IEEE TRANSACTIONS ON COMPONENTS, HYBRIDS, AND MANUFACTURING TECHNOLOGY, VOL. CHMT-9, NO.

3, SEPTEMBER i986 293

Shelf-Life Evaluation of Aluminuin Electrdytic

Capacitcirs
w. D. GREASON, SENIOR MEMBER, IEEE, AND JOHN CRITCHLEY, MEMBER, IEEE

Abstract-Because of a lack of information on the electrical character- The industry type for the capacitors used is generally
istics of alumi&ni eJectrolytic capacitors whenoperated in a nonco&inu- referred to as a low-leakage miniature alutinum electrolytic.
ous manner in low-voltage circuit applications, an accelerated shelf-life
All capacitors tested were of the radial lead configuration
aging test was conducted for 3000 h at 85°C on three types of capacitors
from four vendors. The effects of accelirated conditioning on capaci- ranging in size from 5-mm diameter x 1l-mm length to 6.5-
tance value, effective seriesresistance(ESR), and leakage current ai 1 V, 2 mm diameter x 12.0~mm length. The end consttiction of the
V, and rated voltage are reported. General observations show a decrease units consisted of a nibber bung with the can crimped to
in capacitance and an increase in,ESR and ieakagk current. The results provide an effective seal around the leads and the bung-can
also show that leakage current and ESR cannot be optimized simultane- interface.
ously. Guidelines for the use df aluminum electrolytic capacitors in low-
voltage circuits are also presented. Initial capacitance values were measured using a Hewlett-
Packard 4261A Digital LCR Meter Which applied a 1.5 V dc
I. INTRODUCTION bias to the specimens under test. The same instrument and test
. signals were used to measure the initial dissipation factor (DF)
R ECENT
manufacturing
IMPROVEMENTS in the design and
processes of modern aluminum
electrolytic. capacitors, plus their cost effectiveness when
values at 120 and 1000 Hz, from which the ESR was
calculated using the relationship:
compared to tantalum capacitors is resulting in an increased DF
evaluation of these components for applications in the design
ESR=-.
2ufC
(1)
of electronic circuits. The use of these capacitors in low-
voltage circuitry (l-2 V dc) which are powered in a DC leakage current tias measured by connecting the
noncontinuous manner (such as electrotiic telephones) requires capacitor in series with a 1-MQ resistor to a dc power supply,
special attention since performance characteristics for ex- applying 1 V, 2V, and rated voltage across the capacitor, and
tended operation under these coriditions are generally not measuring the voltage drop across the resistor.. The voltage
available. The circuit designer requires knowledge of the was applied for 2 min at each step, and the three tests were
effbcts of these operating condi&ons on capacitance value, d&e sequentially.
leakage, current, and effective series resistance (ESR). Mea- The 2-min test interval ivas the same as that specified by the
surement of these characteristics before and after an acceler- manufacturers of the capacitors tested atid was chosen to allow
ated shelf-life aging test was considered to be an effective comparison of the results with the ieakage current levels, at
method of predicting their actual operating performance. rated voltage, specified by the vendors. The time interval is
The purpose of: the work reported here tias to determine the sufficiently long to eliminate the effect of the charging
suitability of aluminuni electrolytic capacitors for use in low- currents, stabilize the true leakage current and obtain repro-
voltage electronic circuit applications by using an accelerated ducible results.
shelf-life aging test as a simulation method. A total of 180 The capacitors were then conditioned in an oven at 85 “C foi
capacitors from four different tianufacturers (vendors A, B, 3000 h with no voltage applied. This jtepresents an accelera-
C, and D) were tested, and the results are presented. The tion factor of approximately 22 years at 25 “C or 11 years at
significance of the results will be discussed along with ?5’C shelf storage temperature. The relationship between
application guidelines for the use of these components in low- temperature and life is described in more detail in Section VI
voltage circuits. of this paper. Capacitance, dissipation factor, and leakage
were remeasured, using the same methods, upon completion
II. TEST PROCEDURE
of the tempera&ire conditioning.
Capacitor values tested were 1 PF at 50 V, 10 PF at 35 V,
111. RESULTS
and 22 PF at 10 V. The number of specimens t&ted by vendor
and type is given in Table I. General
Parameters were measured initially and after 3000 h.
Manuscript received Juhe 12, 1985; revised May 1.5, 1986. This paper was Intermediate measurements were not taken in order to simulate
presented iit the Electronic Components Conference, Seattle, WA, May 5-7,
1986. true shelf-life conditions and eliminate any parameter varia-
W. D. Greason is with the Faculty of Engineering Science, The University tions which might result from the periodic application of test
of Western Ontario, London, ON, Canada N6A 5B9. voltages. The initial and final data points are joined, using a
J. Critchley is with Department 2810, Northern Telecom Canada, Ltd.,
P.O. Box 5155, London, ON, Canada N6A 4N3. straight line, to clarify the presentation of the information and
IEEE Log Number 8609813. show the general direction of change.

0148-641 l/86/0900-0293$01.00 0 1986 IEEE

294 IEEE TRANSACTIONS ON COMPONENTS, HYBRIDS, AND MANUFACTURING TECHNOLOGY, VOL. CHMT-9, NO. 3, SEPTEMBER 1986

TABLE I
DISTRIBUTION OF CAPACITORS TESTED, BY VENDOR AND TYPE

2.00
VENDOR 1.09 t

A B C D

1 pF, 5d V 30 10 10 10
10 pF, 35 V 10 10 30 10
22 pF, 10 V 30 10 10 10’

0 0.92 -

Capacitance
a90h TIME
For the three different capacitor types, plots showing the Fig.- 1. Effect of accelerated temperature test on capacitance (I-pF 50-V
change in capacitance value-are presented in Figs. l-3.- The capacitor).

range in capacitance measured for devices from the four

different vendors is indicated on these graphs while a straight
line is used to join the. average of the values for the four
vendors. It can be seen that the accelerated life test causes a
decrease in capacitance.
Table II summarizes the normalized final value of capaci:
tance after 3000 h by vendor and capacitor type. The average
reduction in capacitance appears to increase as the initial
capacitor value, increases. This change in capacitance was
relatively constant for all values supplied by a given vendor,
with the exception of Vendor B, whose change was much less 9.00 ,

than the others. The overall normalized average value of INITIAL

TIME
3000 h

capacitance was 0.94. Fig. 2. Effect of accelerated temperature test on capacitance (IO-pF 35-V
capacitor).
ESR
Table III summarizes the initial ESR values. Figs. 4-6 show
the effect of the accelerated temperature test on ESR,
calculated from, the measured values of dissipation factor and
capacitance:
Tables IV and V summarize the-normalized final values of .
ESR at 120 Hz and 1000‘ I-Iz, respectively. The average
normalized final value of ESR, at .both measurement frequen-
ties, was approxi.mately 1.6. The results did indicate, how-
ever, that for the 120-Hz measurement,’ the relative change in
ESR decreased as the capacitor value increased, while for the
lOOO-Hz measurement; the, relative change in ESR increased I I
INITIAL 3000 h
as the capacitance value increased. A comparison of capacitors TilE
by vendor showed a change in ESR which varied from i .09 to Fig. 3. Effect of accelerated temperature test on capacitance (22~pF 10-V
capacitor).
1.95 times the initial levels. Also note that capacitors supplied
by Vendor B had a significantly smaller change in. ESR
compared to the specimens from the other vendors.
TABLE II
Le&w G / SUMMARY OF NORMALIZED FINAL VALUES OF CAPACITANCE AFTER
3000-h TEST
Table VI summarizes.the average initial leakage currents
measured, while plots showing ‘the change in leakage current
VENDOR
for the three types of capacitors, measured at 1 V, 2 V, and
rated voltage, are given in Figs. 7, 8, and 9. Note that the ‘A it C D Average
leakage currents at the l- and 2-V levels were measured in
nanoamperes while the leakage current at rated voltage was 1 gF, 50 V 0.96 0.99 0.47 0.97 0.97
measured in microamperes. 10 pF, 35 V 0.92 0.98 0.93 0.93 0.94
22 pF, 10 V 0.92 0.99 0.88 0.88 0.92
Table VII summarizes the normalized final value of leakage Average 0.93 0.99 0.93 0.93 0.94
current, measured at rated voltage. The overall normalized
final leakage current was 4.3 times the initial amount.
Wide variations existed in the measurement of leakage
GREASONANDCRITCHLEY:ALUMINUMELECTROLYTICCAPACITORS 295

TABLE III TABLE ‘IV,

AVERAG~INITIALESRVALUESFORCAPACITORS SUMMARY OF NORMALIZED FINAL ESR VALUES AT 120 HZ AFTER
3000-h TEST

VENDOR
Vendor
Parameter A B C D
A B C D Average
ESR at 120 Hz, fl
1 pF, 50 V 36.4 24.4 28.8 50.0 1 /IF, 50 V 2.25 1.33 2.34 1.75 1.92
10 ~JF, 35 V 8.5 6.3 6.2 6.7 10 gF, 35 V 1.67 1.05 1.91 1.74 1.59
22 /LF, 10 V 6.0 4.2 6.9 6.7 22 pF, 10 V 1.53 1.09 1.59 1.58 1.45
Average 1.82 1.16 1.95 1.69 1.65
ESR at I kHz, Q
1 pF, 50 V 10.8 5.9 11.5 15.5
10 pF, 35 V 2.6 2.1 2.1 1.9
22 ~IF, 10 V 2.1 2.0 6.5 2.5 TABLE V
SUMMARY OF NORMALIZED FINAL EsR VALUES AT 1000 HZ AFTER
3CO-~TFST

ESR AT 120 Hz -
100 ESR AT I kHz ---
t A B C D Average

1 pF, 50 V 1.57 1.13 1.72 1.42 1.46

10 pF, 35 V 1.89 1.02 1.87 1.81 1.64
22 gF, 10 V 2.05 1.12 1.88 1.98 1.76
Average 1.84 1.09 1.82 1.73 1.62

TABLE VI
INITIAL 3000 h
TIME AVERAGEINITIALLEAKAGECURRENTFORCAPACITORS

Fig. 4. Effect of accelerated temperature test on ESR (1 -gF 50-V capacitor).

Vendor

Parameter A B C D

I5 ESR AT 120 Hz - Leakage current at 1 V, nA

ESR AT I kHz ---
1 pF, 50 V 0.05 0.24 0.13 0.33
10 pF, 35 V 1.12 1.54 0.52 2.07
I l 22 pF, 10 V 2.98 1.50 2.26 7.98
Leakage current at 2 V, nA
IO- 1 pF, 50 V 2.00 1.49 0.56 0.81
4 10 pF, 35 V 2.42 6.48 0.93 9.71
=; 22 gF, 10 V 12.20 10.60 12.30 50.20
z Leakage current at
5- rated voltage, PA
1 /IF, 50 V 0.05 0.26 0.05 0.07
10 gF, 35 V 0.12 0.34 0.07 0.32
22 pF, 10 V 0.34 0.38 0.14 0.49

INITIAL 3000 h
TIME
Fig. 5. Effect of accelerated temperature test on ESR (10qF 35-V
capacitor).

ESR AT 120 Hz -
IO ESR AT I kHz --

O.ll 1 , (
INITIAL 3000 h
lNlTlAl 3000 h
TIME TIME
Fig. 6. Effect of accelerated temperature test on ESR (22-PF 10-V Fig. 7. Effect of accelerated temperature test on leakage current (1-gF SO-V
capacitor). capacitor).
 296 IEEE TRANSACTIONS ON COMPONENTS, HYBRIDS, AND MANUFACTURING TECHNOLOGY, VOL. CHMT-9, NO. 3, SEPTEMBER1986

TABLE VIII
SUMMARY OF NORMALIZED FINAL VALUE OF LEAKAGE CURRENT AT
1v

Vendor

A B C D Average

1 fiF, 50 V 2.8 136.4 1.6 32.6 43.4

10 pF, 35 V 0.6 2.5 1.3 11.9 4.1
22 gF, 10 V 0.8 1.2 2.1 12.9 4.3
Average 1.4 46.7 1.7 19.1 17.2

TABLE IX
SUMMARY OF NORMALIZED FINAL VALUE OF LEAKAGE CURRENT AT
Fig. 8. Effect of accelerated temperature test on leakage current (lo-PF 35-
2v
V capacitor).
Vendor

A B C D Average

1 pF, 50 V 5.2 8.7 0.9 63.1 19.5

10 gF, 35 V 2.1 15.8 1.1 7.8 6.7
22 fiF, 10 V 0.6 18.2 1.3 5.4 6.4
Average 2.6 14.2 1.1 25.4 10.9

IV. DISCUSSION

The results will be discussed in relationship to parameters of

an equivalent circuit for an electrolytic capacitor and some
guidelines will be formulated to assist circuit designers in their
L I
evaluation of aluminum electrolytic capacitors as circuit
“’ INITIAL 3000 h components. Aluminum electrolytic capacitors can be ana-
TIME
lyzed with equivalent circuits using models with distributed or
Fig. 9. Effect of accelerated temperature test on leakage current (22-PF lo-
V capacitor).
lumped parameters [l]-[7]. The discrete component model,
shown in Fig. 10, will be used in this discussion. RI models
the resistance associated with the electrolyte, foil spacer and
contacts, RL is the resistance which takes into account the
TABLE VII
transport of leakage current through the aluminum oxide while
SUMMARY OF NORMALIZED FINAL VALUE OF LEAKAGE CURRENT AT
RATED VOLTAGE R2 is the resistance which accounts for the dielectric loss of the
aluminum oxide. Inductive effects will be neglected for low-
Vendor frequency operation.
For dc leakage current tests, the equivalent circuit simplifies
A B C D Average while for ac tests, the equivalent circuit can be
to RL,
approximated by R, and R2 in series with C. The measured
1 pF, 50 V 2.6 8.5 0.8 6.9 4.1
10 ~IF, 35 V 2.3 5.5 1.3 9.4 4.6 dissipation factor tan 6 equals wC(R, + R2) from which the
22 pF, 10 V 0.76 5.8 2.2 5.4 3.5 effective series resistance (ESR) can be equated to (RI + Rz).
Average 1.9 6.6 1.4 7.2 4.3 The operating electrolyte and the oxide film are the main
factors responsible for determining the operating performance
of an aluminum electrolytic capacitor [8]. The results will be
analyzed with reference to these two key factors.
currents at the l- and 2-V levels. Tables VIII and IX
summarize the results, which show that leakage currents Capacitance
increased, on average, by 17.2 times at the 1-V test level, The observed decrease in capacitance could be caused by
while an average increase of 10.9 times was measured at the 2- electrolyte dryout which would result in a slight increase in the
V level. It is of interest to note that capacitors supplied by effective plate separation of the capacitor. The larger plate
Vendors B and D exhibited larger changes in leakage currents, surface area associated with a larger capacitance device could
at all three test voltages when compared to capacitors supplied be responsible for the greater change in capacitance noted with
by the other vendors. these devices.
GREASONAND CRITCHLEY: ALUMINUM ELECTROLYTIC CAPACITORS 297

an increase in electrolyte resistivity and a slight decrease in the

electrolyte layer thickness due to electrolyte dryout. The
observed increase in ESR suggests that the former factor
RL
predominates. Since it was observed that the relative capaci-
Fig. 10. Capacitor equivalent circuit.
tance change was greater in the large value capacitors, it
suggests that the electrolyte thickness change was greater in
ESR these units, possibly due to the larger plate surface area. The
The dissipation factor measured can be expressed as observed greater change in ESR for the larger value capacitors
could be the result of a greater change in electrolyte resistivity
tan 6=tan G’+wCR, (2) in these units.

where tan 6’ is the dissipation factor for the A1203 layer. The Leakage Current
ESR can then be calculated as Leakage current is a function of the oxide quality, geome-
tan 6’ + wCRl try, and the applied voltage. The resistance of the oxide is
ESR= _ . (3) RL = p&A where p is the volume resistivity (Q - m), d is
WC
thickness (m), and A is the plate area (m2), and the
It has been shown that tan 6 ’ decreases with frequency in the capacitance is C = Ke,-J/d with K the dielectric constant, e.
range of 10-1000 Hz and is also inversely related to oxide the permittivity of free space (8.85 x lo-l2 F * m-l).
thickness or formation voltage [9]. Therefore, the leakage current 1, for an applied voltage VA is
It was observed from the initial measurements that the given by
capacitor with the lowest capacitance value and highest voltage IL= VA/RL=K, VaC (4)
rating yielded the highest ESR value. As the capacitance value
increases and the voltage rating drops, two opposing effects where K1 = l(pKeo).
can be seen from an examination of (3). If the oxide layer is Thus for a given voltage, the leakage current is proportional
assumed to be the major contribution to ESR, then at a given to the capacitance value. Due to the nonlinear V-Icharacteris-
frequency: tic of aluminum electrolytic capacitors [l 11, leakage current
will decrease at a greater than linear rate as the applied voltage
1) the ESR will increase due to the inverse relationship
is decreased from the capacitor’s rated voltage. Nevertheless,
between tan 6’ and oxide thickness; and
the linear relationship can be used as a first-order approxima-
2) the ESR will decrease due to the inverse relationship
tion when evaluating aluminum electrolytic capacitors.
between ESR and capacitance.
From (4), for a fixed value of capacitance:
The latter effect predominates, and the net result is a decrease
in the ESR which is in agreement with the results obtained. IL=K2VA
It was also observed that the ESR at the lOOO-Hz test where K2 = K1 C. Then
frequency was smaller than that at the 120-Hz test frequency.
This is due to a decrease in tan 6 ’ as frequency increases [9] log IL = log K2 + log VA. (5)
and the presence of w in the denominator of (3). At even
higher frequencies, the wCRl term can be expected to Figs. 11 and 12 give the least-squares straight line plots of the
predominate compared to the tan 6 ’ term [lo]. initial and final leakage currents, respectively. The results are
An interesting observation was made when the results of the in agreement with the linear relationship between leakage
accelerated life test were analyzed with respect to ESR current, capacitance value and applied voltage.
change. The ESR of electrolytic capacitors is the sum of The effect of an accelerated life test causes an increase in
several elements with the electrolyte and the anodic oxide film leakage current which may be due to a decrease in oxide
being the major contributors. It has been noted [4], [lo] that resistivity. The greater relative change at the l- and 2-V test
the contribution of the electrolyte is frequency independent levels may be due to the nonlinear conduction mechanism of
while the contribution due to the series resistance of the film is the oxide layer [ 1l] and the presence of a polarization current
inversely proportional to frequency. component in the currents measured [12]. However, a linear
At 120 Hz, the change. in ESR was greatest for those V-I characteristic can still be used to give an estimate of
capacitors with the higher voltage ratings or thicker oxide. If leakage currents at low operating voltages, compared to
the oxide is the main contributor to ESR at this test frequency, leakage currents at rated voltages after exposure to an
the results suggest that the accelerated life test causes a bulk accelerated life test.
related effect in the aluminum oxide which is more predomi-
nant in the higher voltage rated capacitors. V. VENDOR VARIATIONS
In contrast, the change in ESR at 1000 Hz was greatest for The discussion has been based on average results obtained
lower voltage higher capacitance devices. This suggests that from measurements made on devices supplied by four differ-
the electrolyte is the main contributor to ESR at this ent vendors. It is interesting to note how initial values and
frequency, and that electrolyte dryout may be responsible for relative changes varied among vendors.
the observed results. An accelerated life test could cause both Table III and VI show that Vendor B had units with a
 298 IEEETRANSACTIONSONCOMPONENTS,HYBRIDS,ANDMANUFACTURINGTECHNOLOGY,VOL.CHMT-9,NO.3,SEPTEMBER1986

of the results showed some additional factors which may prove

useful to circuit designers.
At low frequencies, where the oxide predominates, ESR is
inversely proportional to the capacitance value. During the life
of the capacitor, the low-frequency ESR will increase more in
higher voltage rated devices. The high-frequency ESR is
predominantly due to the electrolyte and increases more in
higher capacitance value devices during an accelerated shelf-
life aging test. Also note a general decrease in ESR with
frequency which is the result of the decrease in the oxide
o-10 NF, 35v contribution term in the overall expression for ESR.
x-22pF, IOV Leakage current at a low operating voltage can be estimated
-10 ' assuming a linear relationship between the leakage current at
0 I.0 2.0 the rated voltage and the chosen operating voltage. Leakage
LOG APPLIED VOLTAGE, v current increases as a result of an accelerated shelf-life aging
Fig. 11. Log leakage current versus log applied voltage: initial test and is higher at lower operating voltages.
ments.
Leakage current can be reduced by using a smaller value of
capacitance. It should be noted, however, that the low-
frequency ESR is inversely proportional to capacitance so that
both leakage current and ESR cannot be optimized simultane-
a -5..
ously. Initial selection of a capacitor with a higher voltage
rating can be used to reduce the leakage current, but as noted,
ESR will increase at a greater rate during the life of the device.
The following guidelines are suggested for selection of an
aluminum electrolytic capacitor for use in a low-voltage circuit
application.
a) Select the capacitance value with the voltage rating
closest to the operating voltage of the circuit.
b) By using the vendor’s specifications, check ESR and
calculate the derated leakage current.
c) Increase the capacitance value if ESR is to be reduced.
d) Recalculate the derated leakage current and select a
0 1.0 2.0 device with a higher rated voltage if leakage current is to
LOG APPLIED VOLTAGE, v be reduced.
Fig. 12. Log leakage current versus log applkl voltage: final measure- e) A first-order approximation of the change in the capaci-
ments.
tor’s parameters when operated at an average tempera-
ture of 35°C for 11 years can be estimated from the
results reported in this paper.
consistently low initial value of ESR while initial leakage f) Significant life at a given temperature can be estimated
currents were higher in devices supplied by Vendors B and D. from
After the accelerated life test, capacitors from Vendor B Lz=L,x2(TI-T2)'10
showed the least change in capacitance and ESR but had (6)
relatively large changes in leakage current. Capacitors from where
Vendor D also exhibited relatively smaller changes in ESR and
Ll significant life at temperature T,,
larger changes in leakage current. In contrast, capacitors from
J52 significant life at temperature T2.
Vendors A and C showed the greatest change in capacitance
and ESR but the least change in leakage current. Equation (6) is based on Arrhenius’ equation in the field of
These observations suggest that both leakage current and chemical kinetics [13], using a temperature acceleration
ESR cannot be optimized simultaneously by the manufacturers coefficient of two and assuming an activation energy of 1 eV/
of aluminum electrolytic capacitors. molecule. A temperature change of 10°C causes the life to
change by a factor of two. Failure rates, for a device operating
VI. CAPACITOR APPLICATION GUIDELINES at a given temperature can be predicted from failure rates
The results of accelerated shelf-life aging tests can be used measured at an accelerated temperature [14]. The results of
to estimate performance characteristics of aluminum electro- these tests apply for L, = 3000 h and Tl = 85°C.
lytic capacitors in noncontinuously powered low-voltage
circuit applications. From a general point of view, the VII. CONCLUSION
accelerated tests yielded a decrease in capacitance, and The results of an accelerated shelf-life aging test, run for
increases in leakage current and ESR. However, the analysis 3000 h at 85°C on samples of three types of aluminum
GREASONANDCRITCHLEY:ALUMINUMELECTROLYTICCAPACITORS 299

electrolytic capacitors supplied by four different vendors has 1968.

[2] R. M. Peekema and J. P. Beesley, “Factors affecting the impedance of
been presented. The objective of the tests was to simulate
foil-type electrolytic capacitors,” Electrochem. Technol., vol. 6, pp.
performance characteristics of the capacitors when used in 165-172, 1968.
low-voltage noncontinuously powered circuit applications. [3] D. A. McLean, “The a-c properties of tantalum solid electrolytic
The average decrease in capacitance was six percent while capacitors,” J. Electrochem. Sot., vol. 108, pp. 48-56, 1961.
[4] F. G. Hayatee, “The equivalent series resistance in electrolytic
ESR (at both 120 Hz and 1000 Hz) increased on average by 65 capacitors,” Electrocomponent Sci. Technol,, vol. 2, pp. 67-72,
percent. Leakage current at rated voltage increased by 430 1975.
percent while levels at the 2- and 1-V test voltages increased [5] C. S. Lindquist, “Capacitor DF and Q,” IEEE Trans. Components,
Hybrids, Manuf. Technol., vol. CHMT-1, pp. 115-117, 1978.
by 1090 and 1720 percent, respectively. [6] R. P. McManus, “Aluminum electrolytic capacitors,” Sprague Elec-
The results indicate that ESR and leakage current cannot be tric Co., tech. Paper TP72-2A.
optimized simultaneously in the manufacture of aluminum [7] E. H. L. J. Dekker, D. Fidder, H. Schmickl, and B. R. Spcarman,
“Small electrolytic capacitors-The argument for solid aluminum
electrolytic capacitors. However, guidelines have been pre- types,” Electron. Eng., vol. 54, pp. 75-86, Apr. 1982.
sented which allow a circuit designer to adjust the capacitance [8] W. J. Bernard, “Developments in electrolytic capacitors,” .I. Electro-
value and voltage rating of a device to meet the requirements them. Sot., vol. 124, pp. 403-409, 1977.
[9] S. H. A. Begemann and A. W. Smith, “Dielectric properties of surface
of a circuit application. oxides on aluminum,” .I. Electrochem. Sot., vol. 117, ‘pp. 1440-
1442, 1970.
ACKNOWLEDGMENT [lo] W. J. Bernard, “The equivalent series resistance of anodically formed
oxide films on aluminum,” J. Electrochem. SOL, vol. 108, pp. 446-
This work is the result of a joint project involving the 449. 1961.
Facultv of Eneineering SCienCe. The UBiVtXSitY of Western 1111 J. h. Albella. C. Gomez-Aleixandre, and J. M. Martinez-Duart,
Ontario, and GortheriTelecom’ Canada Ltd., London, ON, _ - “Dielectric characteristics of miniature aluminum electrolytic capaci-
tors under stressed voltage conditions,” J. Appl. Electrochem., vol.
Canada. The authors wish to express their appreciation for 14, pp. 9-14, 1984.
support shown for this joint University-Industry project. [12] A. R. Von Hippel, Dielectrics and Waves. Cambridge, MA: MIT
Press, 1954.
[13] S. W. Benson, The Foundations of Chemical Kinetics. New York:
REFERENCES McGraw-Hill, 1960, p. 66.
[l] R. H. Broadbent, “Alternating-current properties of aluminum foil [14] R. A. Evans, “The -analysis of accelerated temperature-Tests,” in
electrolytic capacitors,” Electrochem. Technol.. vol. 6, pp. 163-165, Proc. 1969 Annu. Symp. Reliability, IEEE Catalog no. 69C8-R.