Advanced Methods in Materials Processing Defects

M. Predeleanuand P. Gilormini(Editors)
9 1997Elsevier ScienceB.V. All rights reserved. 143

Processing o f zinc oxide varistors: sources o f defects and possible measures for
their elimination

A N M Karim, S Begum and M S J Hashmi

School of Mechanical and Manufacturing Engineering

Dublin City University, Dublin 9, IRELAND

The processing of ZnO varistors primarily follows the conventional ceramic route. A
number of dopants are added to zinc oxide powder to prepare the constituent material and some
additional finishing operations are performed fgr securing the varistor action. Thus there involved
a large number of processing operations and there are naturally multiple sources from which
defects can originate. To qualify a varistor performance several parameters are commonly
investigated of which energy absorption capability of the device is a significant one. In this
respect flaws and defects developed in the varistor body during the processing have detrimental
effects and can considerably restrict the energy absorption capability. This paper briefly outlines
the probable sources of defects of the metal-oxide varistor device and their possible solutions
mainly with sintering orientation, passivation thickness and electrode area.

1. I N T R O D U C T I O N

A zinc oxide varistor is an electronic ceramic material which possesses a non-linear

current (I) - voltage (V) characteristics with a symmetrical sharp breakdown ~1 - similar to that of
a zener diode. But unlike a diode, a varistor can limit overvoltages equally in both the polarities,
thus giving rise to an I-V characteristics which is analogous to the two back-to-back diodes. This
has enabled it to provide an excellent transient suppression performance.
The state-of-the-art of the metal-oxide varistor is represented by a broad range of
products manufactured 3 to meet the need of the present day transient voltage suppression. The
products cover from arresters for power systems to the low power and low energy application
such as integrated circuits, automotive systems and other modem electrical and electronic
circuits.
There are several critical application parameters some of which are associated with the
various regions of the I-V curve. These parameters serve functions in the design and operation of
a surge protector. The most desirable device should have a high value of non-linear coefficient or
a low value of protective level, a low value of leakage current, a long varistor life and high energy
absorption capability.
Processing parameters of varistors, though basically follow the conventional ceramic
fabrication route, vary to a certain extent depending upon the configuration of the device. The
144

present study is concerned with the cylindrical varistor discs applied for arrester application. The
manufacturing route of the varistor disc is presented by the block diagram in Figure 1.

I Zincoxide I I Additiveoxido I
I I
I
Mixing and
Milling
I
Spray drying
I
Calcining
I
Milling
I
Spray drying
I
Compaction
I
Sintering
I
Lapping
I
Ultrasonic
cleaning
I
Electroding
!
Passivation
I
Testing

Figure 1. The fabrication procedure of metal-oxide arrester block

The application of zinc oxide varistor is selected on the basis of some critical parameters
such as non-linear coefficient, nominal voltage, leakage current, energy absorption capability etc.
The material and processing parameters such as the green and the fired body, homogeneity, grain
size, porosity, varistor chemistry and sintering parameters are identified to affect the energy
absorption capability remarkably 49. The life of varistors is largely dependent on the leakage
current and the energy absorption capability
 145

1.1 E n e r g y A b s o r p t i o n C a p a b i l i t y

Commercial arresters are rated conservatively according to the lowest level of energy at
which failures initiate. The failures occur randomly in a lot starting from a low level of energy
sustaining up to a high level. The minimum level of energy at which failure starts restricts the
enhanced rating of the arrester block. A typical failure distributionS~ is presented below.

100 "

80- f
60 "

II

I
E~ 4 0 -
rj

20-
.e,----- _____._------

250 300 350 400 450

Energy (J.cnl 3)
Figure 2. A typical failure pattern of arrester block in energy test

It is obvious that few earlier failures occurring at the lower level of energy absorption
capability considerably restrict the rating of the varistor. Prevention of the initial failures would
allow enhanced rating of the discs. In that case it will be possible to apply the arrester blocks for
more demanding applications or their volume could be reduced proportionately. In other words
the same lot will improve the system functional reliability.

2. S I N T E R I N G ORIENTATION AND ORIGIN OF FAILURE

The orientation of the cylindrical arrester discs or blocks during sintering is shown in
Figure 3. It is observed that the face of a disc remaining in contact with the liner material (bottom
face) is not physically as good as the top face. As a result more material needs to be ground off
from this face. But even then this face does not become functionally as good as the top face. In
the experiment the origin of failure was tracked and it was found that the distribution of failure
origin was not same for both the faces. A remarkably higher number of failures were found to
146

Top face

Bottom face
Liner , , ~
J
I

Figure 3. Sintering orientation and identification of different section of an arrester block

originate from the bottom face during the test for the energy absorption capability. The
percentage share of failures according to the destruction mark on the face of a disk is shown in
Figure 4.

Figure 4. Frequency of failure origin as observed in the test for energy absorption capability

This uneven distribution of failure has been correlated by the variation of physical
properties of varistors as affected by the sintering orientation. Tensile strength and density
gradient of the sintered discs were evaluated.
 147

2.1 Sintering Orientation and Tensile Strength

The influence of sintering orientation on the mechanical strength of the arrester blocks
was evaluated by the diametral compression test. Measurement was conducted on five arrester
blocks having a nominal diameter of 32 mm and a height of 34 mm. They were sectioned
perpendicular to the axis by a diamond cutter into three equal disc shaped pieces. Load was
applied on the disc specimen by an Instron machine with a cross-head speed of 1 mm/min. The
breaking load was recorded and the tensile strength of the specimen was calculated. In Figure 5
the tensile strength for the three sections is plotted.

50-

~0
40-
t
30-
t~

o~..q
t~

tD
t--, 20-

10- 1" !

Bottom Middle Top

Section of arrester disc

Figure 5 Variation in tensile strength due to sintering orientation

It is clear that the bottom section, the face which remains in contact with the liner material
during sintering is relatively weaker. The strength of this section is about 40 percent less than that
observed in the case of the middle or the top section. The top and the middle sections appear to
have similar strength. It is clear that at least up to two-thirds of the height of the 34 mm tall disc
is not affected by the contact with the liner material. Though it may not be possible from this
investigation to quantify how far the effect of contact reaches, it is clear that the bottom section is
weaker. This lower mechanical strength of the bottom part of the disc can only be attributed to
the influence of sintering orientation.

2.2 Sintering Orientation and Density Gradient

To further investigate the effect of the sintering orientation, the density gradient of a fired
disc was evaluated. A piece cut from an arrester block was sectioned as shown in Figure 6. The
148

small parallelepiped shaped specimens taken from the eighteen grid points were measured for
density.
I

/ " ..... ~ ...... !

h i
9 ...... :. . . . . . ; ......

I
I

I r
i

Figure 6: Sample preparation for measuring the density gradient

Each specimen was polished with the fine grinding paper to make the surface sufficiently
smooth. These were then washed in deionized water using ultrasonic washing bath so that all the
loose debris were removed from the specimens. They were then dried in an oven at 125 ~ for
more than one hour to expel the trapped water or moisture content completely from the open
pores. Weighing was performed in a high precision laboratory type balance by adopting
Archimedes' principle.
It should be mentioned here that the specimens were wrapped with water-tight masking
tape to prevent water from being entered into the open pores during weighing in water. Care was
taken so that there could not be any air bubble trapped inside the wrapping. In Table 1 the density
of the specimens are given according to the grid points in terms of radius and height. With the
grid point densities the contour lines indicating the constant-density were drawn as shown in
Figure 7.

Table 1 Density (gm/cc) of varistor material at different grid points

Radial Grid number along the height (h)

grid(r) 3 4 5 6
1 5.36 5.55 5.49 5.55 5.57 5.53
2 5.56 5.52 5.57 5.56 5.58 5.61
3 5.59 5.59 5.58 5.62 5.59 5.59

2.3 Alternative Sintering Approach

This investigation includes the scope of alternative sintering configuration. The evaluation
is based on the frequency of regrinding and the energy absorption capability.The objective of this
study was to evaluate the feasibility of alternative liner support and sintering orientation of
arrester blocks. The sintered discs were characterized to evaluate the effect of the new method.
 149

~/ )I~~.. r~~
5.5 5.585
5.60

~ \
5.58

5.50 "

I
, I
Radius

Figure 7: Density gradient as a function of sintering orientation

In addition, to enhance the process capability in terms of the performance the foreseeable
advantages are (i) reducing the problem arising from regrinding (ii) minimizing the level of
bismuth contamination from the liner material due to contact (iii) increasing the scope of
repetitive use of the liner material (iv) lowering the allowance of block height for regrinding and
(v) better geometry of the disc.
The orientation of a varistor disc in conventional sintering operation has already been
shown in Figure 3. The modified arrangement for this experiment is presented in Figure 8. It
should be mentioned here that for the horizontal sintering, the Vee-groove supports were made
from the fired arrester discs.
To prevent sticking of the discs during the sintering operation the supports were covered
by spreading spinel powder. Spinel is known to have an inhibiting effect on the grain growth ~
and its selection was attributed to keep the dimensional elongation along the contact to a
minimum. But the dry powder poured on the surface did not stick to it due to the inclination of
the surface. To ensure proper adhesion of the dry liner power with the inclined surface, it was
necessary to lightly wet the supports by spraying water.
The Vee-groove support facilitates to keep the edges free from any physical contact
during the sintering process. Improved faces with uniform edges thus achieved were helpful in
150

reducing the frequency of regrinding operation. Energy absorption capability was also found to
be consistent having initial failure at a higher level of energy

Green
disc-
" - ......... Fired , . t

:: :
iI / ~i Liner_ I

Sap.~er

I
(a) Front view (b) Side view
Figure 8" Horizontal or Vee-groove
.
sintering orientation

3. P A S S I V A T I O N THICKNESS

Arrester blocks are passivated with glass material to prevent dielectric breakdown when
subjected to stresses due to transient electrical surges. The material property and thickness of the
coating are very critical in this respect. In this study the effect of glass thickness was evaluated
with standard glass material and glassing procedure. Passivation of an arrester block is very
critical to prevent the failure by flashover. Thinner passivation leads to more flashover while the
thicker passivation results in failure due to problem with heat transfer, usually accompanied by
pinholes. Experiment was conducted on 32 mm diameter discs by depositing glass coating of 110,
220 and 340 Ixm.
The glass thickness plays an important role in varistor performance. The influence of glass
thickness on the varistor performance was assessed by evaluating the energy absorption
capability. The plot of the energy absorption capability of the varistors with different glass
thicknesses is presented in Figure 9 with the mean energy along with standard error. The standard
amount of glass taken as the minimum level in the experiment had shown inferior performance
and the varistors started to fail at a very low energy level. Nevertheless, the thickest coating did
not exhibit the best performance.
The coating of glass acts as an insulator and restricts heat transfer. The effect is more
obvious in the second and the third shot in the energy test. The heavier coating acts as higher
insulator and does not dissipate heat as effectively as the lower thickness. Thus, the high
temperature in the ceramics makes it more vulnerable to failure. This feature is supported by the
fact that the discs with the thicker coating failed through the ceramic and most of the discs failed
by electrical puncture (number of pinhole (PH) increased substantially) as presented in Table 2.
Here interface indicates the zone where ceramic and glass coating meet and FO indicates
flashover.
 151

550
500
450
400
t.)
350
300
I= 250
200
150
100 i I i I i

110 220 340

(31ass thickness 0xm)

Figure 9 Effect of passivation thickness on the energy absorption capability

It was also noticed that for the best cell, 47% of the discs survived after 400 J.cm 3. The
mode of failure with various glass thickness was also categorised. It is clear from the figure that
the increase of glass thickness has shifted the failure. More than 85% of the varistors failed
through the ceramic coated with the thicker passivation coating compared to 27% in the case of
the thinner coating.

Table 2 Failure mode with different glass coatings during "strength to destruction" test

Glass Sample Failure mode

thickness size Inter- PH FO Rupture Interface+ Interface+ PH+FO
(~tm) face FO PH
110 15 8 3 - - 3 - 1

220 15 3 4 - 4 3 - 1

340 14 - 8 - 2 2 - 2

4. E F F E C T OF MARGIN ON ELECTRODE

The margin on electrode as shown in Figure 10 appears to be advantageous in one respect

but harmful in the other. This experiment was undertaken to identify the effect of margin on the
energy absorption capability.
The experiment was conducted by covering the top and the bottom face of an arrester
block with aluminium deposited by arc-spray method. Three options were selected (i) control
leaving a margin on electrode on both the faces (ii) one face electroded with margin and the other
face fully electroded leaving no margin and (iii) both the faces were fully electroded. The three
categories of arrester discs were identified by CONT, OFFE and BFFE respectively.
152

(a) Top view (b) Side view

Figure 10 Electrode with margin as shown on the face of an arrester disc

4.1 Evaluation of Energy Absorption Capability

The energy absorption capability was evaluated by the standard testing procedure. The
performance of the three cells is demonstrated in Figure 11.

100% --
CONT(F) --o- CONT(S)
- - ~ OFFE (F) -'- OFFE (S)
80% - BFFE (V) "- BFFE (S)
o

60%"
..

40%--
r~
20% -

0%-
200 250 300 350 400 450 500
Energy absorption capability (J.cm "3)
Figure 11 Effect of margin on electrode on the Energy absorption capability

The sample size was 17, 16 and 15 respectively for CONT, OFFE and BFFE. It should be
mentioned here that the test for energy was initiated with a charging voltage of 24 KV which was
equivalent to a level of 140 J.cm "3. With the increment of 1.2 KV for every subsequent cycle
testing was continued up to 49.9 KV, maximum limit of the generator. But out of 48 discs only
18 failed while the remaining 30 discs survived. Energy absorption capability of the survivor
discs was computed on the basis of the data obtained at the last cycle of test at the charging
voltage of 49.9 KV of the generator. In the legend the letters 'F' and 'S' in parenthesis stand
respectively for the failed and survived disc. From the percentage of discs survived after the
maximum possible energy injection for the three cells it is apparent that the full face electrode is
conducive to energy absorption capability. However, unsatisfactory results with full face
 153

electrode may not be unlikely when the quality of passivation and its thickness are not properly
maintained.

5. C O N C L U S I O N S

Defects generated in arrester blocks due to their sintering orientation have significant
effect than normally perceived. The adverse effects arising from the bottom face cannot be
avoided by simply removing more materials through grinding. Investigation revealed that there
was a considerable density gradient in the sintered body which is influenced by the sintering
orientation resulting in the lowest density at the central part of the bottom face. Horizontal
sintering on the Vee-groove support was found to be advantageous in minimizing the density
gradient and delaying the earlier failures in energy test.
Passivation thickness was found to greatly influence the energy absorption capability of
varistors. Neither too thin nor too thick passivation was found to be favourable for superior
varistor performance. Thus there is scope of optimizing the amount of glass to be deposited on
the side of the arrester block for passivation.
Full-face electrode was found to improve the energy absorption capability. However, if
the passivation is not proper in terms of the quality and thickness this method may result in poor
performance. In such a condition full electrode may result in a preferential path of current flow
causing earlier failure and thus lead to a wrong perception about the effect of full-face electrode.

References

1. M Matsuoka, "Non-ohmic Properties of Zinc Oxide Ceramics", Jpn. J. Appl. Phys., Vol. 10,
No. 6, pp 736-46, 1971
Tapan K. Gupta, "Application of Zinc Oxide Varistors", J. Am. Ceram. Soc. 73(7), pp 1817-
1840, 1990
3. "Transient Voltage Suppression Devices 1995", a Databook published by Harris
Semiconductor, Melbourne, FL 32902, USA
4. K. Eda, A. Iga and M. Matsuoka, "Current Creep in Non-ohmic ZnO Ceramics " Jpn. J.
Appl. Phys., 18 [5], pp. 997-98, 1979.
5. E. Saksaug, J. Burke and J. Kresge, "IEEE Trans. on Powder Del. Vol.4 No. 4, Oct., 1989.
6. W.N. Lawless and T. K. Gupta, "Thermal Properties of Pure and Varistor Zinc Oxide at
Low Temperature", J. Appl. Phys., 60 [2], pp. 607-11, 1986.
7. J. Ozawa, et al, IEEE Transac. on Powder App. and Sys.", VoI.PAS102,No. 5, May, 1993.
8. N. Amiji, et al., Advanced Cearmic Materials, 113], pp. 232-36, 1986.
9. K. Eda "Destruction Mechanism of ZnO Varistors Due to High Currents", J. Appl. Phys.,
56(10), 15 Nov., pp. 2948-55, 1984.
10. A N M Karim, "Effect of Compaction Parameters and Sintering Configurations on the
Performance of ZnO Varistor" PhD thesis,Dublin City University, Ireland, 1996.
11. E. Olsson, G. Dunlop and R. Osterland, "Development of Functional Microstructure during
Sintering of a ZnO Varistor Material", J. Am. Ceram. Soc., 76 [ 1], pp-65-71, (1993)
12. S Begum, "Powder Processing Parameters and Their Influence on the Electrical
Performance of ZnO Varistor", PhD thesis, Dublin City University, Ireland, 1996.