Materials Science and Engineering B106 (2004) 132–140
Structure–property relationship in high-tension ceramic
insulator fired at high temperature
Rashed Adnan Islam a , Y.C. Chan a,∗ , Md. Fakhrul Islam b
b
a Department of Electronic Engineering, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon Tong, Hong Kong
Department of Materials and Metallurgical Engineering, Bangladesh University of Engineering and Technology, Dhaka 1000, Bangladesh
Received 7 March 2003; accepted 2 September 2003
Abstract
Ceramic insulators are widely used in microelectronic devices. In this paper, the mechanical and electrical properties of porcelain ceramic
insulator fired at 1350 ◦ C have been investigated along with microstructural characterization using scanning electron microscopy (SEM) in
order to understand the structure–property relationship of ceramic insulator. The bending and the dielectric strength were measured on various
samples fired at 1350 ◦ C. The bending strength (757.3 kg/cm2 ) and the dielectric strength (28.36 kV/mm) was found short of the desired value.
The microstructural features developed clearly describe why the dielectric strength and the bending strength are not up to the mark. EDAX
analysis, X-ray fluorescence (XRF) and X-ray diffractometry (XRD) techniques were also done to support the results. XRD pattern shows
70% mullite and 20% quartz peak intensity and the XRF results shows 22.64% Al2 O3 that indicates low mullite formation and hence it is
confirmed that it is mullite, the crystalline phase, which contribute together with quartz particle to the dielectric and mechanical strength.
SEM image shows large number of microcracks that also hinder the high electrical and mechanical properties of porcelain ceramic insulator.
© 2003 Published by Elsevier B.V.
Keywords: Bending strength; Dielectric strength; Mullite; Glassy phase; Microcracks
1. Introduction
Insulators are materials, which prevent or regulate current flow in electrical circuits by being inserted as a barrier
between conductors. The properties required being an insulator is high resistivity, high dielectric strength, a low
loss factor, good mechanical properties, dissipation of heat
and protection of conductors from severe environment, like
humidity and corrosiveness. Ceramics are widely used as
insulating materials. The advantage of ceramic insulators
which frequently indicate their use, are superior electrical
properties, absence of creep or deformation under stress at
room temperature and greater resistance to environmental
changes. One of the great advantages of ceramics as insulators is the fact that they are not sensitive to the minor
changes in composition, fabrication, techniques, and firing temperature. Ceramic insulators are widely used in the
microelectronic devices as well as in power transmission
lines. For the electrical insulation application the properties
∗
Corresponding author. Tel.: +852-2788-7130; fax: +852-2788-7579.
E-mail address: eeycchan@cityu.edu.hk (Y.C. Chan).
0921-5107/$ – see front matter © 2003 Published by Elsevier B.V.
doi:10.1016/j.mseb.2003.09.005
most concerned are the dielectric and mechanical strength.
Dielectric strength (represented by kilovolt per millimeter)
measures the ability to withstand large field strength without
electrical breakdown. For high-tension electrical insulation
the dielectric strength has to be above 30 kV/mm [1].
The microstructure of porcelain electrical insulator body
shows mullite needle in both glassy phase and unresolved
clay matrix (cannot be specified by an optical microscope
and need SEM equipped with EDAX to resolve) with some
undissolved quartz particle. The thermal behavior of silica
and alumina shows that the mullite (3Al2 O3 2SiO2 ) phase
starts to form at 1100 ◦ C and above this temperature mullite crystals continue to grow and the needles appear [2].
It has some specific properties like high creep resistance,
low thermal expansion and good thermal and chemical stability which makes it desirable for high tension structural
insulating material [3]. Slow firing and slow cooling favour
the formation of mullite crystals and the size of the mullite
crystals increase with firing [4]. The glassy phase and the
quartz particles are other two important features of the microstructure. But where dielectric performance such as high
dielectric strength is required glassy phase and mobile ion
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
133
Table 1
Composition of different raw materials used in the insulator body
SiO2
TiO2
Al2 O3
Fe2 O3
MgO
CaO
Na2 O
K2 O
Composition
China clay (plastic)
China clay (body)
Ball clay (white)
Bejoypur clay (local)
Ball clay (black)
Feldspar
Quartz
50.18
0.014
33.34
1.88
1.05
0.94
0.81
2.58
11
51.86
0.50
34.72
0.57
0.07
0.07
0.041
0.18
8
54.43
1.84
30.97
1.37
0.34
0.17
0.17
0.81
8
72.09
1.02
18.69
1.01
0.14
0.10
0.11
0.61
32
61.27
1.41
24.85
0.17
0.26
0.65
0.25
0.44
–
74.98
0.08
14.37
0.83
0.08
0.68
3.54
4.91
21
98.17
–
–
0.5
–
–
0.34
0.45
20
content must be minimized [5]. At higher temperatures an
increasing amount of liquid is formed which at equilibrium
would be associated with mullite as a solid phase. The general equilibrium conditions do not change at temperatures
above about 1200 ◦ C so that long firing times at this temperature give results that are very similar to shorter times at
higher temperatures [6]. The initial mix is composed of relatively large quartz and feldspar grains in a fine-grained clay
matrix [7]. Fine mullite needles appear at about 1000 ◦ C but
cannot be resolved with an optical microscope until temperatures of at least 1250 ◦ C are reached. With further increases
of temperature mullite crystals continue to grow [8]. The
solution rim of high-silica glass around each quartz grain increases in amount at higher temperatures. By 1350 ◦ C grains
smaller than 20 ␮m are completely dissolved [9].
The objective of this research is to find out the
structure–property relationship in high-tension ceramic insulators. It is aimed to determine the phases, which are
detrimental to high dielectric and mechanical strength.
2. Experimental
The processing of insulating body was done in the
Bangladesh Insulator and Sanitary ware Factory (BISF)
Ltd. The composition of the insulating body is shown in
Table 1. The flow sheet for the insulating body preparation
is shown in Fig. 1. Two types of samples were prepared,
one for dielectric testing and another for mechanical testing.
The samples are shown in Fig. 2. The dielectric strength
was measured using high voltage dielectric tester (Model
No. DTS—100 D) manufactured by High Voltage Inc.,
having a maximum output voltage of 100 kV. The thickness of the specimen was kept below 3 mm to avoid the
maximum probability of pores and cracks in the measuring
regions. The breakdown voltage of a spark gap between
two metal spheres was used as a measure of voltages up
to the highest encountered in high voltage testing. The test
was followed by the ASTM standard no D3755 [10]. The
bending strength was measured by three-point bending. The
load in Newton at which the insulating body breaks was
converted to kg/cm2 . The result of both bending and the dielectric strength is reported on Tables 2 and 3. The samples
Fig. 1. Flow chart of the insulator body processing.
134
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
Table 3
Dielectric strength of insulating materials fired at 1350 ◦ C
Specimen
no.
Applied
voltage
(kV)
Specimen
thickness
(mm)
Dielectric
strength
(kV/mm)
Average
(kV/mm)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
85
70
75
70
75
82
71
70
65
82
73
70
75
76
2.46
2.54
2.39
2.74
2.86
2.97
2.41
2.65
2.49
2.82
2.28
2.6
2.94
2.61
34.55
27.55
31.38
25.54
26.22
27.45
29.46
26.41
26.1
28.92
32.01
26.92
25.51
29.11
28.36
acterized by a set of line positions 2θ and a set of relative
line intensities, I. The specimens for SEM were prepared
according to the optical ceramography and were coated
by Agar SEM carbon coater. The SEM was done by the
Philips XL30 FEG (Field Emission Gun) Scanning Electron
Microscope. Semi quantitative analysis was performed by
EDAX on four spots shown in Fig. 4. X-ray fluorescence
(XRF) was done on the grounded samples of the fired body.
Fig. 2. Test specimen for (a) dielectric strength test and (b) bending
strength test.
are then prepared for X-ray diffractometry (XRD) and scanning electron microscopy (SEM) and EDAX analysis. For
XRD the fired bodies are ground to −300 mesh and then
examined according to the Hanwalt method. X-ray patterns
were recorded in a JEOL JDX-8P X-ray diffractometer. All
the diffraction patterns were recorded under identical conditions using a beam current of 20 mA at 30 kV. Scanning
speed used in all cases was 1◦ /min. The powder was charTable 2
Bending strength data of different samples fired at 1350 ◦ C
Sample
no.
Firing
temperature
(◦ C)
Load
Diameter
reading (cm)
(N)
Bending
strength
(kg/cm2 )
Mean bending
strength
(kg/cm2 )
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1350
310
275
285
292
311
324
325
311
285
302
307
335
321
341
745
693
689
704
770
793
807
768
701
741
749
844
776
822
757.3
1.026
1.01
1.024
1.025
1.016
1.02
1.015
1.017
1.018
1.019
1.021
1.01
1.024
1.025
3. Results and discussion
The target of our research was to attain the dielectric
strength of 30 kV/mm. The results of dielectric strength measured on various samples are shown in Table 3. Although
some samples attain 30 kV/mm, the average (28.36 kV/mm)
does not meet the target goal. Again the bending strength
was expected to be very high but it reaches the average figure of 757.3 kg/cm2 reported in Table 2. The SEM analysis
(from Figs. 3–6) shows very small size cracks, quartz, mullite needle and unresolved clay matrix (cannot be detected by
the optical microscope and needs EDAX analysis to detect
it). EDAX spectrum is shown from Figs. 7–10. The compositions of elements in different spots in atomic percentage
are given in Table 4. The XRD pattern is shown in Fig. 11
and the analysis is also shown in Table 5. The XRF result
is shown in the Table 6.
SEM micrographs show quartz, mullite needles, glassy
rim and unresolved clay matrix. At the boundary of quartz
Table 4
Atomic percentage of the elements found from EDAX analysis
Elements
Spot 1
Spot 2
Spot 3
Spot 4
Si
Al
O
K
54.66
18.82
26.50
Negligible
48.70
21.61
27.08
2.59
61.96
8.98
22.09
4.03
75.15
3.13
21.71
–
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
135
Fig. 3. SEM image of porcelain ceramic insulator showing quartz grains, glass rim and clay matrix.
small rim of glassy phase is observed which is due to the
partial melting of quartz. The particles observed are in the
size of 1–5 ␮m or even 10-␮m quartz particles can be seen.
Undissolved quartz particle is not coherent with the unresolved clay matrix.
SEM images of the porcelain show mullite needles and
EDAX analysis on that needles support the exact chemical composition of the mullite. Images indicate mullite’s
preferential orientation on the surface of clay or kaolinite matrix. Probably the mullite in the clay matrix is the
seed for the crystallisation of the mullite needles. The
mullite needles observed in the clay matrix are generally
termed as primary mullite. The morphology of the mullite
needle is like acicular crystal. The mullite needles seem
to be interlocked, which actually act as the strength increaser. When mullite crystal forms from the clay matrix
it increases the volume by 10%. The increase in volume
heals up any cavity or porosity or any cracks formed due
to shrinkage and other thermal expansion. The SEM image clearly describes this nature. Where there is a low
mullite content or no mullite the cracks become more
prominent.
Fig. 4. SEM image of porcelain ceramic insulator showing a quartz grain and a glassy rim surrounded by micro cracks and mullite needles in a clay
matrix. Points 1–4 indicates the EDAX spots.
136
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
Fig. 5. SEM image of porcelain ceramic insulator showing acicular shapes of mullite needles and some interconnected cracks in clay matrix.
Table 5
XRD analysis of grounded samples fired at 1350 ◦ C
Diffraction
angle (2θ)
d-Spacing
I/I0
Diffraction
planes, h k l
Phases
present
10.3
11.8
13.6
14.8
17.5
18.2
22.2
26.3
29
3.95
3.45
3.001
2.75
2.33
2.24
1.84
1.56
1.42
30
70
20
20
20
20
15
15
10
200
210
112
220
102
040
311
211
250
Mullite
Mullite
Clay
Mullite
Quartz
Clay
Mullite
Quartz
Mullite
The matrix surrounding the quartz particle and beneath
the mullite needles is supposed to be the unresolved clay
matrix. The glass rim around the quartz particle shown (in
Fig. 4) is due to the partial dissolution of quartz. The reason
behind this is above 1200 ◦ C, the quartz particle starts to
fuse and a rim of glass was formed surrounding the quartz
particle.
Each region was inspected by the EDAX analysis. The
point 1, which was taken in the needle region, is shown in
Fig. 8 shows aluminium content almost three times of the
silicon content. The mullite has a stoichiometric formula
of Al6 Si2 O13 , where the ratio of aluminium-to-silicon is 3.
From this point of view it is clearly understood that the needle shape areas are mullite. Another EDAX result, which
is chosen in the large particle (point 4), shows only silicon
and oxygen. These regions are quartz particle. The regions
surrounding the quartz particle (shown in point 3) show almost all silicon and oxygen with small amount of Na, K and
Al. This is the glassy phase. The matrix shows considerable
amount of aluminium and silicon (shown in point 2). These
are the regions of unresolved clay matrix. The XRF results
show a large amount (70%) of SiO2 and a very small amount
(22.64%) of Al2 O3 (Table 6). The XRD pattern shows the
mullite, clay and quartz peaks. But the intensity of the mullite is the largest among all. The result of the XRD pattern
confirms the phases present in the ceramic insulator.
The fired body also shows 70% silica and 22.64% alumina in sample fired at BISF at 1350 ◦ C according to the
XRF result. Weight ratio corresponding to the stoichiometric formula of mullite shows that there has to be 2.55 times
alumina than silica. So from the results it can be estimated
that there is a very low mullite formed in the fired body.
The bending strength and the dielectric strength of
ceramic insulator at 1350 ◦ C is found 757 kg/cm2 and
28.36 kV/mm respectively, which is not good for ceramic
insulator in service life. There are various reasons behind
these low strengths. Actually the strength both bending and
the dielectric, in the ceramic insulator is given by the mullite phase and the quartz particle. The undissolved quartz
strengthens the rigid skeleton of the ceramic body as well
as disconnects the conductivity. The interlocking system of
Table 6
XRF results of various raw materials and the fired body
Samples
SiO2
TiO2
Al2 O3
Fe2 O3
MnO
MgO
CaO
Na2 O
K2 O
P2 O5
Total
LOI
BISF burned body
70.03
0.81
22.64
1.38
0.02
0.32
0.54
1.17
2.77
0.04
100
0.05
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
137
Fig. 6. SEM image of porcelain ceramic insulator showing wide regions of mullite crystals in a matrix of clay.
the mullite needles is another reason for strengthening the
ceramic insulator. The phase, which reduces the bending
and dielectric strength of ceramic insulator, is the presence
of glassy phase. The glassy phase will have a harmful effect
if the amount is quite high. When there is a low amount of
glassy phase content it is quite advantageous to the ceramic
insulator as it only fills up the pores. But when the amount
gets higher it only just increases the volume and decreases
the density and that’s why decreases the bending and dielectric strength. Increase in glassy phase increases the liquid
viscosity and helps to maintain the shape during firing. On
the other hand mullite, which is a crystalline phase, have
a vital role on mechanical and electrical strength. As the
matrix of the ceramic sample is either clay matrix or the
glassy phase the needle shape mullite maintain the stress
level in a higher order as just in a composite matrix. But
here in our sample the SEM micrograph (shown in Figs. 3
and 6) and the XRF results signifies the lower amount of
mullite formation and that’s why there is a significant effect
on bending and dielectric strength. Again with increasing
the temperature the partial dissolution of quartz phase occurs which starts at 1200 ◦ C. Increased glassy phase and
decreasing the quartz content tends to decrease the bending
and dielectric strength of the porcelain ceramic insulator.
At 1350 ◦ C, there will be a considerable amount of glassy
phase. When there is a large amount of glassy phase present
in the structure the mobile ions such as Na+ , K+ , Al3+ and
Li+ finds an easy path to move and hence increases the
Fig. 7. EDAX analysis of glassy phase showing high amount of silica and other oxides (spot 3 in Fig. 4).
138
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
Fig. 8. EDAX analysis of mullite needles showing stoichiometric amount of alumina and silica (spot 1 in Fig. 4).
Fig. 9. EDAX analysis of clay matrix showing the proportionate amount of silica and alumina (spot 2 in Fig. 4).
Fig. 10. EDAX analysis of quartz showing almost all the silica (spot 4 in Fig. 4).
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
139
Fig. 11. XRD Pattern of powdered fired body showing the intensities of mullite, quartz and clay.
conductivity. On the other hand, the strength can be greatly
increased by undissolved quartz. If there is a large amount
of undissolved quartz then it creates an obstacle in the way
of conductive path that is glassy phase in which mobile ion
can easily move. If it were possible to retain the undissolved
quartz with higher amount of mullite then the strength would
be greatly increased.
Another reason for low dielectric and bending strength is
the presence of microcracks in the ceramic insulator. The
cracks are observed mainly near the quartz particle and those
regions where the crystalline phase or mullite phase is less or
absent. The cracks are peripheral around the quartz grain. At
the crack tip some expansion cracks are also to be observed.
Cracks can occur for various reasons. Cooling through the
quartz inversion temperature of 573 ◦ C results in a quartz
particle volume decrease of 2% which can produce sufficient
strain to cause cracking of the glassy matrix and even in a
rare case quartz itself which is observed in Fig. 4. Transgranular crack is observed in the middle of the quartz particle, which is due the differential thermal contraction in the
quartz inversion point. The cracking severity largely depends
on cooling rate. Slow cooling rate through the transformation zone will produce small strain.
The nature of cracks in porcelain body is dependent on
the expansion coefficient of the matrix and the particle. Circumferential cracking results due to the particles contracting
more than the matrix. This is true for quartz particle in the
feldspathic glass of the porcelain body matrix. The stress
generation and associated cracking due to the presence of
quartz particles tend to be severe because of rapid displacive
phase transformation of quartz during cooling. The evidence
of having these types of cracks is clearly observed in Figs. 4
and 5. The residual cracks results if the matrix contracts
more than the particle resulting in emanating from the parti-
cles. In Fig. 4, the large quartz particle exhibits continuous
peripheral fracture at or near the grain boundaries and interconnected matrix fracture. An effect that can lead to artefacts in microstructural evolution is the release of induced
stresses during specimen preparation like grinding and polishing of samples might lead to some observed cracks on the
surface because of stress release. Crack size can be seen in
the SEM images, which is shown in Fig. 5. The crack width
is very small seems to be 0.1 micron or less which is not
visible in the optical microscope. When such a crack retains
in the structure, air or any type of gas stays in those cracks,
which have a higher conductivity than any other phases of
the structure. Again the crack tips always act as stress concentrators, which, lead to further cracks or failure and that,
reduces the bending strength. So this type of micro cracks
surely deteriorates the dielectric and mechanical properties
of ceramic insulator.
4. Conclusion
From the above discussion, it is clearly understood that
mullite and the quartz are the two phases, which contributes
to the mechanical and electrical properties of the ceramic
insulator. In the reverse, if the glassy phase content gets
higher it no longer contributes to the properties of the
ceramic insulator instead it reduces the mechanical property (bending strength) and dielectric property (dielectric
strength). The mullite formation is less due to the presence
of small amounts of Al2 O3 (22.64%) and a large amount of
SiO2 (70.03%). Again the presence of high amount of SiO2
will lead to the presence of high amount of glassy phase
that is the most detrimental for developing high dielectric
strength. Because if the glassy phase increases it gives a
140
R.A. Islam et al. / Materials Science and Engineering B106 (2004) 132–140
free path to the mobile ions like Na+ , K+ , Al3+ etc. to
move and hence increases the conductivity. EDAX analysis
of glass rim shows considerable amount of Na and K ion.
Another important microstructural features that reduce the
bending and the dielectric strength is microcracks. SEM images clearly show micro sized cracks, which are 0.1 ␮m in
width in general. The presence of those micro sized cracks
surely reduces the mechanical and electrical properties that
are clearly reflected in the bending strength and dielectric
strength data, which are 757 kg/cm2 and 28 kV/mm respectively. XRD pattern describes the presence of mullite, clay
matrix and quartz particle. The presence of low percentage
of mullite can be understood by the intensity of the mullite
peak, which is almost 70%. The formation of low mullite,
presence of micro cracks and the dissolution of quartz particle is the main cause of lacking the dielectric and mechanical
properties of the ceramic insulator. So it can be concluded
that the best mechanical and dielectric properties can be
achieved by higher mullite and quartz content with lower
amount of glassy phase and the absence of microcracks.
Acknowledgements
The authors would like to acknowledge the collaboration
of Electronic Engineering Department of the City University
of Hong Kong specially the Centre for Electronic Packaging
and Assemblies, Failure Analysis and Reliability Engineering and the Materials and Metallurgical Engineering Department of BUET. The authors also like to thank the BISF Ltd.
for supplying the necessary raw materials and Bangladesh
University of Engineering and Technology for allowing
them to conduct some experiments in their laboratory.
References
[1] R.C. Buchanan, Ceramic Materials for Electronics – Processing Properties and Application, second edition, Marcel Dekker, New York,
1991, p. 1.
[2] J.A. McPhee, S.J. Macgregor, J.R. Tidmarsh, Annual report: Conference On Electrical Insulation and Dielectric Phenomena, IEEE
Dielectr. Insulat. Soc. II (2000).
[3] K.H. Schuller, J. Mater. Educ. 4 (1982) 526–529.
[4] S.P. Chaudhury, J. Br. Ceram. Soc. 73 (2) (1974) 37–41.
[5] I.A. Aksay, J.A. Pask, J. Am. Ceram. Soc. 58 (11–12) (1975) 507–
512.
[6] W.M. Carty, U. Senapati, J. of Am. Ceram. Soc. 81 (1) (1998) 3–20.
[7] W.D. Kingery, H.K. Bowen, D.R. Uhlmann, Introduction to Ceramics,
second edition, Wiley, New York, 1975, p. 534.
[8] P.J. Harrop, Dielectrics, first edition, Wiley, New York, 1972 (Chapters 1–6).
[9] F.M. Clark, Insulating Materials for Design and Engineering Practice,
second edition, Wiley, New York, 1962 (Chapter 16).
[10] ASTM (D3755).