621.317.333.

6 The Institution of Electrical Engineers

Paper No. 3723
Feb. 1962

ELECTRICAL BREAKDOWN BY TRACKING

By N. PARKMAN.
{The paper was first received \9th May, and in revised form 28th July, 1961. // was presented at the INTERNATIONAL CONFERENCE ON
COMPONENTS AND MATERIALS USED IN ELECTRONIC ENGINEERING 15th June, 1961.)

SUMMARY
When an electrically stressed insulating material is exposed to con-
ditions which contaminate its surfaces, the tendency for the resultant
transient leakage currents to form conducting tracks on insulator
surfaces is of recognized importance. This form of failure is known as
tracking and the processes leading to its development are very depen-
dent on the ambient conditions. The work reported here uses a
contaminant internationally accepted for assessing the degree of
resistance of insulation to tracking and indicates some of the factors
important in causing breakdown under these conditions. Above a
certain voltage the time to failure depends on the power input to small
arcs on the surface of insulation. At low voltages the dependence on
power input is not so marked and failure is more dependent on
chemical processes. The electric field necessary to produce tracking in i
track-resistant materials falls off much more slowly with increase in
electrode spacing than is the case for materials with poor resistance to
tracking. It is suggested that classification of materials by their
changed susceptibility to tracking with change in electrode spacing
may be advantageous. The values of direct voltage required to pro-
duce tracking are sometimes substantially greater than those for :
alternating voltage. If, however, reversal of polarity occurs the
direct-voltage values are drastically reduced. !

(1) INTRODUCTION
Where a dielectric material is exposed to conditions which
contaminate its surface, leakage currents can flow when a
voltage gradient exists across the surface. In some direct-
voltage cases where the contamination is sufficient to produce a
continuous leakage current, electrochemical processes can
occur, especially at electrodes, leading to various undesirable
effects. In other cases the passage of a leakage current produces
sufficient heating of the surface to remove the conductingfilmof
contamination locally. This sudden introduction of a high-
resistance discontinuity in the conducting film causes an electric Fig. 1.—Tracking on an s.r.b.p. bushing.
spark to occur at the site of the discontinuity. The leakage
current then falls to a very low value until further contamination diverse mechanisms unresolved. The paper deals with a few of
causes the above process to recur. The sparks which occur these mechanisms as they manifest themselves under specific
at each break in the conducting layer can cause degradation of conditions.
the insulator surfaces. This is particularly the case if the
insulator is organic, when the ultimate result is frequently the (2) TEST APPARATUS AND CONDITIONS
formation of a continuous conducting carbon track between
electrodes. This is known as breakdown by tracking, and it The greatest difficulty surrounding the development of tests
should be differentiated from that failure mechanism in which to investigate and assess track resistance is the choice of electrode
dielectric breakdown is due to repeated arcs between electrodes. materials and contaminating conditions. In service these para-
The latter process is known as breakdown by arcing. meters will, in general, vary from one application to another, so
that choice of conditions is bound to be somewhat arbitrary.
Failure by tracking can occur, not only at the highest supply The international (I.E.C.) test has standardized 0 • 1 % ammonium
voltages, but under certain circumstances at a few tens of volts. chloride in water as its contaminant, and this is placed in suc-
Fig. 1 shows an example of tracking in an s.r.b.p. bushing. cessive 20 mm3 drops between brass electrodes resting on the
Thus stated, there appears to be little that is recondite in the surface to be tested. This contaminant has been chosen because
processes leading to tracking. However, the number of inter- it volatilizes during the test, thereby preventing any appreciable
dependent mechanisms is so large that elucidation of the pro- build-up in contaminant concentration as the test proceeds.
cesses has been largely neglected. Treatments of the subject The test frequency is 50c/s. The I.E.C. system is shown
have frequently accepted the fact that the effect is a composite diagrammatically in Fig. 2, and an automatic form1 of the
one and that the only profitable approach is to attempt to apparatus has been used to obtain much of the data provided
classify materials by a standard test, leaving the nature of the here.
Mr. Parkman is at the Electrical Research Association Laboratory. A similar electrode arrangement has been used for photo-
[448]

Authorized licensed use limited to: Indian Instt of Engg Science & Tech- SHIBPUR. Downloaded on March 11,2020 at 11:39:40 UTC from IEEE Xplore. Restrictions apply.
 PARKMAN: ELECTRICAL BREAKDOWN BY TRACKING 449

2±0-l
(3) RESULTS
T Tbf (3.1) Test Specimens
ELECTRODES-

TEST SPECIMEN^
To compare the effects of varying the parameters in the test,
20 five materials were used which had Va between 110 and 260 V.
SUPPORT-
(a) These were all moulding compounds of the following types:
5:O*I 0M-0-2
ADJUSTABLE LOAD. (i) Epoxy resin with short glass-fibre filler,
ELECTRODE (ii) Alkyd resin with mineral filler,
HOLDER (iii) Phenolic resin with asbestos and cotton fillers,
KNEE (iv) Phenolic resin with woodflour filler,
JOINT (v) Polycarbonate.
Because the supply of polycarbonate was limited at the time
of this investigation, comparisons were made with either (i),
(ii), (iii) and (iv) or (i), (ii), (iii) and (v), both (iv) and (v) having
ADJUSTABLE quite poor track resistance. Because the tracking susceptibility
TABLE
of a material can vary with moulding conditions, comparative
tests to assess the importance of a particular parameter were
Fig. 2.—Electrode system. carried out using a single disc of material.
The values of Va quoted are given after neglecting the unit
a) Electrode details.
[b) Test arrangement. figure; e.g., if the curve gave Va as 256 V this would be quoted
as 250 V, and if the curve gave 250 V this would be quoted as
graphic investigation. Drops of electrolyte are placed between Va. This practice was adopted since values of Va with constant
the electrodes until tracking occurs. The work is repeated for a parameters are reproducible only within 10 V.
range of voltage gradients between the electrodes, and a curve (3.2) Typical Form of Results
of voltage against number of drops of contaminant to cause
tracking is derived. A typical plot is shown in Fig. 3, where it The curve shown in Fig. 3 is typical of that obtained for many
materials. Fig. 4 shows similar data plotted on a log/log scale.
too 5O 1

90 \
\

1I
PRODUCE TRACKING

80 \
ro
O

70 \

\
—

y\
o> to O
PS TO

+ \
\ O
\
4O ex
° 4
LL. \
30 \ O
NUMBER

+
\
ro

\
Va
—

IOO 2OO SOO IOOO

O IOO 2OO 3OO 4OO SOO 6OO 7O0 TEST VOLTAGE
VOLTS Fig. 4.—Number of drops of electrolyte required to produce tracking
Fig. 3.—Typical tracking voltage curve. at different voltages.

The slope of the curve is about 2 for voltages above a certain

will be seen that the number of drops required to produce value, but below this the exact slope is difficult to determine
tracking increases rapidly below a certain voltage. This because of the dispersion in the results. The square law which
'asymptotic' value of voltage (Va) may then be taken as the operates above Va suggests that the tracking processes are
measure of a material's resistance to tracking. In some cases simply dependent on the power input at the site of the discon-
the asymptotic value is approximately equal to the voltage tinuity in the film of contaminant. Later results show that this
corresponding to 50 drops. The latter is called the comparative simple concept is complicated by other parameters.
tracking index (C.T.I.) in the I.E.C. test, and for convenience the
50-drop value has also been taken for Va in this work. The (3.3) Effect of Varying the Electrode Spacing
curves were drawn through minimum values obtained in tests The tests reported here were carried out at 2 mm, 4 mm and
at different voltages. 6 mm ± 0 - 1 mm spacing. The results, given in Fig. 5, show

Authorized licensed use limited to: Indian Instt of Engg Science & Tech- SHIBPUR. Downloaded on March 11,2020 at 11:39:40 UTC from IEEE Xplore. Restrictions apply.
 450 PARKMAN: ELECTRICAL BREAKDOWN BY TRACKING

—-j;

E p— (.0)
•—»•,
(d)
>SO \ ,

Ki
*•>

\ (f)
\ V %
(f)
— (c

2 5 IO
ELECTRODE SPACING,mm

Fig. 6.—Electric stress required to produce tracking at different

electrode spacings.
(a) Epoxide
(b) Polycarbonate Using NaCl contaminant.
(c) Phenolic
SO IOO 2OO 3OO 4OO 5O IOO 2OO 3OO (d) Epoxide
VOLTS VOLTS (e) Phenolic I Usin.ig NH4CI contaminant.
(/) Polycarbonate
Fig. 5.—Effect of change in Electrode spacing on resistance to
tracking.
Spacing
variation with electrode spacing. While Va will vary appreciably
Va
mm volts with change of contamination, it seems possible that x, which
(a) Glass-filled epoxide 2 150 determines the dependence of tracking voltage on spacing, may
4 250
6 360 not change greatly. It would be unwise, therefore, to incor-
(b) Unfilled polycarbonate 2 90
porate materials with poor track resistance and large spacing
4 110 coefficients, x, into designs merely (as is done in some cases) by
6 120
specifying increased creepage distances.
(c) Wood-flour-filled phenolic 2 80
4 130
6 160 (3.4) Effect of Electrolyte Conductivity
Mineral-filled alkyd 2 170
4 260 The results given in Fig. 7 were obtained using electrolytes of
6 320 ammonium chloride in concentrations of 0 05%, 0-1% and
that, for materials with poor track resistance, Va does not
increase markedly with increase in electrode spacing; e.g. Va
for the polycarbonate increases by only 30 V as the electrode
spacing increases from 2 to 6 mm. Va for the alkyd material,
however, increases by 150 V in the same range. When two
fairly track-resistant materials are compared, the Va of the more
heterogeneous material may show a greater dependence on
electrode spacing, especially if it contains regions of fairly high
track resistance, e.g. glass fibre. Thus, while the Va of the
mineral-filled alkyd goes from 170 to 320 V when the electrode
spacing is increased from 2 to 6 mm, the Va of the heterogeneous
glass-filled epoxy goes from 150 to 360 V for the same change
in spacing. For the latter material a tolerance of ± 0 1 mm in
spacing would contribute a maximum difference of about 10 V
in Va. For materials with much poorer track resistance the TOO 2OO SOO IOOO
difference is less; e.g. the corresponding difference for the ELECTROLYTE RESISTIVITY, fl'C
polycarbonate is only about 1V. If the field strength Ea corre-
sponding to Va is plotted for different spacings /, as in Fig. 6, it Fig. 7.—Voltage to cause tracking (Va) as a function of electrolyte
is found that the relationship is of the form Ea = kt~x, where x resistivity.
is about 0-4 for the track-resistant materials and about 0-7 for (a) Epoxide. (c) Phenolic.
(b) Alkyd. (d) Phenolic.
materials with poor track resistance. These figures are, of
course, relevant to a specific operating condition, so it may be 0 15%, corresponding to resistivities of 780, 416 and 260Q-cm
argued that the derived relationship between field and electrode at 18° C. It will be seen that Va decreases as the resistivity of the
spacing is not generally valid. Results of tests with sodium contaminant decreases. For the data given the approximate
chloride electrolyte, also plotted in Fig. 6, show that the absolute relationship between Va and resistivity, p, is Va = kp~x, where
values of Va, and hence of Ea, are appreciably changed with the A: is a constant and x lies between 0 • 3 and 0 • 6. The data of
change of electrolyte. The corresponding values of x, however, Mathes and McGowan,2 using a different type of tracking test,
are still approximately the same as with ammonium chloride in give a value for x of 0 • 5.
most cases. Thus there are two factors of importance in deter- At the instant a discontinuity is introduced into the conduct-
mining resistance to tracking, namely absolute values of Va and ing film by evaporation of electrolyte, a discharge bridges the

Authorized licensed use limited to: Indian Instt of Engg Science & Tech- SHIBPUR. Downloaded on March 11,2020 at 11:39:40 UTC from IEEE Xplore. Restrictions apply.
 PARKMAN: ELECTRICAL BREAKDOWN BY TRACKING 451
electrodes. It is reasonable to assume that the running current
for the discharge which bridges the discontinuity will depend on
the series impedance. This is made up of transformer imped-
ance, short-circuit series resistance, which is adjusted to suit
the test voltage, and the resistance of the electrolyte. The effect
of electrolyte resistivity will therefore depend on the diluting
effect of other circuit impedances so that it may vary from one
apparatus to another. Tests carried out with NaCl electrolyte
with the same resistivity as 0 • 1 % NH4C1 give tracking voltages
which differ greatly from the results with the latter electrolyte.

(3.5) Effect of Short-Circuit Current.

Changes in short-circuit current are brought about by varying s
the resistance in series with the test electrodes. It will be seen -
from Fig. 8 that the effect on Va is not very significant. Appar- en

•a
. —1 a
to I
(d)

1 V ft
c

4O
1\(
\
X

tracking.
1
SO IOO
1
2OO
VOLTS

Fig. 9.—Effect of change of electrode material on resistance to

3OO

i °\
\
X + Brass Rh Pd • Au
(a) Glass-filled epoxide.

\y \(o
A
\\(b)
Au ")
Rh >• Would not cause tracking up to 1000V.
Pd J
Brass, Va = 280 V.
(b) Cotton-filled phenolic.
(a) Va, volts
Rh 200
Brass 160
Pd 150
Au 90
(c) Wood-flour-filled phenolic.
6O IOO ISO 2OO Va, volts
VOLTAGE Rh 100
Pd 130
Fig. 8.—Effect of short-circuit current on tracking voltage for a cotton- Brass 130
Au 70
filled phenolic. (d) Mineral-filled alkyd.
(a) o 1 A Short-circuit current. Va, volts
(b) D 0 • 3 A Short-circuit current. Rh 320
(c) x 0-7A Short-circuit current. Pd 310
Au 290
Brass 240
ently the resistance of the thin film of surface contaminant is
much greater than the series resistance, so that changes in the (3.6.1) Changes in Electrolyte Conductivity.
latter do not affect the total resistance appreciably. The epoxide material has a Va of 250 V when using brass
electrodes and ammonium chloride contaminant but cannot be
(3.6) Electrode Effects made to track when platinum electrodes are used. Tests were
The results given in Fig. 9 show that the nature of the elec- therefore repeated on this material using brass electrodes and
trodes plays an important part in determining Va. While there removing electrolyte from around the electrodes after every
is a tendency for the inert metals to give somewhat higher values tenth drop. In this way a sample of the electrode-electrolyte was
of Va than is the case with brass electrodes, there is no invariable built up until sufficient was available to carry out a conductivity
rule, and the values for brass may be higher or lower than those test. A similar sample was obtained using platinum electrodes.
for palladium (say). For fairly track-resistant materials the It was found that, although the conductivity of the electrolyte
change to inert-metal electrodes will frequently make it impos- removed from the region near the electrodes was in each case
sible to cause tracking under the conditions of the present test. higher than for the 0.1% NH4C1 used as contaminant, there was
However, materials with poor track resistance (e.g. phenolics) do no significant difference in the conductivity of samples obtained
not usually change their value of Va by more than a factor of from around the brass and platinum electrodes.
two owing to a change in the metal of test electrode.
Presumably the only ways in which the electrode material can (3.6.2) Tests with Direct Voltage.
influence Va is by entering into chemical reactions which The apparent absence of any significant difference in the
(a) increase the conductivity of the contaminant, and (b) modify change of electrolyte conductivity with different electrode
the surface of the specimen. Increasing the electrolyte con- materials led to other investigations using direct voltage as a
ductivity was shown to decrease Va, so that changes in electrolyte more convenient means of separating the electrode processes.
conductivity due to reaction with electrodes may be expected to Using the same electrode arrangement as for the alternating-
produce similar effects. voltage work, it was found that the epoxide material would not

Authorized licensed use limited to: Indian Instt of Engg Science & Tech- SHIBPUR. Downloaded on March 11,2020 at 11:39:40 UTC from IEEE Xplore. Restrictions apply.
 452 PARKMAN: ELECTRICAL BREAKDOWN BY TRACKING
track at 500 V using either brass or platinum electrodes and
ammonium chloride contaminant. In the tests with brass elec-
trodes, tracking took place fairly readily if the polarity was
reversed after each drop of electrolyte had fallen. If, instead,
the polarity was reversed after 10 drops, tracking took place
either immediately or after one, or at the most two, further
drops. Using other electrode/electrolyte combinations the effect
due to polarity reversal was not always present. Several series of
tests having established that ten drops of ammonium chloride
electrolyte followed by reversal of polarity with brass electrodes
would cause tracking, the following tests were carried out to
determine which electrode processes were operative in this type
of breakdown.
(3.6.3) Electrode Substitution.
Using the test described above, with platinum instead of brass
electrodes, the epoxide material could not be induced to track at
500 V with or without reversal of polarity. Clearly the pheno- Fig. 10.—Dendritic growth on copper electrode after reversal of
mena which occur on reversal of polarity are dependent on the polarity.
electrode material. A further test was carried out in which the
electrodes were wiped clean after 10 drops of electrolyte had
fallen but before reversal of polarity. In this case reversal of
polarity as well as the addition of 20 further drops did not
produce tracking, although a further reversal polarity at this
point, without wiping the electrodes, immediately did so.
To show that the act of removing and replacing the electrodes
was not important, a control test was carried out in which the
electrodes were removed and then replaced without wiping. In
this case reversal of polarity after replacing the electrodes
was sufficient to produce tracking immediately. Subsequent
tests showed that cleaning the anode alone before reversal of
polarity was sufficient to prevent tracking. Applying the same
treatment to the cathode, however, did not prevent tracking on
reversal of polarity. The importance of the deposits on the
anode was further demonstrated by using platinum for one
electrode and brass for the other. When the anode was plati-
num no tracking occurred on reversal of polarity, but when it
was brass, tracking occurred on reversal or soon after. Again,
if the deposits on the anode are taken up in distilled water and
used as the contaminant after 10 drops of the normal NH4C1
contaminant have been used, tracking occurs after one or two
drops without the need for polarity reversal.
Fig. 11.—Copper trees growing on the surface of a sample during a
(3.6.4) Voltameter Tests. tracking experiment.
It is not easy to observe electrode processes in detail using
the above type of tracking test. Experiments were therefore test and that one important way in which the electrode material
made using a voltameter containing the contaminant electrolyte affects the susceptibility of a material to tracking is by causing
and brass electrodes. Observing the anode with a microscope electrode material, e.g. copper, to be deposited in very small
it was found that a brown deposit, similar to that seen on the amounts on the surface of the specimen.
anode in a tracking test, formed on it in the first few minutes of
electrolysis. Bubbles of hydrogen were evolved at the cathode (4) DISCUSSION AND CONCLUSIONS
immediately on application of voltage. When the polarity was As stated in the Introduction it is not possible to extrapolate
reversed after a few minutes, the brown deposit on what was from the results of tests made under specific conditions to cover
formerly the anode was converted into a black dendritic growth, all other operating conditions. Hence, while it is probable that
as shown in Fig. 10. During this conversion no hydrogen was the following considerations will apply qualitatively to any
evolved from the electrode, but when the dendrites were i to i in situation where a tracking hazard exists, they are only quantita-
long, hydrogen was once more evolved. From further chemical tively accurate for the stated conditions of test.
identification tests it appears that in the direct-voltage tests the Above a certain voltage, Va, the minimum number of applica-
anode is first oxidized and then chemical reduction which occurs tions of contaminant required to produce tracking varies
on polarity reversal results in a dendritic growth of black finely approximately with the square of the applied voltage. This
divided copper. suggests that the tracking process in short-time (high-voltage)
Upon carrying out further tracking tests while using cine- tests is dependent on the power input to the small arcs which
photography techniques, the formation of copper trees was occur at the instant a high-resistance discontinuity is introduced
observed on the surface of the epoxide specimens after the in the surface film of conductive contaminant.
reversal of polarity. This is shown in Fig. 11. It seems At voltages around and below Va, the dispersion in the number
likely that these processes are still present in the 50c/s tracking of applications of contaminant required to produce tracking at

Authorized licensed use limited to: Indian Instt of Engg Science & Tech- SHIBPUR. Downloaded on March 11,2020 at 11:39:40 UTC from IEEE Xplore. Restrictions apply.
 PARKMAN: ELECTRICAL'! BREAKDOWN BY TRACKING 453
different voltages makes it difficult to establish a mathematical factor in causing tracking. Nevertheless, tests with different
relationship between the two parameters. It is clear, however, electrolytes of the same conductivity give widely differing track-
that at these low voltages the simple square law does not hold. ing voltages. The electrode/electrolyte interaction must there-
It seems likely that the much longer times required for tracking fore contribute greatly to the effective conductivity in the
at low voltages permit chemical reactions between electrode, longer-time tests.
electrolyte and specimen to play a more important role. In the tests reported here, the increased conductivity does not
For materials which have poor resistance to tracking the arise so much in the electrolyte around the electrodes as from
voltage to cause tracking frequently does not increase greatly with successive processes of oxidation and reduction which occur
increase in electrode spacing. Thus the electric field necessary at the electrodes owing to reversal of polarity. These processes
to cause tracking falls off rapidly as the spacing between elec- result in a dendritic metallic layer being deposited on the surface
trodes is increased. The field necessary to produce tracking in under test. The regions of high conductivity thereby introduced
track-resistant materials falls off much more slowly with increase will permit the passage of current densities sufficient to cause
in spacing. It is thus important to realize that increase in characteristic scintillations (sparks) which carbonize organic
creepage distances while using track-susceptible material will not dielectrics.
necessarily be a sufficient alternative to the use of anti-tracking This work has shown that the direct voltages required to
material. It has already been indicated that the degree of produce tracking are sometimes substantially higher than the
resistance to tracking depends in part on the electrolyte/electrode corresponding power-frequency alternating voltages. However,
combination used for test purposes. However, it seems possible if reversal of polarity occurs, the direct voltages approximate to
that the characterization of materials by their changed sus- the alternating voltages. The phenomena which occur on
ceptibility to tracking with change in electrode spacing will be reversal of polarity are thus of particular significance with
less dependent on the form of test; but many more tests will direct voltage.
have to be carried out before this can be regarded as an (5) ACKNOWLEDGMENT
established fact.
The effect of increasing the conductivity of a given con- Mr. R. Kiernan and Mr. D. Wright carried out much of the
taminant is to decrease the number of drops required to produce experimental work reported here. The author is grateful to
tracking at a given voltage. On the other hand, changing the them for their assistance, and to the Director of the Electrical
size of the drops does not significantly affect the other values Research Association for permission to publish the paper.
required for tracking. Thus the current which flows through
the contaminant while it is being boiled off is less important than (6) REFERENCES
the series resistance of the electrolyte film at the instant of a (1) PARKMAN, N.: 'Design of an Automatic Tracking Com-
break in the film. parator', E.R.A. Report Ref. L/T388, 1959.
The variation of tracking voltage, Va, with contaminant con- (2) MATHES, K. N., and MCGOWAN, E. J.: 'Surface Electrical
ductivity suggests that the power input to small surface arcs Failure in the Presence of Contaminants', American
through the electrolyte resistance is an important controlling Institute of Electrical Engineers, Paper No. 61-21,1961.

DISCUSSION ON
DIELECTRICS
AT THE CONFERENCE 15th JUNE, 1961
Mr. B. Walton: Several papers deal with the subject of thin Mr. R. W. Franklin: First, the anodic oxide films are of a
inorganic dielectric films, particularly oxide films, and it seems glassy or amorphous structure, and if they ever occur in a
clear that these are of increasing technological importance. For crystalline form, something has gone wrong. The oxides are
capacitor applications, probably the two most important useless as a dielectric if we have much crystalline material. This
properties of thefilmare the electric strength and the permittivity. is especially true in the case of tantalum and it seems to be true
To what extent do these parameters depend upon the structure in the case of aluminium too. With anodic oxides we are always
of the material ? working with a glassy state.
The material for which the highest electric strength has been The oxides are grown in fields of the order of 107 V/cm, and
quoted is fused silica; and commercial glasses, while having in these conditions ions are moving through the films. They
strengths lower than that of pure fused silica, presumably are used in fields of about the same order, and so we are using
because of the presence of relatively mobile alkali ions, also have them in capacitors in fields of the order of 107 V/cm.
very high intrinsic strengths. Is this high electric strength in fact The permittivity of anodic oxides is never very high: that of
a function of the glassy state, and have most of the thin oxide aluminium oxide is about 8; that of tantalum about 23-25;
and other inorganic films which have been used in capacitor and that of niobium somewhere in the region of 40.
applications a glassy structure? I know that some are stated to So we have something approaching a glass as a dielectric.
have an amorphous structure, whatever that may imply. In the In our case breakdown generally occurs in holes in the film.
case of crystalline or semi-crystalline films, does electrical break- There is not much indication of current flowing through the
down usually occur at grain boundaries or does it, in practice, compact anodic-oxide film until the field becomes great enough
occur at grosser imperfections in the film? for ions to move. Electronic current can pass at much lower
It is, of course, well known that the very high permittivities fields, but only where there are imperfections.
of some materials are due to polarization dependent on the Mr. H. F. Church: I think Mr. Franklin's statement that these
crystalline structure, but for the lower-permittivity materials, to oxide films are operated at something like their formation
what extent does the permittivity depend upon the structure ? voltage (within 90 % of it) is true only where the counter-electrode

Authorized licensed use limited to: Indian Instt of Engg Science & Tech- SHIBPUR. Downloaded on March 11,2020 at 11:39:40 UTC from IEEE Xplore. Restrictions apply.