Full text of DOI:10.2514/2.3838

JOURNAL OF SPACECRAFT AND ROCKETS
–
Vol. 39, No. 3, May June 2002
Plasma Response to Arcing in Ionospheric Plasma
Environment: Laboratory Experiment
Mengu Cho,¤ Raju Ramasamy,† and Masayuki Hikita‡
Kyushu Institute of Technology, Kitakyushu 804, Japan
and
Koji Tanaka§ and Susumu Sasaki¶
Institute of Space and Astronautical Sciences, Kanagawa 229, Japan
Laboratory experiments are carried out to study the response of plasma to a sudden potential change induced
by arcing on a solar array in low-Earth-orbit plasma environment. A solar array is biased negatively in a plasma
chamber with insulator simulating coverglass or thermal paint located remotely from the arc point. When an arc
occurs, electrons are ejected from the arc point to the surrounding plasma, producing a negative sheath. Shortly
after the arc onset, the in ux of positive charge into the circuit causes a jump of the potential, and the positive
sheath is formed near the insulatorsurface. If the positive sheath meets the negativesheath near the arc point while
the conductivity of the arc plasma is still high, a current path is formed between the arc point and the insulator
surface. The current path supplies the positive charge on the insulator surface to the arc current and feeds energy
to arc plasma. Experimental results show that there is a safety distance beyond which arc plasmas cannot form
the current path.
Nomenclature
magnetic Ž eld, G
capacitance, F
capacitanceof insulator surface to the electrode, F
distance between array and capacitance, cm
neutralizationcurrent, A
peak of discharge current, A
particle mass, kg
Áe
Ás
Á1
Á2
=
=
=
=
electrode potential, V
insulator surface potential, V
potential of the positive end of solar array, V
potential of the negative end of solar array, V
B
C
Cs
d
In
Ip
m
n
Qd
Qs
R
Te
Tp
Tpr
t
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
I. Introduction
(
)
O promote industrial use of low Earth orbit LEO , such as
manufacturing,sightseeing,or power generation,the power of
number density, m¡3
charge  own in one arc, C
charge stored on insulator surface, C
resistance, Ä
electron temperature, eV
peak time of discharge current, s
peak time of probe signal, s
time, s
potential, V
solar array output voltage, V
bias voltage, V
probe signal, V
T
a large LEO platform that follows the International Space Station
(
)
ISS may soon reach the levelof megawatts. In principle,the trans-
missionvoltagescalesto the squarerootofthe powerto be delivered
to minimizethe energylossduringpowertransmissionand the cable
mass. ISS with 100 kW is operated at 120 V. An experimentalsolar
power satellite with 10 MW needs to be operated at least 1000 V
(
)
Ref. 1 . Therefore, a space platform with 1-MW power must be
operated at approximately 400 V. To realize 400-V operation in
LEO, arcing caused by interaction between the spacecraft and the
surroundingLEO plasma must be overcome.
V
Varray
Vb
Vpr
y
z
1
When a spacecraft has a high voltage within its body and the
high voltage surface is exposed, most of the high voltage becomes
negative with respect to the electric potential of the surrounding
plasma. Figure 1 shows a circuit layout of typical spacecraft. The
spacecraftstructuralbody that is made of conductormaterial serves
as the grounding point of the circuit and is usually connectedto the
negative end of the array. Then, the spacecraft structural body has
a high negative potential. The spacecraftbody attracts positive ions
from the surroundingplasma, and the ions impact the body surface
and solar array. Many parts of spacecraftsurfaces are made of elec-
trical insulators, such as solar array coverglass or thermal coating
materials, which are often also electricalinsulatormaterials. There-
fore, when ions impact spacecraft dielectric surfaces, the surfaces
are positively charged. The positive charging continues until the
surface potential reaches a value where the ion current to the sur-
face is equal to the electron current, which is typically comparable
horizontal position, cm
axial position, cm
change of spacecraftpotentialdue to arc onset, V
V
Subscripts
e
i
n
=
=
=
electron
ion
neutral
Presented as Paper 2001-0955 at the AIAA 39th Aerospace Sciences
–
Meeting, Reno, NV, 8 11 January 2001; received 22 March 2001; revision
received 24 August 2001; accepted for publication 30 August 2001. Copy-
°c
right
2001 by the American Institute of Aeronautics and Astronautics,
Inc. All rights reserved. Copies of this paper may be made for personal or
internal use, on condition that the copier pay the $10.00 per-copy fee to
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA
01923; include the code 0022-4650/02 $10.00 in correspondence with the
CCC.
T
e
to electron temperature · and negligible compared to the body
potential . Therefore, the electricalinsulatoracts as a capacitance
that stores the positive charge on the surface facing the plasma.
It is known that arcing can occur once there is an opening in the
insulator where the underlying conductor with a negative potential
V
¤Associate Professor, Department of Electrical Engineering, Tobata-ku;
cho@ele.kyutech.ac.jp. Member AIAA.
2
¡
of 100 V or larger, typically, is exposed to space. The arcing
†Postdoctoral Fellow, Satellite Venture Business Laboratory, Tobata-ku.
‡Professor, Department of Electrical Engineering, Tobata-ku.
§Research Associate, Space Energy Division, Sagamihara.
¶Professor, Space Energy Division, Sagamihara.
is triggered by charging of insulator material via plasma and Ž eld
intensiŽcation at the conductor surface, especially at the junction
with the insulator, which is called a triple junction, because three
392

CHO ET AL.
393
RC
R
discharge, where
the plasma. This current path is essentially
is given by the resistance of plasma and circuit and is coverglass
RC
C
capacitance. This model of
discharge feeding the energy to the
arc plasma has been illustrated in Refs. 5 and 10.
This study is important in the sense that we need to know how
much energy an arc can have. It is also important to know the con-
dition by which an arc becomes more severe. Current paths 1 and
RC
2 in Fig. 2 feed the energy to the arc plasma through the
dis-
charge. As the arc plasma grows and extends over the positive end
of the array string, the plasma can short circuit the positive and
negative ends of the array, shown as current path 3 in Fig. 2. This
path 3 can be attached to the positive end of the same strings, as
shown in Fig. 2, or of a different string and depends on the cell
layout on the array surface. A typical solar array string can gen-
erate a current of several amperes at the solar intensity in LEO. If
the arc plasma has a sufŽ cient conductivity to maintain itself with
that amount of current, the arc current keeps  owing, leading to
destruction of the solar array circuit, a so-called sustained arc.3;4
Initial growth of the arc plasma, therefore, is very important. The
degree of growth determines whether an arc causes a fatal effect to
the spacecraft circuit. Even if arcs end as single pulses, accumula-
tive effects, such as surface deteriorationlike discoloration,peeling
off, damage, and others, are determined by how large each single
pulse is. In the presentpaper,we investigateboth how far the surface
charge of the unarced surface is neutralizedby this mechanismand
how the current path is connectedinto the spacecraftcircuitusing a
model system in the laboratory experiment.
Fig. 1 Schematic of electrical circuit of spacecraft and potential with
respect to the LEO plasma.
Recently, Cho et al.11 found that a current path between the arc
point and remotely located insulatoris formed by arc plasma. They
biased the combination of a solar array and an electrode wrapped
¡
by insulatorŽ lm to a negativepotentialas high as 700 V in a large
vacuum chamber of 2.5-m diam, where the LEO plasma condition
was simulated. They tested two cases of the distance between the
solar array and the electrode, 40 and 80 cm. The arc plasma could
connect the arc point on a solar array and the insulator on top of
the biasedelectrodeeven for the distanceof 80 cm. The arc plasma,
however, often failed to connect for the 80-cm case compared to
the 40-cm case. Also, the time for the arc plasma to connect the
two surfaces was longer for 80 cm than for 40 cm. A langmuir
probe was used to measure the plasma condition at various points
insidethe chamber,and it was found thatthe spacepotentialnearthe
solar array became negative and that the potentialnear the insulator
became positive, once an arc occurred at the solar array.
Based on these results, a model of the discharge path formation
was proposed, which is schematically shown in Fig. 3. Once an
arc occurs, electrons are ejected from the array to the surrounding
plasmaproducinganelectron-richnegativesheath.At thesametime,
injectionof positivechargeintothe circuitcausesa jump in the array
potentialand the electrodepotentialunder the insulatortoward zero
from the highly negative value. Before the arc, the insulatorsurface
potential is nearly zero, comparable to the electron temperature,
typically 1 eV, to have the zero current condition at a steady state.
Fig. 2 Schematic of current paths to the arc point.
(
materials with different electrical conductivity conductor, insula-
)
tor, and vacuum , meet there. There are many ways for a triple
junction to be formed. One example is an interconnector between
solar cells, where the triple junction is formed with the intercon-
nector exposed to space, the adhesive to attach coverglass, and the
vacuum. Another example is a hole in the thermal paint exposing
the underlying conductive surface produced by impact of a small
particleon the surface.Arcing raisesconcernswith electromagnetic
interference,surfacedeterioration,or even destructionof spacecraft
circuits, which is demonstrated in Refs. 3 and 4. How much the
arcing affects the spacecraft system depends on how much current
and energythe arcingcurrentcarriesinto the spacecraftcircuit. Pre-
vious experiments5¡8 indicatethat, once an arc occurson spacecraft
insulator surface, the arc induces a signiŽ cant current on the other
parts of the insulator surface.
The purpose of this paper is to study via laboratory experiments
how the arc current grows by receiving charge from the nearby in-
sulators.In the accompanyingpaper,9 we carry out a basic study on
the growing mechanism via a laboratory experiment and computer
simulation. Once an arc occursat one point on a spacecraftsurface,
eithersolararrayorbodysurface,thearc spotcollectsa largeamount
of positive charge in a short time, driving the spacecraft potential
up. At the same time, electrons are ejected from the arc spot. The
insulating surface becomes positive comparableto the array output
voltage at maximum, exposing its positive charge on the surface to
the electrons. Then, the surface charge is rapidly neutralized by the
electrons, which results in a large negative current  owing through
the spacecraftbody circuit. This phenomenonis the capacitivecou-
pling of the arc plasma with the spacecraftcircuit. The current path
isschematicallyshownin Fig. 2as paths1and2. Path1 is the current
path between the arc point and the insulatorsurfaceon the body, and
the path2 isthe currentpathbetweenthe arcpointandthe solararray
coverglass. The solid lines in Fig. 2 indicate the currents inside the
spacecraft circuit, and the broken lines indicate the current through
Fig. 3 Modelofcurrent pathformationbetween arcpointandunarced
insulator surface.

394
CHO ET AL.
Therefore,acrossthe insulator,thereis a potentialdrop nearlyequal
to the bias voltage,which is of the order of magnitudeof 100 V. The
jump of the electrode potential, which is several microseconds or
less, also causes the jump of insulator surface potential from the
near zero value to a positive value of the order of magnitude of
100 V. Then a positive sheath is formed near the insulator surface
where the electric potential is positive. There is an electric Ž eld
between the array and the insulator, and if the conductivity is high
enough, the current path is formed.
In this study, we carry out new experiments and computer simu-
lations to further validate the model of current path formation. The
experimental setup is basically similar to the previous work,11 but
this time we have installed more test samples inside the chamber so
that we can study how far the arc plasma can in uence the remotely
located insulator.Finding the safety distanceof arc plasmas provide
very important data to the design of high-voltagespace systems. In
theaccompanyingpaper,9 to investigatehowthesurroundingplasma
responds to the sudden change of the insulator surface potential
caused by arcing, we give an artiŽ cial jump of the insulator sur-
face potential and carry out a controlled experiment. We also carry
out computer simulation to visualize the phenomena occurring in
the experiment. In the second section of this paper, we describe the
laboratory experiment, which simulates solar arrays and remotely
located capacitanceunder the LEO space plasma conditions.In the
thirdsection,we showtheexperimentalresults.Finally,we conclude
the paper with suggestionsfor future experimentalwork.
if we look down from the top. The arrays and capacitances were
mounted on acrylic poles at the same height. One array was located
at the chamberaxis, which we call array1. Anotherarraywas placed
70 cm left of array 1, which we call array 2. Capacitance 2 was
placed40 cm right of array 1. Capacitance1 was placed 30 cm right
of capacitance2. Arrays 1 and 2 and capacitances1 and 2 face the
plasma source. Capacitance3 was placed at the axis of the chamber
facing array 1, 270 cm from array 1. To specify the probe position,
we employed a coordinate system whose origin was at the surface
of array 1, as shown in Fig. 4. The axial directiontoward the plasma
z
source is the axis. The horizontal direction toward capacitance 1
y
x
is the axis. In the presentpaper, the axis is the verticaldirection,
x
positive in the upward direction.
The capacitances simulated coverglass of remotely located cells
from the arced interconnector.To increase the capacitanceper unit
(
)
area, thin polyimideŽ lm 7.5 ¹m thick, dielectricconstant3.5 was
(
)
»
used instead of real coverglass 100 ¹m, typically . The Ž lm was
placed on top of a stainless-steelelectrode, which was completely
insulated by the Ž lm and acrylic holder. Capacitances 1 and 2 had
21 nF, and capacitance3 had 5.2 nF because its radius was one-half
of capacitances 1 and 2. Each solar array consisted of six Si cells
connectedin series with a separationdistanceof 0.8 mm. The sizeof
®
£
each cell was 2 4 cm. The cells were glued on Kapton Ž lm, and
the Ž lm was Ž xed on an aluminum frame. The wires of the arrays
were shortened and connectedto a dc power supply to be biased to
¡
up to 800 V.
One array, either array 1 or array 2 and either capacitance1 or 2
or 3 were biasedat the same time by a dc powersupply.The distance
between the array and the capacitancefrom 40 to 270 cm was varied
by changingthe combinationof bias. The current from each biased
sample was measured by a current probe. The arrows at the current
probe in Fig. 4 indicate the deŽ nition of the directionof the positive
II. Experiment
The vacuum chamber we used in the experimentswas 4.0 m long
with 2.5-m diam and was located at the Institute of Space and As-
tronautical Sciences. The axial length is 5.5 m if we include the
semispherical noses at the both ends. The chamber had a diffusive
£
current. The array potentialwas monitored by a 100 high-voltage
argon plasma source located at one end, which generated plasma
11
10 m¡3 and ·
2:4 eV. The chamber
probe. The voltage probe, the current probes, and the high-voltage
probe were connectedto a four-channeldigital oscilloscope,whose
output was transferred to a lap-top computer. The oscilloscopewas
n »
£
T D
conditions of
pressure was 2
5
£
e
e
¡4
»
3
10 torr while the plasma source was oper-
ating. To make the analysis easier, the Earth’s magnetic Ž eld was
¡
triggered when the array potential rose beyond 420 V.
canceledby externalcoilsmost of the time. Unless stated otherwise,
The chamber is equipped with a planar langmuir probe of 6 mm
diam that can be moved in the three axes. The probe is grounded
through a 100-kÄ resistance. The current  owing into the probe is
measured by reading the voltage across the 100-kÄ resistance by a
voltageprobe.The resistanceof 100kÄ was chosenso thatthesmall
amount of current, such as an ion saturation current in the density
measurement, can be measured as the voltage across the resistance.
At the same time, however, the cable connecting the probe has a
capacitance of 600 pF to the ground, and currents with timescales
B
the magnetic Ž eld inside the chamber was kept less than 0.09 G.
We placedtwo small solararraysandthreecapacitancesinsidethe
chamber,asshownin Fig. 4. The coordinateofeachbiasedsampleis
givenin centimeters.The cylindricalchamberis placedhorizontally.
Figure 4 shows the setup inside the chamber from the viewpoint as
£
D
faster than 100 kÄ 600 pF 60 ¹s cannot be measured directly.
For such a fast varying current, only the time-integrated value can
be measured as the voltage across the 600-pF capacitance that is
parallel to the 100-kÄ resistance. The probe signal was also very
(
)
susceptibleto external high-frequencynoise 2 MHz or higher . To
minimize the effect of noise, we carried out the Fourier transform
of all of the waveforms, including the currents and the voltage, and
Ž ltered out frequencies higher than 2 MHz. This Ž ltering limited
the time resolution of the experiment to 0.5 ¹s.
The current owing to the probe consistsof two parts. One part is
the conduction current that is carried by charged particles entering
the probe surface. The other part is displacementcurrent that is due
to the time variation of the surface charge induced on the probe
surface as the electric Ž eld on the surface varies in time. Because
(
)
the probe is Ž xed to the zero potential chamber ground , there is
always a potential differencefrom the surroundingplasma, and the
probe is surrounded by a sheath. The probe surface and the sheath
boundary forms a capacitance, and any change on the potential or
the sheath thicknesscauses the change on the surface charge on the
probe surface. When the potential near the probe changes rapidly,
the probe surfacechargealso changes,and the displacementcurrent
 ows between the probe and the ground. The displacement current
basically correspondsto the time derivativeof the electric potential
near the probe.
In this paper, we mainly show the probe signal varying with
the timescale of the orders of magnitude of 1 ¹s. Therefore, the
probe signal should be regarded as the time-integrated value of the
Fig. 4 Setup of arc experiment.

CHO ET AL.
395
probe current. Because the probe signal is the time integration of
the probe current, the signal directly correspondsto the plasma po-
tential around the probe. Then the probe acts as an ac potential
(
)
probe it cannot measure the dc potential with the bandwidth be-
tween 16.6 kHz and 2 MHz. The positive probe signal means that
the plasma potentialaround the probe is positive with respect to the
probe. In that case, the probe signal mostly consists of the time-
integrateddisplacementcurrent because the conductioncurrent due
to ions is smaller in a short timescale of the order of magnitude of
1 ¹s. Therefore, the positive probe signal should be interpreted as
an indicatorof the plasma potential,thoughthe probe signalvoltage
is not equal to the plasma potential. For the negative probe signal,
the conduction current is as equally important as the displacement
current. The interest is mostly in the positive probe signal.
The plasma uniformity inside the chamber was measured by the
moving probe that measured the ion saturation current. When the
temperature was assumed uniform, the plasma density decreased
y D
less than 10% between
the chamberwall. In the direction,the plasma densityincreasedby
0 and 70 cm as the position approached
z
z D
a factoroftwobetween
toward the plasma source. During the experiment, the plasma con-
10and180cmasthepositionapproaches
(
)
dition was monitored by a spherical 1-cm diam langmuir probe
x D ¡
y D ¡
z D
that was Ž xed at
In a typical case of the experiment, the bias voltage was set to
30 cm,
80 cm, and
163 cm.
»
a Ž xed value for 15 90 min, and we waited for arcs to occur.
»
Arcs occur with a time interval of typically 10 30 s. This time is
long enough for ions to recharge the insulator surface, producing
the potential difference of several hundred volts. After a sufŽ cient
»
number of arcs, usually 5 20, occur, the moving probe is moved
to a different position. The averages of currents and probe signals
are taken from the data taken in one run.
III. Experimental Result and Discussion
In Figs. 5 and 6 examplesof the measured waveforms are shown.
Fig. 6 Example of typical waveforms of 2-peak-type discharge.
¡
In these examples, array 1 and capacitance2 are biased to 500 V,
from the capacitance,which we call neutralizationcurrent. The po-
tentialjumpsto nearzerovalueas a partofdischargecurrent owing
through 10-kÄ resistance causes the voltage drop across the resis-
tance. Once the potentialjumps to near zero, the dc power supply is
effectively decoupled from the array and the capacitance. Then the
dischargecurrentis supplied by the neutralizationcurrent and other
stray capacitancesin the circuit.
z D
y D
and the probe is located at
10 cm and
0. Once an arc oc-
curs, positivecurrent ows from the array toward the ground,which
we call discharge current. At the same time, negative current  ows
The potentials of the insulator surface Ás is given by
( )
1
D
Ás Áe
C Qs
C
=
s
assuming that the insulator surface potential is uniform. Before an
arc, the insulator surface has almost zero potential to attain the
current zero condition. Then there is a potential difference nearly
V
equal to the bias voltage across the insulator. The charge stored
b
t
by the insulator at a time is given by
Z
t
( )
2
Q
t D C ¡V
s . /
C
I
t
n d
.
/
s
b
0
( )
( )
Q
Substituting Eq. 2 into
of Eq. 1 , we obtain
s
R
t
I
t
n d
0
( )
3
D
Ás Áe
C ¡V C
.
/
b
Cs
astheinsulatorsurfacepotential.Whentheelectrodepotentialjumps
to near zero and while the charge own as the neutralizationcurrent
is still small, the insulator surface potential is approximatedby
( )
4
’ ¡Vb
Ás
Similar analysisof the surfacepotentialis presentedin Ref. 5, where
an ion acoustic wave launched from the insulator surface is also
observed. There is also a report12 of the optical measurement of the
insulator surface potential at the arc onset.
The dischargecurrent waveforms are categorizedinto two types.
I
One is the type where the discharge current has only one peak,
,
p1
Fig. 5 Example of typical waveforms of 1-peak-type discharge.
as shown in Fig. 5, which we call 1-peak type. When 1-peak-type

396
CHO ET AL.
pr1
V
discharge occurs, the probe signal Ž rst shows a negative peak
T
whenthedischargecurrentpeaksattime p1 andlatershowspositive
V
pr3
T
at time pr3. The other waveform is the type where the
peaks
discharge current has two peaks, as shown in Fig. 6, which we call
I
the2-peaktype.TheŽ rstpeak p1 issimilarto thatofthe1-peaktype.
I
»
The second peak
occurs 5 15 ¹s later, and it is usually larger
p2
I
than the Ž rst peak p1. The probe signal for the 2-peak type also
V
shows a negativepeak pr1, at the Ž rst peak of the dischargecurrent,
which is similar to the 1-peak type. When the dischargecurrent has
the secondpeak, however, the probe signal differs, dependingon its
position.
In the present experiment, we observed more than 1000 arcs. To
judgewhethereacharc was1-peakor 2-peak,we createda computer
program.In the program, each dischargecurrentwaveformwas Ž rst
smoothed by taking the time average with the window size of 40
neighboring points to reduce the block noise associated with A/D
conversion at the digital oscilloscope. In the smoothed waveform,
the local peaks of 0.05 A or higher were searched. The criteria,
0.05Aand40 datapoints,were determinedaftertrialanderrorofthe
analysis.If therewas a peakin the smoothedwaveform,the program
went back to the original waveform before smoothing and looked
for the local maximum among the 40 points, then recordedits value
Fig. 8 Example of waveforms of probe signal and discharge current
of 1-peak-type discharge measured at different positions along y axis.
I
and time. The program successfullyidentiŽ ed the Ž rst peak
for
p1
all of the waveforms analyzed. If the arc was 2-peak type, however,
the program, occasionally, about 5% of all of the 2-peak types,
I
gave more than two second peaks
if the waveform had a large
p2
 uctuation that was not removed even after the low-path Ž ltering
with the 2-MHz criterion.In that case, the judgmentwas performed
manually and the waveform was looked at with the naked eye.
It was common to 1-peak and 2-peak types that the probe signal
V
shows a negative peak pr1 at the Ž rst peak of the discharge current
Fig. 9 Probe signal Vpr1 at time of the Ž rst discharge current peak Tp1
measured at different positions along y axis.
I
p1. Figure 7 shows the correlation between the two peaks. In this
case, the probe position is at
capacitance1 are biased to 500
biasvoltagesare combinedin Fig. 7. The probenegativepeak pr1 is
y D
z D
0 and
80 cm, and array 2 and
¡
» ¡
800 V. The data at different
V
I
proportionalto the discharge current peak p1. Also, the data for 1-
peaklie on thesamelineasthedatafor2-peak.As farastheŽ rstpeak
I
V
pr1 andtheaccompanyingprobesignal pr1 areconcerned,we seeno
signiŽ cant differencebetween 1-peak- and 2-peak-typedischarges.
Once an arc occurs,the electronsare producedas the result of local-
ized gas discharge near the interconnectorsof the array.13 The ions
produced by the discharge are collected by the array and measured
as the discharge current. The electrons produced by the discharge
either neutralize the charges in the coverglass or are ejected into
I
space. By the time the discharge current has the Ž rst peak p1, a
positive charge of more than 10¡7 C is already  own to the ground.
This means that the same amount of negative charge is ejected to
the chamber. If all of the charges stay in the chamber at the same
time, the number of electrons, approximately1012, is about 10% of
all of the electronsinside the chamber before the arc. This increase
of the negative charge signiŽ cantly lowers the chamber plasma po-
tential. The plasma around the array becomes electron rich, and
the space potential becomes negative due to the space charge of
electrons. While the electrons travel toward the chamber wall or
the capacitance surface to close the current circuit, the probe col-
Fig. 10 Probe peak signal Vpr3 for 1-peak-type discharge and probe
signal Vpr2 for 2-peak-type discharge measured at different positions
along y axis.
lects the electron conduction current giving the observed negative
V
pr1
peak
.
Figure 8 shows examples of how the probe signal of 1-peak-type
y
discharge changes depending on the probe position. The second
row of Fig. 8 shows the discharge current waveforms measured at
y D ¡
the same time. In this case, array 2 located at
70 cm and
y D
¡
capacitance 2 located at
40 cm are biased to 500 V, and the
z D
probe axial positionis
10 cm. The dischargecurrenthas the Ž rst
t D T
’
V
pr1
t D T
peak at
(
(
1 ¹s. The negative probe signal
at
p1
p1
)
see Fig. 5 for deŽ nition is observed even in front of the insulator
)
y D
40 cm , though it becomes smaller.
V
y
Figure 9 shows the probe signal pr1 observedat different posi-
(
)
y D ¡
tions, which are the resultsof the cases where array 2
70 cm
(
)
y D
¡
and capacitance 2
40 cm were biased to 500 V. Figure 10
for 2-peak-type discharge. In Fig. 9 the probe
V
pr3
V
pr2
shows
and
z
z D
axial position is
10 cm. To increase the number of sampling
data, we combined data of the 1-peak- and 2-peak-typedischarges,
and eachpointis an averageof at leasteight measurements.Figure 9
V
pr1
shows that the probe signal
is more negative as the probe po-
sition approaches the array. In Ref. 11, it was shown that the probe
( )
V z
signal pr1 becomesmore negativeas the axialdistance direction
Fig. 7 Correlation between the Ž rst peak of discharge current Ipr1 and
probe signal Vpr1
.

CHO ET AL.
397
Fig. 12 Probe signal Vpr2 at the second peak of 2-peak-type discharge
measured at different positions along y axis.
Fig. 11 Example of waveforms of probe signal and discharge current
of 2-peak-type discharge measured at different positions along y axis.
–
from the array surface decreases. The results shown in Figs. 7 9
indicate that the electron-richnegative sheath is formed around the
array within 2 or 3 ¹s from the arc onset regardlessof the arc being
1-peak or 2-peak.
(
)
Vpr3
In Fig. 10, the positive peak value
see Fig. 5 is shown
y
for different positions, which are the result of the cases where
(
)
(
)
y D ¡
y D
array 2
¡
70 cm and capacitance2
z D
40 cm were biasedto
10 cm. In Fig. 10, we also
z
500 V. The probe axial positionis
V
V
,
pr3
plot the probe signal pr2, which will be explained later. For
Fig. 13 Percentage of 2-peak-type discharge for different distances
between array and capacitance.
each point is an average of at least Ž ve measurements. Even if the
current path is not formed between the array and the capacitance,
the insulator surface potential jumps to a highly positive value as
the electrodepotentialjumps. For the case shown in Fig. 10, array 2
and capacitance2 were biased. At the arc onset the potentialof the
(
)
y D
from the insulator
40 cm exceeds the in uence of the negative
y D
(
)
sheathfrom the array
toarray1 orbeyond.
becausethe in uenceof the negativesheathbecomesgreater.On the
0 cm or not. As the probeisplacedcloser
0/, theprobesignalbecomesmorenegative
¡
y ·
array and the capacitance electrode jumped from 500 V to near
0 V within 3 ¹s. The potential jump produces the positive sheath
y D
(
)
y D
around the insulator surface. In Fig. 10, the position of
40 cm
otherhand,astheprobeisplacedclosertocapacitance2
40cm ,
corresponds to the position in front of capacitance 2. Therefore,
the positive sheath is centered at the insulator surface. The array
y D ¡
the probe signal becomes more positive because the in uence of
the positive sheath becomes greater. Because the probe signal is
an indicator of the plasma potential, Fig. 12 indicates that there is
a gradient in the plasma potential, that is, an electric Ž eld, from
capacitance 2 to array 1, as shown in Fig. 3. If the conductivity of
the plasma between the two surfaces is high enough, the current
path is formed along the electric Ž eld, and the positive charge on
the insulator surface is carried to the array as the arc current. The
2-peak-type discharge is the result of the current path formation,
RC
position is
70 cm.
Figure 11 shows how the probe signal of 2-peak-type discharge
y
changes depending on position. The second row shows the dis-
charge current waveforms measured at the same time. In this
y D
case, the array 1 located at
0 and the capacitance 2 located
y D
¡
at
40 cm are biased to 600 V, and the probe axial posi-
z D
tion is
t D Tp1
10 cm. The discharge current has the Ž rst peak at
0:5 ¹s and has the second peak at
’
t D T ’
7 ¹s for all
y D
which is a
dischargebetween the insulatorand the array, where
p2
(
)
C
R
three examples. When the probe is in front of the array
0 , the
(
see Fig. 5 for notation is negative,
p2
isthe capacitanceof insulatorand istheresistanceof the current
path. At the second peak, we have a current of approximately 1 A
)
V
pr2
t D T
probe signal
at
(
)
see Fig. 6 for an example . When it is assumed that the potential
T
indicating that the in uence of negative sheath formed at
still
p1
T
remains even at time p2. When the probe approachescapacitance2
difference of the order of magnitude of 100 V exists between the
insulator surface and the arc spot, the resistance of the current path
is inferred to be of the order of magnitude of 100 Ä.
(
)
y D
V
40 cm , however, the probe signal pr2 becomes more positive
suggesting the presence of ion-rich positive sheath. Once an arc
occurs, the potential of the solar array increases rapidly from the
negative bias voltage to near zero, as shown in Fig. 6. The potential
ofthe capacitancesurfacealsoincreasessuddenlyfromthenearzero
We now go backto Fig. 10, which showsthe comparisonbetween
V
V
the signal
2-peak-type discharge. For the case of 1-peak-type discharge, the
Vpr2
for 1-peak-type discharge and the signal
for
pr3
pr2
jV j
jV j
V
probe signal
pr3
value to a positive value comparable to
, where
is the bias
increases near the capacitance, similar to
.
b
b
V
V
voltage,as the potentialof underlyingelectrodeincreases.Although
the electrode and insulator surface potential increases within 3 or
4 ¹s from the arc onset, there is a time lag for the positive sheath
to develop. It takes only 2 or 3 ¹s from the arc onset for the nega-
tive sheath to appear in front of the insulator surface. This time is
even faster than the complete jump of the electrode potential. The
t ’ T
However,
The two signals,
sheath near the capacitance surface. For the case of 2-peak-type
does not decrease nor become negative as
does.
pr3
pr2
V
pr3
V
and pr2, indicates the presence of positive
V
discharge,the value of pr2 near the capacitanceis larger,by a factor
V
of 4 or 5, than the value of pr3. This suggeststhat, when the 2-peak-
typedischargeoccurs, the scale of positivesheathis larger.Also, for
V
positive sheath appears at
p2, which is nearly 10 ¹s from the
the case of 1-peak-typedischarge,the signal pr3 does not decrease
arc onset. The mechanism of the positive sheath formation is fur-
nor become negative near the array, indicatingthe disappearanceof
negative sheath near the array. Whether an arc leads to the current
path formation and 2-peak-type discharge or not depends on many
factors. In the following discussion, we Ž rst focus on the distance
between the two points because knowing how far the effect of an
arc extendsis an importanttechnologicalissue. Later we discusson
the dependenceon the bias voltage.
ther discussedin the experimentand the computersimulation in the
companion paper Ref. 9 .
Figure 12 shows the dependence of
(
)
V
pr2
on the probe position
for the combinationof array 1 and capacitance2 at the bias voltage
y D y D
¡
of 600 V. In Fig. 12,
0 correspondsto array 1 and
40 cm
z
correspondsto capacitance2. The probe positionis 10 cm for allof
V
the data points. Whether the probe signal pr2 is positiveor negative
Figure 13 shows the percentageof 2-peak-typedischargefor dif-
ferentcombinationsof biasedarray and capacitance.The horizontal
t D T
at
depends on whether the in uence of the positive sheath
p2

398
CHO ET AL.
to bias the samples, which stores 1:4 10¡6 C before an arc for the
£
¡
bias voltage of 600 V. That the charge for the 1-peak-type dis-
charge does not depend on the distance between the array and the
capacitance suggests that the charge from the stray capacitance is
probably supplied as the arc current for the 1-peak-type discharge.
For the case of 2-peak-typedischarge, the charge stored on the in-
sulator surface is also added to the arc current.
Even if the 2-peak-type discharge occurs, its scale decreases as
the distance between the arced point and the unarced insulator sur-
face becomeslonger. Even if the currentpath is formed between the
arced point and the insulatorsurface and the charge on the insulator
is supplied as the arc current, not all of the charge on the insulator
is released in one arc. Capacitances 1 and 2 have 21 nF,¡a5nd the
Fig. 14 Time delay between the Ž rst and second peaks in 2-peak-type
discharge, Tp2 ¡ Tp1, for different distances between array and capaci-
tance.
£
charge on the insulator surface before arc onset is 1:3 10 C for
¡
the bias voltageof 600 V. Because capacitance3 has 5.2 nF, it has
¡6
£
3:1 10 C. To take into account that the capacitancefor the case
of 270 cm is one-quarter of the other four cases, we multiply the
difference from the value of 1-peak-type discharge 0:8 10
¡7
(
)
£
C
by four and plot the corrected value for the 270-cm case by a tri-
angle in Fig. 15. Then the data points of 2-peak-type discharge
follow a single line. The least-square Ž t to the Ž ve data points
gives
¡6
¡9d
6:96 10
( )
5
Qd
D
£
¡
£
2:60 10
when the line is extended farther, how far the effect of one arc
reaches on the charge of unarced insulator can be estimated. From
( )
Eq. 5 , the distance of 374 cm is obtained as the point where the
line crosseszero, and that is the longestdistance one arc can extend
¡
its effect. The similar Ž t to the four data points for 500 V bias
gives the distance of 297 cm as the safety distance.
(
)
We biasedarray 2 and capacitance1 separationdistance140cm
Fig. 15 Charge  own in one arc, Qd, for different distances d between
array and capacitance.
¡
¡
at voltages between 500 and 800 V. The intentionwas to see the
effectofbiasvoltageonthe probabilityofoccurrenceof2-peak-type
discharge. At the beginning it was thought that the higher the bias
voltage was, the more 2-peak-type discharges would be observed.
However, as shown in Fig. 16, we saw no conclusiveevidenceof the
dependenceof the probabilityon the biasvoltage.The percentageof
2-peak-typedischargeincreasedas the bias voltage was raised from
axis of Fig. 13 is given by the distance between the array and ca-
¡
¡
pacitance. The bias voltages are 600 and 500 V, and each data
point is calculated from at least 32 arcs. Generally, the probability
of occurrence of 2-peak type is less at longer distances because it
is more likely that the negative sheath in front of the array and the
positivesheathin frontof the capacitancefail to meet as the distance
becomes longer.
¡
¡
500 to 600 V. This tendency was observed for all of the other
separation distances, as shown in Fig. 13. However, from 600 to
800 V, the percentage did not change, showing little dependence
on the bias voltage. The data of 500 and 600 V in Fig. 16 were
¡
¡
T
p2 ¡ Tp1
Figure14 showsthe time delaybetween the two peaks,
,
¡
¡
for differentcombinationsof biasedarrayand capacitance.The hor-
izontal axis of Fig. 14 is given by the distance between the array
calculated from 163 and 246 arcs, respectively.On the other hand,
¡
¡
the data of 700 and 800 V were calculatedfrom 58 and 55 arcs,
respectively. To investigate whether or not the bias voltage affects
the occurrence of 2-peak-type discharge, we probably need more
experimental data.
¡
¡
and capacitance.The bias voltages are 600 and 500 V, and each
data point is calculated from at least nine arcs. The error bars in
Fig. 14 indicate the standard deviation. Generally, the time delay
is longer when the distance between the array and capacitance is
longerbecausethe longerthe distance is, the longertime it takes for
the positive and negative sheath to expand to form the current path.
Althoughwe sawnoclearevidenceofthedependenceofthe prob-
ability of occurrence on the bias voltage, the time delay,
had clear dependence on the bias voltage. Figure 17 plots the time
delay between the two peaks,
The bias combination is the same as Fig. 16, that is, array 2 and
capacitance1. In Fig. 17, each data point representsan average cal-
culatedfromatleast21 measurementsof thecurrentwaveforms,and
the error bars indicate the standard deviation. Note that the higher
T
p2 ¡ Tp1
,
¡
The exceptionis the case of 270 cm at 600 V, which corresponds
T
p2 ¡ T
p1, against the bias voltages.
to the bias combination of array 1 and capacitance 3. Array 1 and
capacitance 3 are facing each other along the chamber axis. Then
the sheaths expanding in the direction perpendicularto the surface,
z
thatis, directiondeŽ ned in Fig. 4, meet each other. The other cases
shown in Fig. 14 correspondto the cases when either capacitance1
or 2 is biased with one of the solar arrays, where the sheaths meet
by expanding in the direction parallel to the surface. The speed of
expandingsheath is probably different in the two directions,which
will be shown in Ref. 9 via computer simulation.
Q
Figure 15 shows the charge  own in one arc, d , for different
combinations of biased array and capacitance. The horizontal axis
of Fig. 15 is given by the distance between the array and capaci-
¡
¡
tance. The bias voltages are 600 and 500 V, and each data point
is calculated from at least nine arcs. The error bars in Fig. 15 indi-
catethe standarddeviation.When the 2-peak-typedischargeoccurs,
the charge  own in one arc is much larger than the 1-peak-typedis-
charge. Therefore, the 2-peak-type discharge has more signiŽcant
impact on the spacecraft system. For the case of 1-peak-type dis-
¡7
»
£
charge, the charge is almost the same at 3:5 4:0 10 C for all
¡
of the Ž ve cases of 600-V bias. The experimental circuit has a
stray capacitance of 2.4 nF, mainly due to the coaxial cables used
Fig. 16 Percentage of 2-peak-typedischargefor different biasvoltages
applied to array 2 and capacitance 1.

CHO ET AL.
399
released depends on how far the arc point and the insulatorare sep-
arated. By extrapolatingthe experimentalresult, the safety distance
»
¡
» ¡
of 3 4 m has been derivedfor biasvoltagesof 500
600V. Of
course, this number dependson the experimentalcondition,such as
plasma density,neutraldensity,and others,and must be veriŽ ed in a
futureexperimentwitha largerscale,possiblyina spaceexperiment.
Acknowledgments
The authors thank K. Aihara of the Institute of Space and Astro-
nautical Sciences and T. Matsumoto and K. Shiraishi, students of
Kyushu Institute of Technology,for their help with the experiment.
References
1Sasaki, S., Naruo, Y., and Nagatomo, M., “Engineering Research for
Fig. 17 Time delay between the Ž rst and second peaks in 2-peak-type
discharge, Tp2 ¡ Tp1, for different bias voltages applied to array 2 and
capacitance 1.
SolarPowerSatelliteSPS2000,”ProceedingsSPS’97Conference, Montreal,
–
PQ, Canada, 1997, pp. 73 77.
2Hastings, D. E., Cho, M., and Kuninaka, H., “The Arcing Rate for a
High Voltage Solar Array: Theory, Experiment, and Predictions,”Journal of
–
Spacecraft and Rockets, Vol. 29, No. 4, 1992, pp. 538 554.
the bias voltage is, the shorter time it takes for the positiveand neg-
ative sheath to expand to form the current path. Similar data from
the previous experiment are also shown in Ref. 11.
3Katz, I., Davis, V. A., and Snyder, D. B., “Mechanism for Space-
craft Charging Initiated Destruction of Solar Arrays in GEO,” AIAA Paper
98-1002, Jan. 1998.
4Ferguson, D. C., Snyder, D. B., Vayner, B. V., and Galofaro, J. T., “Array
Arcing in Orbit: from LEO to GEO,” AIAA Paper 99-0218, Jan. 1999.
5Cho, M., Miyata, N., Hikita, M., and Sasaki, S., “Discharge over Space-
craft Insulator Surface in Low Earth Orbit Plasma Environment,” IEEE
Transactions on Dielectrics and Electrical Insulation, Vol. 6, No. 4, 1999,
IV. Conclusions
As the size of spacecraftsystems and the demand for more power
increases, arcing on solar array will become an obstacle against
the development of megawatt-class space platforms that generate
and deliver power at voltages of 400 V or higher. How an arc on
solar array occurs has been well studied in previous work, and it
has been known that the Ž eld intensiŽcation near the triple junction
due to charging of insulator by plasma triggers the arc. To develop
mitigation strategies against arcing on high-voltagespace systems,
both preventing arc onset and preventing its growth have practical
importance. Laboratory experimentshave been carried out to study
how an arc on a solar array in the LEO plasma environment grows
by taking charge from a remotely located insulatorsurface.
Once an arc occurs on a solar array, electrons are ejected from
the arc spot. Initially the charge is suppliedby the stray capacitance
of the external circuit, such as the cable. Electrons diffuse away
rapidly,and on their way, they producea negativesheatharound the
arc point due to their negative space charge. At the same time, the
jump in the circuit potential makes the insulator surface potential
jump fromnearzerotoa highlypositivevalue.As thepositivesheath
reachesthe arc point, there is an electric Ž eld between the array and
the insulator. If the conductivity of the plasma along the Ž eld line
is high enough, a current path is formed between the two points
and carriesthe insulatorchargeand suppliesthe arc current,feeding
energy to the arc plasma.
–
pp. 501 506.
6Vayner, B. V., Doreswamy, C. V., Ferguson, D. C., Galofaro, J. T., and
Snyder, D. B., “Arcing on Aluminum Anodized Plates Immersed in Low-
Density Plasma,” Journal of Spacecraft and Rockets, Vol. 35, No. 6, 1998,
–
pp. 805 811.
7Carruth, M. R., Jr., Vaughn, J. A., Bechtel, R. T., and Gray, P. A., “Exper-
imental Studies on Spacecraft Arcing,” Journal of Spacecraft and Rockets,
–
Vol. 30, No. 3, 1993, pp. 323 327.
8Vaughn,J. A., Carruth, M.R.,Jr., Katz, I.,Mandell,M. J., andJongeward,
G. A., “Electrical Breakdown Currents on Large Spacecraft in Low Earth
–
Orbit,” Journal of Spacecraft and Rockets, Vol. 31, No. 1, 1994, pp. 54 59.
9Cho, M., Ramasamy, R., Hikita, M., Tanaka, K., and Sasaki, K., “Plasma
Responseto Jump of InsulatorSurface Potentialin IonosphericPlasma Envi-
–
ronment,” Journal of Spacecraft and Rockets, Vol. 39, No. 3, 2002, pp. 400
408.
10Carruth, M. R., Jr., Schneider, T., McCollum, M., Ferguson,D., Katz, I.,
Mickaterian, R., Alred, J., and Pankop, C., “ISS and Spaced Environment
Interactions WithoutOperating Plasma Contactor,”AIAA Paper 2001-0401,
Jan. 2001.
11Cho, M., Hikita, M., Tanaka, K., and Sasaki, S., “Transient Phenomena
Induced by Discharge on Solar Array in Low Earth Orbit Plasma Environ-
(
)
ment,” IEEE Transactions on Plasma Science to be published .
12Cho, M., Miyata, N., and Hikita, M., “Effects of Arcing on Insulator
Surface Charging Condition in Plasma Environment,” AIAA Paper 2000-
0872, Jan. 2000.
The experimenthas shown that not all of the arc leads to the cur-
rent path formation. If not, an arc is a just small pulse of current,
that is, 1-peak-type discharge, whose scale is determined only by
the stray capacitanceconnected to the circuit. If the current path is
formed, that is, 2-peak-type discharge, it has a possibility of lead-
ing to a large-scale arc, even short circuiting of positive and nega-
tive ends of the solar array. The experimental results indicate that
whether the current path is formed and how much the charge is
13Cho, M., and Hastings, D. E., “Computer Particle Simulation of High
Voltage Solar Array Arcing Onset,” Journal of Spacecraft and Rockets,
–
Vol. 30, No. 2, 1992, pp. 189 201.
A. C. Tribble
Associate Editor