Session : Hall Thrusters

Hall thrusters are electrostatic ion accelerators in which the grid system (which serves in
classical ion engines to anchor the negative charges used to accelerate the ions) is replaced
with a relatively strong magnetic eld perpendicular to the ow. This magnetic eld impedes
the counterow of electrons in the accelerating eld, and, as will be shown, does away with
the space-charge limitation which restricts the ow and thrust of ion engines.
Hall thrusters have had a curious and long history. Discovered in the early 1960s in the
U.S. [1], and possibly independently in the USSR [2], they were later abandoned in the West
when it became apparent that there were strong instabilities which could not be completely
eliminated, and when some additional work in Germany [3] indicated higher eective plasma
collisionality than had been expected. Contributing to the demise was the simultaneous
successful development of gridded ion engines of the Kaufman type, which appeared to satisfy
most high Isp mission requirements then envisioned, and the promise of ecient hydrogen
arcjets for the low and intermediate Isp ranges. It later developed that many important
missions optimize in the Isp range of 1000 2000 sec, not well covered by either of these
types of engines, and this remained an unmet requirement for many years, at least in the
West.
Development continued, however, in the USSR, particularly at the Kurchatov Institute in
Moscow [2], under A.I. Morozovs leadership. Progressively more ecient congurations
evolved there, and it was realized that the instabilities, while present and annoying, did
not materially interfere with performance. The Soviet generic name for this type of engine
was Thruster with a Closed Electron Drift, and two competing implementations have been
developed, known respectively as the Stationary Plasma Thruster (SPT) and the Thruster
with an Anode Layer (TAL). The general principles are the same in both, and will be
discussed below.
By the early 1980s these engines had achieved operational status in the USSR, and have
since own in many (> 50) missions, which, however, were until recently limited to relatively
small total impulses. Starting in 1991, with the complete removal of the earlier communica
tions barrier, development re-started in the West, in the form of collaborative eorts with
Russian teams aimed at improved life, lighter electronics and ight qualication to the gen
erally more stringent Western standards. The strong interest shown by the user community
stems from the ability of Hall thrusters to operate with fairly good eciency ( 50%) in
the hitherto dicult specic impulse range around 1500 sec, and also from their relative
simplicity compared to ion engines.
Many dierent varieties of Hall thrusters exist, created in several laboratories and companies
around the world. Academic eorts have been directed mostly towards the understanding
of ionization, electron trapping and diusion, and loss mechanisms. These eorts include
exploring the low power regime (< 50 W) with aims in miniaturizing plasma propulsion
for applications in small, or power-limited spacecraft. While able to operate, small Hall
thrusters suer from reduced life and eciency, due for the most part to the violent plasma
environment required to maintain a collisionality level similar to larger devices. Recent work
has been directed towards the design and development of high-power Hall thrusters in the
10-50 kW range, motivated in good measure by the steady improvement of solar power
systems.

 r

B
cathode
magnetic coils
v x
E

CL
j Br
Ex
V
CL

propellant
dielectric
anode walls

magnetic
structure

Hall Thruster Physics

A Hall thruster schematic is shown in the gure above. It consist of a coaxial annular cavity
where plasma is created by passing current between the annular anode on the upstream
end of an otherwise dielectric cavity and the externally located cathode. The propellant
enters this plasma cavity via an annular manifold at the anode. A radial magnetic eld is
applied, either by ring-shaped permanent magnets, or through coils and soft iron yokes. The
magnetic eld greatly slows down the axial mean velocity of the electrons, which, due to the
j B
low collisionality prevailing, are forced to execute mostly E j drift around the annulus,
while being radially conned by sheaths on the insulating walls.
The ions meanwhile, are only weakly aected by the magnetic eld, and, if the density is
low enough that collisions are rare, are simply accelerated by the electrostatic eld to an
exit velocity

2e
vi = (1)
mi
where is the potential at the place where the ion is created (with respect to the outside
potential). Because of the quasi-neutrality facilitated by the presence of the electrons, no
space-charge limitation arises in this type of thruster (as opposed to gridded ion thrusters),
and the acceleration distance can be several cm, compared to the typical 0.51 mm gap used
in ion engines. This exibility is one of the main advantages of the Hall thruster, removing
the strong thrust density limitation dictated by the Child-Langmuir law in ion engines.
To the extent that collisions do occur, but, more importantly, because of electron scattering
by a combination of plasma electrostatic uctuations and wall collisions, electrons also travel
axially across the magnetic eld under the inuence of the applied axial electric eld. They
are then collected by the upstream anode, and pumped by the power supply to an external
cathode. The emitted electrons mainly join the accelerated ions to form a neutralized plasma
beam, but, inevitably, a fraction also diverts upstream into the accelerator section. This

 fraction is to be minimized (this is the role of the magnetic eld) because their acceleration
to the anode potential is one of the devices loss mechanisms. On the other hand, not all of
this energy is lost to the anode, because a good part of it is used to produce ionization of
the injected neutral gas.
The name Hall Thruster arises form the mechanism by
which thrust forces are exerted on the solid parts of the
r engine. As indicated, ions are simply accelerated by the
electrostatic eld, but since the ions are in a quasineutral
B
cathode plasma throughout, an equal and opposite electrostatic
force is exerted on the free electrons in that plasma. How
v x ever, in the presence of the radial magnetic eld, these
E electrons are not free to accelerate towards the anode; in
stead, they drift azimuthally (perpendicular to both E j and
CL j at such a velocity as to generate an equal and oppo
B)
Ex site magnetic force on themselves. If we denote by x the
forward axial direction the electrons end up drifting with
a velocity,
CL
j B
E j
jv = 2
(2)
B
dielectric
walls The ions have no such azimuthal drift (or have a very small one, because their gyro (Larmor)
radius is larger than the devices length), and so a net azimuthal current density arises, called
agnetic a Hall current,
ructure
j j
jj = ene E B (3)
B2

Given this current, the magnetic (Lorentz) force density on it is fj = jj B

j , and an equal
and opposite force is exerted by the plasma currents on the magnetic structure. We then
have,

fj = ene E
j (4)

i.e., the same as the forward electrostatic force on the ions, as it should be. The important
point is that the structure is not electrostatically acted on (electric elds and electric pres
sures are too weak here), but magnetically, through the Hall current - hence the name. The
device is a Hall thruster, but an electrostatic accelerator, a duality which has led to some
confusion.

Electron diusion in a magnetic eld

The description of Hall thruster physics cannot be complete without some discussion of the
diusion of electrons towards the anode, forced by its higher electric potential. Eventually, as
is the case in the ionization chamber of ion engines, all electrons, primaries and secondaries
are collected by the anode. The goal is to maximize the lifetime of electrons to increase the
probability of ionizing collisions with neutrals.

Let us begin by analyzing the electron diusion in the absence of magnetic elds. In the
electron momentum balance, main forces are the pressure gradient, the electric eld and
collisional retardation. Neglecting electron inertia and ignoring the electric eld (since linear
acceleration does not aect the conclusion), this reduces to,

Vpe ne me ejve (5)

Where the collision frequency is e = nj ce Qej . We also have pe = ne kTe , and for near-
isothermal conditions Vpe = kTe Vne . Solving for the electron ux,

Vpe kTe
nejve Vne (6)
m e e m e e
This is Ficks law of diusion, with a diusivity,

kTe
De = (7)
me e
It is clear then that collisions impede the diusion of electrons. In this situation, large
collisions frequencies would result in the longest lifetimes before electrons are collected by
the anode.
Assume now that a magnetic eld (perpendicular to Vpe ) is applied. We add the magnetic
force to the momentum equation,

j
kTe Vne ne me ejve enejve B (8)

j and write,
To solve for for the ux, take the cross product with B

j ne me ejve B
kTe Vne B j enejve B
j B
j

Let us use our tensorial notation to nd a vectorial identity for jv B B,

j B
jv B j = ijk jlm vel Bm Bk = (il km im kl )vel Bm Bk = vek Bi Bk vei Bk Bk = vei B 2
i

j . We then have,
since jve B

j ne me ejve B
kTe Vne B j + enejve B 2 (9)

j between Eqs. (8-9) leads to,

Eliminating jve B

kTe ekTe j
Vne + 2 2 Vne B
me e me e
nejve 2 (10)
eB
1+
me e

Recall that this formulation leaves the electric field out. To include it, just replace ne
ene ~
by ne + E. Including this term in Eq. (6) leads to the usual definition of electric
kTe
conductivity. Define the non-dimensional factor,

eB ce
= = (11)
m e e e
where ce is the cyclotron frequency for electrons. Then,

De ne De ne
ne~ve (12)
1 + 2

Of these two terms, the second is perpendicular to both B~ and ne , and is called the pB
~
drift. The main interest is on the first term, which is along ne and E, ~ as in regular
diffusion and conduction (when there is an electric field).
We see that this cross-field diffusion is governed by ne~ve D ne , with,

De
D = (13)
1 + 2
So, a high Hall parameter can greatly reduce diffusion, compared to that in the absence
of a magnetic field. High means both, high magnetic field and/or low collision frequency.
As an example, in an ion engine or Hall thruster, with Te = 4 eV = 46, 400 K the en and ei
cross-sections are roughly,

Qen 1019 m2 and Qei 4 1018 m2

and,
r
8kTe
ce = 1.34 106 m/s
me
For particle densities,

ne 2.8 1017 m3 and nn 7.4 1018 m3

then,

en 9.9 105 s1 and ei 1.5 106 s1 so that e 2.49 106 s1

At a point in the engine where B = 100 gauss = 0.01 Tesla,

ce 1.76 109 s1

which means that the Hall parameter is,

ce
= = 706  1
e

5
Under these conditions, Eq. (13) reduces to,

De kTe e
D 2
= 2
(14)
me ce

This last form shows that collisions favor diusion, in contrast with Eq. (7), with no magnetic
eld, or 1. Eq. (14) also shows that D scales as B 2 :

De me kTe e
D 2
=
e2 B 2
and so, increasing B should provide very strong connement of electrons. With the given
numbers, we nd,

De = 2.83 105 m2 /s andD = 0.57 m2 /s

A diusing substance spreads (in 1-D) roughly as x 2 Dt. So, to spread by 1 cm, electrons
would require a time t 4.5 105 s.
It turns out, however, that electrons can diuse faster than this in most cases. The physical
reasons are apparently related to the equivalent collisionality produced by scattering of the
electrons by small-scale plasma density uctuations which are almost always present. This
is the same situation that has kept tokamaks from delivering fusion power (only in that case
it is the H+ ions that leak through the conning B j eld).
Bohm obtained an empirical expression (with some theoretical guidance) for this so-called
anomalous or Bohm diusion,

kTe
DBohm = (15)
16eB
and experiments in ion engines and Hall thrusters appear to conrm the B 1 dependence,
but also seem to indicate a somewhat smaller diusivity magnitude.
An often used expression is,

kTe
Danomalous = (cB 16 to 100) (16)
cB eB
It is of some interest to see what collision frequency would produce the same diusivity as
these uctuations,

kTe me kTe anomalous ce

= anomalous = (17)
cB eB e2 B2 cB
and so cB can be thought of as an anomalous Hall parameter. For modeling purposes, one
often adds together the regular and anomalous collision frequencies e +anomalous in calculating
diusivity D by Eq. (14).

 Thrust Capability

Unlike the case of a gridded ion engine, there is not a clear-cut upper limit to the obtainable
thrust density in a Hall thruster. In principle, increasing the mass ow rate through a given
engine would increase thrust by the same factor, provided the same degree of ionization can
be maintained. The voltage is presumed to be kept the same, since this controls specic
impulse fairly directly. The ion beam current and the electron current leaking into the
accelerating channel would increase as the mass ow rate, so that the energetic ability of
these electrons to ionize the increased neutral ow would be maintained. Since no space
charge develops these increases in ow, current and power could go on indenitely.
The above reasoning has ignored the various eects of collisions. As the ow rate, and
hence the gas density increases, so does the scattering collision frequency of electrons with
heavy particles, eH = nH ce QeH . When this frequency becomes comparable to the gyro or
cyclotron frequency ce , electrons can no longer maintain pure azimuthal drift, because each
collision will allow the electric eld to nudge them in the direction of the anode by about one
Larmor radius, correspondingly to a velocity similar to the azimuthal drift velocity. Thus,
electrons are now able to cross the magnetic barrier at a velocity,

E eH
vex (18)
B ce
which increases with eH , hence with nH and ultimately with m . The electron density ne in
the acceleration region is governed by the ow of ions from the upstream ionization region,
as shown in the schematic below, and increases generally in the same proportion as m or IB .
This means that the electron leakage current,

E eH
Ileak = ene (19)
B ce
will increase faster than IB , due to the nH factor in eH .

cathode

anode neutral ionization acceleration

IB
region region region

I leak
Ia IB

V Br

As will be discussed later, one of the factors determining the overall eciency of the device
is the anode eciency,
CL
IB
a =
Ia

propellant
and from the schematic, ignoring wall losses, IB = Ia Ileak , so,

Ileak 1
a = 1 =
Ia 1 + Ileak /IB
which shows how the higher electron collisionality would hurt eciency. By the way, Eq.
(19) also shows that, if this were the only problem with higher ow rates, increasing the
magnetic eld strength (if it were feasible) would be a way to counteract it.
Quantitatively, we would like to keep a low leak ratio, say Ileak /IB 0.2. Assuming the
ions constituting IB cross the magnetic barrier at about 1/2 of their eventual speed vi , this
yields,

Ileak vex 2 E eH vi Bce

= = 0.2 eH 0.1 (20)
IB vi /2 vi B ce E

For the SPT-100 thruster, vi = 16, 000 m/s, B 0.02 T (corresponding to an electron
cyclotron frequency ce 3.5 109 rad/s). The axial electric eld E is about 5000 V/m in
the high B region. Eq. (20) then requires, eH 2.2 107 s1 . At Te 10 eV, we have
ce 2.1 106 m/s, and the scattering cross-section for en collisions is about 3 1019 m2 .
We then obtain a limiting neutral density in this region,

eH
nH = 3.5 1019 m3
ce QeH
Measured neutral densities in this type of thruster are still signicantly lower than this value.
Operation of Hall thrusters in such high-ow regime has recently been considered as an
alternative to MPD propulsion when high power is available (upwards of 50 kW). The physics
of this regime could be, at least in principle, explored at ow rates several times above
the nominal for a given thruster type. Interestingly, no systematic study of this operation
mode has been carried out, in part because very high powers have been associated with the
development of nuclear reactors for space applications, as always a contentious issue. The
limited availability of facilities capable to handle high powers has also played a role in the
lack of data in this regime.

References
[1] Seikel, G., and E. Reshotko. Hall Current Ion Accelerator. Bull, Am. Phys. Soc. Series
11, V. 7, No. 7 (1962): 19. Also Salz. F., R. Meyerand, and E. Larry. Ion Acceleration
in Gyro Dominated Neutral Plasma. Experim. Bull, Am. Phys. Soc.Sec. 11, Vol. 7, No.
7 (1962): 441.

[2] Yushmanov, Ye. Ye. Radial Potential Distribution in the Cylindrical Trap with the
Magnetron Method of Injecting the Ions. Phys. of Plasma on Controlled Thermonuclear
Fusion. Vol. 4. AS of the USSR, 1958.

[3] Krulle, and Zeyfang. Preliminary Conclusions of Applied Field Electromagnetic

Thruster Research at DFVLR. AIAA Paper 75-417. 11th AIAA Electric Propulsion Con
ference, New Orleans, 1975.

[4] Bishaev, A. M., V. M. Gavryushin, A. I. Bugrova, V. Kim, and V. K. Kharchevnikov.

The Experimental Investigation of Physical Processes and Characteristics of Stationary
Plasma Thrusters with Closed Drift of Electrons. RGC-EP 92-08. Also in Bishaev, A. M.,
and V. Kim. Local Plasma Properties in a Hall Current Accelerator with an Extended
Acceleration Zone. Sov. Phys. Tech. Phys. 23, no. 9 (September 1978): pp. 1055-1057.

MIT OpenCourseWare
http://ocw.mit.edu

16.522 Space Propulsion

Spring 2015

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.