
Thus, the enthalpy probe can be used efficiently at temperatures

below 8000 K and relatively high flow speeds of the high-temperature
gas.
The error of enthalpy measurements. This error can be evaluated
by two methods. Firstly, by comparing the mean mass enthalpy of the
flow produced by integration of spatial distributions of the enthalpy in
the flow measured using the probe, and the values determined from
the thermal balance of the plasma flow generator. Secondly, by com-
paring the enthalpy measured by the probe with that determined by
another method.
The authors of Ref.35, 42 and 47 analyse in detail the factors ef-
fecting the measurement error. For example, for a stationary Grey probe
the rms error of enthalpy measurement is ±4 and ±6% for enthalpy
probes with screens (the so-called four-barrel and six-barrel probes).
The measurement error can be reduced by reducing the external
dimensions of the enthalpy flow and by using flow rates of the gas
samples not interfering the flow.
The authors of Ref.49 and 50 evaluated the error of measuring the
gas enthalpy in the flow in relation to the ratio between the unperturbed
flow speed and the speed of the sucked-away gas taking into account
the design dimensions of the sensor. Measurements were carried out
in a free plasma jet of hf discharge excited using a metallic split chamber
with a nozzle with a diameter of 1.4 cm placed in the output part of
the chamber 50 (Fig.14.25). Measurements were taken with a longitu-
dinal enthalpy probe with an outer diameter of 5 mm and with meas-
uring orifices of diameters d i = 0.4; 1.0; 1.6; 2.2 mm. Figure 14.25
shows the calibration curve constructed for a longitudinal enthalpy probe
(h ¥ is the enthalpy of the unperturbed plasma flow; h i is the enthalpy
measured by the sensor; a i = G i/G i¥, where G i = rv ipd i 2/4 is the gas
di2
flow rate of the sucked-away gas, and Gi¥ = rv¥ p is the mass flow
4
rate of the unperturbed plasma flow through the section equal to the
inlet section of the measuring tube). The flow rate of the sucked-away
gas was G i = 0.032 g/s and was maintained constant to reduce the
measurement error, i.e. a i was varied by varying the plasma flow pa-
rameters.
Figure 14.25 shows that v i should not greatly exceed the speed of
the non-perturbed plasma flow. In Ref.28, it was shown that at
v i/v i∞ ¾ 10 the perturbation introduced by the probe propagates up-
wards along the flow to the distance of almost 3d i.
In sampling the gas through the measuring orifice the current lines
are divided into two parts. One part of the flow travels into the measuring
384
(h ∞ h i )/h ∞
αi
Fig.14.25 Calibrating curve of the enthalpy probe. d i = 0.4 (1), 1.0 (2), 1.6 (3), 2.2
mm (4).
tube through which the gas was sampled, the other part flows around
the probe. The current lines are separated at specific points on the
surface of the nose part of the probe, and only the normal component
of the speed varies from zero here. These points on the surface of
the asymmetric enthalpy probe are situated in the top part of the probe
around the circumference55 which at a i = 1 coincides with the circum-
ference of the inlet orifice or the measuring tube. As indicated by
Fig.14.25, the diameter of this critical circumference can exceed the
diameter of the measuring orifice by no more than a factor of 2.
Since the data are limited, it is difficult to analyse the errors. At
present, it is usually possible to carry out only the analysis described
previously it is assumed that the error of enthalpy measurement is not
less than 4%.
14.4 Conclusions
The enthalpy probes can be used for sufficiently accurate and effi-
cient determination of the enthalpy of high-temperature gas flows. The
methods of measurement and the design of the enthalpy probes are
greatly varied. The development of measuring complexes and automated
systems may greatly expand the possibilities of using enthalpy probes
and sensors for enthalpy measurements.
385
Plasma Diagnostics
Chapter 15
PROBE MEASUREMENTS OF HEAT FLOWS IN

PLASMA JETS
15.1 Measurement procedure

Heat flows in a plasma jet are measured by the well-known calorimetric
method. 1 In this method, the heat flow is evaluated on the basis of the
rate of variation of the resistance of the wire introduced into the plasma
flow.
The following disadvantages were revealed in the process of meas-
urement of the heat flows in the plasma jet by this method. Loop os-
cilloscopes do not make it possible to measure with sufficient accu-
racy the variation of the resistance of the wire after its insertion into
the jet. Consequently, the wire fractures after each insertion and the
experiment time is very long due to the need to calibrate the system
and insert the next wire. In addition, movement through the high-tem-
perature medium and vibrations of the wire at the moment of arrest
greatly exceed the systematic error of the method. Therefore, to carry
out the local measurements, the method was improved so it was possible
to measure the heat flow on to the wire which intersect the plasma
jet without failure. The direction of wire entry was selected in such
a manner that the integral heat flow was measured in the fixed sec-
tion of the plasma jet along a chord at distance x from the axis. The
modified method greatly increased the speed and reliability of the meas-
urement.
The principal diagram of equipment is shown in Fig.15.1. The sensor
of the heat flow 2 – tungsten wire 0.1 mm in diameter and 10 mm long
- is secured in thin metallic holders 3 and placed on a long (~1 m) arm
of the rotating system (rod 5). During rotation of the rods the wire in-
tersects, at a linear speed of 4 m/s, the plasma flow along the chord
situated at distance x from the jet axis (Fig.15.2). Changing the mu-
tual position of the wire and the plasma torch, it is possible to meas-
ure the heat flows at different distances from the axis and at differ-
ent sections of the plasma jet. The wire is connected to the measur-
386
Probe Measurements of Heat Flows in Plasma Jets
Fig.15.1 Diagram of equipment for measuring heat flows. 1) plasma jet, 2) tungsten
wire, 3) holder contacts for moving contact of constant resistance, 5) rotating rod.
Fig.15.2 Diagram of the cross-section of the jet of plasma intersected by the probe
along a cord.
ing system (Fig.15.1) through the moving contact of contact resistance

4. The duration of its passage through the plasma jet is 0.005 s; the
variation of resistance in heating the plasma is recorded by the meas-
uring system. This system consists of a Winston bridge with one of
the arms containing the measuring wire. The diagonal of the bridge is
connected to a loop oscilloscope (type H-102). The bridge is powered
by a battery of accumulators with a total voltage of 25 V. Rheostat
R 5 is used to measure the sensitivity of the bridge system. Prior to
measurement, to control balancing, a milliammeter is connected to the
diagonal of the bridge instead of the oscilloscope. The bridge is bal-
anced by the variable resistor R 4. During the period of increase of the
main signal (Fig.15.3) the voltage drop in the wire is recorded as a
result of an inherent electrical field of the jet.2 The oscillogram shows
387
Plasma Diagnostics
a b c
Fig.15.3 Oscillograms of changes of the resistance of wire recorded at different sensitivity

of the measuring system at distances of 2 (a), 15 (b) and 30 mm (c) from the outlet of
the plasma torch.
clearly this additional signal superimposed on the main signal. Special

measurements show that heating of the wire, caused by the electric
field of the jet, is negligible and can be ignored.
To determine the relationship of the local heat flows with the re-
sistance of the wire we shall examine the cross section of the plasma
jet in which the wire is situated at a distance x from the axis (Fig.15.2).
The heat flow to the section of the wire with the length dL during period
dt is equal to q(t)dLdt, where q(t) is the heat flow per unit length and
unit time. The total heat flow on the section of the wire with length
dL during the period of intersection of the plasma is
t0
∫ q(t )dLdt.
0
The total amount of heat received by the entire wire at a distance

x from the jet axis is
L t0 t0
D θ( x) = ∫∫ q(t )dLdt = L∫ q(t )dt,

0 0 0
where dt = dy/v, v is the speed of the wire. At t = 0 y = –y 0, at t =

t 0 y = + y 0 and, consequently,
y
L 0
θ( x) =
v −y ∫
q( y)dy. (15.1)
0
388
On the other hand, the total amount of heat received by the wire
can be determined as follows
q( x ) = mCDT ( x ), (15.2)
where m is the mass of the wire, m = rSL, where r, S, L is the density

of tungsten and the cross section of length along the wire, respectively;
DT (x) is the variation of the wire temperature intersecting the given
cross section at the distance x from the jet axis; C is the heat capacity
of tungsten. Using the well-known dependence of the variation of the
wire resistance with temperature
R ( x ) = R0 [1 + bDT ( x) ],
we obtain
DR ( x )
DT ( x ) = , (15.3)
bR0
where R 0 is the initial resistance of the wire, DR (x) is the variation

of the resistance of the wire during holding in the plasma jet, b is a
temperature coefficient of the resistance of tungsten. Equating the right
hand part of the equation (15.2) and (15.1) and taking (15.3) into account,
we obtain
y0
ρCSv ∆R( x)
∫ q( y)dy =
− y0
β R0
. (15.4)
The experiments were carried out to determine DR (x) and the data
was then used to determine the value in the right hand part of the equal-
ity since b, C, r, S and v are known. Using Abel’s integral transfor-
mation for (15.4), we determine the radial distribution of the heat flow
in the examined cross section of the plasma jet per unit length of the
wire for 1 second. Thus, to determine the heat flow, it is necessary
to record during measurements only the maximum value of the resistance
of the wire during its passage through a specific chord of the exam-
ined cross section of the plasma jet. The absolute value of the vari-
ation of the wire resistance DR (x) is determined using a calibration
curve from the maximum amplitude of deviation of the loop. The re-
389
Plasma Diagnostics
cording system is calibrated using a cartridge with resistances connected

instead of the wire.
In this method, the error of measuring the heat flow consists of the
random error, determined by the scatter of the measured values of the
amplitude of the deviation of the loop and the systematic errors of the
method. The relative error measurements of deviation of the loop is
~2%, in the axial region and ~7% in the peripheral region of the jet.
Since the wire is heated only during its passage through the plasma
jet, there is some period during which part of the wire is still in the
wire and is heated up, whereas the other part, leaving the plasma jet,
already starts to cool down. Consequently, we record the reduced value
of the variation of resistance ∆R. For estimates, we shall assume that
time is equal to L/v, where L and v is the length and linear speed of
the wire, respectively, i.e. for the case in which the entire wire left
the jet and cooled down. The curve shown in Fig.15.3 shows that the
decrease of the amplitude of the loop during this period is ~30% of
the maximum deviation. Evidently, this value should also be regarded
as the maximum estimate of the systematic error typical of the given
method of measuring the heat flows. The error of measuring the heat
flow taking into account the reproducibility of the results and the sys-
tematic error of the method is ~5% in the axial and 10% in the pe-
ripheral regions of the jet. It should be note that, changing the sen-
sitivity of the measuring system, the method can be used to examine
not only plasma flows but also flames.
The results of measurements in the form of the lines of a constant
heat flow are presented in Fig.15.4. The numerical solution of Abel’s
integral was carried out in a computer. As shown in Fig.15.4, the transfer
of heat to the wire takes place mainly in the axial direction and is in
good agreement with the assumptions on the laminar nature of the
plasma flow at the gas flow rate shown in the figure. In the region
away from the nozzle outlet of the plasma torch the distribution of the
heat flow widens as a result of rapid mixing of the jet with a surround-
ing gas and due to the formation of turbulence.
15.2 Determination of plasma temperature

The spectroscopic methods characterised by high accuracy can be usu-
ally used to take measurements of temperature only in the near-axial
part of the plasma flow. The distribution of temperature over the entire
cross section of the plasma jet can be determined by the calorimet-
ric method 3 from the measured values of the heat flow and the flow
speed of the plasma. The heat flow to the solid introduced into the
plasma flow is determined by the equation 4
390
Fig.15.4 Distribution of heat flow on a wire 0.1 mm in diameter in a plasma jet at an

argon flow rate of 0.18 g/s. q = 3500 (1), 3000 (2), 2500 (3), 2000 (4), 1500 (5),
1000 (6), 500 (7), 300 (8), 100 (9), 50 (10), 20 (11), 0 cal/(cm 2 s) (12).
H ¢¢ - H ¢
q=a , (15.5)
Cp
where q is the heat flow, cal/(cm 2 ×s); H²–H¢ is the change of the
plasma enthalpy when cooling down to the temperature of the sensor
wall; C p is the heat capacity of the plasma at constant pressure; a is
the heat transfer coefficient which in transverse flow-around the cylinder
is calculated from the equation 3,5
0.25
κ  Pr′′ 
α = 0.86 Re0.41 Pr 0.35   , (15.6)
d  Pr′ 
391
Plasma Diagnostics
mC p
where Re = rvd/µ is the Reynolds number; Pr = is the Prandtl
k
number; d is the diameter of the cylinder with flow-around (wire); k,
r, µ, H, C p, v is respectively the heat conductivity, density, viscosity,
enthalpy, heat capacity and the flow speed of the plasma. The values
denoted by a single prime correspond to the gas temperature equal to
the temperature of the wire, those denoted by two primes correspond
to the temperature of the incident flow. The quantities without the primes
were calculated at temperatures specifying the heat exchange in the
boundary layer at the wire. This temperature can be represented by
the plasma temperature (incident flow), the mean arithmetic tempera-
ture, the temperature corresponding to the mean enthalpy of the gas
in the boundary layer. The calculation results show that the selection
of the temperature does not have any significant effect on the calculated
plasma temperature (maximum difference with the temperature measured
spectroscopically 2 is 10%). The controlling temperature here was the
temperature corresponding to the mean enthalpy of the gas in the
boundary layer. The plasma temperature values obtained in this case
occupy an intermediate position in relation to those calculated at other
controlling temperatures.
The equations (15.5), (15.6) link the heat flow on to the introduced
solid with the speed and temperature of the plasma jet. Assuming that
the plasma characteristics k, r, H, µ, C p are known temperature func-
tions, equation (15.5) can be written in the following convenient form
qv -0.41 = f (T ). (15.7)
The left part of equation (15.7) includes directly measured quan-

tities whereas the right hand part depends only on temperature. To
calculate f (T) (Fig.15.5), the values of the thermophysical parameters
of argon plasma were taken from Ref.5–8. To determine the plasma
temperature in the near-axial part of the jet we use the results of
velocity measurements.2 The determined temperature agrees in the range
5% with the temperature measured in spectroscopic examination
(Fig.15.6) which indicate that there are no large systematic errors in
the calorimetric method.
Estimates show that the error of calorimetric measurements of the
temperature is associated with the error of measuring the quantities
q and v for the near-axial part of the plasma jet from the relationship
dT » 0.35dq + 0.14dv
392
and equals 3−4%. However, the main error of the method is caused
by the well-known scatter of the values of the thermophysical param-
eters of the plasma (µ, H, C p) and equals approximately 10%.
In the peripheral part of the jet the flow speed was not measured
so that to estimate the temperature in this region on the basis of the
measured heat flows we accepted a parabolic radial velocity distribution.
It is evident that this assumption makes the results of temperature
calculations for the jet periphery (Fig.15.6) less reliable in compari-
son with those for the axis. However, it should be noted that the in-
determinacy in the velocity has a relatively weak effect on the tem-
perature measured by the calorimetric method. Figure 15.5 shows that
dT ~ 0.24dv at a temperature of 1000–8000 K, dT ~ 0.41dv at
8000–10000 K and dT ~ 0.14dv at 10000−14000 K.
Fig.15.5 Function f(T) = qv –0.41 for the case of flow around of the cylinder by the
plasma flow of atmospheric pressure of argon in the LT condition taking into account
the transfer of heat by the atoms, ions and electrons (1) and by atoms and ions only
(2). Wire diameter 0.1 mm, q in cal/cm 2 s, v in cm/s.
Fig.15.6 (right) Radial distribution of spectroscopically (I) and calorimetrically (II)
measured temperature in a plasma jet at distances of 2 (1), 15 (2) and 30 mm (3) from
the outlet of the plasma torch.
393
Plasma Diagnostics
Part 3
MACROSCOPIC DIAGNOSTICS
Chapter 16
METHODS OF EXAMINING THE SPATIAL

STRUCTURE OF RADIO-FREQUENCY CAPACITANCE
DISCHARGES
16.1 Introduction
The radio-frequency capacitance discharge (RFCD) is one of the sim-
ple, reliable and efficient sources of plasma in various regions, for ex-
ample, in plasma chemistry, 1–3 plasma technology, 4–7 and the technol-
ogy of gas lasers, 8–11 and others. Therefore, a large number of pub-
lications concerned with the methods of producing, examining and using
RFCD in a wide range of gas pressures from fractions to hundreds
of torr and frequencies of the RF field from 10 6 to 10 8 Hz is not sur-
prising.
It is well known that the plasma of RFCD in these frequency and
gas pressure ranges may be highly non-equilibrium, capable of filling
large volumes at high specific energy inputs (100 W/cm 3 and higher),
the possibility of controlling these parameters by varying the frequency
of the RF field and the conditions of burning of RFCD.
Interesting results in understanding the mechanism of sustaining sta-
tionary RFCD and in its numerical applications have been obtained by
the discovery of the important role of the near-electrode layers of the
spatial charge (NLSC) and of the secondary emission processes taking
place in them. It was found that there are various forms of the ra-
dio-frequency capacitance discharge which are characterised by the
processes taking place in the NLSC and, for this reason, they have
394
Methods of Examining Spatial Structure of RF Capacitance Discharges
qualitative different spatial structures under the same external conditions:

gas pressure and type, the size of the interelectrode gap and, in cer-
tain cases, at the same RF voltage of the electrodes. In particular, the
density of discharge current in transition from one form of discharge
to another may change by an order of magnitude or more, 12 and the
strength of the electric field and the electron concentration change not
only quantitatively but, more interestingly, their distribution in the
interelectrode gap becomes qualitatively different. 13
These facts indicate that it is promising to develop methods of ob-
taining, in a radio-frequency capacitance discharge, plasma with the
required properties. However, to apply this method it is necessary to
have clear physical assumptions of the mechanism of formation of the
spatial–time structure of each form of the stationary RF capacitance
discharge and have data on the regions of their existence and reasons
for the transition from one form of discharge to another.
In this section, we shall discuss the experimental methods of ex-
amining the structure of RF capacitance discharge which make it pos-
sible to describe in a simple manner and, at the same time, reliably the
specific features of the RF capacitance discharge and the difference
between its forms, obtain information on the evolution of the spatial
structure of the discharge with the change of the external conditions
and the conditions of stationary burning and, consequently, create suitable
conditions for justified selection of a specific form of RFCD when
solving specific problems in practice. This is especially important at
medium pressures where the free path of the electron is considerably
shorter than the amplitudes of its drift vibrations in the plasma sus-
tained by the RF field.
16.2 Specific features of producing, examining and using

radio-frequency capacitance discharge plasma. Forms of
existence of RFCD
As mentioned in the introduction, the plasma devices based on RFCD
are used widely in technology. This is due to the appearance at high
frequencies (f ~ 1 ¸ 100 MHz) of a number of specific features by
which the RFCD differs from direct current discharges, both arc and
glow. A special feature of plasma generators based on RFCD is that
the electrodes are placed outside the discharge chamber. In this case,
the active current, flowing through the plasma, is closed on the elec-
trodes by the bias current. The maximum density of RF current which
can be passed through the dielectric without disrupting it is determined
by the equation
395
Plasma Diagnostics
¶Ebr
jmax £ e d e 0 = e d e 0wEbr , (16.1)
¶t
where e 0 = 8.85·10 –12 F/m; e d is relative dielectric permittivity; E br is

the electric field strength in the dielectric at which its breakdown takes
place; w = 2pf is the RF field frequency. For typical dielectric ma-
terials E br ~ 10 5 V/cm; e d ~ 5. Consequently, the limiting current den-
sities in the frequency range f = 1 ¸ 100 MHz equal 0.5–50 A/cm 2
which is fully sufficient for many applications. Therefore, RF discharges
with current densities in plasma varying from units to hundreds of
millimetres per square centimetre are used in most cases. 1–13
This special feature of RFCD, i.e. the possibility of closing the dis-
charge current by bias currents in the dielectric, coating the electrodes,
has supported the view according to which in these discharges the proc-
esses of the electrodes and the near-electrode regions do not play any
significant role.14,15 Therefore, the radio-frequency capacitance discharge
was regarded, especially in the pressure range p >> 1 torr, as the sim-
plest because of its spatial structure. It was assumed that the stationary
retention of the plasma in RFCD is ensured at every point of the
interelectrode gap by the local ionisation balance in accordance with
the equation
n i ( E pl / N ) = n dif + bne + n p , (16.2)
where n i (E pl /N) are the frequencies of generation (ionisation) and an-

nihilation ndif, bne, np (diffusion, recombination, sticking) of the charged
particles, E pl is the strength of the electric field in plasma, N is the
concentration of the neutral gas particles, b is the recombination factor.
Specifying the value of the electron concentration ne, equation (16.2)
can be used to determine E pl if the mechanism of annihilation of the
charge is known. It is important to stress that the value of n e and the
density of discharge current j (which depends on n e) is linked with the
parameters of the external circuit, assuming
Irf
j = e ne m e E pl = = const, (16.3)
S
where I rf is the intensity of discharge current given by the external

circuit, S is the electrode area, µ e, e is the mobility and charge of the
electron. The contribution of the reactive component of current in plasma
is ignored in (16.3).
396
However, although the equations (16.2) and (16.3) are simple and
obvious, they contradict the following experimental facts established
in Ref.12.
1. At some values of the discharge current intensity I rf the area of
the electrode S pl filled by the discharge plasma is smaller than S. With
increasing I rf S pl also increases and since
Irf
jn1 = = const, (16.4)
S pl
i.e. RFCD not only fills completely the interelectrode gap in the di-
rection transverse to the current but is also characterised by the so-
called effects of normal current density – effect in which the current
density does not depend on the total current flowing through the dis-
charge. Specific value of j n1 is determined by the type and pressure
of the gas p, frequency ω and the width of the interelectrode gap d,16,17
and is almost completely independent of the electrode material.
2. If j > j cr, another form of RFCD appears in which the effect of
the normal current density also operates but with different current
density jn2 (j n2 > jcr > j n1, with other conditions being equal). The elec-
trode material or its dielectric coating influences the values of j cr and
j n2.
3. In the interelectrode gap filled with the RFCD plasma zones with
very low E pl appear at j > j cr. At these values equation (16.2) is not
fulfilled.
4. In addition, the RF discharge always contains near-electrode layers
of the spatial charge (NLSC) (which separate the plasma from the
electrode surface), and the processes in these layers are not described
by (16.2) and can differ qualitatively depending on the RF voltage on
the NLSC.
Another very important special feature of the RF capacitance dis-
charge, established in Ref.12, is the restriction of the region of existence
of the form of the RFCD with respect to pressure and the width of
the interelectrode gap d, with current density j satisfying the relationship
jn1 £ j £ jcr . (16.5)
Following Ref.12, this spatial form of the RFCD will be referred

to as low-current, and the RF discharge with a current density j > j cr
as the high-current discharge.
The difference between these two forms of the RFCD is mainly that
397
Plasma Diagnostics
Fig.16.1 Photographs of low-current (a) and high-current (b) RFCD. Frequency 13.6
MHz, air, p = 10 torr, d = 2 cm, U rf = 320 V, j = 7 (a), 120 mA/cm 2 (b).
I 0 (x), rel. units
d, cm
Fig.16.2 Distribution of glow through the RFCD. Frequency 13.6 MHz, air, p = 10
torr, 1) low-current, 2) high-current discharge.
there are qualitative differences in the distributions of the concentration

of charged particles and the strength of the electric field along the
direction of current passage. Visually, this is reflected in the transfor-
mation of the nature of distribution of the intensity of integral radia-
tion of the RFCD in the visible wavelength range (Fig.16.1) when the
discharge conditions are changed. Attention should be given to the sharp
increase of current density (from 7 to 120 mA/cm 2) and the variation
398
of the distribution of glow intensity I 0 (x), especially the appearance

of dark regions which separate the plasma column in the centre of the
discharge gap from brightly glowing near-electrode zones (Fig.16.2).
In addition, the spectral composition of radiation of the near-electrode
regions also changes: the radiation spectrum in the high-current dis-
charge is close to the radiation spectrum of the negative glow of the
cathode region of the dc glow discharge. The agreement between the
distribution of the glow of the high-current RFCD and the distribution
of glow in the dc glow discharge has been known for some time.
Nevertheless, it was assumed that the processes taking place on the
cathode which lead to the formation of a large number of electrons
are not important for the radio-frequency discharge because the phe-
nomena in both electrodes of RFCD are in this case completely sym-
metric, and the required number of free electrons forms at each electrode
by ionisation of gas particles by collisions of the first kind with the
electrons moving to both sides. 14
The initial information on the possible effect of secondary emission
electrons, formed in near-electrode regions as a result of bombardment
of the electrode surface with positive ions, on the structure of RFCD
appeared in Ref.18. This conclusion was made on the basis of indi-
rect data: similar glow in the cathode regions of the glow discharge
and near-electrode zones of the RFCD, sputtering of the electrode
material and a high constant potential of the plasma in relation to the
electrodes U 0 which reaches hundreds of volts at pressures of p < 1
torr.
An important support of the hypothesis in Ref.18 on the effect of
secondary emission processes on the characteristics of the near-electrode
layers of the spatial discharge and of the entire discharge as a whole
is the existence of high values of U 0 which greatly exceed (by an order
of magnitude or more) the ‘floating’ potential of the plasma correspond-
ing to typical values of the electronic discharge temperature. The effect
of sputtering of the electrodes in the RFCD also indicates the pres-
ence of high constant potentials localising the NLSC. In fact, in the
discharge conditions at frequency f > 1 MHz even a hydrogen ion as-
sumes the energy of the ordered oscillatory motion in the RF field which
does not exceed several electron volts. This is obviously insufficient
for explaining the phenomenon of sputtering of the material on elec-
trodes in RFCD.
Irrespective of the logical nature of the assumption 18 on the role
of secondary-emission processes in RFCD, this approach has not been
developed further. There are several reasons for this. Firstly, the data
on the probe measurements of U 0 presented in Ref.18 and 19 indicate
a large decrease of U 0 (from hundreds to units of volts) with increasing
399
Plasma Diagnostics
gas pressure in the discharge above 1 torr, i.e. they indicate the
disappearance of NLSC with increasing pressure. Secondly, the fact
that the phenomena taking place in RFCD with hollow electrodes and
electrodes coated with the dielectrics are identical casts doubts on the
effect of secondary-emission processes on the conditions of burning
of the RF capacitance discharge. Thirdly, at pressures p < 1 torr the
second-emission electrons (g-electrons) have only a slight effect on the
parameter of the NLSC (especially in inert gases). This will be
examined in greater detail later. Here it should be noted that this situ-
ation forms due to the small number of inelastic collisions of
g-electrons with neutral particles on the characteristic thickness of the
NLSC equal to d l . This means that the near-electrode layers in the
low-pressure RFCD behave at high U 0 as sources of fast electrons
which can influence the ionisation balance in the discharge only out-
side the limits of the layers. 20–22 In other words, at low pressures the
transition to the high-current conditions of RF capacitance discharge
is smooth, is difficult to record and this is the reason why it has not
been noted for a very long period of time.
Recently, the assumptions regarding the nature of the RF capaci-
tance discharge in a wide range of pressures and frequencies of the
RF field have been explained further. Firstly, it was shown in Ref.23
that a high constant potential of RFCD plasma in relation to the electrons
(above 100 V), localised in NLSC, is also observed at pressures p >>
1 torr. It is assumed that near-electrode layers of a spatial discharge
must form unavoidably in RFCD whereas when the pressure is increased
the main reason for their appearance is not eliminated – the need for
the equalisation, during the cycle of the RF field, of the number of posi-
tive ions and electrons of greatly different mobilities falling on the
surface of the electrodes under the effect of the alternating field fre-
quency. Secondly, the results show 12 that the existence of various forms
of the RFCD and of transitions between them does not depend on
whether the electrodes are coated with the dielectric or not. Thirdly,
direct experiments 12,24 show that in transition to the high-current form
the active conductivity of NLSC increases to the values comparable
with the conductivity of the cathode region of dc glow discharge un-
der the same experiment conditions. This means that the transition to
the high-current burning regime can be qualified as a breakdown of
capacitance (non-conducting) near-electrode RFCD layers with par-
ticipation of secondary emission electrons where the Townsend crite-
rion 12
400
 dl

 0
∫
γ exp α( El / N )dx − 1 = 1,

(16.6)
is fulfilled at a specific RF voltage on the electrodes in the NLSC with

thickness d l. Here γ is the coefficient of secondary emission of elec-
trons from the surface of the electrode or of the dielectric coating on
it; α (E l/N) is the volume ionisation coefficient, E l is the strength of
the electric field in the layer. In other words, when (16.6) is fulfilled,
the balance of the charged particles in the NLSC of RFCD is facili-
tated by the ionisation multiplication of g-electrons within the limits of
the layers of the special discharge whereas in the low-current form
of the discharge (is the left part of (16.6) is less than unity) ionisa-
tion is significant only in plasma.
For practical applications it is very important that the spatial structure
in the high-current form of the RFCD is qualitatively different, the active
component of the current in NLSC rapidly increases and at pressures
of tens of torr it usually exceeds the capacitance component, 25 thus
resulting in a rapid increase of the ion flux on the surface of the elec-
trodes, failure of the electrodes and generation of a large amount of
power in the NLSC. If the actual situation in the near-electrode lay-
ers is not taken into account, the results can be unexpected. For ex-
ample, if it is assumed that if the RF electrodes are placed outside the
limits of the discharge chamber, it may be expected that sterile plasma
may be produced. This assumption holds only for the low-current form.
If the pressure, interelectrode spacing and frequency of the RF field
are not correctly selected, the chamber contains RFCD only in the high-
current form and, consequently, the discharge volume will be contami-
nated with the products of sputtering of the walls of the chamber con-
nected with the electrodes. Another example is taken from laser tech-
nology. In 1974 it was attempted to use a RF capacitance discharge,
transverse in relation to the optical axis of the resonator, for pump-
ing a stationary CO 2 laser. 26 The results were unsatisfactory. Never-
theless, at present CO 2 lasers with RF excitation are superior in their
group as regards the specific characteristics 8–11,13 due to the rational
application of the RFCD taking into account the special features of
its spatial structure, although the main elements of the design have not
been changed.
The experimental data on the structure of RFCD and the forms of
its existence have also been confirmed by numerical calculations pre-
sented in Ref.27–29. It is important to note Ref.28 and 29 where the
existence of two forms of RF discharge and of transition between them,
401
Plasma Diagnostics
including under the conditions in which electrodes were coated with

the dielectric, was shown for the first time by numerical modelling.
16.3 Interpretation of volt–ampere characteristics of RFCD

and their relationship with the spatial structure of the
discharge
The volt–ampere characteristics of the radio-frequency capacitance dis-
charge on the whole, i.e. the dependence of the of the RF voltage on
the electrodes U RF on the discharge current I RF or its density j RF, pro-
vide qualitative information on the structural special features of the
RFCD with the variation of I RF (or j RF).
We shall examine static VACs of medium-pressure RFCD together
with the distribution of glow along the direction of the field. Figure 16.3
shows the most general case of the dependence U RF = F 1 (I RF ), and
Fig.16.4 U RF = F 2 (j RF ) obtained in examining a stationary RF capaci-
tance discharge at medium pressure between flat cooled electrodes. 12,30
It can be seen that the VAC contains five different sections: 1) OA
(Oa in Fig.16.4), 2) ABC (ab), 3) CD (bc), 4) DEF (cd) and 5) FG
(df). The first section – interelectrode gap is not ‘ruptured’, the voltage
of the electrodes linearly increases with the increasing I RF . One U RF
becomes equal to the breakdown voltage (points A, a) a discharge forms
whose plasma does not fill the interelectrode gap completely in the
direction normal to the current. The voltage of the electrodes decreases
when discharge appears (section AB, ab). Increase of IRF only increases
the transverse dimension of the glowing discharge zone. It is impor-
tant to note that the structure of the glow along the direction of the
current passage and the value j RF do not change in this case (section
BC in Fig.16.3, point b in Fig.16.4).
The third section of the VAC has a positive derivative with respect
to IRF, jRF. It is usually realised in final filling of the discharge gap with
plasma in the direction normal transverse to the current. The nature
U RF URF
Ubr
U br
U per
U per U min2
U min2 U min1
U min1
j n1 j br j n2 j RF
I RF
Fig.16.3 Typical volt–ampere characteristic of RFCD in coordinates U rf (I rf) (transition

from D is possible to any point EF).
Fig.16.4 Typical volt–ampere characteristic of RFCD in coordinates U rf (j rf).
402
and distribution of the glow along the current lines are here on the whole
the same as in the second section of the VAC; there is only an in-
crease of the total intensity of light radiation, especially at the ends
of the plasma column. The glow in this case is again homogeneous along
the field, with the exception of the several bright interelectrode regions
separated from the electrodes by dark zones with a characteristic di-
mension depending on the nature of the gas and the frequency of the
RF field.
Measurements of the conductivity of NLSC by the method which
will be discussed later showed that the active conductivity of RFCD
layers in the examined section of the VAC (BCD, bc) is not high and
its order of magnitude is close to that of the conductivity determined
by the ionic saturation current with the ion concentration n+ @ ne, where
n e is the concentration of electrons in the discharge plasma.
It should be noted that under the real conditions after the appear-
ance of a discharge the value U RF is not necessarily minimum. This
is determined by the parameters of the external circuit, and the matching
circuit of the RF generator with the discharge chamber because when
plasma appears the total impedance of the discharge gap greatly
changes. Consequently, matching of the RF generator with the discharge
chamber can both improve and become less efficient. It is possible that
after the breakdown of the interelectrode gap an anomalous regime of
low-current RFCD is established there (portions CD, bc). The normal
regime of the RF discharge is obtained by regulation in the external
circuit resulting in a decrease of U RF .
When the voltage of the electrode reaches the value U per (point D,
c) an inflection point or jump is observed. The jump usually occurs in
discharge on molecular gases. In inert gases, for example, in helium,
a distinctive transition to another regime in the VAC is visible only at
relatively high pressures (p ~ 100 torr). The inflection point or jump
in the VAC is accompanied by the distribution of the glow along the
direction of current passage, and the new structure of the glow can
be compared with two normal dc glow discharges which are distrib-
uted in such a manner that the common positive column is in the centre
between the electrodes and separated from them by regions identical
with the Faraday dark space and the negative glow (see Fig.16.1b and
16.2).
The most important special feature of the examined RFCD regime
is the abrupt increase, in comparison with the previous type of discharge,
of the active conductivity of the near-electrode layers to character-
istic values corresponding to the conductivity of the cathode region of
dc glow discharge under the same conditions. Another important typical
feature of the RFCD after the jump on the VAC is the large increase
403
Plasma Diagnostics
of the discharge current density: for example, in air at p = 30 torr and

a frequency of 13.6 MHz it changes in transition to the new regime
from 12 to 240 mA/cm 2. The current density jump is associated not
only with the appearance of a significant active component of current
density ja but also with a rapid decrease, especially in molecular gases,
or the thickness of the NLSC d l which determines the density of bias
current at a fixed voltage in the layer U l
Ul
jcm = ee 0 wEl @ ee 0 w . (16.7)
dl
For example, if in RFCD in air in the regime to the left of the jump
on the VAC at p = 15 torr the value d l1 » 4 mm, measured from the
distribution U 0(x), then in the new regime it does not exceed 0.3 mm.12
This means that, according to equation (16.7), j cm in the regime after
the jump on the VAC increased 13 times. It is important to know that
the thickness of the layer in the new burning regime is almost inde-
pendent of the frequency of the RF field but depends strongly on pres-
sure
C1 T
d l2 @ = C1 , (16.8)
N pT0
where C 1 is a constant determined by the type of gas and electron ma-

terial, determined at T0 = 300 K; T is the gas temperature in the NLSC.
Comparison of the parameter of NLSC to the left and right of the
jump on the VAC indicates that the Townsend condition (16.6) is fulfilled
in the near-electrode layers of the RFCD corresponding to the points
D and c on the volt–ampere characteristics. This means that the VAC
U 0, V
Fig.16.5 Dependences of constant

potential of plasma in relation to
electrodes U 0 on U RF . Low-current
RFCD, frequency 13.6 MHz; 1,7)
helium, p = 75 and 150 torr,
respectively; 2) air, p = 7.5 torr; 3–
U RF , V
6) CO 2 , 7.5; 15; 30; 50 torr.
404
jump, observed in the experiment, is caused by the breakdown of the

NLSC with secondary emissions taking part in the process and by the
transition of the RFCD to the high-current regime, as already indicated
in the previous section.
The results of measuring the constant plasma potential in relation
to the electrodes U 0 in dependence on the value of the RF voltage U RF
applied to the electrodes, are presented in Fig.16.5 and indicate the
increase of U 0 with increasing U RF . They can be used as another ar-
gument in favour of the validity of the assumptions regarding the break-
down of the NLSC because the constant voltage U 0 is localised in the
layers. 12
Other experimental verifications of the conclusions of the breakdown
of the NLSC have also been made. In particular, investigations were
carried out into the effect of electrode material and the type of gas
on the transition voltage to the high-current regime.12 In addition to Uper,
the constant plasma potential U 0per at the moment of the jump of the
VAC was recorded. The results were compared with the data obtained
for the case of breakdown of the constant electric field 31,32 and showed
good agreement. As in the constant field, U per and U 0per are determined
by the combination ‘type of gas–electrode material’. As the breakdown
voltage in the constant field for the specific gas–electrode pair in-
creased, higher U per and U 0per were required for the transition of NLSC
of RFCD to the high-current regime. Analysing the transition of RFCD
to the high-current regime, it is necessary to select, as the parameter
pd, the value of pd l, i.e. examine not the size of the interelectrode gap
but the thickness of the NLSC d l1 of the low-current type of the RF
discharge.
Thus, the breakdown of NLSC of the low-current RFCD with sec-
ondary emission electrons taking part in it with the Townsend condi-
tion fulfilled in the layers restricts the upper permissible values of
current density that are realised in the volume form of this type of dis-
charge.
The expression for the limiting current density in the NLSC of the
low-current RFCD follows from the breakdown condition
U br ee 0w 2U br
jbr @ ee 0w @ , (16.9)
d l1 vdr
where U br is the voltage on the NLSC at which a jump is observed

on the VAC, v dr is the drift speed of the electrons in plasma. The ratio
d l1 ¾ v dr /w is used in this case. 12
405
Plasma Diagnostics
The new stable state of the NLSC of the RFCD in which the
Townsend conditions is satisfied is obtained at the thickness of lay-
ers satisfying the equation (16.8). In this case, the number of ionis-
ing collisions of the electrons with neutral particles in the layer reaches
the maximum value and the voltage in the layers is minimum. 13,32 This
means that the normal current effect (section EF) is realised in the
layers of the high-current RFCD. It is well known 12,25 that the active
component of the current density in NLSC with the Townsend crite-
rion fulfilled is determined by the type and density of gas and elec-
trode material
2
T 
ja = C2 N = C2  0  p 2 .
2
(16.10)
T 
Here C 2 is the tabulated constant, 31,32 which depends on the electrode

material and the type of gas, and is obtained at T = 300 K. In addi-
tion to j a, the NLSC carries the capacitance current, and the expression
for this current can be written in the form taking into account (16.8)
U l ee 0wp(T0 / T )
jcm = ee 0wE l @ ee 0w @ , (16.11)
dl2 C1
Taking into account that the phase shift between j a and j cm is equal
to p/2, we obtain the equation for the normal current density in the
NLSC of the high-current RFCD
2
 εε 0 ω U l 
2
T 
jn 2 = ja2 + jcm
2
= C2  0  p 2 1 +   . (16.12)
T   C1C 2 (T0 / T ) p 
If the current passed through the discharge is higher than IRF = jn2S,
the anomalous regime of the high-current RFCD is realised (sections
FG and df) in which the RF voltage in the electrodes and the NLSC
and the width of the regions of negative glow and the Faraday dark
space l increase and the size of the glowing plasma column along the
direction of d pl decreases. This is indicated by the dependence of l
on I RF obtained in examining RFCD in helium at a frequency of f =
13.6 MHz at p = 15 torr between flat electrodes with a distance be-
tween them being d = 60 mm (it is evident that d = 2l + d pl with d l2
ignored):
406
I RF , A 1 1.2 2 3 4 6 8
l, mm 7 7 10 12 16 20 26
It is evident that this phenomenon is associated with the ionisation

of the gas by the electron beam formed in the NLSC of the anoma-
lous high-current RFCD.
Another experiment was carried out to confirm the assumption on
the controlling role of the electron beam under the examined condi-
tions. A high-current RFCD was excited in the same discharge vol-
ume and under the same interelectrode spacing and the helium pres-
sure. However, in this case, the electrodes were ring-shaped to en-
sure that the electron beam, formed in the cathode layer, propagates
in the radial direction (instead of propagating along the discharge axis)
and cannot have any effect on the structure of the part of the plasma
column which is situated between the electrodes. The results show that
at the I RF values given above there was no reflection of the columns:
on the contrary, l decreased with increasing I RF . Similar phenomena
were observed in Ref.33. At very high current densities the high-current
RFCD collapsed to the arc regime with high electrode erosion. How-
ever, this region of the conditions was not examined in this work.
We shall now return to the BC section of the volt–ampere charac-
teristic of the low-current RFCD, i.e. the region of the effect of normal
current density for this discharge form. The key to understanding the
nature of the effect is the observed dependence of current density j n1
on electrode spacing d which indicates the more important role played
by the plasma column in the low-current RFCD than, for example, in
the high-current form where the positive column in a wide range of
the external conditions is ‘aligned’ under the current density determined
by the NLSC conditions.
A large amount of information on the dependence of the parameters
of the low-current RFCD on the length of the plasma column was pro-
vided by the experiments in Ref.34. The following procedure was used.
A thin dielectric sheet was placed in the interelectrode gap formed by
two flat electrodes with the area S. The dielectric sheet was placed
parallel to the working surface of the electrodes and could move in
the gap. The distance from one electrode to the sheet is denoted D 1,
to the other D 2. It is evident that D 1 + D 2 ¾ d. In ignition of RFCD
the single plasma column was divided into two successive discharges,
and a layer of spatial discharge whose parameter did not differ from
the corresponding NLSC, formed on both sides of the sheet. The re-
sults show that the transverse sections of the plasma columns and, con-
sequently, current densities in them j 1, j 2 coincide if D 1 = D 2. When
407
Plasma Diagnostics
D 1 ¹ D 2, the cross section is smaller and, consequently, higher current

density was observed in the plasma column with higher D, since
j 1 S 1 = j 2 S 2 = I RF .
To explain the effect of normal current density in the low-current
RFCD it is important to have information on the nature of the VAC
or the NLSC and the plasma column. 34 The presence in the electrode
gap of the RF capacitance discharge of zones with different current
passage conditions and self-organisation of stable stationary states in
the NLSC–plasma column system determine the properties of the RFCD
observed in the experiments. 34
Taking into account Ref.34, we shall examine in greater detail the
mechanism of formation of the spatial structure of the low-current form
of the RFCD.
After breakdown of the discharge gap, the different mobility of the
electronic and ionic components of the plasma during the process of
formation of RFCD leads unavoidably to the appearance of near-elec-
trode layers of spatial discharge at the boundaries of plasma with elec-
trons – this is an essential condition for generating stationary discharge
because the number of ions and electrons, hitting the electrodes dur-
ing the cycle of the RF field, can be the same only in the presence
of NLSC. However, of all the permissible stationary states of the NLSC-
plasma column system, only the states that are stable are realised in
the experiment. The stable regime is the stationary regime of burn-
ing of low-current RFCD where the small deviation of current den-
sity from the steady value decrease with time. In the opposite case,
the state of the NLSC–plasma column system is unstable and if no
special measures are taken this discharge burning regime cannot be
realised in practice. The effect of the normal current density in the
low-current RFCD on electrodes not coated with a dielectric mate-
rial, 12,16,17 and observed in experiments confirms the possibility of ex-
istence of stationary regimes of burning of the RF discharge stabilised
by NLSC. It is not possible to produce a stable low-current form with
a current density of j < j n1 on ‘bare’ electrodes. As shown later, the
formation of low-current RFCD with a current density of j > j n1 re-
quires higher U RF and, and if discharge does not fill the electrode gap
in the direction transverse to the current, is also forbidden as a re-
sult of disruption of the ionisation bands on the plasma–neutral gas lateral
boundary. An ionisation wave forms in this case and leads to move-
ment of the lateral boundary of plasma and an increase of the cross
section of discharge until the equality j = j n1 is fulfilled: in this case
U RF on the electrodes decreases to some value U min1 . This effect is
often observed in experiments in attempts to increase j above j n1 in
the normal burning regime. If the RF source operates in the voltage
408
U pl, 2U l , U RF , V d l , mm d, mm
d l , mm
2U l.n., V
p, torr
j, mA/cm 2
Fig.16.6 Dependences of d l ,(1) U l (2), U pl (3), U RF (4) on j. RFCD, air, f = 13.6 MHz,
p = 30 torr.
Fig.16.7 Low-current RFCD in air, f = 13.6 MHz. 1) U l.n (p) at d = 2 cm, 2) U l.n (d)
at p = 10 torr, 3) d l (p).
generator regime, i.e. U RF on the electrodes does not decrease dur-

ing plasma expansion, the plasma completely fills the gap between the
electrodes in the direction transverse to the current and the anoma-
lous burning regime is established.
To illustrate this situation, Fig.16.6 shows the VAC of NLSC U l (j),
plasma U pl (j), the discharge as a whole U RF (j) and the dependence
of the thickness of NLSC on j, dl (j) obtained in the low-current RFCD.
The behaviour of the RF voltage on the NLSC under the normal
burning regime, i.e. the minimum possible RF voltage on the NLSC Ul.n
with a change of the pressure in the discharge chamber p or the width
of the electrode gap d (Fig.16.7), as well interesting and important for
understanding the nature of the low-current RFCD. It can be seen that
the minimum RF voltage on the NLSC becomes higher with increas-
ing pressure and, which is completely unexpected, U l.n is a mon-
otonically increasing function of d. This behaviour of Ul.n with the vari-
ation of p, d explains the restriction of the region of existence of the
low-current form of the RFCD with respect to pressure and the size
of the electrode gap observed in the experiments. 12
In fact, as indicated by Fig.16.7, at a constant electrode gap there
is a value of p* at which the minimum possible RF voltage on the NLSC
can reach the value U br at which equation (16.6) is fulfilled and the
VAC shows an inflection point. This means that at given d and
p > p* the low-current form of an independent RFCD is not realised.
A similar situation is observed if p is fixed and d is varied.
Analysis of the data shown in Fig.16.6 and 16.7 indicates that two
409
Plasma Diagnostics
qualitatively different regions can easily be separated in the spatial struc-

ture of the low-current RFCD: NLSC and the plasma column which
differ by the mechanism of maintenance of the active conductivity in
them. If the active conductivity in the plasma column is ensured mainly
by ionisation in the RF field, then in NLSC it is sustained mainly as
a result of injection of charges from the ends of the plasma column.
The dependence U pl (j) follows from the standard equations of ioni-
sation balance in the plasma columns. 32 However, to simplify consid-
erations, we shall approximate the experimental values of U pl (j). The
validity of this approximation was verified in the pressure range from
units to hundreds of torrs:
C
U pl ( j ) = pd pl , (16.13)
ja
where d pl = d – 2d pl is the length of the plasma column, C, a are the

quantities determined by the type of gas.
On the other hand, the RF voltage U ext which can be applied to the
plasma column depends on U RF , j and NLSC parameters
1/ 2
U ext = U RF
2
- ( jZl ) 2 , (16.14)
where Z l = 2d l /Dee 0w is the capacitance impedance of the NLSC.

Equation (16.14) holds when the phase shift between U l and j in
the NLSC is equal to p/2 which is equivalent to the condition j a <<
j cm which is fulfilled in the NLSC of the low-current RFCD. 12,17
Under a stationary regime U pl = U ext, and from this equality tak-
ing into account (16.13)–(16.14) we can find all j at the given external
parameters (pressure, gas composition, d, w, U RF). However, as already
mentioned, in experiments we can realise only the values of j at which
the NLSC – plasma column system is stable. In this case, for the ef-
fective values of j, U pl and U ext, the stability criterion is written in the
form
dU pl dU ext
£ . (16.15)
dj dj
Equation (16.13)–(16.15) yields an expression for the minimum (nor-

mal) value of j min = j n1 in the low-current RFCD
410
1
 ε ω p d pl C α  α +1
jn1 =  . (16.16)
 2 d l / ε + 2δ / ε D 
 
When deriving equation (16.16), we took into account the possibility

of coating the electrode with a dielectric with a thickness d with di-
electric permittivity d D.
The nature of the dependences U pl (j), U l (j) (see Fig.16.6) indi-
cates that the function
1
U RF ( j ) = (U pl
2
+ U l2 ) 2 (16.17)
has a minimum value U min. Simple considerations show that at U RF =

U min at j = j n1 = j min is determined by equation (16.16). Therefore, for-
mally, the nature of the effect of the normal current density in the low-
current RFCD is the same as in the cathode region of dc glow dis-
charge. 32 However, there is also a principal difference caused by the
specific features of the spatial structure of the low-current RFCD in
which the destabilising factor (region with independent conductivity and
falling VAC) is situated at j < j cr outside the limits of the NLSC,
whereas the cathodic layer of the dc discharge (or the NLSC of the
high-current RFCD) is itself a zone with independent active conduc-
tivity. This also results in differences in the physical mechanism re-
sponsible for the appearance of U min in the cathodic layer of the glow
discharge in the low-current RFCD. In the former case, the value of
U min2 is associated with the non-monotonic form of the coefficient of
ionisation with electrons h i as a function of the parameter E l/N and
does not depend on the plasma column characteristics. On the other
hand, in low-current RFCD the U min1 is determined by the parameters
of the plasma column. This explains the relationship of the parameter
of the NLSC j, U l in the normal regime of the low-current RFCD with
the value of d (or, more accurately, d p) and pressure. In addition, this
results in a conclusion which is of considerable importance for prac-
tice: the region of existence of the lower-current RFCD with respect
to the values of PND can be expanded if the ionisation balance in the
plasma column is influenced by, for example, the electron beam. 35
It was mentioned previously that the burning regime of the RFCD
changes both suddenly and continuously. Reference (12) shows that
in the same gas the transition to the high-current RFCD is continu-
ous at low pressures and sudden at high ones. For the sake of deter-
411
Plasma Diagnostics
minacy, we shall examine the dependence of U RF on j (see Fig.16.4).

In this case, the jump on the VAC is accompanied by: 1) a large in-
crease of current density j from j cr to j n2, 2) decrease of U RF of the
electrodes, 3) the fact that stable burning of the RFCD with interme-
diate current planes j Î (j cr, j n2) is not possible. Since the transition
of RFCD from one form to another is associated with a breakdown
and transformation of NLSC, we shall pay attention mainly to the prop-
erties of the near-electrode layers prior to and after the jump.
In the low-current form of the RF discharge the thickness of the
NLSC d l1 is inversely proportional to the field frequency and depends
only slightly on gas density. 12,34 On the other hand, in the high-current
RFCD d l2 is almost independent of frequency but is inversely propor-
tional to pressure (see equation (16.8)). In a general case d l1 ¹ d l2 .
However, the pressure and frequency of the RF field can be selected
for any type of gas such that d l1 = d l2, i.e. taking into account (16.8)
vdr C1 T w T
= or p= C1 . (16.18)
w p T0 vdr T0
It is clear that in this case the RF voltage will not show any ‘jump’
on the NLSC because at d = d l2 the voltage in the layer corresponds
to the minimum of the Paschen curve U k. There is also no jump of
the capacitance component of the current in the NLSC because nei-
ther the thickness of the layer nor the RF voltage on it have changed.
As regards the active component of the current in the layers j a, af-
ter transition its value will be determined by the expression (16.10),
and the ratio of the current densities in the layer at p, satisfying (16.8),
is determined by the expression
jcm ee 0vdrU l
= . (16.19)
ja C12 C2
Estimates on the basis of (16.19) for typical discharge conditions

give that j cm/j a >> 1. Thus, if the relationship (16.18) is fulfilled the
transition of the RFCD to the high-current regime will take place without
w T
jumps on the VAC. At p >> v C1 T the VAC shows a clearly vis-
dr 0
ible discontinuity, and the sudden decrease of the RF voltage on the
electrodes is caused by at least three reasons: 1) decrease of U l from
U br to U k (breakdown on the right part of the Paschen curve); 2) ap-
412
pearance of regions with weak electric fields (analogues of glow dis-

charge and Faraday dark space) elongated along the direction of RF
current in the electrode gap after transition of the RFCD to the high-
current regime; 3) falling VAC of plasma. T 
A qualitatively different situation arises at p < (ω / vdr )  T  C1 , i.e.
 0
when d l1 < d l2. This is typical of low (usually p < 1 torr) gas pres-
sures, and the breakdown of NLSC takes place on the left (high-voltage)
part of the Paschen curve (since the number of ionisations, carried out
by the secondary emissions electrons on the characteristic thickness
of the NLSC is not high because of the low current density of the gas
density). For this reason, a large part of the secondary-emission elec-
trons, moving in the NLSC without collisions, acquire the energy ¾eU0
and form an electron beam on the side opposite to the electrode. The
energy of these electrons is considerably higher than the mean energy
of the free plasma electrons.
The dissociation of the energy of the electron beams stored in the
LNSC takes place outside the limits of the layers in the plasma col-
umn. This is equivalent to the appearance of an additional (not of the
field type) source of ionisation in the plasma and results in a decrease
of the strength of the electric field there. This situation is similar to
that observed in the plasma of a non-independent discharge sustained
by the electron beam. A specific feature of the RFCD is that the
sources of the electron beams are situated inside the discharge (in
NLSC). This circumstance facilitated the introduction of a transition
criterion to the high-current regime, or the so-called g-discharge 22,43
which differ from that examined previously (16.6). 12 According to
Ref.43, the transition to the regime of the g-discharge takes place at
RF voltages on the electrodes at which the rate of ionisation of the
secondary emissions electrons with the beam is equal to the rate of
ionisation with plasma electrons in the electric field of the plasma
UL
U pl
U l U L U meas
Ul U RF
Fig.16.8 Vector diagram of RF voltage in the circuit of passage of discharge current.
413
Plasma Diagnostics
column. It can easily be seen that the transition criterion, introduced

in Ref.43, is arbitrary to some extent and not equivalent to the crite-
rion proposed in Ref.12. If d l1 << d l2, the effect of the electron beams
on the plasma characteristics can become evident prior to fulfilling the
condition (16.6) and the RFCD transfers to the high-current regime.
Therefore, at a pressure determined from (16.18) or at a lower pressure,
the type of RFCD can be identified using additional diagnostics me-
dia, for example, probing of the RF discharge by direct current (see
below).
We shall now examine the special features of formation of the VAC
of the RFCD. With the variation of the U RF and also the magnitude
of the phase shift between U RF and IRF special attention must be given
to excluding systematic errors determined by the finite length of the
current-conducting parts of the RF electrodes which introduce an ad-
ditional inductance to the discharge circuit. Neglecting this inductance,
especially if the device for measuring the RF voltage is incorrectly con-
nected, can greatly distort the results (Fig.16.8). It can be seen that
the RF voltage measured on the ‘electrodes’ U meas can be consider-
ably lower than the actual voltage U RF applied to the discharge if the
RF voltage in the current conductor from the area of connection of
the RF voltage measuring device to the working surface of the elec-
trode U L is comparable as regards the absolute value with U l. The
typical value is U l ~ 100 ¸ 300 V. At an inductance L of the current
conductors of the electrodes ~10 –6 H, the inductive impedance at a fre-
quency of 13.6 MHz is |Z L| = 2pfL = 80 W because for typical I RF
(several amperes) U L ~ U l. Since |U L| = 2pfL I RF, to obtain U meas close
to U RF it is required either to reduce L or the device for measuring
the voltage should be connected to the electrodes using additional con-
ductors situated in such a manner that the discharge current I RF flows
through them.
It is interesting to note that the results described above can be used
to determine the voltage in the plasma column U pl because the data
on this parameter are extremely important for constructing a RFCD
model adequate for experiments. For this purpose, a conductor was
connected in series with the electrodes to the feed circuit of the RF
discharge. Its inductance exceeded the value
vdr
L= . (16.20)
ee 0w 3S
Moving the point of connection of the device for measuring RF volt-

age along the conductor it is possible to record the minimum reading
414
of the measuring device U min. In accordance with the vector diagram

in Fig.16.8, the determined value of U min determines U pl because in
the examined case the drop of the RF voltage on the NLSC of the low-
current RFCD, characterised by the capacitance conductivity is com-
pensated by the drop of the RF voltage on the inductance of the con-
ductor from the point of connection of the device for measuring the
RF voltage to the working surface of the electrode.
16.4 Constant electric fields in RFCD

The formation of constant electric fields in discharges, sustained by
a sinusoidal radio-frequency field, indicates that in discharges there are
regions with non-linear characteristics and that this can be used for
diagnostics of the spatial structure of the RF capacitance discharge.
The facts confirming the presence in the stationary RFCD of low
pressure (p < 1 torr), constant voltage U 0 localised between the dis-
charge plasma and the electrodes and comparable with the RF volt-
age on the electrodes U RF have been known for a long time. 18,19,36-42
The results show that with increasing U RF the value of U 0 also in-
creases and reaches ~10 3. 41 With increasing pressure U 0 decreases and
at at p > 1 torr it does not exceed several volts (see Fig.16.9). 18,19 As
regards the dependence of U 0 on the frequency of the feed RF volt-
age, the experimental results obtained on this subject are contradicted.
According to Ref.18, with increasing frequency U 0 decreases. The
authors of Ref.9 observed a reversed dependence with other condi-
tions being equal.
There are several viewpoints regarding the reasons of appearance
of high U 0 in the RFCD. One of them, according to which U 0 is the
result of ambipolar diffusion of the charges,42 is eroneous 18 and is only
of historical interest at present.
The explanation proposed in Ref.18 appears to be more acceptable.
According to this explanation, the formation of U 0 in the RFCD is as-
sociated with the formation of near-electrode layers of a positive spatial
discharge caused by the fact that during the cycle of the RF field part
of the electrons are removed from the discharge gap as a result of
high mobility. It was postulated that the characteristic thickness of the
NLSC d l is determined by the vibration amplitude of the electrons in
the discharge, i.e.
m 0 Em
dl = , (16.21)
pw
where m 0 is the mobility of electrons at a pressure of 1 torr, E m is the
415
Plasma Diagnostics
amplitude value of the strength RF field. From Poisson’s equation under

the condition of stationary ions in their layer, an equation was derived
for the constant potential of the plasma of the RFCD in relation to the
electrodes 18
m 20U m2
U 0 = 6pen p d l2 = 6pen p , (16.22)
w 2 p 2d 2
where U m is the amplitude value of the radio-frequency voltage ap-

plied to the discharge, np is the concentration of the positive ions, where
n p ¾ ne.
However, the results of calculations carried out using equation (16.22)
do not agree with the experimental values. For example, at f = 3 MHz,
p = 1 torr, d = 12 cm, U m = 300 V and n p = 10 9 cm –3, U 0 = 650 V,
whereas the value of U 0 measured under these conditions is 65 V. 18
This casts doubts on the whether it would be efficient to examine the
vibrational motion of the plasma electrons in the RFCD.
According to the authors of Ref.40, in the plasma of the station-
ary RFCD where the directional velocity of the electrons in the RF
field is considerably lower than there thermal velocity, the concept of
the amplitude of the vibrations of the electrons in the RF field has
hardly any physical meaning, and the spatial scale of the division of
the charges on the plasma boundaries d l is determined not by the am-
plitude of vibrations of the electrons but by the polarisation length of
the plasma in the electrostatic field. At low potentials (U 0 << V e, V e
is the electronic temperature in potential units), the value dl is evidently
equal to the Debye radius D e
1
 Ve  2
d l = De =   . (16.23)
 4 πene 
At high potentials (U 0 >> V e)
1
 U0 2
dl =   . (16.24)
 4 π ene 
The very fact of appearance of U 0 in the RF capacitance discharge

is interpreted by the authors of Ref.40 as the result of rectification
416
of the RF voltage on the non-linear complex conductivity of the near-

electrode layer of the spatial charge. For the case of low pressures
where it is possible to ignore the collisions of the electrons in the RFCD
and assume that the entire RF voltage applied to the electrode is lo-
calised in the layers, the equation has the form
2 pme U RF
U 0 = Ve ln - , (16.25)
M p
where m e, M is the mass of the electron and the ion, respectively.

However, the conclusion of the authors of Ref.40 according to the
vibrational model of the spatial structure of the low-current RFCD is
not correct has not been confirmed, as indicated by the experimental
data presented in Ref.12. where the experiments have confirmed the
concept of the layer spatial structure of the low-current RFCD, including
near-electrode regions with low active conductivity and the plasma
column. In a wide pressure range from units to hundreds of torr, the
equation of continuity for the discharge current can be written in the
form
m 0 E pl
ene @ ee 0wE l . (16.26)
p
The measured value of U 0 is linked with the RF voltage in the layers

U l by the relationship
U 0 = kU l , (16.27)
where k ¾ 1 is a constant. Assuming that the density of the positive

ions n p is constant to a first approximation along the discharge gap and
equal to ne in the plasma column, from Poisson’s equation, taking into
account equation (16.26) and (16.27), we obtain an equation for de-
termining the thickness of the RFCD d l
m 0 E pl vdr
dl = k @ . (16.28)
pw w
Comparison of (16.28) with (16.21) shows that the characteristic

thickness of RFCD is determined (despite the results published in Ref.40)
417
Plasma Diagnostics
but the amplitude of the drift vibrations of the RFCD electrons in the
plasma field Epl. It is important to stress the large difference of (16.28)
in relation to (16.21) caused by the fact that the valuable vibrations
of the electrons in deriving (16.28) are assumed to be taking place in
the electric field of the plasma column E pl are not in the vacuum field
E m = U m/d, as implicitly assumed in Ref.18. It can easily be seen that
the latter circumstance also leads to the previously noted large differ-
ence (by an order of magnitude) of the experimental values of U 0 from
those calculated from equation (16.22).
The attempt to explain the dependence of U 0 on the voltage of the
electrodes URF, observed in Refs.18, 41, by assuming that Um from equa-
tion (16.22) is identical with the amplitude value U RF is incorrect. In
fact, according to the layer model of the low-current RFCD, 12
U m 4 U pl , but
 E pl   E pl 
U pl =   pd pl ≅   pd , (16.29)
 p   p 
where d pl is the length of the plasma column, and when d >> d l then
d pl ¾ d. Equations (16.29) and (16.22) show that U 0 does not depend
explicitly on the RF voltage of the electrodes, gas pressure, and gives
values of the constant potential of the plasma in relation to the elec-
trodes similar to those observed in practice.
In reality, U 0 is related to U RF with a concentration of charged par-
ticles n p ¾ n e, which can easily be confirmed using the relationships
(16.22), (16.29), (16.26) and (16.17):
 µ e E pl  vdr  ω  dl 3 3
U 0 = 3π  ene  2 = 3π  El  ≅ U l = 2
U RF − U 2pl . (16.30)
 p ω  4π  ω 4 4
In deriving (16.30) the capacitance current in the NLSC is expressed

in the system of CGS units, j l = (w/4p) E l, and the numerical coeffi-
cient in (16.22) is halved in accordance with Ref.32.
Thus, the analysis of the conditions of appearance U 0 taking into
account the layer structure of the low-current RFCD shows that the
constant potential of the plasma in relation to the electrodes U 0 is de-
termined unambiguously by the decrease of the RF voltage on the NLSC
U l. As noted previously, the physical reason for the appearance of the
NLSC with high U 0 in an independent RFCD is the different mobil-
ity of the electron and ion components of the plasma. This circumstance
418
leads unavoidably to the formation of NLSC with high U0 in a stationary

RFCD. This equalises the number of electrons and ions falling of the
electrodes during a cycle of the RF field.
As indicated by (16.30), there is no explicit relation of the constant
potential of the plasma with pressure and frequency RF field. At the
same time, the dependence of the RF voltage on the NLSC on p and
is determined by the equation
vdr ( E pl / p)
U l ( p, w ) = j ( p, w ) , (16.31)
ee 0w 2
where j(p, w) is at the discharge current density. It is well known 16,17

that with increasing p normal, i.e. minimum, density of the discharge
current decreases, and E pl /p, determined by the ionisation balance in
the plasma changes only slightly and, consequently, drift velocity of the
electrons in the plasma also changes only slightly. Therefore, increasing
pressure increases the minimum value of the RF voltage on the NLSC
U l.n at which a low-current RFCD still can exist and, according to
(16.27), U0 (p) will also increase. This contradicts the conclusions made
in Ref.18. However, this contradiction is removed if we take into account
the real layer structure of the RFCD and, as already mentioned, U m
in (16.22) is the decrease of the RF voltage in the plasma and not the
RF voltage in the electrodes.
Thus, if it is assumed that the positive ions are stationary, the equation
of stationarity of the low-current RFCD shows that the value of U 0
is equal to the amplitude value of U l, applied to the NLSC, irrespec-
tive of pressure and with the accuracy equal to the electronic tem-
U cp, V U cp, V
a b
log p, torr log p, torr
Fig.16.9 Dependence of constant potential of plasma on pressure 18(a) and b 19 .
419
Plasma Diagnostics
U RF
Zl U l.p.
E slo
Z bl
Cp
Fig.16.10 Principal (a) and equivalent (b) diagram of circuits of probe measurements
of constant plasma potential. 1) electrode; 2) discharge chamber; 3) probe.
perature of plasma. Taking into account the real movement of the ions
in the layer does not greatly reduce the value of U 0 since µ e/ µ p > 10 2.
However, probe measurements of the constant potential of the RF
plasma (Fig.16.9) 18,19 indicate that U 0 greatly decreases with increasing
pressure p > 1 torr.
We shall examine in greater detail the procedure of probe
measurements U 0 in the RFCD proposed in Ref.44. According to this
proced-ure, for accurate measurements of the constant potential of the
plasma of RFCD it is necessary and sufficient to organise measure-
ments in such a manner that the RF component of the voltage between
the probe and the plasma U l.p is low. It has therefore been proposed
to place a choke coil with a high inductive resistance for RF current
between the probe and the measuring circuit. A similar procedure was
used in Ref.18 and 19. The principal measurement circuit is shown in
Fig.16.10a, and the equivalent circuit in Fig.16.10b, with the follow-
ing notations: U ≡ – the RF voltage between the region of unperturbed
plasma containing the probe (point A) and the earth; U 0 – the constant
voltage between the plasma and the electrode; E slo – the constant
voltage between the plasma and the probe surface; C p – parasitic
capacitance of the probe and the blocking element on the ground; Z bl
– the impedance of the blocking element of the probe with respect to
RF voltage; Z l – the impedance of the plasma–probe layer; C – the
capacitance of the condenser on which U 0p (recorded constant volt-
420
age) generates; L bl the inductance blocking the RF current; U l.p – the

decrease of the RF voltage at the near-probe layer.
Figure 16.10b shows that
U0 p = U0 - Eslo . (16.32)
The procedure error of measurement U 0 according to (16.32) is as-

sociated with E slo .
E l can be written in the form of the sum of two independent terms:
k 1V e which depends on the electronic temperature of plasma, and k 2U l.p
– the component determined by the passage of RF current to the probe
(k 1, k 2 are constant quantities). The term k 1V e is determined by the
properties of plasma and cannot be eliminated if the probe is placed
in the plasma. The second term is the linear function U l.p. It would
appear that by selecting a corresponding value of Z bl , k 2 U l.p can be
reduced to the value comparable with k 1V e. However, the effect of the
parasitic capacitance of the probe and the blocking element C p in re-
lation to the earth makes it irrational to increase the impedance of the
1
blocking element above Zbl = because of the shunting effect of
wC p
C p. Therefore, the only way of increasing the accuracy of measure-
ments of U 0 is to decrease C p which is reduced to the rational posi-
tioning of the probe, the blocking element, the selection of their dimen-
sions, etc.
Additional difficulties in evaluating the accuracy of measuring U 0
are associated with the absence of reliable theoretical models which
will make it possible to evaluate Z l at medium and higher pressures.
The method of theoretical evaluation of the error of probe measure-
ments, proposed in Ref.45, is valid only for low pressures in the dis-
charge (p << 1 torr) when the near-probe layers of the special dis-
charge can be regarded as collisionless. Under other conditions, ex-
perimental verification of the accuracy of the probe measurements of
U 0 becomes important.
It was shown that the accuracy of measurement of U 0 was deter-
mined mainly by the value U l.p which should satisfy the inequality U l.p/
U << 1. This is equivalent, according to Fig.16.10b, to fulfilling the
condition
Zl . p / Z S << 1. (16.33)
421
Plasma Diagnostics
U 03, V U 03, V
log Z bl
C add
U RF , V
Fig.16.11 Dependence of readings of the probe on the size of the blocking element
Z bl. Low-current RFCD, air, p = 10 torr, f = 13.6 MHz, flat disc electrodes diameter
60 mm, d = 30 mm, probe in the centre of the discharge gap; 1,3) blocking element is
made in the form of low-capacitance high-resistance resistor, 2) in the form of a choke
coil.
Fig.16.12 (right) Dependences of U 0p on RF voltage on electrodes. p = 7.5 torr; 1)
RFCD in air, 2) in CO 2 .
Here Z S is the impedance of parallel-connected C p and Z bl. The ca-

pacitance C is not considered in the calculations because its value can
always be selected as satisfying the condition 1/wC << Z bl.
The validity of the previously formulated conclusions on the error
of probe measurements of U 0 can be verified by experiments by ref-
erence measurements of Z bl and C p (Fig.16.11). Curve 1 represents
the dependence U 0p and Z bl on the semi-logarithmic scale, curves 2,
3 the dependence of U 0p and C p, varied by additionally connecting the
capacitance C add .
At low Z bl the probe potential assumes negative values. Optical ef-
fects, similar to near-electrode effects, are observed in the vicinity of
the probe, i.e. at low Z bl the probe represents an additional electrode.
The dependence of U 0p on Z bl is initially monotonically increasing and
then reaches ‘saturation’. This result confirms the correctness of the
equivalent circuit (see Fig.16.10b), but it doesn’t enable any conclu-
sions to be drawn regarding the accuracy of measurements because
the question of the reason for ‘saturation’ of the dependence U 0p (Z bl )
remains open. The latter can occur in two cases: either the the fixed
U 0b actually approaches U 0 as regards its value, or the shunting ef-
fect of C p restricts the increase of the effective blocking impedance
ZS.
The situation can be clarified by analysing curves 2 and 3 in Fig.16.11.
If a decrease of the artificially introduced parasitic capacitance to zero
leads to ‘saturation’ of U 0p, then the error of measuring the data by
the probe under these conditions is close to minimum determined only
422
U 0, V
Fig.16.13 Distribution of constant plasma

potential U 0 (x) along the electrode gap.
p = 7.5 torr. U RF = 305 V, f = 13.6 MHz;
1) low-current RFCD in CO 2 , 2) in air, 3)
x, mm in nitrogen.
by a correction for the electronic temperature of plasma T e. In par-

ticular, Figure 16.11 shows that when Z bl is represented by a low ca-
pacitance resistor, the accuracy of measurement of U 0 rapidly increases
and the value U 0p reaches ‘saturation’ at C add ® 0. This cannot be
said of curve 2. Evidently, this is associated with high parasitic capaci-
tance of the choke coil because the probes themselves were identi-
cal in both cases.
The typical dependences of U 0p on U RF on the electrodes under dif-
ferent conditions of burning of the RFCD are presented in Fig.16.12.
Attention should be given to a dual dependence of U 0b on U RF as a
result of the transition of RFCD to the high-current burning regime.
It should be noted that the increase of U 0p with decreasing U RF again
indicates the localisation of U 0 in NLSC and the redistribution of RF
voltage in the electrode gap in favour of the near-electrode layers in
transition of RFCD to an anomalous high-current regime.
Making the probe movable, it is possible to record the distribution
of the constant potential U 0 in the electrode gap. To confirm the layered
structure of the RFCD, it is of special interest to examine the redis-
tribution of U 0 (x) along the direction of passage of the RF current
(Fig.16.13). It can be seen that U 0 is localised in the vicinity of the
electrode. The same situation exists at the opposite electrode, i.e. U 0
(x) distribution is symmetric and, consequently, Fig.16.13 shows the U0
(x) only at one electrode. The maximum value of U 0 (x) is obtained
in the regions of stronger glow of the plasma column (see Fig.16.2a).
423
Plasma Diagnostics
Cp
Cl Cl
U RF
R act
Fig.16.14 Diagram of the discharge chamber for examining the appearance of constant
voltage on the capicitance C 0 .
The non-monotonic distribution of U0 (x), is clearly visible in Fig.16.13,

has not as been satisfactorily explained.
It should be stressed that the accuracy of measurement of U 0 is
strongly influenced by the dependence of Z l.p on the external experi-
mental conditions, the type and pressure of gas, the frequency of the
RF field, etc. A suitable example are the results of probe measure-
ments of U 0 presented in Ref.18 and 19. The experiments carried out
with the same probe, i.e. at Z bl = const, did not nevertheless make it
possible to record values of U 0 at p > 1 torr, whereas at a pressure
of p << 1 torr the same probe gave satisfactory results which were
in agreement with the results of measurements by other methods (for
example, based on measuring the energy of positive ions leaving the
NLSC of the RFCD). 18 It is evident that with increasing pressure in
the discharge chamber the value of Z l.p also increased, and the ratio
of Z l.p/Z bl , included in the criterion (16.33), decreased together with
the accuracy of measurements. The measures taken in Ref.12 in fact
increased Z bl so that the probe method made it possible to detect (in
contrast to Ref.18 and 19) high values of U 0 in the RF capacitance
discharge and at p > torr.
The effect of Z l.p on the results of measuring the constant poten-
tial of the RF plasma makes it necessary to find other methods of ob-
taining quantitative information on U 0. It should be noted that in many
cases it is interesting to take correct measurements of the maximum
value of U 0m and not its distribution in the electrode gap. In addition,
as shown previously, U 0 is localised in NLSC and the characteristic
424
thickness of the NLSC in the radio-frequency range (~ 1000 MHz) or

high-current RFCD at medium or higher pressures is small and equals
fractions of a millimetre. 13 This complicates the measurements. How-
ever, to obtain information on the maximum value of U 0 it is not nec-
essary to place the probe in the plasma of the RFCD in the vicinity
of NLSC. In fact, we shall examine the experimental results obtained
under the conditions shown schematically in Fig.16.14. 12 The RFCD
was excited between water-cooled electrodes made in the form of discs
with a diameter of 10 cm which were placed in the centre of a me-
tallic chamber with a large volume (60 litres). A reference capacitor
with a capacitance of C 0 was connected in series between one of the
electrodes and the earth to calibrate the RF current measuring device.
A constant voltage UC0 appeared on the reference condenser. This
voltage reached 100 V or higher and depended on the U RF on the elec-
trodes. In discharges in helium and nitrogen this voltage remained at
pressures of up to 50 torr and higher, whereas for CO 2 at p > 5 torr
it was not possible to record high values of UC0 .
Evidently, it may be expected that there is a relationship between
U 0 and UC0 . The attempts to link them through the parasitic capaci-
tance of the plasma C pl in relation to the earthed casing (Fig.16.14)
were not successful because, according to the estimates, the capaci-
tance of the NLSC C l >> Cpl (in the opposite case, the discharge would
be short circuited with the casing and not the earthed electrode).
Better results were obtained in the basis of the assumption on the
existence of a finite active conductivity R act between the RFCD plasma
and the earthed casing of the chamber. The measurements carried out
using the circuit shown in Fig.16.14 to verify this assumption in a quartz
tube, showed that there is no constant component of the RF voltage
on C 0. A similar result was obtained due to an artificial increase of
C pl by placing the quartz tube with a discharge in an earthed metal-
lic screen. However, making a small hole in the wall of the quartz tube
and introducing a thin conductor into it, connected with the earth through
a low-capacitance resistor (1 Mohm), it was possible to record UC0
comparable with that of observed in the metallic chamber.
Thus, a low-conductivity medium exists in the metallic chamber be-
tween the discharge plasma and the walls. In Fig.16.14 this medium
is denoted R act and the capacitance C 0 is charged to UC0 through this
medium. The absence of high values of UC0 in the RFCD on CO 2 at
p > 5 torr can be explained by additional annihilation of the charges
outside the discharge zone, typical of the electronegative gases.
425
Plasma Diagnostics
ε0, V
U RF
Cbl
R bl
U RF , V
Fig.16.15 Coaxial electrode system for examining the battery effect in the RFCD. 1)
internal, 2) outer RF electrode; R bl, C bl – blocking resistor and condensor.
Fig.16.16 Typical dependences of constant EMF ε 0 on RF voltage of electrodes U RF.
f = 13.6 MHz; 1,2) air, p = 7.5; 15 torr, 3) helium, p = 100 torr.
Thus, experiments show that at large distances from the RFCD

plasma (~10 cm) there is still a region of weakly ionised gas with high
conductivity. It can therefore be assumed that a small probe, placed
in this region, receives the potential of the space at a given point.
Actually, if there were no such region, the charges induced by the
measured field on the metallic probe would greatly distort the result.
In the absence of a slightly ionised medium the induced charge of the
probe is compensated by the flux of charged particles from the space
around the probe. Consequently, the probe has the potential of the space.
Measurements of U 0m by the proposed method are now reduced to
placing the probe of the equipotential corresponding to maximum value
of U 0. In practice, measurements are carried out using the following
procedure: the probe is moved at the periphery of the discharge and
the maximum reading of a high-resistance voltage measuring device
is recorded. This device is represented by, for example, an electrostatic
voltmeter protected against RF interference. The value of U0m, obtained
using this procedure, coincides, with the accuracy to several percent,
with the results of measurements of U0m by a ‘floating’ probe immersed
in the plasma. However, in contrast to the latter method, the super-
heating of the probe and the blocking element is prevented, there are
no breakdowns of the surface of Z bl, and the effect of C p is weaker.
Disadvantages of the method of the displaced probe include only the
fact that it is necessary to change the spatial position of the probe when
the burning conditions of the RFCD are changed.
To pull out complete information, it is important to note another method
426
of increasing the accuracy of probe measurements under RFCD con-

ditions based on using resonance filters – plugs as blocking elements. 46
However, the application is restricted by the relatively low frequency
of the RF field (units of megahertz) due to the low efficiency of the
filters–plugs and the fact that the parasitic capacitance cannot be re-
moved.
16.5 Using the battery effect in examining RFCD

It was shown in the previous section that near-electrode layers of the
spatial charge form in the stationary RFCD irrespective of pressure.
The constant potential of the RF plasma in relation to each electrode
is proportional to the RF voltage applied to the NLSC U l. It was also
reported (see Fig.16.6) that the thickness of the NLSC d l in the low-
current form in the RFCD is almost independent of the discharge current
density. Thus, if the RF capacitance discharge burns in stationary man-
ner under the conditions in which the density of the RF current in one
of the NLSC j 1 is higher than the current density j 2 in another near-
electrode layer, then U l1 > U l2 and U 01 > U 02 , according to (16.31),
(16.27) and (16.30). The latter inequality shows that a constant EMF
ε0 appears between the electrode of such a non-symmetric RF discharge.
Its value is
e 0 = U 01 – U 02 (16.34)
These RF electrodes are closed with respect to direct current, for

example, they are joined by a choke coil, a constant electric current
I 0 = (U 01 – U 02 )/r in forms in the circuit. Here r in is the active inter-
nal resistance of the discharge including the in-series connected ac-
tive resistance of the plasma and both NLSC.
From the technical viewpoint, the non-symmetric RFCD can be realised
most efficiently in a coaxial electrode system (Fig.16.15). In this con-
figuration of the electrodes, the RFCD in the region of the pressures
p << 1 torr actually shows a battery effect, i.e. appearance of high
ε 0 and constant currents I 0 when closing the circuits. 47 Because the
existence of the NLSC, which is the reason for the appearance of a
high constant potential of the RF plasma in relation to the electrodes,
have been established without doubt by several methods, the appearance
of ε 0 under these conditions is quite regular.
A different situation exists in the region of pressures higher than
1 torr where the situation is ambiguous and experimental data 18,19 in-
dicating a rapid decrease of U 0 with increasing p have been published.
Therefore, the battery effect should be used for independent verification
of the existence of NLSC and high U 0 in the RFCD at p > 1 torr.
427
Plasma Diagnostics
Figure 16.16 shows typical dependences e 0 (U RF ), 48 obtained in a

coaxial electrode system represented schematically in Fig.16.15. The
constant potential of the smaller electrode was negative in relation to
the larger one.
Attention should be given to two special features in the behaviour
of e 0(U RF): high values of e 0 at relatively high p >> 1 torr; a rapid re-
duction of ε 0 (to almost zero) when the RF voltage on the electrode
reaches some value U br which depends on the type of gas, pressure
and electrode material.
Analysis of the experimental results leads to the following conclusions:
irrespective of the pressure, NLSCs form in the RFCD, and the constant
voltage on the NLSC reaches hundreds of volts. A decrease of ε 0
when the RF voltage on the electrodes exceeds U br does not indicate
the disappearance of NLSC and indicates the formation, in the elec-
trode gap, of a qualitatively other, high-current RFCD regime where
U 01 = U 02 (regardless of the asymmetry of the electrodes).
In fact, at U RF > U br, breakdown of the capacitance NLSC of the
RFCD takes place, and a new spatial structure forms in the electrode
gap. One of the distinguishing features of the latter is that the RF voltage
on each of NLSC U l1 and U l2 is determined, as in the case of the
cathode region of normal glow discharge, only by the nature of the gas
and the material from which the electrodes are made. However, since
the gas, filling the electrode gap, and the material from which elec-
trodes are made, have not changed, U l1 = U l2 in the given burning re-
gime. Therefore, taking into account (16.27) we obtain that U 01 = U 02
and, according to (16.34), e 0 = 0.
When the negative glow discharge fills completely the electrode with
a small area, the value of e 0 differs from 0 because in this case the
electrode with a small area operated under the anomalous regime, i.e.
at higher U l.
The considerations regarding the reasons for the large decrease of
e 0 in these experiments in the transition of the RFCD to the high-re-
gime were verified by the initial experiments. The RF discharge was
ignited in a symmetric electrode system (S 1 = S 2), but one of the elec-
trodes was made of copper and the other one of duralumin. The re-
sults show that in the low-current form of the RFCD, i.e. when no
electric breakdown have taken place in the NLSC, the value of e 0 is
close to 0 at all values of U RF in the range U n £ U RF £ U br (U n is the
minimum RF voltage on the electrodes at which the low-current form
of the RFCD exists under the given specific conditions. However, in
transition to the high-current regime e 0 ¹ 0. For example, for a high-
current RFCD in air e 0 = 70 V, and the potential of the copper elec-
trode is lower.
428
16.6 Effect of the frequency of RF field on the structure of

the discharge and plasma characteristics
The frequency of the RF field is the most important parameter affecting
the main characteristics of the RF capacitance discharge. There have
been a number of studies49–52 in which attempts were made to describe
the mechanism of the effect of frequency on the characteristics of the
discharge plasma, especially on its conductivity spl, the reduced strength
of the electric field E pl/p, and others. The specific feature of the RFCD
is that the kinetic processes, taking place in the plasma of the RF ca-
pacitance discharge, differ from the corresponding processes in the
plasma of the dc discharge. 51,52 However, it has not been possible to
explain the effects observed in the experiments, for example, the in-
crease of the active conductivity of the plasma spl with increasing fre-
quency w, examining only the phenomena taking place in the RFCD
plasma. In fact, in the frequency range f Î 1 ¸ 100 MHz, the gas pres-
sure range p Î 1 ¸ 100 torr, typical of practical application of the
RFCD, 1–13,49–52 the active conductivity of the plasma in the RF field
should not depend on the frequency ω and in accordance with the well-
known equation 32
e 2 ne vm
s pl = . (16.35)
m(w 2 + vm2 )
Since in the examined frequency range of the RF-field w and the gas
pressure range the frequency of collisions of the plasma electrons with
neutral particles is n m >> w, then s pl = e 2n e/mn m.
In reality, the conductivity of the RFCD plasma s pl and the elec-
tron concentration in it are determined not only by the local ionisation
balance of the charged particles but also by the spatial structure of
the discharge, especially by the conditions of closure of the discharge
current at the plasma–electrode boundary where, as mentioned pre-
viously, NLSC with frequency-dependent properties form.
The effect of the NLSC on the conductivity of the RF discharge
plasma is very evident when comparing the two forms of the RFCD
examined previously (Fig.16.1). Depending on whether a breakdown
has taken place in an NLSC (b) or not (a), the conductivity of the
plasma column changes by no more than an order of magnitude, ir-
respective of the completely identical experiment conditions, including
similar values of E pl . 13 The jumps in the parameters of the RFCD
plasma with changes in the discharge burning conditions can be pre-
dicted, with the accuracy acceptable for many practical applications,
429
Plasma Diagnostics
by comparing equations for current densities jn1 (16.16) and jn2 (16.12)
which depend on the field frequency.
We shall examine the low-current from the RFCD and explain the
possibilities of controlling the parameters of this type of discharge by
changing the frequency.
There are a large number of experimental data, discussed previously,
and numerical calculations 28,29 which show that the most characteristic
feature of the low-current RFCD is that the inequality
s l << s pl . (16.36)
is fulfilled. Here s l is the average active conductivity of NLSC. Di-

rect experimental methods of verifying the validity of (16.36) in the
low-current RF discharge will be examined in the following chapter.
When (16.36) is fulfilled, the continuity equation for the discharge current
of the low-current RFCD is written in the form (16.26) which indicates
the dependence of the conductivity of the plasma and of electron con-
centration in it on the frequency of the RF field ω:
El
s pl = ee 0w , (16.37)
E pl
ee 0 E
ne = w l . (16.38)
em e E pl
The equations (16.37) and (16.38) show that s pl and n e are propor-
tional to the frequency of the RF-field, with other conditions being equal.
At fixed frequency, n e is determined by the E l/E pl ratio, and for given
p, d, w the strength of the electric field in the NLSC is restricted at
the bottom by the value E l.n = j n1/ee 0w, where j n1 is given by the ef-
fect of a normal current density in the low-current RFCD (16.16). 34
On the other hand, E l cannot be higher than E br at which the condi-
tion (16.6) is fulfilled, i.e. a breakdown of NLSC takes place in which
secondary emission electrons take part, and the discharge changes to
the high-current regime. E pl is determined from the ionisation balance
equation. Under the typical conditions of application of RFCD the de-
pendence E pl (n e) is ‘drooping’ (see Fig.16.6). 34
Thus, at the fixed frequency of RF field w = 2pf the value of n e
in the plasma of the low-current form of the RFCD can be regulated
by changing the RF voltage at the electrodes in the range
430
ee 0 El .n ee E
w £ ne £ 0 br w. (16.39)
em e E pl1 em e E pl 2
Attention should be given to another important special feature of

the low-current RFCD which is of considerable importance for the ap-
plications, for example, in laser technology. It is well known that the
secondary-emission processes play a secondary role in the mechanism
in the method of sustaining the discharge of this type. 12 Therefore, it
would appear that the emission characteristics of electrodes should have
no effect on the energy characteristics of the CO 2 laser with RF ex-
citation. However, opposite results were obtained in the experiment. 53
Nevertheless, there is no contradiction here. Although the material of
the electrodes has only a slight effect on the structure and parameters
of the low-current RFCD, 12 it does determine the threshold of tran-
sition of the RF capacitance discharge to the high-current burning regime
because the secondary-emission processes determine the value of E br
and, consequently, also (ne)max (see equation 16.39). This leads to a very
important conclusion for practice: coating the electrode with a mate-
rial with low emission properties increases the power supplied to the
plasma of the low-current RFCD without increasing the frequency of
the RF voltage feeding the discharge.
We shall now clarify the dependence of the minimum RF voltage
at the electrodes of the low-current RFCD at the given values of the
electrode gap and the gas pressure on the frequency of the RF field.
The experiments show that U min decreases with increasing frequency. 30
For this purpose, we shall use the considerations regarding the layer
structure of the low-current RF discharge including capacitance near-
electrode layers of the special discharge with the RF voltage on them,
satisfying the equation
vdr
Ul = j , (16.40)
ee 0w 2
and the RF voltage on the plasma column U pl in accordance with equa-

tion (16.13). Taking into account the capacitance nature of the NLSC,
i.e. the fact that the phase shift between U pl and U l is close to p/2,
we can write the RF voltage on the electrodes in the form of (16.17).
It is well known that the U min at the electrodes forms at j = j n1, where
j n1 is the normal current density in the low-current RFCD. 34 Conse-
quently, substituting the equations for U l, U pl and j n1 into the equation
(16.17), in accordance with (16.40), (16.13) and (16.16), we obtain that
431
Plasma Diagnostics
the minimum RF voltage at the electrode of the low-current of the

RFCD as a function of the frequency U min (ω) can be written in the
form
FG IJ
1
1 4Cvdr 2
Umin (w ) @
H
w ee 0
pd pl
K . (16.41)
When deriving (16.41) and (16.16) it was assumed that d = 0 and

a = 1. In fact, the experiments show that a is slightly higher than unity
and depends on the type of gas. However, to a first approximation,
equation (16.41) gives results that are in satisfactory agreement with
the experiments and, most importantly, explains the nature of the de-
pendence of minimum voltage on the electrodes on frequency.
Using the layer model of the low-current RFCD and substituting
j = j n1 into (16.13), we can easily derive the dependence of the re-
duced electric field in the plasma E pl/p as a function of the frequency
of the RF field
E pl C ( 2 d l / e + 2d / e d )
= (16.42)
p e 0wpd pl
or on bare electrodes where d = 0
E pl 1 2Cvdr
@ . (16.43)
p w ee 0 pd pl
Equations (16.42) and (16.43) show that E pl/p decreases with in-
creasing frequency of the RF field, with other conditions being un-
changed. However, detailed analysis shows that this special feature of
the RFCD plasma is based on the ‘drooping’ nature of the VAC of
the plasma (decrease of U pl with increasing j, see Figure 16.6) but
it is not caused by the difference of the elementary processes in the
plasma of the RFCD in comparison with the corresponding processes
in the plasma of the positive column of the dc glow discharge at least
in the examined frequency range of the RF field (1–100 MHz) and the
gas pressure from units to hundreds of torr. To confirm this, it is suf-
ficient to increase artificially the impedance of the NLSC at w = const,
for example, by coating the electrodes with a dielectric with the char-
432
acteristic thickness d > e d d l . In this case, both normal current

density j n1 (16.16) and E pl/p (16.42) can be changed by selecting the
appropriate thickness δ of the dielectric coating on the electrode sur-
face.
The experiments confirming the previous conclusions were carried
out using the following procedure. The discharge chamber was formed
by two flat quartz sheets with a thickness d > e d dl. A metallic coating
was sprayed on one of the sides of both sheets. The minimum cur-
rent density j n1 and the electric power scattered in the discharge were
measured (by calorimetry) at the same values of w, p, d and the type
of gas in two cases: 1) the sprayed sides of the sheets are turned
towards the inside of the discharge chamber, which is equivalent to
d = 0; 2) the sprayed surfaces of the quartz sheets are positioned on
the outer side of the chamber, i.e. d ¹ 0. It appears that j n1 and the
value of the power scattered in the RFCD is always smaller in the sec-
ond case. If the parameters of the discharge were completely deter-
mined by the plasma column, the presence of the dielectric coating would
have no effect on the characteristics of the plasma of the low-cur-
rent RFCD and would depend only on the frequency of the RF field.
Taking into account the actual layer structure of the low-current
RFCD, it is also possible to understand other special features of the
RF capacitance discharges, especially the non-monotonic dependence
of the RF voltage on the electrodes on the pressure in the discharge
chamber U RF (p) at the fixed specific energy input into the plasma
w. 51,52
In fact, the following equation holds for the low-current RFCD
w = jE pl ; (16.44)
Ul
j @ ee 0w , (16.45)
dl
E pl = Ep, (16.46)
where E = (E pl /p).
Consequently, taking into account the layer structure of the RFCD,
i.e. s l << s pl , we have
L F 2wv I 1 O
1
( p) = Md Ed i p + G P
2 2
H ee w E JK
1
MN p P
2 dr
U RF . (16.47)
Q
pl 2 2
0
433
Plasma Diagnostics
Equation (16.47) shows that the minimum value U RF (p) is obtained

at
1 2 wvdr
p* @ . (16.48)
Ew ee 0d pl
Substituting into (16.48) the typical values for the low-current RFCD,
for example w = 10 2 W/cm 3; v dr = 10 7 cm/s; E = 10 V/(cm torr), it
can be seen that equation (16.48) is in quite satisfactory agreement
with the experiment. 51,52
Of special interest is the expression (16.47) because it can be used
to estimate the mean reduced electric field in the plasma on the ba-
sis of the measured values of w (for example, by calorimetric meas-
urements of the discharge) and the RF voltage on the electrodes.
Thus, taking into account the real layer structure of the RFCD is
the essential condition for correct interpretation of the frequency char-
acteristics of the RF capacitance discharge.
16.7 Active probing of RFCD using a constant electric

current
In the group of various methods of examining the spatial structure of
the RFCD, special attention is given to the method of probing the RF
capacitance discharge with a constant electric current which, combined
with calorimetric measurements of the power scattered in the RFCD
makes it possible to identify unambiguously the burning regime of the
discharge, determine the integral characteristics of the near-electrode
layers of the spatial charge (thickness d l , active resistance R l ), the
strength of the electric field in the plasma E pl, etc.
In the method, a source of controlled constant voltage U p is connec-
ted to the RF electrodes if they are not coated with a dielectric (or,
in the opposite case, to the electrodes additionally introduced into the
discharge gap). The volt–ampere characteristic of the circuit includ-
ing the examined RFCD is then recorded. The linear section of the VAC
of the probing circuit is used to evaluate the active resistance of the
object placed between the probing electrodes. If it is necessary to
exclude from examination the near-electrode regions of the spatial
charge which always form at the surface of the probing electrodes,
the volt–ampere characteristics of the probing circuit are recorded at
different distances between the additional electrode, but constant probing
current I p .
The method of active probing with the constant electric current of
434
the radio-frequency discharge is attracted because it is simple to sepa-

rate the discharge current I RF and probing current I p using frequency-
dependent elements (inductance coils, condensers).
We shall discuss the method of determining d l, R l in the two pre-
viously examined burning regimes of the RFCD using active probing.
In this case, it is convenient to use flat cooled electrodes, not coated
with a dielectric, with area S. The distance d between electrodes can
be varied. Since it is required to determine the integral characteris-
tics of the NLSC, to exclude the effect of the plasma column on the
results of measurements the value of d is selected minimum permis-
sible at which the required burning regime still exists.
To apply the method, it is necessary to measure the RF voltage on
the electrodes U RF, discharge current R RF and also record the area of
electrodes occupied by the discharge S pl . The RF generator is con-
nected to the electrodes via the condenser C bl. A stabilised source of
constant voltage with regulated U p and the device for measuring the
current I p are connected to the same electrodes using the blocking in-
ductance coils L bl. As a result of using L bl and C bl, the RF discharge
and constant probing currents are completely separated. I RF is deter-
mined by U RF and by the total impedance of the discharge gap, and
I p depends on U p and the active conductivity of the discharge. As a
result of selecting d the effect of the column of RFCD plasma on the
conductivity of the discharge gap can be minimised. Consequently, the
voltage U pS bl/I p determines the active resistance of the near-electrode
regions R l of unit area, and by analogy U RF S bl/I RF gives the total im-
pedance of the same near-electrode layers Z l, also related to the unit
area.
Knowing Rl and Zl, we calculate the effected value of the capacitance
component of the conductivity of the layers and, consequently, we de-
termine the average thickness of the near-electrode layers d l (for the
period of the RF field) under different burning conditions and deter-
mines its dependence on the gas pressure and frequency, carrying out
these measurements at different values of p and ω.
Assuming that the near-electrode layers in the period of the RF field
are represented in the form of a flat condenser with leakage R l, it is
quite easy to derive an equation which links the effective value of d l
with the values measured in the experiment
ew
dl = .
(16.49)
1 / Zl2 - 1 / Rl2
435
Plasma Diagnostics
The assumption that the value d l from (16.49) is synonymous with

the actual thickness of the near-electrode layers of the spatial charge
is confirmed by the structure of the RFCD in the experiments 12 and
by the fact that the recorded form of U RF (t) is close to sinusoidal.
This indicates that the circuit is linear as a whole, irrespective of the
non-linear characteristics of each layer, as indicated by the formation
in the RFCD of a high-constant voltage between the plasma and the
electrodes (see above). Thus, although the thickness of each layer d l1
(t), d l2 (t) oscillates with time, the total value is d l1 (t) + d l2 (t) @
const.
When examining the procedural errors of the results of measure-
ments it is necessary to take into account mainly the specific features
of each of the forms of the RFCD manifested in this case by the fact
that the NLSC of the high-current RFCD like the cathode region of
the dc glow discharge, independent of the plasma column. This makes
it possible to realise the conditions in which U RF = U l, by decreasing
d. In the low-current discharge the main source of positive ions for
the layers of the special discharge is the plasma column. In other words,
these layers represent zones with non-independent active conductiv-
ity for which the Townsend criterion (16.6) is not fulfilled, in contrast
to the high-current regime. This circumstance results in an error of
determining d l using equation (16.49) (its value is too high) because
U RF in the low-current RFCD always exceeds U l due to the presence
of the RF voltage in the plasma column. To check whether URF is higher
than U l, it is necessary to measure the phase shift j between U RF and
I RF. It can easily be shown, taking into account the inequality s l <<
s pl, that at j > p/3 the relative increase is ( U RF - U l ) / U l < 015
. .
Another source of errors in determining j and d l in the low-current
discharge appears when measuring I RF where due to the low (with
partial filling of the electrode gap by the plasma in the direction nor-
mal to the current the shunting effect of the RF current, passing outside
the discharge, becomes significant. This effect can be minimised by
a smooth increase of I RF to the value at which the discharge fills the
entire electrode, i.e. S pl = pR 2el (where R el is the electrode radius).
Measurements in the dc probing circuit also have their special fea-
tures. The expression R l = U pS pl /Ip is correct only if the following con-
ditions are fulfilled: 1) passage of Ip through the discharge gap has no
effect on the structure of the near-electrode regions; 2) in the proc-
ess of burning of the RFCD no constant emf e 0 forms between the
electrodes, i.e. the discharge is a passive load for the source U p. The
experiments show that these two conditions can be violated in prac-
tice. In particular, at U p > 10 V the VAC of the probing circuit be-
436
d l × 10, cm; R l , Z l ×10 5 , Ohm cm 2 d l × 10, cm; R l , Z l ×10 5, Ohm cm 2

j, mA/cm 2 j, mA/cm 2
p, torr p, torr
Fig.16.17 Low-current RFCD in helium. 1) j (p), 2) d l (p), 3) R l (p), 4) Z l (p).
Fig.16.18 (right) Low-current RFCD in air. For symbols see Fig.16.17.
d l × 10, cm; R l , Z l ×10 5, Ohm cm 2

j, mA/cm 2 d l × 10, cm; R l , Z l ×10 5 , Ohm cm 2
j, mA/cm 2
p, torr
p, torr
Fig.16.19 High-current RFCD in helium. 1) j(p), 2) j a(p), 3) R l(p), 4) Z l (p), 5) d l(p).

Fig.16.20 High-current RFCD in air. 1) j(p), 2) j a (p), R l (p), 4) Z l (p), 5) d l (p), 6)
j(p) in discharge with non-cooled electrodes.
comes non-linear indicating the effect of the probing voltage on the

structure of the NLSC, especially in the low-current burning regime.
The decrease of Up < 5 V imposes more stringent requirements on sup-
pressing RF interference in the measuring circuits. It is almost impossible
to fulfil the second condition because even at U p = 0 the current I p
reaches, in the high-current regime, several milliammeters. This indi-
cates that a constant emf of several volts forms between the RF elec-
trodes. To exclude the procedure error, the measurement of I p in this
437
Plasma Diagnostics
R, ohm
Fig.16.21 Dependences R (1/I RF) for RFCD. 1) in

CO2, p = 30 torr; 2) in air, p = 15 torr; 3) in helium,
1/I RF, 1/A p = 150 torr.
case must be taken twice: at U p ¹ 0, and U p = 0, and in relation to

the currents recorded in this case the true value of I p is represented
by the sum, even the currents are directed against each other, or by
their difference in the opposite case.
To illustrate the possibilities of this method, Figure 16.17–16.20 show
the experimental results of measuring d l, R l, Z l, j in the low-current
(Figure 16.17 and 16.18), high-current regime (Figure 16.19, 16.20)
RFCD in helium (Figure 16.17 and 16.19) and in air (Figure 16.18 and
16.20) at a frequency of 13.6 MHz.
We shall now consider the experimental confirmation of the exist-
ence of the effect of the normal current density in both forms RFCD.
The direct method, based on measuring the cross section area of the
discharge at different currents, is often highly laborious because the
contours of the near-electrode layers, especially in the high-current
discharge, are usually far away from the regular geometrical form and,
in addition, can continuously move on the electrode surface. There-
fore, it is more efficient to use the previously examined method of active
probing of the RFCD or, more accurately, the dependence of R l on the
value reciprocal to IRF. Typical results, obtained in a high-current RFCD
at a frequency of 13.6 MHz, are presented in Fig.16.21. It can be seen
that the experimental points fit with satisfactory accuracy the straight
lines satisfying the equation R =A 0/I RF , where A 0 is a constant deter-
mined by the type of gas and electrode material (U RF for each experi-
438
ment series remained almost constant). It can therefore be assumed

that this specific characteristics of the discharge also remain constant
with increasing I RF , and the results are a consequence of the linear
dependence of S pl on I RF , i.e. in the high-current RFCD at
I RF < j n2pR 2el the values j, U l, d l are independent of I RF . This method
can be used if I RF >> I 0 (I 0 is the RF current passing flowing outside
the discharge) in the opposite case the effect of I 0 must be taken into
account.
We shall now apply the method of active probing of the RFCD in
the direction normal to the RF current. In this case, the main (RF) elec-
trodes are coated with a dielectric material, and the additional elec-
trodes are placed in the electrode gap.. To avoid taking into account
the effect of the layers of the spatial charge formed at the surface
of the probing electrodes, it is necessary to ensure that the distance
l between them varies. If the probing electrodes cannot be moved, their
number should not be less than three.
Let it be that l 12 is the distance between the first and second elec-
trodes, and l 23 is the distance between the second and third electrode,
and Dl = l 12 – l 23 ¹ 0. It is clear that at the same probing current I p,
selected in the linear section of the VAC of both probing circuits, we
can write
I12 ( I p ) ^
I p12 = ( Rl1 + Rl 2 )( I p ) ^ _ ,
s ^ S^
I 23 ( I p ) ^ (16.50)
U p 23 = ( Rl 2 + Rl 3 )( I p ) ^ _ ,
s ^ S^
Here R li are the active resistances of the near-electrode regions of the

additional electrodes, and the first and third electrode can be situated
in such a manner that Rl1 = R l3; S ^ is the cross sectional area of plasma
through which (I p) ^ flows; s ^ is the conductivity of the RF discharge
in probing in the direction normal to the RF current.
From (16.50) taking into account R l1 = R l3, we obtain
( I p ) ^ Dl
s^ = . (16.51)
S ^ (U p12 - U p 23 )
Comparison of s ^ with the active conductivity of the low-current

RFCD, obtained in probing the RFCD along the direction of the RF
current s || , under the same conditions shows that s ^/s || > 10 2 ¸ 10 3
439
Plasma Diagnostics
Fig.16.22 Electrode system for

examining the radial structure of
RFCD.
which again confirms the layer structure of the low-current RFCD

including in-series connected capacitance RFCD and the plasma col-
umn, where s l << s pl .
16.8 The radial structure of the RF capacitance discharge

In previous sections special attention was paid to the special structure
of the RFCD along the direction of passage of RF current. At the same
time, it should be noted that even in the simplest geometry of a flat
condenser, the main features, which are very distinctive but have been
studied insufficiently, are: 1) the property of the discharge to fill only
the part of the electrode gap in the direction normal to the current with
decrease of I RF below some value (the effect of the normal (minimal)
discharge current density), and the RFCD is characterised by the oc-
currence of at least two stationary states of the non-contracted plasma
column with the normal current densities j n1 (16.16) and j n2 (16.12);
2) dependence of the transverse dimension of the plasma on I RF and,
in its specific cases, on the distance to the electrode; 3) the phenomenon
of the spatial modulation of the discharge current density manifested,
for example, in the multichannel discharge structure observed in the
experiments.
Examination of these special features of the RFCD requires, in ad-
dition to using conventional probe and optical methods of diagnostics,
the development of new methods of examining its radial characteris-
tics which do not interfere with the discharge.
We shall examine in greater detail one of these procedures which
enable conclusion to be drawn on the degree of radial heterogeneity
of the stationary nature of RFCD between two flat electrodes
(Fig.16.22). One of the electrodes, representing the plating of the flat
condenser, is in the form of a circular disc assembled from closely
packed metallic rings. Each ring has no galvanic coupling with the
adjacent ones, but the potential of all the rings with respect to the RF
component of the field is the same. The latter, as indicated by Fig.16.22,
is achieved as a result of connecting each ring to the metallic surface
with zero RF potential through a condenser whose capacitance resist-
440
V 0, V
Fig.16.23 Dependence of constant

voltage V 0 (m,n) formed between two
insulated ring-shaped sections of the
electrode in low-current RFCD in air.
p = 7.5 torr; d = 0.7 (1), 1.5 (2), 3 cm
m (n= 10) (3).
ance at the working frequency can be ignored. Experiments show that

the RFCD in this case does not differ by its optical and electrical char-
acteristics from the discharge burning between the continuous elec-
trode. However, when an electrode section with respect to the con-
stant component of the field it is possible to measure the constant dif-
ference of the potential U 0 not only between the plasma and electrode12
but also between any two rings V 0 (m, n) (where m, n are the number
of the rings), and the value of V 0 (m, n) can be used to evaluate the
radial heterogeneity of the discharge.
The mechanism of formation of the link of V 0 (m, n) with the ra-
dial heterogeneity of the discharge can be described as follows. If the
RFCD in the radial direction is homogeneous, then d l and j l do not
depend on radius r. Consequently, since U l ~ j l d l, and U 0 ~ U l (see
equations (16.31) and (16.27)) U 0 is also independent of the radial co-
ordinate, i.e. V 0 (m, n). In the opposite case V 0 (m, n) ¹ 0 and is de-
termined by the degree of radial heterogeneity of the discharge.
Typical results of the experiments carried out using this procedure
are presented in Fig.16.23. Attention should be given to the following.
The degree of radial heterogeneity of the RFCD depends on the size
of the electrode gap d. With increasing d, with other conditions be-
ing equal, the heterogeneity of the discharge in the radial direction
becomes greater. In particular, the values of j l and U l in the centre
of the discharge gap and at the periphery can greatly differ. There-
fore, the experimental results which show that13,62 qualitatively different
forms of RFCD: low-current and high-current, form and co-exist in the
same discharge gap, are not surprising. It was shown previously that
the transition of the RFCD to the high-current regime takes place when
the RF voltage on the layer U l reaches the value U br at which the con-
dition (16.6) is fulfilled in the NLSC. However, according to the quali-
tative data presented in Ref.62, situations can arise in which in the
441
Plasma Diagnostics
U RF , V
3 cords
2 cords
1 cord
I RF , A
Fig.16.24 Volt–ampere characteristic of low-RFCD with appearance of cords.
vicinity of the centre of the discharge U bl > U br , whereas at some

distance from the centre U l < U br, and the RF discharge exists there
in the low-current form.
Analysis of the experimental data presented in Fig.16.23 shows that
at low d ¾ 2d l the radial heterogeneity of the low-current RFCD is
evident only at the periphery of the discharge at the boundary at the
plasma with the neutral gas. With increasing d, i.e. when a plasma
column, elongated along the current direction appears (energy in re-
lation in this plasma column forms the radial profile of the gas tem-
perature) the strength of the electric field E is redistributed in both the
longitudinal and transverse (radial) directions. This is associated with
the fact that the heating of the gas in the centre of the gap creates
more suitable conditions for the passage of the discharge current in
the near-axial zone of the RFCD plasma. However, until U l < U br in
the centre of the RFCD, i.e. the conductivity of the layers is of the
capacitance type and the current density j a is lower than j pl, increase
of d results in additional radial heterogeneity of the RFCD due to the
formation of the radial component of the field ensuring the transfer
charges from the NLSC periphery to the central part of the discharge.
They present the experimental material on the radial structure RFCD
explains the mechanism of formation of the multichannel structure in
the examined discharge. It has been noted that in a low-current RFCD
in heavy inert gases, for example, Ar, Xe, including IRF results initially
in the formation of one plasma cord where j pl >> j l, and the second
one, etc. The results show that the VAC in the examined case is non-
monotonic: the appearance of each subsequent cord is accompanied
by a decrease of the RF voltage at the electrodes. A further increase
of current results in a smooth increase of the area NLSC and U RF up
to the appearance of a new cord (see Fig.16.24). Thus, regardless of
442
the falling VAC of the plasma cord the radial component of RF volt-
age U r, ensuring the transfer of charges from the periphery of the near-
electrode zones to the region of the plasma cord and depending on r
is the increasing function of I RF which not only compensates the de-
crease of U pl but also explains the large increase of U RF . This fact
increases the strength of the electric field at the periphery of the dis-
charge to the values resulting in the ionisation balance. This results
in the formation of a new plasma cord and the characteristic values
of the r and, consequently, U r decreases. U RF also decreases in this
case (Fig.16.24). With a further increase of the I RF the process is re-
Fig.16.25 Effect of branching of the plasma column in the vicinity of NLSC in relation
to value I RF. High-current RFCD at the frequency of 13.6 MHz in a mixture of Xe 30
torr + air (5 torr); I RF = 2.8 (a), 5.7 A (b).
443
Plasma Diagnostics
peated and a multichannel ordered structure forms in the electrode gap.

Similar phenomena take place also in the high-current form of the
RFCD, i.e. in the high-current burning regime of the RF discharge the
NLSC and the plasma column cannot be regarded as completely au-
tonomous because of the high conductivity of the discharge regions -
analogues of the glow and the Faraday dark space, as assumed in
Ref.54. In fact, as shown by the examination of the transverse struc-
ture of the high-current RFCD using the method described previously,
the high-current RF discharge is also characterised by the formation
(during contraction of the plasma column) of the radial fields in the
vicinity of the electrodes due to finite values of the active resistances
of the glow discharge and the Faraday dark space.
These special features of the spatial structure of the high-current
RFCD lead to important consequences, especially the radial depend-
ence of the current density of the NLSC, the possibility of disruption
of the RF discharge in the arc regime even if the electrodes are not
completely filled with plasma because the discharge current density of
the NLSC directly below the plasma cord may greatly exceed the
normal current density j n1 at the periphery of the RFCD. The radial
fluxes of the charged particles, formed in contraction of the plasma
column of the high-current RFCD are also responsible for the phenomena
of branching of the plasma cord in the vicinity of the NLSC (see
Fig.16.25) 55 and other important special features of the RFCD.
Up to now in examining the spatial structure of the RFCD normal
to the direction of RF current it was assumed that the RF electrodes
are equipotential surfaces. However, this assumption holds only in the
case in which the characteristic size of the electrode d el << l = c/f
(where l is the wavelength, c is the speed of light). At frequency
f ~ 10 8 MHz l = 3 m, and in the standing wave regime at the elec-
trodes with the characteristic dimensions of the order of tens of cen-
timetres (this situation is typical of laser technology) there is a prob-
lem with the heterogeneity of the distribution of U RF (z) along the large
size of the electrode. There are various methods of eliminating this het-
erogeneity, especially by connecting the inductance coils at different
points of the electrode gap 56 maintaining a stationary RFCD in the trav-
elling wave regime, etc.
Analysis of the spatial feature of the RFCD makes it possible to
propose another method of equalising the plasma parameters under the
conditions of non-uniform U RF (z): it is necessary to vary the thick-
ness of the dielectric coating d (z) along the electrode in such a manner
that the density of the discharge current is independent of z, with the
given distribution U RF (z) taken into account. This can be realised as
follows: one of the surfaces of the dielectric coating of the electrode
444
facing the internal volume of the discharge chamber is flat, whereas

the opposite surface is sectioned in accordance with U RF (z), and the
surface of the electrode, coupled with the dielectric, repeats the profile
of the dielectric coating. In this case, regardless of the non-uniform
distribution of the RF voltage along the electrodes, the voltage drop
at the discharge, including U l and U pl will the same along the entire
electrode length.
16.9 Optical methods of examining the spatial structure of

the RF capacitance discharge
Examination of the nature of the inherent optical radiation of the RF
capacitance discharge is the most already available and widely used
method of examining its spatial structure.
Already simple comparison of the photographs of the RFCD shown
in Fig.16.1, regardless of the identical experiment conditions (electrode
geometry, type of gas, p, d, f, URF), shows their qualitative differences.
Measuring the spatial distribution of the integral intensity of glow along
the direction of current, we obtain quantitative data (see Fig.16.2) re-
quired in analysing the spatial structure of the discharge. In particu-
lar, the appearance of dark near-electrode zones in the low-current
RFCD whose thickness d l depends on frequency (d l ¾ 8 ¸ 12 mm
for f = 1.3 MHz, 2–4 mm for 13.6 MHz, 12 and 0.2–0.4 mm for 81
MHz,13) shows either a low concentration of electrons n e in these zones
or weak electric fields. Using the assumptions on the plasma column
oscillating in the electrode gap 18,32 it becomes clear that the absence
of glow in the near-electrode zones of the low-current RFCD is in fact
the consequence of both weak fields and low n e in the examined dis-
charge regions.
In fact, when the plasma column moves away from the instanta-
neous cathode and exposes NLSC, a high-strength electric field forms
in the layer but the electrode concentration n e there is not high. In the
following half-cycle where the plasma column approaches the electrode
acting as an instantaneous anode, the strength of electric field in the
column decreases and although the electron concentration n e is high,
the frequency of excitations of the levels of the atoms (molecules) is
not high because of weak fields. It should be noted that at the mo-
ment of time when the plasma column is situated at the minimum dis-
tance (of the order of Debye radius) from the instantaneous anode,
the strength of electric field in the entire plasma column is close to
zero because at this moment of time the speed of the plasma bound-
ary converts to zero. In subsequent moments of time the plasma bound-
ary and the entire plasma column start to move in the direction away
from the examined electrode, the strength of the field in the plasma
445
Plasma Diagnostics
will increase but the plasma column itself, i.e. plasma electrons, will
leave the near-electrode region. Thus, a situation is realised in the
vicinity of both RF electrodes in the low-current RFCD in which the
strength of the electric field and the electron concentration varies with
time in the opposite phase. This leads to the appearance of dark near-
electrode zones observed in the experiments whose thickness within
the framework of the examined mechanism should depend on the fre-
quency of the applied RF field. The weak dependence of the thick-
ness of the examined dark near-electrode zones on pressure (Fig.16.17,
16.18) can be easily explained. In fact, the characteristic thickness of
the NLSC is explained by the drift velocity of electrons in the plasma
v dr (see (16.28)) which, in turn, is a function of the reduced strength
of electric field in the plasma E pl /p, which does not change if the
mechanism of annihilation of the charges continues to operate.
At the same, according to photometric measurements of the spa-
tial distribution of the glow of the discharge in the high-current RFCD,
the thickness of the near-electrode dark zones are almost completely
independent of the frequency but are inversely proportional to the gas
pressure which is in accordance with the measurements of d l2 of the
high-current RFCD using other methods (see Fig.16.19, 16.20).
The absence of the glow in the examined near-electrode zones of
the high-current RFCD is of the same nature as in the low-current
discharge: in the phase of the instantaneous cathode in the vicinity of
the electrode surface the characteristic thickness d l2 = C 1 / p (con-
stant C 1 is determined by the type of gas and electrode material) the
value n e is relatively small whereas in the phase of the instantaneous
anode the strength of the electric field is small.
Differences in the nature of the dependences of the thicknesses of
the dark near-electrode zones on pressure and frequency of the RF
field indicate that the mechanisms of their formation in the low- and
high-current burning regimes of the RFCD differ. This is also indicated
by the results of spectral measurements: the spectral composition of
the radiation of the near-electrode regions of the high-current RFCD
coincides with the spectrum of the negative glow of the conventional
dc glow discharge burning under the same conditions, whereas the spec-
trum of the near-electrode zones of the low-current RFCD similar to
the radiation spectrum of the positive column.
The qualitatively different behaviour of d l (p, w) in low- and high-
current forms of the RFCD can be explained on the basis of the mecha-
nism proposed in Ref.12 which interprets the transition as a breakdown
of the capacitance NLSC with the secondary emission processes on
the electrodes taking place.
According to Ref.12, ionisation of the low-current RFCD by the sec-
446
ondary-emission electrons of the NLSC can be ignored because in non-

glowing near-electrode zones or the high-current discharge there is not
only intensive ionisation with electrons formed as a result of secondary-
emission processes but also the Townsend criterion is fulfilled. The char-
acteristic thickness of the NLSC is determined not by the amplitude
by the drift vibrations of the plasma electrons but by the condition of
the optimum frequency of the ionising collisions, as in the cathode region
of the dc glow discharge.
These considerations show that the additional confirmation of the
mechanism of transition between the stationary conditions of RFCD,
proposed in Ref.12, are the data on the time dependence of the ra-
diation intensity of different discharge regions. It should be noted that
the method of oscillographic recording the glow of the near-electrode
regions and the plasma of the RF capacitance discharge has been used
for a long time for analysis of the spatial structure (see, for example,
RF.19, 21, 57–60). The results show that the integral intensity of ra-
diation in the plasma column is modulated, as expected, with the double
frequency 2w of the applied RF voltage. Nevertheless, near-electrode
regions of the RF discharge showed oscillation of the glow intensity
at frequencies w and 2w, and the amplitude of oscillations of the fre-
quency w exceeded the amplitude of oscillations at the doubled fre-
quency by two orders of magnitude. 21 It is also important to note that
the glow formed in the same area of the discharge gap under the same
conditions, can appear twice during a cycle (this corresponds to the
low-current RFCD) and under different conditions only once during a
cycle (high-current form of RFCD). 59
However, when using the optical methods for determining the burning
regime of the RFCD in the frequency and pressure range where
 ω  T 
p<  C1   (see equation (16.18)) it must be remembered that
 vdr   T0 
the pulsations of the glow with the frequency w can become evident
as a result of the pressure of beams of secondary-emission electrons
also in the highly anomalous low-current RFCD where the criterion
(16.6) is not fulfilled (for more detail see Section 16.2).
 ω  T 
If p >  v  C1  T  , the application of optical methods with the time
 dr   0 
and spatial resolutions makes it possible to draw unambiguous conclu-
sions on the burning regime of the RFCD applied in the experiments,
especially if it is possible to synchronise the examined oscillograms of
the glow with the oscillograms of the discharge current. 60
The optical methods with the spatial resolution of the glow inten-
447
Plasma Diagnostics
sity of the near-electrode layers give good results in examining the

process of transition of the RFCD to the high-current burning regime
when the electrodes are coated with a thick dielectric layer, i.e. when
d > e d d l. As established in Ref.61 varying the value of d is it possi-
ble to realise the falling sections of the VAC of the stationary RFCD
(see Figure 16.3, 16.4, sections VAC DE and cd, respectively). The
dark zones in the vicinity of the electrodes coated with dielectric, which
carry information of the thickness of NLSC, are distinct and their size
varies in relation to d and U RF in the range d l1, d l2. The NLSC side
opposite to the surface of the dielectric is limited by a thin plasma layer
whose radiation spectrum differs from that of the positive column. Thus,
the coating of the electrodes with the dielectric with of different thick-
ness makes it possible to examine the process of transition of the RFCD
to the high-carrying burning regime or transformation of its structure
without using optical methods with time resolution.
16.10 Conclusions
The methods of examining the spatial structure of the RF capacitance
discharges, examined in this section, do not obviously exhaust the entire
range of diagnostic means used for examining the special feature of
the RFCD. It is evident that in measuring the parameter of the RFCD
an important role is played by the conventional method of plasma di-
agnostics: probe, microwave, optical, etc., which makes it possible to
determine the required characteristics of discharge plasma with the
sufficient accuracy and spatial–time resolution: T e, n e, T, etc. However,
the availability of only local discharge characteristics in many cases
doesn’t make it possible to give an unambiguous answer to the problem
why these and not some other values of the measured quantities are
realised in the RFCD. As shown previously, the reason for this am-
biguity is that the RFCD is characterised, during the passage of dis-
charge current in the electrode gap, by the formation of regions which
qualitatively different mechanisms of formation of conductivity in them:
near-electrode layers of the spatial discharge in the plasma column.
In particular, the process of establishment of a stable stationary state
(self-organisation) in the NLSC – plasma column system is the basis
which determines the special feature of the RFCD and stimulates the
search for the corresponding diagnostic means some of which have
already been described.
448
Wave Breakdown in Distributed Systems
Chapter 17
WAVE BREAKDOWN IN DISTRIBUTED SYSTEMS
17.1 Introduction
This chapter is concerned with the properties of the waves of elec-
tric breakdown in gases moving at the velocity close to that of light.
For comparison, we shall discuss schematically the main variants of
electric breakdown of the gas gaps.
The breakdown of short gaps between flat electrodes at moderate
pressures and strength of electric fields by an electron avalanche has
been studied quite extensively. 1 The speed of propagation of the ava-
lanche front is determined in this case by the drift speed v dr of the
electrons in an applied electric field:
v dr = µ eE
where µ e is the mobility of the electrons.

The following data mechanism is associated with the transition of
the avalanche with increasing pressure and length of the gap into a
streamer. 2 It is probable that the streamer channel is of the plasma
type, and the rate of its growth to the electrodes becomes consider-
ably higher than in the avalanche and usually equals 10 8 cm/s. The
increase of the speed of movement of the ‘head’ of the streamer to
the anode is explained by the effect of two factors. The first factor
may be the photoexcitation of the gas in front of the ‘head’ of the
streamer by the resonance radiation of the plasma channel. The re-
action of associative ionisation
A* + A ® A + + e
results in the formation of free (‘seed’) electrons. Finally, the strong

electric field of the spatial charge in the vicinity of the ‘head’ results
in a high ionisation rate. Thus, the existence of local electric fields which
are stronger than the mean field in the gap is the second factor ex-
449
Plasma Diagnostics
Fig.17.1 Geometry of the device for examining

the development of a wave breakdown. I)
slightly ionised plasma, II) dielectric, III) ideal
conducting screen, AB - electrode.
plaining the observed velocity of movement of the ‘head’ of the

streamer. 2
The avalanche (Townsend) and streamer breakdown mechanisms
do not exhaust the entire variety of the physics of electrical break-
downs of the gas gaps. From the retrospective viewpoint, it is evident
that when the length of the gap is greatly increased the distribution
of bias currents in the space becomes far more complicated which, in
turn, affects the dynamics of the ionisation front. We shall assume that
the breakdown of a gas takes place inside a tube made of a dielec-
tric surrounded by a metal screen (Fig.17.1). In other words, we are
discussing here the breakdown of the gas in systems with distributed
parameters whose suitable example is a co-axial line shown in Fig.17.1.
Of course, the special features of the breakdown of this type become
important at relatively rapid changes of the potential at the electrodes.
For example, in this case, a volume charge wave forms and moves at
the velocity close to the velocity of light.
It is important to note that the development of, for example, ball
lightning is modelled more efficiently by a breakdown in a co-axial
system than between flat electrodes in a Paschen system which is now
regarded as traditional.
It is also useful to pay attention to the large difference of the dis-
cussed wave of the electric breakdown in comparison with the wave
of microwave breakdown in the waveguides. In the latter, the energy
flux is directed against the ionisation wave.
17.2 History
In the eighties, the largest number of data on the nanosecond break-
down in the distributed system were obtained in the experiments with
long low-pressure tubes convenient for laboratory investigations. The
interest in experiments of this type appeared already in the previous
century and then periodically in the following years. 3,4 Their main aim
was to examine the nature of transfer of perturbations along the dis-
charge gap. Special attention has been paid to the propagation of rapid
gas ionisation fronts along the pipes. The high velocity v = 10 10
450
cm/s of such a front in the direction of propagation of the gas glow

front was measured by Thompson. 4 Further studies were concerned
with examining the processes accompanying this breakdown stage re-
ferred to by Fowler 5 as the wave breakdown.
In 1926, Beams confirmed 6 the high velocity of the glow wave, re-
corded by Thompson, and 10 years later Snoddy, Deitricht and Beams 7
found, placing two additional electrodes on a tube, that this movement
of the light pulse is accompanied by the wave of the potential with the
same velocity. They showed that the velocity of the ionisation wave
depends on the tube diameter, gas pressure, and the strength and polarity
of the applied potential.
The wave always started propagating from the electrode with a high
potential in relation to the air, irrespective of its polarity. The veloc-
ity at a density of 1.4·10 16 cm –3 increased almost linearly from 1.7·10 9
cm/s at 73 kV to 3.7·10 9 cm/s at 175 kV. The maximum current in the
initial wave is very high, in the range 90–200 A, i.e. current density
90–4000 A/cm 2. The voltage wave propagates through the tube with
some absorption which increases with increasing pressure. At low gas
densities in a tube 12 m long, the pulse amplitude of 194 kV decreased
at the other end to 180 kV, and at a gas density of 3.5·10 17 cm –3 to
100 kV. In their experiments, they recorded also a reversed wave
formed when the first wave reached the earth electrode. The veloc-
ity of the reversed wave was 10 10 cm/s at 132 kV. The velocity of
the reversed wave was slightly higher than the highest limited veloc-
ity of the primary wave and was almost independent of the potential
sign and the pipe diameter.
It is evident that one of the main conditions of formation of ionis-
ing waves is the rapid change of the potential of one of the electrodes
of the discharge gap. This conclusion was made in Ref.8 in examin-
ing the glow pulses in long pipes. A wave formed only at relatively
fast supply of a voltage of 20–40 kV on one of the electrodes of the
pipe. The authors of Ref.9 obtained both slow glow waves in long pipes
and fast ones. However, the fast waves appeared only at a high speed
of increase of high potentials. The required change of the electrode
potential can take place not only when the source is represented by
a pulsed voltage oscillator. In a study by Westberg, 10 perturbation of
the electric field in the vicinity of an electrode forms spontaneously
in breakdown of the oxide film on the cathode of the stationary dis-
charge. It forms when a wave travelling from another electrode ap-
proaches the given electrode, or when the voltage is removed at the
moment of discharge.
Ionising waves are observed in the same experiment at both the posi-
tive and negative polarity of the pulse but their velocity differs. 7,11,12
451
Plasma Diagnostics
As shown in Ref.10, 13, the nature of propagation of the waves is greatly

influenced by the preliminary ionisation of the gap.
The ionising waves transfer the volume charge. This is confirmed
by recording the waves of the potential along long pipes 7,14 and by the
variation of the current on the electrode at the moment of arrival of
the wave. 10
The passage of the wave through the gap is accompanied by gas
ionisation. This is indicated by the increase of radiation behind its front.
The increase of the degree of ionisation of the gas in the discharge
gap and, consequently, its transition from the non-conducting or low-
conducting (glow discharge, corona) state to an efficiently conduct-
ing arc can take place by two methods, according to Loeb. In the first
method, which is used more widely, the discharge current heats the
gas, ionisation takes place and the conductivity of the channel increases.
The rate of this gas is restricted by a velocity of the order of 10 7
cm/s of the ionising electrons intersecting the discharge gap. The second,
faster process, 10,15,16 is based on the ionising waves.
Analysing the studies published by Westerburg,10 Loeb describes the
process as follows. The initial electrical perturbation during the period
shorter than the diffusion period produces a steep potential gradient
and the gradient of the volume discharge at length l f . This is possi-
ble only at the initial concentration n 0 of free electrons. They initially
either change the volume or can be generated by photoionisation as
a result of initial perturbation. The speed of displacement of the po-
tential gradient is described by the equation
v = a v dr l f /(ln n e/n e0)
where a is the Townsend coefficient, v dr is the drift velocity of the

electrons, n e0 and n e is the initial and final concentration of the elec-
trons.
Ionisation takes place in orderless collisions of electrons with at-
oms. If the pressure is too low and the electrons have mainly direc-
tional movement, the equation cannot be used and the velocity of the
wave is determined by the velocity of the electrons moving in vacuum
with the energy which the electrons initially possessed or which is equal
to the potential drop. At a high potential, the formation of the poten-
tial gradient requires a high density of the ions and l f is considerably
lower. In this case, the pulse, generating the initial perturbation must
have steep fronts and photoionisation takes place with fronts with shorter
wavelength.
It should be noted that the movement in slow waves 17–20 formed in
the case of slow supply of a voltage through the electrode, is deter-
452
mined by electron diffusion to the walls and the subsequent charging

of the walls. However, in the case of surface waves 21,22 their move-
ment is determined completely by the change of the charges and the
fields on the side surface of the cylindrical discharge, and at large
diameters of the discharge pipes it affects a relatively small fraction
of the volume.
Summing up these results, the development of a wave breakdown
in the long low-pressure pipes can be described as follows. Irrespective
of polarity, the wave forms at the electrode with a high potential, 8 the
voltage range is 1–200 kV. The wave does not form always immedi-
ately after supplying a voltage pulse to the electrode. Sometimes, a
certain period of time is require for its formation. 23,24 The wave moves
along the pipe with a velocity of 10 8–10 10 cm/s, transferring the po-
tential 8 and causing the glow of the gas in the part of the pipe through
which the wave has travelled. The length of the pipe was changed in
the experiments from several centimetres to 10 m and longer, its di-
ameter from several millimetres 25 to 14 cm, 9,26 gas density 10 16 –3·10 18
cm –3 . The velocity of the wave increased with increasing voltage and
pipe diameter, 7 and also if the screen was removed or preliminary ioni-
sation was induced. 23 With movement of the wave in the pipe the po-
tential at the front decreased. 7,9,27 There are differences in the waves
formed by voltage pulses of positive and negative polarity. In the first
case, the current of the second electrode was non-monotonic. 28
For the majority of experiments whose results, obtained up to the
end of the eighties were summarised in Ref.29, the common measured
quantity was the speed of travel of the front wave v along the dis-
charge pipe. The dependence of the velocity of the front on gas density
v(N) has been studied most extensively. In Ref.27, v(N) was obtained
for Ar, CO 2, N 2, H 2 and air, in Ref.25 for Ne and N 2, in Ref. 9 and
16 for air, in Ref.26 for Ar, H 2 and He. In all gases, the v(N) curve
has a maximum. McGhee explains the behaviour of the v(N) curves
using the following qualitative models. 27 The electric potential at the
front in the case of positive polarisation of the voltage pulse is gen-
erated by the ions whose concentration can be relatively high. From
the estimate of the conductivity of the discharge the concentration
behind the front is 4·10 11 cm –3 and at a relatively high value of the
Townsend coefficient a this concentration can be generated by impact
ionisation at the front with a single photoelectron. The front of the wave
is characterised by intensive excitation of the gas, and primary pho-
toelectrons ahead of the front are generated by the radiation of the
discharge. Therefore, the velocity of the wave is determined by both
the efficiency of photoionisation of the gas ahead of the front and the
intensity of impact ionisation at the front. According to McGhee, the
453
Plasma Diagnostics
velocity of the wave decreases at high pressures as a result of a de-

crease of the effective length on which photoionisation takes place.
At the same time, the shift of the maximum with increasing voltage
is explained by the fact that the maximum velocity of impact ionisa-
tion is displaced to the range of high pressures.
These results relate to different experimental conditions and although
the form of the v(N) curve is the same, they cannot be compared di-
rectly. The results show that the velocity of the wave is also influenced
by the diameter of the pipe, 7 the presence or absence of a metallic
screen at the pipe, 23 and pulse polarity. 23
17.3 Transition processes

Winn proposed an explanation of the role of the screen 23 which forms
some linear capacitance C with plasma. This capacitance is charged
with a wave to the voltage of the breakdown pulse U. If the wave
has travelled the distance x, the total charge in the pipe Qx = CUx
is generated by the current I = CUv, where v is the velocity of the
wave. In Ref.23, C = 0.79 pF/cm, U = 24 kV, v = 1.9·10 9 cm/s and
I = 40 A. Thus, the screen, like the size of the cross section of the
pipe, influences the rate of formation of the volume discharge in the
plasma and, consequently, the entire process dynamics. It is obvious
that the effect of the screen cannot be completely eliminated because
in its absence the role of the screen will be played by the earthed el-
ements of the system.
Winn’s explanation is simplified. It contains, in particular, an assump-
tion according to which the linear capacitance C and the potential U
are constant along the length of the pipe, although the potential be-
hind the wave front decreased in the longitudinal direction. This de-
crease of the potential was measured in Ref.14 and its pressure de-
pendence in Ref.27.
To examine the nature of the phenomenon it is important to study
the problem of the dependence of the velocity of the wave on the
potential drop at the front U f. In Ref.7,14,30, the dependence v (U)
where U is the voltage amplitude at the electrode exciting the wave,
is close to linear, and in Suzuki’s study 16 there was a large deviation
from it.
Presenting the data on the dependence of the velocity v on the initial
concentration of the electrons n e0 ahead of the front wave, Winn in
Ref.23 published an important conclusion according to which the ionising
wave at increasing n e0 should resemble more and more the electro-
magnetic waves propagating along the coaxial transmission line. The
concentration of the electrons at which the ionised channel can be
regarded as a good conductor was determined by Winn by two methods.
454
In the first method, the experimental curve v (n e0) is extrapolated to

the value v = c, i.e. the velocity of light in vacuum. For the wave with
positive polarity, the concentration ne0, derived in this manner, is equal
10 12 cm –3 .
Extrapolation of the experimental dependences v (ne0) in Ref.16 gave
the values of the limiting concentration of the electrons n ef an order
of magnitude lower than those obtained by Winn. In addition, it was
established in Ref.16 that the velocity of the wave depends on the
duration of the front v (t f ) of the voltage pulse. Extrapolation to the
velocity of light in vacuum using this parameter gives the limiting value
dU/dt = 5·10 12 V/s.
In Ref.23, Winn determined the limiting concentration of the elec-
trons at a front n ef by examining the balance of the energy stored and
dissipated in the wave. If it is assumed that ionisation in nitrogen requires
1/3 of this energy, then from the relationship
e dis = e p - e E = IU / v - CU 2 / 2 » CU 2 / 2 = 3nef Ii S ,
where e dis , e p = IU/v, e E is the energy per unit length of the plasma
column from the electrode to the front: e dis – dissipated; e p – intro-
duced to the discharge; e E – the energy stored in the electric field
between the plasma and the screen; C is the linear capacitance; I i is
the ionisation potential of the nitrogen molecule; S is the cross sec-
tional area of the discharge pipe; v is the velocity of the wave; U is
the potential of the plasma in relation to the screen, gives n ef = 1.2·1012
cm –3 which is close to that obtained by extrapolation. Winn also es-
timated the concentration of the electrons at the front of the wave on
the basis of the experimental data for the attenuation for the poten-
tial along the pipe, assuming that the current I between the electrode
and the front in every cross section is the same and that the veloc-
ity of the electrons is equal to the drift velocity:
nef = CUv / evdr S .
This volume was 4.7·10 11 cm –3 at a nitrogen density of 7.8·10 16

cm –3. According to Winn, these values are in good agreement with the
previous estimates and the difference is explained by the fact that the
electrons, which have acquired a high energy in the strong electric field
at the front, are not in equilibrium with the field behind the front.
To determine the relationship between the electron concentrations
ahead and behind the wave front, Suzuki 16 used the experimental data
455
Plasma Diagnostics
on the jump of the current in the front and carried out calculations based
on the model of the discharge as a long condenser charged with an
ionising wave. The model was then developed further in Ref.33 and
34.
The problems of calculating the plasma parameters in the wave are
very important because there are almost no experimental studies where
the electron concentration and temperature were measured. Simulta-
neous measurements of the current jump in the front and the electron
concentration were taken in Ref.35 in examining precursors in impact
pipes. The thickness of the front, determined in the range 0–90% at
the stationary value behind the front was 2–5 cm but it can also be
low because the ‘hot’ electrons carried out ionisation even after passage
of the wave. Measurements of the electron temperature are associ-
ated with considerable difficulties. For example, the measurements of
T e in relation to the intensity of two helium lines 30 are characterised
by a large error because the population of the helium levels in the nano-
second time period is determined by the transfer of excitation32,36 which
was not taken into account by the authors.
In Ref.24, when calculating the current of the electric field in the
discharge and in comparison with the experimental data, the discharge
gap was regarded as a section of the coaxial transmission line with
the losses described by the telegraphic equations
dU/dx = –RI – LdI/dt,

dI/dx = –CdU/dt.
The signals, received from the capacitance voltage dividers, posi-

tioned along the pipe, were processed taking into account that they are
superimpositions of several waves reflected from the electrodes and
from the front as a result of the mismatch with respect to the wave
resistance in these areas. The principal difficulties of using this method
are associated with the electrodynamic description of the first wave
moving from the cathode because the coaxial transmission line in the
pipe for this wave has not yet formed along its entire length and the
wave is transferred by the moving first front. The current in the earth
anode was calculated and was found to be similar to the measured
value. The longitudinal strength of the electric field and current
I (x, t) was calculated at different phases of the discharge as the su-
perimposition of the fields and currents of all waves and this was fol-
lowed by calculating the internal parameters of the plasma: T e, n e, the
rates of excitation and population of the levels assuming that the elec-
tron velocity of the distribution function manages to follow the elec-
tric field.
456
The large number of data obtained in this work indicate that it is

important and essential to develop a similar procedure for examining
the processes taking place in ionising waves. In particular, assuming
that R, C, L are variable, the wave can be regarded as an impact wave.
All features of this wave were examined in several experiments. 31 Of
special importance is the examination of the first ionisation wave, where
the relationship of the electrical state of the electrodes with the proc-
esses inside the gap are of special interest for constructing the elec-
trodynamic model of the wave breakdown.
17.4 Generalised block diagram of experiment

The generalised block diagram of typical experiments is shown in
Fig.17.2. The oscillator 1, shaping a short high-voltage pulse, is con-
nected with one of the electrodes of the discharge pipe 3, the second
electrode is earthed through the resistance 4. To induce preliminary
ionisation in the gap, the dc source 5 and the ammeter 6 can be con-
nected in series with the pipe. The pulsed voltage of the electrode on
the side of the oscillator an the corresponding current in the circuit
in the second electrode are measured using the divider 2, the shunt
4 and the oscilloscope 9. The electric probes 7 and high-speed
photomultipliers 8 are used to measure the electrical and optical pa-
rameters of the breakdown waves, respectively. To examine the wave
breakdown, it is necessary to solve a number of procedural problems.
In Ref.43–51, the problem of the breakdown wave was solved by
synchronised measurement of currents through the electrodes of the
pipe with simultaneous measurement of the electrical parameters and
radiation of the breakdown wave in different areas of the pipe. The
accuracy of synchronisation together with the time resolution deter-
mines the scale of the spatial resolution of the examined distributed
process. To fully utilise the time resolution of the measured parameters
Fig.17.2 Generalised block diagram of experiments for investigating the wave mechanism
of breakdown in long tubes.
457
Plasma Diagnostics
of the wave during their spatial–time comparison, it is necessary to ensure

that the synchronisation of the measurements is not lower than their
time resolution.
In breakdown, the energy flux, supplied from the high-voltage os-
cillator, interacts with the gas. In addition to the incident flux, reflected
flux is always present. These fluxes, propagating in the opposite di-
rections, are superimposed onto each other, in addition to interacting
with the gases. Consequently, it is almost impossible to separate them
in recording directly at the discharge pipe. However, since the amplitude
and the phase of the fluxes do not change during their propagation along
the linear transmission line, i.e. the coaxial cable, separate recording
of the fluxes in the transmission line becomes quite easy.
The same measuring element (shunt) is used to measure the cur-
rent of the high-voltage pulse directed towards the discharge tube (in-
cident pulse) and, after some required for the displacement of the pulse
along the cable from the shunt to the tube and back, the current of
the pulse directed to the opposite side (reflected pulse) is measured.
The amplitude-phase differences between the incident and reflected
pulses are determined by the interaction of the measured energy fluxes
with a gas. Efficient electromagnetic matching of the cable–electronic
section - discharge volume circuit is essential.
To examine the structure of the wave, the dimensions of the dis-
charge tube should be selected in accordance with the time resolution
of the apparatus and procedure used. It should be noted that the au-
thors of Ref.40 and 41, who took synchronised measurements, worked
with gaps not exceeding 4 cm, and the time resolution obtained in Ref.41
was 1–2 ns which did not make it possible to examine the spatial struc-
ture of the breakdown wave. In fact, its spatial scale of resolution of
the measurements determined as vDt (v is the speed of the object, Dt
is the time resolution) at a breakdown wave speed of 4·10 9 cm/s is
greater than the width of the gap. 41 In Ref.23, the breakdown wave
was investigated in relatively long tubes so that the spatial scale of
resolution, corresponding to the time resolution of the order of 10 ns,
was close to the length of the tube but an order of magnitude greater
than its diameter. It is evident that the structure of the breakdown wave
should be examined at the spatial scale of resolution close to the trans-
verse dimension of the discharge tube.
When examining the structure of the breakdown wave using capaci-
tance devices, an important factor in addition to the time resolution is
the presence of the directional diagram of the capacitance divider. The
width of this diagram is equal to approximately the diameter of the
screen jacket. The spatial resolution of the sensor can be improved
by mathematical processing which was used in Ref.49 on the basis of
458
the assumption that the scale of the spatial–time non-reproducibility of

the breakdown wave is considerably smaller than the resolution of
experiment. It was thus possible to examine the evolution of a wave
with a spatial resolution of ¾ 1 cm at a screen diameter of ¾ 10 cm.
The time resolution and synchronisation of measurements of the
wave current obtained in Ref.43–51 were equal to 0.25 ns which at
a speed of the wave of 10 10 cm/s corresponds to a spatial scale of
2.5 cm (the diameter of the discharge tube 4.5 cm). The synchroni-
sation of the electrical and photoelectrical measurements is 0.55 ns which
at the same speed of the wave corresponds to a scale of ¾ 5 cm.
The breakdown was organised in a long (l >> d) discharge tube sur-
rounded by a metallic screen to form a controlled linear capacitance
of the discharge channel. Figure 17.3 shows the discharge device used
in Ref.43–51. The discharge gap represents a uniform coaxial line, and
the break of the central strand of the line includes a glass tube with
a length of 400 mm and an internal diameter 45 mm. The ends of the
tube are closed with flat molybdenum electrodes 50 mm in diameter
which smoothly transfer to the central strand of the coaxial cable, and
the outer screen smoothly changes to the braiding of the cable. The
region of smooth transition from the cable to the electrode was pro-
duced in such a manner that the unit electrical parameters were uni-
form and corresponded to an impedance of 50 W. In the screen along
the entire length of the gap there is a slit whose width is equal to the
diameter of the discharge tube. The slit can be closed by a moving
metallic sheet which carries capacitance sensors and window for a
narrowly-directed photodetector. A high-voltage nanosecond genera-
tor based on ferrite forming lines is used. The main advantage of this
device is the high stability of generated pulses with the possibility of
Fig.17.3 Discharge device. 1) cable RK-50-11-13, 2) internal electrode casing, 3) insulator,

4) outer electrode casing, 5) molybdenum electrode, 6) glass discharge tube, 7) screen,
8) capilliary tube with a pin for connecting the vacuum system.
459
Plasma Diagnostics
operating with the repetition frequency of up to hundreds of pulses per

second.
The pressure range of the working gas (helium) included the range
of the maximum value of the speed at which the electrodynamic ef-
fects should be most evident. The initial ionisation was produced in the
tube and also increased the speed of the wave; at the same time, the
stability of propagation of the wave increased.
To determine the power or other energy characteristics of the break-
down, it is necessary to take independent measurements of the cur-
rent and voltage at electrodes in the discharge gap. The independence
of the measurements is equivalent here to the measurement of the phase
shift between current and voltage.
For the waves propagating along the cable, the currents flowing
through the given cross section of the cable in the braiding and the
central conductor are equal to each other. The current is measured
using a shunt connected to the break of the screen in the braiding of
the coaxial cable. The discharge tube, connected to the break in the
central conductor, causes a heterogeneity in the uniform cable trans-
mission line. This leads to the reflection of electromagnetic waves which,
after reflection, interfere with the incident waves.
For separate recording of the current of the incident and reflected
waves using a single shunt, the position of the shunt in the cable is
selected from the condition
l > Tc/2e ½
where l is the length of the cable between the shunt and the discharge
tube, T is the duration of the electromagnetic pulse, c is the speed of
light, e is the dielectric constant of the filler of the cable.
For the waves propagating in the cable the voltage between the cen-
tral conductor and the braiding is unambiguously linked with the in-
tensity of the current flowing to the given section: 42 U = IZ. If the
direction of current in the central conductor of the cable coincides with
the direction of propagation of the electromagnetic wave, the central
conductor in the given cross section of the cable has a positive po-
tential in relation to the screening braiding. The current and voltage
in the incident and reflected waves are linearly added up in the ca-
ble in the interference zone. To determine the absolute values of the
current and voltage in the cross section of the interference zone and,
in particular, at the electrode of the discharge tube, it is sufficient to
measure (using the shunt) separately the currents of the incident and
reflected waves and also the time delay between them.
460
The measurements of the currents of the incident and reflected

waves are independent measurements and, consequently, the current
and voltage of the central conductor in relation to the screen in the
interference zone calculated from the equation
I = I in + I ref ,
U = I in - I ref Z0 ,
are also independent quantities.

It should be noted that the voltage between the central conductor
and the braiding of the cable can be measured using the capacitance
device built into the cable. However, such a divider requires amplitude
calibration whose error should not exceed the error of calibration of
the shunts. As a result, the total relative error in measurements of current
and voltage would be higher than when calculating the voltage through
the measured current and the wave resistance of the cable. In addi-
tion, the use of capacitance dividers results in an additional error as-
sociated with the fact that their width of the band of working frequencies
is considerably smaller than that of the band of the working frequencies
of the shunt.
In Ref.43–51, when determining the spatial–time correlation of the
electrodynamic quantities inside the gap and at its electrodes, all the
measurements were mutually synchronised. Consequently, for each
moment of time during the breakdown it was possible to show the
voltage and current at the electrodes of the gap, the distribution of the
excess charge and the plasma radiation intensity along the gap. The
time resolution obtained in the experiments of the mutual synchroni-
sation of the electrical measurements were 0.25 ns.
The measurements in Ref.50 were taken in repeated conditions in
operation of the nanosecond generator with a frequency of 80 Hz. The
higher reducibility of the breakdown process enabled all the signals to
be recorded with a single oscilloscope. Mutual synchronisation of the
measured quantities was ensured by connecting the corresponding delay
lines in the synchronisation unit.
The start of counting the time on the oscillograms is the same for
all signals if the electrical length of the cables are synchronised from
the rules formulated in Ref.50. When the rules are adhered to, the time
shifts between the measured signals, counted directly from the oscil-
lograms, are determined only by the breakdown process and the du-
ration of propagation of the electromagnetic wave from its excitation
point to the sensor.
461
Plasma Diagnostics
To eliminate the inaccuracies when determining the electrical length

of the cables caused by processing the effects, the accuracy of mu-
tual synchronisation of electrical signals was verified by experiments.
I inc (incident) and I ref (reflected) were synchronised by combining
their oscillograms on the screen. For this purpose, the signal Iref, having
the reversed polarity, was inverted by changing the polarity of connect-
ing the cable to the shunt. The discharge volume was filled with air
to the atmospheric pressure, no breakdown developed and the shape
of the reflected pulse coincided with that of the incident pulse.
Itrans was synchronised with Iinc also by combining their oscillograms
on the screen. The discharge tube was removed and the cathode and
anode plains were tightly pressed together.
The signal from the capacitance sensor was also synchronised with
I ref by combining their oscillograms. The sensor was placed opposite
the anode and the discharge volume was replaced with a metallic tube
so that a uniform line was obtained instead of the gap.
Recording radiation
In Ref.43–51, the amplitude dynamics of integral radiation during each
pulse was recorded with a 14ELU-FS high-speed photodetector which
could be placed against any cross section of the gap. The time reso-
lution of the photodetector was calibrated with a nanosecond surge of
an LD-15 semiconductor laser. The duration of the leading edge of
radiation of this laser, measured with an HSD-1850 photodiode, was
0.5 ns. The minimum front of the signal, obtained from the calibrated
photodetector, was 1.3 ns.
For comparing the spatial dynamics of radiation discharge with volt-
age on the electrodes in relation to time, it is necessary to know the
time of flight of the electrons in the photodetector (the delay time of
the signal in the device). The error in determining this parameter in
the experiment was less 1 ns. The available methods 52,53 of measur-
ing the time of flight give an error of 2–3 ns. Therefore, a procedure
was developed and the time of flight of 14ELU-FS was measured with
an error of ±0.3 ns. 54 The method is based on comparing the delay
of the nanosecond signal in two photodetectors where the accuracy
of the time of flight of one of them (reference photodetector) does not
exceed the required accuracy for the other one. The time of flight of
HSD-1850 vacuum photodiode used as a reference photodetector does
not exceed 10 –10 s. The light source was a LD-15 semiconductor la-
ser with a surge power of 6 W and a duration of 4 ns. The block dia-
gram of the measurements is shown in Fig.17.4. The generator of
nanosecond pulses (GNP) (120 A, pulse front 2 ns, half-width 4 ns)
excites the semiconductor laser – emitter (LD-15). The reference
462
Synchropulse
Fig.17.4 Diagram of device 14ELU-FS for measuring time of flight. 1) generator of

nanosecond pulses, 2) laser semiconductor emitter, 3) HSD-1850 vacuum photodiode,
4) measuring photodetector, 5) separating capacitances, 6) oscilloscope, 7) two-coordinate
automatic recording device.
photodetector – vacuum photodiode (HSD-1850) has a time resolution

of 0.1 ns and a time of flight of 0.5 ns. The signals from the diode
and from the 14ELU-FS system were recorded in a two-channel strobing
oscilloscope (SAS-500-9B) and then displayed in a two-channel automatic
recording device (WX-431). The error in measuring the time range in-
troduced by this automatic recording device did not exceed 50 ps. The
dividing capacitance (DC) (TBPD) in the upper transmission band of
1GHz transferred the measured pulses with their distortions. The os-
cilloscope was started up by synchropulses from the the GNP. The time
instability of the electronic excitation pulse in relation to the synchropulse
was less than 50 ps and that of the signals of the VD photodiode and
14 ELU-FS was 100 ps.
The delay between the maximum of the signals from the reference
and examined photodetectors was measured. The shapes of the pulses
were also compared (Fig.17.5). The difference in the shape was caused
by distortion of the signal of the transmission characteristic of the
photodetector. To reduce the error in determining the time ranges caused
by the non-linear sweep of the oscilloscope, the signals were placed
as close as possible to each other on the time axes by selecting the
length of the cables connecting the photodetectors with the oscilloscope.
In this case (Fig.17.5), the start of deflection of the signals from the
zero line and the time co-ordinates of the maxima coincided with an
error of 0.2 ns. Thus, the measurement of the time of flight is reduced
to measuring the electric length of the cables connecting the
photodetectors with the oscilloscope. The measured time of flight for
14ELU-FS was equal 9 ± 0.4 ns at a certified feed regime (the voltage
on the photocathode –2.6 kV, the voltage on the collector +1 kV).
The dependence of the time of flight on the feed voltage was not
measured because the certified feed regime corresponds to the maximum
time resolution of the device. The time resolution of the photodetector
463
Plasma Diagnostics
t, ns
Fig.17.5 Synchronised oscillograms. a) signals from 14ELU-FS photodetector, b) from

the vacuum photodiode, c) pumping current of the semiconductor laser (relative units).
in this studied was determined for the minimum duration of the front
of the signal.
Electro-optical recording
The spatial dynamics of the radiation of the breakdown was examined
in detail in Ref.55–57 using the electronic–optical chamber (EOC) based
on the electro-optical converter (EOC) of the UMI-93Sh type which
operated in the regime of single-frame recording with an exposure time
of 1.5 ns. The moment of opening of the gate of the chamber was
accurately (t < 0.1 ns) synchronised with each electric pulse supplied
to the examined discharge gap. Their results show that the reproducibility
of the process is very high and, consequently, in operation with a pulse
repetition frequency of 80 pulses/s the screen of the EOC showed a
stationary image corresponding to a specific stage of development of
the breakdown whose duration was equal to the exposure time of the
individual frame, i.e. 1.5 ns. When the delay between the moment of
opening the gate of the chamber and the moment of arrival of the pulse
on the electrodes of the gap on the screen of the EOC was changed
this was followed by a sequence of events in the discharge tube. The
image from the screen of the EOC was transferred to film.
The nanosecond control circuit of the EOC is shown in Fig.17.6.
The pulse generated by the oscillator 1 is divided into two pulses in
the T-junction box 2. Pulse amplitude 16 kV, duration 20 ns. One pulse
is directed to the examined discharge gap 9 and carries out its break-
down. Part of the second pulse, taken from the divider 3, is directed
464
Fig.17.6 Nanosecond circuit of controlling the electro-optical chamber.
to the special plasma shape 4, 48 which forms a triangular pulse with

an amplitude of 2 kV and a duration at the half height of 1.5 ns. Sub-
sequently, passing through the dividing capacitance 5, the pulse is di-
rected on to the plates of the compensated gate of the EOC and opens
it. The bias potential, closing the gate, is generated by the source 6.
Line 7 is used to synchronise the compensating place with the clos-
ing plates. The length of the line 7 and the damping resistors R 1 and
R 2 are selected for total compensation separately for each EOP. To
avoid undesirable reflections of the closing pulse in the transmission
line, the pulse is directed to the matched load through the capacitance
7 where it is absorbed. The line 10 is used to remove the parasitic
image from the screen formed as a result of the presence, in the force
lines, of parasitic reflected pulses which cause false triggering of the
gate.
The exposure time of the individual frame of the EOC was deter-
mined by experiments. For this purpose, in operation of the gate, the
spot image of the transient of the LD-15 semiconductor laser devel-
oped with time on the screen of the EOC. The duration of the tran-
sient smoothly decreased to the detection of a decrease of the length
of the image of the point developed into a line on both sides. The du-
ration of the transient of the laser was then measured using a HSD-
1850 vacuum photodiode with a time resolution of 10–10 s. The measured
duration of the transient was equated with the time of exposure of the
frame of the EOC.
465
Plasma Diagnostics
Measurement of electric parameters

Nanosecond current pulses were measured in Ref.43–51 in a cable us-
ing wide-band shunts which have the form a ring of titanium approxi-
mately 50 µm thick connected into a break in the cable braiding. The
active resistance is equal to hundredths of an ohm, and the reactive
linear parameters are almost identical with those of the cable. Such
a shunt has only a minimum effect on the measured signal and removes
only a small amount of power from it. The upper boundary frequency
of the working band of the shunt was estimated from the minimum du-
ration of the front of the pulse propagating from the cable in which
the breakdown of the gap with overvoltage between the braiding and
the central strand of the cable took place. The minimum duration of
the dip was 0.5 ns. The amplitude calibration of the shunts was car-
ried out by closing one end of the cable charged to voltage U whose
braiding included a shunt. In discharge of the cable the current passing
through the shunt was expressed in the form I = U/Z, where Z is the
impedance of the cable.
The dynamics of the excess charge, formed in the gap during the
breakdown process, was recorded in Ref.43, 45, 51 using a capaci-
tance sensor placed in the screening jacket of the gap on a moving
sheet which travelled along its length. To increase the division coef-
ficient and expand the working frequency band, ceramic condensers
of the KM type with a total capacitance of C = 3000 pF were placed
between the sheet of the capacitance sensor and the screen of the gap.
The time constant of the sensor with the cable load with an imped-
ance of 50 ohm connected to it is 150 ns which is another magnitude
higher than the duration of the examined processes. The signal from
the sensor is proportional to the potential of the plasma in the tube in
the zone situated opposite the sensor. The quality of the sensor was
verified in passing pulses with short fronts through the gap which in
this case was replaced with a section of the metallic tube in such a
manner that a uniform coaxial line was produced. The signals from the
capacitance sensor almost completely coincided with the signals taken
from the shunts. In addition to the wide-band sensor, a narrow-band
sensor in the form of a pin 3 mm long was also used. The signal from
this sensor was proportional to the rate of variation of the potential
of the plasma situated opposite the sensor.
Recording x radiation
In Ref.61, x radiation was recorded using a detector consisting of a
scintillator–converter of radiation and an FEU-87 photoelectronic mul-
tiplier. Plastic scintillators based on POPOP with a scintillation time
466
no longer than 2 ns and the radiation range of 360–380 nm were used,

together with an SPS B-151 high-speed scintillator. The thickness of
the converters was 1 and 6 cm, respectively, and they absorbed al-
most the entire x radiation in the examined quantum energy range. The
side surface of the converters was painted black to reduce the effect
of multiply reflected photons. The duration of increase of the signal
of the photoelectronic multiplier according to the certificate data is not
greater than 2.0 ns, the duration of the output pulse at an infinitely short
input pulse is 5 ns, multiplication factors 3·10 7. The spectral sensitivity
of the photocathode is 300–600 nm. The signals of all sensors were
supplied to the S7-19 high-speed oscilloscope (5 GHz band) which was
synchronised with the high-voltage breakdown pulse.
To measure the spectrum of x radiation, filters/absorbers of different
thickness were placed in front of the detector. In this case, the sig-
nal of the detector I is the function of thickness d i of both the param-
eter of the required spectral density of x radiation I E and the trans-
fer function of the detector j(E) which takes into account the absorp-
tion of x radiation in the windows and the dependence of the yield of
light quanta from the scintillator on the energy E of the incident x-ray
quanta
E1
I (d i ) = A I E ϕ( E ) exp {−µ( E ) di }dE,
∫
0
where A is a constant, E1 is the maximum energy of x radiation quanta,

µ (E) is the coefficient of absorption of x radiation in the material of
the absorber which depends on the energy of the quanta E.
This equation is an integral Fredholm equation of the first kind in
relation to the required function I E. It’s solution represents an incor-
rect problem because it is unstable in relation to the small changes of
the parameters included in this equation. 59 Therefore, it is necessary
either to make assumptions regarding the type of spectrum or reduce
the problem to a correct one using the regularisation procedure.
A method of obtaining information on the spectrum of x radiation
from the attenuation curve is the reduction of the integral equation to
an algebraic one. For this purpose, taking into account the physical con-
siderations, it is necessary to make some assumption regarding the type
of the required function. Subsequently, the function of this type with
free parameters is substituted into the equation and analytical or nu-
merical integration is carried out. This gives an algebraic equation (or
a system of equations in the case of numerical integration) which is
467
Plasma Diagnostics
then used to determine the values of free parameters at which the

function of a given type satisfies the equation. For the x radiation in
the majority of cases an assumption is made on the purely braking nature
of the radiation spectrum. This procedure was used in, for example,
Ref.60, 61 for measuring the electron energy in almost relativistic pulsed
beams of short duration.
It is assumed that
I E = A 2 (E 1 – E)
i.e. x radiation is of purely braking nature with the minimum energy

of quanta E 1. 62 Therefore, as a result of numerical integration with the
above equation taken into account, we obtain a series of attenuation
curves for the given set of the absorbers with thickness d i and dif-
ferent energies E. Comparing the calculated curves with the experi-
mentally determined attenuation curve (which in the co-ordinates ln [I
(d i )] and d i has the form of a straight line) we select the values at
which the calculated curves coincide within the experiment error range
(3%).
This method, which is the simplest and most readily available, is used
widely. However, it requires selecting the type of function IE, and gives
limited information on the radiation characteristics, for example, only
the value of the maximum energy of the x radiation quanta, as in Ref.61.
The type of the x radiation spectrum IE can also be determined directly
as a solution of the integral equation. For the soft part of the spec-
trum of x radiation (E << 12 kV) this was carried out in Ref.70.
Recording laser radiation

The following method was developed for examining the spatial–time
dynamics of the breakdown wave and radiation waves in a nitrogen
laser in Ref.63. To measure the distribution of the intensity of radia-
tion along the discharge, the discharge tube contains inside eight semi-
transparent mirrors for transferring the laser radiation from the active
medium. The mirrors are made of cover glass for microscopes 150 µm
thick. They are positioned under an angle of 45° to the axis of the tube,
were situated in the same longitudinal section of the tube and occupy
the area equal to 11% of the cross section area of the tube. The mirrors
were oriented alternately in the opposite size to avoid the transverse
displacement of the radiation in passage through the mirrors. The trans-
mission factor of the mirrors was measured in the spectrophotometer
at λ = 337.1 nm and equalled 97%.
The radiation from the mirrors was recorded using an HSD-1850
photoelement with an increased time of 0.1 ns. The signals from the
468
dividers and the photoelement were supplied to a Tektronix-519 oscil-

loscope (band 1 GHz). The accuracy of measuring the time ranges was
increased using a calibrating sinusoid with a frequency of 1 GHz.
17.5 Spatial–time structure of breakdown waves. Effect of

preliminary ionisation
The effect of preliminary ionisation of the speed of the breakdown wave
was investigated in Ref.10,16,23,64. In Refs.10 and 23, it was reported
that the speed of the breakdown waves monotonically increases with
increasing electron concentration, and the speed of the waves with posi-
tive polarity is more sensitive to the changes in the electron concen-
tration. Here, we shall present the results of Ref.64, where the effect
of the electron concentration was investigated at different pressures.
Comparison of Fig.17.7 and 17.8 shows that the speed of the break-
down waves in relation to p and n e0 shows on the whole the same be-
haviour at both polarities of the high-voltage pulse. The main differ-
ence is that the speed of the wave at positive polarity is on average
30% higher than the speed of the wave at negative polarity. At an
electron concentration of n e0 » (3 ¸ 5)×10 8 cm –3, the speed of the wave
shows a sharp jump at both polarities. At positive polarity, this jump
is clearly evident at pressures high than 5 torr, and at negative polarity
at pressures 10 torr. In the range of the variation of the parameters
3·10 8 cm –3 < n e0 < 5·10 8 m –3 and 5 torr < p < 10 torr, where there
v, 10 9 cm/s v, 10 9 cm/s
P,
m P,
m
H 3 m
g cm m 3
, H m
ne 0 g ,c
ne 0
Fig.17.7 Dependence of the speed of breakdown wave v on the initial concentration of

electrons n e0 and helium pressure p at the negative polarity of the breakdown pulse.
Fig.17.8 Dependence of the speed of the breakdown wave v on the initial concentration
of electrons n e0 and helium pressure p at the positive polarity of the breakdown pulse.
469
Plasma Diagnostics
Fig.17.9 Photographs of a wave breakdown in a gap 400 mm long, 45 mm in diameter.

Exposure time of a single frame 1.5 ns, time shift between the frames 1 ns.
is a jump in the speed for both pulse polarities, there was an anomalous
peak of the speed to 5·10 9 cm/s for the breakdown waves initiated by
the positive polarity pulse.
Figure 17.9 shows the evolution of the form of the emitting volume
of the breakdown wave obtained using an electronic-optical camera.
The shape of the front of the breakdown wave depends on the initial
electron concentration. At n e0 < 3·10 8 cm –3 the radiation front of the
breakdown wave is flat, and the wave is of the volume type with the
uniform glow of the discharge volume behind the front. When n e0 ex-
ceeds (3 ¸ 4)×10 8 cm –3 , the radiation front of the breakdown wave
becomes conical and extends during movement of the breakdown wave
through the gap. The emitting volume behind this front of the wave
is as previously uniform.
Detailed electron optical investigations of the spatial structure of the
breakdown wave were carried out in Ref.65–69. The experiments show
that the glow of the gas caused by the breakdown wave assumes various
forms: conical, uniform through the cross section of the discharge tube
and boundary. Each form exists in a specific gas pressure range and
smoothly changes to another with changing pressure. The results show
that there is a correspondence between the nature of the dependence
470
of the speed of the wave on pressure and the form of glow of the
breakdown wave. Regardless of the polarity of the breakdown pulse,
in the pressure range where the rate increases with increasing pres-
sure, the glow has the form uniform throughout the cross section of
the discharge tube. In the region corresponding to the falling part of
the dependence of speed on pressure, the wave propagates in the near-
wall region of the discharge tube. For the positive wave of the break-
down, in the low-pressure region, the falling part of the dependence
v (p) was recorded for the first time. The conical shape at positive
polarity of the breakdown pulse is related to the falling part of the v
(p) curve in the range of low pressures (p £ 3 torr). 68, 69
The formation of the breakdown wave at negative polarity of the
breakdown pulse at all pressures starts with the formation of a dark
near-cathode region whose thickness decreases with increasing gas
pressure in the tube. As regards the external appearance, it corresponds
to the Crooks dark space with a sharp boundary of negative glow. Several
nanoseconds after the formation of the dark region, a spot forms on
the cathode and a new glow wave starts to move along the tube.
17.6 Electrodynamic processes in wave breakdown

Absorption of energy, energy parameters of the breakdown wave
Synchronised measurements of the electric current in the electrodes
of the gap with a wave mechanism of its breakdown make it possi-
ble to obtain the dependence of electric energy parameters of the dis-
charge on time. 64
The electrode receiving a high-voltage nanosecond pulse will be
referred to as the anode (irrespective of pulse polarity) in accordance
with the polarity of the voltage of the stationary glow discharge.
The current flowing into the discharge tube on the side of the an-
ode Ia (anode current) is determined from the oscillograms of the pulses
of incident and reflected current measured separately in the cable and
synchronised at the anode
I a (t) = I in (t) – I ref (t).
Cathode current I c is measured in the coaxial cable connected di-

rectly behind the cathode (second electrode). The voltage on the an-
ode consists of the sum of the voltages of the incident and reflected
currents and, since the anode is part of the linear transmission line with
the impedance Z = 50 ohm, equal to the impedance of the conduct-
ing cable, the voltage in it is calculated from the equation
471
Plasma Diagnostics
U a (t) = [I im (t) + I ref (t)] Z.
The voltage in the second electrode is determined only by one current

I c and impedance of the electrode which is also equal to Z:
U c (t) = I c (t) Z.
The electrical power in the gap is calculated from the equation
P (t) = U a (t) I a (t) – U k (t) I k (t),
where the first term is the power supplied into the gap from the an-
ode and the second term is the power transferred from the cathode
to the cable.
The total energy introduced into the gap is determined by the in-
tegration of the power flux
z
t
E0 = P(t ) dt.
0
The excess charge, built up in the gap, is computed by integrating

the difference between the anode and cathode currents with respect
to time
z
t
Q(t ) = Ia (t ) - Ik (t ) dt.
0
The capacitance of the gap is determined assuming that the charge

behind the front of the breakdown wave is distributed more uniformly
along the length of the tube, and the plasma potential behind the front
is equal to the voltage along the anode:
C (t) = Q (t)/U a(t).
The energy of the charged gap, stored in the electric field, is computed
from the equation
E E (t ) = Q (t ) U a (t ) / 2.
472
The difference between E 0 and E E consists of the energy stored in the

magnetic field of the current E m, flowing behind the wave front, and
the energy dissipated in the discharge E d
E 0 – E E = E m + E d.
The input impedance of the gap, characterising matching of the ca-

ble with the discharge, is calculated as the ratio of the voltage and
anode current
Z in = U a (t)/I a (t).
The time dependences of the electric energy parameters have a com-

mon time axis and, consequently, as a result excluding the time as a
parameter, we obtain the dynamic characteristic dependences of electric
energy parameters on each other: volt–ampere characteristic, ampere–
Coulomb characteristic, the relationship of the discharge energy stored
in different types.
Here, we present the data for a discharge tube 400 mm long whose
cathode is connected in series with an inductance which increases the
impedance of the cathode to more than 10 kW. For pulses of the na-
nosecond duration, this circuit of the cathode is close to break because
the leakage of energy from the gap to the cable behind the cathode
does not exceed Z/Zk = 0.5%. This greatly simplified the general pattern
of the wave breakdown of the gap, excluding the individual stages, whose
formation is strongly influenced by the cathode having an impedance
close to the impedance of the tube and the cable. This complicates
examination of the breakdown wave. Calculations of the electric en-
ergy parameters are also simplified because from the equations described
previously we exclude the terms relating to the cathode.
The experiment described in Ref.64 were carried out at a helium
pressure in the tube of 15 torr at three values of the initial concen-
tration of the electrons n e0 = 2.9·10 8; 1.6·10 9; 4.3·10 9 cm –3 at the posi-
tive and negative polarities of the nanosecond pulse.
Figure 17.10 shows the oscillograms of the incident and reflected
pulses (n e0 = 1.6·10 9 cm –3). The amplitude of the reflected pulse is
shorter and the duration longer than in the incident pulse. Consequently,
the voltage is applied to the anode for a longer time than the dura-
tion of the incident pulse. Figure 17.9 shows the current and voltage
curves measured on the anode. The current oscillates and consists of
two half cycles. In the positive half cycle, the charge flows into the
tube and then, when the incident pulse decreases, flows from the tube
back into the cable. The variation of the charge in the tube is shown
473
Plasma Diagnostics
I, A
Q, µC
t, ns t, ns
Fig.17.10 Oscillograms of pulses of incident (1) and reflected (2) pulses.

Fig.17.11 (right) Dependence of the electrical discharge Q in the discharge tube on
time t.
ε , mJ
P, MW
t, ns
t, ns
Fig.17.12 Dynamics of electrical power P introduced into the discharge tube.

Fig.17.13 (right) Dynamics of energy in the discharge gap.
Z, ohm C, pF
t, ns t, ns
Fig.17.14 Variation of input impedance Z in on the gap.

Fig.17.15 (right) Variation of the capacitance C of the gap.
474
in Fig.17.11. Figure 17.12 shows the dynamics of the electric power

supplied to the tube. The oscillating nature of the graph was the change
of the sign indicates the large fraction of the reactive power in the
gap. The maximum power supplied to the tube is 7.5 MW which reaches
60% of the maximum power of the nanosecond pulse. The power has
this value when the wave front reaches the second electrode. Subse-
quently, the power supplied to the tube decreases and changes its sign.
In this case, the tube operates as an oscillator. The maximum power
transferred to the cable from the tube is 3 MW which equals 40% of
the maximum value of the power supplied to the discharge.
Figure 17.13 shows the dynamics of the energy in the gap. e 0 is the
energy supplied to the discharge, e E is the energy stored in the field
of the excess charge. Both values initially increase, and then with ending
the pulsed energy of the electric field decreases to almost zero indi-
cating the almost complete compensation of the excess discharge in
the tube; the energy e 0, introduced into the gap, decreases only to 65%
of the maximum value.
The maximum of the graph of total energy is shifted to the right
in relation to the maximum of the electrostatic energy. It may there-
fore be assumed that in addition to the reactive component of the energy
in the gap there is also a large dissipative component associated with
the passage of current. The graph of the difference between e 0 and
e E shown in the same figure, indicates that the dissipated energy in-
creases with increasing electronic energy in the gap. When the electric
energy in the gap starts to decrease, the increase of the dissipated
energy slows down and at the end of the pulse the dependence of the
dissipated energy on time reaches a plateau. Consequently, it can be
concluded that the dissipation in the gap takes place far more efficiently
when energy is added into the gap then in the case of generation of
the energy stored in the electric field of the excess charge. Thus, the
main contribution of the power to the discharge is observed during
movement of the breakdown wave of the gap. The input resistance of
the discharge is shown in Fig.17.14. During the period of time whose
duration is determined by the ‘induction time’ of the charge t i, Z in is
infinitely high and then with the start of movement of the breakdown
wave Z in and rapidly decreases to the value of 300 ohm, and when
the wave reaches the cathode Zin rapidly decreases. Subsequently, when
the current of the anode becomes negative, Z in rapidly decreases to
150 W.
Figure 17.15 shows the variation of the capacitance of the gap. Dur-
ing movement of the breakdown wave the capacitance linearly in-
creases. When the front of the breakdown wave reaches the cathode
the capacitance equals 14 pF which equals 60% of the capacitance
475
Plasma Diagnostics
of the tube filled with the metal.

Figure 17.16 shows the volt–ampere characteristic of the gap for
a wave breakdown. The VAC has a hysteresis. The VAC starts at some
voltage, in this case 7 kV, and then rises upwards with an almost con-
stant angle of inclination corresponding to a resistance of approximately
300 ohm; reaching the upper point, the characteristic is then reversed
and continues to the region of negative current with an angle corre-
sponding to a differential resistance of 50 W. The VAC then returns
along a circle to the original co-ordinates.
Figure 17.17 shows the dependence of the energy e E, stored in the
electric field, and the energy e d = e 0 – e E, dissipated in the gap. The
dependence shows three sections characterised by the linear relationship
between e d and e E. In the first section, corresponding to the movement
of the breakdown wave through the gap, e E is 1.6 times higher than
e d. The second section corresponds to the state in which the break-
down wave has already closed the gap but the gap still receives the
energy with constantly decreasing current from the anode which is at
an almost constant potential. In this section, the increase of energy e E
is equal to the increase of e d. In the third section, relating to the stage
of reversed anode current, e E rapidly decreases to zero and the en-
ergy e d slight increases. The ratio of the changes of the energy in this
section e E/e d » 5.
The limiting speed of the breakdown wave, attenuation. The
authors of Ref.72 carried out experiments with voltage pulses of nega-
tive polarity with an amplitude of up to 250 kV, duration 9 ns, with
increase and decrease fronts of 2.5 ns.
The dependences of the speed of the breakdown wave, the amplitude
εE, m J
U a , kV
I a, A
ε 0 εE , mJ
Fig.17.16 Dynamic volt–ampere characteristic of the gap in wave breakdown. U A , I a,

voltage and current in the high-voltage electrode.
Fig.17.17 (right) Relationship of energy, stored in the electric field of the breakdown
wave, with the energy dissipated in the gap.
476
of passed current and the attenuation coefficient of the wave on pressure

and the type of gas were investigated. Figure 17.18 shows the results
of measurements taken for helium. The attenuation coefficient a has
the form:
U ( x ) = U 0e - ax ,
where U (x) is the amplitude of the voltage at the distance x from the
end of the cathode.
The data in Fig.17.18 show the existence of the conditions with mini-
mum attenuation. Previously, they could not be detected because of
narrow ranges of pressure variation in which the measurements were
taken. The results also show that the maximum speed does not depend
on the type of gas (Fig.17.19). Its value, 2·10 10 cm/s is close to the
maximum value for the given voltage (see Fig.17.18), because within
the error range it coincides with the speed of the electron flying in
vacuum and travelling through a potential difference of 250 kV (2.2·1010
cm/s). The graph of the dependence a (v f) in Fig.17.20 gives at ex-
trapolation to a = 0 the expected limiting speed of 2.2·10 10 cm/s.
Extrapolation of the dependence of the passed current on the speed
I (vf) to the determined limiting speed vlim = 2.2·1010 cm/s gives a limiting
current of 1.7 kA for all gases (Fig.17.21). This current is equal to
the current which would flow in the transmission coaxial line without
losses with the geometrical dimensions of the discharge device, i.e.
diameter of the internal and external conductors of 0.4 and 5.0 cm
respectively at a voltage pulse amplitude of 250 kV. 72
These results indicate that the breakdown waves, moving through
α , 1 0 2 c m 1 v , 1 0 10 c m / s v f , 1 0 1 0 c m / s
I, kA
p, torr
p, torr
Fig.17.18 Dependence of the velocity of the breakdown wave v (1), amplitude of passing
current I (2) and the continuation factor α (3) on gas pressure P.
Fig.17.19 Dependence of the velocity of the breakdown wave v on pressure p of a
nitrogen and helium mixture. Helium content 10 (1), 50 (2), 90% (3).
477
Plasma Diagnostics
α , 1 0 2 c m 1 I, kA
v , 1 0 10 c m / s
v , 1 0 10 c m / s
Fig.17.20 Dependence of the attenuation factorα on the velocity of the breakdown wave
v in different gases. 1) nitrogen, 2) helium, 3) air.
Fig.17.21 (right) Dependence of the amplitude of the pulse of past current I on the
velocity of the breakdown wave v in different gases. 1) nitrogen, 2) helium, 3) air, 4)
extrapolation.
the previously non-ionised gas with a limiting speed corresponding to

a voltage pulse of 250 kV are identical with those described previously.
In Ref.103, investigations were carried out into the conditions of
propagation of a breakdown wave and the results confirm the exist-
ence of the ‘limiting’ speed of the wave in the voltage range 100-280
kV for a negative polarity pulse with a duration 35 ns, with a front
of 2 ns, in a coaxial discharge device with the effected directed permitivity
of the insulator of e eff » 25.4. In these experiments, the dependence
of the ‘limiting’ speed on voltages satisfactorily approximated by the
formula v lim = (2eU/m ee eff ) 1/2 , where e, m e is the charge and mass of
the electron.
In Ref.103 it was reported that the pressure p m at which the lim-
iting speed is achieved does not always increase with increasing voltage
amplitude, i.e. for each gas under the given conditions there is minimum
pressure below which the slowly attenuating breakdown wave, have
in the property of ‘limiting’ speed does not form.
The structure of charge density in the breakdown wave

In Ref.43 and 64 it was found that the breakdown wave at positive
polarity of voltage has a stable structure with the harmonic longitu-
dinal modulation of the charge density moving with the speed of the
wave. The reduced structure of the discharge in the breakdown wave
was determined by taking into account the apparatus function of the
capacitance sensor. It should be noted that the electric field at the front
of the breakdown wave can reach the values required for the formation
with ‘running away’ electrons.
478
Fig.17.22 Synchronised oscillograms obtained from

wide-band capicitance transducer moved along the
axis z of the discharge tube with the step 2 cm
t, ns (n e0 = 4.3×10 9 cm –3 ).
The oscillograms, obtained using the capacitance sensor, which

records the propagation of the excess gas through the tube, are char-
acterised by high-frequency modulation. (x–t) diagrams, Fig.17.22, that
the modulation of each separate oscillogram is the result of movement
of the modulation source through the gap. The structure was recorded
with a capacitance sensor which reacts to the electric field generated
by the charges situated not only against a sensor but also by the charges
forming the structure as a whole and distributed in accordance with
the structure along the tube. Therefore, the signal, taken from the sensor,
is determined by the superimposition of the field of charges distributed
along the tube. With increase distance of the charge from the sensor
its contribution to the common signal will decrease in accordance with
the diagram of sensitivity of the capacitance sensor. If it is assumed
that the wave contains only the longitudinal structure of the charge
j (x + vt), the sensitivity diagram can be conveniently expressed by
the longitudinal-coordinate dependence f (x). In this case the signal,
taken from the capacitance sensor can be written as
z
d
U(t ) = 1 / A j ( x + vt ) f ( x ) dx.
-d
The measurements show that the contribution of the charges, situated
479
Plasma Diagnostics
at distances larger than 12 cm from the sensor, to the common sig-

nal is less than 1% and, consequently, the integration limits were se-
z
d
lected as d = 12 cm. The normalisation multiplier is A = f ( x) dx. . The
d
sensitivity diagram f (x) was determined by experiments. For this pur-
pose, a nanosecond pulse was supplied to a narrow (3 mm) ring of
copper foil placed on the top of the discharge tube. The amplitude of
the signal, taken from the capacitance sensor gradually moved from
the end along the axis of the tube, was measured. The resultant de-
pendence was approximated by a corresponding function. The delay
of the signal from the peripheral regions of the sensitivity diagram was
ignored because the strength of its effect on the signal amplitude did
not exceed 2%. The Lorentz reduction of the dimensions of the structure
was also disregarded because it equals ~2% at a speed of the struc-
ture of 6·10 9 cm/s. The structure j (x + vt) was found by selecting
the approximating function which, after numerical integration, was com-
pared with the observed oscillogram of the signal taken from the ca-
pacitance sensor in the centre of the tube.
The depth of modulation of the structure increases with increasing
initial electron concentration. Oscillograms recorded from a capacitance
sensor at the maximum value n e0 = 4.3·10 9 cm –3 (Fig.17.22) were proc-
essed. The speed of displacement of the charged structure was 6·10 9
cm/s.
The amplitude of the first oscillation can be relatively high so that
the electric field of the field in the wavefront exceeds the critical value
required for realisation of the regime of continuous acceleration of the
primary electrons. 70 Estimates show that the field in the front of the
structure, shown in Fig.17.23, curve 1, is
E = s/2e 0 » 3.5·10 4 V/cm
which at a pressure of 15 torr in helium is already sufficient for the

formation of ‘running-away’ electrons. 70 Here s = Q/S is the ‘end’
surface density of the charge in the front wave, S is the cross sec-
tion of the tubes; Q is the charge present in the first peak of the struc-
ture. Under the effect of the field, the electrons assume the energy
close to the potential of the first oscillation of the front of the struc-
ture. This strength of the field is too low because the radial structure,
especially the ‘pointed’ form of the front of the breakdown wave, was
not taken into account in the estimates.
According to the authors of Ref.43 and 64, the formation of the
480
Q, 10 9 C/cm
Fig.17.23 Distribution of excess charge in

the breakdown wave. 1) corrected, 2)
x + vt, cm observed distribution.
charge structure is caused by the interaction of high-energy electrons,

formed in the front of the structure, with the plasma behind the front
of the breakdown wave. The presence of the vibrational structure of
charge density, moving together with the breakdown wave, leads to
vibrations of the current flowing through the high-voltage electrode.
The depth of modulation with respect to the measured current through
the plasma is 15%.
In Ref.71, which deals with the theoretical description of the break-
down wave, it was found that under specific conditions oscillations of
the electric field, current and volume charge form behind the front of
the ionisation wave. The frequency of these oscillations is close to the
plasma frequency. There are conditions under which attenuating current
oscillations are preceded by a finite number of oscillations with increas-
ing amplitude. Calculations were carried out under a hydrodynamic ap-
proximation, and the authors used the complete equation of movement
of the electrons including, in contrast to an early study in Ref.34, in-
ertia terms and the gradient of electronic pressure. The authors noted
that if the inertia of the electrons is taken into account, oscillations of
the volume charge form behind the front of the ionisation wave. The
calculations show that there is a range of electron concentration op-
timum for the formation of oscillations. For example, increase of n e
initially leads to the formation of several increasing oscillations and then
a further increase of ne decreases their number. The calculations carried
out in Ref.71 were conducted for the voltage of a high-voltage
electrode not higher than 30 kV. Under these conditions, as reported
481
Plasma Diagnostics
by the authors, the values of E/p are not too high and the effect of
run-away electrons can be ignored. Taking into account the qualita-
tive agreement of the calculated result of the experimental data, 43,64
the authors of Ref.71 noted that in the experiments relatively high val-
ues of E/p were obtained for a number of regimes in the region of
the front. At these values, the hydrodynamic description of the elec-
tronic gas may prove to be a very rough approximation.
Generation of microwave vibrations

The authors of Ref.33 and 64 produced a 15% depth of modulation
of current in the breakdown wave. In Ref.93 under the experimental
conditions close to those in Ref.43 and 64, 100% harmonic modula-
tion of current was obtained when introducing a feedback (Fig.17.24).
The processes in the discharge take place in the following sequence.
The induction period to the moment t 1 is characterised by the com-
plete reflection of the exciting pulse from the gap. The stage t 1+t 2
corresponds to the displacement of the front of the excess charge along
the plasma accompanied by the ionisation of the neutral gas. Efficiently
conducting plasma, charged to the potential 2U in relation to the reso-
nator walls, is produced at the resonator axis. At the moment t 3 the
plasma is neutralised by the electrons emitted by the second electrode.
Consequently, the electric energy stored by the plasma changes to the
vibrations of electromagnetic field in the resonator containing the plasma.
The excitation of the oscillations was of the impact type and their
frequency w 1 = 2·10 9 rad/s was less than the frequency w 0 of the lower
mode of an empty resonator. The measured current can be represented
in the form I = I 1 + I 2 exp (–qt) sin w 1t, where I 1 is the current of
plasma in the absence of oscillations. The relationship I 2 ~ dI 1/dt was
approximately fulfilled during the period of excitation of the oscilla-
tions. In the examined pressure range of the neutral gas p = 2 ¸ 30
U, rel. units
Fig.17.24 Formation of 100% hf

modulation of the breakdown wave in the
presence of a "feedback". 1) oscillogram
of the exciting high-voltage pulse, 2)
oscillogram of a pulse reflected from the
gap in the absence of oscillations, 3) the
t, ns same, in their presence.
482
torr the speed v of movement of the front of the excess charge and
the value I 2 monotonically decreased with increasing p, and the fre-
quency w 1 remained constant within the limits of the measurement ac-
curacy with the change of both the pressure and the duration of the
excitation pulse. Attenuation q was determined mainly by the removal
of electromagnetic energy to the conducting coaxial cable and did not
depend in the experiments on the duration of the excitation pulse and
slightly increased with increasing pressure.
The authors of Ref.78 recorded microwave oscillations behind the
front of the breakdown wave using capacitance dividers. Evidently, the
observed oscillations are the result of the excitation of the coaxial
resonator at the internal frequency as a result of the resonance inter-
action of the charge wave moving in the front (for the water line of
v » 2·10 9 cm/s, for the line with an oil dielectric v » 1.5·10 10 cm/s)
with the electromagnetic wave in the delaying structure (c/e ½ » 5.5·10 9
cm/s in the water line and 1.6·10 10cm/s in the oil line). On the basis
of the order of magnitude, the frequency spectrum of the observed os-
cillations can be regarded as equal to the internal frequency of the
coaxial resonator w » c/l f, where l f is the length of the front. For the
oil and water lines w » 10 9 s –1 , which is in agreement with the fre-
quencies observed in the experiments.
The experiments with a breakdown of the tube in the oil line show
that the amplitude of oscillations depends in a nonmonotonic manner
on the pressure and is maximum at the maximum speed of the breakdown
wave, in the pressure range where the speed of light in the coaxial
and the speed of the breakdown wave are similar. It should be noted
that at this pressure the current in the tube is maximum.
It was also established that in the pressure range in which the am-
plitude of the oscillations is maximum, the speed of the wave (Fig.17.25)
has a local minimum and the voltage attenuation coefficient has a local
maximum. It can be assumed that as a result of energy losses in the
excitation of the electromagnetic wave in the coaxial line behind the
front of the breakdown wave its speed decreases and the voltage at-
tenuation coefficient increases.
Doppler effect in wave breakdown

The propagation of a breakdown wave is identical to the movement
of a rod with high conductivity from a high-voltage electrode. In fact,
in propagation of a positive polarity wave the drop of the potential
behind the wave front is considerably smaller than the potential jump
in the front; after closure of the gap by the breakdown wave the high-
voltage pulse freely passes through the gap and is recorded in the cable
483
Plasma Diagnostics
α, c m 1 v , 1 0 10 c m / s
P, torr
Fig.17.25 Dependence of the speed of breakdown wave (1) and attenuation (2) on air
pressure.
behind the gap without any decrease of the amplitude in relation to the
incident pulse. 64 During the propagation of the wave through the gap
the duration of the recorded deflected pulse is longer than that of the
incident pulse and the amplitude is smaller. The latter fact is explained
by the Doppler amplitude-frequency change of the pulse in reflection
from the moving front of the breakdown wave.
If the duration of the current pulse in a stationary cable is t, then
after reflection from the open end of the cable moving with the
1+ b
speed v = bc, it’s amplitude decreases 1 - b times and the duration
1+ b
increases correspondingly 1 - b times. This occurs if, using the Lorentz
transformation for the pulse, and transfer into the moving co-ordinate
system, we write the equation of the mirror reflection of the pulse in
the moving system and transfer back to the stationary co-ordinate
system. 73 The voltage and current of the reflected pulse are written
in the form 51
1- b FG
1- b IJ
U ref ( t ) =
1+ b
U in
H
1+ b
t ,
K
1- b FG
1- b IJ
Iref (t ) =
1+ b
Iin
H
1+ b
t .
K
484
This explains the absence of any strong reflected electromagnetic

wave at a voltage of 250 kV. 74 With increasing speed, the amplitude
of the pulse reflected from the front decreases in accordance with the
above equations.
Thus, measuring the speed of the wave, we can easily calculate the
electric energy parameters with a wave breakdown carried out by the
high-voltage rectangular pulse. Thus, the input resistance of the gap
is:
U in + U ref
Zin = = Z / b.
I in - I ref
In the experiments for v f = 4.9·10 9 cm/s b = 0.163. 64 The value Z in,

computed from the above equation, is 306 ohm. The values of Z in, ob-
tained from the oscillograms of the incident and reflected currents, are
also equal to approximately 300 W.
The power introduced into the gap is:
4b
Pintr = ( I in - I ref )( I in + I ref ) Z = Pin .
(1 + b) 2
We assume that the tube is long because the front of the break-
down wave does not reach its end due the short pulse duration. The
distance which the breakdown wave can travel, is expressed in the form
b
Lmax = cTin .
1- b
1- b
We denote 1 + b = g . The energy of the reflected pulse is associ-
ated with the energy of the incident pulse
z
T1
Eref = Zg 2 Iin2 ( g ) dt = gEin ,
0
where T 1 = T incg –1 . The energy of the pulse dissipated at the front is
2b
Ed = Ein - E ref = Ein .
1+ b
485
Plasma Diagnostics
The total dissipation energy
Ein 1 - b
E d .total = E / Lmax = 2
cTin 1 + b
and is equal to the pressure force of the electromagnetic wave
Pin 1 - b
Fem = E d .total = 2 .
c 1+ b
The dissipation power is determined by the ratio of the dissipation

energy to the duration of movement of the front wave t w = L max/v f =
T in/(1 – b) and equals
Ein 2b / (1 + b)
Pd = = Femv f .
Tin / (1 - b)
It can easily be seen that the dissipation power is maximum at

b = Ö2 – 1 and equals 0.343 P in .
In reflection of the electromagnetic wave from the moving mirror,
the power of the incident wave is higher than the sum of the powers
of the reflected and dissipated components: P in > P ref + P d. This is
caused by the difference between the duration of the reflected pulse
T ref and the duration of movement of the front wave t w and the du-
ration of the incident pulse T in. This results in the energy balance
Ein = Pin Tin = Pref Tref + Pd t w .
The measurement of the electromagnetic pulse, reflected from the

front of the breakdown wave, and also the agreement of the above
described expression of the electrode dynamic properties of the wave
with the experiment make it possible to propose the following pattern
of the process.
The plasma column behind the front of the ionisation wave repre-
sents together with the metallic screen a co-axial waveguide whose
properties are almost identical with those of the electric cable. A packet
of electromagnetic waves, representing a high-voltage pulse, propagates
to the plasma waveguide and also in the cable. In the moving zone in
the vicinity of the wave front where the conductivity of the plasma
channel rapidly decreases, the incident electromagnetic packet is re-
486
flected and propagates in the opposite direction. If the losses in the

plasma coaxial line are ignored, the voltage drop taking place in the
front of the breakdown wave is determined by the sum of the ampli-
tudes of the incident and reflected electromagnetic packets. In a mirror
reflection of the electromagnetic packet the voltage drop at the front
of the breakdown wave is associated with the amplitude of the high-
voltage pulse fed to the gap in the following manner:
2
Dj = U in
1+ b
and, consequently, is determined only by the speed of the reflection

zone. Behind the reflection zone, there is a zone occupied by the ioni-
sation wave. The energy ‘feeding’ of the ionisation processes in the
wave takes place as a result of the energy flux of the incident elec-
tromagnetic packet. The intensity of this feeding does not exceed the
work of the force of the electromagnetic pressure in reflection of the
incident electromagnetic packet from the ionisation front.
Impact properties of the breakdown wave

Dissipation of the electromagnetic wave at the front of the breakdown
wave causes a moving conduction jump to form in the front. This phe-
nomenon in the gas-discharge device is identical with the movement
of the impedance jump along the transmission line with non-linear pa-
rameters. An example of powerful transmission lines with magnetic self-
insulation in which the propagation of a powerful electric pulse results
in a decrease of the duration of the pulse front as a result of the
magnetron effect is described in Ref. 75.. This is associated with the
change of leakage and effective reactive parameters of the line. Another
example are lines with a ferrite filling. The waves of the impedance
jump in these lines are referred to as impact electromagnetic waves. 76
They have the following properties. 76 The propagation of a powerful
electromagnetic pulse along the line becomes non-linear due to dissi-
pation in remagnetising ferrite at the pulse front. This results in sub-
sequent shortening of the pulse time. When reaching the minimum time
determined by the properties of ferrite and the signal amplitude, the
propagation of the front becomes stationary and the pulse time linearly
decreases because dissipation still continues at the front. Behind the
front of this ‘impact electromagnetic wave’ the properties of the wave
guide are linear and the propagation speed of the electromagnetic waves
behind the front is higher than the speed of the front. This results in
487
Plasma Diagnostics
the reflection of the electromagnetic signal from the moving imped-

ance jump.
The effects observed in the experiments with the breakdown waves
have direct analogy with the above-mentioned ‘impact electromagnetic
wave’ regardless of differences and nature of the processes in fer-
rite and weakly ionised plasma. The duration of increase of the leading
edge of the pulse of the reversed current of the wave, measured in
Ref.31, proved to be considerably shorter than the duration of the lead-
ing front of the voltage pulse supplied to the tube. Consequently, the
author referred to it as the impact electrical wave. Breakdown waves
with positive polarity characterised by a gradual decrease of the du-
ration of the charge front and the radiation.49,51,64 The speed of the front
of the breakdown wave is constant. The curvature of increase of the
current, recorded behind the discharge tube, is higher than the curvature
of increase of the current supplied to the tube. 64 Movement of the
breakdown wave is characterised by the formation of a pulse reflected
from the front. The amplitude and duration of this pulse change in relation
to the incidence pulse in accordance with the Doppler effect. 64 The
speed of the breakdown wave increases with increasing electron con-
centration in the plasma thus leading to the tip of the wave hitting its
base and to an increase of the curvature of the front of the break-
down wave.
The authors of Ref.37 and 38 assume that the steeper front of the
signal in the second electrode, obtained in their work, is caused by the
formation of an impact electromagnetic wave during a breakdown. A
similar approach to explaining the experimental results was developed
in Ref.39 in which the non-linear element was assumed to be the linear
F I
z
t -1
resistance R = R G1_ q -1
I dtJ
H K
0 0 similar to that proposed in Ref.36.
0
Fast electrons
One of the special features of the wave breakdown is the appearance
of high-energy electrons. Stationary movement of the electrons is dis-
rupted when the condition 80
E > E k .max = 4 pe 2 Nz / 2,72 e ,
is fulfilled, where N is the gas density, z is the number of electrons

in the molecule, e is the average energy of excited electrons enter-
ing the molecule.
488
It is well known that for each gas there is the pressure p m (for air
it is 20 torr, helium 150 torr at a voltage of 250 kV) at which the speed
of the wave front and current are maximum and the attenuation of volt-
age minimum. 72 In Ref.77 on the basis of analysis of the U (pd) de-
pendences it was shown that the typical phenomenon for the break-
down of the gases in the wide range of pd is the ‘runaway’ of the
electrons. High-energy electrons 61,78 were also recorded at the front
of a breakdown wave in air. The aim of the work in Ref.81 was to
verify the assumption according to which the existence of optimum p m
is associated with the formation of ‘runaway’ electrons. These inves-
tigations were carried out in air and helium for which P m and (E/p) c
are greatly different. Single pulses of negative polarity with an am-
plitude of 250 kV, the duration at half height of 35 ns with the front
of 3 ns were supplied to the cathode of a discharge tube 0.8 cm in
diameter and 80 cm long surrounded with an earthed cylindrical me-
tallic screen 5.5 cm in diameter connected to a mesh anode. The space
between the tube and the screen was filled with oil. The voltage along
the tube was measured with capacitance dividers, the currents with
inverse current shunts, the current of ‘runaway’ electrons I e was
measured with a Faraday’s cylinder behind a mesh anode. The duration
of I e in air and helium decreased with increasing pressure from 30 to
4 ns at the half height (Fig.17.26). This difference from the data in
Ref.78 where I e at all pressures was around 3 ns is evidently caused
by the fact that in Ref.81 the speed of the front v f » 10 10 cm/s
(Fig.17.27) was higher than in Ref.78 (2×10 9 cm/s) where the insula-
tor was represented by water, and since electrons with different en-
ergies formed at the front of the wave, part of the electrons do not
keep pace with the front. Figure 17.27 shows the form of the calcu-
lated speed v e of passage of ‘runaway’ electrons to the discharge tube
calculated from the time delay between the start of the voltage pulse
of the cathode at the maximum of the signal I e. They are close to v f
but displaced to the range of higher pressures for He. The signals from
the reverse current shunt have two characteristic phases; first evidently
corresponds to the conduction current of the wave front I f, the sec-
ond to the total current I t because it disappears with disappearance
of I e. The dependences of Ie, I f, It on pressure for helium are displaced
to higher pressures in comparison with air (Fig.17.28); this is in agree-
ment with the displacement of (E/p) cr . The current of the front I f in
a wide range of pressures can be transferred by the current of ‘running-
away’ electrons I e because these electrons have a strong effect on the
characteristics, like the ionisation wave in movement of the electron
beam in a gas. 79
489
Plasma Diagnostics
v, cm/s
10 ns a
ve
b
vf
c ve air
vf
d
p, torr
Fig.17.26 Oscillograms of current pulse Ie of fast electrons at different helium pressures.

p = 20 (a), 33 (b), 64 (c), 168 torr (c).
Fig.17.27 Dependence of the velocity of breakdown layer v f and the velocity of fast
electrons v e on gas pressure p.
17.7 X-radiation during a wave breakdown

In Ref.61, measurements of x-radiation, synchronised with electric
measurements, were taken during the development of a wave break-
down (Fig.17.29). Breakdown in a discharge device was carried out
with pulses of negative polarity with an amplitude of 15 kV. At a break-
down starting at moment t 1 the voltage of the cathode becomes lower
than the open circuit voltage. This moment corresponds to the appear-
ance of current on the cathode. In subsequent stages the current behaves
non-monotonically and has 2 maxima, and its increase correlates in time
with the processes inside the gap.
X-radiation at the cathode (Fig.17.29b) is recorded after a decrease
of the voltage on the cathode to almost zero.
¶U ( x , t )
The signals from the capacitance sensor ~ in different sec-
¶t
tions of the tube are presented in Fig.17.29c. The oscillogram of the
signal, recorded at the cathode is non-monotonic, and its maxima and
minima correspond in time to the ‘fastest’ variation of the current on
the cathode. With increasing co-ordinate x, i.e. with increase of the
distance from the cathode, the general form of the oscillograms re-
mains unchanged. Because of the mutual synchronisation of the os-
cillograms, it can be seen that the perturbation formed initially at the
490
a
15 kV
c
I e, I f, k A I p, k A
p, torr
Fig.17.28 Dependence of the amplitude of

d conductivity current of the wave front If, total
current I t and fast electron current Ie on gas
pressure in the discharge tube.
I, rel. units
t, ns
Fig.17.29 Synchronised measurements of electrical parameters, spatial dynamics of optical

and x-ray radiation of the breakdown layer at p = 1.5 torr. a) oscillograms of voltage
(1), current (2) on the high-voltage electrode voltage of the incident pulse (3); b)
oscillogram of the pulse of x-ray radiation at the cathode of the discharge tube; c)
x–t diagram of the breakdown layer obtained using a differentiating capacitance divider:
4 - first, 5 - second, 6 - third, 7 - reflective waves; d) x–t diagram of x-ray source:
8 - third wave of breakdown; d) 9 - the pulse of ex-radiation at the cathode; 10 -
optical emission of the plasma at the cathode; 11 - optical emission of the cathode jet.
491
Plasma Diagnostics
vicinity of the electrodes or inside the gap is then transferred along

the tube.
The (x – t) diagrams of propagation of the source of x-radiation
(Fig.17.29d) show that the source forms inside the gap at a distance
of x = 24 cm from the anode and then moves at a speed of 4·10 9
cm/s in the direction towards the anode. The duration of the front of
increase of x-radiation, recorded on the side of the tube, remains con-
stant, and the amplitude increases with increasing propagation. The
speed of the source of x-radiation coincides with the speed of the front
of optical radiation and the front of the potential which were deter-
mined from (x – t) diagrams of these parameters measured using a
photoelectronic multiplier and a capacitance sensor.
The time dependence of the intensity and effective energy of x-ray
quanta, recorded at the end of the discharge tube at the anode, are
shown in Fig.17.30.
Optical radiation of plasma in the vicinity of the cathode (Fig.17.29e,
curve 8) forms with the start of passage of current on the cathode and
the radiation intensity then monotonically increases. At the same time,
there is strong optical radiation of the cathode jet which occupies a
small region in the vicinity of the cathode. The dynamics of the lat-
ter is different, and the signal from the photodetector, focused only on
the part of the jet is non-monotonic with time (Fig.17.29e, curve 9).
The oscillograms obtained in this case can be used to describe the
sequence of processes in the discharge gap. Initially, the increase of
voltage does not cause any emission from the cathode, and the capaci-
tance sensor, placed in the vicinity of the cathode, records the signal
¶U k
Id = . The moment of time t 1 corresponds to the start of emis-
¶t
sion current Ic and the start of the wave transferring the potential inside
the gap with the speed v 1 = 2·10 9 cm/s. Its displacement is accom-
E eff , keV
Fig.17.30 Dynamics of effective energy of quanta Eeff

I, ns (1) and intensity of x-radiation (2).
492
panied by excitation and ionisation of the gas. Up to the start of the

wave the optical radiation from the near-cathode zone is almost non-
existent. At the moment t 2 a cathode jet appears. It is preceded by
a decrease of current I c and of optical radiation measured with a
narrow-band photomultiplier. 48 The possible reason for the decrease of
current is the depletion of the near-electrode layer by the electrons
owing to the fact that the emission current is not capable of compen-
sating the movement of the electrons in the direction to the front of
the wave, i.e. I e < I f (I e is the autoelectronic emission current, I f is
the current transferred by the wave). With the formation of the cathode
jet the density of emission of electrons increases, the current through
the electrode I c becomes considerably higher thus resulting in the for-
mation of a second ionisation wave at the cathode. Since the conductivity
of the plasma behind the front of the first wave is higher than in front
of it, the speed of the second wave v 2 >> v 1 and according to the
measurements it is equal to v 2 = 10 10 cm/s. The second wave also ex-
cites and ionises the gas and increases the plasma potential. At a dis-
tance x = 22–24 cm from the cathode the second wave catches up
with the front of the first wave and increases the strength of the electric
field in the front of the first wave. The electrons, accelerated by the
latter to high energies, are ejected forward (without any collisions with
gas atoms) in the direction of movement of the wave and on to the
wall of the discharge tube in the vicinity of the front. This moment
corresponds to the appearance of x-radiation. The front of the new
wave, formed as a result of the interaction of the first and second
waves, moves at the speed v 3 = 4.10 9 cm/s, generating fast electrons
which, in turn, cause breaking x-radiation on the walls of the discharge
tube. When the wave reaches the anode, the front of the reflected
wave forms in the vicinity of the anode. This front moves at the speed
v 4 = 1.10 10 cm/s in the direction to the cathode and discharges the
plasma in the discharge tube because the jump of voltage in the front
of the reflected wave has the sign opposite to the sign of the voltage
jump at the front of the direct wave and is sufficiently large so that
when it approaches the cathode x-radiation also forms there (Fig.17.29b).
In Ref.61 the mean (with respect to the cross section) effective en-
ergy of x-ray quanta reaches 21 keV so that it can be concluded that
the electrons were accelerated to an energy of 41 keV. According to
the results of measurements taken with a capacitance sensor, the voltage
drop at the front is less than 2 kV. This means that the fastest elec-
trons have the energy exceeding the maximum possible drop of the
potential at the front of the breakdown wave. One of the possible ex-
planations of this fact is the assumption that the electrons have ac-
493
Plasma Diagnostics
quired an additional energy as a result of rapid movement of the ac-

celerating front.
17.8 Initiation of laser radiation by the breakdown wave

The spatial–time dynamics of pumping and radiation waves in a nitrogen
laser has been studied in Ref.24,63,82,83. In Ref.63, it was found that
the first peak of laser radiation and the breakdown wave move from
the cathode to the anode at a speed v » 3·10 9 cm/s (voltage ampli-
tude 300 kV). The delay of the first peak of laser radiation from the
front of the breakdown wave is 8–10 ns. The second peak of laser
radiation propagates from the anode to the cathode with a speed of
5·10 9 cm/s which is approximately equal to the speed of the electro-
magnetic signal in the coaxial line with a water dielectric and a dis-
charge tube as a central conductor.
Figure 17.31 shows the dependence of the speed of the ionisation
wave and the peak generation power in air on pressure. In Ref.63
generation was observed in the pressure range 0.8–646 torr in nitro-
gen and 0.8–342 torr in air. The laser operated in the superluminance
regime. The laser radiation time was varied depending on pressure from
1.5 to 10 ns. The maxima of the speed of movement of the wave and
the radiation power are situated in the same pressure range, and the
maximum power of generation is observed at a higher gas pressure.
This is in qualitative agreement with the results in Ref.63. As in
Ref.82,83, the laser radiation pulse at the exit of the discharge tube
has two peaks with time. The maximum value of the peak power was
450 kW in nitrogen and 250 kW in air at a pressure of 30 torr. The
maximum specific power was 15 kW/cm 3.
The speed of the breakdown wave in Ref.63 is 7–8 times less than
in Ref.82 owing to the fact that the space between the discharge tube
v, 109 cm/s P, rel. units
p, mm Hg
Fig.17.31 Dependence of the peak power of generation of a nitrogen laser (1), the
velocity of the front of the ionisation wave (2), the first (3) and second (4) peaks of
laser radiation on air pressure.
494
and the metallic screen was filled with water with high dielectric per-
mittivity e = 80, and in Ref.82 it was filled with nitrogen.
To examine the dynamics of generation in space, investigations were
carried out into the laser radiation generated from a discharge tube using
semi-transparent mirrors described previously. 63 Radiation from these
mirrors have two peaks with time. Figure 17.32 shows an (x–t)
diagram of movement of laser radiation along the tube with air. This
movement can be characterised by the speed of the peaks. This speed
gradually decreases along the tube.
It should also be stressed that the first radiation peak moves from
the cathode to the anode and the second one from the anode to the
cathode, i.e. the second peak initially appears on the mirror close to
the second electrode, and only then on the mirror close to the high-
voltage electrode. Evidently, the speed of the first peak is in good
agreement with the measured value of the speed of the ionisation wave,
and the speed of the second peak greatly exceeds this speed. In Ref.
63, this is explained by the fact that the speed of the second peak
coincides with the speed of the reflected wave. The reflected ionisation
wave propagates along the gas strongly ionised with the first break-
down wave and, consequently, its speed is close to the speed of propa-
gation of the electromagnetic signal in the coaxial system. 24
Figure 17.33 shows the data on the times of appearance of the front
of the ionisation wave and laser radiation on the mirrors in relation to
the distance from the cathode. 63 It can be seen that the delay between
the front of the breakdown wave and the maximum radiation remains
almost constant along the tube and equals 8–10 ns. Within 10 ns the
wave passes approximately 35 cm in this pressure range and does not
manage to reach the second electrode. Thus, at the start of propagation
x, cm
Fig.17.32 (x–t) diagram of movement of the

t, ns peaks of laser radiation through the discharge
tube.
495
Plasma Diagnostics
t, ns
x, cm
Fig.17.33 Time to appearance of the front of the ionisation wave (1), start (2) and the
first peak of laser radiation (3) in relation to the distance to the cathode at an air pressure
of 9.7 (4), 20.7 (5) and 45.6 (6) torr.
of the ionisation wave reflected from the second electrode, the laser
radiation already exists and the reflected ionisation wave, exciting the
gas for the second time, results in the formation of the second peak
in laser radiation. The time delay between the front of the breakdown
wave and the appearance of radiation on the mirror changes along the
tube. At the start of the tube it equals approximately 4 ns which is
equal to the duration of the front of the exciting pulse. At the end of
the tube the delay time decreases to zero, i.e. the laser photons pass
to the second electrode almost simultaneously with the front of the
breakdown wave. The appearance of laser radiation at the exit from
the tube simultaneously with the breakdown wave was also reported
in Ref.82. In Ref.63, it was assumed that this long delay time of the
maximum of radiation in relation to the front of the breakdown wave
is essential for converting the energy of the electrons, accelerated at
the front wave to an energy of ~100 kV, to the laser radiation energy.
The measurement show24,63 that the removal of excitation of the me-
dium by laser radiation takes place by a wave mechanism and the dy-
namics of laser radiation with time and in space is determined by the
nature of excitation of the gas by the breakdown wave.
17.9 Effect of the type of gas

In Ref.103, detailed investigations were carried out into the conditions
of propagation of the breakdown wave in different gases and mixtures:
496
He, Ne, Ar, Kr, Xe, H2, air, CO2 propane–butane, acetone vapours, CCl4,
SF6, N 2, Ne–Ar. Analysis of the experimental data shows that both the
elementary processes in the plasma formed behind the front and the
kinetics in the gas at the head of the wave front can have a strong
effect on the conditions of propagation of the breakdown wave.
Processing of the experimental dependences v(p) in the inert gases
was carried out using an empirical equation for the pressure optimum
for the propagation of the breakdown wave: p m = CI ai M b, where C is
a constant which does not depend on the type of gas. At a = 5 and
b = 0.5, the value of C is the same for all examined inert gases, with
the error not exceeding 10%. Consequently, according to a prediction,
for radon p m = 1.3 torr.
17.10 Current state of the theory of wave breakdown

The most extensive total system of equations was derived by A.N.
Lagar’kov and I.M. Rutkevich. 34,84 They used the approximation of the
surface wave from the theory of plasma waveguides but took into
account the collision processes and the kinetics of ionisation in the
electric field. Figure 17.1 shows the schema of the examined electro-
dynamic system. The main equations have the following form:
for plasma (region I)
¶ne 1
+ div j = ane , (17.1)
¶t e
¶n1
= ane , (17.2)
¶t
j = sE , (17.3)
div E = 4pe(ne - ni ), (17.4)
E = -Ñj, (17.5)
for the dielectric (region II)
Ñ 2 j = 0.
It is assumed that the initial concentrations of the charges and the bound-
ary conditions are known: j = 0 on the metallic screen (region III),
on the electrode AB, j = j 0 (y, t) on the end surface of the dielec-
tric a < |y| < a + h. The following condition must be fulfilled at the
boundary of plasma with the dielectric |y| = a
497
Plasma Diagnostics
¶q
jy = , where q = ( e - E y-w - e + E y+w ) / 4 p,
¶t
E –y, E +y are the normal components of the field in the plasma and the
dielectric, respectively.
It is evident that the system of equations (17.1)–(17.5) holds
under the condition V<<C so that we can ignore the vortex compo-
nent of the field in plasma in comparison with the potential component.
The self-modelling solution of the system of equations (17.1)–(17.5),
obtained in Ref.84, explained in particular the non-monotonic dependence
of the current on time in the circuit of an earthed electrode 28 by the
existence of an ionising soliton of the electric field at the wave front.
O.A. Sinkevich and Yu.V. Trofimov 33,85 examined the non-station-
ary boundary-value problem of the propagation of a breakdown wave
in a pre-ionised gas. In contrast to equation (17.1), it was assumed that
in the internal part of the channel n e = n 1 and div j = 0. This means
that the excess charge is distributed over the surface of the plasma
column. Thus, the problem was solved for the case ka << 1 (a is the
radius of the tube, k is the wave number).
It was shown in Ref.85 that the long-wave approximation corresponds
to the description of the ‘diffusion’ process along the electric condenser
formed by the screen and the plasma column. The corresponding ‘dif-
fusion’ equation has the form
¶j ¶ ¶jFG IJ
=
¶t ¶x
ne
¶x H
.
K
The role of the ‘diffusion’ coefficient is played by the electron con-
centration. In Fig.17.34, the calculation results are compared with the
results of measurements of the distribution of the potential along the
channel at different moments of time carried out by Westberg. 10
In conclusion, it should be noted that in electrodynamics similar sys-
tems are referred to as distributed RC structures. 86 At a constant linear
resistance R, the phase velocity of the wave would be 86
F 2w I 1/ 2
Vphase =
H RC K ,
where w is frequency, C is the linear capacitance. The breakdown in

the coaxial line using the RC structure was modelled in, for example,
498
Fig.17.34 Distribution of potential

along the discharge tube at different
times. 1) experiment10; 2) calculation85;
the numbers of the curves are the values
of t.
Ref.87. The variation of R in this work was taken into account using
the following procedure. It is assumed that the strength of the field
in the conductor E is linked with the current by the relation i = k 1n eE,
i.e. we ignore the bias current in the plasma. At a penetration depth
of the field in the conductor larger than the radius of the latter the
multiplier k 1 is equal to µ es (s = pa 2). Using the Rompe–Weitzel ap-
proximation, we accept that ¶n e/¶t= k 2iE, and consequently
F I
z
t 1/ 2
R = k1k2 /G 2
dt J
H K
i (Rompe - Weitzel equation). .
0
This model is relatively approximate. However, it makes it possi-
ble to examine the effect of the main factors (screen, initial electron
concentration, the curvature of the pulse front) on the breakdown dy-
namics.
Thus, at present, the physics of non-relativistic breakdown waves
(v << c) on the condition that the distribution function of the electrons
is determined by the local values of the electric field, can be regarded
as basically sufficiently developed. (Experimental and theoretical ex-
amination of the role of radial components was carried out in: E.I.
Asinovskii, et al., A similarity criterion for the speed of electric
breakdown waves in screened tubes of different diameter, Pre-print
IVTN, No.3-50, Moscow (1992)).
At breakdown wave speeds close to the speed of light, the non-
potential nature of the electric field which influences the formation of
the wave and restricts its speed is very strong.
In contrast to Ref.33,85, the authors of Ref.88 examined a numerical
model with a non-potential electric field. The procedure of averaging
over the cross section of the waveguide was used. 33,34 Taking into
account ¶/¶z << 1/r 0 (r 0 is the radius of the discharge channel, z is
the longitudinal co-ordinate) and ¶/¶t << 4 ps 0 (s 0 is the initial con-
ductivity), the equations for the propagation of the breakdown wave
499
Plasma Diagnostics
have the form
C / pr02 × ¶j( z, t ) / ¶t + ¶jz ( z , t ) = 0, (17.6)
E z ( z, t ) = -¶v / dz - Lpr02 ¶jz / ¶t , (17.7)

j z = s( z , t ) E z , (17.8)
where s, j, j z, E z are the values of the conductivity, potential, cur-

rent density and the longitudinal electric field averaged over the cross
section of the waveguide, C, L are the linear capacitance and inductance
of the waveguide.
The equations (17.6)–(17.8) are closed by the equations for the con-
ductivity of slightly ionised plasma
s = m e ( E z / P )ne ( z , t ), ¶ne / ¶t = a (iz / P)ne ,
where m e, a is the mobility of the electrons and the constant of the

ionisation rate which depend on E z /p, p is the gas pressure. The
initial-boundary conditions are s 0 (z) = s (z, 0) and j 0 (t) = j (0, t).
Figure 17.35 shows the dependence of the speed of propagation of
the breakdown wave on ¶U/¶t = ¶U 0 /t f and s 0. Agreement between
the experimental and calculated data was obtained at s 0 = 1.2 W –1 ×
cm –1. According to estimates, this level of conductivity in front of the
wave front can be generated as a result of the flux of ‘runaway’ elec-
trons.
The authors of Ref.89 constructed a theoretical model taking into
the ‘running-away’ electrons and ionisation of the gas by them. The
equations describing the electrodynamics of the breakdown wave (tel-
egraph equations) were solved numerically together with the
non-stationary Boltzmann equation in the two-dimensional phase space
v , 1 0 10 c m / s
Fig.17.35 Dependence of the velocity of the

∂ U / ∂t , 1 0 14 V / s breakdown wave on the rate of voltage increase.
500
for the distribution function of the ‘runaway’ electrons with respect

to the energies with the integral of the inelastic collisions in the right
hand part. The results of calculations carried out in Ref.89 describe
satisfactorily the experimental data in Ref.72 for the speed of the wave
and attenuation. The ‘runaway’ electrons are characterised by a wide
spectrum whose upper boundary exceeds eU 0. 89
17.11 Wave breakdown - a method of examining processes in

plasma
Examination of the lifetime of excited states
At present, the main methods of perturbation of the population of the
examined level are either the monoenergetic electron beam or optical
pumping. The gas discharge has until now been neglected because of
a number of reasons, especially the heterogeneity and poor reproducibility
of excitation. In Refs.28, 90, 91 it was established that these short-
comings can be avoided, at least in the pressure range 0.5–30 torr, if
the gas is excited by means of a wave breakdown. This also results
in a high degree of excitation thus simplifying the problem of recording
radiation and processing the results. The method is relatively simple
and universal.
The 3 1D level of helium 90 and the states C 3Π u of the nitrogen mol-
ecule were investigated. 90,91
The lifetime of the C 3P u state (V = 0) was measured using vari-
ous methods, and the scatter from 30 to 46.3 ns exceeded the meas-
urement error. The aim of the investigations in Ref.90 and 91 was to
measure the lifetime of the state C 3P u (V = 0) of the N 2 molecule in
nitrogen and when measuring the partial pressure of oxygen. In process-
ing the results, the level not excited by the electrons was ignored
because of the low degree of ionisation. According to the estimates,
the gas temperature is 320 K.
The radiation lifetime of the C 3P u state (V = 0) (37 ns), measured
in Ref.90 and 91, and the cross section of quenching of this state by
the nitrogen molecules (1.9 Å 2 ) coincide, in the measurement error
range, with the data from the majority of studies, and the quenching
cross section of the state by the oxygen molecules (52 Å 2) is larger
than the data in other studies. 90 The results can be used to explain the
decrease of the peak power of the laser in air in comparison with the
laser in pure nitrogen92 by quenching of the upper laser level by oxygen
molecules.
501
Plasma Diagnostics
Hydrogen dissociation, measurement of the diffusion coefficient

of atomic hydrogen
The efficiency of the dissociation process in the electric discharge is
determined by the energy spectrum of the electrons. Dissociation in
hydrogen takes place mainly through the excitation of the electronic
state with an excitation threshold of 8.8 eV. At a higher electron energy
a large part of this energy is used for the excitation of the electronic
states with the gase heated to a relatively low temperature. The de-
velopment of a discharge in the shape of a breakdown wave enables
it to be used to form the required fields in the plasma and makes it
a suitable means for the dissociation of molecular hydrogen.
In Ref.97, wave breakdown was used to measure the diffusion co-
efficient of hydrogen atoms. The following measurement procedure was
used. After the controlled delay time t d after the nanosecond break-
down a low-current rectangular probing pulse of microsecond duration
was supplied to the discharge gap to excite the hydrogen atoms. How-
ever, because of additional dissociation the pulse does not change the
concentration of the atoms. The dependence of the integral intensity
of the line H a with respect to time on the delay time t d is measured;
the dependence is then used to determine the diffusion coefficient.
In Ref.98, the relatively course of radiation H a was used to esti-
mate the concentration of atomic hydrogen after the discharge. The
absolute concentration method is based on the fact that the processes
of direct and dissociative excitation prevail at different delay times t d.
In Ref. 8, the maximum number of dissociation acts per 1 eV of
the supplied energy was h = 0.018 eV –1 . If it is assumed that a sin-
gle dissociation act requires 8.8 eV (excitation energy of the level 3S+u),
then it can be seen that the electric efficiency is 16%. Comparison
of the nanosecond discharge developed in the form of a wave break-
down, with other known plasma sources of hydrogen atoms shows that
it is characterised by the highest efficiency at such a high dissocia-
tion rate.
Relaxation of the plasma loaded with direct current

In Ref.99 and 100 investigations were carried out into relaxation proc-
esses in helium in exciting the plasma with a nanosecond discharge de-
veloped in the form of a breakdown wave, especially the special fea-
tures of the breakdown of helium plasma loaded with direct current.
The region of the breakdown, characterised by a low radiation yield,
‘glow break’, was observed and examined. The results show that the
nanosecond discharge generating additional ionisation in the plasma
makes it possible to regulate the temperature of electrons in the range
from 0.05 to 0.5 eV during hundreds of microseconds. 100 In Ref.101
502
a method was developed of diagnostics of ionisation in the nanosec-

ond discharge on the basis of the duration of ‘glow break’. The re-
sults show that there are conditions under which the breakdown wave
results in general ionisation in the nanosecond discharge.
Examination of the fast electrons shows that the front of the break-
down wave contains regions with the strength of the electric field of
E/N ³ 10 –14 V cm 2. At these values of the strength of the electric field
at the front of the breakdown wave the ionisation processes are highly
effective. A very large increase of pressure reduces the efficiency of
ionisation but the latter remains sufficiently high in relation to the losses
in heating the gas.
This makes it possible to use the wave breakdown for sustaining
a plasma with a high electron concentration (n e » 10 11 –10 12 cm –3 ) at
a relatively low temperature of the working gas (T » 100 ¸ 300°C).
Of special interest is the combined discharge 102 in which repeating na-
nosecond pulses with high amplitude are superimposed on a dc dis-
charge. The conditions of the optimum sustainment of ionisation were
examined in Ref.61.
Examination of ion–ion recombination

Analysis of the studies carried out in recent years shows that the prob-
lem associated with the measurement of the recombination coefficient
with the ion density increasing to 10 14 –10 15 cm –3 has not yet been
solved.
In Ref.95 and 96, it was proposed to examine recombination in F 2
and SF 6 at high ion concentrations on the basis of the afterglow of a
powerful nanosecond discharge developed in the form of a wave break-
down. Ion–ion plasma forms in the presence of effective ‘sticking’ in
the afterglow, and when a relatively short current pulse is fed to this
plasma the voltage drop in the near-electrode regions may be greatly
lower than the voltage drop in the plasma column. Defining the size
of the layer as r = µ i/lt, where µ i is the mobility of the ions, l is the
length of the discharge tube, U is the voltage drop in the plasma, and
calculating the voltage drop in the near-electrode layer from Poisson’s
equation, we obtain a criterion for the duration of the probing current
pulse
. l / J m i 10-7 ,
t < 14
where t is in s, l in cm, µ i in cm/(V s), J in A/cm 2.

To explain the conductivity of the plasma of the afterglow of the
nanosecond breakdown, the authors of Refs.95 and 96 carried out
503
Plasma Diagnostics
numerical modelling taking into account the main special features of

the elementary processes in F 2 and SF 6. The main annihilation chan-
nel of the ions at pressures in the range p = 60 ¸ 400 torr was as-
sumed to be the three-particle recombination. The effect of a possi-
ble slightly non-ideal nature of the plasma was taken into account by
analogy with electrolytes by introducing the activity coefficient and a
relaxation correction for mobility. The results show that 2–3 µs after
completing the discharge the ion concentration was 10 12–10 14 cm –3. In
Ref.95 and 96 a delay in the decrease of the conductivity of the plasma
in F 2 and SF 6 in an early afterglow was detected. This delay depended
on gas pressure.
The experimental results and the results of modelling plasma break-
down make it possible to conclude that non-ideal ion–ion plasma forms
in the early stages of breakdown. The effective constant of ion re-
combination in this plasma decreases.
Nitride synthesis
In Ref.94 the wave breakdown was used for synthesis of phosphorus
nitrides. The measurement of the amount of absorbed nitrogen for dif-
ferent durations and pulse repetition frequency shows that the rate of
synthesis is directly proportional to the repetition frequency of the
pulses. In the experiments, the rate of combination of nitrogen decreased
with time. It is evident that this was associated with the change of the
conditions in the gas-discharge device: formation of a nitride film on
the surface of phosphorus, formation of microarcs, etc. Figure 17.36
shows the dependence of the energy used for bonding the nitrogen
molecule on the inverse duration of the current pulse flowing through
the anode of the discharge device and averaged out for the first 30
minutes of synthesis during which there were no saturation effects.
The slope of the straight lines in Fig.17.36 shows that a decrease
ε /N, eV/mol
Fig.17.36 Energy used for binding one

molecule of nitrogen in relation to the
1/ τ , ns 1 reverse duration of current pulse.
504
of the duration of energy supply results in a large increase of the

efficiency of using this energy in synthesis. The dependence in Fig.17.36
is efficiently approximated by the formula E/N = A exp (– t/T), where
E is the energy used up for the absorption of N molecules of nitro-
gen, T is the nanosecond current pulse time, t and A are constants:
t = 17 ± 3 ns, A = 6000 eV/mol for a pipe with a diameter of 5 cm
and A = 365 eV/mol for a pipe with a diameter of 12 cm.
Comparison of the efficiency of synthesis of phosphorus nitride in
the wave breakdown with other types of gas discharge shows that the
efficiency of the wave breakdown is higher.
17.12 Conclusion
This review of the investigations of wave breakdown can be summarised
by the following comments.
1. The sub-nanosecond level of synchronisation of the channels of
recording the signals of electromagnetic radiation in a wide range of
the variation of the wavelength (10 –8 –10 2 cm) has been reached in
diagnostics in the last 10–15 years.
2. The complicated structure of the breakdown wave has been
established. In particular, it has been shown that at positive polarity
the front is conical, and at negative polarity it approaches sliding
discharge along the tube walls. The Doppler broadening of the
current pulse reflected from the front has been recorded. The vibra-
tional structure of the charge behind the front of the breakdown wave
was detected. The presence of fast electrodes in the vicinity of the
front of the wave with an energy corresponding to the double ampli-
tude of the voltage pulse was confirmed, together with the existence
of the limiting speed of the breakdown wave. This speed coincides with
the speed of the fast electrons.
3. It has been shown possible to use the wave breakdown as a means
of examining the elementary processes in non-equilibrium plasma
occupying an intermediate position between the glow discharge and the
electron beam. For example, measurements were taken of the diffu-
sion coefficient of atomic hydrogen, the energy efficieny of synthe-
sis of the molecule of phosphorus nitride in relation to the duration of
the nanosecond discharge, and the recombination coefficient in dense
ion plasma was also measured.
In conclusion, the authors would like to express their gratitude to
A.E. Sheindlin for attention and support of work in the area of the wave
breakdown and also would like to note the significant contribution to
these investigations from A.V. Kirillin, A.G. Abimov, L.N. Vasilyak,
S.V. Kostychenko, A.B. Matstenko, Yu.M. Tokunov, A.M. Ul’ynov and
I.V. Filyugin.
505
Plasma Diagnostics
Chapter 18
PROBE MEASUREMENTS OF POTENTIAL

DISTRIBUTION IN DENSE PLASMA
18.1 Special features of the method

To measure the potential in the arc discharge, a probe is introduced
into the discharge and the voltage in the probe is measured in rela-
tion to one of two electrodes. 1 The probe has the form of a section
of a thin non-insulated metallic wire which is moved in a strictly hori-
zontal plane at a certain distance from the electrode in the vertical arc
column normal to the discharge axis. The potential difference between
the probe and one of the electrodes is transferred to the deflecting plate
of a two-beam oscilloscope. The speed of the probe is varied from 40
to 200 cm/s. An inductanceless resistor is connected in parallel to the
probe measuring circuit. The resistor is used to shunt voltage surges
formed as a result of transient processes. Changing the nominal of the
resistor, it is possible to vary the probe current and measure almost
electrostatically the potential distribution in the cross section of the arc.
The sweep of the oscilloscope is synchronized with the movement of
the probe by an auxiliary probe which moves together with the measuring
probe at some distance from it. Selecting the sweep frequency and the
delay time of the signal from the auxiliary probe, it is possible to produce
a sufficiently stable image on the screen of the oscilloscope.
The main problem in probe measurements is associated with the per-
turbation of the discharge by the probe. This perturbation is of the ther-
mal and electrical nature. Perturbation of the arc by the probe was
investigated by Muller 2 using oscillographic and optical methods. The
results show that the extent of thermal perturbation depends on the
thickness of the probe and the characteristics of the arc region through
which the probe passes: as the thickness of the probe increases and
the current density in the plasma region intersected by the probe be-
comes higher, the intensity of perturbation increases. Perturbation in
the vicinity of the arc axis is the consequence of partial overlapping
of the main discharge channel by the probe and by the surrounding zone
506
Probe Measurements of Potential Distribution in Dense Plasma
with reduced electrical conductivity. The insertion of the probe into the
arc column changes stationary current and arc voltage. In contrast to
the thermal perturbation, formed at any polarity of the probe irrespective
of the presence of insulation of the probe, the electric perturbation is
observed only at the positive potential of the probe in relation to the
surrounding plasma when the plasma receives high current. Therefore,
the perturbation of the arc voltage increases with increasing probe cur-
rent. In the vicinity of the arc axis it is superimposed on the thermal
perturbation which does not depend in a wide range on the electric
perturbation intensity. The calculations determined the stationary strength
of the field in the arc and the width of the perturbation zone. The
strength of the axial electric field in the arc proved to be close to that
obtained from the results of probe measurements of the potential, and
the length of the perturbation zone was close to the width of the visible
dark space around the probe measured using photographs of the probe
trajectory in plasma.
In Ref.3, it was shown that the total perturbation, introduced by the
probe, can be evaluated from the oscillograms of the voltage and arc
current which are recorded together with the probe voltage. Consecutive
passage of the probe through one and the same cross section of the
arc column should generate the same potential at different passage
rates. Consequently, it is concluded that neither the thermal excitation
nor the turbulence of the gas, caused by the passage of the probe
through the arc column, has any significant effect on the measured
results.
The perturbation of the arc column by the probe was examined in
Ref.4. Firstly, the arc voltage and arc current were inspected prior to
and after entry of the probe into the arc column; secondly, the same
measurements were taken in the conditions of probing the arc with three
probes. Two probes were used to measure the strength and the third
probe, entering the arc column after the first two probes, was used
to measure the arc perturbation. The measurements show that the
strength of the field did not change after the entry of the perturbing
probe. The authors of Ref.4 concluded that probing is the most suit-
able method of measuring the voltage gradient in the arc.
The dependence of the nature of perturbation of the arc by the
probe on the parameters of the probe has been quite widely studied
because by selecting their probe it is possible to minimize the pertur-
bation of the arc. To minimise perturbation, it is recommended to use
a thin probe that rapidly travels through the arc column. The mate-
rial of the probe can be characterised by specific thermal conductiv-
ity and heat capacity. The results show that a tungsten probe approxi-
507
Plasma Diagnostics
mately 0.2 mm wide with the travel speed higher than 100 cm/s is most
suitable.
Muller and Finkelnburg 5 measured for the first time the distribution
of the potential along the arc axis and its peripheral part using a probe.
They show that the potential can be varied only when the resistance
of the external probe circuit is considerably higher than the resistance
of the plasma column between the probe and the corresponding elec-
trode. The latter depends on the position of the probe in the arc, be-
cause the conductivity of the plasma rapidly decreases from the axis
to the periphery. The authors of Ref.5 obtained the following results:
1. When the probe speed is higher than 100 cm/s, the probe cur-
rent through the perturbed area of the plasma is determined exclusively
by charge carriers drifting from the plasma, and the thermal emission
of the ions and electrons from the external surface of the probe is
insignificantly low;
2. As a result of the difference of the mobility of the electrons and
the ions, the resistance of the layer between the probe and the un-
perturbed plasma in contact of the probe with the anode is lower than
in contact with the cathode. Therefore, at a given resistance of the
circuit of the probe the region of measurement of the potential is al-
ways wider for the case of contact with the anode;
3. Regardless of the fact that the plasma potential changes along
the probe, the measured potential as a function of the distance from
the arc axis remains constant with the variation of probe current and
the probe requires the potential that corresponds to the point with the
highest conductivity of the arc column that is in contact with the probe;
4. Calculations and experimental examination show that the contact
difference of the potentials of the probe and the plasma is constant
along the arc axis and does not depend on plasma temperature; for the
argon arc this value does not exceed 2 V;
5. The probe method of measuring the potential using an electron
beam oscilloscope for recording is far more efficient than the method
in which the data are recorded using a loop oscilloscope.
Ringler 6 measured the radial distribution of the potential in the arc
column and proposed the following explanation. Investigations were car-
ried out for the arc examined extensively by other methods. The arc
was stabilised by the plasma jet leaving the cathode (current intensity
200 A). In addition to the probe potential, oscillographic recording was
made of the changes of the arcing voltage and arc current formed as
a result of perturbation of the arc by the probe. The method enabled
the author to determine the radial distribution of the arc potential at
any distance from the cathode with sufficient accuracy. The results
508
show that the size of the perturbation zones within the range of the
measurement error is in agreement with the diameter of the main cur-
rent-carrying arc channel. Starting with the equation for the density
of electric current through the plasma, the author of Ref.6 calculated
the radial distribution of the strength of the field for the examined arc.
Calculations confirm the experimental data.
The authors of Ref.7 measured the anodic voltage in the arcs in
relation to arc current and arc length. It is well known that the anodic
voltage drop is of considerable interest for the theory of near-anode
processes. Therefore, the results of measurement of the anodic volt-
age by the probe method are of considerable importance for developing
the theory of the anodic voltage drop. Up to now it was generally rec-
ognised that the anodic and cathodic volt drops remain constant when
the electrode gap is increased. This assumption was used as a basis
for developing the well-known method which makes it possible to de-
termine the strength of the field in the column from the increase of
arcing voltage with increasing arc length. However, Finkelnburg de-
termined 7 that the anodic voltage in the argon arc depends (in addi-
tion to current intensity) on the length of the arc which decreases with
a decrease of the arc length. With increasing plasma temperature in
the vicinity of the anode its potential decrease is linearly irrespective
of arc length and current density. To confirm this, two independent
methods were used to determine the strength of the field and the po-
tential drop along the arc column of different length. The results show
that the anodic drop which is associated with the density of anodic
current and plasma temperature increases with increasing arc length.
To confirm the dependence of anodic drop U A on arc length in argon,
an independent method was used to measure the axial distribution of
the strength of the field in the column. This was carried out by cal-
culating the distribution of electrical conductivity in the column using
the well-known distribution of temperature in the arc at a current of
200 A in argon. The results show that the anodic drop in the arc, burn-
ing in nitrogen at a current intensity higher than 50 A, decreases lin-
early with increasing current intensity. At the same time, the anodic
voltage in the high-temperature argon arc depends also on the arc length
and temperature, and when the arc length increases this drop becomes
larger, and decreases when the plasma temperature in front of the anode
increases.
The problem of the length of the section of the arc over which the
potential at the anode rapidly decreases has been examined in Ref.8.
The length of the region of the anodic drop was measured using the
following procedure: a tungsten probe, insulated with quartz, with a very
509
Plasma Diagnostics
short bare end was ejected with a special device at a speed of 10-
20 cm/s in the direction of the surface of the anode of a carbon arc.
Touching the anode surface the probe rapidly returned. The voltage
of the probe in relation to the anode was recorded with an electronic
oscilloscope. The results show that the axial length of the region of
the anodic drop in the low-current carbon arc is approximately equal
to the free path of the electron and amounts to 2 µm for the given
experiment conditions. 8
Finkelnburg and Segel 9 measured the distribution of potential inside
and around the column of the electric arc. It is well known that the
distribution influences both the axial and radial movement of the elec-
trons and the arcs and also the arc characteristics. The results show
that the data obtained in the probe measurements of the potential in
low- and high-current carbon arcs are in good agreement with the
conclusions of theoretical analysis and indicate that the length of the
transition region between the distorted potential field in the arc and
the non-distorted potential field outside the discharge is quite small.
When measuring the anodic and cathodic drops the probe was moved
manually or with an pneumatic mechanism up to contact with the cath-
ode or anode. The results of probe measurements of the potential inside
the arc column enabled the authors to find equipotential surfaces. The
appearance of the region with a large decrease of the potential in front
of the electrodes is explained by the presence of an excess charge in
them. As a result of the redistribution of the charges in the vicinity
of the electrodes, there is always an electric field with higher strength.
In the vicinity of the cathode under the effect of the radial compo-
nent of this field the electrons are displaced in the direction towards
the axis of the arc column. Thus, the normal mechanism of ambipo-
lar diffusion of charge carriers changes. This results in a decrease of
the width of the arc column in the near-cathode region. The radial elec-
tric field in the vicinity of the anode displaces the electrons to the column
periphery so that the column widens in the near-anode region.
18.2 Electric field of the high-current arc

Measurements were taken of the axial and radial distribution of the
strength of the electric field in the freely burning arc in the gas be-
tween cooled metallic electrodes. The effect of the resistance in the
circuit of a single probe from the results of potential measurements
has also been investigated. The arc diagram is shown in Fig.18.1.
The anode was represented by the bottom of a steel vessel cooled
inside with water, the cathode was a thoriated tungsten rod with a
diameter of 2.4 mm also intensively cooled with water. The arc was
510
water
gas
water
Fig.18.1 Diagram of the arc with argon shielding. 1) tungsten cathode, 2) copper anode.
Rp
Fig.18.2 Diagram of connection of a probe into the measuring circuit. 1) probe, 2) arc,
O – oscilloscope, R p is the variable resistance.
powered with a smooth voltage from a rectifier. The arc was stable
for a long period of time as a result of stabilising with an argon flow
(flow rate 5 l/min). The photographs show that the cross section of
the arc smoothing increases from the cathode to the anode.
The probe was in the form of a section of tungsten wire 0.2 mm
in diameter with the non-isolated section 2 cm long. The probe was
moved using a special device in the direction normal to the actual line
of the arc column strictly in one plane. The travel speed of the probe
was varied from 50 to 250 cm/s, with the measurements usually taken
at a speed of 220 cm/s. The diagram of connection of the probe to
the measuring circuit is shown in Fig.18.2. A variable resistor was used
to vary the intensity of probe current. The difference of the potential
511
Plasma Diagnostics
a b c
Fig.18.3 Oscillograms of the probe signal in the near-cathode region of the arc for
values of the variable resistor. R p = 1.5×10 6 (a), 5×10 6 (b), 1.5×10 7 ohm (c).
of the probe and one of the electrodes was supplied to the input of
the electronic oscilloscope. To estimate the extent of perturbation of
the arc by the probe, voltage at the arc electrode was also recorded.
Probing of the arc in its different sections was carried out by mov-
ing the entire arc device using a stand in which the electrodes were
secured. The distance between the probe and the cathode was con-
trolled in an area of ± 0.05 mm. The potential was measured in re-
lation to the cathode. The electrode spacing was 5 mm and was set
using a special template.
Typical oscillograms of the probe signal (Fig.18.3) were obtained
by intersecting the plasma column by the probe at a distance of 1 mm
from the cathode. At resistor resistances higher than 5·10 6 ohm the
oscillograms in Fig.18b, c, show a small increase of the potential in
the near-axial region of the arc column caused by the perturbation
caused by the probe. The intensity of the perturbation in the poten-
tial units is equal to the change of arc voltage. At low resistances of
the probe circuit, i.e. in this case less than 1.5·10 6 ohm, there are no
potential surges in the peripheral zones of the arc on the oscillogram
(1). The surges on the oscillogram appear only under conditions in which
the resistance of the probe circuit is considerably higher than the re-
sistance between the probe and the corresponding electrode (it will be
referred to as the arc resistance and denoted R a). The region between
these surges is the region in which the potential is measured (see
above). The width of the region of measurement of the potential in-
creases with increasing resistance of the probe circuit, i.e. the distortion
of the measured distribution of the potential in the peripheral regions
of the plasma decreases with increasing resistance of the probe cir-
cuit. The increase of the resistance of the probe circuit is restricted
by the sensitivity of the oscilloscope. It should be noted that the
oscillogram is used for determining the potential of the near-axial region
of the plasma because due to higher conductivity of the plasma in this
region of the arc the condition R p >> R a is fulfilled.
To determine the range of measurement of the potential, investigations
were carried out into the dependence of the half width of the meas-
512
V, V S, mm
log R a , ohm
Fig.18.4 Distribution of the potential along the arc axis (1) measured using a probe
and the width of the region of variation of the potential (2) in relation to resistance Rp
of the probe circuit.
ured radial distribution of the potential on the resistance of the probe

circuit (Fig.18.4, curve 2). These dependences are explained by the
rapid drop of the conductivity of the arc plasma in the radial direc-
tion. In our case at a probe resistance of R p = 10 7 ohm the width of
the region of measurement of potential is ~28 mm, although the ‘light’
diameter of the widest part of the arc does not exceed 6 mm.
To determine the minimum resistance in the probe circuit in which
it is still possible to measure reliably the potential of the near-axial regions
of the arc we constructed the graph of the dependence of the measured
potential of the plasma on the arc axis on R p (curve 1). The results
show that the latter increases linearly with the resistance of the probe
circuit and at some values of these resistance it ceases to depend on
it. The increase of the potential measured at the arc axis continues
until the resistors of the plasma between the probe and the cathode
is higher than the resistance in the probe circuit. Comparison of these
two graphs shows that at a resistance lower than ~10 3 ohm the re-
gion of measurement of the potential has a constant width of approxi-
mately 3 mm, and the measured potential at resistances higher than
~10 3 ohm tends to a constant value. It can therefore be concluded that
the resistance of the plasma column between the probe and the cathode
does not change when the probe moves within the limits of the cross
section 3 mm in diameter.
The estimates of the resistance of the arc column presented here
are intentionally too high because examination of the probe determines
the resistance of a narrow section of the arc column corresponding
to the diameter of the probe to which the resistance of the ‘cold’ part
of the arc column is then added. In addition, it must be taken into ac-
513
Plasma Diagnostics
count that only some part of the probe surface ‘takes part’ in the meas-
urements. Therefore, a more accurate value of the arc resistance can
be obtained using the ratio of the cross sections of the arc and the
probe and also the resistance of the ‘cold’ part of the arc. It should
be noted that Muller and Finkelnburg5 published too high values of the
resistance of the arc column (300 ohm). We believe that after con-
sidering the previous comments, the given method of determining the
minimum resistance of the probe circuit can be used to determine the
resistance of the arc column and, consequently, the electrical conduc-
tivity of the plasma.
Thus, to ensure that the probe can be used in the regime of meas-
urement of the potential, it is necessary to fulfil the condition
aR p >> R a. The radial distribution of the potential was examined at a
probe circuit resistance of 10 7 ohm. The measured distribution of the
potential was corrected by extrapolating it from the non-perturbed section
of the oscillograms to the axis of the arc along a parabola. This ex-
trapolation was used to determine the potential of the near-axial sections
of the plasma. The radii of the zones of perturbation of the arc, measured
for three cross sections, situated at distances of 1 and 2.5 and 4 mm
from the cathode, were respectively, 3, 4, 5 and 6 mm, i.e. they in-
crease in the direction from cathode to anode. This is consistent with
the observed form of the arc. Figure 18.5 shows the radial distribu-
tion of the potential measured by the probe in relation to the cathode.
The measured potential value includes the axial voltage drop in the arc
from the cathode to the examined cross section and the radial com-
ponent of the potential which depends on the distance from the probe
to the axis. For sections close to the cathode the potential increases
more rapidly. This is explained mainly by the nature of the radial dis-
tribution of the concentration of charged particles in the arc. The equi-
V, V
r, mm
Fig.18.5 Radial distribution of the probe potential at the resistance in the probe circuit
of 10 7 ohm for sections at a distance of 1 (1), 2.5 (2) and 4 mm (3) from the cathode.
514
Fig.18.6 Equipotential lines of arc discharge. V = 4.5 (1), 5 (2), 6 (4), 6.5 (5), 7 (6),
7.5 (7), 8 V (8).
potential V-lines of the arc column are shown in Fig.18.6. The dotted
line shows the observed contour of the arc column. These data are
in agreement with the results published in Ref.9. The anodic drop of
the potential between the cathode and probe where the latter inter-
sects the arc in the vicinity of the anode is equal to U a = U arc – U p
where U a is the anodic drop, U arc is the voltage on the arc electrodes,
Up is the difference of the potential between the cathode and the probe.
The anodic voltage in the examined arc is 6.5 V at an arcing voltage
of 13 V which is in good agreement with the data in Ref.9, 10, ac-
cording to which in examining an arc with a current of 50 A and an
electrode gap of 5 mm the anodic drop was 6 V at a total arc volt-
age of 15–16 V.
18.3 Electric field in the laminar plasma jet of the electric

arc plasma torch
Measurements of the potential in the flow of the plasma in the ‘laminar’
plasma torch 11 were taken at an arc current of 80 A and an arc voltage
of 80 V, the flow rate of plasma-forming argon was 0.18 g/s. The probe
was in the form of a section of tungsten wire 0.2 mm in diameter, the
length of the non-insulated part was 2 cm, and the section periodically
intersected the plasma jet in the direction normal to its axis. The speed
of the probe (~4 m/s) was such that the temperature of the probe
remained almost constant during the intersection of the plasma flow,
and electron emission from its surface can be ignored. The probe signal
was recorded using an oscilloscope through an in-series connected
515
Plasma Diagnostics
resistance of 5 MΩ. The potential received by the probe in plasma in

relation to an earthed cathode was measured. The tungsten wire had
no insulating coating so that the probe was in contact with areas of
plasma having different potential at the same time.
Typical oscillograms of the probe signal are shown in Fig.18.7. Probe
measurements at the outlet of the nozzle did not show any high-fre-
quency pulsations of the electric signal. In addition, when the orien-
tation of the probe in relation to the cathode spot of the arc was changed
electric signal returned it’s symmetry. This indicates that electric arc
is not blown away by the gas flow from the nozzle of the plasma torch.
The results obtained in processing the oscillograms were used to
plot the distribution of the potential acquired by the probe in the plasma
jet (Fig.18.8). The results show that the electric field is observed at
relatively large distances from the outlet of the nozzle (see also Ref.12
and 13). When explaining the resultant dependence, it should be taken
into account that the strength of the probe signal depends not only on
the plasma potential but also on the contact difference of the probe-
plasma potentials and on the ratio of the resistance of the cathode-
probe R c.p and the resistance in the probe circuit 9. The probe oper-
ates as a voltmeter only when R p >> R c.p . For near-axial regions of the
plasma jet this condition is obviously fulfilled. 14 We introduced a cor-
rection for the contact difference of the probe–plasma potentials whose
value is a function of temperature and is determined by the follow-
ing expression: 1 where T is plasma temperature, M is the ion mass,
m, e is the mass and charge of the electron, respectively, k is the
Boltzmann constant:
kT M
DU = - ln ,
2e m
The corrections for the measured values of the potential were cal-
culated taking into account the known distribution of the temperature
in the jet. The corrected distribution of the potential along the axis of
5 V
a b c
Fig.18.7 Typical oscillograms of the probe potential at a distance of 5 (a), 30 (b) and
60 mm (c) from the outlet of the nozzle.
516
U, V U, V
b
a
z, mm z, mm
Fig.18.8 (Left) Distribution of the electric

potential of the probe in the plasma jet. U =
6.25 (1), 6 (2), 5.75 (3), 5.5 (4), 5.25 (5), 5
(6), 4.75 (7), 4.5 (8), 4.25 V (9); broken line
shows the growing corona of the jet.
Fig.18.9 Measured and calculated – on the basis
of the ambipolar diffusion (1) taking into
account the contact difference of the probe-
plasma potentials (2) – axial (a) and radial (b)
distribution of the potentials of the electric field
in the plasma jet.
the jet and in the direction of the radius at the outlet of the nozzle from
the plasma jet represented in Fig.18.9. The plasma potential at the outlet
of the nozzle is U 0 = 12.8 V. Since the results of probe measurements
show that in the plasma torch with a narrow nozzle the electric arc
is not blown out in the form of a loop, this potential is evidently de-
termined by the cathode drop in the electric arc.
It will be assumed that the variation of the potential in the plasma
jet is caused by ambipolar diffusion. The gradient of the potential formed
in the plasma as a result of ambipolar diffusion is 1
1
grad U amb = grad ( N e kT ). (18.1)
eN e
Integrating equations (18.1) from the point with the temperature T 0 to

the point with the temperature T gives
517
Plasma Diagnostics
z
T
k ¶ ln Ne kT
Uamb (T ) = U0 - dT.
e ¶ ln kT (18.2)
T0
Here U 0, T 0 is the potential and temperature of the plasma at the axis
of the jet at the outlet of the nozzle. The distribution of the potential,
calculated from equation (18.2), and determined by ambipolar diffu-
sion, is shown in Fig.18.9 which indicates that there is satisfactory
agreement between the measured U and calculated U amb potential dis-
tributions. It can therefore be assumed that the potential of the plasma
jet is determined by the cathode drop in the electric arc and ambipolar
diffusion and the plasma jet itself is without current.
18.4 The electric field of the plasma flow of a two-jet

plasma torch
The measurements were taken at the following parameters of the two-
jet plasma torch: arc current 105 A voltage 145 V, total flow rate of
the plasma forming argon for both heads 0.12 g/s, initial angle of con-
vergence of the plasma jet 60°. 15
Experimental equipment consisted of a horizontal table with a
device for securing and rotating the probe and a C1-55 two-beam
oscilloscope. The probe was in the form of a tungsten wire section 0.2
mm in diameter with the length of the non-insulated part being 2 cm.
To increase its stiffness, the wire was placed in a thin quartz capil-
lary tube. The non-insulated part of the wire was inserted into the
plasma. The capillary was secured to a metal rod 70 cm long. The probe
signal was fed to the input of the oscilloscope through a mercury con-
tact. To vary the probe current, an alternating resistor with a maxi-
mum resistance of 1 MΩ, equal to the input resistance of the
oscilloscope was connected in parallel to the input terminals of the
oscilloscope.
The linear speed of the probe was 5 m/s. This speed was selected
on the basis of experiments taking into account requirements on the
absence of thermal emission, evaporation of the probe material and the
minimum dynamic disruption of the plasma. In accordance with Ref.3,
the absence of thermal emission of electrons and ions was inspected
on the basis of degree of symmetry of the probe signal. The probe
potential was measured in relation to the cathode (when connected to
the anode, the signal amplitude was not measured). The multiple passage
of the probe in one cross section also did not cause any changes in
the pattern of the radial distribution of the potential. The form of the
oscillograms for the jets and the plasma flow is shown in Fig.18.10.
The oscillograms of arc voltage are also given in here. To construct
518
1 ms
5 V
1 V
1 ms
20 V b
1 V
Fig.18.10 Typical oscillograms of

2 ms c the probe potential (1) and arc
10 V voltage (2) for the cathode (a) and
anode (b) jets at a distance of zero
from the outlet of the nozzle of the
1 V plasma torch, for the plasma flow
after merger of the jet (c) 30 mm
above the line connecting the
outlets of the nozzles of the plasma
torches.
the field of the equipotentials of the jet the plasma flow was probed
in different sections. The measurement of the potential in the anode
and cathode jet was carried out separately, and the plane of displacement
of the probe was always normal to the direction of the discharge of
the jets and the plasma flow. At the same time, arc voltage was
inspected when the arc column was intersected by the probe. The resistor
in the probe circuit was represented by the input resistance of the
C1-55 two-beam oscilloscope (R p = 1 MW). A V7-16 digital volt meter
was used to measure the distribution of the potential in the diaphragms
of both heads of the plasma torch in relation to the cathode. The
results of measurements in the form of the field of equipotential are
presented in Fig.18.11. The distribution of the density of the electric
current between the plasma jets was found from the measured strength
of the electric field and the electrical conductivity of the plasma
calculated from the measured temperature (~7000 K). The results are
presented in Fig.18.12 which shows that the current density does not
exceed 1.5 A/mm 2 .
519
Plasma Diagnostics
Fig.18.11 Distribution of the electric potential of the probe (in volts) in the plasma
flow (1). 2 – current passage area, 3 – line of the lowest electrical resistance of the
plasma.
Fig.18.12 Distribution of the density of electric current in the gap between the plasma
jets (coordinate z – see Fig.18.11).
520
The experimental results show that the arc discharge in the con-
vergence zone is stationary and of the diffusion type up to a height
of approximately 25 mm with a current density of 1 A/mm 2. Here the
discharge is evidently non-independent because the ionisation condi-
tions are determined to a large degree by the external inflow of heat.
The zone of convergence of the jet is heated by the plasma jet by
conductive and convective heat flows ensuring the diffusion passage
of currents advantageous from the energy viewpoint. The volt equivalent
of this zone is 5–7 V which equals 3–5% of the total arc voltage. This
is also indicated by the absence of contracted plasma channels between
the jets. The absence of high voltage jumps in the arc also indicates
the stationary nature of current passage. The authors of Ref.16 and
17 examined the behaviour of the arc in a long cylindrical channel for
turbulent arcing. Film frames of the arc show clearly the areas of rup-
tures of the arc cord. It can therefore be assumed that a similar mecha-
nism of the diffusion passage of current can also be realised in this
case.
On the basis of the results of probe measurements, the potential in
the region of current passage at Z = 59 mm is equal to 66 V. At
Z = 64 mm and higher, the potential decreases to 64 V and remains
unchanged up to Z = 230 mm.
521
Plasma Diagnostics
Part IV
METHODS OF PROCESSING RESULTS
AND AUTOMATION OF DIAGNOSTICS
Chapter 19
REDUCTION TO A UNIFORM LAYER IN

AXISYMMETRIC OBJECTS
In optical diagnostics of electric arcs, pulse discharges and plasma flows
measurements are taken of certain integral quantities which can then
be used to determine the local optical parameters of the examined object.
These objects are usually characterised by rotational symmetry, and
by taking into account the non-uniformity (reduction to the uniform
layer), the problem is reduced to the numerical solution of Volterra
integral equations, especially of the Abel’s type. The solution of these
equations with the right part unperturbed does not cause any princi-
ple difficulties. However, in reality, the experimental data are always
loaded with random errors. Consequently, the results of the solution
will contain very large errors. Therefore, special attention should be
paid to smoothing the initial data.
It is also important to develop a rational logarithm with a minimum
computation time. This is determined by the complicated expression of
the Volterra kernel. Finally, to apply the results in practice, it is nec-
essary to evaluate their accuracy. Here we describe a method 1,2 de-
termining the spectral coefficients of emission and absorption in non-
uniform axisymmetric objects that satisfies these requirements to a large
degree.
19.1 Smoothing experimental data

The experimental function y(x) will be approximated by expansion to
a generalised Fourier series
522
Reduction to a Uniform Layer in Axisymmetric Objects
M
y ( x) » Y ( x) = å Ak P2k ( x), (19.1)
k =0
where P 2k (x) are the even Chebyshef polynomials, orthogonal with the
weight ρ (x) on the system of points x i, i = D 0, n , x Î -1, 1 and de-
termined by recurrent relationships
P2 k ( x ) = P2 k - 2 ( x ) - (q2 k - 2 / q2 k -1 ) xP2 k -1 ( x ),
P2 k +1 ( x ) = xP2 k ( x ) - (q2 k / q2 k - 2 ) P2 k -1 ( x ),
P0 ( x ) = 1, P1 ( x ) = x , - 1 £ x £ 1,
n n n
qk = å ri Pki2 , å ri Pki Psi = 0, å ri = n + 1, (19.2)
i =0 i =0 i =0
ri = r( xi ), Pki = Pk ( xi ).
In accordance with the principle of least squares, the Fourier coeffi-

cients are
n
Ak = q2-k1 å ri P2 k ,i yi , yi = y ( xi ). (19.3)
i =0
The measure of scattering of the initial data y(x i) in relation to the ap-
proximating function Y(x) is the residual dispersion
n F ry - q
n M I.
DM = (n - M ) -1
å ri ( yi - Yi )
i=0
2
= (n - M ) -1
GH å å
i=0
2
i i
k =0
2
2 k Ak JK (19.4)
The number of the terms of series (19.1) is optimised in accordance

with the functional dependence and the accuracy of the initial data.
The following criterion is highly suitable for selecting M irrespective
of the number of points, dispersion and weight of the measurements.
The number of the terms of the series (19.1) is restricted by the mini-
mum value of M at which the residual dispersion (19.4) ceases to de-
crease, namely as soon as the conditions
D M £ D M +1 , D M £ D M + 2 .
523
Plasma Diagnostics
are fulfilled. The approximating function Y(x) (19.1), representing the

smoothed-out value of the experimental function y(x), is used in further
calculations.
The advantage of Chebyshef’s polynomials in addition to the pos-
sibility of taking into account the weight of measurements is that the
initial data can be represented with an arbitrary step along the x axis.
In the absence of information on the weight of measurements, the ex-
perimental data are assumed to be of equal accuracy.
19.2 Numerical solution of Abel’s integral equation

From the solution of the integral Abel equation we determined the radial
distributions of the refractive index, the absorption factor and the emis-
sion factor in an optically thin source and a source with moderate self-
adsorption or with spatially distributed emission and adsorption zones.
The integral Abel equation (using an example of the required coeffi-
cient of emission in an optically thin source) has the form
z
1
e(r )rdr
2R = I( x) (19.5)
x r2 - x2
Here and later we shall use the reduced co-ordinates 0 £ r £ 1, –1

£ x £ 1, and e(r ³ 1) = 0, I (|x| ³ 1) = 0.
The integration range in (19.5) 0 £ r £ 1 will be divided into N zones
of equal size and it will be assumed, to simplify considerations, that
e(r) in the zones is constant. The integral equation (19.5) acquires the
form
z
p rj -1 p
å 2 Re j 2
(r - x12 )1/ 2 rdr å apj e j = I p ,
=R p = 1, N (19.6)
j =1 rj j =1
where
a pj = 2 (r j2-1 - x 2p )1/ 2 - (r j2 - x 2p )1/ 2 ,

e j = e(r j -1/ 2 ), I p = I ( x p ), r j = x j = (1 - j ) / N .
The system of algebraic equations (19.6) with a triangular matrix has

the solution
524
p -1
e p = R I p / a pp -
1
å (a pj / a pp )e j , j = 1, N . (19.7)
j -1
Equation (19.7) can be written in the form
p
e p = R -1 å bpj I j , j = 1, N , (19.8)
j =1
where
p -1
b pj = - å (a ps / a pp )bsj , b pp = 1 / a pp .
s =1
Instead of the experimental values of I j the corresponding values

of the approximating function Y(x j) (see (19.1)) are substituted into
equation (19.8).
We shall estimate the random error of the calculated results. We
shall write the solution of the integral Abel equation (19.8) taking into
account (19.1) and (19.3) (y j will be replaced here by I i) in the form
F I
å G år JK
n M p
ep = 1/ 2 -1
å bpj P2k , j r1i / 2 Ii / R.
H
i =0 k =0
i P2 k , i q2 k
j =1
(19.9)
Assuming that the errors of the measurements are random and inde-
pendent, in accordance with the linear equation (19.9) we obtain the
following expression for the dispersion e (r):
F I 2
D(e ) = å G år JK b g
n M p
1/ 2 -1
åbpj P2k, j ri D Ii / R .
p
H i =0 k =0
i P2k ,i q2 k
j =1
Taking into account that r i×D (I i/R) = D (I/R), where D (I/R) is the
weighted dispersion of the initial data, and taking into account the dis-
tribution of q2k and the orthogonality condition in accordance with (19.2),
we obtain
525
Plasma Diagnostics
F
D(e ) = D( I / R)å q G å b
I,
2
JK
M p
-1
p = 1, N.
p
Hk =0
2k
j =1
pj P2 k , j (19.10)
As a measure of the accuracy of the initial data it is recommended

to use the dispersion of the value I/R because the intensity I is pro-
portional to the radius of the source R, and the measurement error
represents some fraction of the intensity. When using non-smoothed
values of Ij in equation (19.8) for measurements of equal accuracy we
directly obtain
d i
p
D e p = D( I, R) åbpj2 . (19.11)
j =1
We determine the coefficient of transfer of the error S (r) of the initial

data to the calculated results by the equation
S 2 (r ) = D( e(r )) / D( I / R ). (19.12)
Calculations carried out using equation (19.10) show that for the axis
of the source (r = 0) we can approximately write
S 2 (0) » ( M + 1) 3 / 3(n + 1). (19.13)
The interpolation method for the case when the initial data are not
n
smoothed out corresponds to the condition N = = M , and
2
S 2 (0) » (1 / 6) N 2 , (19.14)
which also follows from the calculations carried out using equation
(19.11). On the basis of the equations (19.13), (19.14) and Fig.19.1,
we can draw the following conclusions. The number of zones N in con-
version of (19.18) with the presmoothed initial data has only a slight
effect on the stability of the solution. This makes it possible to use a
simple quadrature (constancy of e in the zones) and use a relatively
large value N (~50–100) for reducing the systematic error. The use
of the Fourier series results in the stability of the solution whose ac-
curacy in comparison with the interpolation method increases with in-
526
Fig.19.1 Coefficient of transfer of the error in conversion of Abel transformation calculated

from equations (19.10), (19.12). n = 100, p = 1, N =50 (1), 100 (2), N = 25 (3),
M = 5 (4), 10 (5), 20 (6).
creasing number of the points for which the initial information is specified
(under the condition of non-correlation of their errors) and with a de-
crease of the number of the terms of the series required for approxi-
mating the experimental function.
19.3 Numerical solution of the Volterra integral equation

In the case of an optically dense source the coefficients of emission
e(r) and absorption k(r) are linked with the observed intensity I (x)
and transparency w(x) by the integral relationship
F I
z z ch G JJ
1 r
k(r ¢)r ¢dr ¢ e(r )rdr I( x)
2P
x
GH x r¢ - x
2 2
K r¢ - x
2 2
=
w( x )
, (19.15)
z
1
k(r)rdr
2R = - ln w( x) º t( x), (19.16)
x r2 - x2
where k(r ³ 1) = 0; w (|x| ³ 1) = 1. For the maximum value of the

hyperbolic cosines we can write
527
Plasma Diagnostics
LM OP 1 + w( x)
z
1
k(r )r dr
ch
MN x r2 - x2 PQ = 2 w( x) .
Estimates carried out using this equation show that the hyperbolic
cosine can be assumed to be equal to unity with the error smaller than
6% at w > 0.5 and less than 1% at w > 0.8. If self-absorption in the
central emitting part of the source is ignored, the hyperbolic cosine is
equal to unity, regardless of the level of absorption on its periphery.
This is indicated by the fact that the integral below the sign of the hy-
perbolic cosine in (19.15) is taken in the region where e (r) differs
from zero. Thus, with moderate self-absorption and in the case of spa-
tially distributed zones of emission and absorption, the equation (19.15)
assumes the form of the integral Abel equation
z
1
2 R e(r) (r 2 - x 2 )-1/ 2 rdr » I( x) / w( x).
x
The solution of this equation in relation e(r) can be determined from

the previously described algorithm with the approximation of the ex-
perimental function I ( x ) / w ( x ) by the Chebyshef polynomials. If these
approximations are not uniform, it is necessary to solve the equation
(19.15) in relation to e (r) determining in advance k (r) from the in-
tegral Abel equation (19.16). We shall write the integral equation (19.15)
in the form
LM OP
z z
1 r
e( r )
2R
k (r )
x
ch
MN x
PQ
k(r ¢)(r ¢ 2 - x 2 ) -1/ 2 r ¢dr ¢ k (r )(r 2 - x 2 )-1/ 2 rdr = I ( x )w -1/ 2 ( x )
The integration range 0 £ r £ 1 will be divided into N zones of equal

size and it will be assumed that e (r) and k (r) in the zones are constant.
Consequently, we obtain
528
LM OP
z z
j rp -1 r
å 2( e p / k p ) ch
MMN k (r ¢ )(r ¢ - 2
x 2j ) -1/ 2 r ¢dr ¢
PPQ k(r)(r
2
- x 2j ) -1/ 2 rdr =
p =1 rp xj
z
j e j ; p -1 j
å 2( e p / k p ) ch tdt = å 2(e p / k p )(sh t j; p-1 - sh t jp ) =
p =1 t jp p =1 (19.17)
j
å c jpe p = R -1I j w -j 1/ 2 , j = 1, N ,
p =1
where
c jp = (2 / k p )(sh t j ; p -1 - sh t j; p ),
j
t jp = å (rt2-1 - x 2j )1/2 - (rt2 - x 2j )1/2 , k p = k (rp -1/ 2 ),
t = p +1 (19.18)
e p = e(rp -1/ 2 ), I j = I ( x j ), w j = w ( x j ), rp = x p = 1 - p / N .
The system of algebraic equations (19.17) with a triangular matrix has

the solution
j
ej = R -1
I j w -j 1/ 2 / c jj - å (c jp / c jj )e p , j = 1, N . (19.19)
p =1
Prior to calculating e(r) from (19.19), the experimental function

I ( x ) / w ( x ) is smoothed out using the Chebyshef polynomials. Using
the procedure identical with that described previously for the integral
Abel equation, we obtain the following estimate of the dispersion of
the emission coefficient in the case of an optically dense source:
F
)å q G å e
I,
2
JK
M j
D(e j ) = D( R-1Iw -1/ 2 -1
j = 1, N,
k =0 H 2k
p=1
jp P2 k , p
where
529
Plasma Diagnostics
Fig.19.2 Test functions (19.22).
j -1
e jp = å dc js / c jj i esp , e jj = 1 / c jj,
s= p
and the coefficients c jp, are determined by equation (19.18). Equation

(19.20) gives the lower estimate of the dispersion of the emission fac-
tor, because it does not take into account the effect of the absorption
factor included in cjp. However, as shown by the calculations, the effect
of this error is not large because in calculations of ε we use the in-
tegral from κ. The absorption factor is calculated from Abel's inte-
gral equation, and we have
F I 2
GH å b JK
M j
D (k j ) = D ( R -1
ln w ) å q2-k1 pj P2 k , p , j = 1, N . (19.21)
k =0 p =1
The results of calculating the variance factor for test functions (Fig.
19.2) are:
d
e( r ) = ( -15 p ) -1 1 - r 2 i 1/ 2
(38 - 176r 2 + 288 r 4 ),
k (r ) = g e(r ),
I ( x ) = g -1 1 - w ( x ) , (19.22)
w ( x ) = exp - g (1 - 3 x + 8 x - 6 x )
2 4 6
are presented in Fig.19.3. The value of the factor g was used to specify
530
Fig.19.3 Relative error of restoration of the coefficients of absorption (1) and emission
(2,3).
different values of transparency w (0) = 0.1 ¸ 0.9. It was assumed

that the relative random error of experimental functions I (x) and w
(x) is 3%, N = 80, N = 100. As indicated, it can be seen that the trans-
mission coefficient in this example is determined with satisfactory
accuracy up to w (0) = 0.1. The absorption coefficient is determined
with a large error.
We assume that for the examined object we know the dependence
k = k ( e), (19.23)
and for example, the function e (T) or the radial distribution at tem-
perature T (r) are specified. Consequently, on the basis of the
Kirchoff’s law we can write k = e/B, where B is the Planck’s func-
tion. Therefore, it is not necessary to measure the transparency and
the emission coefficient is determined from the measured intensity from
the solution of the equation (19.15) with an allowance made for (19.23).
Equation (19.19) is transformed with this purpose for the iteration proc-
ess. For each zone, starting with the first one (j =1), the accuracy of
the solution is improved by iteration taking into account the depend-
ence (19.23) and is then used as the zero approximation for the fol-
lowing zone. The corresponding algorithm has the form
j -1
e vj = R -1 (w vj ) -1/ 2 / d vjj - å (d vjp / d vjj )e p , j = 1, N ,
p =1
(19.24)
d vjj = (2 / k vj ) sh t vjj , d vjp = (2 / k p )( sh t vj , p -1 - sh t vjp ),
531
Plasma Diagnostics
w vj = exp( -2 t vjp ),
j -1
t vjp = k vj (r j2-1 - x 2j )1/ 2 + å k t (rt2-1 - x 2j )1/2 - (rt2 - x 2j )1/2 ,
t = p +1
k vj = k (e vj -1 ),
v = 0, j = 1: d11
0
= 2( R 2 - x12 )1/ 2 , t10, 0 = 0, (19.25)
v = 0, j = 2, N : k 0j = k j -1 ,
I j = I ( x j ), k j = k (r j -1/ 2 ), e j = e (r j -1/ 2 ), x j = r j = ( N - j ) / N.
As previously, the experimental distribution of the intensity is

smoothed out by Chebyshef’s polynomials. The dispersion of the cal-
culated results when taking into account self-absorption by iteration is
evaluated as follows
F f I 2
GH å JK
M j
D(e ) = D( I / R)å
j q2-k1 jp P2 k , p , j = 1, N, (19.26)
k =0 p =1
where
j -1
f jp = å (d js / d jj ) f sp , f jj = w -j 1/ 2 / d jj ,
s= p
and the values d jj , d js , ω j are given after completing iterations. Fig-

ure 19.3 (curve 3) gives the relative error of iteration restoration of
e (0) on the axis of the source for the example (19.22) calculated using
(19.26). The accuracy of restoration of the emission coefficient with
self-absorption taken into account by iteration is satisfactory but on
the whole is lower than in the case in which we measure not only the
intensity (curve 3) but also transparency (curve 2).
19.4 Simplified reduction algorithms

The previously described algorithm of reduction to the optical thin layer
is characterised by high stability in relation to random measurement
errors, relatively small systematic error and moderate requirements on
the calculation time using computers. At the same time, the method
532
of numerical solution of the integral Abel equation are used widely, they
suitable both for application in a computer and manual calculations. These
methods ensure sufficient accuracy of the results of the calculations
for the experimental data obtained with a small error.
Numerical conversion of the Abel transformation

To calculate e (r), we shall use the conversion of the Abel transfor-
mation
z
1
1 dI dx
e(r ) = - , (19.27)
pR dx x - r2
2
r
0 £ |x| £ 1, 0 £ r £ 1, (r ³ 1) = 0, I (|x| ³ 1) = 0. The integration

region in (19.27) 0 £ x £ 1 will be divided into N zones of equal size
[x j, x j+1], x j = ( N - j ) / N , j = 1, N . I (x) in the zones will be approxi-
mated by the interpolation polynomial of the third degree with the ex-
ception of the first and third zones where we shall use the interpo-
lation polynomials of the second and fourth (with respect to even ex-
ponents x) degree
R| Ax 2
+ Bx + C, x Î x2 , x1 ,
I ( x ) » P( x ) = S Ax 3
+ Bx 2 + Cx + D, x Î x j , x j -1 , j = 2, N - 1,
|T Ax 4
+ Bx 2 + C, x Î x N , x N -1
(19.28)
The coefficients of the polynomial are determined from the condition

of coincidence of values of the functions of I (x j) and P (x j) at the
nodes x j. The integral in (19.27) will be represented by a sum of the
integrals with respect to the zones and I (x) will be replaced by cor-
responding polynomials P (x):
e(rp ) = -
1
å
pR j =1 z
p x j -1
xj
dP
dx
dx
x - rp2
2
.
We shall then carry out analytical integration and reducing with respect
to I (x j), we obtain
533
Plasma Diagnostics
N
1
ep =
R å g pj I j , p = 1, N , (19.29)
j =1
where
N-p N-j
e p = e(rp ), I j = I ( x j ), rp = , xj = .
N N
The coefficient g pj can be computed in advance and tabulated. The

dispersion of the results of the calculations used in (19.29) for meas-
urements of equal accuracy without correlations is
N
D ( e p ) = D( I / R ) å g 2pj . (19.30)
j =1
The coefficient of transfer of the error (19.12) for this method is rela-
tively high (19.4). Its value for the axis of the source (r = 0) is close
to the estimate given previously (equation (19.14)). To reduce the effect
of random errors without complicating the processing method, we shall
carry out local smoothing of the experimental data. We approximate
each group of seven points by a polynomial of the third degree using
the method of least squares. The corresponding smoothed-out values
are
R| 2 I - 1 I - 2 I + 1 I + 8 I + 19 I , j = 1,
|| 211 I - 62 I +211 I +72 I +2119 I +428 I , j = 2,
6 5 4 3 2 j
I ( x ) = S 42 6 5 4 3 2 1
i
|| - 2 I 21+ 1 I 21 + 2 I7 + 142I + 2 I21 + 1 I - 2 I j +3 , j = 3, 7,
|T I 21 = I 7, I 7= I , 3I =7 I . 7 21
j -3 j -2 j -1 j j +1 j +2
N +3 N -3 N +2 N -2 N +1 N -1
(19.31)
I j in equation (19.29) will be substituted by the corresponding smooth

values of I j from (19.31) and reducing the like with respect to I j, we
obtain
N
e p = R -1 å G pj I j , p = 1, N . (19.32)
j =1
534
Fig.19.4 Coefficient of transfer of the error for the interpolation method (19.29) of
numerical conversion of Abel's transformation (1) and for the improved algorithm (19.33)
(2) at N = 40.
Consequently, the smoothing procedure is included in the coefficients

G pj and this is carried out simultaneously with conversion of the Abel
transformation. The values of the coefficients G pj for N = 40 are
presented in Ref.1. The tabulated coefficients can be used at any
number N ¢ £ N = 40 of zones of uniform division of the region
[0, R]. Consequently
N¢
N¢
ep = -
40 R å G pj I j ,
j =1
(19.33)
x j = ( N ¢ - j ) / N ¢ , rp = ( N ¢ - p ) / N ¢ , j , p = 0, N ¢ - 1.
The dispersion of the calculated results using equation (19.33) for non-
correlated measurements of equal accuracy is
N¢
D( e p ) = D( I / R ) å G pj2 , (19.34)
j =1
The corresponding coefficient of transfer of the error (19.12) is shown

in Fig.19.4. The addition of the smoothing procedure to the calcula-
tion method decreases the error of the calculated results 2–3 times.
Processing schlieren measurements

The radial distribution of the refractive index in schlieren measurements
is calculated using the equation
535
Plasma Diagnostics
z
1
1 q( x )dx
n(r) - nR = , (19.35)
p x2 - r2
r
where
0 £ r £ 1, 0 £ x £ 1, n(r ³ 1) = n R , q( x ³ 1) = 0.
The integration region 0 £ x £ 1 will be divided into N zones of equal

size. q (x) in the zones will be approximated by interpolation polyno-
mials (19.28). Consequently, after the corresponding transformations
and calculations we obtain
p +1
n p - nR = å C pj q j , p = 1, N - 1, (19.36)
j =1
where
N-p N-j
n p = n(rp ), q j = q( x j ), rp = , xj = .
N N
The coefficients C pj at N = 40 are presented in Ref.1. The coef-

ficients of transfer of the error are given in the same reference
S 2p = D(n p - nR ) / D(q),
where the dispersion of the calculated results follows from equation

(19.36) and is equal to
p +1
D( n p - n R ) = D( q) å C pj2 . (19.37)
j =1
The numerical values of S p are lower than unity even without using
the smoothing procedure of the initial data. This is due to the fact that
in the schlieren method we measured the angle of deviation of the light
beam whose value is determined by the radial gradient of the refractive
index. The result is that in the conversion of the Abel transformation
(19.35) there is no differentiation of the experimental function. Con-
536
sequently, it is the presence of the derivative from the experimental

function that determines the instability of the numerical conversion of
the Abel transformation (19.24). Naturally, the same instability is also
manifested when solving the Abel integral transformation (19.5).
537
Plasma Diagnostics
Chapter 20
RECONSTRUCTION OF VELOCITY DISTRIBUTION

FUNCTIONS OF EMITTING PARTICLES FROM THE
SHAPE OF THE CONTOUR OF SPECTRAL LINES
Recently, the attention of the investigators has been attracted by ex-
amination of various types of non-equilibrium processes and gases in
plasma. This is due to both the development of a large number of
applications where non-equilibrium phenomena are observed to vari-
ous degrees (physics of gas phases, dynamics of rarefied gases, plasma
chemistry, astrophysics, etc.) and also to the solution of purely fun-
damental problems in the nature of various physical phenomena. 1,2 In-
vestigations of this type relate to both the kinetics of formation of the
energy distributions of the internal state of the molecules – electronic,
vibrational and rotational, and also the kinetics of translational energy.2
It should be noted that whereas the experimental facts of manifesta-
tion of strong nonequilibria in internal degrees of freedom of molecules
are already well known (see, for example, (Ref.23)), the investigations
of nonequilibria in the velocity distributions have been carried out only
recently. Obviously, this circumstance reflects the fact that the processes
of the translational relaxation of heavy particles takes place with the
highest speed in comparison with the speed of all previously mentioned
internal degrees of freedom and, usually, is already completed when
the latter start to relax. 2,3 It is therefore interesting to examine the
special features of the translational non-equilibrium distributions and
the rates of their relaxation to the equilibrium Maxwell distribution. At
the same time, there are two main experimental methods of measur-
ing the velocity distribution functions of heavy particles: the method
of Doppler broadening of the spectral lines, and the time-of-flight
method. Without discussing the advantages and disadvantages of these
methods, we shall mention only the special features of obtaining the
velocity distributions from the values mentioned in the experiments.
1. The method of Doppler broadening of the spectral lines is based
on extracting information from the measurements of the shape of the
538
Reconstruction of Velocity Distribution Functions of Emitting Particles...
contours of the lines of atoms and molecules. It is well known that

the distribution of intensities in the Doppler contour of the line for an
homogeneous isotropic medium is associated with the function of
distribution of emitters with respect to the absolute values of the ve-
locity P (v) by the following integral equation: 4,5
z
¥
P(v)
dv = j( n), (20.1)
v
m
n - n0
where m = c , n 0 is the emission frequency of a stationary par-
n0
ticle, c is the velocity of light.
However, in practice, the actual contour of the spectral line ϕ(v)
is not measured in the explicit form but as a convolution with some
apparatus function a (v) of the measuring system (as, for example, in
examining the contours of the lines using a Fabry–Perot interferom-
eter 6
z
¥
a( n - n¢) j ( n¢) dn¢ = f ( n), (20.2)
-¥
where f (n) is the measured form of the contour of the spectral line.
2. Within the framework of the time-of-flight method of measuring
the distribution functions of the velocities of the atoms and molecules
the true distribution function P (n) is also treated 7 as only a convo-
lution with the apparatus function of the formation system and recording
of the beam H (n) and, consequently, the task of determining this function
is reduced to solving the integral equation of the type (20.2) 7,8
z
¥
H( n - n¢) P( n¢) dn¢ = R( n), (20.3)
-¥
where R (n) is the measured time-of-flight signal.

Thus, it can be seen that when using both methods of measuring
the distribution function it is necessary to solve the integral equations
of the first kind: in the first method – both equation (20.1), (20.2) and
in the second method the equation (20.3). It is important to note that
539
Plasma Diagnostics
in the experiment the functions f (v) (in the first method) and R (n)
(in the second method) are measured unavoidably with an error which
is different from zero, i.e. determination of the function P (n) from
the equations (20.1)–(20.3) is an inverse incorrect problem. 9 The so-
lution of these problems is especially complicated when for the velocity
distribution function it is not possible to show in advance 9 the case in
which the shape of P (n) differs from the Maxwell distribution.
At the same time, these situations often occur in practice in experi-
ments in measuring the velocity distribution function of heavy parti-
cles in different non-equilibrium objects (see, for example, Ref.10, 13).
In recent years, to solve the inverse incorrect problems of the type
(20.1) and (20.2), (20.3) various investigators have used successfully
the regularisation methods. 9,14 The success with application of these
methods is attributed to a large degree to the use of the existing
apriori information on the required solution, for example, the properties
of smoothness of the required solution.9,14 However, in complicated cases,
this information may be insufficient to find the solution with the re-
quired accuracy. Additional information from these problems can be
represented by the information indicating that the required solution is
similar to some unknown function.
In particular, the calculations carried out in Ref.15 show that in a
number of cases it is highly efficient to use the multiplicative repre-
sentation of the required function in the form j (n) = j 0 (n)j 1, where
j 0 (n) is some zero approximation for j (n). As indicated by Ref.15,
this approach gives more accurate results in comparison with the con-
ventional method.
However, the application of the multiplicative representation in the
form of the difference kernel of the initial equation (as was done in,
for example, (20.2)) does not make it possible to solve the problem
by the Fourier analysis method. In fact, this gives a new integral equa-
tion of the first kind with a kernel a 1 (n, n¢) = a (n – n¢)j(n¢) which
is no longer of the difference type.
The authors of Ref.16 proposed not a multiplicative but additive ap-
proximation of the required solution in the form j(n) = y(n) + j 0(n),
where j0 (n) is some function known from apriori assumptions. It can
be shown that this approximation, in contrast to Ref.15, does not change
the difference nature of the kernel of the initial integral equation. This
is important for solving the problem by the Fourier analysis methods.
At the same time, application of the Fourier variant of the regulari-
sation of the logarithm in practice is highly efficient for solving the
integral equations of the first kind with a difference kernel (see, for
example, Ref.9).
540
It can easily be seen that the additive approximation of the func-

tion j(n) is equivalent to the method of a test solution (see, for ex-
ample, Ref.17) when in searching for the regular solution of equation
(20.2) we use the stabilising potential not with respect to the j(n)
function but the difference of the function j(n) – j 0 (n).
In this work, we consider the additional apriori information according
to which the examined distribution is similar to Maxwell’s distribution
(which in a number of cases is confirmed in practice 16 ): the required
solution of P (v) is presented in the form of a sum
P(v) = P 0(v) + Q(v), (20.4)
where P 0 (v) is the given Maxwell distribution, Q (v) is some ‘addi-

tion’ to P0 (v). After substituting (20.4) into (20.1) and (20.2), we obtain
a system of equations
z z z
¥ ¥ ¥
P0 (v)
a ( n - n1 )y (v1 ) dv1 = f ( n) - a( n - n1 ) dvdv, (20.5)
v
-¥ -¥ m1
z
¥
Q(v)
dv = y( n). (20.6)
v
m
To explain the efficiency of the additive approximation of (20.4), we

carried out comparative modelling calculations in two modifications of
the Fourier-variant of the Tikhonov’s method: 9
1) Normal modification where the required distribution P (v) is de-
termined directly from the equations (20.1) and (20.2);
2) Consecutive, using the additive approximating function P0 (v) from
(20.4).
The modelling problems were solved using the following schema:
– we specify the velocity distribution of particles P (v);
– we determine the distribution of the intensities in the corresponding
z
¥
P(v) n - n0
contour of the spectral line j( n) = const × dv; m = c;
v n0
m
- this distribution was ‘rolled up’ together with the known apparatus
541
Plasma Diagnostics
z
¥
function a (n) and the function f0 ( n) = a( n - n¢) j ( n¢)dv¢ was cal-

-¥
culated;
– function f 0 (n) was added to the random function e (n) (model-
ling noise) distributed with dispersion s 2f in accordance with the nor-
mal law f 0 (n) + e (n) = f (n).
The inverse problem was solved directly for this purpose:
– at a known f (n) and a (n), we used (20.2) to calculate the func-
tion j a(n);
– j a(n) was differentiated with respect to gap in accordance with
dj a ( n)
(20.1) thus giving the function Pa ( n) = const × v × 1 .
dn n= n0 ×
1- v / c
The error of restoration of the solution was calculated from the de-
viation of the numerical solution P a (n) from the given P (n)*; (in
reality, functions P (v) and P (v)/v are compared instead of P (v) and
P (v) because the similarity of these functions indicates the quality of
restoration in both regularisation stages)
The modelling distribution P (v) was represented by the distribution
consisting of the sum of Maxwell’s distributions with differing tempera-
tures and amplitude of the maxima
RS 1 n2UV RS F I UV
1 v 1
P( n) = A1v 2 exp -
T 2 T
+ A2 v2 exp -
W T HK W
2 3 T
. (20.7)
This case was examined to model the presence of the possible struc-
ture in the distributions.
To determine the function P (v) (or Q (v)) from the available values
of j(n) (or y(n)), we used the differential variant of equation (20.1)
(or (20.6)): calculation of derivatives j¢(n) (or y¢(n)) was carried out
using smoothing cubic splines with the selection of the smoothing pa-
rameter using the discrepancy. 18,19
It is interesting to compare the efficiency of restoring the veloc-
ity distribution function using the conventional and consecutive restoration
procedures. Figures 20.1 and 20.2 show the modelling and restored ve-
locity distribution functions for different numbers of counting N and
noise levels s 2f. The ratio of the widths of functions j(n) and a(n)
in all cases is D 1 : D 2 = 1.8, the ratio of the amplitudes A 1 : A 2 = 10.
Comparison shows that, with other conditions being equal, the procedure
542
P(v)/v, rel. units
v, rel. units
Fig.20.1 Modelling (1) and restored (2,3) distribution functions. 2) using additive
approximation, 3) conventional restoration; a) ∆ 1 : ∆ 2 = 1.8; A 1 :A 2 = 1 : 10; N = 32;
σ f = 5%. b) ∆ 1 :∆ 2 = 1.8; A 1:A 2 = 1:10; N = 32; σ f = 2%.
P(v)/v, rel. units
v, rel. units
Fig.20.2 Modelling (1) and restored (2,3) distribution of functions. 2) using additive
approximation, 3) conventional restoration; a) ∆ 1 :∆ 2 = 1.8; A1 :A 2 = 1:10; N = 128;
σ f = 2%. b) ∆ 1 : ∆ 2 = 1.8; A 1:A 2 = 1:10; N = 128; σ f = 5%.
543
Plasma Diagnostics
of successive restoration of the ‘structural’ distributions of the type

(20.7) is far more efficient that the conventional procedure, especially
in the relevant region of high velocities. The controlling factor in this
case is that it was possible to use more efficiently the apriori infor-
mation on the initial solution.
The application of the method of successive restoration enabled the
authors 12,16 to restore efficiently the velocity distribution function of
the excited nitrogen molecules 16 and oxygen atoms. 12
In conclusion, it should be stressed again that the application of the
proposed method of additive approximation is especially efficient when
restoring various types of non-equilibrium velocity distribution functions,
for example, functions with singularities at the ‘tails’, functions with
discontinuities in the derivatives, etc. However, as already mentioned,
the restoration of these function under the non-equilibrium conditions
is a very important task.
544
Automation of Measurements in Plasma Diagnostic
Chapter 21
AUTOMATION OF MEASUREMENTS IN
PLASMA DIAGNOSTICS
Recently, a number of publications concerned with the development of
automated systems of experimental investigations of physico-chemical
processes under low-temperature plasma conditions has increased. 1-10
The main tasks of using these systems are automated collection,
processing, build-up and imagining of information and, if necessary,
controls of the experiments.
Original investigations were carried out in this country into the systems
of collecting experimental information and examining the flows of high-
temperature gases generated by electric arc heaters. 1–3 The experi-
ence obtained in constructing and using these systems has made it pos-
sible to formulate a number of important requirements of measuring
apparatus, a selection of noise-resistant temperature, pressure and flow
rate sensors ensuring reliable measurements under the conditions of
the effect of strong electric and magnetic fields on them. It was thus
possible to develop effective electronic-measuring systems using mi-
cro- and mini-computers for automating multi-functional apparatus and
organising the multi-channel collection of information under the con-
ditions with more stringent requirements on measurement accuracy. In
addition, the application of computers in the composition of measur-
ing complexes makes it possible to operate under their real time scale
which opens wide possibilities of development and improvement of the
measuring systems.
For example, the authors of Ref.4 described a measuring system
for recording and processing the spectra of spatially non-uniform objects
in examining stationary and high-rate processes. This system can also
be used in spectroscopic investigations and plasma diagnostics of con-
tinuous and pulsed electric and optical discharges.
The automated system of photoelectric recording on the basis of the
IVK-1 measuring-computing complex has been developed in the con-
ditions of time correlation and spatial-time measurements and was
545
Plasma Diagnostics
described in Ref.5. For time correlation measurements, the radiation

receiver is connected to the output of several monochromators or to
a quantometer. For spatial–time measurements, the radiation from dif-
ferent areas of the plasma object is fed to the input of the receiver
using a set of lightguides.
The authors of Refs.6 and 7 report on the results of development
and application of a measuring system for determining the main thermo-
physical and gas-dynamic characteristics of the plasma flows and electric
arcs.
The authors of Ref.8 described a simple system of collecting the
data using a microcomputer for recording the volt–ampere character-
istics of a Langmuir’s probe in stationary glow discharge plasma. Evi-
dently, it can be used as a basic diagnostic complex by investigating
the stationary plasma. It makes it possible to control automatically the
plasma parameters during measurements.
The authors of Ref.9 describe equipment based on expanded complex
of the M400 control and computing system and communication means
in the KAMAK standard for automating the collection and process-
ing diagnostic information and also the programmed control of the
plasma formation conditions in the experiments with the interaction of
intensive electromagnetic waves and electron beams with low-tempera-
ture plasma. Equipment makes it possible to carry out automated ex-
periments in examining the development of parametric instabilities in
plasma and the dependence of the spatial–time structure of microwave
fields on the plasma parameters.
In Ref.10, the authors developed a system of controlling the gen-
erator of low-temperature plasma and carried out experiments aimed
at controlling and regulating the regime parameters of equipment, re-
cording and processing the experimental data. This system was used
for investigating the pulsation characteristics of the current and volt-
age of the plasma torch and turbulent pulsations in the plasma jet using
stationary and flight-path electrostatic probes.
Diagnostic equipment developed in line a computer is used at present
in laboratory investigations of processes in electric arcs, plasma jets
of complicated chemical composition (both single-phase and those car-
rying particles of inertia impurities) and other plasma objects may be-
come an efficient means of inspection of complicated plasma and plasma
chemical technological processes. This is convincingly confirmed by
Ref.11 in which the authors described an optical apparatus fitted with
microprocessor technology and used for monitoring the plasma jet with
a dispersed phase and the electrons of the plasma torch on the real
time scale.
Until recently a computer was a very expensive tool for automat-
546
Automation of Measurements in Plasma Diagnostic
ing ‘routine’ investigations; 12 the situation has greatly changed because

of the appearance of relatively cheap and compact microprocessor de-
vices having the possibilities of minicomputers of previous generations;
combination of microprocessor technology and KAMAK apparatus has
opened new prospects for the mass automation of laboratory investi-
gations. The production of the first Russian microcomputer Elektronika-
60 created suitable conditions for the development and application of
automated systems of a new type on the basis of microprocessor tech-
nology and equipment of the KAMAK type (Micro-Kamak-lab), 13
designed for complex automation of research laboratories. 12 These sys-
tems offer the investigator in the working area various possibilities,
e.g. collection of experimental data, primary processing of these data,
547
Plasma Diagnostics
548
References
REFERENCES
Preface
1 G. Grim, Spectroscopy of plasma, Atomizdat, Moscow (1969).
2 R. Hoddlestone and S. Leonard (eds), Plasma diagnostics, Mir, Moscow (1967).
3 V. Locht-Holtgreven, Methods of plasma examination, Mir, Moscow (1971).
4 S.E Frish (ed), Spectroscopy of as-discharged plasma, Nauka, Leningrad (1970).
5 L.A. Dushin and O.S. Pavlichenko, Examination of plasma using lasers, Atomizdat,
Moscow (1968).
6 N.G. Preobrazhenskii, Spectroscopy of optical dense plasma, Siberian division of
Nauka Publishing House, Novosibirsk (1971).
7 L.N. Pyatnitskii, Laser plasma diagnostics, Atomizdat, Moscow (1976).
8 Yu.A. Ivanov, Yu.A. Lebedev and L.S. Polak, Contact methods of diagnostics in non-
equilibrium plasma chemistry, Nauka, Moscow (1981).
9 P. Chang, L. Tolbot and K. Turyan, Electrical probes in stationary and moving
plasma, Mir, Moscow (1978).
10 V.K. Zhivotov, V.D. Rusanov and A.A. Fridman, Diagnostics of non-equilibrium
chemically active plasma, Energoatomizdat, Moscow (1985).
Chapter 1
1 L.T. Lar'kina and V.S. Engel'sht, Reduction to a homogenous optically thin layer in
axisymmetric objects, VINITI, Moscow (1973), Manuscript deposited at VINITI
No.6917-73 (1973).
2 A.S. An'shakov, G.Yu. Dautov and A.P. Petrov, Generators of low-temperature
plasma, Energiya, Moscow (1969).
3 N.G. Kolesnikov, L.T. Lar'kina and V.S. Engel'sht, Izv. SO AN SSSR, Ser. Tekhn.
Nauk., No.3, Issue No.1, 31-35 (1974).
4 A. Locht-Holtgreven (ed), Methods of plasma examination, Mir, Moscow (1971).
5 G. Grim, Plasma spectroscopy, Atomizdat, Moscow (1969).
6 Zh. Zheenbaev and V.S. Engel'sht, Laminar plasma torch, Ilim, Frunze (1975).
7 L.A. Spektorov and L.A. Sinel'nikova, Determination of arial distribution of temperature
in axisymmetric plasma sources, Moscow (1972), Manuscript deposited at VINITI
No.4902-72 (1972).
8 V.S. Engel'sht, Distribution of atoms of sodium and copper in the cloud of a dc
plasma, works of the Department of Physics, Kirgiz State University, Frunze (1961).
9 Examination of the electric arc in argon, Ilim, Frunze (1966).
10 N.G. Preobrazhenskii, Nukleonika, 20, No.5, 429-438 (1975).
11 I.F. Garapova, A.I. Dautov, N.G. Sabitova, et al, Fiz. Khim. Obrab. Mater., No.6, 50-
55 (1977).
12 M.Kh. Salakhov, I.D. Grachev and R.Z. Lapitov, ZhPS, 41, No.4, 558-560 (1984).
13 G.P. Luzunkov, V.P. Kabashnikov, A.A. Kurskov and V.D. Shimanovich, Methods of
spectroscopy diagnostics of non-stationary arc plasma on the basis of modelling its
oscillations, Preprint of the Institute of Physics of the Belrussian Academy of
Sciences, Minsk (1986).
Chapter 2
1 V.K. Zhivotov, V.D. Rusanov and A.A. Fridman, Diagnostics of non-equilibrium
2 T.T. Karasheva, D.K. Otorbaev and V.N. Ochkin, et al, Electronically excited molecules
in non-equilibrium plasma, Nauka, Moscow (1985).
3 L.A. Vainshtein, I.I. Sobel'man and E.A. Yukov, Excitation of atoms and broadening
of spectral lines, Nauka, Moscow (1979).
4 V.A .Dudkin, Optika i Spektroskopiya, 24, 367-371 (1968).
5 G.N. Polyakova and A.I. Ranyuk, Prepr. KhFTI, No.81, Kar'kov (1981).
6 A.N. Tikhonov and V.Ya. Arsenin, Methods of solving incorrect problems, Nauka,
Moscow (1979).
549
Plasma Diagnostics
7 V.F. Turchin, V.P. Kozlov and M.S. Malkevich, Usp. Fiz. Nauk, 102, 345-386 (1970).
8 A.V. Kryanev, ZhVMiMF, 14, 25-35 (1974).
9 D.K. Otorbaev, V.N. Ochkin, N.G. Preobrazhenskii, et al, Prepr. FIAN SSSR, No.39,
Moscow (1981).
10 D.K. Otorbaev, V.N. Ochkin, N.G. Preobrazhenskii, et al, ZhETF, 81, 1626-1638
(1981).
11 L.D. Landau, E.M. Lifshits, Quantum mechanics, non-relativistic theory, Nauka,
Moscow (1974).
12 S.Yu. Savinov and S.N. Tskhai, Krat. Sobshch. Fiz., No.9, 3-7 (1983).
13 E.G. Whipple, J. Chem. Phys., 60, 1345-1351 (1974).
14 T.T. Karasheva, M. Malikov, D.K. Otorbaev, et al, Prepr. FIAN SSSR, No.193,
Moscow (1986).
15 T.T. Karasheva, D.K. Otorbaev, V.N. Ochkin, et al, Krat. Soobshch. Fiz., No.10, 53-
58 (1982).
16 M.S. Ivanov, T.T. Karasheva, D.K. Otorbaev, et al, Prepr. FIAN SSSR, Moscow,
No.80 (1987).
17 A.M. Sravath, Phys. Rev., 36, 248-250 (1930).
18 I. MacDaniel', Processes of collision in ionized gases, Mir, Moscow (1962).
19 V.N. Ochkin, S.Yu. Savinov, N.N. Sobolev, ZhETF, 75, 463-472 (1978).
20 D.I. Slobetskii, Mechanisms of chemical reactions in non-equilibrium plasma, Nauka,
Moscow (1980).
Chapter 3
1 L.S. Polak, D.I. Slovetskii, A.S. Sokolov and T.V. Fedoseeva, Non-equilibrium
chemical kinetics and its application, Nauka, Moscow (1979).
2 J.W. Coburn and M. Chen, J. Appl. Phys., 51, No.6 (1980).
3 R. D'Agostino, F. Cramarossa, S. Benedictis and G. Ferraro, Ibid, 52, No.3, 1259-
1265 (1981).
4 L.A. Vainshtein, I.I. Sobel'man and E.A. Yukov, Excitation of atoms and broadening
of spectral lines, Nauka, Moscow (1979).
5 D.I. Slobetskii, Mechanisms of chemical reactions in non-equilibrium plasma, Nauka,
Moscow (1980).
6 A.J. Williams and J.P. Doering, Planet. Space Sci., 17, No.11, 1527-1537 (1969).
7 V.V. Skubenich, M.V. Povch and I.P. Zapesochnii, Optika i Spektroskopiya, 11, No.3,
116-120 (1977).
8 l.S. Polak, D.I. Slovetskii, A.S. Sokolov and T.V. Fedoseeva, Experimental and
theoretical examination of non-equilibrium physico-chemical proceses, Vol.3, Nauka,
Moscow (1974).
9 C.J. Mogab, A.C. Adams and D.L. Flamm, J. Appl. Phys., 49, No.7, 3796-3803
(1978).
10 V.M. Donnelly, D.L. Flamm, W.S. Dautremont-Smith and D.J. Werder, Ibid, 55, No.1,
242-252 (1984).
11 D.E. Ibbotson, D.L. Flamm and V.M. Donnelly, Ibid, 54, No.10, 5974-5981 (1983).
12 R.A. Gottscho, G.P. Davis and R.H. Burton, Vac. Sci. Technology, 1A, No.3, 622-
624 (1983).
13 R.A. Gottscho and V.M. Donnelly, J. Appl. Phys., 56, No.2, 245-250 (1984).
14 R. d'Agostino, V. Colaprico, F. Cramaross, et al, 7th ESCAMPIG, Abstracts of
papers, Bari, Italy (1984), pp.107-108.
15 A. Ricard, 17th ICPIG, Abstracts of invited lectures, Budapest (1985).
16 R. d'Agostino, F. Cramarossa, V. Colaprico and R. d'Ettole, J. Appl. Phys., 54, No.3,
1284-1288 (1983).
17 R.A. Gottscho, G. Smolinsky and R.H. Burton, Ibid, 53, No.8, 5908-5918 (1982).
18 D.L. Flamm, V.M. Donnelly and J.A. Mucha, Ibid, 52, No.5, 3633-3639 (1981).
19 T.T. Karasheva, D.K. Otorbaev, V.N. Ochkin, et al, Tr. FIAN, Vol.157, Nauka,
Moscow (1985).
20 V.I. Gorokhovskii, G.N. Kurtynina and D.K. Otorbaev, Khim. Vysok. Energii, 23, No.5,
450-455 (1989).
21 V.I. Gorokhovskii and D.K. Otorbaev, Fiz. Khim. Obrab. Mater., No.2, 51-54 (1989).
550
References
22 N.Zh. Zheenbaev, M.Z. Mamytbekov and D.K. Otorbaev, ZhPS, 51, No.1, 12-16
(1989).
23 A.T. Baiterekov, K.A. Imankulov and D.K. Otorbaev, Production, examination and
application of plasma in microwave fields, Ilim, Frunze (1987).
Chapter 4
1 L.N. Pyatnitskii, Laser plasma diagnostics, Atomizdat, Moscow (1976).
2 A.N. Zaidel' and G.V. Ostrovskaya, Laser methods of plasma investigation, Nauka,
Leningrad (1977).
3 Yu.I. Ostrovskii, M.M. Butusov and G.V. Ostrovskaya, Holographic interferrometry,
Nauka, Moscow (1977).
4 G.V. Ostrovskaya and Ju.I. Ostrovsky, Progress in optics, 22, 199-270 (1985).
5 A.N. Zaidel', UFN, 149, No.1, 105-138 (1986).
6 G.T. Razdobarin and D.A. Shcheglov, Plasma diagnostics, No.6, Energoatomizdat,
Moscow (1989), pp.88-101.
7 Kh.P. Alum, Yu.V. Koval'chuk an G.V. Ostrovskaya, ZhTF, 54, No.5, 896-903 (1984).
8 K.W. Allen, Astrophysical quantities, Mir, Moscow (1977).
9 G.V. Ostrovskaya, Kh.P. Alum, Yu.V. Koval'chuk, Dispersion interferrometer, Author's
certificate No.864942 SSSR, Published priority since, 29.04.80, Bull. Izobr., No.35
(1981).
10 F.A. Hopf, A. Tomito and G. Al-Jumaily, Opt. Lett. , 5, No.5, 386-388 (1980).
11 Kh.P. Alum, Yu.V. Koval'chuk and G.V. Ostrovskaya, Pis'ma ZhTF, 7, No.22, 1359-
1364 (1981).
12 Yu.I. Ostrovskii, Author's certificate No.268732, Published priority since 16.01.61,
Bull. Izobr., No.14 (1970).
13 A.F. Belozerov, K.S. Mustafin, A.I. Salykova, et al, Optika and Spektroskopiya, 29,
No.2, 384-389 (1970).
14 K.M. Measures, Appl. Opt., 9, No.3, 737-741 (1970).
15 G.V. Dreiden, A.N. Zaidel', G.V. Ostrovskaya, Fiz. Plazmy, 1, no.3, 462-482 (1975).
16 G.V. Dreiden, G.V. Ostrovskaya, et al, Fiz. Plazmy, 1, no.3, 462-482 (1975).
17 G.V. Ostrovskaya and N.A. Pobedonstseva, ZhTF, 45, No.7, 1462-1469 (1975).
18 B.E. Golant, Super-high frequency methods of plasma examination, Nauka, Moscow
(1968).
19 D.E.T.F. Ashbyand D.F. Jephcott, Appl. Phys. Lett., 3, No.1, 13-16 (1963).
20 E.B. Hooper and G. Bekefi, Ibid, 7, No.5, 133-135 (1965).
21 H. Herold and F.C. Jahoda, Rev. Sci. Instr., 40, No.1, 145-147 (1969).
22 J.B. Gerardo and J.T. Verdeyen, Appl. Phys. Lett., 3, No.6, 121-123 (1963).
23 J.B. Gerardo and J.T. Verdeyen, Proc. IEEE, 52, No.6, 690-698 (1964).
24 J.B. Gerardo, J.T. Verdeyen and M.A. Gusinov, J. Appl. Phys., 36, No.7, 2146-2151
(1965).
25 D.E.T.J. Ashby, D.F. Jephcott, A. Malein and F.A. Raynor, Ibid, 36, No.1, 29-34.
26 P.A. Miller, J.T. Verdeyen and B.E. Cerrington, Phys. Rev., A4, No.2, 692-700
(1971).
27 L.A. Schlie and J.T. Verdeyen, IEEE J. Quant. Electr., QE-5, No.1, 21-29 (1969).
28 V.V. Korobkin, Plasma diagnostics, No.1, Atomizdat, Moscow (1963), pp.36-41.
29 J.T. Verdeyen, B.E. Cherrington and M.E. Fein, Appl. Phys. Lett., 9, No.11, 360-361
(1966).
30 L.A. Dushin, O.S. Pavlichenko, V.I. Privezentsev and V.G. Yakovlev, Plasma Diagnostics,
No.2, Atomizdat, Moscow (1968).
31 L.A. Dushin and V.I. Privezentsev, Teplofiz. Vysokikh Temp., 8, No.1, 159-162
(1970).
32 L.A. Dushin and O.S. Pavlichenko, Examination of plasma using lasers, Energiya,
Moscow (1968).
33 G.V. Ostrovskaya and Yu.I. Ostrovskii, Pis'ma ZhTF, 4, No.4, 121-124 (1966).
34 A.N. Zaidel', G.V. Ostrovskaya, Yu.I. Ostrovskii and T.Ya. Chelidze, ZhTF, 36,
No.12, 2208-2210 (1966).
35 P.W. Chen and C.S. Lee, Phys. Lett., 62A, No.1, 33-35 (1977).
36 A.G. Smirnov, Optical holography and its applications, LDNTP, Leningrad (1974).
37 S.M. Belotserkovskii, V.S. Sukhorukikh and V.S. Tatarenchik, ZhPMTF, No.3, 95-99
551
Plasma Diagnostics
(1964).
38 R.D. Matulka and D.J. Collins, J. Appl. Phys., 42, No.3, 1109-1119 (1971).
39 K. Iwata and R. Nagata, JOSA, 60, No.1, 133-135 (1970).
40 D.W. Sweeney and C.M. Vest, Appl. Opt., 11, No.1, 205-207 (1972).
41 D.W. Sweeney and C.M. Vest, Ibid, 12, No.11, 2649-2664.
42 V.M. Ginzburg and B.M. Stepanov (eds), Holography. Methods and equipment. Sov.
Radio, Moscow (1974).
43 C.M. Vest, Holographic Interferometry, Wiley, New York (1979).
44 R.E. Brooks, L.O. Heflinger and R.G. Wuerker, IEEE J. Quant. Electr., QE-2, No.8,
275-279 (1966).
45 C.M. Vest and D.W. Sweeney, Appl. Opt., 9, No.10, 2321-2325 (1970).
46 M.M. Butusov, Proc. of the 3rd Nat. School of Holography, LIYaF, Leningrad (1972),
pp.195-203.
47 F. Grible, W.E. Quinn and R.E. Siemon, Phys. Fluids., 14, No.9, 2042-2047 (1971).
48 K.S. Thomas, C.R. Harder, W.E. Quinn and R.E. Siemon, Ibid, 15, No.10, 1658-1666
(1972).
49 L.V. Dubovoi, A.G. Smirnov, V.G. Smirnov and D.I. Stasel'ko, Optical holography and
its applications, LDNTP, Leningrad (1974).
50 Yu.V. Ascheulov, A.A. Mymnikov, Yu.I. Ostrovsky and A.N. Zaidel, Phys. Lett., 25A,
No.1, 61-62 (1967).
51 V.A. Nikashin, G.I. Rukman, V.K. Sakharov and V.K. Tarasov, Teplof. Vysokikh
Temp., 7, No.6, 1198-1200 (1969).
52 V.M. Ginzburg, B.M. Stepanov and Yu.I. Filenko, RE, 17, No.10, 2219-2220 (1972).
53 J.M. Gates, Nature, 220, 473-474 (1968).
54 A.K. Beketova, L.T. Mustafina and A.Ya. Smolyak, Optika i Spetroskopiya, 39, No.2,
336-339 (1975).
55 G.V. Dreiden, Yu.I. Ostrovsky, E.N. Shedova and A.N. Zaidel, Opt. Comm., 46, No.3,
209-213 (1971).
56 G.V. Dreiden, A.N. Zaidel', Yu.I. Ostrovskii and E.N. Shedova, ZhTF, 43, No.7, 1537-
1542 (1973).
57 D.W. Koopman, H.J. Siebeneck, G. Jellison and W.G. Niessen, Rev. Sci. Instrum.,
49, No.4, 524-525 (1978).
58 A.M. Mirzabekov, N.K. Mitrofanov, Yu.I. Ostrovskii and E.N. Shedova, ZhTF, 51,
No.10, 2038-2042 (1981).
59 A.F. Belozerov, A.N. Berezkin, L.T. Mustafina and A.I. Razumovskaya, Pis'ma ZhTF,
4, No.9/10, 522-525 (1978).
60 F. Friedrich, O.M. Weigl and A.A. Dougal, IEEE J. Quant. Electr., QE-5, No.6, 360-
361 (1969).
61 F. Weigl, O.M. Friedrich and A.A. Dougal, Ibid, QE-6, No.1, 41-49 (1970).
62 I.I. Komissarova, G.V. Ostrovskaya, V.N. Filippov and E.N. Shedova, Author's
certificate No.1028152, Bull. Izobr., No.32 (1985).
63 I.I. Komissarova, G.V. Ostrovskaya, V.N. Filippov and E.N. Shedova, ZhTF, 57,
No.2, 377-380 (1987).
64 O. Bryngdahl and A.W. Lohman, JOSA, 58, No.10, 1325-1334 (1968).
65 K.S. Mustafin, V.A. Seleznev and E.I. Shtyrkov, Optika i Spektroskopiya, 28, No.6,
1186-1189 (1970).
66 G.I. Mishin (ed), Holographic interferometry of phase objects, Nauka, Leningrad
Division, Leningrad (1979).
67 D. Apostol, D. Barbulesky, I.I. Komissarova, et al, ZhTF, 58, No.11, 2156-2161
(1988).
68 A.A. Bugaev, B.P. Zakharchenya and Chudnovskii, Phase transition between the
semiconductor and metal and its applications, Nauka, Moscow (1979).
69 B.P. Zakharchenya, F.A. Chudnovskii and Z.I. Shteingol'ts, Pis'ma ZhTF, 9, No.2,
76-78 (1983).
70 G.F. Frasier, T.D. Wilkerson and J.M. Lindsey, Appl. Opt., 15, No.6, 1350-1352
(1976).
71 G.R. Mitchel, B. Grek, T.W. Johnston, et al, Ibid, 18, No.14, 2422-2426 (1979).
72 D. Naor, A. Flusberg and I. Itzkan, Ibid, 20, No.14, 2574-2584 (1981).
73 H.F. Dobele, Proc. 15th Intern. Conf. Phenomena Ionized Gases, Pt.3, Minsk (1981),
552
References
pp.135-144.
74 G.V. Ostrovskaya, 3rd Intern. Conf. IR Physics, ETH, Zurich (1984), pp.133-141.
75 P.R. Forman, S. Humphries and R.W. Peterson, Appl. Phys. Lett., 22, No.10, 537-
539 (1973).
76 I.I. Komissarova, G.V. Ostrovskaya, V.N. Filippov and E.H. Shedova, ZhTF, 53,
No.2, 251-257 (1983).
77 E.M. Barkhudarov, V.R. Berezovskii, T.Ya. Chelidze, et al, Proc. 13th Intern. Conf.
Phenomena Ionized Gases, Pt.2, Berlin (1977), pp.625-626.
78 E.M. Barkhudarov, V.R. Berezovskii, T.Ya. Chelidze, et al, Proc. 9th Europ. Conf.
Contr. Fusion Plasma Phys., Oxford (1979), p.35.
79 I.I. Komissarova, G.V. Ostrovskaya, Yu.I. Ostrovsky, et al, 7th Intern. Conf. Phenomena
Ionized Gases, Pt.2, Budapest (1985), pp.1099-1101.
80 A.B. Ignatov, I.I. Komissarova, G.V. Ostrovskaya and L.L. Shapiro, ZhTF, 41, No.4,
701-708 (1971).
81 G.V. Ostrovskaya and Yu.I. Ostrovskii, ZhTF, 40, No.11, 2419-2422 (1970).
82 A.B. Ignatov, I.I. Komissarova, G.V. Ostrovskaya and L.L. Shapiro, ZhTF, 41, No.2,
417-422 (1971).
83 I.I. Komissarova and G.V. Ostrovskaya, ZhTF, 48, No.10, 2062-2067 (1978).
84 A. Kakos, G.V. Ostrovskaya, Yu.I. Ostrovsky and A.N. Zaidel, Phys. Lett., 23, No.1,
81-83 (1966).
85 I.I. Komissarova, G.V. Ostrovskaya and L.L. Shapiro, ZhTF, 38, No.8, 1369-1373
(1968).
86 T. Tschudi, C. Yamanaka, T. Sasaki, et al, J. Phys. D., 11, No.2, 177-180 (1978).
87 G.V. Dreiden, I.I. Komissarova, V.S. Markov, et al, ZhTF, 51, No.9, 1850-1857
(1981).
88 S.Yu. Bogdanov, G.V. Dreiden, I.I. Komissarova, et al, Proc. 14th Intern. Congr.
Phenomena Ionized Gases, Pt.2, Minsk (1981), pp.973-974.
89 S.Yu. Boldanov, G.V. Dreiden, I.I. Komissarova, et al, Proc. of the 14th Int. Conf. on
High-speed Photography and Photonics, Mashinostroenie, Moscow (1980).
90 A.P. Burmakov, V.A. Zaikov and G.M. Novik, Theoretical physics. Physics of plasma,
Institute of Physics, Academy of Sciences of the Belorussian SSSR, (1975).
91 A.P. Burmakov, A.A. Labuda and V.M. Lutkovskii, IFZh, 29, No.3, 499-503 (1975).
92 A.E. Dabydov, V.S. Abrukov, S.A. Abrukov, et al, Fiz. Gor. Vzryva, 14, No.3, 78-82
(1978).
93 G.K. Wessel, H.C. Rothenberg and B. Zendle, Am. Astr. Soc., No.5, 10 (1962).
94 T.P. Hughes, Nature, 194, No.4, 268-269 (1962).
95 G.M. Mal'shev, ZhTF, 35, No.12, 2129-2143 (1965).
96 D.E. Evans and J. Katzenstein, J. Rep. Prog. Phys., 32, No.2, 207-212 (1969).
97 A.W. de Silva and G.G. Goldenbaum, Methods of Experimental Physics, Vol.9A,
Academic Press, New York (1970), pp.61-93.
98 G.T. Razdobarin, Proc. 7th Conf. Phys. Ionized Gases. Invited Lect., Zagreb (1974),
pp.809-849.
99 N. Sheffield, Plasma Scattering of Electromagnetic Radiation, Wiley, New York
(1975), p.305.
100 A. Lochte-Holtgreven, Methods of plasma examination, Mir, Moscow (1971).
101 S.Yu. Luk'yanov, Hot plasma and controlled nuclear synthesis, Nauka, Moscow
(1975).
102 E.E. Salpeter, Phys. Rev., 120, No.5, 1528-1535 (1960).
103 J.A. Fejer, Can. J. Phys., 39, No.5, 716-740 (1961).
104 F.W. Perkins and E.E. Salpeter, Phys. Rev., 139A, No.1, 55-62 (1965).
105 A.I. Akhiezer, I.A. Akhiezer, R.V. Polovin, et al, Electromagnetic fluctuations in
plasma, Atomizdat, Moscow (1964).
106 A.G. Sitenko, Electromagnetic fluctuations in plasma, Publishing House of the
Khar'kov University, Khar'kov (1965).
107 G.M. Malyshev, G.V. Ostrovskaya, G.T. Razdobarin and L.V. Sokolova, DAN SSSR,
168, No.3, 554-555 (1966).
108 G.M. Malyshev, G.V. Ostrovskaya, G.T. Razdobarin and L.V. Sokolova, Plasma
diagnostics, Atomizdat, No.2, Atomizdat, Moscow (1968), pp.53-58.
109 U. Ascoli-Bartoli, J. Katzenstein and L. Lovisetto, Nature, 204, No.11, 672-673
553
Plasma Diagnostics
(1964).
110 A.N. Zaidel', G.V. Ostrovskaya, Yu.I. Ostrovskii, Technology and practice of spectroscopy,
111 G.M. Malyshev, G.T. Razdobarin and V.V. Semenov, ZhTF, 42, No.7, 1429-1431
(1972).
112 G.M. Malyshev, G.T. Razdobarin and V.V. Semenov, Plasma diagnostics, No.3,
Atomizdat, Moscow (1973), pp.206-210.
113 S.A. Ramsden, P.K. John, B. Kronast and R. Benesch, Phys. Rev. Lett., 19, No.12,
688-689 (1967).
114 V.V. Aleksandrov, A.I. Gorlanov, N.G. Koval'skii, et al, Plasma diagnostics, Vol.3,
Atomizdat, Moscow (1973), pp.200-210.
115 E.A. Burunov, G.M. Malyshev, G.T. Razdobarin and V.V. Semenov, ZhTF, 44, No.1,
113-118 (1974).
116 E.A. Burunov, G.M. Malyshev, G.T. Razdobarin, et al, ZhTF, 45, No.9, 1878-1883
(1975).
117 D. Johnson, D. Dimock, et al, Rev. Sci. Instrum., 56, No.5, 1015-1017 (1985).
118 D. Meisel, H. Murmann, H. Rohr, et al, Rep. IPP/110, 140-143 (1986).
119 H. Murmann and M. Huang, Plasma Phys. Control. Fus., 27, No.2, 103-114 (1985).
120 V.K. Gusev, G.M. Malyshev, G.T. Razdobarin and L.V. Sokolova, ZhTF, 42, No.2,
340-343 (1972).
121 H.C. Koons and G. Fiocco, Phys. Lett., 26A, No.12, 614-615 (1968).
122 K. Huyase and T. Okuda, J. Phys. Soc. Japan, 30, No.6, 1768-1771 (1971).
123 G. Fiocco and E. Thompson, Phys. Rev. Lett., 10, No.3, 89-91 (1963).
124 S.A. Ramsden and W.E. Davies, Ibid, 13, No.7, 227-229 (1964).
125 S.L. Mandel'shtam, P.P. Pashinin, A.M. Prokhorov, et al, ZhETF, 49, No.1(7), 127-
133 (1965).
126 V.V. Korobkin, S.L. Mandel'shtam, P.P. Pashinin, et al, ZhETF, 53, No.1(7), 116-123
(1967).
127 A. Unzold, Physics of star atmospheres, Monthly for atomic-fluorescent analysis.
Physical fundamentals of the method, Moscow (1949).
128 H.J. Kunze, E. Funfer, B. Kronast and W.H. Kegel, Phys. Lett., 11, No.1, 42-43
(1964).
129 S.A. Ramsden and W.E. Davies, Phys. Rev. Lett., 16, No.8, 303-306 (1966).
130 B. Kronast, H. Rohr and E. Glock, Ibid, No.24, 1082-1084.
131 L. Keller, Z. Phys., 239, No.2, 147-161 (1970).
132 C. Chen, Phys. Fluids, 14, No.12, 2787-2788 (1971).
133 C. Chan and J.W. Daley, J. Quant. Spectr. Rad. Trans., 21, No.6, 527-531 (1979).
134 A.N. Zaidel', Atomic-fluorescent analysis, Physical fundamentals of the method,
135 G.T. Razdobarin and I.G. Folomkin, ZhTF, 49, No.7, 1353-1372 (1979).
136 D.R. Olivares and M. Hieftie, Spectrochim. Acta., 33B, No.3/4, 79-99 (1978).
137 V.I. Balykin, G.I. Bekov, V.S. Letokhov and V.I. Mishin, UFN, 132, No.2, 293-344
(1980).
138 M.A. Bol'shov, A.V. Zybin and V.G. Koloshnikov, Kvant. Elektronika, 7, No.8, 1808-
1812 (1980).
139 E. Hinnov, J.G. Hinschberg, F.W. Hoffman and N. Rynn, Phys. Fluids, 6, No.12,
1779-1780 (1963).
140 F.W. Hoffman, Ibid, 7, No.4, 532-536 (1964).
141 M.B. Denton and H.V. Malmstadt, Appl. Phys. Lett., 18, No.11, 485-487 (1971).
142 R.M. Measures and A.B. Rodrigo, Ibid, 20, No.3, 102-104 (1972).
143 J. Lipson, Phys. Fluids, 22, No.11, 2236-2240 (1972).
144 N. Omenetto, S. Nikdel, R.D. Reeves, et al, Spectrochimica Acta, 35B, No.8, 507-
517 (1980).
145 A.E. Elbern, Appl. Phys., 15, No.1, 111-112 (1978).
146 A.E. Elbern, E. Hurtz, B. Schweer, J. Nuclear Mat., 76/77, 143-148 (1978).
147 C.H. Muller and K.H. Burrell, Bull. Am. Phys. Soc., 24, No.8, 971-976 (1979).
148 C.H. Muller and K.H. Burrell, Ibid, 25, No.8, 977-981 (1980).
149 B. Schweer, D. Rusbuldt, Hints E, et al, J. Nucl. Mater., 93/94, No.3, 357-363 (1980).
150 C.H. Muller and K.H. Burrel, Phys. Rev. Lett., 47, No.5, 330-333 (1981).
554
References
151 S. Suckewer, Phys. Scripta, 23, No.1, 72-77 (1981).

152 K. Bergstedt, G. Himmel, F. Pinnekampf, Phys. Lett., 53A, No.3, 261-263 (1975).
153 D.D. Burgess, V.P. Myerscough, C.H. Skinner and J.M. Ward, J. Phys., B13, No.8,
1675-1701 (1975).
154 R.A. Burrell and H.J. Kunze, Phys. Rev. Lett., 28, No.25, 1548-1551 (1975).
155 R.A. Stern and J.A. Johnson, Ibid, 34, No.25, 1548-1551 (1975).
156 V.S. Burakov, P.Ya. Misakov, P. Naumenkov, et al, Plasma diagnostics, Vol.6,
Energoatomizdat, Moscow (1989), pp.105-108.
157 D.A. Evans, M.J. Forrest, M.G. Nickolson, et al, Rev. Sci. Instrum., 56, No.5 (Pt2),
1012-1014 (1985).
158 P. Bogen and F. Mertens, Plasma diagnostics, No.5, Energoizdat, Moscow (1986),
pp.200-205.
159 S.A. Batishe, V.S. Burakov, V.I. Gladushchak, et al, Pis'ma ZhTF, 9, No.23, 1461-
1465 (1983).
160 T.W. Hansch, S.A. Lee, R. Wallenstein and C. Wieman, Phys. Rev. Lett., 34, No.6,
307-309 (1975).
161 B.I. Troshin, V.P. Chebotaev and A.A. Chernenko, Pis'ma ZhETF, 27, No.5, 293-296
(1978).
162 R.A. Stern, D.L. Correll, H. Bohmer and N. Rynn, Phys. Rev. Lett., 37, No.13, 833-
836 (1976).
163 G. Himmel, F. Pinnekampf, J. Phys. B., 10, No.8, 1457-1464 (1977).
164 G.T. Razdobarin, V.V. Semenov, L.V. Sokolova, et al, Nucl. Fus., 19, No.11, 1439-
1446 (1979).
165 M.M. Larionov, G.T. Razdobarin and I.P. Folomkin, Pis'ma ZhTF, 6, No.22, 1375-
1379 (1980).
166 H.C. Meng and H.J. Kunze, Phys. Fluids, 22, No.6, 1082-1088 (1979).
Chapter 5
1 J. Steinfeld, Laser and coherent spectroscopy, Mir, Minsk (1982).
2 J. Helmcke, S.A. Lee, J.L. Hall, Appl. Opt., 21, 1686-1694 (1982).
3 W.J. Witteman, Appl. Phys. Lett., 10, 347-350 (1967).
4 Yu.N. Dubnishchev and B.S. Rinkevichyus, Methods of laser Doppler anemometry,
5 L.A. Dushin and O.S. Pavlichenko, Examination of plasma using lasers, Atomizdat,
Moscow (1968).
6 L.N. Pyatnitskii, Laser diagnostics of plasma, Atomizdat, Moscow (1976).
7 A.A. Besshaposhnikov, V.N. Blokhin, V.B. Voronin and V.A. Myslin, ZhPS, 46, No.5,
723-727 (1987).
8 A.I. Ferguson, Phil. Trans. R. Soc. Lond., A307, 645-660 (1982).
9 V.N. Ochkin, N.G. Preobrazhenskii, N.N. Sobolev and N.Ya. Shaparev, Usp. Fiz.
Nauk, 148, No.473-507 (1986).
10 D.R. Crosley and G.P. Smith, Opt. Eng., 22, No.5, 545-553 (1983).
11 V.V. Kosichkin and A.I. Nadezhdinskii, Izv. AN SSSR. Ser. Fiz., 47, 20-37 (1983).
12 A.V. Dem'yachenko, I.I. Zasavitskii, S.Yu. Savinov, et al, Kvant. Elektronika, 47,
No.4, 851-859 (1987).
13 C. Dang, J. Reid and B.K. Garside, Appl. Phys., B27, 145-151 (1982).
14 I.I. Zasavitskii, R.Sh. Islamov, M.A. Kerimkulov, et al, Examination of the active
media of the wave CO 2 laser, Preprint FIAN, Moscow, No.18, (1988).
15 D.K. Otorbaev, V.N. Ochkin, P.L. Rubin, et al, Electronically excited molecules in
non-equilibrium plasma, Nauka, Moscow (1985).
16 B.F. Gordiets, A.I. Osipov and L.A. Shelepin, Kinetic processes in gases and
molecular lasers, Nauka, Moscow (1980).
17 I.I. Zasavitskii and A.I. Nadezhdinskii, Spectral gas analysis of polyatomic molecules
using tunable diode lasers, Preprint IOFAN, Moscow, No.50 (1987).
18 J.A. Mucha, Appl. Spectr., 36, No.4, 393-400 (1982).
18a A.I. Nadezhdinskii, Proc. of the 28th Nat. Conf. on Spectroscopy, Naukova dumka,
Kiev (1988), p.138.
19 M. Polack, M. Gruebele and R. Saykally, J. Chem. Phys., 87, No.6, 3352-3356
(1987).
555
Plasma Diagnostics
20 F. Pan and T. Oka, Phys. Rev., A36, No.5, 2297-2310 (1987).

21 I.I. Zasavitskii, M.A. Kerimkulov, A.I. Nadezhdinskii, et al, Optika i Spektroskopiya,
65, No.6, 1198-1201 (1988).
22 S.A. Akhmanov and N.I. Koroteev, Methods of non-linear optics in spectroscopy of
light scattering, Nauka, Moscow (1981).
23 Kh. Shrettner and Kh. Kikner, Spectroscopy of Rama and scattering of light in gases
and liquids, Mir, Moscow (1982).
24 M.A. Yuratich and D.C. Hanna, Mol. Phys., 33, No.3, 671-682 (1977).
25 W.F. Murphy, W. Holzer and H.J. Bernstein, Appl. Spectr., 23, No.3, 211-218 (1969).
26 A.F. Bunkin, S.G. Ivanov and N.I. Koroteev, Pis'ma ZhETF, 24, No.8, 468-472
(1976).
27 B.U. Asanov, V.N. Ochkin, S.Yu. Savinov, et al, Krat. Soob. po Fiz., No.9, 26-27
(1988).
28 V.N. Ochkin, S.Yu. Savinov, N.N. Sobolev and S.N. Tskhai, ZhTF, 58, No.7, 1283-
1290 (1988).
29 V.V. Krunetsky, L.A. Kulevsky and V.A. Mishin, Opt. Commun., 21, 225-228 (1977).
30 J.A. Shirely, R.J. Hall and A.S. Eckbreth, Opt. Letts., 5, No.9, 380-382 (1980).
31 W.B. Roh, P.W. Schreiber, J.P.E. Taran, Appl. Phys. Letts., 29, No.3, 174-176
(1976).
32 B.I. Greene, R.B. Weisman and R.M. Hochstrasser, Chem. Phys. Lett., 59, no.1, 5-
9 (1978).
33 S.Yu. Volkov, D.N. Kozlov, A.M. Prokhorov, et al, High-resolution CARS-Spectroscopy
of molecular gases, Nauka, Moscow (1986).
34 B. Massabieaux, G. Gousset, M. Lefebre and M. Pelat, J. Phys., 48, No.11, 1939-
1949 (1987).
35 S.I. Valyanskii, K.A. Vereshchagin, A.Yu. Volkov, et al, Examination of the kinetics
of the function of distribution of nitrogen molecules in vibrational and rotational states
in excitation with plasma discharge in a biharmonic field. Measurements of the
constant of the exchange rate, Prepr. IOFAN, No.109, Moscow (1984).
36 A.A. Devyatov, S.A. Dolenko, A.T. Rakhimov, et al, ZhETF, 90, No.2, 429-436
(1986).
Chapter 6
1 L.D. Landau and E.M. Lifshits, Electrodynamics of solids, GIFML, Moscow (1959).
2 I.L. Fabelinskii, Molecular scattering of light, Nauka, Moscow (1965).
3 R.D. Mountain, Rev. Mod. Phys., 38, No.2, 205-214 (1966).
4 S. yip, Journ. Acoust. Soc. Amer., 49, No.3 (Pt3), 941-949 (1971).
5 G.J. Gabriel, Phys. Rev., A8, No.3, 963-990 (1973).
6 R.W. Pitz, R. Cattolica, F. Robben and L. Talbot, Combust. Flame, 27, No.3, 313-320
(1976).
7 L.N. Rozanov, Vacuum technology, Vyssh. Shkola, Moscow (1982).
8 W. Childs, Physical constants, GIFML, Moscow (1961).
9 I. Namer, R.W. Schefer and M. Chan, Proc. Spring 1980 Meeting, Western States
Section of Combustion Inst., Univer. California, Irvine (1980) (1980). Pap. WSS/80-
18.22.
10 I. Namer, R.W. Schefer and M. Chan, Lawrence Berkeley Laboratory Rep. LBL-1065
(1980).
11 R.G. Bill Jr, I. Namer and L. Talbot, Combust. Flame, 43, No.2, 229-242 (1981).
12 R.G. Bill Jr, I. Namer and L. Talbot, Ibid, 44, No.2, 277-285 (1982).
13 M. Namazian, L. Talbot, F. Robben and R.K. Cheng, Proc. 19th Sympos. (Intern)
Combustion, Combustion Inst. (1987), pp.487-493.
14 M.B. Long, P.S. Levin and D.C. Fourguette, Optics Letters, 10, No.6, 267-269
(1985).
15 D.C. Fourguette, R.M. Zurn, M.B. Long, Combust. Sci. Technol., 44, No.3, 307-317
(1986).
16 S.C. Graham, A.J. Grant and J.M. Jones, AIAA Journ., 12, No.8, 1140-1142 (1974).
17 T.M. Dyer, Ibid, 17, No.8, 912-914 (1979).
18 A.J.D. Farmer and G.N. Haddad, J. Phys. D: Appl. Phys., 21, No.3, 426-431 (1988).
556
References
19 P. Kovitya and L.E. Cram, Amer. Welding Journ., 65, No.1, 34-39 (1986).
20 L.E. Cram, L. Poladian and G. Roumeliotis, J. Phys. D: Appl. Phys., 21, No.3, 418-
425.
21 H.G. Green, J. Phys. E: Sci. Instrum., 20, No.7, 670-676 (1987).
Chapter 7
1 Zh. Zheenbaev and V.S. Engel'sht, Laminar plasma torch, Ilim, Frunze (1975).
2 G. Grim, Plasma spectroscopy, Atomizdat, Moscow (1986).
3 R. Huddlestone and S. Leonard (eds), Plasma diagnostics, Mir, Moscow (1967).
4 L.A. Vasil'ev, Schlieren methods, Nauka, Moscow (1968).
5 L.A. Vasil'ev and I.V. Ershov, Interferrometer with a diffraction grating, Mashinostroenie,
Moscow (1976).
6 M.M. Skotnikov, Schlieren quantitative methods in gas dynamics, Nauka, Moscow
(1976).
7 M.F. Zhukov (ed), Properties of light temperature plasma and methods of
diagnostics,Siberian division of Nauka, Novosibirsk (1977).
8 S.A. Abrukov, Schlieren and interference methods of examination of optical
heterogenieties, Publishing House of the Kazan' University, Kazan (1962).
9 A.K. Beketova, O.P. Klochkova and N.V. Ryabova, Optiko-Mekhan. Prom., No.3, 11-
14 (1962).
10 Zh. Zheenbaev and V.S. Engel'sht, Two-jet plasma torch, Ilim, Frunze 1983.
Chapter 8
1 Proc. of the 4th Nat. Symp. on Plasma Chemistry, Vol2. DKhTI, Dnepropetrovsk
(1984), p.254.
2 B.Y.H. Lin (ed), Fine particles (Aerosols Generation, Measurements, Sampling and
Analysis), Academic Press, New York (1976), p.889.
3 Y. Mizutani, H. Kodama and K. Miyasaka, Combustion and Flame, 44, No.1, 85-95
(1982).
4 F. Durst, Trans. ASME, 104, No.3, 284-296 (1982).
5 Kh. Chen and E. Pfender, Plasma Chem. Plasma Process., 2, No.2, 185-212 (1982).
6 Kh. Chen and E. Pfender, Ibid, No.3, 293-316.
7 E. Pfender, Pure Appl. Chem., 57, No.9, 1179-1195 (1985).
8 M.F. Khukov, V.P. Lyagushkin and O.P. Solonenko, Automated experimental stand
for complex examination of high-temperature heterogeneous jets, No.145-86, Preprints
of the Institute of Thermophysics, Siberian Division of the USSR Academy of
Sciences, Novosibirsk (1986).
9 F. Durst, A. Melling and J.H. Whitelaw, Principles and Practice of Laser Doppler
Anemometry, Whiley, London and New York (1981).
10 G.A. Gousbet, Plasma Chem. Plasma Process., 5, No.2, 91-117 (1985).
11 W.D. Buchalo and M.J. Houser, Opt. Engn., 23, No.5, 695-697 (1984).
12 Particle Dynamics Analyzer, Dantec Publication N 6605E, Elsevier, Denmark (1986),
p.12.
13 J. Mishin, M. Vardelle, J. Lesinski and P. Fauchais, Proc. 7th Intern. Sympos.
Plasma Chem., Eindhoven, Netehrlands, Vol. 3 (1985), pp.724-729.
14 G. van de Khyulst, Scattering of light by small particles, Izd-vo Inostr. Lit., Moscow
(1961).
15 Yu.N. Dubnishchev and B.S. Rinkevichyus, Methods of laser Doppler anemometry,
16 A. Jyle, F. Born, D. Swirl, et al, J. Energy, 1, No.2, 220-225 (1977).
17 V.P. Yankov, Tr. TsAGI, 1755, 83-93 (1976).
18 Y. Mizutani, H. Kadama and K. Nivasoka, Combust. Flame, 44, No.1, 85-95 (1982).
19 W. Mayr, 16th Intern. Conf. Phenomena Ionized Gases, Dusseldorf (1982), pp.412-
413.
20 D.J. Holve and S.A. Self, Appl. Opt., 18, No.10, 1632-1652 (1980).
21 D.J. Holve and K.D. Annen, Opt. Engng., 23, No.5, 591-603 (1984).
22 N. Chiger and G. Stewart, Ibid, pp.554-556.
23 C.F. Hess, Appl. Opt., 23, No.23, 4375-4383 (1984).
557
Plasma Diagnostics
24 C.F. Hess and V.E. Espinoza, Opt. Engng., 23, No.5, 604-609 (1984).
25 K.S. Shifrin and V.I. Golikov, Tr./GGO, 152, 18-21 (1964).
26 R.N. Sokolov, G.D. Petrov and F.A. Kudryavitskii, Izv. AN SSSR, Fiz. Atmosfery i
Okeana, 6, No.6, 214-215 (1970).
27 K.S. Shifrin, Theoretical and applied problems of light scattering, Nauka and Tekhnika,
Minsk (1971).
28 D.E. Hirleman, Opt. Engng., 23, No.5, 610-619 (1984).
29 E.P. Zimin and V.V. Forfutdinov, Gas dynamics of multiphase flows in power
systems, Publishing House of the Khar'kov University, Khar'kov (1978).
30 E.P. Zimin, L.B. Levin and V.I. Chursin, Physical methods of examination of transparent
heterogenieties, Mashinostroenie, Moscow (1987).
31 R.S. Valeev, R.N. Gizatullin, A.G. Golubev and V.I. Yagodkin, Izmeritel., Tekhnika,
No.5, 71-74 (1983).
32 D.A. Tichenor, R.E. Mitchell, K.R. Hencken and S. Niksa, SANDIA Rep. SAND 84-
8628, Albuquerque, New Mexico and Livermore, California (1984), p.29.
33 B.B. Weiner, Modern Methods of Particle Size Analysis, Vol.73, Pergamon Press,
New York (1984), pp.93-116.
34 P.K. Pusey and W. von Megen, J. Phys. Chem., 80, No.8, 3513-3520 (1984).
35 S.C. Graham, J.B. Homer and J.L.J. Rosenfield, Proc. Roy. Soc. (London), A344,
No.2, 259-265 (1975).
36 J.J. Frurip and S.H. Bauer, J. Phys. Chem., 81, No.10, 1007-1015 (1977).
37 P. Chylek, V. Ramaswamy, A. Ashkin and J.M. Dziedzic, Appl. Opt., 22, No.15, 2302-
2307 (1983).
38 T.T. Charalampopoulos, Rev. Sci. Instrum., 58, No.9, 1638-1646 (1987).
39 D.M. Benenson and H.C. Kuck, Proc. 16th Intern. Conf. Phenomena Ionized Gases,
Dusseldorf (1983), pp.473-475.
40 G. Gousbet, Proc. 14th Inter. Conf. Phenomena Ionized Gases, Grenoble (1979),
pp.789-790.
41 M.R. Barrault, Pure Appl. Chem., 52, No.10, 1829-1835 (1980).
42 V.F. Klimkin, A.N. Papyrin and R.I. Soloukhin, Optical methods of recording high-
speed processes, Siberian Division of Nauka, Novosibirsk (1980).
43 P.W. Forder, J. Phys. E: Sci. Instrum., 14, No.7, 1014-1018 (1980).
44 P.W. Forder and D.A. Jackson, Ibid, 15, No.4, 555-557 (1982).
45 R.N. James, W.R. Babcock and H.S. Seifert, AIAA Journ., 6, No.1, 160-162 (1968).
46 H.L. Morse, B.I. Tullis, H.S. Seifert and W.R. Babcock, J. Spacecraft Rockets, 6,
No.2, 264-272 (1969).
47 S.A. Self and K.Kh. Kryuger, Raket. Tekhn. Kosmonavtika, 16, No.5, 160-186 (1978).
48 L. Mannik, S.K. Brown and F.Y. Chu, Proc. 7th Intern. Sympos. Plasma Chem., Vol.2,
Eindhoven, Netherlands (1985), pp.704-709.
49 A.N. Korotkov, S.I. Kruglyi and A.P. Nefedov, Avtometriya, No.3, 27-33 (1982).
50 A.N. Korotkov and A.P. Nefedov, Teplof. Vysokikh Temp., 23, No.4, 792-797 (1985).
51 V.G. Goloborod'ko and Yu.M. Kiselev, Teplof. Yysokikh Temp., 9, No.6, 1248-1252
(1971).
52 V.V. Pikalov and N.G. Preobrazhenskii, Properties of low-temperature plasma and
methods of its diagnostics, Siberian Division of Nauka, Novosibirsk (1977).
53 R.A. Stern and J.A. Johnson, Phys. Rev. Lett., 34, No.25, 1545-1551 (1975).
54 A.A. Knyasev, N.B. Lerner and K.I. Svinolupov, Proc. of the 2nd Nat. Conf. ,
Leningrad (1984), pp.27-29 and Leningrad (1984), p.43.
55 G. Couesbet, C.R. Academ. Sci. Paris, 280, No.19, 597-600 (1975).
56 M.R. Barrault, G.B. Jones and T.R. Blackburn, J. Phys. E: Sci. Instrum., 7, No.5,
663-666 (1974).
57 L. Mannik and S.K. Brown, Appl. Opt., 25, No.5, 649-652 (1986).
58 V.V. Krasovskii, T.P. Kuyanova and E.I. Palagashvili, High-temperature synthesis
and properties of refractory metals, Zinatne, Riga (1979).
59 B.A. Pavlovskii and B. Ruk, Prib. Sistem. Upravlen., No.3, 27-29 (1986).
60 M.C. Bleiweiss, P.N.P. Chang and S.S. Penner, J. Quant. Spectr. Radiat. Transfer.,
26, No.3, 273-275 (1981).
61 J.R. Hunt, J. Fluid Mech., 122, No.5, 933-942 (1982).
558
References
62 D.Ya. Svet, Optical methods of measuring true temperatures, Nauka, Moscow

(1982).
63 G.A. Lyzenga and J.T. Ahrens, Rev. Sci. Instrum., 50, No.1, 1421-1427 (1979).
64 F.H.A. Jorgensen and M. Suiderwyk, J. Phys. E: Sci. Instrum., 18, No.4, 486-491
(1985).
65 M.L. Eider and J.D. Winefordner, Progr. Analyt. Atom. Spectr., 6, No.2, 293-314
(1983).
66 J. Mishin, M. Vardelle, J. Lesinski and P. Fauchais, Proc. 7th Inter. Sympos. Plasma
Chem., Vol.3, Eindhoven and Elsevier, Netherlands (1985), pp.724-729.
67 V.P. Lyagushkin, O.P. Solonenko, Yu.L. Stankevich and V.A. Starikov, Tez. Dokl.,
Proc. of the 11th Nat. Conf. on Low-temperature plasma generators, Ilim, Frunze
(1983), pp.272-273.
68 V.P. Lyagushkin and O.P. Solonenko, Proc. 7th Intern. Sympos. Plasma Chem.,
Eindhoven and Elsevier, Netherlands (1985), pp.730-736.
69 T. Sakuta and M.I. Boulos, Proc. 8th Inter. Sympos. Plasma Chem., Vol.1, Whiley
Interscience, Tokyo (1987), pp.371-376.
70 M.A. Sheindlin, A.V. Kirillin, L.M. Kheivets and K.A. Khodakov, Teplofiz. Vysokikh
Temp., 19, No.4, 839-848 (1981).
Chapter 9
1 N. Kimura and A. Kanzava, Raket. Tekhnika Kosmon., No.2, 25-28 (1963).
2 V.M. Gol'dfarb and A.M. Uzdenov, Izv. CO AN SSSR, Ser. Tekhn. Hauk., 3, No.1, 81-
84 (1967).
3 D. Bannenberg and J. Inzinger, Pribor. Nauch. Issled., No.7, 10-13 (1964).
4 W. Rother, Wiss. Z. TH. Ilmenau, 19, No.3/4, 219-226 (1973).
5 W. Rother, Beitr. Plasma Phys., 10, No.6, 497-506 (1970).
6 B. Bowman, J. Phys. D: Appl. Phys., 5, No.8, 1422-1432 (1972).
7 E. Seymour, Pap. ASME, 1-10 (1971).
8 A. Abdrazakov, Zh. Zheenbaev, R.I. Konavko, et al, Proc. of the 5th Nat. Conf. on
Low-temperature Plasma Generators, Vol. 1, Siberian Division of Nauka, Novosibirsk
(1972), pp.141-144.
9 M.K. Asanaliev, Zh. Zheenbaev and K.K. Makesheva, IFZh, 57, No.4, 554-562
(1989).
10 J. Levis and W. Gauvin, AIChE Journ., 19, No.5, 982-990 (1973).
11 D. Kassoy, T. Adamson and A. Messiter, Phys. Fluids, 9, No.4, 671-681 (1966).
12 M.K. Asanaliev, V.S. Engel'sht, E.P. Pakhomov, et al, Proc. 5th ICPIG, Part 2, Minsk
(1981), pp.959-960.
13 V.V. Kabanov and V.S. Klubnikin, Proc. of the 9th Nat. Conf. on Low-temperature
Plasma Generators, Ilim, Frunze (1982).
14 E. Pfender and Y. Lee, Plasma Chem. Plasma Proc., 5, No.3, 211-236 (1985).
15 Zh. Zheenbaev and V.S. Engelsht, Laminar plasma torch, Ilim, Frunze (1975).
16 G.N. Abramovich, Applied gas dynamics, Nauka, Moscow (1969).
Chapter 10
1 Yu.M. Kagan and V.I. Perel', UFN, 81, No.3, 409-452 (1963).
2 O.V. Kozlov, Electric probe in plasma, Atomizdat, Moscow (1969).
3 F. Chen, Electric probes, (edited by R. Huddlestone and S.M. Leonard), Mir, Moscow
(1967).
4 L. Schott, Electric probes (edited by V. Lochte-Holdgreven), Mir, Moscow (1971).
5 J.D. Swift and M.J.R. Schwar, Electrical probes for plasma diagnostics, Iliffe books,
London (1970).
6 P. Chan, L. Telbot and K. Turyan, Electric probes in stationary and moving plasma,
Mir, Moscow (1978).
7 Yu.A. Ivanov, Yu.A. Lebedev and L.S. Polak, Methods of contact diagnostics in non-
8 V.K. Zhivotov, V.D. Rusanov and A.A. Fridman, Diagnostics on non-equilibrium
9 H. Mott-Smith and I. Langmuir, Phys. Rev., 28, No.5, 727-763 (1926).
559
Plasma Diagnostics
10 M.J. Druyvesteyn, Z. Phys., 64, 781-798 (1930).

11 J.D. Swift, Proc. Phys. Soc., 79, 697-701 (1962).
12 A.I. Lukovnikov and M.Z. Novgorodov, Krat. Soobshch. Fizike, No.1, 27-34 (1971).
13 M.A. Mal'kov, Plasma diagnostics taking into account the sink of electrons on the
probe and in the case of the anisotropy of the velocity of the distribution of electrons,
Author's dissertation, M.V. Lomonosov Moscow State University (1985).
14 S. Matsumura and S-L. Chen, J. Appl. Phys., 43, No.8, 3357-3361 (1972).
15 E. Berger and A. Heisen, J. Phys. D: Appl. Phys., 8, No.6, 629-639 (1975).
16 J-S. Chang, Ibid, 6, No.12, 1674-1683 (1973).
17 A. Boschi and F. Magistrelly, Nuovo Cimento, 29, No.2, 487-493 (1963).
18 S. Klagge and M. Maass, Beitr. Plasmaphys., 23, No.4, 355-368 (1983).
19 N. Hershkowitz, M.H. Cho, C.H. Nam and T. Intrator, Plasma Chem. Plasma Proc.
8, No.1, 35-52 (1988).
20 H. Sabadil, S. Klagge and M. Kammeyer, Ibid, No.4, 425-444.
21 D. Maundrill, J. Slatter, A.T. Spices and C.C. Welch, J. Phys. D: Appl. Phys., 20,
No.7, 815-819 (1987).
22 T.I. Cox, V.G.I. Deshmukh, D.A.O. Hope, et al, Ibid, 820-831.
23 A.P. Ershov, Examination of the kinetics of electrons in the plasma of high-frequency
low-pressure discharge in inert gases, Author's abstract of a disseration, M.V.
Lomonosov Moscow State University (1982).
24 K.D. Asvadurov and I.A. Vasil'eva, ZhTF, 45, No.7, 1558-1559 (1975).
25 Yu.M. Kagan, N.B. Kolokolov, P.M. Pramatarov and M.A. Petrun'kii, ZhTF, 47, No.6,
1160-1166 (1977).
26 H. Amemiya and K. Shimizu, Jpn. J. Appl. Phys., 13, No.6, 1035-1036 (1974).
27 H. Amemiya, Ibid, 14, No.1, 165-166 (1975).
28 A.B. Blagoev, Yu.M. Kagan, N.B. Kolokolov and R.I. Lyagushchenko, ZhTF, 45, No.3,
579-585 (1975).
29 V.I. Demidov, N.B. Kolokolov and O.G. Toronov, Proc. of 6th Nat. Conf. on Low-
temperature Plasma Physics, Leningrad (1983), pp.309-310.
30 S.I. Kryvosheev, V.N. Makarchuk and O.I. Fisun, Teplofiz. Vysokikh Temp., 25,
No.4, 791-793 (1987).
31 L.M. Volkova, V.I. Demidov, N.B. Kolokolov and E.A. Kral;kina, Teplofiz. Vysokikh
Temp., 22, No.4, 757-763 (1984).
Moscow (1986).
33 E.K. Eroshchenkov, Physics, technology and application of low-temperature plasma,
Alma-Ata (1970).
34 L.M. Volkova, A.M. Debyatov, A.S. Mechenov, et al, Vestn. MGU, Ser.3, Fizika,
astronomiya, 16, No.3, 371-374 (1975).
35 Yu.B. Golubovskii, V.M. Zakharova, V.N. Pasunkin and L.D. Tsendin, Fiz. Plazmy, 7,
No.3, 620-628 (1981).
36 N.A. Gorbunov, N.B. Kolokolov and A.A. Kudryavtsev, Fiz. Plazmy, 15, No.12, 1513-
1520 (1989).
37 A.B. Blagoev, N.B. Kolokolov, V.M. Milenin, ZhTF, 42, No.8, 1701-1704 (1971).
38 K. One and T. Kimuar, Jpn. J. Appl. Phys., 28, No.10, 1997-2003 (1989).
39 T.L. Thomas and E.L. Battle, J. Appl. Phys., 41, No.8, 3428-3432 (1970).
40 D.G. Bulls, R.B. Holt and B.T. McClure, Ibid, 33, No.1, 29-33 (1962).
41 Yu.I. Chutov, A.I. Kravchenko and Yu.V. Vovchenko, Proc. of the 6th Nat. Conf. on
Low-temperature Plasma Physics, Leningrad University, Leningrad (1983), pp.320-
322.
42 J. Andrew, G. Sardin, A. Lloret, et al, J. Appl. Phys., 63, No.4, 1230-1232 (1988).
43 W.D. Bunting and W.J. Heikkila, Ibid, 41, No.5, 2263-2264 (1970).
44 J. Laframboise, Univ. Toronto, Inst. Aerospace Studies, Rept. No.100 (1965).
45 J. Laframboise, Rarefield Gas Dynamics, Vol.2, J.H. De Leeuv (ed), Academic
Press, New York (1966), pp.22-44.
46 A.A. Sonin, AIAA Journal, 4, No.11, 1588-1593 (1966).
47 A.I. Babarickij, S. Klagge and M. Maass, Beitr. Plasmaphys., 22, No.2, 181-194
(1982).
560
References
48 E.O. Johnson and L. Malter, Phys. Rev., 76, No.10, 1411-1417 (1949).
49 Yu.F. Nasedkin, G.V. Levadnyi, A.A. Serov, et al, Teplofiz Vysokikh Temp., 23, No.1,
156-162 (1985).
50 S. Klagge, M. Maass and A. Serov, Beitr. Plasmaphys., 25, No.3, 255-262 (1985).
51 R.L.F. Boyd adn J.B. Thompson, Proc. Roy. Soc., A252, No.1268, 102-119 (1959).
52 J.B. Thompson, Ibid, 262, No.1311, 503-528 (1961).
53 M.J. Kusner, J. Appl. Phys., 53, No.4, 2939-2946 (1982).
54 V.L. Fedorov, ZhTF, 55, No.5, 926-929 (1985).
55 A.P. Mezentsev, A.S. Mustafaev and V.L. Fedorov, ZhTF, 55, No.3, 544-549 (1985).
56 A.P. Mezentsev, A.S. Mustafaev, V.F. Lapshin and V.L. Fedorov, ZhTF, 56, No.11,
2104-2110 (1986).
57 V.L. Fedorov and A.P. Mezentsev, ZhTF, 57, No.3, 595-597 (1987).
58 A.P. Mezentsev, A.S. Mustafaev and V.L. Fedorov, ZhTF, 58, No.6, 1096-1101
(1988).
59 A.I. Lukovnikov, ZhTF, 43, No.7, 1478-1483 (1973).
60 V.L. Granovskii, Electric current in gas, GITTL, Moscow and Leningrad (1952).
61 R.V. Radchenko, A.G. Shtoik and A.S. Zel'skii, ZhTF, 45, No.10, 2225-2228 (1975).
62 A.M. Devyatov and M.A. Mal'kov, Proc. of 6th Nat. Conf. on Low-temperature Plasma
Physics, Leningrad University, Leningrad (1983), pp.314-316
Chapter 11
1 I. Langmuir, Collected works, C.G. Suits (ed), Pergamon Press, Oxford (1961), pp.3-
5.
2 Yu.M. Kagan and V.I. Perel', UFN, 81, No.3, 409-452 (1963).
3 F. Chen, Plasma diagnostics (Edited by R. Huddlestone and Leonard), Mir, Moscow
(1967).
4 O.V. Kozlov, Electric probe in plasma, Atomizdat, Moscow (1969).
5 L. Schott, Methods of plasma examination (edited by V. Lochte-Holdgreven), Mir,
Moscow (1971).
6 P. Chan, L. Telbot and K. Turyan, Electric probes in stationary and moving plasma.
Theory and application, Mir, Moscow (1978).
7 Yu.A. Ivanov, Yu.A. Lebedev and L.S. Polak, Methods of contact diagnostics in non-
8 B.V. Alekseev and V.A. Kotel'nikov, Probe methods of plasma diagnostics,
Energoatomizdat, Moscow (1988).
9 M.J.M. Parrot, L.R.O. Storey, L.W. Parker and J.G. Laframboise, Phys. Fluids, 25,
No.12, 2388-2400 (1982).
10 Yu.B. Golubovskii, V.M. Zakharova, V.N. Pasunkin and L.D. Tsendin, Fiz. Plazmy, 7,
No.3, 620-628 (1981).
11 N.A. Gorbunov, N.B. Kolokolov and A.A. Kudryavtsev, Fiz. Plazmy, 15, No.12, 1513-
1520 (1989).
12 B.I. Davydov and L.I. Zmanovskaya, ZhTF, 6, No.7, 1244-1255 (1989).
13 C.H. Su and S.H. Lam, Phys. Fluids, 6, No.10, 1479-1491 (1963).
14 I.M. Cohen, Ibid, pp.1492-1499.
15 P.R. Smy, Adv. Physics, 25, No.5, 517-553 (1976).
16 V.Z. Kompaniets, A.A. Ovsyannikov and L.S. Polak, Chemical reactions in turbulent
gas and plasma flows, Nauka, Moscow (1979).
17 M.S. Benilov, Teplofiz. Vysokikh Temp., 26, No.5, 993-1004 (1988).
18 M. Mitchner and Ch. Kruger, Partially ionized gases, Mir, Moscow (1976).
19 I. MacDaniel' and E. Meeon, Mobility and diffusion of ions in gases, Mir, Moscow
(1976).
20 I.L. Pankrat'eva and V.A. Polyanskii, Izv. AN SSSR MZhG, No.2, 103-112 (1979).
21 B.V. Alekseev and A.M. Grishin, Physical cast dynamics of reacting media, Vyssh.
Shk. (1985).
22 A.F. Kolesnikov and G.A. Tirskii, Molecular gas dynamics, Nauka, Moscow (1982).
23 D. Girshfel'der, Ch. Kertiss and R. Berd, Molecular theory of gases and liquids, IL,
Moscow (1961).
24 V.M. Zhdanov, Transport phenomena in multicomponent plasma, Energoizdat, Moscow
561
Plasma Diagnostics
(1982).
25 V.V. Gogosov and V.A. Polyanskii, Itogi Nauki Tekhn. Mekhan. Zhidkosti Gaza,
Vol.10, VINITI, Moscow (1976).
26 J. Fertsiger and G. Kaper, Mathematical theory of transport processes in gases, Mir,
Moscow (1976).
27 B.M. Smirnov, Physics of slightly ionized gas in problems with solutions, Nauka,
Moscow (1985).
28 A.A. Radtsig and B.M. Smirnov, Plasma chemistry, Vol.11, Energoatomizdat, Moscow
(1984), pp.170-200.
29 H.W. Ellis, R.Y. Pai, E.W. McDaniel, et al, Atom. Nucl: Data Tabl., 17, No.3, 177-210
(1976).
30 H.W. Ellis, E.W. McDaniel, D.L. Albritton, et al, Ibid, 22, No.3, 179-217 (1978).
31 N.L. Aleksandrov, A.M. Konchakov, A.P. Napartovichand A.N. Starostin, Plasma
chemistry, Vol.11, Energoatomizdat, Moscow (1984), pp.3-45.
32 B.Ya. Moizes and G.E. Pikus (eds), Thermal emission converters and low-temperature
plasma, Nauka, Moscow (1973).
33 F.G. Baksht and V.G. Yuriev, ZhTF, 49, No.5, 905-944 (1979).
34 F.G. Baksht, V.G. Ivanov and V.G. Yuriev, Inv. Pap. 18th Intern. Conf. Phenomena
Ionized Gases, Swansea (1987), pp.206-218.
35 L. Huxley and R. Crompton, Diffusion and drift of electrons in gases, Mir, Moscow
(1977).
36 L.M. Biberman, V.S. Vorobiev and I.T. Yakubov, Kinetics of non-equilibrium low-
temperature plasma, Nauka, Moscow (1982).
37 J. Dutton, J. Phys. Chem: Ref. Data, 4, No.3, 577-856 (1975).
38 J.W. Gallagher, E.C. Beaty, J. Dutton and L.C. Pitchford, Ibid, 12, No.1, 109-152
(1983).
39 J. Iticawa, Atom Data Nucl: Data Tabl., 14, No.1, 1-10 (1974).
40 J. Iticawa, Ibid, 21, No.1, 69-75 (1978).
41 S.R. Hunter and L.S. Christophorou, Electron-molecule interactions and their applications,
L.G. Christophorou (ed), Vol.2,1.6, Academic Press, Orlando (1984), pp.89-219.
42 I.B. Chekmarev, ZhTF, 50, No.1, 48-53 (1980).
43 M.S. Benilov, V.I. Kovbasyuk and A.V. Lyashko, 10th Intern. Conf. MHD Electrical
Power Generation, Vol.2, Tiruchirapally, India (1989), pp.X105-X109.
44 P.M. Chung, Phys. Fluids, 12, No.8, 1623-1634 (1969).
46 Lem, Raket. Tekhnika Kosmonavtika, 2, 43-51 (1964).
47 C.H. Su, Raket. Tekhnika Kosmonavtika, 3, No.5, 52-61 (1965).
48 M.S. Benilov, G.G. Bochkarev, A.E. Buznikov, et al, Izv. AN SSSR. MZhG, No.1, 150-
160 (1983).
49 W.B. Bush and F.E. Fendell, J. Plasma Phys., 4, Pt.2, 317-334 (1970).
50 M.S. Benilov, Izv. AN SSSR. MZhG, No.5, 145-152 (1982).
51 P.C.T. de Boer and G.S.S. Ludford, Plasma Phys., 17, No.1, 29-43 (1975).
52 M.S. Benilov, V.F. Kosov, B.V. Rogov and V.A. Sinel'shikov, Teplofiz. Vysokikh
Temp., 25, No.3, 573-581 (1987).
53 M.S. Benilov, G.G. Bochkarev, V.I. Kovbasyuk and E.P. Reshetov, Diagnostics of
plasma in MHD generators, Electrophysical processes in MHD channels, IVTAN,
Moscow (1990).
54 A.A. Yastrebov, ZhTF, 42, No.6, 1143-1153 (1972).
55 A.A. Yastrebov, ZhTF, 4, No.6, 809-820 (1972).
56 F.H. Dorman and J.A. Hamilton, Inter. J. Mass Spectrom. Ion Phys., 24, No.4, 359-
361 (1977).
57 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, Chemical reactions in non-equilibrium
plasma, (edited by L.S. Polak), Nauka, Moscow (1983).
58 N.N. Barano, M.S. Benilov, G.G. Bochkarev, et al, ZhPMTF, No.3, 13-19 (1983).
59 V.A. Kotel'nikov and V.P. Demkov, IFZh, 52, No.2, 224-227 (1987).
60 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Izv. AN SSSR, MZhG, No.6, 114-
119 (1987).
61 G.S. Aravin, R.A. Vlasov, Yu.K. Karasevich, et al, Izv. AN SSSR, MZhG, No.5, 163-
562
References
169 (1988).
62 P.A. Vlasov, Yu.K. Karasevich, I.L. Pankrat'eva and V.A. Polyanskii, Teplof. Vysokikh
Temp., 26, No.6, 1047-1056 (1988).
63 V.P. Demkov, Mathematical modelling of the transport processes in plasma taking
surface effects into account, MAI, Moscow (1989).
64 V.A. Kotel'nikov and T.A. Gurina, IFZh, 52, No.5, 856 (1989).
65 T.A. Gurina, Flows of particles and the structure of the perturbed zone in the vicinity
of charged solid moving implants, Dissertation, MAI, Moscow (1987).
66 M.S. Benilov and G.A. Tirskii, PMM, 43, No.2, 288-304 (1979).
67 J. Stratton, Electromagnetism phenomena, Gostekhizdat, Moscow and Leningrad
(1948).
68 M.S. Benilov, B.V. Rogov and G.A. Tirskii, ZhPMTF, No.3, 5-13 (1982).
69 I. Khoisler, Magnitnaya Gidrodinamika, No.2, 103-110 (1967).
70 M.A. Benilov and Z.G. Kamalov, Magnitnaya Gidrodinamika, No.1, 41-49 (1988).
71 R.E. Kiel, J. Appl. Phys., 40, No.9, 3668-3673 (1969).
72 K.N. Ul'yanov, ZhTF, 40, No.4, 790-798 (1970).
74 E.F. Prozorov, Ya.I. Londer, K.P. Novikova and K.N. Ul'yanov, Teplofiz. Vysokikh
Temp., 18, No.1, 164-168 (1980).
75 K.N. Ul'yanov, Teplofiz. Vysokikh Temp., 16, No.3, 492-496 (1978).
76 M.S. Benilov, V.I. Kyuvbasyuk and G.A. Lyubimov, DAN SSSR, 266, No.4, 812-816
(1982).
77 N. Stahl and C.H. Su, Phys. Fluids, 14, No.7, 1366-1376 (1971).
78 B. Tolbot, Raket. Tekhnika Kosmonavtika, 14, No.1, 95-104 (1976).
79 T. Hirano, Bull. JSME, 17, No.107, 625-632 (1974).
80 R.M. Clements, B.M. Oliver, A.I. Noor, P.R. Smy, Electron. Lett., 12, No.11, 274-275
(1976).
81 A.I. Noor, P.R. Smy and R.M. Clements, J. Phys. D., 10, No.12, 1643-1651 (1977).
82 D. de Boer, Raket. Teknika Kosmonavtika, 15, No.5, 106-115 (1977).
83 R.M. Clements and P.R. Smy, J.Phys. D., 14, No.6, 1001-1008 (1981).
84 P.R. Smy, Can. J. Phys., 54, No.15, 1627-1636 (1976).
85 M.S. Benilov, B.V. Rogov and G.A. Tirskii, Teplofiz. Vysokikh Temp., 19, No.5, 1031-
1039 (1981).
86 D.A. Frank-Kamenetskii, Diffusion and heat transfer in chemical kinetics, Nauka,
Moscow (1987).
87 A.A. Zhukauskas, Convective transport in heat exchanges, Nauka, Moscow (1982).
88 B.S. Petukhova and V.K. Shikova, A handbook of heat exchangers (translated from
English), Vol.1, Energoatomizdat, Moscow (1987).
89 H. Sudzi, Raket. Tekhnika Kosmonavtika, 11, No.1, 119-121 (1973).
90 A. Kanzawa and S. Monouchi, Intern. Chem. Eng., 16, No.1, 184-189 (1976).
91 M. Sudzuki and A. Kanzava, Raket. Tekhnika Kosmonavtika, 17, No.12, 26-33
(1979).
92 Yu.P. Gupalo, A.D. Polyanin and Yu.S. Ryazantsev, Heat and mass exchange of
reacting particles with the flow, Nauka, Moscow (1985).
93 M.S. Benilov and B.V. Rogov, Proc. of the Nat. Conf. on the Electrophysics of
Combustion, Karaganda State University, Karaganda (1990), pp.37-38.
94 S.. Chang, Raket. Tekhnika Kosmonavtika, 3, No.5, 22-33 (1965).
95 M.S. Benilov, Aerodynamics of hypersonic velocities in the presence of blow, (edited
G.A. Tiriskii), Publishing House of the Moscow State University, Moscow (1979),
pp.167-173.
96 M.V. Zakharov, Aerophysics and geocosmic investigations, MFTI, Moscow (1983).
99 O.V. Dobrocheev and V.P. Motulevich, Proc. of Nat. Symp. on the Methods of
Aerophysical Investigations, ITPM SO AN SSSR, Novosibirsk (1976).
100 E.K. Chekalin, L.V. Chernikh, M.A. Novgorodov and N.V. Khandurov, Ibid, ITPM SO
AN SSSR, Novosibirsk (1976).
101 F.G. Baksht and O.G. Rutkin, ZhTF, 49, No.12, 2549-2558 (1979).
563
Plasma Diagnostics
102 E.K. Chekalin and L.V. Chernikh, ZhPMTF, No.1, 41-47 (1981).
103 X. Chen, J. Phys. D., 16, No.2, L29-L32 (1983).
104 M.S. Benilov and G.A. Tirskii, ZhPMTF, No.6, 16-24 (1979).
105 S.V. Bushuev, E.F. Prozorov and K.N. Ul'yanov, Teplof. Vysokikh Temp., 26, No.3,
582-588 (1988).
106 M.B. Hopkins and W.G. Graham, Rev. Sci. Instrum., 57, No.9, 2210-2217 (1986).
107 M.B. Hopkins, W.G. Graham and T.J. Griffin, Rev. Sci. Instrum., 58, No.3, 475-476
(1987).
108 A.A. Vanin, M.G. Kasparov, A.V. Mokhov, et al, Prepr./IVTAN, Moscow, No.3-286
(1990).
109 R.M. Clements and P.R. Smy, J. Energy, 2, No.1, 53-58 (1978).
110 I.A. Vasil'eva, A.P. Nefedov and S.E. Self, Magneto-hydrodynamic energy conversion.
Open cycle, (edited by B.Ya. Shumyatskii and M. Petrik), Nauka, Moscow (1979).
111 V.M. Batenin, I.A. Vasil'eva and V.F. Kosov, Teplofiz. Vysokikh Temp., 20, No.2,
229-235 (1982).
112 I.A. Vasil'eva, Magneto-hydrodynamic energy conversion. Physio-technical aspects,
(edited by V.A. Kirillin and A.E. Sheindlin), Nauka, Moscow (1983).
113 A. Anders, A.P. Ershov, K.Sh. Isaev and I.B. Timofeev, Teplofiz. Vysokikh Temp. 25,
No.4, 743-747 (1987).
114 G.A. Batyrbekov, E.A. Belyakova, M.S. Benilov, et al, Contr. Pap. 19th Intern. Conf.
Phenomena Ionized Gases, Vol. 4, Belgrade (1989).
115 R.M. Clements and P.R. Smy, J. Phys. D., 7, No.4, 551-562 (1974).
116 M.I. Ashin, I.A. Vasil'eva, V.F. Kosov and A.P. Nefedov, Fiz. Plazmy, 1, No.3, 483-
487 (1975).
117 M.V. Zake and K.S. Landman, Izv. AN Latv. SSR. Ser. Fiz i Tekhn. Nauk., No.2, 79-
85 (1977).
118 N. Negishi and I. Kimuar, Bul. JSME, 23, No.181, 1171-1179 (1975).
119 T. de Boer, Raket. Tekhnika Kosmonavtika, 13, No.8, 182-184 (1975).
120 C.S. Maclatchy, Combust. Flame, 36, No.2, 171-178 (1979).
121 C.S. Maclatchy and R. Didsbury, Can. J. Phys., 57, No.3, 381-384 (1979).
122 I.A. Vasil'eva and V.F. Teplofiz. Vysokikh Temp., 19, No.5, 1022-1030 (1981).
123 K.D. Annen, P.J. Kuzmenko, R. Keating and S.A. Self, J. Energy, 5, No.1, 31-38
(1981).
124 J. Lachmann and G. Winkelmann, Beit. Plasmaphys., 22, No.1, 55-63 (1982).
125 M. Rosenbaum and J.F. Louis, J. Appl. Phys., 53, No.6, 4088-4092 (1982).
126 M.S. Benilov and B.V. Rogov, Proc. 2nd National seminar: Elementary processes in
plasma of electronegative gases, EGU, Erevan (1984), pp. 35-37.
127 P.M. Hierl and J.F. Paulson, J. Chem. Phys., 80, No.10, 4890-4900 (1984).
128 M.S. Benilov, V.M. Grine, A.A. Lash, et al, Teplof. Vysokikh Temp., 28, No.3, 620-
622 (1990).
129 B.M. Oliver, R.M. Clements and P.R. Smy, J. Phys. D., 9, No.12, 1715-1718 (1976).
130 F.G. Baksht, ZhTF, 48, No.10, 2019-2026 (1978).
131 C.R. Giles, R.M. Clements and P.R. Smy, J. Phys. D., 12, No.10, 1685-1697 (1983).
132 E.F. Prozorov and K.N. Ul'yanov, Teplof. Vysokikh Temp., 21, No.3, 538-543 (1983).
133 E.F. Prozorov and K.N. Ul'yanov, Teplof. Vysokikh Temp., 22, No.6, 1179-1185
(1983).
134 Yu.S. Akishev and A.P. Napartovich, DAN SSSR, 242, No.4, 812-815 (1978).
136 V.A. Rozhansky and L.D. Tsendin, ZhTF, 48, No.8, 1647-1653 (1978).
137 F.G. Baksht, ZhTF, 49, No.2, 439-441 (1979).
138 B.V. Kuteev, V.A. Rozhansky and L.D. Tsenin, Beit. Plasmaphys., 19, No.2, 123-126
(1979).
139 V.P. Demkov, V.A. Kotel'nikov and A.A. Moskalenko, IFZh, 57, No.4, 692-693 (1989).
140 Ya.F. Volkov, V.G. Dyatlov and N.I. Mitina, Diagnostics of turbulent flows, Naukova
dumka, Kiev (1983).
141 M.S. Barad and I.M. Cohen, Phys. Fluids, 17, No.4, 724-734 (1974).
142 F.G. Baksht, G.A. Dyuzhev, V.B. Kaplan, et al, Problem diagnostics of low-temperature
plasma, Preprint of the A.F. Ioffe Institute of Technical Physics, AN SSSR,
564
References
Leningrad, No.532, 57 (1978).

143 F.G. Baksht, G.A. Dyuzhev, Tsirkel', et al, Problem diagnostics of low-temperature
plasma, Preprint of the A.F. Ioffe Institute of Technical Physics, Leningrad, No.533,
52 (1978).
144 G.A. Dyuzhev, Near-electrode phenomena in high-current arc and thermoemission
cathodes, A.F. Ioffe Institute of Technical Physics, AN SSSR, Leningrad (1980).
145 E.M. Stepanov and B.G. D'yachkov, Ionization in the flame and the electric field,
Metallurgiya, Moscow (1968).
146 J. Lauton and F. Bainberg, Electric aspects of combustion, Energiya, Moscow
(1976).
147 S. Matsumura, J.S. Chang and S. Teii, Contr. Pap. 18th Intern. Conf. Phenomena
Ionized Gases, Swansea, Vol.4, (1987), pp.620-621.
148 V. Roter and F. Bergman, Physics and technology of low-temperature plasma, ITMO
im A.V. Lykova AN BSSR, Minsk (1977).
149 H.F. Calcote and I.R. King, 5th Sympos. Intern. Combustion, Pitsburg, USA (1954),
pp.423-434.
150 M. Mackey, Raket. Teknika Kosmonavtika, 6, No.8, 186-188 (1968).
151 R. Carabetta and R.P. Porter, 12th Sympos. Intern. Combustion, Poiters, France
(1968), pp.423-434.
152 D.E. Jensen and S.C. Combust. Flame, 13, No.2, 219-222 (1969).
153 H. Tsudzi, Raket. Tekhnika Kosmonavtika, 11, No.6, 153-154 (1973).
154 E.M. Egorova, A.V. Kashevarov, E.M. Fomina and N.S. Tskhai, Teplofiz. Vysokikh
Temp., 26, No.3, 577-581 (1988).
155 L. Katona and Gh. Nichifor, 10th Intern. Conf. MHD Electrical Power Generation,
Tiruchirapalli, Vol.3 (1989), pp.X127-X131.
Chapter 12
1 P. Chan, L. Telbot and K. Turyan, Electric probes in stationary and moving plasma,
Mir, Moscow (1978).
2 I.L. Pankrat'eva and V.A. Polyanskii, Izv. AN SSSR. MZhG, No.2, 102-120 (1979).
3 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Fiz. Gor. Vzriva, 18, No.1, 49-57
(1982).
4 R. Huddlestone and S. Leonard (eds), Plasma diagnostics, Mir, Moscow (1967).
5 Yu.N. Belyaev, V.A. Polyanskii and E.G. Shapiro, Aerodynamics of high velocities,
Publishing House of the Moscow State University, Moscow (1979).
6 V.A. Polyanskii, ZhPMTF, No.5, 11-17 (1964).
7 I. MacDaniel and E. Mezon, Mobility and difusion of ions in gases, Mir, Moscow
(1976).
8 D.E. Golden and H. Bandel, Phys. Rev., 149, No.1, 58-59 (1966).
9 I.L. Pankrat'eva and V.A. Polyanskii, Aerodynamics of hypersonic flows in the
presence of blow, Publishing House of the Moscow State University, Moscow (1979).
11 T.A. Cool and P.J.H. Tjossem, Gas-Phase Chemiluminescence and Chemi-Ionization,
A. Fontijn (ed), Amsterdam, North Holland (1985).
12 B.V. Alekseev and V.A. Kotel'nikov, Probe method of plasma diagnostics,
Energoatomizdat, Moscow (1988).
13 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Experimental and theoretical
investigations of plasma chemical processes, (edited by L.S. Polak), Nauka, Moscow
(1984).
14 D. Hasted, Physics of atomic collisions, Mir, Moscow (1965).
15 V.D. Rusanov and A.A. Fridman, Physics of chemically active plasma, Nauka,
Moscow (1984).
16 A.V. Eletskii and B.M. Smirnov, Modelling and methods of calculating chemical
processes in low-temperature plasma (edited by L.S. Polak), Nauka, Moscow (1974).
17 P. Bailey and K. Turyan, Raket. Tekhnika i Kosmonavtika, 11, No.9, 12-13 (1973).
18 T. Luzzi and R. Jenkins, Raket. Teknika i Kosmonavtika, 9, No.12, 126-132 (1971).
19 B.V. Alekseev, V.A. Kotel'nikov and V.V. Cherepanov, Teplofiz. Vysokikh Temp., 22,
No.2, 395-396 (1984).
565
Plasma Diagnostics
20 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Izv. AN SSSR. MZhG, No.5, 163-
169 (1988).
21 P.A. Vlasov, Yu.K. Karasevich, I.L. Pankrat'eva and V.A. Polyanskii, Teplofiz.
Vysokikh Temp., 26, No.6, 1047-1056 (1988).
22 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Chemical reactions in non-
equilibrium, (edited by L.S. Polak), Nauka, Moscow (1983).
23 B.M. Smirnov, Atomic collisions and elementary processes in plasma, Atomizdat,
Moscow (1978).
24 G. Messi, Negative ions, Mir, Moscow (1981).
25 F. Bruair, D. MacGowen (eds), Physics of ion- and electron-ion collisions, Mir,
Moscow (1986).
26 J. Lauton, F. Bainberg, Electric aspects of combustion, Energiya, Moscow (1976).
27 P.A. Vlasov, Probe diagnostics of non-stationary chemically reacting high-pressure
plasma, Dissertation, IKhF AN SSSR, Moscow (1983).
28 B.V. Alekseev and V.A. Kotel'nikov, Experimental and theoretical investigations of
plasma chemical processes, (edited by L.S. Polak), Nauka, Moscow (1984).
29 G.S. Aravin, Y.K. Karasevich, P.A. Vlasov, et al, Proc. 15th Intern. Conf. Phenomena
Ionized Gases, Pt.2, Minsk (1981), pp.957-958.
30 S.A. Losev and V.A. Polyanskii, Izv. AN SSSR. MZhG, No.1, 176-183 (1968).
31 C.F. Hansen, Phys. Fluids, 11, No.4, 904-908 (1968).
32 I. McDaniel, Collisions processes in ionised gases, Izd-vo Inostr. Lit., Moscow
(1966).
33 V.K. Dushin, Tr. Inst. Mekh. MGU, No.21, part 2, 35-43, Moscow (1973).
34 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Fiz. Gor. Vzryva, 20, No.4, 70-77
(1984).
35 B.M. Smirnov, Ions and excited atoms in plasma, Atomizdat, Moscow (1974).
36 J. Tarstrup and W.J. Heikkila, Radio Science, 7, No.4, 493-502 (1972).
37 F.G. Baksht, hTF, 48, No.10, 2019-2026 (1978).
38 E.F. Prozorov and K.N. Ul'yanov, Teplofiz. Vysokikh Temp., 21, No.3, 538-543
(1983).
39 E.F. Prozorov and K.N. Ul'yanov, Teplofiz. Vysokikh Temp., 21, No.6, 1179-1185
(1983).
40 G.S. Aravin, P.A. Vlasov, Yu.K. Karasevich, et al, Izv. AN SSSR. MZhG, No.6, 114-
119 (1987).
Chapter 13
1 J.R. Hollahan and A.T. Bell (eds), Techniques and Application of Plasma Chemistry,
Wiley Interscience Publication, New York (1974), p.403.
2 B.V. Tkachuk and V.M. Kolotyrkin, Production of thin polymer films from gas phases,
3 N. Isbrooke and D. Braun (eds), Plasma technology in the production of SBIS, Mir,
Moscow (1987).
4 H. Deutsch and S. Klagge, Ernst-Moritz-Arnold Univ. Section Physik/Electronik.
Rep. 1977/78, Greifswald (1978), pp.1-26.
5 H. Deutsch and S. Klagge, Beit. Plasma Phys., 19, No.1, 49-57 (1979).
6 G.K. Vinogradov and Yu.A. Ivanov, Khim. Vys. Energ., 12, No.6, 542-546 (1978).
7 G.K. Vinogradov, Yu.A. Ivanov and L.S. Polak, Khim. Vys. Energ., 13, No.1, 84-85
(1979).
8 Yu.A. Ivanov, Yu.A. Lebedev and L.S. Polak, Methods of contact diagnostics and
non-equilibrium plasma chemistry, Nauka, Moscow (1981).
9 I.I. Amirov, G.K. Vinogradov and D.I. Slovetskii, Mechanisms of plasma chemical
reactions of hydrocarbons and carbon-containing molecules, Ch.1,(edited by L.S.
Polak), INKhS AN SSSR, Moscow (1987).
10 Yu.A. Ivanov, Physico-chemical processes in non-equilibrium plasma of hydrocarbons
and carbon-containing molecules, IKhF AN SSSR, Moscow (1989).
11 V.A. Pliskin and S.Zh. Zanin, Thin film technology, a handbook, Vol.2 (edited by L.
Maissel and R. Gleng ), Sov. Radio, Moscow (1977).
12 G.K. Vinogradov, Yu.A. Ivanov and L.S. Polak, Khim. Vys. Energ., 14, No.2, 174-178
566
References
(1980).
13 G.K. Vinogradov, Yu.A. Ivanov, L.S. Polak and V.N. Timakin, Khim. Vys. Energ., 14,
No.5, 461-465 (1980).
14 S.G. Garanin, Yu.A. Ivanov and V.N. Timakin, Khim. Vys. Energ., 15, No.2, 183-184
(1981).
15 Yu.A. Ivanov, L.S. Polak and V.N. Timakin, Khim. Vys. Energ., 15, No.2, 181-182
(1981).
16 P.R. Emtage and W. Tantraport, Phys. Rev. Lett., 8, No.7, 267-268 (1962).
17 A. Bradley and J.F. Hammes, J. Electrochem. Soc., 110, No.1, 15-22 (1963).
18 H.T. Mann, J. Appl. Phys., 35, No.7, 2173-2179 (1964).
19 R.W. Christy, Ibid, 2179-2184.
20 M. Gazicki and H. Yasuda, Plasma Chemistry and Plasma Processing, 3, No.3, 279-
329 (1983).
21 D.G. Simmons, Thin film technology, a handbook, Vol.2, (edited by L. Maissel and
R. Gleng), Sov. Radio, Moscow (1977).
22 H. Carchano and M. Valentin, Thin Solid Films, 31, No.3, 335-349 (1975).
23 F.-W. Breitbarth and H.-J. Tiller, Wiss. Z. Fridrich-Schiller Univ., Jena, Math.-Nat.
R (1976).
24 F. Gronlund, Prib. Nauch. Issled., No.10, 100-102 (1982).
25 G.K. Vinogradov, G.Zh. Imanbaev and D.I. Slovetskii, Synthesis of compounds in
plasma containing hydrocarbons (edited by L.S. Polak), INKhS AN SSSR, Moscow
(1985).
26 G.K. Vinogradov, G.Zh. Imanbaev and D.I. Slovetskii, Khim. Vys. Energ., 19, No.5,
461-464 (1985).
27 G.K. Vinogradov, G.Zh. Imanbaev and D.I. Slovetskii, Mechanism of plasma chemical
reactions of hydrocarbons and carbon-containing molecules, Ch.1, ( edited by L.S.
Polak), INKhS AN SSSR, Moscow (1987).
28 T. Harada, K. Gamo and S. Namba, Jap. J. Appl. Phys., 20, No.1, 259-264 (1981).
29 J. Wood, J. Phys. Chem., 67, No.7, 231-239 (1963).
30 D.I. Slovetskii, Mechanisms of chemical reactions in non-equilibrium plasma, Nauka,
Moscow (1980).
31 G.K. Vinogradov, Yu.A. Ivanov and Yu.A. Lebedev, Plasma chemical reactions and
processes, (edited by L.S. Polak), Nauka, Moscow (1977).
32 C.M. Ferreira and A. Ricard, J. Appl. Phys., 54, No.5, 2261-2271 (1983).
33 A.A. Labrenko, Recombination of hydrogen atoms on the surface of solids, Naukova
dumka, Kiev (1973).
Chapter 14
1 R. Friedman and J.A. Cyphers, J. Chem. Phys., 23, No.10, 1875-1880 (1955).
2 J. Grey, P.F. Jacobs and M.P. Sherman, Prib. Nauch. Issled., No.7, 29-33 (1962).
3 J. Grey, Rev. Sci. Instrum., 34, No.8, 857-859 (1963).
4 J. Cordero, F.W. Diederich and H. Hurwicz, Aero-space Eng., 22, No.1, 166-191
(1963).
5 F. Fruchtman, Raket. Tekhnika i Kosmonavtika, No.8, 186-187 (1963).
6 J. Grey and P.F. Jacobs, Raket. Tekhnika i Kosmonavtika, No.3, 25-31 (1964).
7 J.L. Potter, G.D. Arney and M. Kinslow, IEEE Transactions Nuclear Science, NS-
11, No.1, 145-157 (1964).
8 J. Grey, Instr. Soc. Am. Trans., 4, No.2, 102-114 (1965).
9 J. Grey, USA, Cooled calorimetric probe, Patent No.3167956 USA MKI 2 F 02 S 12/
03, Claim 16.01.64, Published 02.02.65.
10 V.L. Sergeev, IFZh, 9, No.5, 657-666 (1965).
11 J. Grey, M.P. Sherman, P.M. Uil'yams, et al, Raket. Tekhnika i Kosmonavtika, 4,
No.6, 36-45 (1966).
12 U.G. Karden, Raket. Tekhnika i Kosmonavtika, 4, No.10, 12-22 (1966).
13 U. Sprengel, Atompraxis (Direct Information Raumfahrtforschung und Technik, 1/
66), 12, No.2 (1966).
14 G.F. Au and U. Sprengel, Z. Flugwissenschaften, 14, No.4, 188-194 (1966).
15 I.T. Alad'ev, I.G. Kulakov, O.L. Magdasiev, Izv. SO AN SSSR. Ser. Tekh. Nauk.,
567
Plasma Diagnostics
No.10, issue 3, 50-55 (1966).

16 J. Grey and P.F. Jacobs, Raket. Tekhnika i Kosmonavtika, 5, No.1, 98-106 (1967).
17 D.W. Esker, AIAA J., 5, No.8, 1504-1506 (1967).
18 M. Krutil, Raket. Tekhnika i Kosmonavtika, 6, No.1, 124-133 (1968).
19 F.A. Vassalo, AIAA Paper, No.68-391, 1-12 (1968).
20 J. Grey, Space Aeronautics, 50, No.5, 82-86 (1968).
21 R.B. Poup, Raket. Tekhnika i Kosmonavtika, 6, No.1, 124-133 (1968).
22 V.S. Klubnikin and S.V. Dresvin, Uchen. Zap. LGPI im A.I. Gertsena, 384, No.2,
46-62 (1968).
23 J. Lot, Proc. of Int. Symp. in Salzburg, Austria, MHG-Generators, VINITI (1969).
24 A.A. Voropaev, V.M. Gol'dfarb, A.V. Donoskoi, et al, Uchen. Zap. LGPI im A.I.
Gertsena (low-temperature plasma), 384, No.2, 109-115 (1968).
25 A. Vitte, T. Kubota and L. Liz, Raket. Tekhnika i Kosmonavtika, No.5, 91-101
(1969).
26 V.S. Klubnikin, Transfer phenomena in low-temperature plasma, Nauka i Tekhnika,
Minsk (1969).
27 P.F. Massey, L.G. Bek and E.D. Roshke, Teploperedacha, 91, Ser. C, No.1, 77-85
(1969).
28 R.M. Fristrom and A.A. Vestenberg, Structure of the flame, Metallurgiya, Moscow
(1969).
29 R. Neubeck, Angew. Chem., 81, No.22, 936-942 (1969).
30 G.A. Andreev, S.V. Dresvin and V.S. Klubnikin, Physics, technology and application
of low-temperature plasma, Alma-Ata (1970).
31 M.D. Petrov and V.A. Sepp, Teplof. Vys. Temp., 8, No.4, 868-874 (1970).
32 J. Cheylan, J.P. Dauvergne and P. le Goff, Chim.et ind., Gen chim., 103, No.20,
2679-2686 (1970).
33 V.S. Klubnikin, Problems in the physics of low-temperature plasma, Nauka i
Tekhnika, Minsk (1970).
34 M.M. Abu-Romia and B. Bhatia, AIAA Pap., No.81, 1-6 (1971).
35 V.K. Mel'nikov and R.Kh. Takhtaganov, Izv. AN Latv. SSR, Ser. Fiz. Tekhn. Nauk,
No.2, 57-64 (1971).
36 L.A. Anderson and R.E. Sheldahl, AIAAJ., 9, No.9, 1804-1810 (1971).
37 A.V. Donskoi and V.S. Klubnikin, Physics and technology of low-temperature
plasma, Atomizdat, Moscow (1972).
38 A.A. Nekrasov and A.D. Rekin, Izmerit. Tekhnika, No.4, 47-48 (1972).
39 M.D. Petrov and V.A. Sepp, Izmerit. Tekhnika, No.4, 49050 (1972).
40 A.I. Abrosimov, V.A. Pechurkin and S.I. Shorin, Thermophysics, Vol.4: Heat
exchange in the high-temperature gas flow, Mintis, Vilnyus (1972), pp.34-42.
41 V.I. Gudkov, V.P. Motulevich, A.S. Sergeev, et al, Proc. 5th Nat. Conf., Vol.2,
IT SO AN SSSR, Novosibirsk (1972).
42 M.D. Petrov and V.A. Sepp, Thermophysics, Vol.4: Heat exchange in the high-
temperature gas flow, Mintis, Vilnyus (1972), pp.20-33.
43 J. Grey (USA), Enthalpy probe with screened thermally measuring tube, US Patent
No.3665763 USA MKI 2 F 02 S 12/06. Claim 27.06.72, Publ. 30.05.72.
44 K.N. Chakalev, V.I. Gudkov and A.S. Sergeev, Thermophysical properties and gas
dynamics of high temperature media, Nauka, Moscow (1972).
45 S. Katta, J.A. Lewis and W.H. Gauvin, Rev. Sci. Instr., 44, No.10, 1519-1523
(1973).
46 J. Lachmann, Exp. Techn. Phys., 21, No.4, 361-367 (1973).
47 V.I. Gudkov, V.P. Motulevich and K.N. Chakalev, Proc. 6th Nat. Conf. on Generated
Low-temperature Plasma, Ilim, Frunze (1974), pp.424-427.
48 L.A. Gus'kov and Yu.V. Yakhlakov, Equipment for determining the entalpy of the
gas flow, Author's certificate No.459714 SSSR, No.5, Publ. Byuss. Izobr. (1975).
49 S.V. Dresvin and V.S. Klubnikin, Teplofiz. Vys. Temp., 13, No.2, 433-435 (1975).
50 V.S. Klubnikin, Teplof. Vys. Temp., 13, No.3, 473-482 (1975).
51 H. Hoffman, Exp. Techn. Phys., 23, No.4, 389-396 (1975).
53 C.P. Polyakov, O.V. Ryazantsev and V.I. Tverdozhlevov, Teplof. Vys. Temp, 14,
No.2, 423-424 (1976).
568
References
54 A.S. Sergeev, Yu.N. Vorontsov and O.V. Inozemtsev, Method and equipment for
determining gas enthalpy, Author's certificate No.731322, USSR, No.3, Published
in Byul. Izobr. (1976).
55 A.S. Sergeev, Equipment for measuring the heat content of gases, No.17, Author's
certificate No.828048, USSR, Published in Byul. Izobr. (1981).
56 Yu.N. Vorontsov, O.V. Inozemtsev and A.S. Sergeev, Equipment for caliometry
measurements of the gas flow, No.28, Author's certificate No.851228, USSR,
Published in Byul. Izobr. (1981).
57 S.N. Aksenov, A.V. Donskoi, S.Kh. Il'yasova, et al, Proc. of the 9th Nat. Conf. on
Low-temperature plasma generators, Ilim, Frunze (1983), pp.244-245.
58 S.N. Aksenov, S.Kh. Il'yasova, S.Kh. Klubnikin, et al, Teplofiz. Vys. Temp., 23,
No.1, 196-198 (1985).
59 A.G. Zavarzin, G.M. Krylov, V.P. Lyagushkin, et al, Proc. of the 10th Nat. Conf:
Generators of Low-temperature plasma, Ch.2, Nauka i Tekhnika, Minsk (1986),
pp.141-142.
60 M.F. Zhukov, V.P. Lyagushkin and O.P. Solonenko, Automated experimental
stands for complex examination of high temperature heterogeneous jets, No.145-
86, Preprint of Institute of Thermophysics, Siberian Division of the Russian
Academy of Sciences, Novosibirsk (1986).
Chapter 15
1 N. Kimura and A. Kanzava, Raket. Tekhnika i Kosmonavtika, No.3, 120-127 (1965).
2 A. Abdrazakov and Zh. Zheenbaev, Examination of the electric arc in argon, Ilim,
Frunze (1966).
3 V.M. Gol'dfarb and A.M. Uzdenov, Izv. SO AN SSSR, Ser. Tekhn. Nauk, No.3,
Issue No.1, 81-84 (1967).
4 E. Eckert, Introduction into the theory of heat and mass exchange, Goseneroizdat,
Moscow (1957).
5 S.S. Kutateladze, V.M. Borishanskii, a handbook of heat transfer, GITTL, Moscow
and Leningrad (1959).
6 J. Bues, H.J. Patt and J. Richter, Z. Ang. Phys., 22, No.4, 345-352 (1967).
7 K.S. Drellischak, Phys. Fluids, 16, No.5, 616-623 (1963).
8 R. Devoto, Phys. Fluids, 16, No.5, 616-623 (1973).
Chapter 16
1 V.D. Rusanov and A.A. Fridman, Physics of chemically active plasma, Nauka,
Moscow (1984).
2 S.A. Krapivina, Plasma chemical technological processes, Khimiya, Leningrad (1981).
3 V.A. Legasov, V.K. Zhivotov, E.G. Krasheninnikov, et al, DAN SSSR, 238, No.1, 66-
69 (1978).
4 IEE Transations on Plasma Science. Special issue on the physics of RF discharges
for plasma processing. PS-14, No.2, 77-197 (1986).
5 D.B. Graves, AIChEJ, 35, No.1, 1-29 (1989).
7 B.S. Danilin and V.Yu. Kireev, Using low-temperature plasma for etchingn and
cleaning materials, Energoatomizdat, Moscow (1989).
8 V.I. Myshenkov and N.A. Yatsenko, Kvant. Elektron., 8, No.10, 2121-2129 (1981).
9 N.A. Yatsenko, Gas lasers with high-temperature excitation, No.381, Preprint of the
Institute of Powder Metallurgy, AN SSSR, Moscow (1989).
10 D.R. Hall and H.J. Baker, Laser focus world, No.10, 77-80 (1989).
11 H.E. Hugel, SPIE, 650, 2-9 (1986).
12 N.A. Yatsenko, ZhTF, 51, No.6, 1195-1204 (1981).
13 N.A. Yatsenko, Special structure of the hf capacitance discharge and prospects for
its application in laser technology, No.338, Preprint of the Institute of Powder
Metallurgy, AN SSSR, Moscow (1988).
14 N.A. Kaptsov, Electric discharges in gases and vacuum, GITTL, Moscow and
Leningrad (1950).
15 B.E. Bragin and V.D. Matyukhin, 8th Nat. Conf. on Low-temperature Plasma Generators,
IT SO AN SSSR, Novosibirsk (1980).
569
Plasma Diagnostics

17 N.A. Yatsenko, Teplof. Vys. Temp., 20, No.6, 1044-1051 (1982).
18 S.M. Levitskii, ZhTF, 27, No.5, 970-977 and 1001-1009 (1957).
19 A.D. Andreev, ZhPS, 5, No.2, 145-147 (1966).
20 N.A. Kuzovnikov and V.P. Savinov, RE, 18, No.4, 816-822 (1973).
21 A.A. Kuzovnikov and V.P. Savinlov, Vestn. MGU, Fizika, No.2, 215-223 (1973).
22 V.A. Godyak and A.S. Khanneh, IEEE Transactions on Plasma Science, PS-14,
No.2, 112-123 (1986).
23 N.A. Yatsenko, Tr.MFTI, Series General and Molecular Physics, Dolgoprudniy,
Moscow (1978), pp.226-229.
24 N.A. Yatsenko, Tr.MFTI, Ibid (1978), pp.166-170.
26 V.Kh. Goikhman and V.M. Gol'dfarb, ZhPS, 21, No.3, 456-459 (1974).
27 A.S. Kovalev, A.I. Nazarov, A.T. Rakhimov, et al, Fiz. Plazmy, 12, No.10, 1264-1268
(1986).
28 Yu.P. Raizer and M.N. Shneider, Fiz. Plazmy, 13, No.4, 471-479 (1987).
29 Yu.P. Raizer and M.N. Shneider, Fiz. Plazmy, 14, No.2, 226-232 (1988).
30 P. Vidaud, S.M.A. Durrani and D.R. Hall, J. Phys. D: Appl. Phys., 21, No.1, 57-66
(1988).
31 A. Engel', Ionized gases, GIFML, Moscow (1959).
32 Yu.P. Raizer, Physics of gas discharge, Nauka, Moscow (1987).
33 V.I. Myshenkov and N.A. Yatsenko, Fiz. Plazmy, 8, No.3, 543-549 (1982).
35 A.S. Kovalev, E.F. Muratov, A.A. Ozerenko, et al, Pis'ma ZhTF, 10, No.18, 1139-
1142 (1955).
36 D. Banerji and R. Ganguli, Phil. Mag., 13, 494-498 (1932).
37 Kh.A. Jerpetov and G.M. Pateyuk, ZhETF, 28, No.3, 343-351 (1955).
38 H.S. Butler and G.S. Kino, Phys. Fluids, 6, No.9, 1052-1065 (1963).
39 V.A. Godyak, A.A. Kuzovnikov, V.P. Savinov, et al, Vestn. MGU, Fiz., No.2, 126-127
(1968).
40 V.A. Godyak and A.A. Kuzovnikov, Fiz. Plazmy, 1, No.3, 496-503 (1975).
41 A.A. Kuzovnikov, V.L. Kovalevskii, V.P. Savinov and V.G. Yakunin, Proc. 13th Intern.
Conf. Phenomena Ionized Gases, Berlin, DDR (1977), pp.343-344.
42 J. Hay, Canad. J. Res., A16, 191-200 (1938).
43 Ph.Belenguer and J.P. Boeuf, Proc. 19th Intern. Conf. Phenomena Ionized Gases,
Belgrade (1989), pp.394-395.
44 H. Beck, Z.Phys., 97, No.4, 766-775 (1935).
45 V.A. Godyak and O.N. Popov, ZhTF, 47, No.4, 766-775 (1977).
46 R.R.J. Gagne and A. Cantin, J. Appl. Phys., 43, No.6, 2639-2651 (1972).
47 A.V. Aleksandrov, V.A. Godjak, A.A. Kuzovnikov, et al, Proc. 8th Intern. Conf.
Phenomena Ionized Gases, Vienna (1967), p.165.
48 N.A. Yatsenko, Reasons and conditions for formation of constant emf in a capacitance
hf medium pressure discharge, Tartu State University, Tartu (1989), pp.208-210.
49 D. He and D.R. Hall, IEEE Journal Quantum Electronics, QE-20, No.5, 509-516
(1984).
50 P. Hoffman, SPIE, 650, 23-29 (1986).
51 A.G. Akimov, A.V. Koba, N.I. Lipatov, et al, Kvant. Elektron., 16, No.5, 938-949
(1986).
52 N.I. Lipatov, P.P. Pashinin, A.M. Prokhorov, et al, Gas discharge and waveguide
molecular lasers, Vol.17, Tr. IOFAN, Nauka, Moscow (1989).
53 S. Yatsiv, Gas Flow Chemical Laser Conf, S. Rosenwarks (ed), Springer, Berlin,
Germany (1987), pp.252-257.
54 V.I. Myshenkov and N.A. Yatsenko, ZhTF, 51, No.10, 2055-2062 (1981).
55 N.A. Yatsenko, Proc. of the 2nd Nat. Conf: High frequency discharge in wave fields,
Kuibysh Polytechnic Institute, Kuibysh (1989), p.13.
56 D. He and D.R. Hall, J. App. Phys., 54, No.3, 4367-4375 (1983).
57 A.D. Andreev, ZhPS, 5, No.2, 145-147 (1966).
58 A.D. Andreev, Vest. BelGU, Ser.1, No.2, 78-80 (1969).
570
References
59 A.D. Andreev, V.D. Il'in and Yu.N. Lobanov, ZhTF, 36, No.9, 1636-1638 (1966).
60 A.V. Kalmykov, B.Yu. Nezhentsev, A.S. Smirnov, et al, ZhTF, 59, No.9, 93-97 (1989).
61 N.A. Yatsenko, Proc. of 3rd Nat. Conf. on the Physics of Gas Discharges, Ch.1,
KGU, Kiev (1989).
62 N.A. Yatsenko, XX ICPIG. Contrib. Papers, Vol.5, Pisa (1991), pp.1159-1160.
Chapter 17
1 G. Peter, Electronic avalanches and breakdown in gases, Mir, Moscow (1968).
2 E.D. Lozanskii and O.B. Firsov, Theory of the spark, Atomizdat, Moscow (1975).
3 L.B. Loeb, Science, 148, No.3676, 1417-1426 (1965).
4 J.J. Thomson, Recent Researches in Electricity and Magnetism, Clarendon, Oxford
(1893), pp.115-118.
5 R.G. Fowler, Adv. Electron. Phys., 35, 1-84 and 41, 1-78 (1976).
6 J.W. Beams, Phys. Rev., 28, 475-481 (1926).
7 L.B. Snoddy, J.R. Dietrich and J.W. Beams, Ibid, 52, No.4, 739-746 (1937).
8 J.W. Beams, Ibid, 36, No.4, 997-1001 (1930).
9 F.H. Mitchell and L.D. Snoddy, Ibid, 72, 1202-1208 (1947).
10 R.G. Westberg, 114, No.1, 1-17 (1959).
11 M.R. Amin, J. Appl. Phys., 25, 358-363 (1954).
12 G.G. Hudson and L.B. Loeb, Phys. Rev., 123, No.1, 29-43 (1961).
13 H. Tholl, Z. Naturforsch, 19a, 346-359 and 704-706 (1964).
14 A.H. Amin, Proc. 5th Intern. Conf. Phenomena Ionized Gases, Vol.1, Munich (1961),
pp.1003-1016.
15 T. Suzuki, J. Appl. Phys., 42, 3766-3771 (1971).
16 T. Suzuki, Ibid, 44, 4534-4544 and 48, No.12, 5001-5007 (1973).
17 W. Bartolmeyezyk, Ann. Phys., 36, No.6, 485-520 (1939).
18 A.V. Nedospasov and A.E. Novik, ZhTF, 30, 1329-1336 (1960).
19 A.V. Nedospasov, G.M. Sadykhzade and K.I. Efendiev, Teplof. Vys. Temp., 16, No.4,
673-676 (1978).
20 V.P. Abramov and I.P. Mazan'ko, ZhTF, 50, 749-754 (1980).
21 A.W. Trievelpiece and R.W. Gold, J. Appl. Phys., 11, 1784-1793 (1959).
22 P.S. Bulkin, V.N. Ponomarev and G.S. Solntsev, Vestn. MGU, 3, 93-95 (1967).
23 W.P.P. Winn, J. Appl. Phys., 38, No.2, 783-790 (1967).
24 H.E.B. Andersson and R.S. Tobin, Physica Scripta, 9, No.1, 7-14 (1967).
25 H.E.B. Andersson, Ibid, 4, 215-220 (1971).
26 N.I. Vinokurov, V.A. Gerasimov, V.V. Zaponchkovskii and Yu.F. Fomenko, ZhTF, 47,
2512-2521 (1977).
27 F.M. McGehee, Virginia J. Science, 6, No.1, 39-45 (1955).
28 E.I. Asinovskii, L.M. Vasilyak, A.V. Kirillin and V.V. Markovets, Teplof. Vys. Temp.
13, No.1, 40-44 and 195-198, and 13, No.6, 1281-1282 (1975).
29 E.I. Asinovskii, L.M. Vasilyak and V.V. Markovets, Teplof. Vys. Temp., 21, No.3, 577-
590 (1983).
30 R.G. Fowler and J.D. Hood, Phys. Rev., 128, No.3, 991-992 (1962).
31 E.I. Asinovsky, A.V. Kirillin, V.V. Markovets and L.M. Vasiljak, 7th Intern. Sympos.
Discharges Electrical Insulation Vacuum, Publishing House of the Novosibirsk
University (1976).
32 E.I. Asinovsky, L.M. Vasilyak, A.V. Kirillin and V.V. Markovets, 3rd Nat. Conf. on
Plasma Accelerators, Belrussian University, Minsk (1976).
33 O.A. Sinkevich and Yu.V. Trofimov, DAN SSSR, 249, No.3, 597-600 (1979).
34 A.N. Lagar'kov and I.M. Rutkevich, DAN SSSR, 249, No.3, 593-596 (1979).
35 M.J. Lubin and E.R. Resler, Phys. Fluids, 10, No.1, 1-7 (1967).
36 E.I. Asinovskii, L.M. Vasilyak, A.V. Kirillin and V.V. Markovets, Teplof. Vys. Temp.,
18, No.4, 668-676 (1980).
37 V.V. Basov, I.G. Kataev and D.P. Kolchin, RE, 28, No.11, 2206-2210 (1983).
38 V.V. Basov, i.G. Kataev and D.P. Kolchin, Proc. of the 6th Nat. Conf. on the Physics
of Low-temperature plasma, Vol.2, Leningrad University, Leningrad, (1983), pp.423-
425.
39 I.P. Krasyuk, N.I. Lipatov and P.P. Pashinin, Kvant. Elektron., 3, No.11, 2384-2391
571
Plasma Diagnostics
(1976).
40 P. Felzenthal and J. Praud, Electron avalanches and breakdown in gases, Mir,
Moscow (1968).
41 L.M. Vasilyak, Electric and spectral characteristics of nanosecond discharge in
helium, Dissertation, MFTI/Nauk, Moscow (1975).
42 I. Lewis and F. Wells, Millimicrosecond pulsed technology, Moscow (1956).
43 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Teplof. Vys. Temp., 20, No.6,
1189-1191 (1982).
44 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Problems of physics and technology
of nanosecond discharges, IVTAN, Moscow (1982).
45 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Proc. of 6th Nat. Conf. on Physics
of Low-temperature Plasma, Vol.1, Frunze (1983), pp.372-374.
46 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Proc. of 9th Nat. Conf. on Low-
temperature Plasma Generators, Frunze (1983), pp.184-185.
47 E.I. Asinovsky, V.V. Markovets, I.S. Samoilov and A.M. Ul'yanov, Teplofiz. Vys.
Temp., 16, No.6, 1309-1311 (1978).
48 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Proc. of 8th Nat. Conf. on Low-
temperature Plasma Generators, Novosibirsk (1980), pp.217-220.
49 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Formation of the shockwave of the
potential gradient in breakdown of the discharge gap, No.6-067, Preprint IVTAN,
Moscow (1981).
50 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Teplofiz. Vys. Temp., 19, No.3,
587-594 (1981).
51 E.I. Asinovsky, V.V. Markovets and I.S. Samoilov, Proc. 16th ICPIG, Vol.5, Dusseldorf
(1983), pp.758-759.
52 V.V. Kovalev, F.M. Subbotin and E.I. Shubnikov, PTE, No.1, 158-159 (1972).
53 G.S. Vil'dgrubbe, Zh.M. Ronkin and Yu.A. Kolosov, Izv. AN SSSR, Ser. Fiz., 28,
No.2, 377-383 (1964).
54 E.I. Asinovsky, V.V. Markovets, I.S. Samoilov and A.M. Ulianov, PTE, No.1, 222-223
(1982).
55 E.I. Asinovsky, V.V. Markovets, I.S. Samoilov and A.M. Ulianov, Proc. 15th Intern.
Conf. Phenomena Ionized Gases, Minsk (1981), pp.961-962.
56 E.I. Asinovsky, V.V. Markovets and A.M. Ulianov, Proc. 16th Intern. Conf. Phenomena
Ionized Gases: Contributed papers, Vol.2, Dusseldorf (1983), pp.148-149.
57 E.I. Asinovsky, V.V. Markovets and A.M. Ul'yanov, Teplof. Vys. Temp., 22, 667-671
(1984).
58 E.I. Asinovsky, V.V. Markovets, I.S. Samoilov and A.M. Ul'yanov, PTE, No.5, 113-
115 (1984).
Moscow (1986).
60 Smit, Svannak, Fleishman, et al, PNI, 48, No.12, 94-102 (1977).
61 R.Kh. Amirov, E.I. Asinovsky, V.V. Markovets, A.S. Panfilov and I.V. Filyugin, No.3-
183, Preprint IVTAN, Moscow (1986).
62 M.A. Blokhin, Physics of x-rays, GITTL, Moscow (1953).
63 A.G. Abramov, E.I. Asinovskii and L.M. Vasilyak, Kvant. Elektron., 10, No.9, 1824-
1828 (1983).
64 I.S. Samoilov, High-speed breakdown waves in long screen tubes, Dissertation,
IVTAN, Moscow (1985).
65 E.I. Asinovsky, V.V. Markovets, I.S. Samoilov and A.M. Ulianov, Proc. 15th Intern.
Conf. Phenomena Ionized Gases, Minsk (1981), pp.961-962.
66 E.I. Asinovsky, V.V. Markovets and A.M. Ulianov, Proc. 16th Intern. Conf. Phenomena
Ionized Gases: Contributed papers, Vol.2, Dusseldorf (1983), pp.148-149.
67 E.I. Asinovsky, V.V. Markovets, I.S. Samoilov and A.M. Ulianov, Breakdown of
dielectrics and semiconductors, DGU University, Makhachkala (1984).
68 A.M. Ulianov, Spatial-time structure of excitation of gas by a breakdown wave,
Dissertation, IVTAN, Moscow (1985).
69 E.I. Asinovsky, V.V. Markovets and A.M. Ulianov, Teplofiz. Vys. Temp., 22, No.4,
667-671 (1984).
572
References
70 E.I. Asinovskii, V.V. Markovets and A.S. Panfilov, Proc. National symposium on
High-frequency gas breakdown, Tartu State University, Tartu (1989), pp.155-157.
71 A.N. Lagarkov, S.E. Rasponomarev and I.M. Rutkevich, 12th Sympos. Physics
Ionized Gases: Contr. pap., Belgrade (1984), pp.552-555.
72 E.I. Asinovskii, L.M. Vasilyak, V.V. Markovets and Yu.M. Tokunov, DAN SSSR, 263,
No.6, 1364-1366 (1982).
73 V. Paulin, Theory of negativity, OGIZ, Gostekhizdat, Moscow (1941).
74 E.I. Asinovsky, Proc. 16th ICPIG, Invited papers, Dusseldorf (1983), pp.223-231.
75 E.I. Baranchikov, A.V. Gordeev, V.D. Korolev and V.P. Smirnov, Pis'ma ZhTF, 3,
No.3, 106-110 (1977).
76 I.G. Kataev, Shock electromagnetic waves, Sov. Radio, Moscow (1963).
77 A.I. Pavlovskii, L.P. Babich, T.B. Loiko and L.B. Tarasova, DAN SSSR, Vol.281
(1985).
78 A.G. Abramov, E.I. Asinovskii, M.G. Bryukov, L.M. Vaslyak, et al, Effect of fast
electrodes on the development of a wave breakdown in air and the generation of a
nitrogen laser, Preprint of IVTAN, No.6-161, Moscow (1985).
79 S.G. Arutyenian, O.V. Bogdankevich, Iu.F. Bondar, et al, Plasma Phys., 25, 11-24
(1983).
80 G.A. Mesyats, Yu.I. Bychkov and V.V. Kremnev, UFN, 107, No.2, 201-228 (1972).
81 L.M. Vasilyak and V.A. Doinikov, Proc. of 4th Nat. Conf. on Physics of Discharged
Gases, Ch.2, DGU University, Makhachkala, (1988), pp.21-22.
82 Yu.M. Tokunov, E.I. Asinovsky and L.M. Vasilyak, Teplofiz. Vys. Temp., 19, No.3,
491-496 (1981).
83 E.I. Asinovsky, A.G. Abramov and L.M. Vasilyak, Proc. of 4th Nat. Symp. on High-
current Electronics, Ch.2, ISE SO AN SSSR, Tomsk (1982), pp.212-215.
84 A.N. Lagar'kov and I.M. Rutkevich, Problems of the physics and technology of
nanosecond discharges, IVTAN, Moscow (1982).
85 O.A. Sinkevich and Yu.V. Trofimov, Problems of the physics and technology of
nanosecond discharges, IVTAN, Moscow (1982).
86 S.I. Baskakov, Radioelectronic circuits with distributed parameters, Vyssh. Shkola,
Moscow (1980).
87 E.I. Asinovsky, L.M. Vasilyak and V.V. Markovets, Problems of physics and technology
of nanosecond discharges, IVTAN, Moscow (1982).
88 B.B. Slavin and P.I. Sopin, Proc. of 4th Nat. Conf. on the Physics of Gases and
Discharge, Vol.2, DGU University, Makhachkala (1988), pp.43-44.
89 B.B. Slavin ad P.P. Sopin, Proc. of 4th Nat. Conf. on the Physics of Gases and
Discharge, Vol..2, DGU University, Makhachkala (1988).
90 E.I. Asinovsky, L.M. Vasilyak and V.V. Markovets, First Soviet–French Seminar:
Plasma Physics, IVTAN, Moscow (1979).
91 E.I. Asinovsky, L.M. Vasilyak and Yu.M. Tokunov, Teplofiz. Vys. Temp., 17, No.4,
858-860 (1979).
92 E.I. Asinovsky, L.M. Vasilyak and Yu.M. Tokunov, Teplofiz. Vys. Temp., 19, No.4,
873-875 (1981).
93 R.Kh. Amirov, E.I. Asinovsky, A.A. Krolin and V.V. Markovets, Proc. of 4th Nat. Conf:
Interaction of Electromagnet Radiation with Plasma, Fan, Tashkent (1985).
94 R.Kh. Amirov, E.I Asinovsky, A.D. Lukin, et al, Khim. Vys. Energ., 21, No.1, 79-82
(1987).
95 R.Kh. Amirov, E.I. Asinovsky and S.V. Kostyuchenko, Teplofiz. Vys. Temp., 25,
No.6, 793-794 (1987).
96 R.Kh. Amirov, E.I. Asinovsky, S.V. Kostyuchenko, et al, Special features of the
breakdown of plasma after nanosecond discharge in F 2 and S 6, Preprint of IVTAN,
No.8-200, Moscow (1986).
97 E.I. Asinovskii, R.Kh. Amirov, L.M. Vasilyak and V.V. Markovets, Teplofiz. Vys.
Temp., 17, No.5, 912-915 (1979).
98 R.Kh. Amirov, E.I. Asinovsky and V.V. Markovets, Teplofiz. Vys. Temp., 22, No.5,
1002-1005 (1984).
99 R.Kh. Amirov, E.I. Asinovsky and V.V. Markovets, Teplofiz. Vys. Temp., 19, No.1,
47-51 (1981).
573
Plasma Diagnostics
100 R.H. Amirov, E.I. Asinovsky and V.V. Markovets, Teplofiz. Vys. Temp., 2, Pt.2, 424-
425 (1981).
101 R.H. Amirov, E.I. Asinovsky, V.V. Markovets and V.P. Fomin, Proc. 15th ICPIG, Pt.2,
Minsk (1981), pp.669-670.
102 Yu.P. Raizer, Fundamentals of modern physics of gas discharge processes, Nauka,
Moscow (1985).
103 S.V. Kostyuchenko, Formation and breakdown of cold iron - iron plasma of fuel gas
and fluorine, Dissertation, MFTI, Moscow (1987).
Chapter 18
1 V.L. Granovskii, Electric current in gases, GITTL, Moscow (1952).
2 G. Muller, Z. Phys., 151, No.4, 460-482 (1958).
3 W. Finkelburg and S.M. Segal, Phys. Rev., 80, No.2, 258-260 (1950).
4 S. Bennet and J. Connors, IEEE Trans. Nucl. Sci., NS-11, No.1, 109-118 (1964).
5 G. Muller and W. Finkelburg, Z. Angew. Phys., 8, No.6, 282-287 (1964).
6 H. Ringler, Z. Phys., 169, No.2, 273-285 (1962).
7 G. Busz-Peukert and W. Finkelburg, Ibid, 140, No.5, 540-545 (1955).
8 M.J. Block and W. Finkelburg, Z. Naturforsch, 8A, No.11, 758-759 (1953).
9 W. Finkelburg and S.M. Segal, Z. Phys., 83, No.3, 582-585 (1951).
10 V. Finkelburg and G. Mekker, Electric arcs and thermal plasma, IL, Moscow (1961).
11 Zh. Zheenbaev, Laminar plasma torch, Ilim, Frunze (1975).
12 G.P. Petrov, Physical gas dynamics and properties of gases at high temperatures,
13 M.G. Morozov, Teplofiz. Vys. Temp., 5, No.1, 106-114 (1967).
14 A. Abdrazakov, Zh.Zh. Zheenvaev, R.I. Konavko, et al, Using the plasma torch in
spectroscopy, Ilim, Frunze (1970).
15 Zh.Zh. Zheenbaev and V.S. Engel'sht, Two-jet plasma torch, Ilim, Frunze (1983).
16 M.F. Zhukov, A.S. Koroteev and B.A. Uryukov, Applied dynamics of thermal plasma,
Siberian Division, Nauka, Novosibirsk(1975).
17 A.D. Lebedev and N.M. Shcherbik, Izv. SO AN SSSR, Ser. Tekhn. Nauk., No.3, 33-
37 (1979).
Chapter 19
1 L.T. Lap'kina and V.S. Engel'sht, Reduction to a homogeneous optically thin layer in
axisymmetric objects, Manuscript No.6917, Deposited at VINITI, Moscow (1973).
2 A.S. Anyshakov, G.Yu. Dautov and A.P. Petrov, Generators of low temperature
plasma, Energiya, Moscow (1969), pp.394-406..
Chapter 20
1 L.M. Biberman, V.S. Vorob'ev and I.T. Yakubov, Kinetics of non-equilibrium low-
temperature plasma, Nauka, Moscow (1982).
2 B.F. Gordiets, A.I. Osipov and L.A. Shchelepin, Kinetic processes in gases and
molecular lasers, Nauka, Moscow (1980).
3 E.V. Stupochenko, S.A. Losev and A.I. Osipov, Relaxation processes in impact
waves, Nauka, Moscow (1965).
4 V.A. Dudkin, Optika i Spektroskopiya, 24, No.3, 367-371 (1968).
5 G.N. Polyakova and A.I. Ranyuk, Extraction of determination from the velocity
distribution of excited particles from the Doppler probing of spectral lines, Preprint
No.81-1, KhFTI, Khar'kov (1981).
6 V.I. Malyshev, Introduction into experimental spectroscopy, Nauka, Moscow (1979).
7 O.F. Hagena and A.K. Varma, Rev. Sci. Instrum., 39, No.1, 47-52 (1968).
8 A.E. Zarvich, Non-equilibrium processes in flows of rarefied gas, S.S. Kutateladze
and A.K. Rebrova (eds), Siberian Division, Nauka , Novosibirsk (1977), pp.51-78.
9 A.N. Tikhonov and V.Ya. Arsenin, Method of solving incorrect problems, Nauka,
Moscow (1979).
10 E.P. Pavlov and V.D. Perminov, ZhPMTF, No.4, 57-61 (1972).
574
References
11 E.K. Kraulinya, S.Ya. Liepa, A.Ya. Skudra and A.E. Lezdin', Optika i Spektroskopiya,
50, No.1, 50-54 (1979).
12 T.T. Karasheva, M. Malikov, D.K. Otorbaev, 13th Intern. Sympos. Rarefield Gas
Dynamics. Book of Abstracts, Vol.2, Novosibirsk (1982).
13 N. Takahashi, T. Moriya and K. Teshima, Ibid, 491-493.
14 V.F. Turchin, V.P. Kozlov and M.S. Malkevich, UFN, 102, No.3, 345-386 (1970).
15 G.A. Vedernikov and N.G. Preobrazhenskii, Optika i Spektroskopiya, 44, No.1, 204-
205 (1979).
16 D.K. Otorbaev, V.N. Ochkin, N.G. Preobrazhenskii, et al, ZhETF, 81, No.5, 1626-
1638 (1981).
17 V.A. Morozov, Regular methods of solving incorrect problems, Moscow (1974).
18 G.I. Marchuk, Methods of computing mathematics, Nauka, Moscow (1980).
19 V.A. Morozov, Zhurn. Vychisl. Matematiki Mat. Fiz., 11, No.3, 545-558 (1971).
Chapter 21
1 V.B. Neshukaitis, F.F. Belinskis, R.V. Marchyulene and G.I. Batsavichyus, Pribory
i Sistemy Upr., No.10, 16-21 (1976).
2 G.I. Batsyavichyus, R.V. Marchyulene, T.P. Vasilyauskas, et al, Pribor. Tekhn.
Eksperiment., No.4, 238-240 (1977).
3 A.B. Ambrazyavichyus, G.I. Batsyavichyus, A.A. Spudis and V.Yu. Stasyukaitis,
Metallurgical facilties for measurements of high temperatures and plasma parameters,
Khar'kov University Publishing House, Khar'kov (1979), pp.49-51.
4 S.V. Goncharik, A.M. Grigorenko, E.A. Ershov-Pavlov, et al, Pribor. Tekhn. Eksperiment.,
No.5, 223-225 (1985).
5 R. Buteikis, G. Davidavichyus adn I. Lberauskas, Proc. of the 9th Nat. Conf. on Low-
temperature Plasma Generators, Ilim, Frunze (1983), pp.242-243.
6 S.N. Aksenov, A.V. Donskoi, S. Kh. Il'yasova, et al, Proc. of the 9th Nat. Conf. on
Low-temperature Plasma Generators, ilim, Frunze (1983), pp.244-245.
7 S.N. Aksenov, S.Kh. Il'yasova, V.S. Klubnikin, et al, Teplofiz. Vys. Temp., 23, No.1,
196-198 (1985).
8 C.S. Wond, Rev. Sci. Instrum., 56, No.4, 632 (1985).
9 I.V. Knyazhechenko, S.V. Lukoshkov, V.I. Maryin, et al, Pribor. Tekhn. Eksperiment.,
No.2, 260-260-261 (1983).
10 E.Kh. Krieger and U.L. Pikhlak, Proc. of the 9th Nat. Conf. on Low-temperature
Plasma Generators, Ilim, Frunze (1983), pp.240-241.
11 F.Y. Chu and R.M. Cilic, Proc. 7th Intern. Sympos. Plasma Chemistry, Vol.3,
Eindhoven Netherlands (1985), pp.754-757.
12 Yu.E. Nesterikhin, Yu.N. Eolotukhin and Z.A. Lifshits, Avtometriya, No.4, 3-14
(1984).
13 O.Z. Gusev, Yu.N. Zolotukhin, O.V. Prokhozhev and A.P. Yan, Avtometriya, No.4, 15-
20 (1984).
14 M.F. Zhukov, V.P. Lyagushkin and O.P. Solonenko, Automated experimental stand
for the detailed examination of high temperature heterogeneous jets. Current state
and prospects, Preprint No.145-86, ITF, Novosibirsk (1986).
15 M.F. Zhukov and O.P. Solonenko, Izv. SO AN SSSR, Ser. Tekhn. Nauk., No.11,
Issue No.3, 69-86 (1987).
16 A.P. Zinov'ev, O.P. Solonenko and B.V. Tarasov, Subsystem for colour computer
graphics for Elektronika-60 computer, Preprint No.129-85, ITF, Novosibirsk (1985).
17 A.P. Zinov'ev adn O.P. Solonenko, Information-guidance subsystem based on Fortin
for micro and minicomputers of the SM series, Preprint No.147-86, ITF, Novosibirsk
(1986).
18 V.P. Lyagushkin, O.P. Solonenko, Yu.L, Stankevich and V.A. Starikov, Proc. of 11th
Nat. Conf. on Low-temperature Plasma Generators, Ilim, Frunze (1983).
19 A.G. Zavarzin, G.M. Krylov, V.P. Lyagushkin, et al, Proc. of 10th Nat. Conf. on Low-
temperature Plasma Generators, Vol.2, Nauka i Tekhnika, Minsk (1986), pp.141-
142.
20 O.P. Solonenko, Proc. Intern. Conf. Fluid Mechanics, Beijing, China (1987), pp.800-
806.
575
Plasma Diagnostics
21 M.F. Zhukov, V.P. Lyagushkin and O.P. Solonenko, Proc. 8th Intern. Sympos.
Plasma Chemistry, Vol.4, Tokyo, Japan (1987), pp.1995-1999.
22 S.M. Gusel'nikov, A.G. Zavarzin, V.P. Lyagushkin, et al, Proc. of the 1st Conf. on
Mechanics. Results of Scientific Investigations and Achievements in Cooperation of
The Academy of Sciences of Socialist Countries, Vol.1, Prague (1987), pp.27-30.
576
References
Index
Abel integral transformation 5 diffraction grating 170

Able integral equation 168 diode spectroscopy 127
absolute intensity of continuum 27 dispersion holography 87
absorption coefficient 4 dispersion interferogram 65
absorption factor 74 dispersion interferometer 63
actinometer 50 dissipation energy 486
aerodynamic drag coefficient 214 dissipation power 486
Angström system 56 Doppler broadening 23, 32
angular enthalpy probe 370 Doppler circuit 22
anomalous dispersion zone 161 Doppler contour 39
anomalous Doppler broadening 35 Doppler half width 22
anti-Stokes spectroscopy 127 Doppler’s effect 22
atomic–molecular transformations 313 drag coefficient 207
drift velocity of the electron 452
B Druyvesteyn equation 221
battery effect 427 E
Blanc’s law 261
Born’s approximation 38 effective scattering cross section 97
Brillouin component 148 Einstein coefficient 120
Brillouin components 152 electrogasdynamic gun 208
buffer gas 45 electron charge 2
electron concentration 246
C electron energy distribution function 220
electron mass 2
Calprobe 357
electronic plasma frequency 104
CALS 136
electronic saturation current 228
CALS spectrometer 140
electronically excited particles 34
CALS spectroscopy 136
emission coefficient 4
CALS spectrum 139
energetic threshold of the reaction 2
classic electron radius 97
enthalpy probe with a cooled screen 368
coefficient of ambipolar diffusion 279
enthalpy probe with an uncooled screen
coefficient of error transfer 18
368
coherence length 138
excitation threshold of the emitting state
collective scattering 103
51
D
F
Debye length 302
Fabry–Perot cavity 71
Debye radius 102
Fabry–Perot interferometer 35
Debye screening length 222
Faraday dark space 403, 413
Debye sphere 2
Faraday’s cylinder 489
differential enthalpy sensor 376
577
Plasma Diagnostics
fluorescence signal 119 local thermodynamic equilibrium 1

Frank–Condon factors 52 Lochsmidt number 151
Frank-Condon factor 52 longitudinal enthalpy probe 365
Fredholm equation 239 Lorentz approximation 253
Lorentz contour 161
G Lorentz halfwidth 29
g-discharge 413 Lorentz method 21
gas temperature 32 Loschmidt number 61
Grey-rad 357 LTE model 154
Grey’s probe 364 M
ground electronic state 52
group velocity 160 Mach number 270
growth curves 29 Mach–Zender interferometer 166
Mach-Zender interferometer 68
H mass of the emitter 33
harmonic oscillator 139 Maxwell distribution 36
helium–neon laser 73 Maxwell law 1
high-speed enthalpy probe 379 mean electron energy 245
holography 74 mean polarisability 139
homophase plasma flows 196 microwave vibrations 482
mobility of the electrons 449
I multicolour pyrometry 206
multicomponent diffusion 261
initial size distribution of the particles 197
inner and outer 364 N
ion charge 2
ion–ion recombination 503 nanosecond breakdown 450
ion–ion recombination coefficient 314 nitride synthesis 504
ionisation energy 2 non-linear dispersion interferometer 62
non-linear interferometer 60
K Nusselt criterion 362
Kerr cell 10 O
kinoholographic system 90
Kirchhoff 4 optical actinometry 53
Kirchhoff law 4 optical heterodyning 73
Kirchof law 27 optical thickness 3
Knudsen criterion 148 optically thin plasma 3
Knudsen layer 263, 300 optimum temperature 21
Kramers–Kronig relationship 161 Ornstein method 19
L P
Laframboise theory 248 parameter of the sink 239

laminar plasma torch 185 partial local thermodynamic equilibrium 1
laminar’ plasma torch 185 Paschen curve 412
Langmuir probe 221 Peclet number 284
Langmuir–Mottsmith probe 340, 342 phase velocity 160
laser diodes 127 Plachek–Teller coefficient 139
laser Doppler anemometry 200 planar enthalpy probe 377
laser heterodyne system 73 Planck’s radiation 16
laser interferometry 60, 70 plasma flow velocity 207
law of acting masses 2 plasma refraction 63
578
References
Index
plasma temperature 1 sheath–convection regime 282

polarisability of the atoms 114 Sherwood number 284
Poole–Frenkel mechanism 344 sink parameter 222
population of the energy levels 1 sliding effect 205
primary intensity standard 15 spatial resolution power 10
probability of spontaneous emission 57 Stark broadening 23, 33
probe characteristic 223 Stark half width 23
pulse conservation law 37 Stark modulation 133
Stefan–Maxwell equations 261
Q streamer breakdown mechanism 450
Q-branch 139 T
Q-factor 67
q-pinch plasma 77 theory of wave breakdown 497
quantum number 2 thickness of the irradiated layer 61
quantum yield of the photocathode 108 Thomson scattering 100
quasineutrality of plasma 2 Thomson scattering cross section 96
quenching rate of the excited state 57 Tikhonov’s method 541
time-of-flight anemometer 204
R time-of-flight method 203
radiation intensity 4 Tokamak 111
radiation lifetime of the excited state 44 Townsend coefficient 452
Raman light scattering 127 Townsend criterion 400
Raman scattering 136 translational energy 46
rate constants of excitation 51 translational movement of neutral particles
Rayleigh region 198 32
Rayleigh scattering 109, 147 transmittance coefficient 5
recombination coefficient 290 transverse enthalpy probe 371
reduced molecular mass 139 Trinor model 142
refractive index of plasma 160 tunable lasers 127
regularisation methods 34 two-jet dc plasma torch 181
relaxation model 45 two-tube enthalpy probe 365
resonance interferometry 65 U
rigid rotator 139
Rompe–Weitzel approximation 499 ultrasound frequency 74
rotation energy of the molecule 46
Rozhdestvenskii ‘hook’ method 161 V
S vector of the electron velocity 98

velocity modulus 33
scattering factor 147 vibrational temperature 131
schlieren circuit 76 Volterra equation 5
Schmidt ambipolar number 284 volume ionisation rate 290
Schmidt number 266, 285
Schottky mechanism 344 W
secondary intensity standards 15 Wollaston prism 171
sensitivity of holographic interferometry 82
sensitivity of the sensor 381 Z
shadow method 173
shearing interferometer 170 Zeeman modulation 133
579

B 2000 (Ovsyannikov) Plasma Diagnostics

Uploaded by

Copyright:

Available Formats

B 2000 (Ovsyannikov) Plasma Diagnostics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

B 2000 (Ovsyannikov) Plasma Diagnostics

Uploaded by

Copyright:

Available Formats

Plasma diagnostics

Ovsyannikov and Zhukov

Cambridge International Science Publishing

CAMBRIDGE INTERNATIONAL SCIENCE PUBLISHING

First published March 2000

© Cambridge International Science Publishing

British Library Cataloguing in Publication Data

Translated by Victor Riecansky

Chapter 5 ................................................................................................... 127

Chapter 7 ................................................................................................... 160

Chapter 8 ................................................................................................... 196

Chapter 9 ................................................................................................... 207

Chapter 11 ................................................................................................. 258

Chapter 12 ................................................................................................. 299

Chapter 13 ................................................................................................. 338

Chapter 14 ................................................................................................. 356

Chapter 15 ................................................................................................. 386

Chapter 17 ................................................................................................. 449

Chapter 18 ................................................................................................. 506

Chapter 19 ................................................................................................. 522

Chapter 20 ................................................................................................. 538

Chapter 21 ................................................................................................. 545

SPECTRAL DIAGNOSTICS METHODS

where N n is the concentration of particles of the energy level n with

r D is the radius of the Debye sphere; e is the electron charge; z is the

where N a, N b, N ab is the concentration of particles of components a,

where m e is the electron mass.

where Ni,z x is the concentration of ions of component x with the charge

purpose, the examined object is irradiated with an auxiliary source or

In order to ignore absorption, the following condition must be fulfilled

1.1 Integral relationships. Optical measurement circuits

– the transmittance coefficient – the fraction of radiant energy passing

We set e (r ³ R) = 0 and, consequently, I(|x| ³ R) = 0. In the coordi-

The order of integration on the right hand side will be changed

F ig .1.1 Cross-section of an axisymmetric object.

to a tabulated integral and is equal to p. We shall write the result for

Per parts integration of (1.12a) and subsequent differentiation with re-

This is a linear differential equation of the first order whose solution

Transferring to integration in the ranges [–y 0, y 0] (Fig.1.1) and setting

Here I (x) is the intensity of inherent radiation of the source; w (x) is

can be written in the form of Abel's integral equation

which takes into account the equality

which represents the integral Volterra equation of the first kind.

Measurement of radiation and transmittance. Figure 1.2 shows some

F ig .1.2 Optical circuit of measurement of radiation and transmittance. A) auxilliary

then transferred to the intensity distribution I (x) on the basis of the

where I (x) is the intensity of radiation, I (x) is the natural radiation

F ig .1.3 Diagram of recording radiation with a finite appature of the device.

Replacing I (x) over the averaging region by a segment of the straight

The difference between the measured intensity I ( x) and true intensity

F ig .1.4 Coordinate system for examining an oscillating source.

instantaneous lateral intensity distributions. This requires using

The total probability α(I, x') dI of passage through the section of

At fixed values of I we exclude I'(x i) from the system (1.25) and

Dividing the first equation (1.25a) by terms by the second equation, we

j( x1¢ - x ) + j( x1¢ + x ) a ( I , x1¢ )

We introduce the notations

device in such a manner that the axis of the source is perpendicular

dq ( x ¢) = I ( x )j( x ¢ - x )dx. (1.26)

is 5 1 0%. The modern methods make it possible to measure the mini-