polymers

Article
Influence of Repetitive Square Voltage Duty Cycle on
the Electrical Tree Characteristics of Epoxy Resin
Peng Wang 1 , Suxin Hui 1 , Shakeel Akram 1, * , Kai Zhou 1 , Muhammad Tariq Nazir 2 ,
Yiwen Chen 1 , Han Dong 1 , Muhammad Sufyan Javed 3 and Inzamam Ul Haq 4
1 College of Electrical Engineering, Sichuan University, Chengdu 610065, China; pwang@scu.edu.cn (P.W.);
hsx1759499817@163.com (S.H.); zhoukai_scu@163.com (K.Z.); ceven96@163.com (Y.C.);
major2011@163.com (H.D.)
2 School of Mechanical and Manufacturing Engineering, University of New South Wales,
Sydney, NSW 2052, Australia; muhammadtariq.nazir@unsw.edu.au
3 Department of Physics, Jinan University, Guangdong 510632, China; safisabri@gmail.com
4 State Key Laboratory of Power Transmission Equipment & System Security and New Technology,
Chongqing University, Chongqing 400044, China; inzimam.324@gmail.com
* Correspondence: shakeel.akram@scu.edu.cn;




Received: 31 August 2020; Accepted: 22 September 2020; Published: 27 September 2020


Abstract: The application of wide band-gap power electronic devices brings more challenges to
insulating packaging technology. Knowing the influence of applied voltage parameters on insulation
performance is helpful to evaluate the insulation condition of electric power equipment. In this paper,
the effect of repetitive square wave voltage duty cycle on the growth characteristics of electrical
trees in epoxy resin was studied. The experimental results show that the square wave voltage duty
cycle has a significant influence on treeing features. The electrical tree proportion initiation has
shown a decreasing trend, and the shape of the electrical tree changes from pine-like to branch-like
by increasing the duty cycles. The length and damaged area of electrical tree increased with the
increase in the duty cycle up to 10% and then decrease by increasing the duty cycle higher than 30%.
It indicates that a low duty cycle will enhance the electron injection and accumulate space charges
and thus accelerate electrical tree development. Under short duty cycles, the electric field due
to the shielding effect near the needle tip suppresses the electrical tree growth, which results in
treeing growth stagnation. The obtained results are helpful to keep these parameters in mind during
the design of epoxy-based insulation such high-voltage rotating machines and power electronic
device packaging.

Keywords: epoxy resin; electrical tree; duty cycle; power electronic devices; square wave voltage;
electric machines insulation

1. Introduction
Electric power equipment is the expensive and core parts of electric power systems. The application
of third-generation wide band-gap semiconductor devices, represented by silicon carbide (SIC)
and gallium nitride (GAN), enables the existing power electronic system to work at the high
switching frequency and voltage level, thereby requiring high insulation reliability for such power
electronic devices [1–3]. Packaging technology determines the reliability of power electronic devices.
The high-frequency and high-voltage stresses generated by wide band-gap devices bring more
challenges to the traditional insulation technology and materials used in silicon-based semiconductor
devices [4–6].
Epoxy resin has been used as packaging materials in power electronic devices because of
its excellent physical and mechanical properties, chemical resistance and electrical insulation [7–9].

Polymers 2020, 12, 2215; doi:10.3390/polym12102215 www.mdpi.com/journal/polymers

Polymers 2020, 12, 2215 2 of 11

Pulse width modulation (PWM) technology is a standard transformation technology of power electronic
devices that can produce sinusoidal current waveform on the load by changing the duty cycle. The duty
cycle is the primary parameter of PWM technology, and thus the performance of insulation packaging
materials under different duty cycles should be considered carefully. The electrical stresses on
epoxy resin generated by the repetitive impulsive voltage are different from traditionally operated
sine wave voltage and DC voltage conditions. Therefore, the performance of epoxy resin under
square wave voltage by varying duty cycles is the primary consideration for its application as the
insulation packaging of power electronic devices to improve the safety, stability, and reliability of the
insulation system.
The electrical tree is an essential phenomenon of physical defects in epoxy resin, which can lead
to the electrical breakdown of insulating materials under high electric field conditions. To avoid the
micro defects entirely in the solid insulation packaging material is difficult because of the limitations
in the industrial production process [10,11]. These physical defects may undergo mechanical and
chemical changes under electrical, thermal, and mechanical stresses. Once the electric field is
localized at a certain point, the partial discharge may be produced to generate further electrical
trees [11,12]. The growth of the electrical tree eventually leads to the breakdown of insulating materials,
which seriously threatens the security and stability of the power electronic system.
The initiation and growth characteristics of the electrical tree in epoxy resin have been investigated
in recent years. In Reference [13], the experimental results indicate that, when the DC and pulse voltage
are of the same polarity, the rate of tree growth increases with the amplitude of DC voltage rising
but would slow down after a certain period. When the polarity of DC and pulse voltage is opposite,
the growth rate decreases with the DC voltage amplitude increasing. A polarity dependence of treeing
at DC voltage was studied in [14], showing that the morphology of the initial trees, either bush-like or
branch-like, does not have a pronounced effect on the structure of the subsequent DC tree. The impact
of combined DC-harmonic voltage ranging from −20 to +20 kV was investigated in epoxy resin [15].
It was reported that the tree structure tends to be branch-like with positive DC voltage and has a
relatively high probability of being bush-branch like with negative DC voltage. None of the above
research can fully reflect the insulation performance of epoxy resin under high-frequency square wave
voltage conditions by the varying duty cycle. Therefore, it is necessary to investigate the growth
characteristics of the electrical tree under repetitive PWM voltage.
From the previous research, we have found that the square wave voltage parameters have a
significant influence on the failure process of turn to turn insulation for inverter-fed motors [16–18].
The partial discharge tested performed on the turn to turn insulation at square wave voltages with
different duty cycles, showing that the smaller or higher duty cycle would cause the asymmetry of the
partial discharge patterns, thereby affect the insulation life. This conclusion is only applicable to the
evaluation of corona resistance turn-to-turn insulation of inverter-fed motors.
This paper aims to report our study on the influence and mechanism of repetitive square wave
voltage duty cycles (100%× positive voltage duration/square wave voltage cycle time) on the electrical
tree growth characteristics of epoxy resin while keeping the peak-to-peak magnitude, frequency, and rise
time of square wave voltage at 12 kV, 1 kHz, and 70 ns, respectively. The mechanism of the electrical tree
under square wave voltage is proposed by analyzing the treeing initiation proportion, shape, length and
damaged area under different duty cycles. The results obtained in this study are expected to provide
a reference for the design of insulation package, the evaluation of insulation reliability, and the
improvement of insulation performance in power electronic systems.

2. Experimental Setup

2.1. Specimen Preparation and Treeing Test Platform

E-51 epoxy resin and methyltetrahydrophthalic anhydride curing agent are mixed and solidified
in a 2:1 mass ratio. The epoxy resin and curing agents are preheated before mixing. The two agents
Polymers 2020, 12, x FOR PEER REVIEW 3 of 13

2. Experimental Setup

2.1. Specimen Preparation and Treeing Test Platform

Polymers 2020, 12, 2215 3 of 11
E-51 epoxy resin and methyltetrahydrophthalic anhydride curing agent are mixed and solidified
in a 2:1 mass ratio. The epoxy resin and curing agents are preheated before mixing. The two agents
◦ C. The mm33
aremixed
are mixedandandstirred
stirredclockwise
clockwiseforfor 3030 min
min at
at 60
60 °C. The mould with the size of 120 ×× 120 120 ××10
10 mm
wasused
was used for
for sample
sample preparation
preparation (see(see Figure
Figure 1a).
1a). The
Theneedle
needleelectrode
electrodeandandaluminum
aluminumplate platewere
were
embeddedininthe
embedded theepoxy
epoxyresin
resinininadvance
advanceto toprevent
prevent bubbles
bubblesin in the
the sample.
sample. TheThemixed
mixedliquid
liquidafter
after
vacuum defoaming is poured into the mould using the needle electrode placed
vacuum defoaming is poured into the mould using the needle electrode placed in advance and placedin advance and placed
intothe
thevacuum
vacuumdrying
dryingoven.
oven.TheThecuring
curingtime
timeisis24
24h,h,keeping
keepingthethetemperature
temperatureatat60 ◦ C. In Figure 1b,
60°C.
into In Figure 1b,
theneedle
the needleelectrode
electrodesize
size 33
33 mm
mm length and 180 μm diameter with
µm diameter with model
modelspecification
specificationof of06cr19ni10
06cr19ni10
(SUS304) stainless-steel wire is used. The curvature e radius of the needle tip is
(SUS304)stainless-steel wire is used. The curvature e radius of the needle tip is 3 ± 0.2 μm, the3 ± 0.2 µm, thesurface
surface
roughnesslesslessthan
than0.63
0.63μm,
µm,thetheincluded
includedangle
angleof ofthe
theneedle
needletip
tip30°, ◦
30 ,and
andthe
thedistance
distancebetween
betweenthethe
roughness
needletip
needle tipand
andthethegrounding
groundingaluminum
aluminumplate 1.8±±0.1
plateisis1.8 0.1mm.
mm.

Unit：mm
Needle electrode
(connect to high voltage)

Epoxy resin

10 15
1.8 4
110 Aluminum plate electrode
(connect to ground)
(a) (b)
Figure
Figure1.1.(a)
(a)Epoxy
Epoxyresin
resincasting
castingmold;
mold;(b)
(b)needle-plate
needle-plateelectrode
electrodesample.
sample.

The test platform

The platformfor forelectrical tree
electrical growth
tree growthis shown in Figure
is shown 2. The 2.
in Figure square
The wave
square voltage
wavegenerator voltage
consists ofconsists
generator a positiveof aand negative
positive andDC powerDC
negative supply,
power signal generator,
supply, high-voltage
signal generator, solid-state solid-
high-voltage switch
state switch and the low voltage control part. The parameters of the square wave voltage can be seen1,
and the low voltage control part. The parameters of the square wave voltage can be seen in Table
and
in all parameters
Table can be adjusted.
1, and all parameters The peak-to-peak
can be adjusted. magnitude
The peak-to-peak and polarity
magnitude of the output
and polarity of the outputvoltage
could becould
voltage adjusted by the positive
be adjusted by theand negative
positive and DCnegative
power supply. The duty
DC power cycleThe
supply. andduty
frequency cycle could and
be adjusted
frequency by changing
could the parameter
be adjusted by changing of the
thesignal generator,
parameter of thewhich was
signal used to control
generator, which the wasswitching
used to
on and the
control off ofswitching
the high-voltage
on and solid-state
off of theswitch modulesolid-state
high-voltage to produce switch
high-voltage
module square wave voltages.
to produce high-
If we change the charging resistance and energy storage capacitor, the rise time
voltage square wave voltages. If we change the charging resistance and energy storage capacitor, the of the output voltage
cantime
rise be changed
Polymers of the
2020,
from ns
12, xoutput
to ms can
voltage
FOR PEER REVIEW
level.
be changed from ns to ms level. 4 of 13
In this experiment, the needle-plate sample was immersed in electric insulation oil to avoid
corona discharge at the high-voltage output. The needle electrodes were connected to the high-
voltage output terminal, and the aluminum plate electrode was grounded. The output voltage Low voltage
control part
signal
was monitored by Computer using a high-voltage probe
Microscope (Tektronix P6015A, 50 MHz frequency bandwidth,
Beaverton, OR, USA) and transmitted to a digital oscilloscope for observation. The parameters of the
Charging resistance

Solid state switch

electrical tree were recorded and observed by a high-precision metallographic microscope Energy storage
(Leica
DC Power + High 1000:1 capacitor
DVM6,Supply
Wetzlar, Germany). High
Voltage Discharge resistance
Solid Voltage
DC Power
Supply
- Switch Probe High voltage probe

DC input
Signal Digital Electric Insulation Oil
Square voltage
Generator Oscillo Scope Needle-plate Sample output

(a) (b)
Figure Testingplatform
Figure 2. Testing platformfor
forthe
the growth
growth of the
of the electrical
electrical tree.tree. (a) Schematic
(a) Schematic diagram
diagram and
and a (b) a (b)
physical
physical
picture ofpicture
squareofwave
square wave
power power supply.
supply.

Table 1. Parameters of square wave voltage.

Table 1. Parameters of square wave voltage.
Peak-to-Peak Magnitude
Peak-to-Peak Magnitude Polarity
Polarity Frequency
Frequency Duty
DutyCycle
Cycle RiseTime
Rise Time
0–15 kV
0–15 Unipolar/Bipolar 0–5
Unipolar/Bipolar kHz
0–5 kHz 0–100%
0%–100% ns-ms level
ns-ms level

2.2. Experimental Procedure

Fifteen needle-plate electrode samples were selected to measure the initiation proportion and a
total of 10 pin electrode samples were tested the propagation under the same duty cycle to reduce
the possible influence of sample randomness on the experimental results. As shown in Figure 2,
Polymers 2020, 12, 2215 4 of 11

In this experiment, the needle-plate sample was immersed in electric insulation oil to avoid
corona discharge at the high-voltage output. The needle electrodes were connected to the high-voltage
output terminal, and the aluminum plate electrode was grounded. The output voltage signal was monitored
by using a high-voltage probe (Tektronix P6015A, 50 MHz frequency bandwidth, Beaverton, OR, USA)
and transmitted to a digital oscilloscope for observation. The parameters of the electrical tree were recorded
and observed by a high-precision metallographic microscope (Leica DVM6, Wetzlar, Germany).

2.2. Experimental Procedure

Fifteen needle-plate electrode samples were selected to measure the initiation proportion and a
total of 10 pin electrode samples were tested the propagation under the same duty cycle to reduce
the possible influence of sample randomness on the experimental results. As shown in Figure 2,
treeing tests were performed at square wave voltages with an adjustable duty cycle from 5% to 90%
while keeping the peak-to-peak voltage, frequency, and rise time at 12 kV, 1 kHz, and 70 ns, respectively.
Then, the treeing parameters (such as length and shape) were recorded by the microscope.

3. Results
The length of the electrical tree is defined in this paper as the longest distance in parallel direction
from the needle tip to the end of electrical tree and the width of the electrical tree as the horizontal
longest distance from one end of the electrical tree to the other end.

3.1. Effect of Duty Cycle on the Treeing Initiation Proportion

The electrical tree initiation is defined when the tree length is longer than 20 µm [19]. Table 2
is showing the probability of treeing initiation at two levels of time under different duty cycles.
The number of electrical tree initiations changes by varying duty cycles, as shown in Table 1. At the
short duty cycle, the chances of tree initiation are higher than that under high duty cycles with the treeing
time increasing. From Figure 3, we can see a falling slope of the electrical tree initiation proportion.
Taking 0.5 as the reference line, it can be concluded that when the initiation proportion of electrical
tree is greater than 0.5, there would be a higher possibility of initiating electrical trees in a short time.
When the electrical tree initiation proportion is less than 0.5, there would be less possibility of initiating
electrical trees in a short time. Therefore, square voltage with short duty cycle (long duration of
negative voltage) is more likely to cause the initiation of the electrical tree.
Table 2. Number of tree initiation.

Duty Cycle Number (3 min) Number (8 min) Total Number

10% 8 10 15
30% 9 10 15
50% 2 6 15
70% 3 6 15
Polymers 90%
2020, 12, x FOR PEER REVIEW 5 7 15 5 of 13
Electrical tree proportion of initiation

1.0
0.9 8min

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
10% 30% 50% 70% 90%

Duty cycle (/%)

Figure 3. Electrical tree initiation proportion versus testing time under square wave voltage with
Figure 3. Electrical tree initiation proportion versus testing time under square wave voltage with
different duty cycles (12
different kV/1
duty cycleskHz).
(12 kV/1 kHz).

3.2. Electrical Tree Morphology and Growth Characteristics

Figure 4 shows the typical shape of the electrical tree under square wave voltage with different
duty cycles. The number and length of branches increased with treeing time under the same duty
cycle. The typical shapes of the electrical tree change from pine-like to branch-like.

10min Unit:μm 30min Unit:μm 10min Unit:μm 30min Unit:μm

 Electrical t
0.2
0.1
0.0
10% 30% 50% 70% 90%

Duty cycle (/%)

Polymers 2020, 12, 2215 5 of 11
Figure 3. Electrical tree initiation proportion versus testing time under square wave voltage with
different duty cycles (12 kV/1 kHz).

3.2. Electrical Tree Morphology and Growth Characteristics

3.2. Electrical Tree Morphology and Growth Characteristics
Figure 4 shows
Figure 4 the typical
shows shape
the typical of of
shape thetheelectrical tree
electrical tree under
under square
square wave wave
voltagevoltage with different
with different
duty
duty cycles. Thecycles.
numberThe number and length
and length of branches
of branches increasedwith
increased with treeing
treeing time
timeunder
underthethe
same dutyduty cycle.
same
The typicalcycle. The typical shapes of the electrical tree change from pine-like to branch-like.
shapes of the electrical tree change from pine-like to branch-like.
10min Unit:μm 30min Unit:μm 10min Unit:μm 30min Unit:μm
Branch-like
Pine-like 159 Pine-like 162 Branch- 386
like 591
209 210
393

951
500 μm 500 μm 500 μm 500 μm

(a) (b)
10min Unit:μm 30min Unit:μm 10min Unit:μm 30min Unit:μm

Branch-
Branch- 795 Branch-
Branch-like like 278
like like
509
403 185

458 933 440

500 μm 500 μm 500 μm 500 μm

(c) (d)
10min Unit:μm 30min Unit:μm 10min Unit:μm 30min Unit:μm

Branch- Branch- Branch-

Branch- like 244
like 263 like 384
like 213
188 589
492

413
500 μm 500 μm 500 μm 500 μm

(e) (f)
Figure 4. Typical morphology of electrical tree under square wave voltage with different duty cycles.
Figure 4. Typical morphology of electrical tree under square wave voltage with different duty cycles.
(a) 5% duty cycle, (b) 10% duty cycle, (c) 30% duty cycle, (d) 50% duty cycle, (e) 70% duty cycle, (f)
(a) 5% duty cycle,
90% duty (b) 10% duty cycle, (c) 30% duty cycle, (d) 50% duty cycle, (e) 70% duty cycle, (f) 90%
cycle.
duty cycle.

At 5% duty cycle, the electrical tree branches are dense in the form of pine-like structure and
basically did not change in shape within 10–30 min. At 10% duty cycle, many branches of the electrical
tree and the end of the electrical tree tend to be hemispherical with the increase in time. Under 30%,
50%, 70%, and 90% duty cycles, the channel of the electrical tree decreased, and the electrical tree
became sparse with the increase in duty cycle.
Figure 5 shows the relationship between the length and width of the electrical tree and the square
wave voltage duty cycle. It can be seen that the electrical tree length tends to decrease by increasing
the square wave voltage duty cycles higher than 10%. Both the growth rate of length and width are
linear within 30 min. And the growth rate of the electrical tree at 30% duty cycle is the fastest, and the
average length of the electrical tree reached 712 µm and the maximum length was 983 µm at 30 min.

Figure 5. Electrical tree growth versus testing time under square wave voltage with different duty cycles.
(a) Electrical tree length versus testing time and (b) Electrical tree width versus testing time.
Polymers 2020, 12, 2215 6 of 11

However, the growth rate of the electrical tree for 5% and 10% duty cycles began to slow down
after 10 min. The length and width of the electrical branch are significantly reduced when the duty
cycle was reduced to 5%. Comparison of Figure 4a,b shows that electrical branches are dense under
10% and 5% duty cycles, the shape of the electrical branches did not change much within 10–30 min,
and a growth stagnation period occurred. The growth of electrical tree could be inhibited to a certain
extent when the duty cycle is smaller than a specific value. However, the threshold value of duty cycle
(duration of positive and negative voltage) for the growth stagnation period of electrical tree remains
to be further studied.

3.3. Electrical Tree Damaged Proportion

The area of the electrical tree reflects the degree of damage to the epoxy resin caused by electrical
tree branches. A fixed-size area (188 pixels × 151 pixels) is selected, where the damage proportion is
defined as the proportion of the area of electrical tree to the area of the selection. As shown in Figure 6,
a fixed size is selected for the electrical tree image under each duty cycle, and then
Polymers 2020, 12, x FOR PEER REVIEW
the image is
7 of 13
binarized
Polymers to12,
2020, calculate
x FOR PEER theREVIEW
proportion of electrical tree damage [20,21]. 7 of 13

(a) (b)
Figure 6. Example of electrical tree image binarization processing. (a) Electrical tree original image,
(b) Electrical tree after image binarization processing.

(a)
Figure 7 is the mean deviation diagram of the damaged area proportion(b) of electrical tree. The
treeing damaged proportion gradually decreases with the increase of duty cycle when the square
Figure 6. Example
Example of of electrical
electrical tree image binarization
binarization processing.
processing. (a)(a) Electrical
Electrical tree
tree original
original image,
image,
voltage duty cycles are higher than 10%, and the largest value of the damaged area lies in the 30%
(b) Electrical tree after image binarization processing.
duty cycle group. Based on Section 3.2 and 3.3, a conclusion can be drawn that a small duty cycle
(long duration of negative voltage) has a greater destructive effect on epoxy resin insulation
Figure
Figure 77isisthe
materials.themean
mean deviation
deviation diagram
diagram of the
of damaged
the damaged area proportion
area proportionof electrical tree. The
of electrical treeing
tree. The
damaged proportion
However, gradually
the damaged decreases
area of thewith the
electrical increase
tree under of
5% duty
duty cycle
cycle
treeing damaged proportion gradually decreases with the increase of duty cycle when the squareis when
smaller. the square
Figure 4a voltage
shows duty
cycles are that
higherthe electrical
than treeand
10%, has athe
hemispherical pine-like structure at 5% area
duty lies
cyclein
and so 30%
manyduty
electric
voltage duty cycles are higher thanlargest
10%, andvalue theoflargest
the damaged
value of the damaged the area liescycle
in thegroup.
30%
branches have a certain shielding effect on the electric field intensity of the needle tip, hindering the
Basedcycle
duty on Sections
group. 3.2 andon3.3,
Based a conclusion
Section 3.2 and can
3.3, be
a drawn thatcan
conclusion a small
be duty that
drawn cyclea (long
small duration
duty of
cycle
growth of electrical tree. Therefore, the tree grows slowly and the area of damage to the material is
negative voltage)
(long duration has a greater destructive effect on epoxy resin insulation
small. of negative voltage) has a greater destructive effect on epoxy resin insulation materials.
materials.
6
However, the damaged area of the electrical tree under 5% duty cycle is smaller. Figure 4a shows
Mean
Damaged area proportion(/%)

that the electrical tree has a hemispherical

5 pine-like structure at 5% duty cycle and so many electric
branches have a certain shielding effect on the electric field intensity of the needle tip, hindering the
growth of electrical tree. Therefore,
4 the tree grows slowly and the area of damage to the material is
small. 3.070 3.180
3 2.598
2.171
6
2
Mean
Damaged area proportion(/%)

1.268 1.414

5 1

0
4 0 10 20 30 40 50 60 70 80 90 100
Duty cycle (%)
3.180
3.070
Figure 7.area 3 area proportion of electrical tree
Damaged versus dutycycle
cycle under
Figure 7. Damaged proportion 2.598
of electrical tree versus duty undersquare wave
square voltage
wave at
voltage at 30 min.
30 min. 2.171
2
4. Discussion
1.268 1.414
To estimate the electric field intensity of the needle-tip electrode sample is crucial for electrical
1
tree initiation, a simulation of the tip electric field was tested, as shown in Figure 8. The model
parameters are consistent with the sample parameters used in the experiment, and the voltage of the
needle electrode is +60 kV. It can be seen from Figure 8a that the electric field intensity near the needle
electrode is obviously0large, 10the 20
maximum
30 position
40 50 is at60the 70
needle80tip (see
90 Figure
100 8b), and the
8
Polymers 2020, 12, 2215 7 of 11

However, the damaged area of the electrical tree under 5% duty cycle is smaller. Figure 4a shows
that the electrical tree has a hemispherical pine-like structure at 5% duty cycle and so many electric
branches have a certain shielding effect on the electric field intensity of the needle tip, hindering the
growth of electrical tree. Therefore, the tree grows slowly and the area of damage to the material
is small.

4. Discussion
To estimate the electric field intensity of the needle-tip electrode sample is crucial for electrical
tree initiation, a simulation of the tip electric field was tested, as shown in Figure 8. The model
parameters are consistent with the sample parameters used in the experiment, and the voltage of
the needle electrode is +6 kV. It can be seen from Figure 8a that the electric field intensity near the
needle electrode is obviously large, the maximum position is at the needle tip (see Figure 8b), and the
maximum
Polymers 2020,value is about
12, x FOR 5.50 × 108 V/m.
PEER REVIEW 8 of 13

(a) (b)
Figure 8. Electric field intensity distribution diagram of the needle-plate electrode. (a) Equipotential
surface distribution and (b) maximum field intensity position.

The electric field intensity at the tip of the needle used in the experiment can also be obtained
from Equation (1) [22]:
2V
E= 2V , (1)
E= r ln(4d/r) , (1)
r ln(4𝑑/𝑟)
where E is the electric field near the needle tip, V the voltage amplitude at the needle tip of the sample,
where E is the electric
r the curvature of thefield neardthe
tip, and theneedle tip,between
distance V the voltage amplitude
the needle at the
tip and theneedle tip of the sample,
plate electrode. In this
rstudy,
the curvature of the tip, and d the distance between the needle tip and the
V, r and d are ± 6 kV, 3 ± 0.2 µm and 1.8 ± 0.1 mm, respectively. The electric field intensity plate electrode. In this
study, V, r and d are ± 6 kV, 3 ± 0.2 μm and 1.8 ± 0.1 mm,
around the needle tip obtained by Equation (1) is approximately 5.14 × 10 V/m. respectively. 8The electric field intensity
around the needle
Equation (1) istipanobtained by Equation
approximate (1) isequation.
calculation approximately It can5.14 108 V/m.
be ×seen that the calculated value
Equation
5.14 × 108 V/m (1) is an
of the approximate
equation is quitecalculation
close to theequation.
simulatedItelectric
can be field
seen intensity
that the calculated
value 5.50 × value 5.14
108 V/m,
×indicating
108 V/m that
of the equation is quite close to the simulated electric field intensity value 5.50
the extremely high electric field may reduce the potential barrier on the tip surface and result × 10 8 V/m,

indicating that the extremely

in the Fowler–Nordheim higheffect,
tunneling electric fieldcan
which may reduce
induce the potential
electron emission barrier
from theonneedle
the tip
tipsurface
[23,24].
and result in the Fowler–Nordheim tunneling effect, which can induce
The electron emission current density (j) under high electric field can be obtained by Equation (2)electron emission from
[25]:the
needle tip [23,24]. The electron emission current density (j) under high electric field can be obtained
by Equation (2) [25]:
r
ϕ− e3 E
4πε0 εr
[− ]
j = AT2 e (kT) , (2)
[ ]
(2)
where A, T, ϕ, E, ε0 εr and k are the constant 𝑗 = 𝐴𝑇 e
depending( ) ,
on metal parameters, metal temperature in
Kelvin, metal work function, electric field
where A, T, φ, E, ɛ0ɛr and k are the constant
√ of the needle tip, dielectric constant andmetal
depending on metal parameters, Boltzmann’s constant.
temperature in
Obviously, the ln j is linearity to the E and thus the electron emission current density
Kelvin, metal work function, electric field of the needle tip, dielectric constant and Boltzmann’s increases with
increasingObviously,
constant. electric fieldtheintensity. Therefore,
ln𝑗 is linearity a higher
to the √𝐸 andelectric
thus field will giveemission
the electron rise to a current
higher electron
density
emission density.
increases with increasing electric field intensity. Therefore, a higher electric field will give rise to a
higher electron emission density.
Figure 9 is showing the relation between space charges and the electric field near the tip of needle
electrodes, where a is a negative voltage, b the rising flank, c the positive voltage and d the falling
flank. E0 is the internal electric field due to external voltage and Eq the electric field produced by the
space charge around the needle. Ei(t) is the total internal electric field, and e is an electron unit charge.
Polymers 2020, 12, 2215 8 of 11

Figure 9 is showing the relation between space charges and the electric field near the tip of needle
electrodes, where a is a negative voltage, b the rising flank, c the positive voltage and d the falling flank.
E0 is the internal electric field due to external voltage and Eq the electric field produced by the space
charge around the needle. Ei(t) is the total internal electric field, and e is an electron unit charge.
Polymers 2020,
2020, 12,
12, xx FOR
FOR PEER REVIEW
REVIEW 99 of
of 13
13
Figure 9 isPolymers
explaining the PEER
flow chart of these processes as mentioned in Figure 10.

bb bb cc cc
dd dd
aa aa

- -/+ + +/ -
ee e ++
--
e ++
++ --
ee ++
--
①
① -- ②
② ee ③
③ ②
ee ②
EEi(t) E00 EEqq
i(t) E EEi(t)
i(t) E0 Eq
E0 Eq EEi(t) E00 EEqq
i(t) E
④
④ EEi(t)
i(t) E0 Eq
E0 Eq ⑤
⑤
Aluminum plate
Aluminum plate Aluminum plate
Aluminum plate Aluminum plate
Aluminum plate Aluminum plate
Aluminum plate

(a)
(a) (b)
(b) (c)
(c) (d)
(d)

Figure 9. Change of space charge and electric field under square wave voltage. (a) Negative voltage,
Figure 9.
Figure 9. Change
Change ofof space
space charge
charge and
and electric
electric field
field under
under square
square wave
wave voltage.
voltage. (a)
(a) Negative
Negative voltage,
voltage,
(b) rising flank,
(b) rising
rising(c)flank,
positive voltage,
(c) positive
positive (d)(d)
voltage, falling
falling flank (elliptical
flank (elliptical
(elliptical green
green spheres:
spheres: polymerpolymer
molecules;molecules;
(b) flank, (c) voltage, (d) falling flank green spheres: polymer molecules;
green spheres:
green small molecule
green spheres:
spheres: small chains;
small molecule
molecule yellow
chains;
chains; yellowballs:
yellow photons;
balls: photons;
balls: photons; redred
red balls:
balls:
balls: hot electrons).
hot electrons).
hot electrons).

Fowler-Nordheim tunnel
Fowler-Nordheim tunnel effect
effect

Negative
Negative emission and
①Electron emission
①Electron and
voltage
voltage collide with
collide with molecules
molecules

Rising
Rising ions and
②Positive ions
②Positive and electrons
electrons
flank
flank recombine and
recombine and release
release photons
photons

Positive
Positive induce ionization
③Photons induce
③Photons ionization Photons induce
④ Photons
④ induce transition
transition to
to an
an excited
excited
voltage
voltage on other
on other molecules
molecules or
or atoms
atoms state on
state on other
other molecules
molecules oror atoms
atoms

The molecules
④ The
④ molecules or
or atoms
atoms The molecules
④ The
④ molecules oror atoms
atoms
return to
return to the
the ground
ground state
state in excited
in excited state
state emit
emit electrons
electrons
with radiating
with radiating photons
photons leading to
leading to hot
hot electrons
electrons

Falling
Falling ions and
②Positive ions
②Positive and electrons
electrons electrons by
⑤Emit electrons
⑤Emit by hitting
hitting
flank
flank recombine and
recombine and release
release photons
photons the needle
the needle tip
tip

Figure 10. The flow chart of the process in Figure 9.

Figure
Figure 10. 10.
The The flowchart
flow chart of
ofthe
theprocess in Figure
process 9.
in Figure 9.
In Figure
In Figure 9a,
9a, the
the EE00 higher
higher than
than 10
1088 V/m
V/m could
could lead
lead to
to aa considerable
considerable Fowler–Nordheim
Fowler–Nordheim tunnel
tunnel
In Figure
effect 9a,
effect (see (1)) E
(see the
(1)) 0 higher
and
and induce
than
induce the 108 V/m
the emission
emission could lead
of electrons
of electrons fromto
from theatip
the
considerable
tip of the
of needle Fowler–Nordheim
the needle [24–27]. Ionization
[24–27]. Ionization tunnel
effect (seeoccurs
Equation
occurs when (1))
electronsand induce
with high the
energy emission
collide with of electrons
molecules in the from
polymerthe tip of
(process①).
when electrons with high energy collide with molecules in the polymer (process①). In Figure the
In needle
Figure [24–27].
9b,
Ionization9b, the direction
occurs
the direction
whenof of the electric
electricwith
electrons
the field of
field of the space
high
the space
energy charge EEqq in
collide
charge in the
the polymer
withpolymer cannotin
molecules
cannot change immediately
the polymer
change immediately (process O).
1
just after
just after the
the voltage
voltage polarity
polarity change
change at at the
the rising
rising flanks
flanks of of square
square wave
wave voltages.
voltages. At
At this
this time,
time, the
the
In Figure 9b, the direction of the electric field of the space charge Eq in the polymer cannot change
immediately just after the voltage polarity change at the rising flanks of square wave voltages. At this
time, the applied electric field intensity E0 will have the same direction as the Eq and will result in the
substantial increase of the total electric field Ei(t) [17]. As shown in Figure 9b,c, the electrons move
Polymers 2020, 12, 2215 9 of 11

toward the needle tip, causing the positive and negative ions to recombine and release the energy in
the form of photons (process O). 2 These photons may induce more ionization (process O) 3 or transition
to an excited state on other molecules or atoms near the tip. The unstable excited state would return
to the ground state with radiating photons after around 10−8 s, or emit electrons by the Auger effect
(process O)4 and transfer energy to other electrons, leading to some of the electrons in the normal state
changing to hot electrons [24,25].
In Figure 9c, the positive ions and electrons recombine and release photons and the needle tip
emits a considerable number of electrons under negative voltage. On the one hand, it collides with
polymer molecules and induces ionization (process O) 1 [28]. On the other hand, the photons generated
by the recombination of positive ions and electrons could generate more emitted electrons (process O) 5
by hitting the needle tip (cathode) [29]. Both of these aspects damage the polymer molecular chain.
Energetic hot electrons may hit the polymer molecular chain, destroy and put the molecular
chain into a free radical. The formation of free radicals in polymer would lead to a free-radical chain
reaction and subsequently the generation of low-molecular products (e.g., hydroxyl, epoxy, etc.) [30,31].
The local area with the low-molecular products is called low-density regions [32]. Then, the average
free path of electrons in the low-density region increases [13] and the small-molecule products in
low-density regions are more prone to collision ionization. Finally, the ionization of polymer molecules
is enhanced [33], which further aggravates the destruction of polymer molecular chains.
Accordingly, it can be summarized that the formation of hot electrons, ions recombination,
and ionization by collision in low-density regions could damage the polymer molecular chain and
form a conductive channel inside the polymer. Therefore, the electrical tree is generated near the
tip of the needle. It can be seen from Table 2 that square wave voltages with smaller duty cycles
(the duration of the negative voltage is greater than the positive voltage) lead to higher electrical tree
initiation probability. A longer duration of negative voltage means more electrons will be injected into
the polymer. As small-size electrons have a longer free path compared with ions, the kinetic energy
of electrons is much greater than that of ions. Hence, electrons increase the probability of ionization
by colliding with the polymer molecular chain, which would break the polymer molecular chain.
Therefore, the electrical tree is easier to form under smaller duty cycles.
As the electrical tree branches of epoxy resin are conductive trees below the glass transition
temperature [15], the electric field near the needle tip becomes extremely complicated as the electric
branches grow. Equation (1) shows that, on the one hand, the existence of electric branches at the
needle tip can shorten the distance between the high-voltage and ground electrode, resulting in an
increased electric field intensity. On the other hand, these small branches would lead to a uniform
electric field near the needle tip, weakening the electric field intensity. Eventually, the electrical tree
grows substantially linearly, as shown in Figure 5.
As longer negative voltage duration (smaller duty cycle) induce more electrons near the needle tip.
The electrons would collide with the epoxy resin molecular chain or its small molecule products and
generate smaller molecular chains or free radicals, expanding the ionization area and forming a larger
low-density region near the needle tip. Therefore, the length and the damaged area of electrical tree
branches will increase under small duty cycles.
However, denser electric branches have a stronger weaken effect on the electric field [34], which is
equivalent to increasing the needle tip curvature r in Equation (1). Thus, the growth of the electrical
tree in the direction of the electric field was suppressed, and a growth stagnation period occurred to a
certain extent under 5% and 10% duty cycles. And the range of damage to the epoxy resin is shortened
under these duty cycles.

5. Conclusions
The statistical characteristics of electrical trees for epoxy resin using needle-plate electrodes were
investigated experimentally at repetitive square wave voltage. The square wave voltage duty cycle has
a considerable effect on the initiation and growth of electrical tree branches due to the total electrons
Polymers 2020, 12, 2215 10 of 11

emitted from the needle tip under positive and negative voltages. The following conclusions can be
drawn from the study.
1. The number of electrical tree initiations varies by varying the duty cycles. The probability of tree
initiation under small duty cycles is higher than that under high duty cycles with the treeing
time increasing.
2. Repetitive square wave voltage duty cycle has a significant influence on the electrical tree
morphology in epoxy resin. More electron emission and accumulation at square wave voltage
with smaller duty cycles (longer negative voltage duration) could break molecular chains in
epoxy with a higher probability. Moreover, the electrical trees develop from a pine-like into a
dense branch-like structure and finally develop into a sparse branch-like structure at square wave
voltages with increasing duty cycle.
3. When the duty cycle of the square wave voltage is higher than 30%, the electrical tree grows
linearly and the electrical tree length, width and damage area decrease when increasing the
duty cycles. However, the growth of the electrical tree may be suppressed when the duty
cycle is less than 10%, due to the shielding effect of pine-like and dense branch-like structures,
which generate hemispherical distribution near the needle-tip.
Author Contributions: Methodology and idea by P.W.; Conducted experiments by S.H.; software, and validation by
S.H. and P.W.; formal analysis and investigation by S.A. and K.Z.; resources and data curation by P.W.; writing—original
draft preparation, S.H. and S.A.; writing—review and editing by M.T.N.; Y.C.; H.D.; I.U.H. and M.S.J. All authors
have read and agreed to the published version of the manuscript.
Funding: This work was supported by the National Natural Science Foundation of China under Grant No. 51977134.
Conflicts of Interest: The authors declare no conflict of interest that they have no known competing financial
interests or personal relationships that could have appeared to influence the work reported in this paper.

References
1. Akram, S.; Wang, P. Impact of impulse voltage frequency on the partial discharge characteristic of electric
vehicles motor insulation. Eng. Fail. Anal. 2020, 116, 104767. [CrossRef]
2. Tousi, M.M.; Ghassemi, M. Electrical Insulation Packaging for a 20 kV High Density Wide Bandgap Power
Module. IEEE Energy Convers. Congr. Expo. 2019, 4162–4166. [CrossRef]
3. Van Erp, R.; Kampitsis, G.; Matioli, E. Efficient microchannel cooling of multiple power devices with compact
flow distribution for high power-density converters. IEEE Trans. Power. Electron. 2020, 35, 7235–7245. [CrossRef]
4. Akram, S.; Gao, G.Q. Degradation mechanism of Al2 O3 nano filled polyimide film due to surface discharge
under square impulse voltage. IEEE Trans. Dielectr. Electr. Insul. 2015, 22, 3341–3349. [CrossRef]
5. Shang, C.; Sun, H.; Hao, Y. Investigation of the Propagation of Electrical Trees in a Polymer Matrix in the
Corona Condition. Polymers 2018, 10, 984. [CrossRef] [PubMed]
6. Li, M.; Kai, Z.; Li, Y.; Xie, M.; Huang, Y.L. An Improved De-Noising Method for Partial Discharge Signals
Contaminated by White Noise Based on Adaptive Short-Time Singular Value Decomposition. Energies 2019,
12, 3465.
7. Li, C.Y.; He, J.L. Surface morphology and electrical characteristics of direct fluorinated epoxy-resin/alumina
composite. IEEE Trans. Dielectr. Electr. Insul. 2016, 23, 3071–3077. [CrossRef]
8. Li, C.Y.; Hu, J. Surface charge migration and dc surface flashover of surface-modified epoxy-based insulators.
J. Phys. D 2017, 50, 65301. [CrossRef]
9. Li, C.Y.; Hu, J. Fluorine gas treatment improves surface degradation inhibiting property of alumina-filled
epoxy composite. AIP Adv. 2016, 6, 25017. [CrossRef]
10. Rodríguez-Serna, J.M.; Albarracín-Sánchez, R.; Carrillo, I. An Improved Physical-Stochastic Model for
Simulating Electrical Tree Propagation in Solid Polymeric Dielectrics. Polymers 2020, 12, 1768. [CrossRef]
11. Sheng, B.; Catterson, V.M. Investigation into pulse sequence analysis of PD features due to electrical tree
growth in epoxy resin. IEEE Int. Conf. Dielectr. 2016, 2, 744–747.
12. Nakamura, S.; Kumada, A. Electrical treeing in silicone gel under repetitive voltage impulses. IEEE Trans.
Dielectr. Electr. Insul. 2019, 26, 1919–1925. [CrossRef]
Polymers 2020, 12, 2215 11 of 11

13. Du, B.X.; Su, J.G. Tree growth characteristics of epoxy resin in LN2 under DC superimposed pulse voltage.
IEEE Trans. Appl. Supercond. 2018, 28, 1–5. [CrossRef]
14. Iddrissu, I.; Zheng, H.L.; Rowland, S.M. Electrical tree growth in epoxy resin under DC voltages. IEEE Int.
Conf. Dielectr. 2016, 2, 820–823.
15. Tian, M.; Du, B.X.; Su, J.G. Electrical tree in epoxy resin under combined DC-harmonic voltage. Cond. Monit. Diagn.
2018, 1–4. [CrossRef]
16. Wang, P.; Xu, H. Effect of repetitive impulsive voltage duty cycle on partial discharge features and insulation
endurance of enameled wires for inverter-fed low voltage machines. IEEE Trans. Dielectr. Electr. Insul. 2017,
24, 2123–2131. [CrossRef]
17. Akram, S.; Wu, G.N. Cavity and interface effect of PI-Film on charge accumulation and PD activity under
bipolar pulse voltage. J. Electr. Eng. Technol. 2015, 10, 2089–2098. [CrossRef]
18. Akram, S.; Wu, G.N.; Gao, G.Q. Effect of surface discharge on nano filled polyimide film under square
voltage. Electr. Insul. Conf. 2015, 226–229. [CrossRef]
19. Tanaka, T. Charge transfer and tree initiation in polyethylene subjected to AC voltage stress. IEEE Trans.
Dielectr. Electr. Insul. 1992, 27, 424–431. [CrossRef]
20. Chen, G.; Tham, C.H. Electrical treeing characteristics in XLPE power cable insulation in frequency range
between 20 and 500 Hz. IEEE Trans. Dielectr. Electr. Insul. 2009, 16, 179–188. [CrossRef]
21. Du, B.X.; Tian, M. Temperature gradient dependence on electrical tree in epoxy resin with harmonic
superimposed DC voltage. IEEE Trans. Dielectr. Electr. Insul. 2020, 27, 270–278. [CrossRef]
22. Han, T.; Zhu, L. Magnetic-field-dependent electrical tree under impulse-superimposed DC voltage at
low temperature. IEEE Trans. Appl. Supercond. 2019, 29, 1–5. [CrossRef]
23. Bhutta, M.S.; Yang, L. Simulation of thickness controlled DC breakdown of XLPE regulated by space charge
& molecular chain movement. IEEE Trans. Dielectr. Electr. Insul. 2020, 27, 1143–1151.
24. Tahir, F.; Butt, T.; Nazir, M.T.; Hussain, H.; Phung, B.T.; Akram, S.; Bhutta, M.S.; Yasin, G. Physical, thermal and
partial discharge evaluation of nano alumina filled silicone rubber in inclined plane test. CSEE J. Power
Energy Syst. 2020, 1–8. [CrossRef]
25. Zhang, Y.X.; Zhang, L. DC electrical tree initiation in silicone rubber under temperature gradient. IEEE Trans.
Dielectr. Electr. Insul. 2018, 25, 1142–1150. [CrossRef]
26. Akram, S.; Castellon, J.; Agnel, S.; Khan, Z.M. Space charge analysis of multi-structure polyimide films
using TSM. In Proceedings of the 2018 IEEE Conference on Electrical Insulation and Dielectric Phenomena
(CEIDP), Cancun, Mexico, 21–24 October 2018; pp. 34–37.
27. Akram, S.; Ruijin, L.; Nazir, M.T.; Yang, L. Space Charge Characteristics of LDPE/MMT Nano-Composite
Insulation Material under different Fields and Temperatures. Int. J. Electr. Comput. Sci. IJECS-IJENS 2012,
12, 50–56.
28. Akram, S.; Yan, Y. Influence of nano layer interface on space charge behavior and trap levels using polyimide
sandwich structure. IEEE Trans. Dielectr. Electr. Insul. 2017, 24, 3505–3514. [CrossRef]
29. Wang, Y.; Li, G.; Wu, J. Effect of temperature on space charge and periodic grounded DC tree in
cross-linked polyethylene. Cond. Monit. Diagn. 2016, 48–51. [CrossRef]
30. Du, B.X.; Han, T. Electrical tree characteristics in silicone rubber under repetitive pulse voltage. IEEE Trans.
Dielectr. Electr. Insul. 2015, 22, 720–727. [CrossRef]
31. Wang, Y.; Feng, C. The Development of electrical tree discharge in epoxy resin impregnated paper insulation.
IEEE Access. 2020, 8, 69522–69531. [CrossRef]
32. Iddrissu, I.; Zheng, H. DC electrical tree growth in epoxy resin and the influence of the size of inceptive
AC trees. IEEE Trans. Dielectr. Electr. Insul. 2017, 24, 1965–1972. [CrossRef]
33. Zydron, P.; Bonk, M.; Roehrich, J. Observation of electrical treeing processes in silicon insulation samples
using of transformed PD phase-resolved patterns. Prog. Appl. Electr. Eng. 2019, 1–4. [CrossRef]
34. Zeng, J.; Song, J.; Lei, Z. The effects of cavity on the growth of electrical trees in XLPE. IEEE Int. Conf. High
Volt. Eng. Appl. 2016, 1–4. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).