IEEE Transactions on Electrical Insulation Vol. EI-15 No.

3, June 1980 25 9

BREAKDOWN OF COMPOSITE DIELECTRICS:

THE BARRIER EFFECT

Teruya Kouno
University of Tokyo
Tokyo

ABSTRACT
The breakdown voltage of a rod to plate is largely
affected by the insertion of an insulating gap
paper in the
gap and this is known as the barrier effect. The
cated characteristics of the breakdown were studied compli-
ex-
perimentally for dc, ac, lightning and switching impulse
voltages. The explanation of these effects
by considering the space charge distribution were
on
attempted
insulating paper. The trial to measure the spacethecharge
distribution was also made using an optical method.

INTRODUCTION ARRANGEMENT OF EXPERIMENT

The breakdown of composite dielectrics has many Fig. 1 shows the arrangement of electrodes and
features. One special characteristic is that the in-
space charge on the boundary surface of the dielectrics
sulating paper. As the high voltage electrode, a
needle electrode having a sharp point is
strongly influences the breakdown. and a rod electrode having a usually used
used for comparison. An semispherical end is also
It is known from the early work of Marx [1] and area larger than the
insulating paper having an
Roser [2] that the breakdown voltage plate electrode is placed a (mm)
of a rod to plate apart from the grounded plate electrode. The insu-
gap in air is remarkably affected by the insertion of lating paper is replaced by new one after each
an insulating paper in the gap; this is called the voltage application.
barrier effect.
When the barrier is an insulation board, for in- DC VOLTAGE APPLICATION
stance, and has itself some insulating strength, a
practical application arises which makes possible the When the dc
voltage is
trode, the charge carried applied
to the needle elec-
design of compact boards and switchgears. by the
when the barrier is a thin insulating paper,However,
it acts
corona discharge
with the same polarity as that of the
to control the space charge and it is thought of as moves toward the plate electrode and applied voltage,
useful model to consider the space charge effect on a to the insulating paper. The becomes attached
the breakdown of composite dielectrics, because buted on charge which is distri-
the paper makes the electric
this case the space charge distribution is simplein and on the needle side and stronger in the field weaker
it is not impossible to measure the space charge plate region.
distribution quantitatively. Fig. 2 is the typical result
The paper used is 0.15 mm thick given by Roser [2].
and the gap length d
In this paper the study on the barrier effect made is 10 cm. The change in the breakdown
recently in Japan, using a thin insulating paper, is
voltage de-
termined by the position of the paper has
discussed as a simple space charge model of composite plained as follows. When the paper is notbeen ex-
close to
dielectrics. either the needle or to the plate, the
a plate electrode paper acts as
having nearly the same potential as
the needle electrode. Thus the breakdown voltage
approximates that of a plate to
strates no polarity effect. Whenplate gap and demon-
the position of the
paper is close to either of the electrodes, the in-
fluence of the paper becomes small and
voltage approaches a value found withoutthepaper.
breakdown

0018-9367/80/0600-0259$00.75 0 1980 IEEE

260 IEEE Transactions on Electrical Insulation Vol. EI-15 No.3, June 1980

In the region (A) in Fig. 3, the breakdown voltage

rod or needle electrode of the negative polarity is lower than the value
without paper. Fig. 5 shows the results of the de-
tailed experiments in this region. When the break-
down voltages, determined for three kinds of gap
length d, are plotted against b, they fall on a
single curve. This suggests that the potential dif-
ference between the paper and the needle is not small
and this potential difference influences the break-
down, but it is difficult to explain quantitatively
why the breakdown voltage is lower than the value
without paper.

b
AC VOLTAGE APPLICATION
insulating paper
d With ac voltage, the breakdown voltage is thought
to coincide with the lower value of both polarities
of the dc voltage application. Therefore the explana-
tion for the breakdown mechanism is in principle the
a same as that for dc. A detailed study, however, has
yet to be carried out.

IMPULSE VOLTAGE APPLICATION

The study of the barrier effect for impulse voltage
has also been reported [1-3]. Only the results ob-
tained by Hidaka and Kouno [4] will be discussed.

Fig. 1: Arrangement of electrodes and insuZating

paper
kV
200
The latter case, when the paper is close to an elec-
trode, has recently been studied by Okumura and
Inuishi [3]. These results are shown in Fig. 3 with 160
the corresponding results obtained by Roser shown in
-

Fig. 2. The thickness of the paper is 0.08 mm and

the gap length is 3 cm in this case. The change of
the breakdown voltage for three different locations
of the paper with respect to the electrodes, (A), (B), 120
and (C) are discussed.
In the region (B), the charge is considered to be
distributed on a broad area of the paper and it acts
as a plate electrode, similarly to the results 80
described by Roser. However, when the paper approaches
the needle electrode, the equivalent diameter of the
distribution of charge becomes smaller and the break-
down voltage drops rapidly in the region (C). In
this region, it is expected that the breakdown voltage 40
will be higher if the diameter of the distributed
charge on the paper can be made larger.
Fig. 4 shows the results using a wetted paper. 0-
Because of the conductivity of the paper surface, the 0 20 40 60 80 100
charge was distributed more broadly and the breakdown a(mm)
voltage of the positive polarity in region C was higher
than the corresponding voltage for dry paper.
Fial.
L 2: TupiKcal
ap- results mngiven
--
thick
by Roser for dc voltage
-
using r o
a paper of 0. 15 mm thic1c.
Kouno: The Barrier Effect in Composite Dielectrics 261

kV l dh
80 - positive
*--)( negative
kV
60 Xd=30 mm
601 (C) Ad=20 mm
without paper I O d=16mm

(A) 40
40
(B)

L
/ without paper
zuil--
aru 7 20

plate to plate
U.-L
n 0
0 10 20 30
0I 10 20 30
a(mm) b(mm)

Fig. 3: Barrier effect for dc voZtage by the

Fig. 5: Breakdown voltage, when the paper is cZose
to the pZate.
insertion of 0.08 mm thick paper in a 3 cm gap

kV Fig. 6 shows the normalized breakdown voltages. The

80 breakdown voltage without paper is taken as unity for
each case. In this experiment, the paper is 0.05 mm
thick and the gap length is 10 cm. As the high vol-
tage electrode, a cylindrical rod of 6 diameter is
mm

used. In Fig. 6 "needle" means that the end of the

rod is formed into a cone having a sharp point and
"semisphere" is the rod, which has a hemispherical
end of the same radius of the rod.

When the positive switching impulse voltage (S. Imp.

in Fig. 6) of the waveform 360/6000 ps is applied,
there is no difference between the needle and the
semisphere. It is well known that the breakdown vol-
tage of a non-uniform gap is small especially for the
positive switching impulse; however, insertion of the
paper causes a large increase in the breakdown voltage.

In the case of a positive lightning impulse voltage

of the waveform 1.5/48 ps, the barrier effect is also
large for the needle, but small for the hemispherical
0 10 20 30 end.
a(mm)
Fig. 7 shows the typical oscillogram of the charge
Q, which is given by integrating the current through
the rod. In the four cases given in Fig. 6, only the
Fig. 4: Change of breakdown voltage by the use of a
case of a hemispherical end for lightning impulse
wetted paper
shows the character of Fig. 7(a) and the three cases
show Fig. 7(b).
262 IEEE Transactions on Electrical Insulation Vol. EI-15 No.3, June 1980

In Fig. 9, the experiment to measure the charge

S.Imp. Semisphere distribution is pictured. A crystal of ZnS is placed
on the back of the barrier. The refractive index of
this kind of crystal varies according to the applied
- 2.5 electric field. Therefore the electric field in the
crystal can be obtained by detecting the change of
the refractive index using an optical method.
Fig. 10 shows the output of the photo-multiplier,
which corresponds to the electric field in the crystal.
From this result, the electric field near the distri-
buted charge is thought to change linearly.
a1)
These experiments and calculations of breakdown
0 -2.0 voltages are now being made.
._
3

CONCLUSION
a)
_
(1) The study of the barrier effect using a thin in-
sulating paper is useful in analyzing the effect of
-

0
space charge on the breakdown mechanism.
CZ - 1.5 (2) When dc voltage is applied and the paper approaches
the needle, the breakdown voltage decreases rapidly.
This is because the equivalent radius of the space
0
(3
-
a
charge distribution on the paper becomes small. The
experiment using a wetted paper supports the interpre-
sv
m
tation.

- 1.0

0 10 20 30 40 50 60 70 80 90 100 (a)
a(mm) breakdown

tI
Fig. 6: Barrier effect for impulse voZtage of
positive polarity Q
A
-
)__- non breakdown

0.
t~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In the former case, the breakdown usually occurs at
the second rise of the charge and only the first rise
of the charge is observed in the case of non-breakdown. (b).
In the latter case, the breakdown occurs after multiple breakdown
small stepwise increases of the charge. It can be
said from Fig. 6 and 7 (for lightning impulse), with t
the exception of the semispherical end, that the
space charge is gradually supplied on the paper and
makes the electric field near the high voltage elec-
Q iJ_ -
non breakdown
trode weak, which acts to increase the breakdown vol-
tage.
Fia. 8 shows the value of the total charge observed
in the case of non-breakdown, when the positive
switching impulse voltage was applied to the needle.
If it can be assumed that all of the measured charge
is attached to the paper, and if the distribution of
the charge can be given, the electric field can be
calculated and the breakdown voltage can also be Fig. 7: TypicaZ osciZZogram of measured charge Q
deduced.
Kouno: The Barrier Effect in Composite Dielectrics 26 3

(3) When the switching impulse voltage is applied, or

a lightning impulse voltage is applied to an electrode
having a sharp point, the space charge is gradually mV a 70mm
supplied to the paper and the breakdown voltage shows 50 -t x
a large increase.
a
(4) An experiment to measure the distribution density 40
of charge on the paper using an optical method is -s
._
described. V=161kV
30
-

0
20 \I X,ll,V 123kV

4-
a=
10

1.0 F- 0
0 40 80 120mm
a 50mm x

0.8 h a =70mm Fig. 10: Change of eZectric fieZd by the position

x on the back of the barrier

a 10mm
/
0.6 h
/ a=90mm

0.4 REFERENCES
[1] E. Marx, "der Durchschlag der Luft in unhomogenen
elektrischen Felde bei verschieden Spannungsarten,"
ETZ-A, Vol. 51, p. 1161, August 1930.
0.2 _
[2] H. Roser, "Schirme zu Erhohung der Durchschlag-
spannung in Luft," ETZ-B Vol. 53, p. 411,
April 1932.
0
50 100 150 kV
[3] N.
Okumura Inuishi, "Barrier Effect of

Insulating Paper in Non-uniform Electric Field,

applied voltage JIEE of Japan, Vol. 96-3, Vol. 96-3, p. 127,
March 1976.
Fig. 8: Measured totaZ charge by needZe elZe strode [4] K. Hidaka and T. Kouno, "Space Charge and Barrier
for switching impulse at non-breakdown. Effect for Positive Impulse Voltage," JIEE of
Japan, Vol. 99-7, p. 313, July 1979.

Ne-He barrier U ZnS photc This paper was presented at the US/Japan Seminar on
laser / multi Lplier EZectricaZ Conduction and Breakdown in Dielectrics,
Gainesville Fla., 29 October - 2 November 1979.

polarizer C detector Manuscript was received 2 November 1979.

Fig. 9: Arrangement to measure the charge distributed

on the paper by optical method.