IETE Journal of Research

ISSN: 0377-2063 (Print) 0974-780X (Online) Journal homepage: http://www.tandfonline.com/loi/tijr20

Frequency Dependence of Junction Capacitance

Balepur S. Murthy & P. R. S. Rao

To cite this article: Balepur S. Murthy & P. R. S. Rao (1971) Frequency Dependence of Junction
Capacitance, IETE Journal of Research, 17:7, 273-276, DOI: 10.1080/03772063.1971.11486806

To link to this article: http://dx.doi.org/10.1080/03772063.1971.11486806

Published online: 21 Aug 2015.

Submit your article to this journal

Article views: 2

View related articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tijr20

Download by: [University of California, San Diego] Date: 29 June 2016, At: 12:41
 Frequency Dependence of Junction· Capacitance
BALEPUR S. MURTHY (Assoc. member) & P. R. S. RAO
Solid State Physics Laboratory, Luc.know Road, Delhi 7

Manuscript received on 30 July 1970

Diode junction capacitance is usually assumed to be a frequency independent

parameter, varying only with applied reverse bias. This frequency independence
is true in the case of ideal junction diodes up to very high frequencies above which
dispersion at the space-charge edge becomes appreciable. Traps or deep centres
in a p-n junction will cause effects similar to the free carrier dispersion. The
junction capacitance of diodes fabricated with material containing traps or deep
impurity levels will show a strong frequency dependence at low frequencies. The
paper (tives the junction capacitance values observed on certain diodes as a function
of frequency. The results have been analysed to obtain the total junction capaci-
Downloaded by [University of California, San Diego] at 12:41 29 June 2016

tance C, as a series combination of a frequency independent capacitance C0 (which

is voltage dependent), and a frequency dependent capacitance Cr It is proposed
in the paper that such a study of the capacitance frequency behaviour of a junction
diode is a practical way of determining the suitability of a given fabrication pro-
cedure (I.e. starting material, method of handlin~ and creation of junction) for the
given objective.

N the course of our work on the fabrication Device Fabrication

I of tuning varactors at the Solid State Physics

Laboratory, some of the diodes exhibited a
frequency dependence of small signal junction
Epitaxial slices (n on n+) with 1 ohm em phos-
phorus layers on 0·01 ohm em antimony doped
capacitance at room temperature, when biased substrates from M/s Wacker-Chemic have been
in the reverse direction. This behaviour has been used in the fabrication of the diodes. No further
studied in some detail and the results are reported information was available from the manufacturers
in this paper. as to the qualitative and quantitative impurity.
Frequency dependence of the reverse-biased content, as these slices were from their develop-
capacitance on gold doped silicon P+N step junc- mental lots. Diodes of both plain and planar
tions have been studied in detail by Sah and Reddil varieties were fabricated by diffusion techniques
who have also given a simple physical model which using industrial gases like nitrogen (from M/s Modi
explains the frequency dependence. Effects of Oxygen) and oxygen (from M/s Indian Oxygen)
deep impurities (like indium in silicon at 77°K) straight from the cyliders. The dopant source
on n+P junction capacitance has been investi- used was boron tribromide of 99·99 per cent
gated by Schibli and Milnes2 who have interpreted purity from M/s Wacker Chemic. The reagents
their results on the basis of a series combination used in photolithography, nickel plating and clean-
of two capacitors, one dependent on d.c. bias and ing operations were AR grade chemicals. Distilled,
another dependent on frequency. deionized water has been used in cleaning the
Since no intentional doping of deep impurities slices before critical operations like diffusion, etc.
was done on the starting material in the present
investigation, it is felt that such impurities were Capacitance Measurements
either present in the imported epitaxial slices
andfor were introduced during the high tempera- The low frequency capacitance of the diodes
ture fabrication processes. The results of the have been assessed on a Boonton capacitance
measurements have indicated that it is possible bridge at 100 kHz. The capacitance variation
to characterize the entire fabrication procedure with d.c. bias was obtained by applying increasing
on the basis of a characteristic frequency. This reverse bias up to the breakdown voltage. The
method of characterization is considered to be capacitance at 4 volts of these diodes was then
of use to semiconductor device industry as it measured on a Wayne Kerr admittance bridge over
provides for a practical figure of merit of the the frequency range S-100 MHz to study the
fabrication procedure. variation with frequency. From the measurements,

273
 J. INSTN TELECOM. ENGRS, Vol. 17, No. 7, 1971

TABLE 1 - VARIATION OF CAPACITANCE IN PF WITH FREQUENCY, BIAS 4 V

(Room temperature v~ues)

Behaviour Diode Frequency of measurement in MHz

type No. ......
0·1 5 10 30 50 80 100
A P205 18·0 18·0 14·0 1f"5 15·0 15·0 16·0
P219 18·6 19·0 15·0 15·5 16·0 16·5 17·5
R401 17·0 16·5 15·5 13-0 1Q-5 6·0 5·0
B R406 19·4 19·0 19·0 17·0 15·0 12·0 11·5
R412 18·8 18·0 18·0 16·5 H·O 12·0 11·5
R424 19·4 19·0 17·5 17·2 14·5 13·0 13·0
1804 20·0 19·5 16·5 11·0 5·0 4·0
c 1409 12·0 11·2 7·0 1·7 1·5 1·5 1·5
1415 12·8 12·5 12·0 5·0 3·0 1·5 1·5
e 1418 12·6 13·0 12·0 7·8 5·0 3·5 3·5
1419 12·2 12·5 11·0 6·0 3·5 3·0 3·0
1406F 24·2 18·0 8·5 5·2 3·5 3·5
Downloaded by [University of California, San Diego] at 12:41 29 June 2016

we have observed three types of behaviour of where R 1 +R 2 = I? includes the losses due to the
capacitance variation in the frequency. In the neutral regions, contacts and losses associated
first type of behaviour, the capacitance falls down with cf"
to a lower value and is maintained constant, the (A) Constant bias case - When we fix the reverse
actual variation of capacitance with frequency bias across the diode and vary the frequency, then
being small. In the second type of behaviour, effectively we are maintaining C0 constant and
the capacitance shows a gradual reduction as the varying C1. Thus, we can obtain a frequency
frequency is increased over the range 5-100 MHz. dependence as shown in Fig. 2(a). One can define
In the third type of behaviour, the capacitance ./,. as the frequency at which the low frequency
falls down to very low values as the frequency is value intersects (straight line approximation) the
increased and this low value of capacitance is high frequency value and / 0 as the frequency at
maintained constant for further increases in fre- which the capacitance has become half the low
quency of measurement. These three types of frequency value, i.e. C1 = C0 /2. The frequency
behaviour termed Types A, B and C are indicated ./,. can be termed as break-over frequency and / 0
in Table 1. as the break-even frequency for purpose of charac-
terization.
Electrical Model of the Diode When the deep impurity concentration is very
high, then it is possible for the depletion layer
The actual capacitance C1 of the junctions at width to become equal to the total width from
any frequency f can be thought as the series com- junction to contact at high enough frequencies ..
bination of a voltage dependent capacitance C0 In the case of diodes fabricated using epitaxial
and a frequency dependent capacitance C1. The material, the width from junction to contact would
-electrical model of the diode is shown in Fig. 1, be equal to left over epitaxial layer thickness,

Co(V) Cf(f) Ct

--
R

Fig. 1 -The electrical model of the diode

274
 TABLE 2 - CHARACTERISTIC FREQUENCIES
C:o= Cf I
I
r- ltE610N 2 _,
I' Run Diode /b fo fc
Ce >> Cf No. No.
Co I
AE610N 3
R4 R401 30 52·5
Co R406 30
T R412 30
BIAS 18 1804 30 so
1810 30 52
1810C 30 60
14 1413 18 40 80
1415 15 25 80
1418 15 4o 75
fit '·
FRUIUINC'Y - - -
Fig. 2(a) as to dominate the capacitance behaviour. This
run was a plain diffusion run designed to obtain
a deep junction 6·0 !L· The time of diffusion for

r run 14 was larger than for the other runs like P2,
Downloaded by [University of California, San Diego] at 12:41 29 June 2016

1 FIXI.D BIAS R4 and 18 where the junction depth was designed

co for 1 !L· The deterioration in the diodes from run
fiii610NJ number 14 may be a direct consequence of this
I Co"~ Cf time temperature cycle. Run number 18 with a
Cc - - - - - - -1- - - - _a.; shallow junction of 1 !L is far superior to run num-
I I
ber 14, both being plain diffusions. The planar
diffusion run R4 is better than .the plain diffusion
run 18 of comparative junction depth. Planar run
f~t fc: P2 is the best from fabrication standpoint amongst
~R£GtU£NCY-
all the three.
Fig. 2(b) (B) Varying bias - In case we maintain the
frequency constant and vary the bias applied, we
should get a situation where C1 is fixed and C0 is
since very high conductivity material is used for varied. If the model we have assumed for C1 is
the substrat;e. In such a case, the high frequency correct then we should be in a position to predict
capacitance will be constant beyond a particular the variation of C1 with bias at any given high
frequency. This situation is shown in Fig. 2(b). frequency, if the bias dependence low frequency
The value of this constant capacitance C, would capacitance C0 is known for the diode at various
be given by C,= (EA/d). Where A is the junction voltages.
area and d is the distance from junction to contact. For example, if we consider a particular frequency
Under .such conditions one can define a frequency j 1 at which we want to know the capacitance as
j, as shown in Fig. 2(b) and this frequency can the bias is varied, we should know the value of C1
be termed as the cut-off frequency for purpose at any bias at j 1 and value of C0 at various voltages.
of characterization. Then one can compute the frequency dependent
We have considered above the behaviour of capacitance component C1, using the expression
the diodes with a fixed bias. Results of the ana-
lysis of the measurements conducted on the diodes
' C C, 0 .
Cf, = C -C at the fixed btas
are given in Table 2, with a fixed bias of 4 volts 0 I

in the reverse direction. and then employing the expression

From Table 2 it can be easily seen that run
number R4, 18 and 14 can be characterized by C _ C0 CJ,
,-Co+CJ,
their break-over frequencies as .30, 30 and 15 MHz
respectively. The break-even frequencies in the and using appropriate C0 values one can calculate
case of these runs are beyond 100, at 50 and around the bias dependence of C1 at that frequency f.
30 MHz respectively. The cut-off frequency has This computation was done on several diodes
bt:en observed mainly in the case of run number and the variation with bias at 80 MHz was cal-
14 and is around 70 MHz. culated according to the above equations. Next,
We can conclude that the run number 14 was capacitance of the diodes was measured at the
the worst from the fabrication standpoint, since same bias points at 80 MHz. Results are tabu-
the deep level impurity density has been so high lated in Table 3.

275
 J. INSTN TELECOM. ENGRS, Vol. 17, No. 7, 1971.

TABL,E 3

Diode Measured values at 4 V Calculated Capacifance in PF at 80 MHz

No. Ct(80 MHz)
Ct(80 MHz) C0 (100 kHz) 6V 8V 10 v 14 v
P205 15·0 18·0 90·0 Calc. 12·7 11·35 10·5 9·0
Measured 13·0 10·2 11·5 10·5
P219 16·5 18·7 140·0 Calc. 14·3 12·5 11·6 10·2
Measured 14·5 13-0 12·0 10·5
1406F 3·5 24·2 4·1 Calc. 3·44 3·38 3·3 3·26
Measured 3·0 3·0 3·0 3·0
1402 3·5 12·6 4·85 Calc. 3·36 3·3 3·24 3·18
Measured 3·0 3·0 3·0 3·0
1409 1·5 11·2 1·73 Calc. 1·5 1·47 1·47 1-45
Measured 1·5 1-5 1·5 1·5
1415 1·5 12·8 1·7 Calc. 1·53 1-475 1·455
Measured 1·5 1·5 1·5 1·50
R424 13-0 19·0 41·0 Calc. 11·7 10·8 9·88 8·9
Measured 12·0 11·0 10·0 9·2
Downloaded by [University of California, San Diego] at 12:41 29 June 2016

From Table 3 one can see that there is a remark- dependent and a voltage dependent component; and
able correlation between the calculated and (3) if sufficient care is taken in the fabrication of
measured values of capacitance when C1 ~ C0 and diodes, it is possible to avoid the frequency depen-
the correlation is within experimental errors of dent component altogether.
measurement when C1 ~ C0 . Hence, the model
assumed for the analysis that the total capacitance Acknowled~ement
is made up of a series combination of a frequency
dependent and a voltage dependent component The authors are thankful to the Director, Solid
has been proved to be valid. State Physics Laboratory, for giving permission
to present this paper at the Thirteenth Technical.
Convention of ITE. The authors wish to express
Conclusion their thanks to Shri K. R. Wadhera for fabricating
From the measurements conducted on different these diodes and to Shri M. G. Rao for keen in-
batches of plain and planar diffused diodes, we terest he evinced in the present study.
can conclude that (1) it is possible to characterize
fabrication processes by characteristic frequencies References
based on capacitance versus frequency plots; (2) the 1. C. T. & REDDI, V. G. K., IEEE Trans. on ED, 11 (7)
SAH,
total capacitance of the diode can be considered (1964). 345.
2. SCHIBLI, E. & MILNF.S, A. G., Solid slate e[,:c•l'onics, 11 (3}
as due to a series combination of a frequency (1968), 323-334.

276