688
J
OURNALOF
P
ROPULSIONAND
P
OWER
Vol. 1 4, N o. 5, September
–
O ctober 1 998
Spacecra ft Electric Propulsion— An Overview
M. Martinez-Sanchez*
Massachusetts Institute of Tec hnology, Cambridge, Massachusetts 02139
and
J. E. Pollard†
The Aerospace Corporation, El Segundo, California 90245
A short review of the status of electr ic propulsion
(
E P
)
i s presented to serve as an introduction to the
more specialized technical papers also appearing in this Special Issue
(
J ournal of Propulsion and Power,
Vol. 14, No. 5, Sept.
–
Oct. 1998
)
. The principles of oper ation and the severa l types of thrusters that are
ei ther operational or in advanced development are discussed rst, follo wed by so me consi derations on
the necessary power sources. A few prototypical missions are then described to highlight the operational
p eculiarities o f EP, including spacecraft i nteractions. We conclude with a historical summary of the
ac cumulated ight experience using this technology.
I . I ntroduction
T
HIS paper is intended to serv e as a ge neral overview of
t he technology of electric propulsion
(
E P
)
and its appli-
cations, and to lead the interested reader to the more speci c
t echnical p apers on the topic that are a lso included in this
S pecial Issue. It is hoped that this se ries of papers will be of
u se not only to the propulsion specialist, but also to spacecraft
(
S /C
)
d esigner s seeking to familiar ize themselves with a tech-
n ology that is now seeing rapid introduction.
E P is by no means a new concept, having rst been tested
i n ight in the 1960s. However, its introduction as a p ractical
alternative to chemical thrusters for S/C propulsion has been
slow in developing, owing to a combination o f insuf cient
o nboard elecrical power on most S/C, and a reluctance by
many mission planners to abandon tried and true solutions. The
p otential performance advantage of primary EP for space mis-
sions with large
D
V requirements was recognized from the
b eginning, and m uch of the early research and development
work addr essed this type of mission. Yet, it has been the grad-
u al application of the simpler forms of EP to secondary pro-
p ulsion tasks that has led to its acceptance, with the long-
envisioned d eep-space applications only now beginning to
materialize.
We rst review the existing and emerging typ es of EP de-
v ices and their power sources, with some comments about their
relative position in the mission spectrum. The m issions them-
selves are examined to highlight the differences in planning
t hat EP introduces. This is ampli ed in a bri ef review of the
n ew S/C integration issues brought about by these thrusters. A
summary is included of the ight experience accumulated up
t o the present time.
I I. EP Systems
A n EP system is a set of components arranged so as to
eventually convert electrical power from the S/C power s ystem
i nto the kinetic energy of a propellant jet.
1
F igure 1 shows in
schematic form the principal elements of an EP system and its
i nterfaces with other S/C systems. Typically, the power system
supplies regulated dc bus power to a power processor unit
(
P PU
)
, as well as to other auxiliary elements, such as v alves,
Received March 3 , 1998; revision received June 15, 1998; accepted
for publication June 24, 1998. Copyright Q 1998 by the American
Institute of Aeronautics and Astronautics, Inc. All rights reserved.
*Professor, Department of Aeronautics and Astronautics.
†Senior Scientist, Technology Operations.
h eaters, etc . The PPU processes this raw power into the spe-
ci c form required by the thrusters and is usually one of the
most complex and chall enging EP c omponents, as will be seen
i n subsequent sections. A regulated, pressure-fed fuel system
i s shown for illustration, although simple blowdown supplies
can sometimes be used. No detail is shown of the plumbing,
which often includes series and parallel valves, pyrote chnically
o pening or closing valves, etc. The ows to be handled are
u sually very small, but occur for very prolonged periods of
t ime
(
m onths
)
, which presents sp ecial challenges for the design
o f precise ow controllers and leak-free valving. Commands
t o the various power switch es, valves, etc., are supplied by the
S /C computer, which also receives and processes a variety o f
status signals from sen sors
(
on ly a pair of pressure signals are
i llustrated
)
.
T he hear t of the system is, of course, the thruster itself, and
t his paper will concentrate on thrusters. I t must be understood,
h owever, that a large proportion of the propulsion e ngineer’s
effort must be devoted to the balance of the EP system, which
i n the end is also usually heavier, bulkier, and more expensive
t han the thruster
(
s
)
. Fortunately, aside from the PPU peculi-
arities, the rest of the syste m is not drastically different from
more familiar cold-gas or monopropellant systems, and indeed,
E P has bene ted in its gradual introductio n from this existing
experience base.
I II. Electric Thrusters
T he common feature of all EP schemes is the addition of
energy to the working uid from some electrical source. This
h as been accomplished, however, i n a lar ge variety of physi-
cally different devices. Operation can be steady or pulsed; gas
acceleration can be t hermal, electrostatic, electromagnetic, or
mixed; the prop ellant can be a noble gas, a chemical mono-
p ropellant, or even a solid. Of the many combinations tested
o ver the years, a reduced, but still large number have reached
maturity, or ar e approaching it. These are listed in Table 1 ,
where a few of their principal attributes are also given. Sim-
p li ed schematics of these thrusters are shown in Fig. 2 to
i llustrate their operating principles. The numerical values listed
i n Table 1 are not to be taken as recommended design values,
b ut only as indicative of a range; for more detailed informa-
t ion, the reader is referred to the quoted literature.
2
–
20
A. R esistojets
A s indicated in Fig. 2a, resistojets operate by passing the
g aseous propellant around an electrical heater
(
which coul d be
t he inside of tubes heated radiatively from the outside
)
, then
MARTINEZ-SANCHEZANDPOLLARD
689
F ig. 1 Schematic of a typical EP syste m.
using a conventional nozzle to generate thrust. The heating
r educes the gas ow rate from a given ups tream pressure
t hrough a given nozzle area, thus leading to the familiar in-
c rease in speci c imp ulse as . Nearly any gas could beT
Ï
used
(
as long as it is compatible with the high -temper ature
heater
)
, and this may be dictated b y considerations such as
w aste disposal on manned S/C. The most successful applica-
t ion has been based on the superheating of catalytically de-
c omposed hydrazine, whic h has the advantage of common ality
w ith familiar fuel systems used in hydrazine monopropellant
a pplications. The hea ters can operate over the wide pressure
r ange encountered with blowdown systems, and their input
vo ltage is low eno ugh to requ ire no special power condition-
i ng, except for current surge protection. An exception to this
w ould be a S/C power system that would allow more than
a bout 20% voltage variations, i n which case a dedicated reg-
ul ator would become necessary. Operation can continue in a
no nsuperheated mode in case of heater failure. The plume is
not i onized and poses no unusual S/C interaction problems.
B ecause the molecular mass of the gas
(
N
2
/H
2
/NH
3
)
i s rel-
a tively high, an d because the heating wall is limited by ma-
t erials
(
W
–
Re or something simi lar
)
t o about 2000 K, t he spe-
c i c impulse
(
I
sp
)
achieved is only mod est, of the order of
30 0
–
3 10 s. This is 40% better t han that without superheating,
a nd the improvement comes at a very small cost in complexity,
i f power is available. A favorable situation
(
f or this and also
f or other EP techniques
)
o ccurs in geostat ionary communica-
t ions satellites, in which excess power is indeed available most
of the time
(
S ec. V
)
, and this prompted the early commercial
i ntroduction of hydrazine resis tojets
(
s tarting with Intelsat V,
19 80
)
for the north
–
south stationkeeping
(
NSSK
)
f unction. A
m ore recent application is for orbit insert ion, control, and de-
or bit of the Iridium low Earth orbit
(
L EO
)
constellation. One
of the few technical problems posed by these thrust ers is the
t endency of the hydrazine to produce no nvolatile deposits at
t he hot inlet to the catalytic chamber; thi s is common to all
hy drazine thrusters, but the problem is made more critical by
t he reduction in ow rate because of the h igher I
sp
. Solutions
have included t he use of ultrapure hydrazi ne and thermal
s hunts to reduce heat ux at the critical points.
A mmonia resistojets have also been used for higher speci c
i mpulse
(
l ighter gas
)
, at some cost in complexity.
A s shown in Table 2, resistojets have own on Intelsat V,
S atcom 1-R, GOMS, Meteo r 3-1, Gstar-3, and Iridium S/C, in
a ddition to some older satellites and test ights.
B . Arcjets
A rcjets are, like resistojets , electrothermal devices, but the
w all temp eratur e limitation of the resistojet is overcome here
by depositing power internally, in t he form of an electric arc,
t ypically between a concentric upstream rod cathode and a
do wnstream anode that also serves as the supersonic nozzle
(
Fi g. 2b
)
. The ow structure at the throat is extremely non-
u niform, with th e arc co re at temperatures of 10,000
–
20 ,000
K, and the buffer layer near the wall at no more than 2000 K.
Because of this there is prac tically no ow through the arc
core, which can be thought of as an effective uid plug; this
reduces the ow, without reducing the pressure integral, and
l eads to the h igh speci c impulse. On the other hand, some
i ntrinsic loss mechani sms are now int roduced.
1
)
C ompared with a uniformly heated stream, the nonun i-
formity reduces by itself the propulsive ef ciency because, as
i n any thermal propuls ion device, maximu m thrust for a given
p ower and ow is obtained when the heat is added uniformly
across the ow.
2
)
T he power invested in ionizing the arc gas is mostly lost
b ecause of the small recombination time available
(
i n addition,
i n molecula r gases, there is a substantial dissociation loss as
well
)
.
3
)
T here are near-el ectrode voltage drops, which mainly
constitute a local heat loss to the electrodes.
A s with resistojets, the choice of gas is often dictated by
considerations unrelated to the thruster itself. Once again, the
rst practical implementation has bee n with hydrazine, as the
n ext logical evolutionary ste p f rom monopropellant systems.
However, arcjets do pose some restrictions
(
r elated to arc sta-
b ility
)
on the range of stagnation pressures used, which tends
t o limit blowdown systems to those in which the arcjet portion
o f the t otal fuel is relatively small [as in the Lock heed Martin
Telstar-4, where most of the hydrazin e is used in a bipropellant
chemical thruster for geos ynchronous Earth orbit
(
GEO
)
or bit
i nsertion, the balance going to a 1.8-kW arcjet for NSSK].
5
Hydrazine arcjets achieve I
sp
’
5 00
–
6 00 s, a signi cant im-
p rovement from resistojets; on the other hand, the ef ciency
i s, for the reasons stated, no higher than 35
–
40 %, which may
o r may not be critical, depending on the relative importan ce
o f fuel vs power savings. Further performance advances wil l
b e paced b y materials and by thermal design, the goal being
t o allow the arc to ll most of the throat, without damaging
i ts wall.
O ther working gases of potential interest for arcjets are hy-
d rogen and ammonia. Hydrogen has a clear advantage in per-
formance
(
I
sp
c an easily exceed 1000 s
)
b ut suffers from the
l ow storage density and the cryogenic nature of the fuel; it
might become practical for missions with continuous thrusting,
where the tank could be cooled by the evaporatio n of the feed.
Ammonia yields sp eci c impulses I
sp
’
80 0
–
9 00 s with a
l iquid fuel , but the propellant system is relatively complex.
T he PPU for an arcjet is signi cantly more complex than
t hat for a resisto jet. The discharge voltage i s higher than most
b us voltages, e.g., 80
–
1 20 V, requiring at a minimum dc
–
dc
conversion. In addition, the negative-impedance characteristic
o f an arc must be handled without the interposition of a dis-
sipative ballast, and special transient modes must be provided
for star tup. As a consequence the PPU can be several times
h eavier than the thruster itself
(
Table 1
)
. The plume is still
relatively benign, with divergence angles similar to those of
conventional thrusters, and n o more than a few percent ioni-
zation.
A rcjets occupy an intermediate place in the I
sp
s cale, and
will remain a viable option for missions where there is some
p reminum on short-burn dur ation
(
l ower speci c impulse im-
p lies higher thrus t for a given power
)
, or on thrust over only
a portio n of the orbit
(
S ec. V
)
. They will also continue to
compete with plasma thrusters in geostationary applications,
where they present fewer S/C integr ation problems
(
S ec. VI
)
.
The Telstar 4 and GE-1 satellite series have feat ured arcjets
(
Table 2
)
.
C. H all Thrusters
T he generic con guration of a Hall thruster is shown in Fig.
2 c. Gas
(
usually xenon
)
i s injected through the anode into an
annular space and is ionized by counter owing electrons,
which are part of the current injected through an external hol-
l ow ca thode
(
t he rest neutralizes the ion beam
)
. The ions ac-
690
MARTINEZ-SANCHEZANDPOLLARD
Table 1 Typical electric thruster features
Thruster
type
Res istojet
(
N
2
H
4
)
Res istojet
(
NH
3
)
N
2
H
4
arc H
2
arc NH
3
arc Xe Hall Xe ion
(
Pu lsed
)
Te on PPT
Cs
FEEP
MPD
(
applied
eld
)
MPD
(
self- eld
)
Po wer range, W 500
–
1500 500 300
–
2000 5
–
100 K 500
–
30 K 300
–
6000 200
–
4000 1
–
200 10
2
5
–
1 1
–
100 K 200
–
4000 K
I
sp
, s
(
typical
)
300 350 500
–
600 1000 500
–
800 1600 2800 1000 6000 2000
–
5000 2000
–
5000
h
(
typical
)
80% 80% 35% 40% 27
–
36% 50% 65% 7% 80% 50% 30%
Plume divergence <20 deg —— <20 deg <20 deg <20 deg 30
–
40 deg <20 deg —— —— —— ——
Peak voltage 28 28 100 200 —— 300 900 1000
–
2000 6000 200 100
Thruster mass, kg /kW 1
–
2 —— 0.7 0.5 0.7 2
–
3 3
–
6 120 —— —— ——
PPU mass, kg/kW 1 —— 2.5 2.5 2
–
3 6
–
10 6
–
10 110
(
in-
cludes ca-
pacitor
)
—— —— ——
Miscellaneous
a
(
per
thruster
)
1 —— 5 5 —— 10 10 Small —— —— ——
Tankage fraction 0.05 —— 0.05 0.15 —— 0.12 0.12 —— —— —— ——
Propellant management Standard
blowdown
Regulated Regula ted or
low -range
blowdown
Regulated or
low-range
blowdown
—— Regulated Reg ulated —— Cs boiler —— ——
Propellant stora ge Stable liquid Liquid Stable liquid Cryogenic H
2
Stable liq-
uid
Su percritical Xe
(
r > 1.5 g/
cm
3
)
Supercritical Xe
(
r > 1.5 g /
cm
3
)
Solid Solid/liquid NH
3
, H
2
, A NH
3
, H
2
, A
Lifetime, h 500 —— >1000 —— 1500 >7000 10,000 >10
7
pulses
4000 N-s
—— —— ——
S/C interaction con-
cerns
—— —— Thermal ra-
diati on
Thermal ra-
diation
—— Wide plume,
ion back ow,
torque
Ion back ow,
nonpropellant
ef uent
Plume con-
densation
Cs contami-
nation
—— ——
Status Operational —— Operational Lab Quali ed Operational Operational Operational Development Lab Lab/ develo p-
ment
Typical mission NSSK EWSK
orbit inser-
tion, deor-
bit
Orbit cor-
rections
NSSK Orbit trans-
fer
(
me-
dium DV
)
NSSK, orbit
raising
NSSK orbit
raising
(
me-
dium DV
)
NSSK orbit
transfer
(
large DV
)
Small or bit
corrections
(
precision
)
Small orbit
corrections
(
precision
)
Large DV,
medium
power
Large DV,
high power
References 2, 3 3, 4 5, 6 7 8, 9 10, 11 12
–
14 15, 16 17 18 19
a
Cable, gimbal, thermal control, s tructure, propellant feed
(
power range 0.5
–
2 kW
)
.
MARTINEZ-SANCHEZANDPOLLARD
691
F ig. 2 Operating principles of a
)
re sistojets, b
)
a rcjets, c
)
Hall thruste rs, d
)
i on engines, e
)
pulsed plasma thrusters, f
)
eld-effect
el ectrostatic propulsion thrusters, and g
)
s elf- eld magnetoplasmadynamic thrus ters.
c elerate under the electrostatic eld impressed by th e negative
c athode, and are only weakly de ected by the imposed radial
m agnetic eld. The electrons, however, are strongly magne-
t ized, and are forced to execute an azimuthal drift
(
H all cur-
r ent
)
, whil e slowly diffusi ng axially across the eld toward the
a node
(
f rom where the power supply pumps them to th e cath-
od e
)
. Thus, Hall thrust ers belong, with ion engines, to the class
of electrostatic ion accelerators; on the other hand, the thrust
i s transmitted to the magnetic coils through their interaction
w ith the electron Hall current, justifying the name. The gas
density is low enough to ensure near collisionless ion ow,
a nd so a Hall thruster is many times wider than an arcjet of
s imilar power
(
b ut still more compact than ion engines
)
.
T he plume is h ighly
(
s ometimes fully
)
i onized, but there are
no dissociation losses, and the ionization loss is somewhat di-
l uted by the higher operating voltage
(
about 300 V for I
sp
>
15 00 s
)
. There is
;
5% I
sp
l oss as a result of the gas ow in
t he hollow cathode; otherwise the el ectrode losses are small.
T hese thrusters achieve ef ciencies in the 45
–
55% range, in-
c reasing slowly with voltage
(
s peci c impulse
)
. Because most
of the ow leaves as ions, the current is controlled by t he ow
r ate; owing to the very low ows involved
(
a few mg/s
)
, this
ow control is a nontrivial task and is accomplished either by
t hermally varying the gas viscosity in the feed capillar ies, or,
m ore recently, by precision elect romechanical valves. For a
gi ven ow, thrust, speci c impulse, and power c an be con-
t rolled through the voltage. Some additional control can be
exerted through the magnetic eld, although for simplici ty, the
magnetic coils tend to be placed in series with the main dis-
charge. The PPU is here even more complex and heavy than
for arcjets
(
s ee Tabl e 1
)
, as plasma uctuations have to be
accommodated, a nd the magnet current and ow controls co-
o rdinated as well.
I n addition to the ceramic-lined stationary plasma thruster
(
SP T
)
ty pe, a competing Hall thruster design that offers about
t wice the thrust density
(
b ut very similar performance
)
i s the
so-called ‘‘ thruster with an anode layer’’
(
TAL
)
, in which the
walls are metallic, and the channel is sh orter and narrower.
F light development and quali cation of TAL thrusters is still
u nder way.
B ecause of their relatively h igh ef ciency at moderately
h igh sp eci c impulses, Hall thrusters are found to be optimal
for many applicati ons and, having a substantial orbital pedi-
g ree in Russia
(
s ee Sec. VII
)
, are now available for many
We stern missions. These include NSSK applications for sev-
eral satellite series, as well as plans for deployment and orbit
control for some l ow Earth constellations. In a series of qual-
i cation tests to Western st andards, the particular SPT type
h as demonstrated 7000 h of operational life
(
l imited by erosion
o f the con ning ceramic walls
)
, which is suf cient for most
o f these missions. TAL thrusters have not undergone this kind
o f life testing, but the fact that the denser part of the plasma
692
MARTINEZ-SANCHEZANDPOLLARD
Table 2 Operational ights of electric thrusters
First launch
Vehicle type /
number on-orbit
Thruster type /
number per vehicle
Kilowatts per
thruster
Thruster
function Sponsor/builder
Attitude control:
1964 Zond-2 /1 Te on-pulsed plasma/6 <0.10 Sun pointing Russia
1987 Kosmos /2 Xenon plasma, SPT-70/6 0.75 Attitude control, orbit ad-
justment
Russia/Fake l
Geosynchronou s stationkeeping:
1968 LES-6/1 Te on-pulsed plasma/4 <0.01 EWSK USAF/FHC 1 MIT
1980 INTELSAT-V/13 Hydrazine resistojet/4 0.35 NSSK Ford /TRW
1982 Kosmos, Luch /13 Xenon plasma, SPT-70/4 0.75 EWSK, reposition NPO PM /Fakel
1983 Satcom-1R, etc./32 Hydrazine resistojet/2,4 0.60 NSSK RCA /RRC
1993 Telstar-4, etc./12 Hydrazine arcjet/4 1.8 NSSK LM/OAC
1994 Gals, Express /5
a
Xe plasma, SPT-100/8 1.35 NS and EWSK NPO PM /Fakel
1994 ETS-6, COMETS Xenon ion, IES 12 cm /4 0.73 NSSK NASDA/MELCO
1994 GOMS
(
Electro
)
/1 Ammonia resistojet/16 0.45 EWSK, attitude control Russia/NIIEM
1996 GE-1, etc./9
a
Hydrazine arcjet/4 2.0 NSSK LM/OAC
1997 PAS-5, Galaxy 8-i/2 Xenon ion, XIPS-13/4 0.44 NSSK, attitude control Hughes
1998
b
Galaxy-10/1 Xenon ion, XIPS-25/4 4.5 NSSK, attitude control Hughes
2000
b
ARTEMIS/1 Xe ion, UK-10, RIT-10/4 0.60 NSSK ESA/RAE 1 MBB
2000
b
Stentor/1 Xe plasma, PPS-1350/4 1.5 NSSK CNES/SEP 1 Fakel
2000
b
DRTS/2 Hydrazine arcjet/4 1.8 NSSK MELCO /NASDA
Other orbit adjustments:
1965 Ve la/ 2 Nitrogen resistojet/1 0.09 Phase adjustment USAF /TRW
1965 U.S. Navy satellite/5 Ammonia resistojet/2? 0.03 Orbit adjustment U.S. Navy/GE
1967 Advanced Vela/6 Nitrogen resistojet/2 0.03 Phase and spin adjustment USAF /TRW
1971 U.S. Navy satellite/4 Ammonia resistojet/2 0.01 Orbit adjustment U.S. Navy/AVCO
1974 Meteor-18, etc./4 Xenon plasma, SPT-60/2 0.45 Orbit adjustment Russia/Fakel
1976 TIP/2, NOVA /3 Te on-pulsed plasma/2 0.03 Drag compensate d U.S. Navy/RCA 1 APL
1981 Meteor 3-1, etc./10 Ammonia resistojet/4 0.45 Orbit adjustment Russia/NIIEM
1988 Gstar-3/1 Hydrazine resistojet/2, 4 0.60 Orbit transfer RCA /RRC
1997 Iridium/72
a
Hydrazine resistojet/1 0.50 O rbit adjustment Iridium / OAC
1998
b
AMSAT P3D/1 Ammonia arcjet/1 0.70 Orbit adjustment AMSAT /IRS
1998
b
New Millenium, DS-1/1 Xenon ion, 30 cm /l 2.4 Orbit transfer NASA/Hughes
2000
b
MightSat II.1/1 Te on-pulsed plasma/1 0.10 Orbit transfer USAF/Primex
a
Total expected by 1998.
b
Planned launch date.
i n these engines is formed outside the annular channel may
favor longer life. Some of the remaining concerns are as fol-
l ows.
1
)
A wide plume dispersion angle, which forces alignment
at angles of 30
–
4 0 deg to solar arrays or other sensitive areas.
2
)
Under some operating conditions
(
and mostly in SPTs
)
,
d eep current uctuations at a few tens of kHz, togethe r with
a variety of higher frequency
(
b ut less deep
)
pl asma uctua-
t ions; this may pose communications interference problems if
t he plume intersects the antenna pattern.
3
)
A small torque about the axis because of magnetic forces
o n the ions; this amounts t o about 2% of the product of thrust
and diameter, and is easily countered by pair arrangements or
small auxiliary thrusting.
H all thrusters have own
(
Table 2
)
on the Kosmos, Luch,
Gals/Express, Meteor 18, and Meteor 3-1 satellites in addition
t o severa l test S/C.
D. I on Engines
I n gridded electrosta tic ion accelerators
(
F ig. 2d
)
, ions are
p roduced in a separate, magnetically con ned ionization cham-
b er, usually by an auxiliary dc discharge, although one can
also use radio-frequency power [as in the European radio-
frequency for thrusters
(
R ITs
)
] , or tuned electron cyclotron res-
o nance
(
E CR ionizer
)
. One side of this chamber is covered by
a double-grid structure, with spacing o f the order of
–
1 mm,
1
–
2
across which the ion acceleration voltage is applied. Ions
t hat wander into the thin sheath covering the inner
(
s creen
)
g rid fall through and ar e extracted and accelerated, the ion
o ptics geometry being desig ned to avoi d impact on the accel-
erating outer grid. Electrons are absent in the grid gap, whic h
l imits the extracted current below that level for which repul-
sion by ions in that gap would keep new ions from entering
(
s pace charge limit
)
. The electric eld at the inner surfac e o f
t he accel grid is strong, and its pull on th e grid transmits the
t hrust to the structure. Electrons are extracted from the ioni-
zation chamber through an anode
(
s o that at least one dc elec-
t rode is present, even in radio frequency devices
)
, and pumped
b y the power supply to an ext ernal hollow cathode/neutra-
l izer held slightly above the potential of the accel grid. Unlike
Hall thrusters, the magnetic e ld in ion engines plays a sec-
o ndary role, limited to delaying the loss of primary ionizing
electrons.
B ecause of the space charge limitation, ion thrusters are
b ulkier for the same thrust than Hall thrusters, in which elec-
t rons neutralize the plasma everywhere. Because they optimize
at high speci c impulse, they p rovide less thrust per unit
p ower. They also have a more complex set of power supplies
and controls, with PPU masses of the order of 7 kg /kW
(
Table
1
)
. On the other hand, t hey offer substantially better control
o f the plasma l ocation and operating parameters, which trans-
l ates into 1
)
l onger life, with 10,000 h demonstrated; and 2
)
b etter ef ciency, at l east at speci c impulses above
’
2 500 s
(
b ut Hall thrusters may evolve in that direction and compete
)
;
and 3
)
l ess beam divergence, but still requiri ng beam canting
o f the order of 30 deg away from solar arrays. One further
advantage of ion engines is their technologic al maturity; de-
v elopment challenges tend to be limited to life extensions
t hrough better materials, and simp ler and lighter PPUs.
I on engines are the thrusters of choice for deep missions,
r equiring high
D
V and tolerating long thrusting times, such as
i nterplanetary or bit transfers. The high speci c impulse advan-
t age may be partially lost in NSSK applications if low avail-
able power forces nearly continuous thruste r oper ation, includ-
i ng inef cient orbital locations
(
S ec. V
)
, but they are still quite
competitive for this application. Ion engines have own op-
erationally on the Japanese ETS-6 a nd COMETS satellite, plus
PAS-5 and Galaxy 8-I
(
f rom Hughes Space and Communica-
t ions Co.
)
, and additional ights are forthcoming.
E. Pulsed Plasma Thrusters
P ulsed p lasma thrusters
(
P PTs
)
d iffer from all other concepts
d iscussed here in two fundamental ways: they operate in short
