You are on page 1of 22

Magnetoplasmadynamic Thrusters

Mariano Andrenucci
Department of Aerospace Engineering, University of Pisa, Pisa, Italy

1 Introduction
2 The Nature of the Lorentz Force
3 The Ideal Self-Field MPD Thruster
4 Real Self-Field MPD thrusters
5 The Onset Riddle
6 Applied-Field MPD Thrusters
7 Lithium Propellant MPD Thrusters
8 Survey of Major R&D Efforts
9 Future Prospects
References

1
2
5
7
10
13
14
15
18
20

1 INTRODUCTION
The essence of what was later to become known as the magnetoplasmadynamic, or MPD, thruster, emerged from the flurry
of research and development activities that characterized the
field of propulsion among others in the feverish, postSecond World War era. Research on arc thrusters came about
almost naturally from work on conventional rockets, as an
alternative way to heat the propellant, as opposed to the use
of the heat released by chemical reactions. Heating the working gas by means of an electric arc offered the additional
bonus of making it possible to adjust the power input independently of the mass flow rate. Extensive research activities
were started in many public and private laboratories, which
brought, in a relatively short time, to the experimentation of
a wide variety of configurations and operating regimes.
Encyclopedia of Aerospace Engineering.
Edited by Richard Blockley and Wei Shyy
c 2010 John Wiley & Sons, Ltd. ISBN: 978-0-470-68665-2


It was just in the midst of one such arcjet-related activity that the evidence of an acceleration mode differing from
the expected conventional gasdynamic mechanism was gathered, quite serendipitously, by Adriano Ducati at the Giannini
Scientific Corporation of Santa Ana, California (Ducati,
Giannini and Muehlberger, 1965). In the words of one of
its major discoverers (Jahn, 1968), in an empirical series
of experiments with a conventional short arcjet device it
was found that by drastically reducing the propellant gas
flow . . . the exhaust velocity of the hydrogen flow could be
increased to values of the order of 100 000 m s1 , and the
overall efficiency reached 50%. The ensuing supposition
was that the high current densities in the arc were generating
self-magnetic fields within the chamber sufficiently intense
to produce substantial electromagnetic acceleration of the
flow.
The device experimentally demonstrated by Ducati (Figure 1) was the MPD arc thruster with a self-induced magnetic
field. This discovery led to a burgeoning of activity in plasma
thruster research. The new acceleration mode was referred
to with a variety of names such as the high-impulse arc,
thermoionic accelerator, magnetic annular arc, and Hall arc
accelerator, and it took some time for the term Magnetoplasmadynamic to become accepted as the standard name for this
new class of device.
This is how MPD thruster work began in the USA. Activities along similar lines were sprouting up in the meantime in
the former Soviet Union, and based on what would become
known only decades later, the dimension of these efforts
soon exceeded the levels reached in the USA and, later on,
in Germany and other western countries. The main lines of
the subsequent evolution of the MPD concept and the main
results achieved are shortly reviewed later. First we shall
focus on the concept itself and its physical bases, which

2 Alternative Propulsion

Anode
Cold propellant
inlet
Cathode

1958

1959

Cold
propellant
inlet

1960

Cathode

1961
(a)

(b)

Plasma
Uniform exhaust
stream

Anode
Very high Isp core

Low Isp envelope

1963

(c)

Figure 1. (a) The elimination of the supersonic nozzle has been our first effort. This was a difficult idea to accept at that time;
however, nozzles are gradually disappearing as one can observe in a comparison of contemporary geometries used in the adoption
c
of this principle; (b) uniformity of thermo-ionic vs conventional arc-jet. Adopted from Ducati, Muehlberger and Giannini (1964) 
AIAA.

were for some time considered rather elusive. The discoverers themselves noticed (Ducati, Muehlberger and Giannini,
1964): Many questions still remain unanswered. One can
call the thruster thermo-ionic, electro-thermal, J-cross-B,
Hall-Current, or cyclotron resonance, or any other descriptive
name, but still no one can explain completely its mechanism.
It is to the clarification of this mechanism that the next section
is dedicated.

2 THE NATURE OF THE LORENTZ


FORCE
Under the physical conditions typical of high-power arc
devices we can assume the working fluid to be in the state
called plasma. The most important implication of this is for
such a fluid to behave as an electrically conductive medium
that remains quasi-neutral at all scales comparable with the
size of the device or experiment of interest. This can be
stated in terms of number densities of the component charges,
electrons and ions, as:
|ne ni |  ne ni = n

(1)

The consequences of this assumption, as well as a number


of other features that are usually associated with the term
plasma, are extensively covered in many excellent textbooks
(Chen, 2006; Bittencourt, 1986; Lieberman and Lichtenberg,
1994; Spitzer, 1964; Mitchner and Kruger, 1992) and will not
be dealt with here. MPD thrusters, as well as other types of
electric thrusters such as Hall-effect thrusters, fit in a category
that can be designated as plasma thrusters. This definition
entails the idea that apart from local effects such as the

sheaths positive and negative particles never get separated


throughout all phases of the acceleration process (differently
from what happens in gridded ion thrusters).
To analyze the nature of the MPD acceleration process,
we shall start by describing the dynamical equilibrium at
any point of the flowfield produced in a generic thruster
microscopically. As is largely known, the analysis of the
motion of an ensemble of particles is the realm of kinetic
theory. The behavior of an ensemble of particles can be thoroughly described by the kinetic equation known as Boltzmann
equation. But as we are interested in the global, collective
behavior of the various components of the working medium,
a description in terms of average behavior of particles of any
species is normally sufficient. This is usually done by taking
the first three velocity moments of the Boltzmann equation,
thus obtaining the mass, momentum, and energy conservation
equations for each species.
As we deal with a general problem of thrust generation,
for the purpose of the present discussion we shall focus on
the momentum equation for each of the species constituting
the working medium. We shall limit our attention to a simple
case that will permit us to reach some general conclusions
without unnecessary complications.
Let us hence adopt the following main simplifications (other assumptions should become obvious from the
context):

r we shall assume the working medium to be composed of two species only: electrons and singly-charged
ions;
r as mentioned, we shall assume the fluid to remain quasineutral at all times: ne ni = n;

Magnetoplasmadynamic Thrusters 3

r we shall neglect viscous effects which are usually very


small in typical situations of our interest;

r we shall neglect the momentum storing capability of the


electron fluid due to the smallness of the electron mass as
compared to that of the ions.
Under the above assumptions the momentum conservation
equations for the ionic and the electronic components at any
point of the thruster channel can be simply stated as
mi n

dui
= n e (E + ui B) pi + Pi e
dt

0 = n e (E + ue B) pe + Pe i

(2)
(3)

where mi is the ion mass, n the common number density


of electrons and ions, ui and ue the ion and electron fluid
velocities in the laboratory frame, E and B the local electric
and magnetic induction field vectors, pi and pe the ion and
electron pressures, and Pie and Pei the momentum gain of the
ion fluid caused by collisions with electrons and vice-versa.
The term on the left in the ion equation describes the time
change in momentum of the ion fluid in a frame moving with
the fluid. It represents the convective derivative

involving neutrals, in that the interaction takes place at a


distance, by means of the Coulomb electric field forces surrounding all the nearby charged particles (glancing collisions
or Coulomb collisions). It takes a large number of such glancing collisions combining casually to produce effects similar
to those induced by a head-on collision. This process can be
described in terms of random walk, so as to define an equivalent collision cross section, a collision frequency, a mean free
path, and so on, allowing collisions between charged particles
to be described in analogy with ordinary strong collisions.
To understand the nature of the Lorentz force it is not
necessary to enter into the details of the collision terms. It is
sufficient to recognize that under the two-fluid idealization
assumed here it is simply
P i e = P e i

(5)

so that we can cancel the collision terms between equations


(2) and (3), to find
mi n

dui
= n e (E + ui B)
dt
pi n e (E + ue B) pe

(6)

which, with the following further definitions


d

=
+ u
dt
t

(4)

which combines the time change in momentum seen by a


static observer plus the change produced as the observer
moves with the fluid into a region of different momentum. The
terms on the right side relate such total momentum change
with the effects of the forces applied. Under the simplifying
assumptions listed above, only the electromagnetic force and
those associated with pressure gradients and collisions are
accounted for.
This is where the Lorentz force comes into play. Named
after the Dutch physicist who discovered it, the Lorentz force
law states that a charge q moving with velocity u in the presence of an electric field E and a magnetic field B will not
only feel a force qE due to the electric field but also a force q
(u B) associated with the magnetic field. Alternately, we
could say that the charge will feel an overall electric field
differing in magnitude and direction with respect to the field
E seen by a static charge by a component u B. This gives
for both electrons and ions the expressions given in equations
(2) and (3).
As for the collision terms, they describe in this case only
collisions between electrons and ions. The characteristics and
effects of collisions between charged particles in a plasma are
very different from the strong, typically inelastic, collisions

p = pe + pi

= mi n

(7)

gives

d ui
= n e (ui ue ) B p = j B p
dt

(8)

where the difference between electron and ion velocities has


been expressed in terms of current density
j = n e (ui ue )

(9)

Thus, in equation (8) everything finally reduces to the familiar


Lorentz force term j B (apart from the pressure gradient
contribution).
The situation can be illustrated as shown in Figure 2. Leaving aside the effects of pressure gradients, the primary cause
of acceleration is the electric field. Electrons are accelerated
by the field but transfer all of the momentum acquired to ions
through collisions. Ions, in turn, are also accelerated by the
electric field and the ensuing momentum increase combines
with that received from the electrons. Also evident is the fact
that the increase in momentum felt by the electrons can be
subdivided in a part that would be felt if the electrons where
moving at the same velocity of the ions and a second part due

4 Alternative Propulsion

Figure 2. The Lorentz force.

to their differential velocity with respect to ions. The former


part is equal and opposite to the momentum increase imparted
by the electric field on ions so that when this is transmitted by
the electrons to ions through collisions the two terms cancel
each other. The only effect left is therefore the effect of the
electric field on the electrons due to the velocity difference
of electrons with respect to ions externally seen as current
and transferred to the ions themselves (i.e., to the fluid)
through collisions. By choosing to represent the electric field
in a frame moving with the ions, E , we are dispensed from
referring to any specific ion velocity, thus making the picture
more general.
To obtain further insight into the character of the acceleration process we need to be more specific about the form
of the collisional term. Assuming the electron-ion collision
process to correspond to an equivalent collision frequency
i e , in the two-fluid idealization assumed here we can write
P i e = P e i = i e me n (ue ui ) =

ne
j

(10)

where we have introduced the conductivity


=

n e2
i e m e

(11)

Making use of equation (10), (2) and (3) can be restated as


mi n

d ui
ne
= n e (E + ui B) pi
j
dt

(12)

0 = n e (E + ue B) pe +

ne
j

(13)

The latter can also be written as:





1
j = E + ue B +
pe
ne


1
1
pe
jB
= E + ui B +
ne
ne

(14)

which can be recognized as the generalized Ohms law


describing the relationship between fields and current in the
plasma. Solving the above equation for the electric field E
we obtain
E = ui B +

1
1
j
jB
pe +
ne
ne

(15)

where we can recognize, from right to left, the Ohmic component (last), the field-equivalent of the pressure gradient, the
field associated with the electron relative motion (current) in
the presence of the magnetic field (Hall term) and the field
associated with the magnetic force exerted on the ions. This is
the so-called self-consistent electric field expressing an equilibrium that must exist at any point of the channel between
the local values of the fields and the other physical quantities.
To see how effectively the momentum exchange between
electron and ions can result in increasing the flow directed
kinetic energy let us derive the dot product of the momentum
equations for the two species, equations (8) and (9), with ui

Magnetoplasmadynamic Thrusters 5
and ue respectively:
ui

we can write

dui
ne
= pi ui + n e Eui
jui
dt

0 = pe ue n e Eue +

(16)

ne
jue










 ne

u2i
jui 

= pi ui + n e Eui 
2



(18)











ne
j 2 

0 = pe ue n e Eue + jui 














(19)

where the dashed boxes now highlight the collisional terms


describing the frictional power exchange between electrons
and ions associated with the collisional friction force density.
Not surprisingly, the rate at which directed energy is
acquired by the electrons due to collisions with the ions is
simply minus the rate at which energy is acquired by the ions
due to collisions with the electrons. But the electron energy
change includes another term, j 2 /, which represents the
conversion of the ordered motion of the electrons, relative to
the ions, into random motion (i.e., heat) via collisions with
the ions. Note that this term is positive definite, indicating
that the randomization of the electron ordered motion gives
rise to irreversible heat generation. This is the term usually
called ohmic or Joule heating term.
Adding up equations (18) and (19) and remembering equation (9) the collisional terms cancel out, and we are left with
d
dt

u2i
2

= pi ui pe ue + Ej

j2

(20)

If we want to make the role of the Lorentz force in equation


(20) more explicit, we can go back to equation (15), and
scalarly multiply with j, thus obtaining
Ej = (ui B) j

1
j2
pe j +
ne

(21)

Considering that it is
(ui B) j = (j B) ui

(23)

so that equation (20) can be finally put in the form









d
dt

1
j2
= (j B) ui
pe j

en
= (j B) ui pe (ui ue )

(17)

being the work of the magnetic force on the moving particles


of course equal to zero. The last term in equation (17) can
now be decomposed by use of equation (9). With obvious
further passages we obtain


Ej

(22)

d
dt

u2
2

= pu + (j B) u

(24)

where u ui is the mass-averaged plasma velocity. Equation


(24) could also be obtained directly from equation (8) by
scalar multiplication with u.
The above analysis shows that the acceleration mechanism
based on the Lorentz force is inherently dissipative in that it
is based on a collisional momentum transfer between electrons and ions that inherently entails frictional dissipation. In
this regard MPD thrusters are necessarily less efficient than
thrusters in which the acceleration of the ions is obtained
from electrostatic forces, and hence conservatively (apart
from other real-life loss mechanisms). The above analysis,
as noted before, described the equilibrium at a generic point
of the acceleration channel of a generic thruster. To correlate
this with the behavior of the thruster as a macroscopic device
implies integrating fluid equations under appropriate boundary conditions expressing the operating conditions applied to
the thruster. Although this could only be made on the basis
of a detailed description of any specific device, some important scaling laws can be obtained that express quite general
behavioral trends.

3 THE IDEAL SELF-FIELD MPD


THRUSTER
In its basic form, the MPD thruster consists of two metal
electrodes separated by an insulator: a central rod-shaped
cathode, and a cylindrical anode that surrounds the cathode
(Figure 3). A high-current electric arc is driven between the
anode and cathode so as to ionize a propellant gas to create
plasma. A magnetic field is generated by the electric current returning to the power supply through the cathode. This
self-induced magnetic field interacts with the electric current
flowing from the anode to the cathode (through the plasma)
to produce the electromagnetic Lorentz force that pushes the
plasma out of the engine, creating thrust. MPD thrusters are
usually classified either in the self-field variety, which is fully
based on the pure self-field mechanism said above, or in
the generally lower-power applied-field version, where

6 Alternative Propulsion
with the inclusion of a corrective term as follows
T =

0 J 2
ln
4

re
+A
ri

(26)

For instance, in the case of a conical cathode tip involving


a combination of radial and axial current attachment (Figure
4b and c) one would find A = 3/4.
In more realistic configurations the relationship between
thrust and current squared would depend on the details of
electrode geometry and current attachment; but the electromagnetic component of the thrust would still follow a law of
the type
Figure 3. Schematic of MPD thruster.

T = b J2

(27)

an external coil is used to provide additional magnetic field


to help stabilize and accelerate the plasma discharge. For the
moment we shall concentrate on the basic, selffield version
of the concept.
The basic analysis of MPD thruster operation is usually
prompted by simple one-dimensional idealizations (Figure
4a). For a coaxial channel of external radius re and internal
radius ri , integration of the distributed Lorentz body-force
over the discharge volume leads to the following expression
for the thrust (Maecker, 1955):

with b representing a factor of a mainly geometrical character. Values of b for typical geometries are about (23)
107 N/A2 .
Based on the above analysis, in an ideal device the thrust
would appear to depend on the discharge current only, regardless of the propellant mass flow rate m.
The effective exhaust
velocity ve would therefore scale with the inverse of m

1  2
0 re 2
LJ =
ln J
2
4 ri

where we have introduced the characteristic parameter k =


J 2 /m,
which is reminiscent of the electrical power deposited
in the channel per unit propellant mass flow-rate. As we shall
see later, the importance of this parameter in characterizing
an MPD device cannot be overemphasized. Equation (28)
shows that, apart from the b factor, the J 2 /m
ratio is equivalent to the effective exhaust velocity; that is, to the specific

T =

(25)

where L is the channel inductance per unit length and J is


the thruster current. In a more complex channel geometry and
taking into account finite cathode length and pressure effects
on the cathode tip, the above expression can be generalized

ve =

T
J2
=b
= bk
m

(28)

Figure 4. Idealized MPD channel models: (a) uniform radial current; (b) radial current into conical cathode; (c) uniform axial current.
c McGraw Hill.
Modified from Jahn (1968) 

Magnetoplasmadynamic Thrusters 7
impulse. Attempts to obtain higher Isp are therefore equivalent to trying to operate the thruster at larger ratios J 2 /m.
As
will be discussed later, beyond a certain limit this turns out
to be prohibitively difficult.
Based on equation (28), the ideal kinetic power associated
with the thrust can be written as
PT =

1
b2 k 2
m
u2e =
J
2
2

(29)

so that we can define a dynamic impedance associated with


the useful power spent in accelerating the fluid as
ZT =

b2 k
2

(30)

Finally, we can express the overall input power as the sum


of the useful power associated with the thrust plus losses
Pi = PT + PL

(31)

The power associated with losses can also be related to an


equivalent impedance
ZL =

PL
J2

(32)

We can therefore write a general expression for the thrust


efficiency as follows
PT
ZT
=
=
T =
PT + P L
ZT + Z L

b2 k
2
b2 k
2

+ ZL

1
(33)
L
1 + 2Z
b2 k

An ideal MPD thruster with thrust scaling quadratically


with the current would therefore obey the following laws of
dependence of power and voltage with the current
b2 4
J
2m
T

(34)

b2 3
Pi
=
J
J
2m
T

(35)

Pi =

V =

In conclusion, the behavioral trends of an ideal MPD


thruster could be summarized as
T J2

V J3

P J4

(36)

4 REAL SELF-FIELD MPD THRUSTERS


Information on how real thrusters behave is obtained through
experimental activities. Self-Field MPD thrusters are naturally relegated to high power operation, as the self-induced
magnetic field is relatively week unless very high currents
of O (10 kA) are applied. Unfortunately, steady-state testing
at the MW level is difficult, and the most experimental data
collected over decades in various laboratories have been gathered with the thruster working in the Quasi-Steady (QS) mode
(Clark and Jahn, 1970). In this mode, the thruster is operated
for current pulse lengths of O (1 ms), and data so obtained
are expected to be representative of its steady-state performance. Unfortunately, this may appear questionable. From
direct comparison of geometrically identical thrusters operated in continuous mode and QS pulsed mode, Auweter-Kurtz
et al. (1994) have drawn indication that results of QS thrusters
cannot be plainly extrapolated to the steady operation case.
Were this so, most QS results obtained in the past decades
would be irrelevant to characterizing the real behavior of
steady-state high-power thrusters. However, this is the data
available at present and nothing better can be expected until
MW level steady-state testing becomes feasible or practical.
Let us return to the ideal MPD thruster model outlined
in the previous section. Given its high degree of idealization, some discrepancies in the behavior of real thrusters
with respect to the model presented above were of course
to be expected. Several factors conspire to make the real situation different and in particular: the geometrical shape of the
thruster, the pattern of current flow lines and fields, various
subtle aspects of ion and electron dynamics not included in
the simple model, losses taking place at various levels.
Even in a simple coaxial configuration the situation would
depart from the assumed patterns of orthogonal electromagnetic fields, currents and gas flow pictured in Figure 4. Due to
the Hall effect, under typical conditions existing in an MPD
thruster channel and especially in the anode sheath region,
the current tends to flow with a strong axial component (Figure 5). In addition to complicating the current flow pattern,
this also brings about a radial component of the Lorentz force
resulting in a depletion of charge carriers near the anode, with
detrimental effects that we shall discuss later.
Another factor that complicates the picture is the backEMF due to plasma motion through the self-field. This
voltage gradient given by the vector product of the flow velocity and the magnetic field strength u B, tends to discourage
the current from flowing in the intermediate region of the
channel, where both u and B are large (Figure 6). This results
in a current density increase at the two ends of the electrodes
with possible consequences in terms of enhanced erosion, and
which can even entail a full conduction crisis in the event

8 Alternative Propulsion

200

Voltage, V

Mass flow rate, g s1

4 5

6
(III)

100

(II)

(I)
Full ionization

Figure 5. Conceptual illustration of current flow in an MPD thruster


c IEPC.
with Hall effect. Modified from Hoyt (2005) 

0
0

10000

20000

30000

Discharge current, A

the back-EMF becomes comparable to the thruster driving


voltage (we shall return to this later).
Such and other effects concur in making the real situation
different from the idealized one. This implies, in particular, that the thrust formulas presented in the previous section
are inadequate to provide anything better that an order of
magnitude appraisal of the expected thrust level. Various
attempts have been made to work out more complex expressions enabling to improve the thrust prediction capability
(Choueiri, 1998), but the expressions worked out seem hardly
applicable to different configurations or operating regimes, so
that in the end the simple expression of equation (26) remains
preferable for a general use.
Unfortunately, depending on the thruster operating point,
other real-world effects of a more elusive and malign nature
come into play to complicate the picture. This can be better illustrated by looking at the electrical characteristic; that

Figure 6. Effects of the back-EMF (a) vs. idealized model (b).

Figure 7. Voltage-Current characteristic of a self-field MPD


Thruster.

is, the curve describing the terminal voltage as a function


of the arc current (Figure 7). Based on the ideal model, at
constant mass flow rate this curve should display a cubic
dependence on current. But since the earliest experiments
with MPD thrusters it was shown (Boyle, Clark and Jahn,
1976) that, at lower current regimes, for all mass flow rates
the voltage tends to scale linearly with the current and the
exhaust velocity remains nearly constant. It is only beyond a
certain point that the dependence of the thrust on the square of
the current starts to be recognizable, giving the characteristic
curve the expected cubic shape.
Unfortunately, at yet higher currents a new unexpected
deviation from the normal behavior is encountered that

Magnetoplasmadynamic Thrusters 9

Figure 8. Experimental V -I characteristics for typical self-field MPD thrusters (a) comparison between different anode shapes; and
(b) comparison between different cathode lengths. Reproduced from Andrenucci et al. (1992); see also Figure 17.

vac = (2 e i /mi )1/2

(37)

(being i is the ionization potential of the involved species)


and all of the excess power fed into the thruster goes into
ionizing the remaining low-velocity neutrals rather than
further accelerating the ionized fraction. Values of the critical ionization velocity for various substances are shown in

100000
H2
He
Alfven CIV (m s1)

appears to be associated with the onset of a variety of


disturbing phenomena, including severe fluctuations of the
terminal voltage (voltage hash) and increased electrode erosion. Simultaneously, the anode losses tend to increase,
leading to a reduced efficiency. Once this condition is reached
the characteristic curve tends to revert to a linear dependence
on J.
This behavior, which in time became known as the onset
phenomenon, or simply onset, is confirmed by a large amount
of experimental data gathered in many laboratories worldwide. For example, in Figure 8 the voltage vs. current data
referring to self-field MPD thruster prototypes of different
configurations and operating conditions are shown. Such data
were obtained in Pisa in a series of experimental activities
carried out in the early 1990s (Andrenucci et al., 1992).
How can we explain such deviations from the theoretical
cubic dependence? As to the linear dependence observed at
lower currents, experiments showed that the range over which
this behavior takes place coincides with current regimes
insufficient for full ionization of the propellant flow. This
has prompted a physical interpretation related to the Critical
Ionization Velocity (CIV) phenomenon described by Alfven
(Alfve n, 1960; Choueiri, Kelly and Jahn, 1985; Turchi, 1986)
according to which, as long as ionized particles move in
the presence of significant amount of non-ionized particles,
the maximum velocity that can be achieved by the ionized
component is limited to

N2
10000

Ne
A

Li

Kr

Na

Xe
K
Cs

1000
1

10

100
Atomi weight

1000

Figure 9. Alfv`en CIV for various substances.

Figure 9.
It is only after reaching the full ionization condition that
the thruster starts complying with the cubic voltage law. But
when the onset phenomenon starts manifesting itself the characteristic swerves again toward a linear dependence. This is
easily interpreted as correlated with the entrainment of eroded
mass adding to the discharge, possibly as a consequence of
heavy erosion. Eroded mass can be expected to ablate at a rate
proportional to the square of the current, so that the self-field
thrust relation of equation (28) implies that exhaust velocities
remain constant; and indeed velocity measurements at those
regimes indicate that the exhaust velocity is independent of
current.
This phenomenon was first reported by Malliaris et al.
(1972) at the AVCO Corporation. In the attempt to increase
the current level at constant mass flow rate, they identified a critical value, (J 2 /m)
, above which the thruster
started exhibiting a noisy voltage signal and enhanced ero-

10 Alternative Propulsion

sion of thruster components. They also determined (J 2 /m)

to depend on propellant atomic weight as M1/2 and to be


smaller for larger values of the anode-to-cathode radius ratio.
Boyle, Clark and Jahn (1976) were the first to use the term
onset.
Based on a long series of experiments with argon propellant carried out mostly in Princeton on the so called Full Scale
Benchmark Thruster (FSBT) a lower bound for the onset criterion was initially estimated to be (Choueiri, Kelly and Jahn,
1987):

k =

J2
m

40kA2 s g1

(38)

Values more than 2.5 times as large for thrusters of the


same type were documented in later studies. A more general
onset criterion including the dependence on propellant atomic
weight was proposed by Hugel (1980) on the basis of different
sources:
 2 1/2 
J Ma
k =
(39)
(15 33)1010 A2 s kg1
m

This expression, graphically represented in Figure 10, is in


good agreement with the previous one for argon propellant.
The limit on the viable (J 2 /m)
in real thrusters is a
problem in many senses. First of all, as already noted, it
limits the specific impulse attainable. In addition, this limit
implies being confined to low efficiency operation, a problem that has plagued MPD thrusters for decades hindering
their introduction into flight applications. Most experimental
MPD thrusters have typically exhibited efficiencies of 25
35%, particularly at the moderate (2000 s) specific impulses
40
Babkin

K* (1010 A2 s kg 1)

20

Cory
Malliaris
Hgel

10
8
6

33.1010

IRS

15.1010

2
He Li

1
1

Ne

Ar

10

Kr Xe

100

1000

Atomic weight

Figure 10. Onset criterion. Reproduced with permission from


c DFVLR.
Hugel (1980) 

of interest to most near-term missions. This low thrust efficiency results primarily from frozen flow losses and from the
power fraction deposited in the anode voltage drop that develops in the vicinity of the anode surface (Gallimore, 1992;
Myers and Soulas, 1992). Exceedance of this limit is typically associated with increased anode losses, that for typical
MPD devices can reach as much as 50 to 90% of the input
power (Gallimore, Kelly and Jahn, 1993), not to mention
the erosion effects which would curtail the thruster lifetime.
As we shall see, frozen flow losses can be reduced by using
low ionization energy propellants such as lithium. However,
enabling an MPD thruster to provide the high-efficiency operation needed for real mission usage will require methods to
significantly reduce the fraction of power wasted in the anode.
This explains why so much time and ingenuity was dedicated over the years in the attempt to clarify and overcome
the onset problem. A brief review of these efforts is made in
the following sections.

5 THE ONSET RIDDLE


Following the work of Malliaris, contributions to the clarification of the onset phenomena came from a host of authors
in the subsequent decades. A detailed review of the enormous body of literature that was developed on the onset
over the years can be found in Uribarri (2008), Appendix D.
The sections below summarize the most significant findings
regarding onset phenomenology and the theories proposed to
explain its nature.

5.1 Onset phenomenology


Once the onset threshold is exceeded the magnitude of the
voltage noise (hash) increases slowly at first and then more
conspicuously with rising (J 2 /m).
At even higher currents the
hash is also noted to fall again (Rudolph et al., 1978; Rudolph,
1980). The characteristic frequency of the hash has been frequently described as hundreds of kHz (Hugel, 1973; Kuriki
and Iida, 1984; Kurtz et al., 1987). The erosion of all thruster
components, and in particular that of the anode, rises steadily
with increasing current, not exhibiting the rise-and-fall trend
of the voltage hash (Ho, 1981). Spots apparently associated
with current concentration and local melting appear on the
anode at discrete points.
The most prominent phenomenon signaling the onset is the
voltage hash. An example of voltage trace taken at different
current levels is given in Figure 11. The presence of characteristic frequencies in the noisy voltage traces associated
with the onset has been taken for granted until recently. The

Magnetoplasmadynamic Thrusters 11

Figure 11. The quasi-steady voltage traces for m = 3 g s1 argon, at two currents, showing the emergence of the voltage hash at higher
current, and a 100 s portion of the same traces. The currents correspond to k = 26 and 123 kA2 s g1 , respectively. Modified from Uribarri
(2008).

first author to relate anode spots to voltage oscillations was


Hugel (1973) who estimated the main frequency of voltage fluctuations at approximatley 230 kHz. Similar results
were documented by many other authors afterwards (Boyle,
Clark and Jahn, 1976; Vainberg, Lyubimov and Smolin, 1978;
Kuriki and Iida, 1984; Wagner, Kaeppeler and AuweterKurtz, 1998).
Recently Uribarri (2008) has questioned this picture. First
he has proved theoretically and experimentally (Uribarri and
Choueiri, 2008) that MPD thruster voltage measurements can
be affected by resonance of the electrical feeding lines; voltage measurements should be taken as close as possible to
thruster body in order to avoid corruption of the real signal. In addition he has shown that power spectra of voltage
measurements taken close to the thruster do not show any
preferred frequency of oscillation, but reveal that the voltage signal has the nature of a Brownian motion; that is, it is
the time integration of a random signal (Figure 12). What is
even more important is that voltage hash statistics are very
similar for anodes made of deeply different materials (lead,
copper and graphite were used), thus showing that voltage
fluctuations are presumably driven by a fundamental plasma
mechanism and not by anode erosion.
Thus, according to Uribarri (2008), previous detections of
peculiar frequencies in the voltage hash are to be attributed to
either a misinterpretation of the fluctuations, or . . . a source
of corruption such as the power supply.
As regards the thruster components erosion phenomena,
starting at onset conditions all thruster components suffer
from intense ablation and degradation, particularly the anode,
thus reducing thruster lifetime. Anode damage, melting and
discoloration, are traces of the transition taking place in the
current pattern from a diffuse fashion to a spotty one. Uribarri has shown that the severity of anode damage depends
essentially on anode material, even if a general increase in

Figure 12. Power spectrum of voltage signals taken on the


Princeton Benchmark Thruster revealing a 1/f trend operation at
k = 69 kA2 s g1 , being k* 60 kA2 s g1 . Modified from Uribarri
(2008).

damage severity is observed with increasing (J 2 /m).


Indeed,
lead anodes show severe damage also at (J 2 /m)
values much
lower than those at which intense voltage hash begins to be
observed, while anodes made of graphite present no evident
marks of damage also after several firings at (J 2 /m)
values
much greater than critical ones.

5.2 Onset theories


The majority of theories developed to explain the onset phenomena fall into two categories: anode starvation and plasma
instabilities. These two perspectives are indeed compatible
to some extent, in that starvation is often seen as a triggering
mechanism for plasma instability.

12 Alternative Propulsion

5.2.1

Anode starvation

By anode starvation or anode crisis we mean a decrease in the


density of charge carriers near the anode up to a point at which
the anode can no longer collect the total current imposed
by the external source. The anode starvation model argues
that with increasing current levels, a condition is reached in
which the current collected at the anode becomes sheathlimited. The value of the sheath-limited current is taken to
correspond to the random thermal flux of electrons across the
sheath. Attempts to conduct current greater than the sheathlimited current result in onset phenomena (Baksht, Moizhes
and Rybakov, 1974; Korsun, 1974; Vainberg, Lyubimov and
Smolin, 1978; Kurtz et al., 1987).
The total current collected at the anode is given by the
integration of charge carrier fluxes; that is, current densities,
over its surface. At low current operation, well below k ,
the anode sheath is slightly electron-repelling and the local
current density can be expressed as
j=



en
esh
vth exp
4
kB Te

(40)

n being the particle density in the neutral region outside the


sheath, a the magnitude of the anode sheath potential barrier,
Te the electron temperature, kB the Boltzmann constant and
vth the electron average thermal velocity


vth =

8 kB Te
me

(41)

As long as the anode barrier is retarding, the current can be


increased if the barrier is lowered. But once the barrier has
vanished, the local current density cannot exceed the random thermal flux of electrons, usually called the electron
saturation current:
jsat =

en
vth
4

(42)

Trying to drive more current than that resulting from the


integration of the electron saturation current density over the
entire anode surface leads to a reversal in the sign of the
anode sheath from negative, or electron repelling, to positive, or electron attracting. If the particle density near the
anode decreases, a large anode fall voltage develops because
the anode potential needs to increase to the level required for
ion generation. But under such conditions a diffuse anode
attachment becomes impossible and the current breaks down
to discrete anode spots with local anode vaporization. This
transition, with all its associated detrimental phenomena,

which include various possible instabilities in addition to spot


formation, is identified with the onset.
The decrease of particles density near the anode, that the
model indicates as the root cause for the onset, is mainly due
to the Hall effect, that is, to the Lorentz-force pinching component deriving, as noted in Section 4, from the interaction
of the axial component of current density with the azimuthal
component of self-induced magnetic field. This also prompts
the idea that any increase in the anode-adjacent particle density, through propellant species or geometry changes, should
delay starvation.

5.2.2

Plasma instabilities

The second great branch of theoretical models trying to


explain onset phenomena is related to plasma instabilities.
These theories basically state that, at critical operation, conditions are created in thruster channel for the development of
a variety of unstable oscillation modes.
One such type of instabilities most frequently evoked are
the so-called drift instabilities, which are excited by large
relative velocities between electrons and ions; that is, large
currents. The criterion for this instability is taken to coincide
with a critical drift velocity which the electrons attain when
the driven current exceeds a threshold (Shubin, 1976; Wagner,
Kaeppeler and Auweter-Kurtz, 1998).
Other authors have also shown that MPD thrusters are
prone to the development of a variety of microinstabilities, among which the Bunemann instability, the generalized
lower hybrid drift instability, the electron cyclotron drift
instability, the ion-acoustic instability and the drift cyclotron
instability (Tilley et al., 1996; Choueiri, Kelly and Jahn,
1990, 1991, 1992; Choueiri, 2001) the space charge or
Pierce instability (Maurer, Kaeppeler and Richert, 1995;
Wagner, Kaeppeler and Auweter-Kurtz (1998, 1998)), the
Wardle instability (Di Vita et al., 2000). Actually most
of such results, while not particularly enlightening about
onset phenomena, seem much more useful to the explanation
of a variety of anomalous transport and energy absorption
effects.

5.2.3

Other onset theories

Besides the onset theories reviewed above, a number of additional theories exist in the literature.
In some of these theories onset is induced by back EMF. As
was shown earlier (Section 4), the back electro-motive force is
responsible for reducing the effective electric field seen by the
plasma in the central part of the acceleration channel, and consequently the electrode current attachment zone. According
to Lawless and Subramaniam it is possible for the acceler-

Magnetoplasmadynamic Thrusters 13
ator plasma to flow quickly enough to impede current from
flowing between the electrodes; they hence hypothesize that
this mechanism is at the base of the onset phenomenon (Lawless, 1987; Subramaniam and Lawless, 1987; Subramaniam,
1991).
Some of the theories developed to explain the onset
have put the blame on macroscopic rather than microscopic instabilities. The onset of rotating disturbances in the
interelectrode region and exhaust jet of an MPD arc had
been experimentally observed since early studies (Larson,
1968;Allario, Jarrett Jr. and Hess, 1970). Schrade, AuweterKurtz and Kurtz (1985) and Schrade, Wegmann and Rosgen
(1991) have suggested that onset may result from a macroscopic instability in a current-carrying channel originating at
the tip of the cathode.
Joint work along similar lines was carried at Centrospazio
(now Alta), Pisa, and at Consorzio RFX, Padova (Zuin et al.,
2004a, 2004b). They attributed the observed oscillations in
terminal voltage as well as in temperature and magnetic field
measurements above certain values of total current to the
inception of MHD kink instability, both in self-field and
applied-field MPD thrusters.
These theories are generally lacking in one way or another,
in that they seem applicable to specific configurations or
operating conditions rather than addressing the fundamental origin of onset in the most general sense. In addition,
although sometimes proving reasonably capable at predict
ing values of (J 2 /m)
, none of the above theories can
fully explain the appearance of the voltage hash or the
spotty current attachment taking place near or beyond the
onset.
Sometimes, the existence of anode spots is simply
assumed without attempting to explain their origin; the
voltage hash is then explained as a result of the formation, extinction, and movement of anode spots. The work of
Diamant, Choueiri and Jahn (1998) provided useful insights
along this line of thought.

Figure 13. (a) Self-field MPD thruster; (b) applied field MPD thruster.

More recently, Di Vita et al. (2000) and Uribarri (2008)


have hypothesized that spot generation can follow from a
plasma instability known as the filamentation instability,
which causes the current to fragment into many channels,
irrespective of the anode material. Current filamentation is
strongly reminiscent of the anode spots phenomenon, and
it has been observed in other plasma-pinch devices that
present analogies with MPD thrusters (Feugeas and Pamel,
1989;Milanese, Niedbalski and Moroso, 2007). Thus, the filamentation approach may represent a promising clue to the
understanding of the onset.

6 APPLIED-FIELD MPD THRUSTERS


A related technology, perhaps more amenable to near-term
application, is the so-called applied-field MPD thruster (Figure 13). In this type of thruster, an external solenoid produces
a field with meridional lines of force, arranged so as to diverge
in a nozzle fashion toward the exit (Krulle, 1998; AuweterKurtz and Kurtz, 2002). The self-induced field is often of the
same order of magnitude as the field applied, so that the magnetic field lines are twisted in a helical fashion. The strong
axial component of the magnetic field hinders the electron
flow to the anode forcing the current to follow trajectories far
downstream of the thruster exit. The thrust fraction generated
within the channel is therefore quite small and Lorentz actions
mainly result here in a swirling effect. In the region where
current stream lines bend to assume a more marked radial
component, the Lorentz actions exhibit an azimuthal component which sustains the swirling and a meridional component
which provides a blowing and a pumping contribution, both
contributing directly to the thrust.
The thrust in an Applied-Field MPD thruster can thus be
visualized as a combination of different components:

14 Alternative Propulsion

r the interaction of the azimuthal (Hall) component of the


discharge current with the applied magnetic field yields
axial and radial Lorentz forces, that can both provide a
direct or indirect contribution to the thrust, T ;
r the interaction between the radial componentH of the discharge current and the self-induced azimuthal magnetic
field results in a thrust component Tsf similar to that occurring in self-field MPD devices;
r the interaction of the radial component of the discharge
current with the axial component of the magnetic field
results in an azimuthal force component that causes the
plasma to rotate. The energy recovered from this swirl
motion can partially give rise to an axial thrust component
Tsw ;
r finally,
a gasdynamic component similar to that found in
arcjets, Tgd is generally present.

The overall thrust produced by an applied-field MPD


device can thus be expressed as

Taf = TH + Tsf + Tsw + Tgd

(43)

The azimuthal electron drift current is akin to that found


in Hall thrusters, although here the collisionality is higher.
Typical values of the Hall parameters in this type of thruster
are about 3 to 5.5. This type of thruster therefore exhibits a
behavior that is intermediate with respect to self-field MPD
and Hall thrusters and may justify expectations for more efficient operation and a lesser sensitivity to instabilities and
erosion compared to the former. Because of this, efficient
operation at lower powers is easier to obtain. On the other
hand, the combination of several types of effects makes the
physics of this thruster more difficult to understand and to
optimize. In addition, the fact that the discharge extends
considerably downstream does not favor accurate vacuum
chamber testing. Development has been hindered as a consequence. Test results obtained with noble gases have not
been encouraging, while hydrogen (again, at high specific
impulses) has provided levels of efficiency of over 50%
(Krulle, Auweter-Kurtz and Sasoh, 1998). Recent work on
lithium-fed AF-MPD thrusters has yielded over 40% at only
130 kW, with Isp up to 3500 s. A critical review of the state
of the art of Applied-Field MPD thrusters, with a detailed
compilation of the performance levels attained by AF-MPD
devices of many different types and propellants, has been
performed by Kodys and Choueiri (2005).

7 LITHIUM PROPELLANT MPD


THRUSTERS
Lithium Lorentz Force Accelerator (LiLFA) is the name
adopted to designate a variety of MPD thruster that has come
of age in the mid-nineties (Figure 14). Its operating principle is essentially identical to that of the self-field MPD
thruster. The new designation was probably intended as a
way to refresh the image of this type of device, weary with
prolonged and sometimes frustrating development efforts.
But the use of lithium vapor as a propellant, and the hollowcathode design of the center electrode may perhaps justify
the adoption of a specific name.
The choice of a low-ionization energy propellant (lithium)
in place of inert gas propellants as used by traditional MPD
thrusters, such as argon, helium, and hydrogen reduces the
power loss associated with propellant ionization, which can
represent almost 50% of the total input power especially for
power levels lower than 200 kW, and is therefore beneficial in
terms of thrust efficiency. The use of lithium also offers additional advantages in terms of reduced system complexity and
storing capability. However, no space-qualified feed system
exists for lithium propellant. As for the multi-channel design
for the central electrode, this has been proved to improve efficiency and increase thruster life-time by reducing electrode
erosion (Ageyev and Ostrovsky, 1993).
The LiLFA concept has also been implemented in the
applied-field version (AF-LFA), which aims to increase
the efficiency of Lithium-fed MPD thrusters at power levels lower than 200 kW. With the addition of an external
solenoid to enhance the magnetic field, efficient electromagnetic acceleration can be obtained at current levels too
low to induce a sufficiently large magnetic field. The AFLFA offers the advantage higher efficiencies ( 40%) at
lower power (<200 kW) compared to MPDT and LiLFA,

Figure 14. The lithium lorentz force accelerator.

Magnetoplasmadynamic Thrusters 15
while maintaining exhaust velocities (1035 km s1 ) that are
comparable. Potential applications of the AF-LiLFA include
missions requiring relatively high thrust-to-power ratios,
such as orbit transfer, N-S stationkeeping, and drag compensation (Sankaran et al., 2004).

8 SURVEY OF MAJOR R&D EFFORTS


Initially investigated in the 1960s, MPD thrusters have been
the objects of periodically funded research in the USA,
achieving slow but significant improvements in performance.
Most of the related activities were initiated and conducted for
many years at the Electric Propulsion Laboratory of Princeton University (now EPPDyL). A multi-decade experimental
activity undertaken in the early 1960s was focused on a
basic model of self-field, gas-fed, coaxial, quasi-steady MPD
thruster that came to be known as the benchmark thruster
(Figure 15).
Activities carried out in Princeton have provided most
of the available knowledge on this class of device. This
information has been collected and made available to the
community in the form of a Quasi-steady Magnetoplasmadynamic Thruster Performance Database (Choueiri and Ziemer,
2001). Also from Princeton came fundamental insights on the
involved physical phenomena, starting from the seminal work
of Robert G. Jahn (1968), through the efforts of a generation
of EP specialist graduated there, up to more recent contributions to the clarification of the onset phenomena. Many significant examples of these are cited in the previous sections.
Another huge contribution to this field since the early years
of development was given by German researchers, especially
from Stuttgart University. In time, the Institute of Space

Systems (IRS) group at Stuttgart performed testing activities on a large class of devices, ranging from simple arcjets,
to Applied-field and Self-field MPD thrusters. Steady-state
MPD Arcjets were extensively studied and tested at power
levels ranging from a few kilowatts to several hundred kilowatts, providing valuable insight on the operation of this type
of devices. Figures 16 and 17 show two of the thrusters tested
at IRS. For the ZT3 thruster, no indication of instability could

be detected up to 12700 A, where a (J 2 /m)


value of more
than 8 1010 A2 s kg1 was reached, whereas for the nozzle
type MPD thruster a critical value of ca 2.7 1010 A2 s kg1
had been found, with argon propellant, with all thrusters of
the DT series (Auweter-Kurtz and Kurtz, 2008).
Researchers from Stuttgart also carried out extensive
theoretical work on the onset problem, with important contributions on the anode-starvation theory and both microscopic
and large-scale instabilities; some of the most relevant of
these are included in the cited references.
In Japan, research on MPD/QS devices became very active
since the late seventies. Important contributions were given
on the theoretical ground (e.g., Kuriki, Kunii and Shimizu,
1983) while R&D activities quickly achieved the space
demonstration level. An MPD thruster was tested onboard
the Japanese Space Flyer Unit (Figure 18) as a part of electric propulsion experiment (EPEX) launched in 1995 and
retrieved by space shuttle mission STS-72 in 1996 (Toki,
Shimuzu and Kuriki, 1997). To date, this is the only operational MPD thruster to have flown in space. A database of
measured quasi-steady thruster performance has been compiled in Japan by Sasoh and Arakawa (1992).
In Italy, work on MPD thrusters was started in the eighties focusing on experiments on ring-anode thrusters similar to
Princetons benchmark thruster. Test campaigns on geometry

Figure 15. The princeton benchmark MPD thruster: rc = 0.95 cm, ra = 5.1 cm, rao = 9.3 cm, rch = 6.4 cm, ta = 0.95 cm, and lc = 10 cm.
c AIAA.
Reproduced with permission from Burton, Clark and Jahn (1983) 

16 Alternative Propulsion

Figure 16. Schematic and test firing of the DT2 nozzle type MPD thruster of the IRS, Stuttgart. Adopted from Auweter-Kurtz and Kurtz
(2008).

Figure 17. Schematic and test firing of the ZT3 cylindrical thruster of the IRS, Stuttgart. Adopted from Auweter-Kurtz and Kurtz (2008).

Figure 18. Integration of the EPEX experiment on the Space Flyer Unit and the MPD thruster.

and scale effects (Figure 19) were carried out with heated
cathode quasi-steady MPD thrusters. Cathode heating was
aimed at assessing the impact of cathode temperature on
cathode phenomena, onset characteristics and performance
levels of the thrusters tested. Joint work on a Hybrid Plasma
Thruster - an MPD thruster with a pre-ionization chamber,
windowed anode and short cathode was carried out in Pisa in

collaboration with the Moscow Aviation Institute (Tikhonov


et al., 2000; Paganucci et al., 2001). Also, the Pisa group at
Centrospazio/Alta and Consorzio RFX, Padova, jointly performed theoretical and experimental work for the study of
macroscopic instabilities of the helical kink type.
A huge variety of MPD thruster concepts were investigated
in the then Soviet Union, starting in the late fifties (Gorshkov

Magnetoplasmadynamic Thrusters 17

Figure 19. Geometries of the Pisa thrusters and one of the thrusters during test.

et al., 2007). The scale of the efforts produced there is imposing and the number of contributors so large to defy any
attempt to cite them here. R&D work was conducted at different institutions, and in particular at the Keldish Research
Center (Figure 20), RSC Energia and DB Fakel (Figure 21)
and the Moscow Aviation Institute (Figure 22). Thrusters
tested included Self-field and Applied-field devices, steadystate devices with power levels up to MW and all types of
propellants, with lithium vapor providing the most efficient
performance (Table 1).
It was the Russians who demonstrated the advantages
obtainable by the use of Lithium. Lithium-fed MPD thrusters
were operated at power levels of several hundred kilowatts,
with efficiencies of 45 percent and plasma exhaust velocities approaching 50 000 m s1 . Tests of up to a 500 h firing
duration at 500 kW were successfully completed. A severalthousand hour life capability was projected, sufficient for
most of the space missions this thruster was cenceived for.

In 1996 the RIAME/MAI team lead by Professor Viktor


Tikhonov started a new investigation of Li-MPD thrusters
under NASA contract to demonstrate the level of the Russian
technology for further research. Laboratory model, appliedfield Li-MPD thrusters with power levels of 30 kW and
200 kW were built and tested. Following this activity, facilities to investigate lithium-fed MPD thrusters were established
in the United States at Princeton University and the NASA
Jet Propulsion Laboratory (Goebel et al., 2005). A 200 kW
version of the lithium-fed MPD thruster called the lithium
Lorentz force accelerator (Li-LFA, Figure 23) tested at the
EPPDyL laboratory in Princeton, has been claimed to have
achieved erosion-free operation over 500 h of steady thrusting at 12.5 N, 4000 s Is, and 48% effciency (Choueiri et al.,
1996).
Based on such activities, JPL started a program to develop
a 500 kW Li-LFA. The conceptual design of the thruster,
the 250 kW ALPHA2 thruster, is illustrated in Figure 24.

18 Alternative Propulsion

Figure 20. Self-field (a) and applied field (b) MPD thrusters of the Keldish Research Center. Reproduced with permission from Gorshkov
c IEPC.
et al. (2007) 

c IEPC.
Figure 21. Lithium thrusters tested at Energiya (a) and Fakel (b). Adopted from Gorshkov et al. (2007) 

This thruster, featuring a flared anode geometry incorporating Lithium heat pipes, a multichannel hollow cathode and
applied-field solenoid was targeted at achieving an efficiency
level in excess of 60% at Isp of 6200 s for a projected lifetime
of more than 3 years (Goebel et al., 2005).

9 FUTURE PROSPECTS
As of the end of the first decade of the twenty-first century
and almost fifty years after its conception, MPD propulsion
can hardly be said to have fulfilled the expectations of its
inventors. This is certainly due to a variety of adverse circumstances. Since high efficiencies (>30%) are only reached at

high power (>200 kW), MPD thrusters require power levels


that are an order of magnitude higher than those typically
available on current spacecraft in order to be competitive
with other propulsion concepts. Therefore, research on MPD
propulsion has been left aside in recent years, in favor of
thrusters offering higher efficiencies at lower power levels.
As a technology inherently suited for high power applications, it could hardly find opportunities in the scant mission
scenarios of the post-Apollo era.
But apart from external factors, it must be said that this
so promising concept has shown in time its own drawbacks. The factors preventing achievement of performance
levels suitable for mission usage, onset in particular, have
proved particularly impervious to penetrate, understand and

Magnetoplasmadynamic Thrusters 19

c IEPC.
Figure 22. A 200 kW thruster tested at RIAME/MAI. Reproduced with permission from Gorshkov et al. (2007) 

Table 1. Russian experience in Li-fed MPD thrusters (Gorshkov et al. (2007)).


Organization
NIITP
Fakel
Energiya
Energiya
Energiya
MAI

Power (kW)

Current (kA)

Specific Imp. (s)

Efficiency (%)

Typical Duration

3001000
300500
300500
500
250500
300500

615
69
69
9
58
69

35005000
35004500
35004500
4500
30004500
35004500

4060
4060
4060
55
3555
4060

5 min
30 min
30 min
500 hours
3060 min
30 min

Notes
NIITP design
Energiya design
Endurance test
Cathode failure
Energiya design

c
Figure 23. The Li-LFA thruster tested in Princeton and at JPL. (a): Reproduced with permission from Choueiri and Ziemer (2001) 
AIAA. and (b): Goebel et al (2005).

20 Alternative Propulsion
In conclusion, while no present operational spacecraft
employs MPD propulsion systems, ongoing and future R&D
activities may result in further improvements in the performance and lifetime of steady-state MPD thrusters. As
research continues, the efficiency of MPD thrusters will
gradually increase, hopefully achieving levels compatible
with the requirements of future space missions. Once higher
power levels are available in space, MPD thrusters could then
become the method of propulsion that carries humans to other
planets in our solar system.

REFERENCES

Figure 24. Conceptual design of the ALPHA2 LFA thruster. Reproduced from Goebel et al. (2005).

circumvent, despite an impressive amount of efforts invested


in this attempt.
Testing is another critical issue. Steady-state testing at the
megawatt level is difficult, and to date all data in the 1
6 MW range has been taken in quasi-steady mode. So far,
steady-state data is limited to less than 1 MW. The NASAGRC test facility had the capability to operate at steady-state
power level of up to 600 kW. Facilities to investigate lithiumfed MPD thrusters have been established in the United
States at the NASA Jet Propulsion Laboratory and Princeton
University.
Despite all shortcomings, the MPD thruster has proved to
be the only type of electric thruster capable of processing
megawatts of electrical power in a small, simple, compact
device with thrust densities of the order of 105 N m2 . NASA
is currently researching both pulsed and continuous forms
of MPD thrusters with hydrogen or lithium as a propellant.
Lithium-fed thrusters in the power range of 0.5 to 1 MW
would be ideal for near-term applications requiring Isp levels of 40006000 s, such as orbit transfer and Mars cargo
applications. One to 5 MW lithium thrusters may be suitable
to fulfill mid-term propulsion requirements, such as initial
piloted Mars missions. For even higher power levels, the terminal voltage with lithium seems too low to process the high
power levels involved; hydrogen should be capable of providing the required efficiency at Isps of 1000015000 s, paving
the way for piloted missions to Mars and the outer planets
(Polk, 2005).

Ageyev, V.P. and Ostrovsky, V.G. (1993) High-current stationary


plasma accelerator of high power. 23rd International Electric
Propulsion Conference, Seattle, WA, USA. IEPC-93-117.
Alfven, H. (1960) Collision between a nonionized gas and a magnetized plasma. R. Mod. Phys., 32(4), 710713.
Allario, F., Jarrett, O. Jr.,and Hess, R.V. (1970) Onset of rotating
disturbance in the interelectrode region and exhaust jet of an MPD
arc. AIAA J., 8(5), 902907.
Andrenucci, M., Paganucci, F., Grazzini, P. and Pupilli, F. (1992)
Scale and Geometric Effects on the Performance of MPD
Thrusters. 28th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, Nashville, TN, USA. AIAA Paper 92-3159.
Auweter-Kurtz, M., Boie, C., Kaeppeler, H.J., Kurtz, H.L., Schrade,
H.O., Sleziona, P.C., Wagner, H.P. and Wegmann, T. (1994)
Magnetoplasmadynamic thrusters: design criteria and numerical
simulation. Int. J. Appl. Electromagn.Mater., 4, 383401.
Auweter-Kurtz, M. and Kurtz, H. (2002) Arc devices for space
propulsion. ISU/AAAF Short Course: Introduction to Space
Propulsion, Versailles, France.
Auweter-Kurtz, M. and Kurtz, H. (2008) High-power and highthrust-density electric propulsion for in-space transportation, in
Nuclear Space Power and Propulsion Systems, (ed. C. Bruno)
(2008) AIAA Progress in Astronautics and Aeronautics Series,
vol. 225, Chapter 4, Reston, VA.
Baksht, F.G., Moizhes, B.Y. and Rybakov, A.B. (1974) Critical
regime of a plasma accelerator. Sov. Phys. Tech. Phys., 18(12),
16131616.
Bittencourt, J.A. (1986) Fundamentals of Plasma Physics. Pergamon Press, Oxford.
Boyle, M.J., Clark, K.E. and Jahn, R.G. (1976) Flowfield
characteristics and performance limitations of quasi-steady
magnetoplasmadynamic accelerators. AIAA J., 14(7), 955
962.
Burton, R.L., Clark, K.E. and Jahn, R.G. (1983) Measured performance of a multimegawatt MPD thruster. J. Spacecraft Rockets,
20(3), 299304.
Chen, F.F. (2006) Introduction to Plasma Physics and Controlled
Fusion. 2nd edn, vol. 1, Springer, New York.
Clark, K.E. and Jahn, R.G. (1970) Quasi-steady plasma acceleration. AIAA J., 8(2), 216220.

Magnetoplasmadynamic Thrusters 21
Choueiri, E.Y., Kelly, A.J. and Jahn, R.G. (1985) The Manifestation of Alfvens Hypothesis of Critical Ionization Velocity in
the Performance of MPD Thrusters. 18th International Electric
Propulsion Conference, Alexandria, VA., USA, SeptemberOctober 1985. AIAA Paper 85-2037.
Choueiri, E.Y., Kelly, A.J. and Jahn, R.G. (1987) MPD Thruster
Plasma Instability Studies. 19th International Electric Propulsion Conference, Colorado Springs, CO, USA. AIAA Paper
87-1067.
Choueiri, E.Y., Kelly, A.J. and Jahn, R.G. (1990) Current-Driven
Plasma Acceleration Versus Current-Driven Energy Dissipation
Part I: Wave Stability Theory. 21st International Electric Propulsion Conference, Orlando, FL, USA. AIAA Paper 90-2610.
Choueiri, E.Y., Kelly, A.J. and Jahn, R.G. (1991) Current-Driven
Plasma Acceleration Versus Current-Driven Energy Dissipation
Part II: Electromagnetic Wave Stability Theory and Experiments.
22nd International Electric Propulsion Conference, Viareggio,
Italy. AIAA Paper 91-100.
Choueiri, E.Y., Kelly, A.J. and Jahn, R.G. (1992) Current-Driven
Plasma Acceleration Versus Current-Driven Energy Dissipation
Part III: Anomalous Transport. 28th AIAA/ASME/SAE/ASEE
Joint Propulsion Conference, Nashville, TN, USA. AIAA Paper
92-3739.
Choueiri, E.Y., Chiravalle, V., Miller, G.E., Jahn, R.G., Anderson, W. and Bland, J. (1996) Lorentz Force Accelerator with
an Open-ended Lithium Heat Pipe. AIAA 32nd Joint Propulsion
Conference, Lake Buena Vista, FL, USA. AIAA Paper 96-2737.
Choueiri, E.Y. (1998) Scaling of thrust in self-field magnetoplasmadynamic thrusters. AIAA J. Propul. Power, 14(5), 744753.
Choueiri, E.Y. (2001) Instability of a current-carrying finite-beta
collisional plasma. Phys. Rev., E, 64(6), 066413.
Choueiri, E.Y. and Ziemer, J. (2001) Quasi-steady magnetoplasmadynamic thruster performance database. AIAA J. Propul. Power,
17(4), 520529.
Diamant, K.D., Choueiri, E.Y. and Jahn, R.G. (1998) Spot mode
transition and the anode fall of pulsed magnetoplasmadynamic
thrusters. AIAA J. Propul. Power, 14(6), 10361042.
Di Vita, A., Paganucci, F., Rossetti P. and Andrenucci, M.
(2000) Spontaneous Symmetry Breaking in MPD Plasmas.
AIAA/ASME/SAE/ASEE 36th Joint Propulsion Conference,
Huntsville, AL, USA. AIAA Paper 2000-3538.
Ducati, A.C., Muehlberger, E. and Giannini, G.M. (1964) High Specific Impulse Thermo-ionic Acceleration. AIAA 4th International
Electric Propulsion Conference, August-September. AIAA Paper
No. 64-668.
Ducati, A.C., Giannini, G.M. and Muehlberger, E. (1965) Recent
Progress in High Specific Impulse Thermo-ionic Acceleration.
AIAA 2nd Aerospace Sciences Meeting. AIAA Paper 65-95.
Feugeas, J.,and Von Pamel, O. (1989) Current distribution during
the breakdown in a coaxial electrode system. J. Appl. Phys., 66(3),
1080.
Gallimore, A.D. (1992) Anode power deposition in coaxial mpd
thrusters. PhD thesis, Princeton University.
Gallimore, A.D., Kelly, A.J. and Jahn, R.G. (1993) Anode power
deposition in magnetoplasmadynamic thrusters. AIAA J. Propul.
Power, 9(3), 361368.

Goebel, D.M., Katz, I., Ziemer, J., Brophy, J.R., Polk, J. and Johnson, L. (2005) Electric Propulsion Research and Development
at JPL. 41st AIAA/ASME/SAE/ASEE Joint Propulsion Conference, Tucson, AZ, USA. AIAA Paper 2005-3535.
Gorshkov, O.A., Shutov, V.N., Kozubsky, K.N., Ostrovsky, V.G. and
Obukhov, V.A. (2007) Development of High Power Magnetoplasmadynamic Thrusters in the USSR. 30th International Electric
Propulsion Conference, Florence, Italy. IEPC-2007-136.
Ho, D. (1981) Erosion studies in an MPD thruster. Masters thesis,
Princeton University.
Hoyt, R.P. (2005) Magnetic Nozzle Design for High-Power MPD
Thrusters. 29th International Electric Propulsion Conference,
Princeton, NJ, USA. IEPC-2005-230.
Hugel, H. (1973) Flow Rate Limitations in the Self-Field Accelerator. AIAA 10th Electric Propulsion Conference, Lake Tahoe,
NEV, USA. AIAA Paper 73-1094.
Hugel, H. (1980) Zur Funktionsweise der Anode im Eigenfeldbeschleuniger, DFVLR-FB 80-30.
Jahn, R.G. (1968) Physics of Electric Propulsion, McGraw-Hill,
New York.
Kodys, A.D. and Choueiri, E.Y. (2005) A Critical Review of the
State of the Art in the Performance of Applied-field Magnetoplasmadynamic Thrusters. 41st AIAA Joint Propulsion Conference,
Tucson, AZ, USA. AIAA Paper 2005-4247.
Korsun, A. (1974) Current limiting by self magnetic field in a plasma
accelerator. Sov. Phys. Tech. Phys., 19(1), 124126.
Krulle, G., Auweter-Kurtz, M. and Sasoh, A. (1998) Technology
and application aspects of applied field magnetoplasmadynamic
propulsion. J. Propul. Power, 14(5), 754762.
Kuriki, K., Kunii, Y. and Shimizu, Y. (1983) Idealized model for
plasma acceleration in an MHD channel. AIAA J., 21(3), 322
326.
Kuriki, K. and Iida, H. (1984) Spectrum analysis of instabilities in
MPD arcjet. 17th International Electric Propulsion Conference,
Tokyo, Japan. IEPC-84-28.
Kurtz, H.L., Auweter-Kurtz, M., Merke, W.D. and Schrade, H.O.
(1987) Experimental MPD Thruster Investigations. 19th International Electric Propulsion Conference, Colorado Springs, CO,
USA. AIAA-87-1019.
Larson, A.V. (1968) Experiments on current rotation in an MPD
engine. AIAA J., 6(6), 10011006.
Lawless, J. and Subramaniam, V.V. (1987) Theory of onset in
magnetoplasmadynamic thrusters. J. Propul. Power, 3(2), 121
127.
Lieberman, M.A. and Lichtenberg, A.J. (1994) Principles of Plasma
Discharges and Materials Processing, Wiley, New York.
Maecker, H. (1955) Plasmastromungen in lichtbogen infolge eigenmagnetischer kompression. Zeitschrift fur Physik, Bd. 141,
198216.
Malliaris, A., John, R., Garrison, R. and Libby, D. (1972) Performance of quasi-steady MPD thrusters at high powers. AIAA J.,
10(2), 121122.
Maurer, M., Kaeppeler, H.J. and Richert, W. (1995) Calculation of
nonlinear drift instabilities in magnetoplasmadynamic thruster
flows. J. Phys. D: Appl. Phys., 28, 22692278.

22 Alternative Propulsion
Myers, R.M. and Soulas, G.C. (1992) Anode Power Deposition in
Applied-Field MPD Thrusters. 28th AIAA/ASME/SAE/ASEE
Joint Propulsion Conference, Nashville, TN, USA. AIAA Paper
92-6463.
Milanese, M.M., Niedbalski, J.J. and Moroso, R.L. (2007) Filaments in the sheath evolution of the dense plasma focus as applied
to intense auroral observations. IEEE Trans. Plasma Sci., 35(4),
808.
Mitchner, M. and Kruger C.H., Jr. (1992) Partially Ionized Gases,
Wiley, New York.
Paganucci, F., Rossetti, P., Andrenucci, M., Tikhonov, V.B. and
Obukhov, V.A. (2001) Performance of an Applied Field MPD
Thruster. 27th International Electric Propulsion Conference.
Pasadena, CA, USA. IEPC Paper 01-132.
Polk, J. (2005) Lithium-fuelled electromagnetic thrusters for robotic
and human exploration missions. Space Nuclear Conference, San
Diego, CA, USA.
Rudolph, L. (1980) The MPD thruster onset current performance
limitation. PhD thesis, Princeton University.
Rudolph, L., Jahn, R.G., Clark, K. and von Jaskowsky, W. (1978)
Onset Phenomena in Self-Field MPD Arcjets. 13th International
Electric Propulsion Conference, San Diego, CA, USA. IEPC-78653.
Sankaran, K., Cassady, L., Kodys, A.D. and Choueiri, E.Y. (2004)
A survey of propulsion options for cargo and piloted missions to
Mars. Astrodyn., Space Missions, and Chaos. The Annals of the
New York Academy of Science, 1017, 450467.
Sasoh, A. and Arakawa, Y. (1992) Electromagnetic effects in an
applied-field magneto plasmadynamic thruster. AIAA J. Propulsion Power, 8(1), 98102.
Schrade, H.O., Auweter-Kurtz, M. and Kurtz, H. (1985) Stability Problems in Magneto plasmadynamik Arc Thrusters. AIAA
18th Fluid Dynamics, Plasma dynamics and Lasers Conference,
Cincinnati, Ohio, USA. AIAA-85-1633.
Schrade, H.O., Wegmann, T. and Rosgen, T. (1991) The Onset
Phenomena Explained by Run-away Joule Heating. 22nd International Electric Propulsion Conference, Viareggio, Italy. AIAA
Paper 91-022.
Shubin, A. (1976) Dynamic nature of critical regimes in steadystate high-current plasma accelerators. Sov. J. Plasma Phys., 2(1),
1821.
Spitzer, L. Jr. (1964) Physics of Fully Ionized Gases, Wiley, New
York.
Subramaniam, V.V. (1991) Onset and Erosion in Self-field MPD
Thrusters. 22nd International Electric Propulsion Conference,
Viareggio, Italy. AIAA Paper 91-021.

Subramaniam, V.V. and Lawless, J.L. (1987) Onset in Magnetoplamadynamic Thrusters with Finite Rate Ionization. 19th
International Electric Propulsion Conference, Colorado Springs,
CO, USA. AIAA-87-1068.
Tikhonov, V.B., Antropov, N.N., Dyakonov, G.A., Obukhov, V.A.,
Paganucci, F., Rossetti, P. and Andrenucci, M. (2000) Investigation on a New Type of MPD Thruster. 27th EPS Conference
on Controlled Fusion and Plasma Physics. Budapest, Hungary,
ECA Volume 24B, pp. 8184.
Tilley, D.L., Choueiri, E.Y., Kelley, A.J. and Jahn, R.G.
(1996) Microinstabilities in a 10-kilowatt self-field magnetoplasmadynamic thruster. J. Propul. and Power, 12(2), 381
389.
Toki, K., Shimuzu, Y. and Kuriki, K. (1997) Electric Propulsion
Experiment (EPEX) of a Repetitively Pulsed MPD Thruster System Onboard Space Flyer Unit (SFU). 25th International Electric
Propulsion Conference, Cleveland, OH, USA. IEPC-97-120.
Turchi, P.J. (1986) Critical speed and voltage-current characteristics
in self-field plasma thrusters. AIAA J. Propul. Power, 2(5), 398
401.
Uribarri, L. (2008) Onset voltage hash and anode in quasi-steady
magnetoplasmadynamic thrusters. PhD thesis, Princeton University.
Uribarri, L. and Choueiri, E.Y. (2008) Corruption of pulsed
electric thruster voltage fluctuation measurements by transmission line resonances. AIAA J. Propul. Power, 24(3), 637
639.
Vainberg, L., Lyubimov, G. and Smolin, G. (1978) High-current
discharge effects and anode damage in an end-fire plasma accelerator. Sov. Phys. Tech. Phys., 23, 439443.
Wagner, H.P., Kaeppeler, H.J. and Auweter-Kurtz, M. (1998a) Instabilities in MPD thruster flows: 1. space charge instabilities in
unbounded and inhomogeneous plasmas. J. Phys. D: Appl. Phys.,
31, 519528.
Wagner, H.P., Kaeppeler, H.J. and Auweter-Kurtz, M. (1998b)
Instabilities in MPD thruster flows: 2. investigation of drift and
gradient driven instabilities using multi-fluid plasma models. J.
Phys. D: Appl. Phys., 31, 529541.
Zuin, M., Cavazzana, R., Martines, E., Serianni, G., Antoni, V.,
Bagatin, M., Andrenucci, M., Paganucci, F. and Rossetti P.
(2004a) Critical regimes and magnetohydrodynamic instabilities in a magnetoplasmadynamic thruster. Phys. Plasmas, 11(10),
47614770.
Zuin, M., Cavazzana, R., Martines, E., Serianni, G., Antoni, V.,
Bagatin, M., Andrenucci, M., Paganucci, F. and Rossetti P.
(2004b) Kink instability in applied field MPD thrusters. Phys.
Rev. Lett., 92, 225003.

You might also like