UCLA
UCLA Previously Published Works
Title
Ion ejection from a permanent-magnet mini-helicon thruster
Permalink
https://escholarship.org/uc/item/75j607q1
Journal
Physics of Plasmas, 21(9)
ISSN
1070-664X
Author
Chen, FF
Publication Date
2014-09-01
DOI
10.1063/1.4896238
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
Ion ejection from a permanent-magnet mini-helicon thruster
Francis F. Chen
Electrical Engineering Department, University of California, Los Angeles 90095-1594, USA
(Received 18 July 2014; accepted 8 September 2014; published online 26 September 2014)
A small helicon source, 5 cm in diameter and 5 cm long, using a permanent magnet (PM) to create
the DC magnetic field B, is investigated for its possible use as an ion spacecraft thruster. Such
ambipolar thrusters do not require a separate electron source for neutralization. The discharge is
placed in the far-field of the annular PM, where B is fairly uniform. The plasma is ejected into a
large chamber, where the ion energy distribution is measured with a retarding-field energy
analyzer. The resulting specific impulse is lower than that of Hall thrusters but can easily be
increased to relevant values by applying to the endplate of the discharge a small voltage relative to
spacecraft ground.
V
C
2014 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4896238]
I. INTRODUCTION
Spacecraft thrusters now in orbit are mostly of two types:
electrostatic ion thrusters and Hall thrusters. These are much
more efficient than the chemical or MHD thrusters used in
early days of space exploration. Ion thrusters create a plasma
with a voltage applied between a thermionic cathode and an
anode grid. The ions passing through that grid are accelerated
to high velocity by a second grid and form the ion beam that
pushes the spacecraft forward by momentum conservation.
Xenon is normally used because of its inertness and high
mass. A source of cool electrons is needed to neutralize the
space charge of the ion beam. Hall thrusters produce a denser
plasma by holding the electrons back with a B-field as the
ions are accelerated by an electrostatic voltage. An electron
source for neutralization is also needed here.
One measure of the effectiveness of a thruster is the spe-
cific impulse, I
sp
, defined as
I
sp
v
ex
g
m=s
m=s
2
s; (1)
where v
ex
is the ion exhaust velocity. The speed which the
spacecraft can reach depends on the amount of xenon avail-
able for ejection. This consideration of mass is out side the
scope of this paper but can be found in standard references.
1
Helicons can produce higher plasma density n than most
other discharges at the same power and therefore have been
suggested for use in thrusters. Since the plasma is not fully
ionized, neutral atoms entrained by the ions will also be
ejected, thus increasing the effective I
sp
. Thus, only a lower
limit to I
sp
can be found from our measurements. A large lit-
erature has arisen on helicon thrusters, and reviews of this
subject have been published by Charles.
2,3
Helicons require
a DC magnetic field, and almost all experiments so far have
produced the B-field with heavy copper coils driven by a DC
power supply. By designing a small but dense helicon source
using permanent magnets, we have greatly reduced the size
and weight of the thruster. In this paper, we show how the
design was made and how I
sp
was measured using an ion
energy analyzer. We have increased I
sp
by applying an elec-
tric potential, as it is done in Hall thrusters.
II. DESIGN OF THE HELICON SOURCE
A. Tube design
Among plasma discharges, helicons are probably the
most complicated, since they involve magnetic fields, neu-
tral collisions, fancy antennas, and RF coupling mecha-
nisms that include parametric instabilities. Fortunately, the
design of an experiment has been fa cilitat ed by the pro-
gram HELIC written by Arnush.
4
After setting the RF
frequency, the gas, and the discharge dimensions, the
most useful plot that one can obtain is that of R(n,B), w her e
R is the plasma resistance, proportional to the RF energy
absorption. A typical result is shown in Fig. 1. To over-
come circuit r esistanc e, R should be at least 1 X.Stable
operation occurs on the downward slope of each curve,
where a drop in n leads to higher absorption, and vice
versa . R(n,B) plots for various discharge tube radii a and
height h led to the tube shown in Fig. 2. The top plate
(endplate) has to be conducting (aluminum), so that the
reflected back-wave interferes constructively with the
downward forward wave when h has the proper value.
The antenna is placed as close as possible to the exit to
avoid losing plasma to the sidewalls. The “skirt” on the
bottom is to minimize induced currents in the flange on
which the tube is mounted.
FIG. 1. A HELIC plot of plasma resistance R vs. n and B for the tube dimen-
sions in Fig. 2.
1070-664X/2014/21(9)/093511/5/$30.00
V
C
2014 AIP Publishing LLC21, 093511-1
PHYSICS OF PLASMAS 21, 093511 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.67.194.141 On: Mon, 29 Sep 2014 17:21:57
B. Magnet design
Annular magnets polarized vertically have a strong field
inside the hole, but plasma produced there will flow side-
ways along field lines instead of downward. The discharge is
therefore placed below the stagnation point, where the
B-field is weaker but is quite uniform and extends to infinity.
The bottom rectangle in Fig. 3 shows a sample position of
the discharge tube. The B-field can be varied by moving the
magnet vertically relative to the tube. Thus, the common
notion that PMs cannot provide variable B-fields is not a real
restriction. The magnet shown in Fig. 3 is a neodymium
(NdFeB) magnet of 3 in. (7.6 cm) inner and 5 in (12.7 cm)
outer diameters and is 1 in. (2.54 cm) thick. Previous experi-
ments with this magnet showed that its strong B-field was
not necessary, and it has been replaced with a commercially
available Nd magnet of 2 in. ID and 4 in. OD,
1
=
2
in thick. A
picture of the discharge with this smaller magnet is shown in
Fig. 4.
Permanent magnets have also been used for helicon
thrusters by Takahashi et al.
5,6
In a much larger experimen t,
the magnets are linear rather than the annular ones used here
in their remote, reverse fields. Takahashi measured the thrust
directly by suspending the entire magnet þ discharge system
and measuring its displacement when the RF is turned on.
III. EXPERIMENTAL SETUP
A. Plasma chamber
The helicon source is mounted on the large chamber
shown in Fig. 5. The retarding-field energy analyzer (RFEA)
is mounted on a vertical shaft that can position it at variable
distances z below the source. Three ports on the side permit
the insertion of probes for radial measurements at three z’s.
Unless otherwise specified, “standard conditions” are with
the large magnet of Fig. 3, a pressure of 15 mTorr of argon,
and 400 W of 27.12 MHz RF power.
B. Ion analyzer
The RFEA is the SEMion
TM
made by Impedans, Ltd.
7,8
The system is designed to work in a strong RF environment.
The sensor is shown in Fig. 6. Its aluminum case is approxi-
mately 10 cm in diameter and 1 cm thick, covered with alu-
minum oxide. Its capacitance to ground is small, so that the
FIG. 2. Final design of the discharge tube. The antenna has three turns for
13 MHz and one turn for 27 MHz.
FIG. 3. Field lines of an annular permanent magnet and possible position of
the discharge tube.
FIG. 4. Photo of the helicon discharge with a commercial Nd magnet at its
optimum height. The top plate (endplate) of the discharge is aluminum and
can be grounded or biased with a battery relative to machine ground.
093511-2 Francis F. Chen Phys. Plasmas 21, 093511 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.67.194.141 On: Mon, 29 Sep 2014 17:21:57
entire grid assembly, shown in Fig. 7, follows the RF at the
local floating potential. The orifices are 800 lm in diameter,
and gr id G
0
has 20 lm holes, smaller than the local Debye
length. The grid G1 repels electrons; G2 provides the swept
retarding voltage on the ions; and C is the ion collector. The
insulated cable carries the voltages to a filter box, which
removes most of the RF and provides DC voltages through a
vacuum seal to the control unit. The latter displays I(V) and
differentiates it to show the ion distribution. The DC voltage
of G
0
was carefully measured with an oscilloscope probe to
obtain the floating potential to relate the RFIDs to machine
ground.
C. Langmuir probes
The probe tips are of 5-mil (0.127 mm) diameter tung-
sten, 1 cm long. The IV curves are obtained with Hiden’s
ESP
TM
hardware and analyzed with the author’s own soft-
ware. The probe results verified that the plasmas in this
experiment are the same as those generated in previous years
and published previously.
9
However, when the RFEA sensor
is close to the discharge tube, where there is no probe, it
forms an endplate for the discharge and under normal cir-
cumstances would change the helicon into a shorter dis-
charge. In this case, the RFEA is a floating, so it should not
affect the discharge as much as a grounded endplate. To
check this, we swung the sensor away from under the tube
and checked the RF matching. Only a small change in the
tuning was needed.
IV. MEASUREMENTS
The retarding-field ion distributions (RFIDs) obtained
are summarized in Fig. 8. The voltages set by the sensor are
relative to the floating potential, and these have been con-
verted to voltage relative to ground by subtracting the float-
ing potential. The voltage then represents the ion energy in
eV. This process yields negative voltages, which are mean-
ingless, since ion traveling upwards toward the back of the
sensor cannot be collected. The negative voltages arise from
FIG. 5. Diagram of the experimental chamber. The magnet is the one shown in
Fig. 3; its support is not shown. The top plate of the discharge tube is normally
grounded but can be biased with a battery as shown. Dimensions are in cm.
FIG. 6. View of the RFEA from the top.
FIG. 7. Diagram of the RFEA.
093511-3 Francis F. Chen Phys. Plasmas 21, 093511 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.67.194.141 On: Mon, 29 Sep 2014 17:21:57
the heavy filtering done by the filter unit should be neglected.
The RF oscillations in floating potential can be as large as
660 V. The sensor is set at Port 2, 16.9 cm below the tube,
as seen in Fig. 5 . Since the ion-neutral mean free path
15 mTorr of argon pressure is approximately 1 cm, the ion
velocities in Fig. 8 are lower than their exit velocities from
the source. Unfortunately, the sensor cannot be placed at
Port 1 because it is overheated there. Note that the ion veloc-
ity increases very little with RF power, a consequence of the
fact that KT
e
does not change much with power. At 200 W in
Fig. 8, the helicon may have gone into a different mode.
The change of RFIDs with distance from the source is
shown in Fig. 9 for P
rf
¼ 400 W. The ions slow down from
neutral collisions and lose density due to diffusion, as
expected.
To calculate the specific impulse, one should plot the
data of Fig. 8 against ion velocity and average over velocity,
since I
sp
is proportional to velocity. However, since filtering
has extended the curves into negative velocities, this would
not yield accurate results. The curves are fairly symmetric,
and it is sufficient to measure only the voltages (i.e., ion
energies) at their peaks. Inserting these values into Eq. (1)
yields the lower curve for I
sp
in Fig. 10. This raw value of I
sp
is too low to be useful, but it can be increased by adding a
bias to the endplate of the helicon source. Because of the
severe RF environment near the helicon discharge, an elec-
tronic power supply cannot be used. We have used small Pb-
acid (car) batteries to apply the bias. Applying þ24 V to the
endplate with two 12 V batteries in series, we obtained the
I
sp
shown by the top curve in Fig. 10. The value of 1200
there is comparable to what is currently obtained in existing
Hall thrusters. For instance, a geostationary satellite operat-
ing at 200 V has a v
ex
of 15 km/s, giv ing I
sp
1500 s.
10
This can be achieved with much lower voltage in helicon
thrusters.
To show the presence of a strong ion beam even in the
absence of a battery, the RFIDs were measured with the sen-
sor facing down, away from the source. The up-down asym-
metry is shown in Fig. 11. The source fills the large chamber
with plasma, and the backflow due to collisions is measured
when the sensor is reversed. In the UP position, the RFID
amplitude increases by only about 10% when a þ24 V bias
is applied to the endplate, indicating that plasma density is
not greatly affected by the bias.
At the much lower pressures encountered in space, the
ions are not slowed down by collisions as in the above data,
and the I
sp
’s are much higher. Figure 12 shows the RFIDs vs.
pressure. The ion density goes down but the energy increases
with pressure. For ions of order 2 eV (accelerated by E-fields
FIG. 9. Z-scan of ion distributions under standard conditions using the small
magnet of Fig. 4. The curves are in the order of the legend. The power de-
pendence is illustrated by the dashed curve (1000 W), compared with the
bottom curve (400 W).
FIG. 10. Specific impulse from a PM helicon thruster at 15 mTorr: intrinsic
(w) and with þ24 V bias (•).
FIG. 11. Ion distributions with the sensor facing the source (up) or away
from it (down).
FIG. 8. Ion energy distributions at Port 2 with the large magnet.
093511-4 Francis F. Chen Phys. Plasmas 21, 093511 (2014)
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
164.67.194.141 On: Mon, 29 Sep 2014 17:21:57
