20
Brazilian Journal of Physi s, vol. 32, no. 1, Mar h, 2002
Industrial Appli ations of Plasma
Fo us Radiation
C. Moreno, M. Venere, R. Barbuzza, M. Del Fresno, R. Ramos,
H. Bruzzone, Florido P. J. Gonz
alez, and A. Clausse
Interinstitutional Program of Dense Magnetized Plasmas,
CNEA-CONICET-CIC-INFIP-UNMP-UNICEN, Argentina
Re eived on 3 July, 2001
Appli ations of a small- hamber Plasma Fo us used as portable radiation generator is presented.
The devi e was designed to maximize the uen e. The mean neutron yield was 3 x 108 neutrons of
2.45 MeV per shot, orresponding to a 106 neutrons/ m2 uen e on the external surfa e of the hamber. A te hnique to dete t the presen e of water in the neighborhood of a ompa t Plasma Fo us
is presented. The measuring system is omposed by two neutron dete tors operated simultaneously
on every shot. The rst dete tor is used to register the PF neutron yield in ea h shot; whereas the
other one was designed for dete ting neutrons s attered by the blanket. The results indi ate that
the system is able to dete t water ontents of few per ents in volume. The orrelation of the ounts
re orded by the dete tors lo ated at di erent positions was mapped with the water distribution
around the neutron sour e. The omplete dete ting system is very simple and inexpensive. Among
many other potential appli ations, the te hnique is spe ially suited for soil humidity prospe tion.
X rays radiation emitted by the ompa t Plasma Fo us operated in Deuterium has been used for
introspe tive radiographi imaging of metalli obje ts. The samples were lo ated about 1 m away
from the PF hamber wall. High-sensitivity, fast-response ommer ial radiographi lm was used
as x-ray dete tor. A set of experimental images is presented demonstrating a very high penetration
power of the x-ray beam. Among many other appli ations, the presented te hnique is spe ially
suited for introspe tive visualization of pie es manufa tured on metal. Radiographi proje tions of
a stainless steel BNC elbow taken at 8 di erent angles were pro essed to re onstru t transversal
uts of the pie e. A omputer te hnique for 3D re onstru tions was ombined with radiographi
images of obje ts X-rayed with a ompa t plasma fo us. The te hnique is able to automati ally
determine the position of the rotation axis, re onstru t the 3D-attenuation map, and display inner
uts. The system was demonstrated in introspe tive tomographi imaging of a stainless steel BNC
elbow.
I
Introdu tion
Plasma Fo us (PF) devi es ourished in the 70's
and 80's as nu lear fusion devi es based in the pin h
phenomenon o urring during the path of high ele tri
urrents through the working gas. The operation of
PF has been extensively studied by resear h laboratories around the world, where several PF on gurations
has been developed over the years aiming to in rease
the neutron emission [1-4℄. Currently, PF pulsors are
among the heapest available neutron generators, with
unique features of extremely short pulses (hundreds of
ns) that suit them for a number of interesting appliations. There are also interesting possibilities to take
advantage of x-rays (1-100 keV), ele tron and ion beams
emitted during PF shots.
The plasma-fo us phenomenon o urs at the open
end of oaxial ele trodes when an intense ele tri al disharge between them is indu ed by external means. The
oaxial ele trodes are lo ated inside a va uum hamber
lled with deuterium gas at low pressure. A harged
apa itor bank is onne ted to the losed end of the
ele trodes through a swit h. After losing the swit h, a
gas dis harge starts in the gap between the ele trodes
forming an umbrella-like plasma layer. The azimuthal
magneti eld lo ated in the toroidal volume en losed
by the urrent, produ es a Lorentz for e that pushes
the sheath toward the open end of the ele trodes. The
run-down of the urrent sheath is a sweeping supersoni
sho k that propagates olle ting the gas parti les ahead
of the front. On its arrival at the open end (some s
after triggering), the magneti eld starts to ontra t,
21
C. Moreno et al.
a elerating the plasma toward the axis. Finally, the
sheath lashes on the axis in the form of a small dense
plasma ylinder (fo us). The lifetime of the fo us is
about 300 ns.
The emitted neutrons an be applied to perform radiographs [5℄ and substan e analysis, taking advantage
of the penetration and a tivation properties of neutral
radiation [6℄. Likely, the intense x-ray pulses produ ed
by fo alized ele tron bremstralung are ex ellent andidates for radiography of moving and soft obje ts and
for mi roele troni lithography [7℄.
Had small portable PF devi es been available, the
added value of the emissions would substantially inrease, for larger uen es an be provided in wider domains of appli ations. However, due to the strong intera tion between the hot plasma and the va uum hamber, the ele trode housing is usually big, leaving room
for the plasma blast. The main trouble with having
the ele trodes too lose to the ontainer wall is the gas
ontamination with impurities, whi h onspires against
performan e and regularity of the emissions [8℄.
Along this arti le, appli ations of the ompa t PF
devi e, pulsing at one shot per minute is presented.
The asso iated x-rays emissions were applied to obtain
three-dimensional introspe tive images of small metalli omponents, and the emitted neutrons were used to
dete t water by neutron inelasti s attering.
II
The plasma fo us GN1
The Plasma-Fo us GN1 is a ompa t version of a
Mather-type ma hine. Fig. 1 shows a diagram of the
va uum hamber and the ele trodes. The anode is an
ele trolyti - opper ylinder, 38 mm diameter, 1.5 mm
thi k, 87 mm long. The athode is formed by 12
bronze bars, 3 mm diameter, 100 mm long, ylindri ally
pla ed, and welded at the end to a bronze ring 72 mm
diameter. The ontainer is a stainless-steel ylinder
157 mm long, 96 mm diameter with a lateral NW25
va uum port for pumping and gas loading. The insulator is a Pyrex glass ylinder 35 mm long and 4 mm
thi k.
Using a me hani al pump and an oil di user, base
pressures down to 10 8 mbar an be rea hed. The
external ir uit is a apa itor bank, divided in three
dis harging modules; ea h of them omposed by ve
Maxwell type 31161 ondensers. The total apa itan e
is 10.5 F and the harging voltage is 30 kV. The three
modules are red simultaneously and peak urrents of
350 kA are attained in a quarter of period (1.1 s).
The system operates between 1 to 8 mbar of Deuterium.
CHAMBER
CATHODE
ANODE
INSULATOR
Figure 1. S hemati s of GN1 hamber (sizes are detailed in
the text).
After ea h shot, the lling pressure in reases about
0.05 mbar due to the release of impurities from the
hamber, ele trodes and insulator walls. Consequently,
the hamber is pumped down (me hani ally) after ea h
shot in order to assure onstant pressure onditions.
The maximum shot frequen y was one shot per minute,
limited by the harger. Under these onditions, the
frontal wall temperature (top in Fig. 1) in reases about
20C over the ambient temperature after 30 shots,
ooled passively by air natural onve tion and heat ondu tion through the metalli stru ture. The working
gas is renewed after 10 shots.
Upon ommissioning, the equipment was tested in a
series of 1000 shots at di erent load pressures. The time
derivative of the urrent owing to the anode, dI/dt,
and the voltage a ross the ele trodes, V , were monitored for ea h shot by a Rogowski oil and a resistive
voltage divider, and were registered using a 500 MHz,
1 Gs/s digitizing os illos ope. Fig. 2 shows typi al signals for lling pressures of 2; 4 and 6 mbar. Very intense voltage spikes (120 kV) are obtained at the time
where the maximum ompression takes pla e, thus indiating good fo using. Su h peaks impose severe design
onditions on the insulator, whi h should be onsiderably thi k in order to stand the stress (2 mm of Pyrex
glass is destroyed with few dis harges). It was observed
in Fig. 2 that the fo us o urs later for higher pressures.
Fig. 3 shows the average dependen e of the fo us timing with the deuterium pressure. The dimensions of
the ele trodes were determined using a omputer aided
22
Brazilian Journal of Physi s, vol. 32, no. 1, Mar h, 2002
design system whi h is based in a thermonu lear PF
model [9℄. The ontinuous tra e is the fo ussing time
derived from an analyti al snowplow des ription oupled with an isoentropi plasma ompression model.
1.4
tf (µ sec)
3
2 mbar
2
1.6
1.2
1.0
1
0
0.8
-1
2
4
6
8
p o (m b)
3
Figure 3. Variation of the fo using timing with lling pressure. The line orresponds to predi tions of a snowplow
model [9℄.
4 mbar
2
1
0
3
-1
3
6 mbar
2
Yn / 10
8
2
1
0
1
-1
0.0
0.5
1.0
t (µs)
1.5
2.0
Figure 2. Signals proportional to the voltage a ross ele trodes (solid line) and urrent time derivative (dotted line).
The signals units are Volts measured in a Rogowski oil and
a voltage divider respe tively.
The time integrated neutron yield was also measured for ea h shot by silver a tivation response as a
fun tion of pressure. The neutron dete tor was pla ed
at 60 m from the fo us in a line perpendi ular to the
axis. The sensitive area was 900 m2 . The neutron
measuring system was alibrated by omparison against
TLD dete tors. Fig. 4 shows the dependen e of the average neutron yield per pulse with the lling pressure.
The ontinuous tra e on Fig. 4 orresponds to same
model of Fig. 3. The optimum average yield, 3 x 108
neutrons per shot, o urs at 4 mbar. This yield orresponds to 106 n/ m2 per shot in the external frontal
wall.
Di erent insulator lengths were tested in preliminary series of measurements aimed to investigate ea h
design performan e. It was found that variations of
5 mm in that length lead to noti eable degradation of
the neutron yield at all the explored pressures.
0
0
2
4
6
8
Po (mb)
Figure 4. Neutrons yield at 90Æ o axis as fun tion of lling
pressure. The line orresponds to predi tions of a snowplow
model [9℄.
III Defe tos opy
PF an be advantageously used as high-intensity
short-duration x-ray sour es. Due to the very fast
plasma ompression attained in these devi es, parti ularly intense soft and hard x-rays pulses are emitted.
High repetition rate small PF are urrently used for
SXR lithography [10℄. High-Z working gases are used
in these appli ations to enhan e the x-ray yield at redu ed wavelengths [5℄. We de ided to use the ompa t
PF operated in Deuterium for making non- onventional
radiographs. Our aim is to nd new appli ation elds
for the PF as x-ray sour e, and to allow for the possibility of obtaining, in a future experiment, a simultaneous
radiographi and neutrographi image of the same obje t.
23
C. Moreno et al.
The samples to be imaged were pla ed outside
the stainless steel hamber, on the ele trodes symmetry axis, and 83.5 m away from the fo ussing region
(Fig. 5). Commer ial radiographi lm, Curix ST-G2
from AGFA was used together with AGFA suggested
developer and xer for this lm. No spe ial pro edures
were needed other than those re ommended by the supplier, to manipulate and develop the lms.
Object to be tomographied
X-Ray
Film
Device for
angular variation
Figure 5. Setup of the tomographi
GN 1
Neutron
Generator
te hnique.
A stainless steel BNC elbow was used as a sample.
All the shots were made at lling pressures of 4-5 mbar
pure deuterium. The sample was mounted on a small
a ryli platform that rotates to allow for taking images
at di erent viewing angles. The rotating axis was set
verti ally and 8 viewing angles were used: 0, 30, 60, 75,
90, 105, 120, and 150 degrees. The axis was marked
with a sharp metalli needle.
Fig. 7 shows the set of radiographs of the pie e at
di erent angles. It an be seen that, even ontrolling
arefully the operation, there are always di eren es in
brightness and fo alization. Therefore, in order to be
useful in a tomographi system, we should be able to
re onstru t inner uts pro essing imperfe t proje tions.
Figure 6. Angular radiographs of a BNC.
Figure 7. Tomographi
re onstru tions using the
radiographs shown in Fig. 6.
The Monte arlo method is a exible te hnique than
an to handle quality di eren es of the information input. Basi ally, Monte arlo tomographi re onstru tion
is a sto hasti sear hing pro ess, where the omputer
boun es randomly in the set of all possible digital 3D
images ( alled instan es ), guided with a sele tion riterion that ensures the onvergen e toward the a tual
inner attenuation eld. The re onstru tion algorithm
is based on the omparison of the a tual radiographs
with the proje tions that would produ e the instan e
image. The general pro edure is as follows:
Starting with an initial guess instan e, modify
slightly the tone of a pixel hosen at random.
Cal ulate the proje tions of the instan e in every
dire tion.
Cal ulate an error indi ator averaging the square
deviations of the instan e proje tions respe t to the
a tual proje tions.
If the error of the new instan e is lower than the
previous one, the modi ation is a epted.
24
Brazilian Journal of Physi s, vol. 32, no. 1, Mar h, 2002
The method was implemented using obje t oriented
programming, in visual C++ . A visualization system
ompletes the tool, allowing the fast inspe tion of inner
uts of the attenuation eld. The hardware requirement is just a Pentium personal omputer, and the system runs in MS Windows environment. Fig. 7 shows
the display of di erent uts of the BNC on the ontrol
panel of the appli ation. The visualization of the inner
uts shows details down to 0.3-mm resolution.
in position inside the an. This dete tor was lo ated
side-on, approximately at 30 m from the PF hamber.
WATER
50 cm
DETECTOR 1
75 cm
7 cm
50 cm
20 cm
PLASMA
FOCUS
DETECTOR-2
Neutron e hography
Neutrons an be used for dete ting Hydrogen in
metals or for dete ting substan es by a tivation analysis
[11℄. Either huge ssion rea tors or powerful deuteron
a elerators are used in these appli ations as neutron
sour es, whi h share the disadvantages of being nonportable and very expensive. On the ontrary, substan e interrogation system should be portable, reliable, e onomi ally ompetitive, and minimize the environmental impa t. PF devi es lled with deuterium
gas, omplemented with a ouple of neutroni dete tors
are a valid alternative, as they ful ll all the requirements mentioned before.
We ondu ted an experiment to dete t water or
other substan es ontaining low Z elements, by neutron
inelasti s attering. The method is based on the same
prin iple of the sonar or the e hographs, registering the
neutrons s attered by the interrogated substan e, omparing the dete tor response when the interrogated substan e is absent.
The measuring system is omposed by a neutron
pulsed generator and two silver a tivation dete tors operated simultaneously on every shot. The rst dete tor is used to register the PF neutron yield; whereas
the other one is used to dete t neutrons s attered by a
water blanket. The ompa t PF devi e lled with deuterium was used to provide neutron pulses. The basi
idea of the method is to use a suitable dete tor to register the neutrons s attered by the interrogated substan e
(if present) and to ontrast the dete tor response when
the interrogated substan e is absent. Shots were made
in presen e of the interrogated substan e, then the proedure was repeated after removing the substan e, and
the neutrons registered in both situations were ompared.
Fig. 8 shows the experimental setup. A standard silver a tivation ounter pla ed side-on, 60 m away from
the PF hamber was used as the referen e dete tor to
take the shot to shot variation of the neutron yield into
a ount. This dete tor was overed with paraÆn to
moderate the fast neutrons. The se ond dete tor is
omposed by three Geigers, ea h of them overed by a
0.3 mm thi k silver foil and pla ed inside a metalli an
(10 m in diameter and 13 m in length). Some pa kaging plasti was used to keep the Geigers assembly
Figure 8. Neutron e ho interrogation setup.
140 liters of water were arranged in a wall of 11 plasti
ontainers. The water wall was put in di erent positions respe t of the dete tor systems. For every on guration the PF was shot about fteen times re ording
the ounts measured in ea h dete tor.
Fig. 9 shows the results obtained along the PF axis.
Ea h point in the graphi orresponds to one shot, the
ounts of ea h dete tor being the oordinates of the
plot. The relative ounting of the moderated dete tor
in reases as the water gets loser to the dete tion system. This e e t an be attributed to the hange in the
solid angle o ered by the water. The ounts olle ted
by the s attered neutrons dete tor when no water is
present, orrespond to neutrons re e ted by the laboratory building. The slope of the least-squares lines
orresponding to ea h position an be used to hara terize the sensitivity of the system. Fig. 10 shows the
ontour map of the relative ounting slope. The oordinates orrespond to the lo ation of the water wall
respe t to the PF head. The x- and y- oordinates are
measured along lines perpendi ular and parallel to the
PF symmetry axis respe tively. It an be seen that
the relative ounting de reases faster when the water is
moved away along the symmetry axis.
1000
Scattered neutrons (a.u.)
IV
7 cm
800
20 cm
600
y = 50 cm
400
200
0
water absent
0
500
1000
1500
2000
2500
Reference neutrons (a.u.)
Figure 9. Neutroni re e tion in along the PF axis.
25
C. Moreno et al.
the position of the rotation axis, re onstru t the 3Dattenuation map, and display inner uts.
50
0.27
y (cm)
30
0.35
20
Referen es
0.23
40
0.31
0.39
0.43
10
0.47
0
20
40
60
80
x (cm)
Figure 10. Contour map of the relative neutron ounting.
Every (x,y) oordinate represent the position of the water.
V Con lusions
Two feasibility studies of industrial appli ations of
the X and neutron radiation produ ed by a ompa t
Plasma Fo us were presented.
A te hnique to dete t the presen e of water was developed using two neutron dete tors operated simultaneously on every shot. The results indi ate that the system is able to dete t water ontents of few per ents in
volume. The omplete dete ting system is very simple
and inexpensive. Among many other potential appli ations, the te hnique is spe ially suited for soil humidity
prospe tion.
On the other hand, X-rays from the PF has been
used for introspe tive radiographi imaging of metalli obje ts. The te hnique is spe ially suited for introspe tive visualization of pie es manufa tured on metal.
Radiographi proje tions of a stainless steel BNC elbow
taken at 8 di erent angles were pro essed to re onstru t
transversal uts of the pie e. A omputer te hnique
for 3D re onstru tions was ombined with radiographi
images of obje ts X-rayed with a ompa t plasma fous. The te hnique is able to automati ally determine
[1℄ A. Bernard, Nu lear Instruments Methods, 145, 191
(1977).
[2℄ H. Conrads, Pro . 3rd Latinameri an Worshop Plasma
Phys., Santiago, Chile, July 18-29 (1989).
[3℄ A. Serban and S. Lee. Experiments on speed-enhan ed
neutron yield from a plasma fo us. J. Plasma Physi s,
60, part 1, 3-15 (1998).
[4℄ G. De ker, R. Wiene ke, Physi a 82C, 155-164 (1976).
[5℄ M. Gibbons, W. Ri hards and K. Shields. Optimization of neutron tomography for rapid H on entration
inspe tion of metal astings, LLNL Rep UCRL-JC129723, (1998).
[6℄ E. Hussein and E. Waller. Review of one-side approa hes to radiographi imaging for dete tion of explosives and nar oti s, Radiation Measurements, 29 (6),
581, (1998).
[7℄ S. Lee, V. Kudryashov, P. Lee, G. Zhang, A. Serban, M.
Liu, X. Feng, S. Springham, T. Wong and C. Selvam.
SXR Lithography Using a High Performan e Plasma
Fo us Sour e, 1998 ICPP and 25 EPS Conf on Contr.
Fusion and Plasma Physi s, 22C, 2591 (1998).
[8℄ J. Gratton, M. Alabraba, A. Warmate and G. Giudi e,
Deterministi dynami s of Plasma Fo us dis harges.
Chaos, Solitons & Fra tals, 3, 343 (1993).
[9℄ C. Moreno, H. Bruzzone, J. Martnez and A. Clausse.
Con eptual engineering of plasma-based neutron pulsors. IEEE Transa tions on Plasma S ien e, 28, 17351741 (2000).
[10℄ S. Lee, P. Lee, G. Zhang, X. Feng, V. Gribkov, M. Liu,
A. Serban, and T. Wong. High Rep Rate High Performan e Plasma Fo us as a Powerful Radiation Sour e.
IEEE Trans on Plasma S i, 26, 1119, Aug (1998).
[11℄ F. Brooks, A. Buer, M. Allie, K. Bharuth-Ram, M.
N hodu and B. Simpson, Determination of HCNO onentrations by fast neutron s attering analysis, Nu lear
Instruments and Methods Se t. A (410) 2, 319-328
(1998).