Electrodeless plasma acceleration system

using rotating magnetic field method

Cite as: AIP Advances 7, 115204 (2017); https://doi.org/10.1063/1.4998248
Submitted: 29 July 2017 . Accepted: 26 October 2017 . Published Online: 03 November 2017

T. Furukawa , K. Takizawa, D. Kuwahara, and S. Shinohara

COLLECTIONS

This paper was selected as Featured

ARTICLES YOU MAY BE INTERESTED IN

Study on electromagnetic plasma propulsion using rotating magnetic field acceleration

scheme
Physics of Plasmas 24, 043505 (2017); https://doi.org/10.1063/1.4979677

Tutorial: Physics and modeling of Hall thrusters

Journal of Applied Physics 121, 011101 (2017); https://doi.org/10.1063/1.4972269

Space micropropulsion systems for Cubesats and small satellites: From proximate targets
to furthermost frontiers
Applied Physics Reviews 5, 011104 (2018); https://doi.org/10.1063/1.5007734

AIP Advances 7, 115204 (2017); https://doi.org/10.1063/1.4998248 7, 115204

© 2017 Author(s).
 AIP ADVANCES 7, 115204 (2017)

Electrodeless plasma acceleration system using rotating

magnetic field method
T. Furukawa,1,a K. Takizawa,1 D. Kuwahara,2 and S. Shinohara2
1 The Graduate School of Engineering, Tokyo University of Agriculture and Technology,
Tokyo 184-8588, Japan
2 Institute of Engineering, Tokyo University of Agriculture and Technology, Tokyo 184-8588,

Japan
(Received 29 July 2017; accepted 26 October 2017; published online 3 November 2017)

We have proposed Rotating Magnetic Field (RMF) acceleration method as one of

electrodeless plasma accelerations. In our experimental scheme, plasma generated by
an rf (radio frequency) antenna, is accelerated by RMF antennas, which consist of
two-pair, opposed, facing coils, and these antennas are outside of a discharge tube.
Therefore, there is no wear of electrodes, degrading the propulsion performance. Here,
we will introduce our RMF acceleration system developed, including the experimental
device, e.g., external antennas, a tapered quartz tube, a vacuum chamber, external mag-
nets, and a pumping system. In addition, we can change RMF operation parameters
(RMF applied current I RMF and RMF current phase difference φ, focusing on RMF
current frequency f RMF ) by adjusting matching conditions of RMF, and investigate
the dependencies on plasma parameters (electron density ne and ion velocity vi ); e.g.,
higher increases of ne and vi (∼360 % and 55 %, respectively) than previous exper-
imental results were obtained by decreasing f RMF from 5 MHz to 0.7 MHz, whose
RMF penetration condition was better according to Milroy’s expression. Moreover,
time-varying component of RMF has been measured directly to survey the penetra-
tion condition experimentally. © 2017 Author(s). All article content, except where
otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license
(http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1063/1.4998248

I. INTRODUCTION
In the field of a space propulsion, electric propulsion systems have been studied due to an
advantage of the higher fuel efficiency than that of chemical ones. However, conventional electric
thrusters, e.g., gridded ion thruster,1 Hall thruster,2 and Magneto-Plasma-Dynamics (MPD) thruster,3
have electrodes, which contact with plasma directly, and this interaction makes an operating time
limited and a long space exploration difficult because of the erosion of the electrodes.
In order to accomplish a longer operating time, which is in proportion to a specific impulse
(exhaust velocity divided by gravitational acceleration), and high thrust performance for a next gen-
eration space propulsion scheme, some electrodeless plasma thrusters have been proposed and studied
on, e.g., Electron Cyclotron Resonance (ECR) plasma thruster,4 Variable Specific Impulse Magneto-
plasma Rocket (VASIMR),5 ratio frequency (rf) plasma thruster.6 In these acceleration schemes, the
plasma production and heating are conducted by external rf antennas, employing microwave ECR, Ion
Cyclotron Heating (ICH), and a magnetic expansion scheme (so-called a magnetic nozzle). Among
all, helicon electrodeless plasma acceleration,6–8 using a helicon wave9 excitation to generate plasmas
(electron density ne is up to ∼ 1013 cm-3 ) with additional acceleration methods have been studied
under the Helicon Electrodeless Advanced Thruster (HEAT)7,8 project.
Here, we have proposed one of the electrodeless plasma accelerations, i.e., Rotating Magnetic
Field (RMF)10,11 acceleration scheme.7,8,12,13 This RMF method comes from the field of magnetically

a
Electronic mail: [email protected]

2158-3226/2017/7(11)/115204/13 7, 115204-1 © Author(s) 2017

115204-2 Furukawa et al. AIP Advances 7, 115204 (2017)

confined plasma fusion research in order to induce an azimuthal current to maintain the Field Reversed
Configuration (RFC). In this acceleration scheme, an rf antenna, which is outside of a discharge tube,
is used to produce a plasma, and the plasma acceleration is conducted by two pairs of opposed, facing
RMF coils, wound around the discharge tube in the downstream of the plasma generation region.
In our acceleration stage, the RMF coils are expected to induce an azimuthal current jθ , whose
mechanism will be shown in the next section, and plasmas can be accelerated by the axial Lorentz
force f z by the product of induced jθ and a radial component of an applied external magnetic field
Br . Here, the process of generating f z is described in Sec. II. It is very novel to apply this method
to electromagnetic plasma acceleration under the divergent magnetic field, and this method will be a
next generation electric propulsion scheme, if promising.
Our RMF experiment has been conducted using the Large Mirror Device (LMD),14 and we can
install RMF system and diagnostic instruments, which will be described in Sec. III. In our previous
study on RMF acceleration, we applied RMF with RMF current frequency f RMF = 5 MHz,12 and
surveyed the operation parameters (rf power Prf , gas flow rate fr, and external magnetic field).15 Here,
we found that RMF could penetrate fully at a radial position of discharge tube r = 60 mm, where
plasma radius R is 85 mm, and partially at r = 0 ∼ 45 mm into plasma, respectively, according to the
Milroy’s expression of RMF penetration condition.11 Recently, dependences of an electron density ne
and ion velocity vi on f RMF have been investigated by changing f RMF form 5 MHz to 3 MHz.13 Higher
increasing rates of ne and vi denoted as ∆ne /ne and ∆vi /vi , respectively, have been obtained: up to
∼ 70 % and 28 %, respectively with f RMF = 3 MHz than these with 5 MHz.10 We have also examined
the RMF operation conditions, e.g., RMF current I RMF and RMF current phase difference φ. Here,
φ represents the rotating direction of RMF, and φ = 90 deg. and -90 deg. denote an acceleration and
deceleration phases, respectively. Moreover, we have found that ∆ne /ne and ∆vi /vi were increased in
proportion to I RMF ,2 and they behaved nearly sinusoidal curves by changing φ.13
Here, we have further decreased f RMF from 3 MHz to 1 MHz as well as 0.7 MHz with applying
higher I RMF , since we can easily change the rf matching conditions, e.g., series/parallel capacitances
and series inductance in matching boxes in a wide range of RMF excitation frequency, and made
RMF penetration condition better, at the expense of lower induced jθ , which is proportional to
f RMF if a full penetration condition is satisfied. We have also investigated the dependencies of ne
and vi on I RMF and φ, where radial measurement positions were the same as previous experiments
with f RMF = 3 MHz. The present results show that the maximum ∆vi /vi and ∆ne /ne were ∼ 55 %
and ∼ 360 % with f RMF = 1 MHz (I RMF = 46 App ), respectively, and these results showed that
decreasing f RMF with an increment of I RMF can enhance the thermal thrust, which is represented
as surface integrals of a static pressure ne k B T e and a dynamic pressure ni mi vi ,2 where ni is
ion density (= ne ), k B is Boltzmann constant, T e is electron temperature, ni is ion density, and mi
is ion mass. This is caused by higher ne and vi obtained by the RMF method. Moreover, radial
distributions of time-varying component of RMF, b̃⊥ , which is orthogonal to the axial direction of
the discharge tube, were measured to examine the RMF penetration condition with f RMF = 0.7 MHz
and 1.5 MHz by using a magnetic probe, and full and partial penetrations were consistent with
expectation.11
This paper is organized as follows. Theoretical background of the RMF acceleration method,
i.e., the process of inducing the azimuthal current jθ , and experimental setup, including our plasma
source, RMF coils, and plasma diagnostics, are described in Sec. II and III, respectively. In
Sec. IV, the present RMF experimental results with f RMF = 0.7 MHz and 1 MHz by changing external
parameters are shown, accompanied by a better plasma performance than those with f RMF = 5 MHz.
Discussion, in particular, about RMF penetration condition and the thermal pressure, and conclusion
of our present research is described in Sec. V and VI, respectively.

II. RMF ACCELERATION METHOD

Originally, inducing an azimuthal current jθ by using RMF has been performed in the field
of nuclear fusion, as was mentioned, and we applied this method to space plasma propulsion with
higher current frequency than that of general FRC operation by one order of magnitude, considering
a higher jθ to achieve a high axial thrust effectively. Here, f RMF is in the range f ci << f RMF << f ce ,
115204-3 Furukawa et al. AIP Advances 7, 115204 (2017)

FIG. 1. Schematic diagram of the RMF coils to induce an azimuthal current jθ , generating the axial Lorentz force f z
[(a) bird’s eye view and (b) cross section view].

where f ci and f ce are an ion and electron cyclotron frequencies, respectively, and electrons can
rotate in the azimuthal direction, as shown in Fig. 1(b). We will theoretically explain the proposed
RMF acceleration process as follows. First, Rotating Magnetic Field BRMF can be given as Eq. (1),
considering a cylindrical plasma:
BRMF = Bω cos (ωRMF t − θ) er + Bω sin (ωRMF t − θ) eθ . (1)
Here, ωRMF is RMF current angular frequency (= 2πf RMF ), and Bω is an amplitude of the rotating
magnetic field, and θ is an arbitrary initial phase angle. Second, an axial oscillating electric field Ẽz
is induced due to the faraday’s electromagnetic induction law:
∂BRMF
∇ × E=− . (2)
∂t
Then, a time-varying axial current j̃z is induced by the Ohm’s law, as shown in Eq. (3):
1
E+ ( j × BRMF ) = ηj. (3)
ne e
Here, e and η are elementary charge and plasma resistivity, respectively. Third, the Hall term <j̃z b̃r >
in the second term on the LHS of Eq. (3) is proposed with the radial component b̃r [= Bω cos(ωRMF
t - θ)] of BRMF . Therefore, jθ term, which is on the RHS of Eq. (3), has a steady and a time-varying
(2ωRMF component) parts. Electrons can rotate in the azimuthal direction with the velocity veθ , which
is derived from a force balance between the drag force due to particle collisions and the driving force,
and dc component of jθ is generated as shown in Eq. (4):
jθ = −ne eveθ . (4)
Finally, the axial Lorentz force f z is generated by the product of this jθ and an external applied radial
magnetic field Br , as shown in Eq. (5), and this schematic diagram is shown in Fig. 1(a):
fz = jθ × B r . (5)
In the full penetration condition of RMF,8 the electrons rotate in the azimuthal direction like a
rigid body rotation, and veθ can reach the maximum value veθ max , as shown in Eq. (6):
veθ max = rωRMF . (6)

III. EXPERIMENTAL SETUP

A. Large mirror device
In our experiment, we use Large Mirror Device (LMD) at Tokyo University of Agriculture
and Technology, as shown in Fig. 2. The device has a 1,700-mm-long and 445-mm-inner diameter
cylindrical vacuum chamber with a 1,000-mm-long quartz tube, and is evacuated by two turbo-
molecular pumps, whose pumping speeds are 1,000 l/s and 2,400 l/s, respectively, with a typical back
115204-4 Furukawa et al. AIP Advances 7, 115204 (2017)

FIG. 2. Schematic diagram of LMD and peripheral apparatuses.

ground pressure of ∼10-4 Pa. This quartz tube has a tapered shape (100 ∼ 170-mm-inner diameter),
expected to decrease an interaction between plasmas and the tube wall, considering a divergent
magnetic field generated by permanent magnets and electromagnets. As a propellant gas, argon gas
is injected continuously from a flange port located on an upstream side of the quartz tube via a mass
flow controller, where the gas flow rate can be controlled up to 100 sccm. A half-helical antenna,
whose azimuthal mode of m = 1,16 is wound around the quartz tube to generate a plasma, whose ne can
be raised up to ∼ 1013 cm-3 if we use electromagnets as well as permanent magnets,13 whose grade is
N35 made by NeoMag Co. Ltd, located as shown Fig. 2. In our present experiment, only permanent
magnets are used, which has non-uniform magnetic field profile, so-called magnetic nozzle.17 This
field profile makes the axial thrust through the product of the radial component of the external
magnetic field Br and the induced diamagnetic current in the azimuthal direction jθ . In our previous
experiment, the magnetic nozzle effect was investigated by the use of Laser Induced Fluorescence
(LIF) method and thrust measurement by means of a target type thrust stand w/o RMF method.18,19
An excitation rf frequency for plasma production is 7 MHz for a 75 ms pulsed discharge (5 % duty)
at intervals of 1.5 s, which is controlled by a function generator. With respect to the rf discharge, a
high power source (Thamway, T162-6746A), whose maximum power is ∼ 5 kW for the frequency
with a range of 3 - 15 MHz, and used with an impedance matching box (Thamway, T020-6326AK),
which can change values of parallel/series capacitances, are utilized. Here, the typical input rf power
is ∼ 1 kW (3 kW is described only in Sec. E).

B. RMF coils
As an RMF acceleration antenna, two sets of 5 turns, opposing coils with 168-mm-axial and
115-mm-azimuthal length are at axial position of z = - 330 ∼ - 162 mm in a downstream region of
fr antenna, considering the divergent magnetic field shape to have the Br component [see Eq. (5)], as
shown in Fig. 2. The material of this RMF coils is oxygen-free copper, which has a 15-mm-width and
a 0.5-mm-thickness, shielded by an insulation tube. Phase difference φ between the two-set coils’
current I RMF can change the rotation direction of BRMF : φ = 90 and -90 degrees denote axial plasma
acceleration and deceleration phases, respectively, as was mentioned. Here, I RMF can be controlled
by using two sets of rf power supplies (Thamway, T145-6326CK and T145-6326B), and φ and f RMF
are adjusted by a function generator, as shown in Fig. 3(a). Depending on f RMF and target plasma
parameters, RMF matching condition can be adjusted by using matching boxes, as shown in Fig 3(a).
We apply 30 ms RMF from t = 35 to 65 ms for the 75 ms rf pulsed discharge. This RMF operation
enables us to compare the plasma performance between the cases of w/and w/o RMF easily in one
shot, and it was confirmed that the rf discharge w/o RMF does not change compared to the one after
finishing RMF application (t = 65 - 75 ms), as shown in Fig. 3(b).

C. Plasma diagnostics
In our plasma diagnostic of local plasma flow velocity in addition to plasma density, Mach probe
method has been adopted to estimate the RMF acceleration effect. A straight type Mach probe is
at z = - 130 mm to measure vi and ne in the downstream of RMF antenna, as shown in Fig. 4(a).
115204-5 Furukawa et al. AIP Advances 7, 115204 (2017)

FIG. 3. Schematics of (a) RMF antenna and peripheral apparatuses and (b) example of vi (red line) and ne (blue line) with
30 ms RMF application (green hatched region) during 75 ms rf discharge.

FIG. 4. Schematics of (a) a straight type Mach probe and (b) a magnetic probe.

Concerning electrodes to capture ion saturation current, two tungsten rods with a diameter of 0.8 mm
are inserted into holes of an insulation Al2 O3 tube, and an exposed length of both electrodes to a
plasma is 4 mm, considering the sheath thickness.20 The insulation tube has a separating barrier to
divide a plasma flow into an upstream and a downstream sides of the electrodes. Thus, ion Mach
number can be calculated by a ratio of ion saturation current densities between the two electrodes,

FIG. 5. Measured ∆ne /ne (open red circle) and ∆vi /vi (open blue diamond) by changing I RMF at r = 60 mm and z = -130 mm
with (a) f RMF = 0.7 MHz and (b) f RMF = 1 MHz (Prf = 1 kW, fr = 40 sccm, and φ = 90 deg.).
115204-6 Furukawa et al. AIP Advances 7, 115204 (2017)

employing an unmagnetized model, whose model number κ is 1.26,19,20 and then a local vi can be
derived, using a typical T e of 5 eV in our experimental scheme. Generally, vi data by a Mach probe12
agree with the ones by Laser Induced Fluorescence (LIF) method within a factor of ∼ 1.5.22 A local
value of ne can be also calculated by an average of ion saturation currents from both electrodes, and
these experimental values are derived from time average of each discharge shot by shot, using eight
discharges.
In addition, a measurement of time-varying component of RMF, b̃⊥ , which is orthogonal to the
axial direction, has been conducted by using a magnetic probe, whose loop diameter and number
of the turn are 10 mm and 2 turns, respectively. The probe consists of a co-axial cable inserted into
insulation Al2 O3 tube and a probe tip is coated with ceramic bond, as shown in Fig. 4(b). The coil
surface, which detects the magnetic field, is vertical to the radial direction of the quartz tube, and can
be scanned to obtain the radial distributions of b̃⊥ at z = -250 mm, which is about the same position
of the axial center of RMF coils.

IV. MEASUREMENTS RESULTS OF RMF ACCELERATION

To examine the RMF acceleration method, following experiments were performed under the
experimental conditions, as shown in Table I, and experimental results presented as follows: Sec. A.
∆ne /ne and ∆vi /vi I RMF with f RMF = 0.7 MHz and 1 MHz, Sec. B. Radial distributions of ne and vi
by changing fr with f RMF = 0.7 MHz and 1 MHz, Sec. C. Comparisons of ∆ne /ne and ∆vi /vi between
f RMF = 0.7 MHz, 1 MHz, and 5 MHz, Sec. D. Dependences of ∆ne /ne and ∆vi /vi on φ with
f RMF = 0.7 MHz, and Sec. E. Radial distributions of radial component of RMF, b̃r .

A. ∆ne /ne and ∆v i /v i by changing I RMF with f RMF = 0.7 MHz and 1 MHz
Figures 3(a) and (b) show the plots of the measured ∆ne /ne and ∆vi /vi with f RMF = 0.7 MHz and
1 MHz, respectively, as a function of I RMF at r = 60 mm and z = -130 mm for the case of Prf = 1 kW,
fr = 40 sccm, and φ = 90 deg. The maximum value of I RMF was not same between f RMF = 0.7 MHz
and 1 MHz due to a different impedance matching of an rf circuit, and I RMF = 0 App means the case
of without applying RMF to plasma. Here, ∆vi /vi as well as ∆ne /ne increased almost quadratically
by increasing I RMF . When I RMF is ∼ 60 (46) App with f RMF = 0.7 (1) MHz, the maximum ∆ne /ne and
∆vi /vi were ∼ 200 (360) % and ∼ 27 (55) %, respectively. The increase of ne was much higher than
that of vi in the present experimental conditions in both cases of f RMF , and increasing rates of ne and
vi were higher in the case of f RMF = 1 MHz than that of f RMF = 0.7 MHz at the same I RMF . Therefore,
a further increase of I RMF with f RMF = 1 MHz is expected to have the better plasma performance.
These high increases of ne contribute to the thermal thrust: dynamic pressure ni mi vi 2 (the product
of ion flux ni vi and momentum mi vi ), and static pressure ne k B T e increased by increasing I RMF ,
which will be described in Sec. V.

B. Radial distributions of ne and v i by changing fr with f RMF = 0.7 MHz and 1 MHz
Figures 6(a)-1 and 2 show radial distributions of ne with f RMF = 0.7 MHz (I RMF = 50 App ) and
1 MHz (I RMF = 30 App ), respectively, and Figs. 6(b)-1 and 2 show those of vi , w/o and w/RMF,
respectively. Here, Prf = 1 kW, φ = 90 deg., and fr was changed: 20, 40, and 60 sccm.

TABLE I. Experimental conditions of RMF application.

f RMF 0.7 (1) MHz

Phase difference φ of I RMF -180 ∼ 180 deg.

RMF current I RMF ∼ 10 App ∼ 15 App ∼ 36 App ∼ 60 App

Total input ∼ 91 W ∼ 300 W ∼ 1.1 kW ∼ 2.6 kW
RMF power (∼75 W) (∼165 W) (∼1.5 kW) (-)

Bω (center) ∼ 2.0 G ∼ 3.2 G ∼ 6.1 G ∼ 12 G

Gass flow rate fr 20, 40, 60 sccm

115204-7 Furukawa et al. AIP Advances 7, 115204 (2017)

FIG. 6. (a)-1 and 2 show radial distributions of ne with f RMF = 0.7 MHz and 1 MHz, respectively, and (b)-1 and 2 show those
of vi , w/(solid lines) and w/o (dotted lines) applying RMF, respectively [z = -130 mm for the case of Prf = 1 kW, fr = 20 sccm
(square), 40 sccm (triangle), and 60 sccm (circle), and φ = 90 deg.].

For both cases of f RMF = 0.7 MHz and 1 MHz, ne increased with applying RMF, and the highest
values are at r = 0 mm (center of the quartz tube), and the decreases of ne toward the outer radial
direction are found in both cases of w/and w/o RMF regardless of fr. The maximum ∆ne /ne is
∼ 220 %, obtained at r = 15 mm for the case of fr = 60 sccm with f RMF = 0.7 MHz.
With respect to vi , it also increased with applying RMF in the entire radial positions, and
the maximum value of ∆vi /vi was ∼ 37 %, at r = 0 mm for the case of fr = 40 sccm with
f RMF = 0.7 MHz. Compared to ne , vi did not decrease appreciably toward the outer radial direc-
tion. This maximum increment of vi corresponds to the high ∆ne /ne around plasma center (pressure
gradient along the z axis is also high near this region). As mentioned in Sec. A, ∆ne /ne and ∆vi /vi
with f RMF = 1 MHz were higher than those with f RMF = 0.7 MHz with I RMF = 30 App in all radial
positions.
By increasing fr, ∆ne /ne increased with both f RMF , and the maximum ∆vi /vi was obtained at
r = 0 mm for the case of fr = 40 sccm with both f RMF . However, the maximum thermal thrust was
obtained for the case of fr = 60 sccm with both f RMF , since ∆ne /ne is higher than ∆vi /vi by ∼ one-order
of magnitude even though the maximum ∆vi /vi were obtained for the case of fr = 40 sccm.
On the other hand, in terms of RMF acceleration, the degradation of an ionization rate or the colli-
sions between an electron-ion and an electron-neutral particles by increasing fr should be considered,
since these factors affect the RMF penetration and electromagnetic acceleration. In our gas pressure
conditions, collision frequency between an electron-neutral particles, ν en , which is much larger than
Coulomb collision frequency, i.e., collision frequency between electron and ion, in our experimen-
tal condition, should be calculated in order to estimate the collision effects. In this calculation, the
following relations can be utilized;

νen = 1.4 × 106 Te P0 . (7)

Here, P0 is neutral pressure [mTorr], and T e is taken in unit of eV. Equation (7) comes from approx-
imation of Brown’s experimental data.23 For typical values, ν en is 5.5×106 s-1 , assuming that neutral
density does not change in the pressure of the plasma, and the electron mean free path is ∼ 17 cm
for fr = 40 sccm (P0 = 0.73 mTorr at r and z of 0 and -130 mm, respectively). Needless to say, this
115204-8 Furukawa et al. AIP Advances 7, 115204 (2017)

collision frequency is related to resistivity η, which affects the thruster performance in this RMF
method.
As discussed above, for an optimal thrust performance, we should consider these target plasma
conditions, also including Prf as well as RMF operation conditions. As for a thrust estimation from
these radial distribution results, we will discuss this in Sec. V.
C. Comparisons of ∆ne /ne and ∆v i /v i between f RMF = 0.7 MHz, 1 MHz, and 5 MHz
Figures 7(a) and (b) show radial distributions of ∆ne /ne and ∆vi /vi , respectively, for the case
of Prf = 1 kW and fr = 40 sccm with f RMF = 0.7 MHz, 1 MHz, and 5 MHz. With the decrease of
f RMF , ∆ne /ne and ∆vi /vi increased in the entire radial positions, although I RMF = 50 App is higher
than other two cases of 30 App . In particular, ∆vi /vi with f RMF = 0.7 MHz and 1 MHz correspond
to high ∆ne /ne at r = 0 mm, comparing those with f RMF = 5 MHz. Decreasing f RMF , increases of
∆ne /ne and ∆vi /vi were found near the plasma center, and a change of ∆ne /ne was higher than that
of ∆vi /vi . Nearly constant ∆vi /vi for the case of f RMF = 5 MHz suggests a partial penetration of
RMF, which will be discussed later. These decreases of f RMF are better (f RMF = 1 MHz may be
best if I RMF is same among three frequencies) compared to the case of f RMF = 5 MHz with respect
to plasma thrust, as was mentioned in Sec. B. Note that, as one of the important factors of these
differences experimentally, the antenna-plasma coupling is also important: The RMF net power,
which is absorbed by plasma, and calculated by plasma and antenna resistances, will be discussed in
Sec. V.
We need to investigate experimental conditions of acceleration mechanism in addition to
the electromagnetic acceleration derived from using the RMF method; e.g., the better effi-
ciency of the axial Lorentz force, produced by diamagnetic current induced by the radial den-
sity gradient and axial force by a pressure gradient, in addition to the RMF method proposed.
For future works, in order to clarify original electromagnetic acceleration, we plan to measure
two-dimensional profiles of ne and vi to derive the density gradient in broader region of our

FIG. 7. Measured radial distributions of (a) ∆ne /ne and (b) ∆vi /vi , comparing f RMF = 0.7 MHz (I RMF = 50 App , filled circle),
1 MHz (30 App , filled triangle), and 5 MHz (30 App , filed square) at z = -130 mm for the case of Prf = 1 kW, fr = 40 sccm,
and φ = 90 deg.
115204-9 Furukawa et al. AIP Advances 7, 115204 (2017)

experimental scheme. In addition, in order to distinguish the original RMF electromagnetic acceler-
ation from that derived from diamagnetic current, as well as a measurement of time-varying 2ωRMF
component of magnetic field of RMF (or ωRMF component), introducing jθ , as will be described
in Sec. V.
D. Dependences of ne and v i on φ with f RMF = 0.7 MHz
Figures 8(a) and (b) show the dependences od ne and vi on φ at r = 60 mm for the case of
Prf = 3 kW, fr = 60 sccm, and I RMF = 50 App with f RMF = 0.7 MHz. As was mentioned above,
φ = 90 deg. and -90 deg. indicate acceleration and deceleration phases, respectively. The highest
∆ne /ne was ∼ 33 % for φ = 90 deg., and the lower was ∼ 1.7 % for φ = -90 deg.. Here, the nearly
sinusoidal curves of ne and vi w/RMF was obtained as a function of φ, as is expected from the RMF
scheme10,13 (see Sec. II), especially, concerning vi . The RMF coils’ current might also contribute
to plasma generation for this experimental condition as well as plasma acceleration, since ne was
increased with applying RMF in any phase except for φ ∼ -90 deg., and it is useful from a viewpoint
of the thrust increment: Practically, in the case of φ = 90 deg., where ne and vi were the highest values.
Here, clear wave excitation condition, such as a helicon wave to contribute to a plasma production, was
not found in a previous experiment,12 but we need to examine the dependence of plasma acceleration
on φ in more detail, trying other experimental conditions to be checked.
E. Measurements of the orthogonal component of RMF
Measured distributions of b̃⊥ with f RMF = 0.7 MHz (I RMF = 55 App ) and 1.5 MHz (20 App ),
considering the full and partial penetration conditions od RMF, are shown in Fig.9(a) and (b), respec-
tively. Figure 7(a) shows that the field may penetrate almost fully into a plasma, since the magnitude
w/plasma is slightly higher (lower) than that w/o plasma near the plasma center (outer region). On
the other hand, the tendency of partially penetration of RMF can be seen in Fig. 9(b) due to the
decrease (increase) of this field in the center (outer region), compared to the case w/o RMF. There-
fore, this experimental results indicate that the RMF coils can work as a RMF generation antenna
to induce jθ due to the full penetration condition. Note that we have lowered I RMF to 20 App with
f RMF = 1.5 MHz, in order to examine the b̃⊥ distributions in partial RMF penetration condition. Here,

FIG. 8. Measured (a) ne and (b) vi by changing φ at r = 60 mm w/(filled circle) and w/o RMF (dotted line) for the case of
Prf = 3 kW, fr = 60 sccm, and I RMF = 50 App with f RMF = 0.7 MHz.
115204-10 Furukawa et al. AIP Advances 7, 115204 (2017)

FIG. 9. Measured radial distributions of b̃⊥ with (a) f RMF = 0.7 MHz (I RMF = 55 App ) and (b) 1.5 MHz (20 App ), which
represent full and partial penetration of RMF, respectively, for the case of Prf = 1 kW and fr = 40 sccm. Open circle and square
symbols represent w/and w/o plasma, respectively.

the total magnetic flux may be conserved, since the field magnitude was increased (decreased) in the
outer (inner) region of r > (<) 30 mm.

V. DISCUSSION
Dependencies of plasma parameters on RMF operation conditions were presented before. More-
over, both decreasing of f RMF and increasing I RMF under the present experimental conditions are also
expected theoretically to enhance the RMF penetration condition, which is discussed in this section
by using the Milroy’s expression:11
(
1.12f g (λ ≤ 6.5)
γc /λ = (8)
1.12 1.0 + 0.12(λ − 6.5)0.4 (λ > 6.5).
Here, λ is a plasma radius normalized by a typical skin depth and γ c is a critic value to determine
RMF penetration boundary, i.e., Hall parameter: an electron cyclotron angular frequency, using the
rotating magnetic field, normalized by a sum of an electron-ion and an electron-neutral collision
frequencies.
The value of λ and γ can be calculated from the experimental results/conditions, and a full
(partial) RMF penetration condition can be satisfied when γ/λ is larger (lower) than γ c /λ, and these
values with four f RMF in the graph of Milroy’s expression are shown in Fig. 8. Generally, central
(edge) region shows the lower (high) value of γ/λ, leading to a worse (better) penetration condition.
With f RMF = 5 MHz, this field penetration condition cannot be satisfied, since λ and γ/λ are 7.5 and
0.5, respectively, at r = 0 mm in our experimental condition, as shown in Fig. 8, considering lower
γ/λ value than γ c /λ one. On the other hand, RMF could fully penetrate with only f RMF = 0.7 MHz,
since λ and γ/λ are 3.8 (8.2) and 1.6 (0.6), respectively, with f RMF = 0.7 (1) MHz at r = 0 mm. On the
other hand, λ and γ/λ are 3.0 (5.4) and 4.6 (2.2), respectively, at r = 60 mm with f RMF = 0.7 (1) MHz,
and γ/λ are increased with increasing r, as shown in Fig. 10. Here, in spite of a rough estimation,
the full (partial) penetration condition by measuring radial distributions of RMF b̃r with f RMF = 0.7
(1.5) MHz, as shown in Fig. 10, is consistent with the data in Fig. 9.
115204-11 Furukawa et al. AIP Advances 7, 115204 (2017)

FIG. 10. Calculated γ/λ and λ with f RMF = 0.7 MHz (I RMF = 50 App , circle), 1 MHz (46 App , triangle), 1.5 MHz (20 App ,
diamond), and 5 MHz (30 App , square), in the graph of Milroy’s expression. Filled symbol represents r = 60 mm and the
colors become thin gradually with approaching r = 0 mm.

In order to clarify this penetration more experimentally, a measurement of RMF oscillating

component by using a magnetic probe is required. In addition, from a basic point of clarifying
acceleration phenomena, jθ component of the axial Lorentz force f z = jθ ×Br can be estimated by
a Rogowski coil or a magnetic probe using ∇× B = µ0 j. Here, measuring 2ωRMF component of jθ
induced by the non-linear effects (mentioned in Sec. II), which is the same amplitude of dc component
we are expecting theoretically. This can distinguish the dc component of jθ by a radial pressure gradient
from the RMF induced jθ , as was mentioned above.
In terms of plasma acceleration, a thermal thrust F, which is represented as a sum of surface
integrals of a static and a dynamic pressures, can be calculated from radial distributions of ne and vi ,
which are shown in Figs. 6(a) and (b):
 R 
F= ne mi vi 2 + ne kB Te 2πr dr. (9)
0

Here, R = 85 mm is an inner radius of the quartz tube at z = -130 mm, T e is a typical value of 5 eV,
and experimental ne values are taken. By applying RMF, the plasma thrust F are calculated to be
increased form ∼ 1.0 mN w/o RMF to ∼ 2.4 mN, ∼ 5.1 mN, and ∼ 1.4 mN with f RMF = 0.7 MHz
(I RMF = 50 App ), 1 MHz (46 App ), and 5 MHz (30 App ), respectively. The thrust to net power ratio
are changed from ∼ 1.7 mN/kW w/o RMF to ∼ 3.2 mN/kW, ∼ 6.7 mN/kW, and ∼ 1.9 mN/kW,
with f RMF = 0.7 MHz, 1 MHz, and 5 MHz, respectively. Here, net RF power is 580 W for
Prf = 1 kW, and total net RMF power were ∼ 300 W, ∼ 170 W, and ∼ 200 W with f RMF = 0.7 MHz,
1 MHz, and 5 MHz, respectively. Here, we have introduced these ratios by using the net rf and RMF
powers, which is important in a physical point of view. The maximum thermal thrust is obtained
with f RMF = 1 MHz, but the maximum total net power was the case with f RMF = 0.7 MHz, and
the maximum thrust to net power ratio is in the case with f RMF = 1 MHz. Based on these thrust
performance, we must examine the operation parameters to achieve the better performance with
f RMF = 1 MHz, also considering RMF penetration condition, for a future study. The present experimen-
tal conditions were not actually optimum ones: Lower electron density, comparing to previous helicon
thruster experiments w/o additional acclerations,19 in order to demonstrate the RMF acceleration
effect on plasma parameters, considering RMF penetration condition. For the purpose of increasing
the thrust, in particular, with f RMF = 1 MHz, the optimal experimental conditions, e.g., I RMF and a
number of turns of RMF coils, i.e., total ampere-turns, which increase BRMF , considering optimum
f RMF and the inductance of RMF coils, and also a target plasma, e.g., Prf and fr, should be surveyed
continuously.
Theoretically, the ideal maximum axial thrust derived from the RMF electromagnetic acceleration
in a full penetration condition with f RMF = 0.7 MHz can be calculated roughly by Eqs. (5) and (6),
and the value is ∼ 40 mN with an axial length of 168 mm, which is the same as the present RMF
coils’ length, and radial integral region is 0 ∼ 70 mm, considering the tapered shape at z = -230 mm.
115204-12 Furukawa et al. AIP Advances 7, 115204 (2017)

This is very high from a view point of a practical electric propulsion, but the particle confinement in
the external divergent, open magnetic field is poorer than that in a closed magnetic field shape like
FRC, due to a short particle axial transit time. In addition, by applying electromagnets, which have
not been used in present experiment, magnetic field lines do not touch the inner surface of a discharge
tube,15 and a wall loss is expected to be reduced.

VI. CONCLUSION
As an electrodeless plasma acceleration scheme, we have developed the RMF acceleration
system and have carried out the experiment successfully. Here, we have utilized two pairs of 5
turns, opposed facing coils as RMF acceleration antennas with LMD, and changed RMF operation
conditions including to f RMF , I RMF , and φ, in addition to target plasma conditions.
The effects of RMF current value I RMF on plasma parameters, e.g., ne and vi , changing the current
frequency f RMF , have been experimentally investigated, and higher I RMF leads to the higher ∆ne /ne
as well as ∆vi /vi : I RMF contributes to plasma generation in addition to an increment of the rotating
magnetic field, leading to a velocity increment [see in Figs. 5(a) and (b)]. Moreover, the maximum
∆vi /vi was ∼ 55 % for the case of Prf = 1 kW, fr = 40 sccm, φ = 90 deg., and I RMF = 46 App with
f RMF = 1 MHz, caused partly by the higher increase of ne [see Figs. 5(a) and (b)].
The penetration conditions could be also improved by a decrease of f RMF and an increase of I RMF .
In a full penetration condition, axial Lorentz force f z can be also saturated theoretically, which is in
proportion to f RMF . Moreover, time-varying component of RMF has been measured directly to verify
the penetration condition experimentally. Therefore, the tradeoff of f RMF between RMF penetration
conditions and acceleration should be considered; i.e., lowering f RMF is expected to improve the RMF
penetration into a plasma, and on the other hand, a further increase of f RMF is needed to increase the
azimuthal current jθ induced by RMF method when the RMF full penetration condition is satisfied.
From the experimental results, the thermal thrusts increased with decreasing f RMF , and also increasing
I RMF , but the absolute values have not been so high compared to the high-dense helicon experiments
with electromagnets21 due to the present experimental conditions (without these magnets), which
have not been optimized.
The dependences of ne and vi on RMF current phase φ between two-facing RMF coils, in
particular ne , have been found, as a sinusoidal curve, which can be expected theoretically, and RMF
coils’ current could also contribute to a plasma generation. Although the deceleration and acceleration
phases have been clearly demonstrated, the wave excitation mechanism, which was discussed in a
previous our experiment,12 have not been investigated this time.
In order to increase jθ , a further increase of Bω is expected, i.e., increases of I RMF and RMF
coils’ turns, with an optimum coil shape, and so on, regardless the tradeoff of f RMF between RMF
acceleration and its field penetration. In terms of I RMF , we plan to make a high-current power supply,
e.g., I RMF is ∼ 300 App 24 changing f RMF between 0.1 - 1 MHz. Measurements of jθ by a Rogowski
coil, or time-varying component of axial magnetic field b̃z by a magnetic probe to deduce jθ , are also
helpful to estimate the axial Lorentz force and estimate the additional plasma acceleration by using
RMF method.
Plasma diagnostics in addition to our present probe method, e.g., as non-contact measurements,
Laser Induced Fluorescence (LIF) method,18,22 which can measure an ion velocity and temperature
without disturbing plasma, and other spectroscopies, are effective to crosscheck our previous mea-
surement. In addition, using a thrust measurement method, e.g., a target stand type, which has a
cylindrical shape, considering the collision between the target and particle19,25 is an effective method
to estimate this RMF acceleration effect.

ACKNOWLEDGMENTS
We appreciate the prior research of RMF acceleration scheme by Messrs. H. Ishii and S. Otsuka
in our group, and useful discussions made by HEAT project members. This study has been partially
supported by Grand-in-Aid for Scientific Research (S: 21226019 and B: 17H02995) from the Japan
Society for the Promotion of Science.
115204-13 Furukawa et al. AIP Advances 7, 115204 (2017)

1 D. M. Goebel, IEEE Trans. Plasma Sci. 36, 211 (2008).

2 V. V. Zhurin, H. R. Kaufman, and R. S. Robinson, Plasma Source Sci. Technol. 8, R1 (1999).
3 M. Martinez-Sanchez and J. E. Pollard, J. Propul. Power 14, 5 (1998).
4 F. Cannat, T. Lafleur, J. Jarrige, P. Chabert, P. -Q. Elias, and D. Packan, Phys. Plasmas 22, 053503 (2015).
5 F. R. Chag-Dı́az, Sci. Am. 283, 90 (2000).
6 C. Charles, J. Phys. D: Appl. Phys. 42, 163001 (2009).
7 S. Shinohara, H. Nishida, T. Tanikawa, T. Hada, I. Funaki, and K. P. Shamrai, IEEE Trans. Plasma Sci. 42, 1245 (2015).
8 S. Shinohara, J. Plasma Fusion Res. 91, 412 (2015) (project review paper) [In Japanese].
9 R. W. Boswell, Phys. Lett. 33A, 457 (1970).
10 I. R. Jones, Phys. Plasmas 6, 1950 (1999).
11 R. D. Milroy, Phys. Plasmas 6, 2771 (1999).
12 S. Otsuka, K. Takizawa, Y. Tanida, D. Kuwahara, and S. Shinohara, Plasma Fusion Res. 10, 3401026 (2015).
13 T. Furukawa, K. Takizawa, D. Kuwahara, and S. Shinohara, Phys. Plasmas 24, 043505 (2017).
14 S. Shinohara, S. Takechi, and Y. Kawai, Jpn. J. Appl. Phys. 35, 4503 (1996).
15 S. Otsuka, T. Nakagawa, H. Ishii, N. Teshigahara, H. Fujitsuka, S. Waseda, T. Ishii, D. Kuwahara, and S. Shinohara, Plasma

Fusion Res. 9, 3406047 (2014).

16 F. F. Chen, Plasma Phys. Control. Fusion 33, 339 (1991).
17 A. Fruchtman, K. Takahashi, C. Charles, and R. W. Boswell, Phys. Plasmas 19, 033507 (2012).
18 Y. Tanida, D. Kuwahara, and S. Shinohara, Trans. JSASS Aerospace Tech. Japan 14, Pb 7 (2016).
19 D. Kuwahara, S. Shinohara, and K. Yano, J. Propul. Power 33, 420 (2017).
20 M. Hudis and L. M. Lidsky, J. Appl. Phys. 41, 5011 (1970).
21 K. S. Chung, I. H. Hutchinson, B. Labombard, and R. W. Conn, Phys. Fluids B1, 2229 (1989).
22 D. Kuwahara, Y. Tanida, M. Watanabe, N. Teshigahara, Y. Yamagata, and S. Shinohara, Plasma Fusion Res. 10, 3401057

(2015).
23 S. C. Brown, Basic Data of Plasma Physics, 2nd ed. (1967).
24 M. Inomoto, K. Kitano, and S. Okada, Plasma Fusion Res. 3, 004 (2008).
25 D. Kuwahara, Y. Koyama, S. Otsuka, T. Ishii, H. Ishii, H. Fujitsuka, S. Waseda, and S. Shinohara, Plasma Fusion Res. 9,

3406025 (2014).