Electrical Engineering in Japan, Vol. 193, No.

2, 2015
Translated from Denki Gakkai Ronbunshi, Vol. 134-A, No. 11, November 2014, pp. 578–584

Transient Charging Characteristics in Surface Flashover Development Process due

to Explosive Electron Emission in Vacuum

YUSUKE NAKANO,1 HIROKI KOJIMA,1 KENJI TSUCHIYA,2 and NAOKI HAYAKAWA1

1 Nagoya University, Japan
2 Hitachi Ltd., Japan

SUMMARY are sensitive to electrode surface conditions; in particular,

surface discharge inception in an electrode configuration
For the development of higher voltage vacuum circuit with an electrode–insulator gap is believed to be caused
breakers and vacuum interrupters, surface discharge in a mainly by electron emission from the electrode. That is,
vacuum is one of the most significant issues. We mea- the development of surface discharge in a vacuum involves
sured the charge activity on alumina dielectrics in a vacuum complex charging phenomena on the solid insulator surface
through ultrahigh-speed optical and electrical measure- depending on the electrode’s geometrical shape, the mate-
ments. We found that surface flashover in a vacuum starts rial of the insulator, its surface conditions, and so on [4, 5].
with an explosive electron emission (EEE) with higher In this connection, there is a need for explanation of the
current density, which consists of electron emission from mechanisms of surface discharge in a vacuum with regard
the cathode, and results in surface flashover development. to charging phenomena, and to establishment of techniques
The transient charge on the alumina dielectrics formed by to suppress surface discharges based on such mechanisms.
electrons emitted by EEE influences the distortion of elec- We investigated the evolution of surface discharges in
tric field distribution around the cathode rod and then the a vacuum under a negative impulse voltage, and classified
subsequent flashover development. In particular, in the case the discharge mechanisms with respect to charge behavior
of a shorter gap between the cathode rod and the alumina on alumina dielectrics [6]. We were able to clarify the
dielectrics, the electron emission may be suppressed by the process of predischarge caused by electron emission from
relaxation effect of the electric field due to transient charge electrodes, and found that the inception of the subsequent
accumulation. C⃝ 2015 Wiley Periodicals, Inc. Electr Eng flashover (FO) is due to the transition from field electron
Jpn, 193(2): 10–17, 2015; Published online in Wiley Online emission in the predischarge to explosive electron emission
Library (wileyonlinelibrary.com). DOI 10.1002/eej.22739 (EEE) [7] caused by local heat-up of electron emission
sites on the cathode [6]. In this paper, we consider the
Key words: vacuum; surface discharge; flashover; mechanism of impulse surface FO in a vacuum with a focus
electron emission; charging; alumina dielectrics. on charge behavior on dielectrics caused by EEE, and in
particular, transient charging in the course of surface FO
development.
1. Introduction

The vacuum has attracted attention as an excellent

electrical insulator, and vacuum circuit breakers (VCBs) 2. Experimental System
are widely used electrical devices due to this high dielectric
strength. As regards vacuum interrupters (VIs), an alterna- 2.1 Experimental system
tive to SF6 gas interrupters from the viewpoint of environ-
ment friendliness, higher voltage and more compact design The experimental system is shown schematically
are required [1]. in Fig. 1. The vacuum chamber was a cylinder made up
One of the vacuum discharge phenomena that present of stainless steel (SUS304) with a diameter of 700 mm, a
problems in VI is surface discharge on solid insulators height of 500 mm, and a volume of 275 𝓁. The vacuum
[2, 3]. The discharge inception characteristics in vacuum chamber was connected to a three-stage glass vacuum
C⃝ 2015 Wiley Periodicals, Inc.

10
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 1. Experimental procedure and measurement system. [Color figure can be viewed in the online issue, which is available
at wileyonlinelibrary.com.]

bushing with a height of 1073 mm and an outer diameter of (SUS304). The creeping distance between the electrodes
240 mm. The pressure inside the chamber was maintained was set to 75 mm. A grounded electrode (100 × 100 mm)
at approximately 10−5 Pa. with electrostatic probes (A to C) was attached to the back
A negative standard lightning impulse voltage of the alumina dielectric. The probes (10 × 10 mm) were
(–1.2/50 μs) was generated by an impulse generator and ap- placed at distances of 0, 20, and 40 mm below the rod
plied to an electrode configuration inside the vacuum cham- electrode, and the current waveforms induced by charge be-
ber, thus producing surface discharge. The applied voltage havior on the dielectric were measured through 50 Ω match-
and current induced by charge behavior on the dielectric ing resistors connected to each probe. Different electrodes
were recorded by an oscilloscope, using, respectively, a and alumina dielectrics were used in every experiment. The
voltage divider and an electrostatic probe placed behind the absence of static charge on the dielectrics prior to voltage
dielectric. application was confirmed by a surface potential meter. The
In addition, light emission images in the course of up-and-down method was employed in voltage application,
surface discharge development were acquired by a digital but no increase in the withstand voltage associated with
camera provided with an image intensifier. The exposure a cathode conditioning effect due to repetitive breakdown
time and shot timing could be controlled arbitrarily by was observed.
means of a pulse generator, thus making possible ultrahigh-
speed shooting of light emission images.

3. Experimental Results and Discussion

2.2 Electrode configuration and measurement
method 3.1 Development of negative impulse surface
discharge
The electrode system arranged inside the vacuum
chamber is shown in Fig. 2. The dielectric samples were Examples of light emission images and waveforms of
alumina ceramics (Al2 O3 , HA-92) of a type widely em- surface discharges in a vacuum are shown in Fig. 3. As can
ployed for insulating tubes in VIs. The alumina dielectrics be seen from Fig. 3(a), FO occurs on the alumina dielec-
had a size of 150 × 150 × 5 mm, a relative permittivity tric surface between the high-voltage electrode (cathode)
εr of 9.0, and an average surface roughness R of 0.69 μm. and opposing grounded electrode (anode). In Fig. 3(b), the
The high-voltage electrode (cathode) was a rod electrode surface discharge phenomena can be divided into predis-
made up of stainless steel (SUS304, machined finish) 2 mm charge and subsequent FO development [6]. Predischarge,
in diameter; a vacuum gap g = 2, 5, and 10 mm was pro- a precursor phenomenon of surface FO, originates in elec-
vided between the electrode and alumina dielectric. The tron emission from the cathode. Surface FO starts when
opposing grounded electrode (anode) was a plate electrode field electron emission or thermionic emission transitions

11
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 3. Still image and voltage/current waveforms of
surface flashover. [Color figure can be viewed in the
online issue, which is available at wileyonlinelibrary.com.]

Fig. 2. Electrode configuration. [Color figure can be

viewed in the online issue, which is available at
wileyonlinelibrary.com.]

Fig. 4. Peak current of probe A as a function of square

to EEE [7], when predischarge causes local heat-up of mi-
root of maximum electric field at EEE inception. [Color
croasperities on the cathode.
figure can be viewed in the online issue, which is available
Surface FO starts with a sharp increase in the elec-
at wileyonlinelibrary.com.]
tron current density caused by EEE, as can be seen from
the buildup of the probe current in Fig. 3(b). Because the
current waveforms are measured by all probes A, B, and
C, we may assume that the emission current due to EEE on the electrode surface at EEE inception and the peak
spreads to the whole surface of the alumina dielectric. After current measured by probe A at gap sizes of g = 2, 5, and
that, the FO process develops for about 1000 ns, after which 10 mm between the rod electrode and alumina dielectric.
the voltage falls to zero and breakdown occurs between the Here, E a was calculated by the charge simulation method
cathode and anode. (CSM). As can be seen from the diagram, the peak probe
current increases with the EEE inception field. EEE occurs
due to local heat-up of microasperities on the cathode [8];
3.2 EEE inception field and EEE current therefore, we can envision a kind of thermionic emission.
That is, the electric field dependence of EEE can be rep-
The Schottky plot in Fig. 4 illustrates the relationship resented by the following Richardson–Dushman equation
1∕2
between the square root E a of the maximum electric field with respect to the decrease Δϕ in the potential barrier due

12
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
to a strong electric field (Schottky effect):
[ ]
4πmk 2 e 2 − (ϕ − Δϕ)
J = T exp , (1)
h3 kT
√
e3 E
Δϕ = ,
4πϵ0
where J is the emission current density, m is the electron
mass, e is the elementary charge, h is Planck’s constant,
k is the Boltzmann constant, ϕ is the work function, T is
the temperature of the metal, and E is the electric field on
the metal surface.
Equation (1) indicates that the EEE current density
J varies exponentially in proportion to the square root of
the electric field E on the electrode surface. In the diagram,
the plotted points lie close to the respective approximation
lines. The approximation lines for g = 5 mm and 10 mm
almost coincide (solid lines). But the approximation line
for g = 2 mm (dashed line) is shifted downward, which is
contradictory to Eq. (1), suggesting that the current density
J is uniquely determined by the relationship with the elec-
tric field E on the electrode surface (exp(E 1∕2 )). We discuss
this issue in Sections 3.3 and 3.4 based on the probe current
waveforms.

3.3 Electron emission from cathode

We first consider the EEE current waveforms. Ac-

cording to the mechanism of EEE, the emission current
from the cathode increases due to electrode heat-up. The
metal temperature rises because of Joule heating, and we
may expect an increase in the current density as a result
of positive feedback. However, multiple pulses due to EEE
are observed in the probe current waveforms in Fig. 3. This
means that EEE has a pulsed emission mechanism, and that
Fig. 5. Current waveforms of probe A for different gaps.
the pulse duration is about 50 ns to 100 ns.
[Color figure can be viewed in the online issue, which is
The pulsed shape of the EEE current can be explained
available at wileyonlinelibrary.com.]
by fusion or annihilation of microasperities (electron emis-
sion sites) on the electrode surface and by reemergence of
the sites. EEE exhibits high current density, which suggests Figure 5 shows the applied voltage waveforms and
fusion of microasperities on the electrode. The fusion of the current waveforms of probe A at g = 2, 5, and 10 mm.
microasperities might result in a decrease in the electric The maximum electric fields E a on the cathode surface at
field at electron emission sites, or discontinuation of elec- EEE inception at the various gap sizes are, respectively,
tron emission because of annihilation of electron emission –31.9, –30.6, and –32.1 kV/mm. The rise rate of the probe
sites. Thereafter, two or three EEE pulses were observed, current (di∕dt) is often 0.03 A/ns to 0.04 A/ns regardless
but these can be attributed to new emission of electrons of E a , as shown in Fig. 6. The diagram suggests that the
from adjacent microasperities when the electron emission mechanism of FO inception does not change with the length
area expanded due to heat conduction on the electrode sur- of the gap between the rod electrode and the alumina dielec-
face, or the formation of a plasma layer there [8]. In such tric, and that FO is initiated by EEE.
new electron emission sites too, fusion or annihilation oc- On the other hand, as shown in Fig. 7, the time to
curs repetitively. The annihilation of microasperities on the peak of the probe current at g = 2 mm is shorter than at g =
cathode was confirmed by the extinction of light emitting 5 mm and 10 mm. In addition, comparing the probe current
points on the alumina dielectric corresponding to electron waveforms in Figs. 5(a) to 5(c), current rise is inhibited in
emission sites [8]. midcourse at g = 2 mm (Fig. 5(a)). This may explain why

13
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 6. Histogram of di∕dt of probe current. [Color
figure can be viewed in the online issue, which is available
at wileyonlinelibrary.com.]

Fig. 7. Histogram of rise time to peak of probe current.

[Color figure can be viewed in the online issue, which is
available at wileyonlinelibrary.com.]

the peak current at g = 2 mm is smaller than at g = 5 and

10 mm in Fig. 4.
EEE current depends on the electric field as indicated
by Eq. (1), and therefore, the decrease in the electric field
at the electron emission sites can be attributed to fusion
of microasperities on the cathode as well as to the field
relaxation effect of the electrostatic charge that occurs tran-
siently on the alumina dielectric due to EEE. The electric Fig. 8. Light emission images at EEE inception (t =
field relaxation caused by this transient charging may have 0 ns) with exposure time of 20 ns. [Color figure can be
influenced the probe current waveform at g = 2 mm in viewed in the online issue, which is available at
Fig. 5(a). Below, we consider transient charging on alumina wileyonlinelibrary.com.]
dielectrics.

exposure time of 20 ns. The light emission is cathode

3.4 Transient charging on alumina dielectrics luminescence that occurs when electrons emitted from
the electrode collide with the alumina dielectric. As
In order to consider the possible influence of transient can be seen from the diagram, the range in which light
charging on current suppression at g = 2 mm, we examine emission is observed expands with increasing gap length.
the transient charge area on the dielectric and evaluate the Assuming that the electrons produced by EEE have a
electrostatic charge density. constant emission angle (solid angle), this expansion can
Discharge light emission images at different gap be attributed by an increase in the electron collision area on
lengths are shown in Fig. 8. The images were captured the alumina dielectric with increasing gap length between
immediately after EEE inception, with an image intensifier the rod electrode and the dielectric.

14
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 1. Surface potential with transient charging at each A transient negative charge is created below the rod
gap length for FEM calculation electrode (cathode) due to EEE. This is because in this
electrode configuration, a grounded electrode is attached
Gap [mm] 2 5 10 to the back of alumina dielectric, so that electrical lines
Applied voltage Va at EEE −26.6 −32.8 −34.5 of force are normally incident on the dielectric. When
inception [kV] electrons incident on alumina dielectrics have energies of
EEE inception electric −31.9 −30.6 −32.1 several tens of keV, the secondary electron emission coef-
field Ea [kV/mm] ficient is smaller than 1 [4], and secondary electrons are
Transient charge Q [nC] −1.40 −4.02 −4.35 not easily emitted from the dielectric. Therefore, electrons
Charging area S [mm2 ] 5.73 19.6 69.4 emitted from the cathode are trapped on the surface and
Transient charge density σ −0.24 −0.21 −0.06
inside the alumina dielectric at the light emission points,
[nC/mm2 ]
Capacitance C [pF] 0.092 0.266 1.11
thus producing a negative charge.
Surface potential Vs [kV] −15.2 −15.0 −3.92 We determined the surface potential Vs on the surface
(Vs /Va [%]) (57.2%) (45.8%) (11.4%) of the alumina dielectric in the presence of a transient
charge at g = 2, 5, and 10 mm as described in Table 1,

Fig. 9. Potential distributions with and without transient charge on alumina dielectrics. [Color figure can be viewed in the
online issue, which is available at wileyonlinelibrary.com.]

15
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and obtained the potential distributions shown in Fig. 9 4. Conclusions
by the finite element method (FEM). The transient charge
Q was found by time integration from EEE inception We investigated the process of surface FO develop-
(t = 0 ns) in the probe current waveforms (Fig. 5) to the ment on alumina dielectrics in a vacuum in terms of the
end of exposure of the light emission images (Fig. 8) so as EEE mechanism and transient charge, and arrived at the
to match the probe current waveforms with the exposure following conclusions:
timing. In addition, based on the luminescence intensity (1) The EEE current density depends on the cathode
distributions in Fig. 8, the charging area S was assumed surface field, and a transient charge is generated on the
to be the area of a circle with the diameter being the full dielectric.
width at half-maximum of the luminescence intensity, and (2) The transient discharge produced by EEE dis-
the electrostatic capacitance C was calculated from the torts the electric field on the dielectric surface from the
thickness of the alumina dielectric (5 mm) and its relative vertical to the horizontal direction, thus promoting FO.
permittivity (9.0). In Fig. 9, diagrams (a-1), (b-1), (c-1) (3) When the vacuum gap between the rod elec-
and (a-2), (b-2), (c-2) represent results obtained without trode and the alumina dielectric is narrow, an influence of
and with transient charge, respectively. The instantaneous transient charge on the dielectric is likely to appear and
voltage Va of EEE inception was set on the high-voltage to inhibit the emission current from the cathode that is
electrode. responsible for FO.
Under the above conditions, the transient charge den-
sity σ on the dielectric is –0.24 nC/mm2 at g = 2 mm,
REFERENCES
–0.21 nC/mm2 at g = 5 mm, and –0.06 nC/mm2 at g =
10 mm: that is, the transient charge density increases with 1. Okubo H. Development of electrical insulation tech-
decreasing gap length. niques in vacuum for high voltage vacuum interrupters.
Comparison of diagrams (a-1) and (a-2), (b-1) and 22nd International Symposium on Discharges and Elec-
(b-2), or (c-1) and (c-2) shows that the potential distribu- trical Insulation in a Vacuum, p 7–12, 2006.
tion near the rod electrode is affected, and that the electric 2. Latham R. High voltage vacuum insulation: Basic con-
field is distorted by transient charging. In particular, at g = cepts and technological practice, p 300–306. Academic
2 mm in diagram (a-2), the electric field around the tip Press, 1995.
of the rod electrode is relaxed due to the transient charge, 3. Miller HC. Flashover of insulators in vacuum. IEEE
which may inhibit electron emission from the cathode. On Trans Electr Insul 1993;28(4):512–527.
the other hand, at g = 5 mm in diagram (b-2) and g = 4. Kato K, Kato H, Ishida T, Okubo H, Tsuchiya K. In-
10 mm in diagram (c-2), the electric field around the tip of fluence of surface charges on impulse flashover charac-
the rod electrode undergoes little distortion when the poten- teristics of alumina dielectrics in vacuum. IEEE Trans
tial distribution changes due to the transient charge; thus, Dielectr Electr Insul 2009;16(6):1710–1716.
we may assume that EEE is not suppressed, and that the 5. Ishida A, et al. Influence of static electrification of alu-
EEE temporarily attenuates due to fusion of microasperities mina insulator on surface discharge path in vacuum.
on the rod electrode. Annual Conference of the IEEDJ, A-2-4, 2009. (in
Thus, it can be concluded that the approximation line Japanese)
(dashed line) for g = 2 mm in Fig. 4 moved downward 6. Nakano Y, Kojima H, Hayakawa N, Tsuchiya K, Okubo
from the lines for g = 5 mm and 10 mm because the EEE H. Pre-discharge and surface flashover of impulse sur-
current from the cathode was inhibited by the transient face discharge in vacuum. IEEE Trans Dielectr Electr
charge above probe A. In addition, the emission current, Insul 2014;1:403–410.
while being inhibited, increases with the EEE inception 7. Fursey GN. Field emission in vacuum micro-
field, because the transient charge required to relax the electronics. Appl Surf Sci 2003;215:113–134.
electric field increases with the applied voltage. The above 8. Mesyats GA, Proskurovsky DI. Pulsed electrical dis-
results suggest that when the gap between an electrode and charge in vacuum. Springer-Verlag, 1989.
a solid insulator is narrow, the emission current may be 9. Nakano Y, Kojima H, Hayakawa N, Tsuchiya K, Okubo
inhibited due to electric field relaxation near the cathode H. Transition mechanism from pre-discharge to surface
caused by the transient charge. flashover in vacuum. 18th International Symposium on
High Voltage Engineering, p 1366–1370, 2013.

16
 15206416, 2015, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/eej.22739 by University Of Liverpool, Wiley Online Library on [01/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
AUTHORS (from left to right)

Yusuke Nakano (student member) completed the first part of the doctoral program in electronic and information systems
at Nagoya University (Graduate School of Engineering) in 2013 and he is now in the second stage of the program. His research
interests include insulation properties in vacuum. He is a member of IEEE.

Hiroki Kojima (member) completed the doctoral program at Nagoya University (Graduate School of Engineering) in
2004 and joined the faculty of the university as a research associate in the EcoTopia Science Institute, becoming an associate
professor in 2007. Since 2014, he has been an associate professor in the Graduate School of Engineering. D.Eng. His research
interests include insulation and discharge in vacuum, gas insulation technology, and superconducting power applications. He is
a member of IEEDJ, JSPF, JPS, and IEEE.

Kenji Tsuchiya (member) received a bachelor’s degree from Waseda University in 1981. Since 2001 he has been employed
with Hitachi, Ltd. His current research interests include midvoltage switches and vacuum interrupters.

Naoki Hayakawa (member) completed the doctoral program at Nagoya University (Graduate School of Engineering)
in 1990 and joined the faculty of the university as a research associate. Since 2008 he has been a professor in the EcoTopia
Science Institute. He was a visiting researcher at Karlsruhe Research Center from 2001 to 2002. D.Eng. He is a member of
IEEE, CIGRE, CSJ, and IEEDJ.

17