IL-5 / G-47

2012 IEEE International Conference on Condition Monitoring and Diagnosis

23-27 September 2012, Bali, Indonesia

Surface Trapping Parameters of Solid Dielectrics:

Novel Measurement Method and Insulation Condition
Characterization
Guan-Jun Zhang*, Wen-Wei Shen, Hai-Bao Mu, Jun-Bo Deng
(State Key Lab of Electrical Insulation and Power Equipment, School of Electrical Engineering, Xian Jiaotong University, Xian, Shaanxi 710049, China)
*E-mail: [email protected]

long run [1-3]. Space charge is closely related to traps existing

in the materials, and trap characteristic provides an important
way to have a more thorough understanding of space charge
properties of insulating materials, thus giving rise to widespread interest in the past several decades.
Many kinds of trap measurement methods have been developed, such as the thermally stimulated currents (TSC) [6], the
photon stimulated current (PSC) [7-8] and the isothermal relaxation current (IRC) [9], etc. Although the TSC method is
widely used, this method can not distinguish the electron or
hole trap and also the actual states of traps may be destroyed
by the motivation of surrounding moleculars under high temperature [8]. The PSC method is based only on the single excitation of charge carrier by the photon, but in the actual case
large quantities of double excitation processes exist simultaneously. The IRC method proposed by Simmons at 1970s is in
the formula that the relaxation current is measured under a
constant temperature ambient and this method can be used to
distinguish hole and electron traps.
In this paper, a novel measurement system of surface trap
parameters is developed, which consists of a setup of corona
triodes used for charging the surface of insulating material
uniformly and an electrostatic voltmeter instrument for noncontact surface potential measurement. Based on the IRC
theory, the energy distribution of electron or hole traps can be
obtained. The trap characteristics of on-site aged high temperature vulcanized (HTV) silicone rubber (SIR) and machinable
ceramics (MCs) are measured, and the relations to their aging
condition and electric withstanding strength are discussed.

AbstractCharge traps greatly affect the charge transport, trapping and recombination in solid dielectrics such as ceramics and
polymers. This paper presents a novel method of measuring the
surface trapping parameters such as trapping density and energy
level of insulating materials. Based on the isothermal relaxation
current (IRC) theory, the trap measurement and calculation
formula is deduced and a non-contact measurement setup is specially designed, and electron or hole trap parameters can be determined by choosing negative or positive charging, respectively.
Two kinds of materials are selected as the samples, i.e., high temperature vulcanized (HTV) silicone rubber and machinable ceramics (MCs). Experimental results indicate that, with extending duration of aging time, electron and hole trap density both increase,
and gradually saturate. SEM and XPS analyses reflect that specimens after aging process appear many physical defects as holes
and flaws and chemical defects as strong polar groups, which
causes significant influence on surface trap characteristics. It is
shown that the glass phase on the surface of MCs greatly influences the existing shallow traps, which in turn degrade the flashover characteristics. Removing the glass phase with HF treatment reduces the concentration of shallow traps and stabilizes
the flashover voltage. It is considered that charge trap is expected
to be as a novel indicator for effective evaluation of aging status
and electric withstanding strength of insulating materials.
Keywords- Solid dielectrics; corona charging; surface potential
decay; electron trap; hole trap

I.

INTRODUCTION

It is well known there are a lot of traps inside solid insulating materials, which are formed by the existing defects in the
amorphous region and chemical tails of macromolecule chain,
especially on the surface of a material. The surface atoms of a
solid are particularly active due to the presence of unsaturated
chemical bonds resulted from the sudden termination of lattice
structures. Thus the solid surface is generally covered with
one or more layers of impurity or atmosphere atoms or molecules on account of chemisorption and physisorption [1-2].
Surface states are created by the above imperfection on the
surface. They act as trap centers of electrons and holes injected from electrodes, and the density of these centers near
the surface is much higher than that in the bulk. These traps
have significant influences on charge transport, trapping, accumulation, detrapping and recombination inside the material
[3-5]. Meanwhile, space charge formation especially under
high dc electric fields may cause the electric field severely
distorted and thus the material overstressed, and this even induces the degradation and breakdown of the material in the

978-1-4673-1018-5/12/$31.00 2012 IEEE

II. SAMPLE PREPARATION AND EXPERIMENTAL SETUP

Two kinds of samples are prepared, one is HTV SIR sample
cut from field-aged insulators with the size of 30303mm3,
and the other is circular MC wafers with a diameter of ~80 mm
and thickness of ~3 mm. All the samples are cleaned with anhydrous alcohol, raised with deionised water in an ultrasonic
cleaner and dried in the air at the temperature of ~60qC in a
baking oven.
Samples are firstly charged under a setup of corona triodes
which generates uniform corona discharge and then moved to
the electrostatic voltmeter for surface potential measurement.
The corona triodes consist of a needle, a mesh grid and a
grounded planar electrode. The needle electrode generates
corona discharge when applied with f10kV dc high voltage
and the mesh grid is applied with f2kV dc bias voltage. The

480

Surface potential (V)

-1400
-1200
-1000
-800
-600
-400

2 years
5 years
10 years
15 years

(b)

1400
1200
1000
800
600
400
200

-200
0

500

1000

1500

2000

2500

3000

3500

4000

500

1000 1500 2000 2500 3000 3500 4000

Time (s)

Time (s)

Figure 2. Surface potential decaying characteristic curves of HTV SIR samples with different operating years
(a) under negative corona charging, (b) under positive corona charging

B. Surface trap energy distribution

Based on numerical differential calculation, the current density
can be obtained. From Eqs. (1) and (2), the electron and hole trap
parameters of these samples can be achieved. The trap characteristic of site aged samples is shown in Fig. 3, reflecting in the
bell-like shape. For the electron traps, the shape of trap distribution does not change a lot, only becoming thinner with
operational time. With the extension of operational time, there
are more traps generated, and it is conformed that the peak of
electron trap density increases significantly from 1.151019 at
2 years to 2.251019(eV)-1m-3 at 15 years.
Compared with that of electron trap, the hole trap distribution presents some new features. The shape of distribution is
also in the bell-like shape but flatter. The hole trap density
increases slowly with the peak density increasing from
1.251019 at 2 years to 1.851019(eV)-1m-3at 15 years.

H 0H r d Vs (t )
dt

(1)
The relationship among the trap energy Et, the trap density Nt
and the isothermal decay current density J is given as [9]
Et kT ln(Jt )

(2)
J (t ) qdkT f ( E ) N ( E )
0
t
t
t

2t

2.00E+019

2.4

(a)

2 years
5 years
10 years
15 years

2.2
2.0
1.8
1.6

1.80E+019

Hole trap density/(eV)-1m-3

Electron trap density/*1019*(eV)-1 m-3

Based on the isothermal decay current theory, after the applied

voltage is removed, the charge carriers inside shallow traps in
dielectric firstly release, and the deeper trap charge carriers release later. Under a constant temperature condition, the released
current changes with time, and this current can describe the trap
energy distribution pattern. The relationship between the surface
potential decay Vs and the current density J is

1.4
1.2
1.0
0.8
0.6
0.4

2 years
5 years
10 years
15 years

1.60E+019
1.40E+019

(b)

1.20E+019
1.00E+019
8.00E+018
6.00E+018
4.00E+018
2.00E+018

0.2

0.00E+000

0.0
0.70

0.75

0.80

0.85

0.90

0.75

0.95

0.80

0.85

0.90

0.95

Trap energy/eV

Trap energy/eV

Figure 3. Trap energy distribution of site aged samples

(a) Electron trap; (b) Hole trap

III. EXPERIMENTAL RESULTS AND DISCUSSION: PART I- HTV

SILICONE RUBBER

Fig. 4 shows the relationship of trap density peak and operational years, and also the fitting formula are given. We can find
the basic linear relations. Electron and hole trap density is expected to be as a novel indicator for effective evaluation for
aging status of HTV SIR material.
1.9

(a)
-3

2.2

(b)

1.8

-1

-1

Density of Hole Trap /*10 (eV) m

-3

2.4

2.0

1.7

19

19

Where Et is the trap energy, k is the Boltzmanns constant, T is

the absolute temperature, J is the electron vibration frequency, set
as 1012s-1 [10], q is the electron charge, d is the sample thickness, f0(Et) is the initial occupation probability of traps, set as 1/2
[10], and Nt(Et) is the trap density corresponding to the energy
level of Et.
According to Eqs. (1) and (2), the electron and hole trap energy
distribution of polymers can be calculated. The injected homocharge exists in a surface layer just as thin as 1~2Pm [11], which
dominate the surface potential, so the measured trap parameters
mainly reveal the trap distribution in the surface layer of a sample.

Density of Electron Trap /*10 (eV) m

2 years
5 years
10 years
15 years

(a)

-1600

Figure 1. Experimental setup for trapping parameter measurement

J (t )

1600

-1800

Surface potential (V)

initial surface potential is determined by the bias voltage applied across the mesh grid and the grounded electrode, as
shown in Fig. 1. After charging the sample for 1min, a piece
of aluminium foil is laid onto the surface of the sample and
shorted for 5min, to remove the free carriers (not trap charge
carriers) off the surface of the sample. The sample then is
moved to the probe of electrostatic voltmeter via a rotatable
design. The distance between the probe and the surface of
sample is kept as ~2mm. The experiments are carried out in an
adiabatic airtight chamber in which the temperature and humidity
is approximately kept as constant.

1.8

1.6

1.4

Ne(t)=1.241+0.062t

1.2

1.0

10

Operational time /year

12

14

16

1.6

1.5

1.4

Np(t)=1.283+0.035t

1.3
0

10

12

14

16

Operational time /year

Figure 4. Relationship of trap density peak and operational years

(a) Electron trap; (b) Hole trap

A. Surface potential decay of HTV SIR

The negative and positive potential decay for aged HTV
SIR samples under the temperature of ~40qC are shown in Fig.
2, respectively. It takes several tens minutes for the isothermal
potential decay. Longer aging time, a quicker potential decay.
There is no significant polarity effect observed.

C. Surface microstructure observation

Changes of surface physical properties imply modification
of the surface. It is supposed that aging generates lots of electrons and ions of high energy which accelerate across the elec-

481

tric field and then bombard the surface, causing erosion in the
material. In order to investigate the correlation between physical characteristic and microstructure of material, surface microstructure is measured by using SEM analysis, as shown in
Fig. 5.
The photographs clearly show that surface microstructure of
site aged sample is greatly eroded and there are some novel
features during various aging period. The surface firstly is
hardened during short aging period, which is easier to be destroyed by the high energy particles and chemical contamination, and then as the aging process going on, the hardened surface is cracked and further erosion is ready to develop into the
material, thus inducing deeper holes and even cracks across
the surface as shown in Figs. 5c and 5d.
(a)

(b)
Figure 6. Surface potential decaying characteristic curves
(a) before treatment with HF, (b) after treatment with HF

(c)

(d)

Figure 7. Surface trap density distributions before and after HF treatment

(a) electron trap before treatment; (b) electron trap after treatment;
(c) hole trap before treatment; (d) hole trap after treatment.

Figure 5. Surface microstructure of site aged samples

(a) 2 years; (b) 5 years; (c) 10 years; (d) 15 years

IV. EXPERIMENTAL RESULTS AND DISCUSSION: PART IIMACHINABLE CERAMICS

According to Fig. 7, different samples show quite different trap

density distributions. When compared with the no-doped MC
sample, while adding some low SEE yield metal oxide, the trap
densities of the samples are reduced significantly, the sample
doped with 3% Cr2O3 shows the lowest trap density. After the HF
treatment, the trap density and energy distribution of samples
change obviously. For the original samples with no HF treatment,
both the electron and hole trap distributions have an obvious density peak at the energy interval of about 0.770.81 eV, and after a
minimal value at the energy interval of about 0.820.85 eV, the
trap density increases appreciably, as shown in Figs. 7a and 7c.
After getting eroded by HF, as shown in Figs. 7b and 7d, the peak
value of trap density corresponding to the shallow traps decreases
significantly.
Table 1 summarizes the changes of trap density peak value
of MC samples near the energy level of 0.770.81 eV before
and after HF treatment. For the MC samples with 3% Cr2O3,
no obvious trap density peak is observed after HF treatment.

A. Trap parameter measurement of MCs

For each MC sample, the surface trap measurement is performed at room temperature before and after HF erosion, respectively. After charging the sample with positive or negative corona
discharges, the surface potential decay curve is recorded immediately, as shown in Fig. 6. It usually takes only about several
minutes for the potential to decrease to zero, much shorter than
that of HTV SIR materials. It can be seen that after treatment with
HF, the surface potential decay of the samples is much quicker
than that of untreated ones.
The electron and hole trap parameters of these samples before and after eroding can be achieved, as shown in Fig. 7. The
trap energy interval of MCs is about 0.710.87 eV, which belongs to the relatively shallow traps. It is believed that the
shallower or deeper traps can also be reflected under lower or
higher temperature, respectively.

Table 1. Comparison of trap density peak for different samples

Trap density (eV)1/m3
Samples
Electron trap Net
Hole trap Nht

482

No doped
1% Cu2O
1% Cr2O3
3% Cr2O3

Before
erosion
1.421018
1.031018
4.621017
2.261017

After erosion
1.811017
6.511016
1.221017
2.931016

Before
erosion
1.901018
1.341018
9.491017
4.211017

After erosion
8.701016
6.271016
1.751017

B. Flashover characteristics of MCs

In the flashover experiment in vacuum, the pulsed voltage of
0.42/2.5 s double exponential pulse is applied. Several voltage
levels are used to characterize the flashover phenomena for each
sample, i.e., the first breakdown voltage Ufb, the conditioned voltage Uco, and the hold-off voltage Uho, as shown in Tables 2 and 3.
After HF eroding, a significant increase of flashover voltage is
achieved. Alumina ceramic (99% purity) is also tested with same
electrode configuration as comparison. There is no significant
difference observed before and after HF treatment, possibly because of chemical stability of alumina.
Table 2. Flashover performance of samples before HF treatment
Samples
Ufb (kV)
Uco (kV)
Uho (kV)
Al2O3
27.3
41.1
32.2
No doped
27.5
45.6
39.7
1% Cu2O
22.2
43.5
41.0
1% Cr2O3
18.1
43.2
37.0
3% Cr2O3
20.9
41.2
40.2
Table 3. Flashover performance of samples after HF treatment
Samples
Ufb (kV)
Uco (kV)
Uho (kV)
Al2O3
28.4
40.5
31.3
No doped
29.0
42.3
40.2
1% Cu2O
33.2
45.0
44.5
1% Cr2O3
36.9
45.8
42.9
3% Cr2O3
41.6
41.6
41.3

C. SEM analysis
The no-doped MC and 3% Cr2O3 doped sample before and after HF treatment were selected to observe their surface microstructure, as shown in Fig. 8. The photographs clearly indicate
that after HF erosion, the glass phase components on the surface
of the samples have been eroded, and the crystal phase components, displayed as the sectional shape of sticks in Figs. 8b and 8d,
emerge on the surface. It is believed that further erosion will bring
more crystal phase component appearance in Fig. 8d. It is considered that the surface glass phase structure has a definite relationship with the trap parameters in the surface layer of a sample.
Reduction of glass phase components induces the decrease of trap
density, as in Table 1.

Figure 8. Typical SEM microstructure of MC samples

(a) original no-doped sample; (b) no-doped sample after HF treatment;
(c) original 3% Cr2O3 doped sample; (d) 3% Cr2O3 doped sample after HF
treatment

Under an applied voltage even lower than the flashover voltage, there are SEE phenomena and the generation of holes
with positive charges on the sample surface. The generated
electrons can be captured for a short time or caught forever by
electron traps. When the holes have not been captured by hole
traps, they may easily recombine with electrons, but while
parts of the holes are captured by hole traps, it will be difficult
for their recombination, leading to more positive charges on
the surface. And based on our proposed charge injection model, charge carriers injected from electrodes can be trapped in
trap centers [12-13]. The trapped charges will nonuniformly
accumulate on the surface between the electrodes, leading to
the apparent distortion of local surface electric field, and make
the flashover easier under the next higher impulse. Thus, it is
easy to understand that a higher surface trap density corresponds to a lower flashover voltage.
V. CONCLUSIONS
Based on the experimental results and analysis, for HTV SIR,
it is considered that aging makes surface holes and cracks, and
produces many strong polar and hydrophilic chemical groups
in the material, leading to the increase of trap density. For
MCs, the glass phase on the surface of MC materials is an important reason for the traps existing on the sample surface. The surface glass phase components can be effectively eroded off by HF

483

[1]
[2]
[3]
[4]

treatment. Surface traps can be considered as the reason for the

lower flashover voltage. Eliminating the shallow trap distribution
with HF erosion treatment can improve the stabilization of flashover characteristics across a material. Trap distribution is expected to be as a novel indicator for effective evaluation of
aging state and electric withstanding strength of solid dielectrics. It is still needed to clarify the exact influences of traps with
different energy levels, and also the effects of traps under different applied waveforms. Much work is being implemented.

[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]

ACKNOWLEDGMENTS
This work is in part supported by the China National Funds
for Distinguished Young Scientists (Grant No. 51125029) and
National Natural Science Foundation of China (Grant No.
50937004 and 51107096).

[13]

REFERENCES

484

K. C. Kao, W. Hwang, London: Pergamon Press Ltd., 1981.

G. Teyssedre, C. Laurent, IEEE TDEI, vol. 12, no. 5, pp. 857-75, 2005.
T. J. Lewis, IEEE TDEI, vol. 9, no. 5, pp. 717-29, 2002.
Z. Q. Xu, L. W. Zhang, G. Chen, J. Phys. D: Appl. Phys. vol. 40, pp.
7085-89, 2007
G. Chen, ZQ Xu, J. Appl. Phys., vol. 106, 1237071-5, 2009.
T. Hino, IEEE Trans. Electr. Insul., vol. 21, no. 6, pp. 1007-10, 1986
J. D. Brodribb, D. Ocolmain, D. M. Hughes, J. Phys. D: Appl. Phys.,
vol. 8, no. 7, pp. 856-62, 1975.
L. J. He, Y. Chen, L. Dai, IEEE 9th ICPADM, China, pp. 19-23, July
2009.
J. G. Simmons, M. C. Tam, Phys. Rev. B, Vol. 7, pp. 3706-13, 1972.
G. S. Nadkarni, J. G. Simmons, Phys. Rev. B, vol. 7, pp. 3719-24, 1972.
P. K. Watson, J. Phys. D: Appl. Phys. vol. 23, pp. 1479-84, 1990.
G. J. Zhang, W. B. Zhao, Z. Yan, Y. S. Liu, IEEE TDEI, Vol. 11, pp.
498-505, 2004.
G. J. Zhang, Z. Yan, Y. S. Liu, K. Yasuoka, S. Ishii, Appl. Phys. Lett.,
vol. 78, pp. 625-27, 2001.