Nefedtsev (2020)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

DOI 10.

1007/s11182-020-02062-y
Russian Physics Journal, Vol. 63, No. 3, July, 2020 (Russian Original No. 3, March, 2020)

PLASMA PHYSICS
OPTIMIZATION OF ELECTRIC THRUST BASED ON PULSE ARC
PLASMA USING PULSED MAGNETIC FIELD
E. V. Nefedtsev and L. A. Zjulkova UDC 537.84, 533.95

Based on the results of a numerical simulation, a possibility in principle is shown to achieve an effective
magneto-induction correction of high-gradient plasma expansion from a localized arc source and a respective
severalfold increase in the thrust pulse with a steepness of the induction rise in the accelerator on the order of
104 T/s. In these conditions, a pulse-periodic mode of the device operation is required with a delay of the
current pulse front in the magnetic field source relative to the ignition voltage pulse in the plasma source,
which is determined by the expansion velocity of the plasma cloud within the device.

Keywords: hydrodynamic plasma model, pulsed magnetic field, high-current vacuum-arc discharge, plasma-jet
thrust of a spacecraft.

INTRODUCTION

Within recent time a large number of engine designs have been proposed for spacecraft applications, which are
based on electric thrust. Each of these projects claims to occupy a distinct functional niche (main engines, vernier and
stabilization thrusters), depending on the thrust parameters (traction thrust, pulse, energy and thrust efficiency, angular
diagram of the rocket propellant expansion, stability, functioning time, ecological requirements, etc.). Frequently the
recoil pulse carrier in electric thrusters is plasma produced by a variety of methods, including reactive heating,
discharges across polymer surfaces, gas breakdown in an inductive field, induced external magnetic field, laser ablation,
vacuum arc ignition, etc. Most of the plasma sources are equipped with electromagnetic acceleration systems.
The potential of a pulsed arc discharge [1] for thrust improvement in small and ultra-small spacecraft has not
been studied in sufficient detail, while it possesses a number of advantages. Among its important merits is a possible
use of a condensed working medium (with a large atomic mass), high supersonic plasma jet velocities, and a possibility
of controlling the thrust within the range of ~1–10 µN to a high accuracy using several independent parameters:
frequency, duration and power of initiation pulses.
A commonly noted disadvantage of the vacuum-arc plasma sources is electrode erosion reducing the service
life of the thruster [2] and requiring a reliable system of a working medium supply into the discharge chamber. More
recently it was demonstrated [3, 4] that this problem could be solved by using a liquid-phase metal propellant
automatically fed into the discharge zone via a steel capillary. The metal fluid meniscus, which is reconstructed under
the surface tension forces within a short period of time, generates the reproducible conditions both for laser-ablation [3]
and arc-based plasma sources [4] in a wide range of pulse repetition frequencies and within long operation times.
A principal disadvantage of the high-current vacuum-arc plasma used as a recoil pulse carrier is its wide spray
pattern, which results in a large pulse loss in the radial direction and, hence, a low thrust efficiency. This disadvantage

Institute of High Current Electronics of the Siberian Branch of the Russian Academy of Sciences, Tomsk,
Russia, e-mail: nev@lve.hcei.tsc.ru; lorik@lve.hcei.tsc.ru. Translated from Izvestiya Vysshikh Uchebnykh Zavedenii,
Fizika, No. 3, pp. 137–144, March, 2020. Original article submitted December 5, 2019.

498 1064-8887/20/6303-0498 2020 Springer Science+Business Media, LLC


can be ruled out by plasma confinement using an external magnetic field. Considering a pulsed nature of the vacuum-
arc plasma generation, an attractive method is the use of an interaction of the external pulsed magnetic field with the
plasma eddy currents induced by this field. It is known that this interaction is utilized to accelerate plasma bunches in
pulsed inductive thrusters [5]. Their design involves a synchronization of the valve of the pulsed working gas supply
with the magnetic field pulse. The most optimal option is in this case the magnetic field of a disc coil whose winding is
perpendicular to the system’s axis [2]. The high-current vacuum-arc plasma itself is generated in a pulsed mode and
does not require any mechanical supply devices. Moreover, it is characterized by a high supersonic expansion velocity
and high concentration gradients. In this connection, the question is a possibility of correcting the angular plasma
expansion pattern rather than inductive acceleration, in which case in addition to the attempts to optimize the spatial
magnetic-field configuration we have to solve the problem of the magnetic pulse delay with respect to the arc initiation
pulse, acquiring a large importance.
In this work, relying on numerical calculations, we provide the estimates of the configurations and parameters
of the magnetic field pulse ensuring a considerable improvement of the thrust of a vacuum-arc plasma source. These
estimates are also applicable to a plasma source based on pulsed superheating of a material by a laser beam.

CALCULATION PROCEDURES AND MATHEMATICAL MODEL

Numerical simulation of plasma expanding into vacuum can be performed using a basic mathematical model
[6], representing a system of equations composed for the first moments of the distribution function for every plasma
component, including the electronic component. This approach is reasonable in the case where the purpose is to study
the structure of charge separation of the near-boundary region of the expanding plasma and the influence of a strong
external electric field on the shape of the plasma boundary under the conditions where the electron flow separates from
this plasma (which is typical for the commutation stage of a vacuum breakdown) [7, 8]. In the majority of cases it is
convenient to use simplified versions of the basic model, specifically, a magneto-hydrodynamic (MHD) model;
a transfer to the latter was demonstrated in [6]. In this model, plasma is treated as a single quasi-neutral fluid having
a certain effective volume density , directional velocity V (determined by the density and average velocity of the ion
ensemble flow) and effective temperature T (determined by the electron temperature). This approach considerably
reduces the time of a numerical calculation of plasma evolution, since the selected time step corresponds to the ion
subsystem dynamics. In order to estimate the influence of the magnetic field, described by magnetic flux density B, let
us disregard the inter-electrode component of the current density concentrated practically entirely in the small region of
the spark, and take that the current density J in the plasma at a distance from the source and the electric field strength E
are primarily determined by the induced eddy components of the current and the field strength.
The use of variational finite-element solutions of equations [9] is also viable for reducing the time of
computational modeling of the expanding plasma dynamics. Recently a Comsol Multiphysics (CM) computational
environment has become very popular; it allows applying a time and effort saving method of computation, based on
representing the equations in an integral simplified finite-element form (weak form). The CM has a package of special
application interfaces simplifying the preparation for computations. However, the available Plasma module relies on an
oversimplified diffusion-drift model, which is more suitable for the description of technological plasma properties but is
not useful for calculations of microexplosive plasma, characterized by high density gradients and absence of the dense
background of neutral particles. Therefore, in this study we used templates of the Computation fluid dynamics (CFD)
module designed for solving a wide range of gas-hydrodynamic problems. By combining the following blocks: Non-
Isothermal Flow from the CFD-module and Magnetic Fields, we obtain a system of equations corresponding to a quasi-
stationary MHD-model of ordinary plasma [6]


   V   0 , (1)
t

499
 
V  T 2 
   V   V    PI   V   V       V  I   F , (2)
t  3 

 T 
C   V  T     k T   Q , (3)
 t 

MP
 , (4)
ZkT

  B  0 J , (5)

J  E  v  B  J e , (6)

B   A , (7)

A
E . (8)
t

Equations (1)–(3) are typical equations describing the variation in the state of compressed gas-dynamic media
in time t. The continuity (1), Navier–Stokes (2) and heat balance (3) equations describe the dynamics and heat content
of every unit of the medium volume. In the case under study, an external Lorentz force is added into the right-hand part
of equation (2) to the pressure and viscosity forces, acting on the unit volume of plasma

F  JB. (9)

The equation of state (4) describes plasma as an ideal neutral gas, in which pressure P is proportional to
temperature: P = nekT, where k is the Boltzmann constant, ne = Zni = Z/M is the electron concentration, ni is the ion
concentration, Z and M are the average values of the plasma ion charge number and mass, respectively.
The heat-balance equation (3) includes the Joule heat release

Q  J2  , (10)

where  is the plasma conductivity mainly determined by the scattering of electrons on ions in the magnetic field

1
e2 ne   '1  x 2   '0 
  1   . (11)
m ei  x 4
  1  x 2
  0 

Since the azimuthal current component in the model was thought to dominate, we took into account only the
conductivity component orthogonal to the magnetic field lines. The term in brackets in expression (11) is a correction to
the Spitzer conductivity (first term), taking into account the magnetic field action, and quantity x= B / ei = (eB/m)/ei
represents a ratio of the Larmor frequency of an electron to the average frequency of its scattering on the Coulomb
centers

4  2 
1/ 2 4
e Z 2 m1/ 2 ni
 ei   ei . (12)
3  40 
2
 kT 3 / 2

500
Z
Z0

Z1

0 R0 r

Fig. 1. Computational cell with a point plasma source and electromagnetic coil
located on the side surface of the accelerator. Arrows schematically show the
directions of plasma expansion, dotted lines indicate the magnetic field forces.

Coefficients '1 = 5.523, '0 = 0.5956 and 1 = 10.8, 0 = 1.0465 correspond to the case where Z = 2 [6]. The
Coulomb logarithm ei can be expressed as follows:

   
 ei  ln 2.415 1041  1.5ln  kTe  J   0.5ln ni  m 3  . (13)

The system of equations describing the magnetic field includes the Ampere law (5), the generalized Ohm law
(6) and the Faraday law (7), (8). These equations can be re-formulated with respect to the vector potential components
A. Quantity Je in the Magnetic Field interface describes the external currents. In the combined plasma model this
quantity could be conveniently related to the component of the current due to pressure gradient, and we can take that

ne kT kT
Je    ln P . (14)
ene e

INITIAL AND BOUNDARY CONDITIONS. OTHER PARAMETERS OF THE PROBLEM

The problem was solved in an axial-isotropic approximation considering the appearance of the azimuthal
current J and eddy electric field E components. The computational cell represented a cylindrical region with a radius
of R0 = 30 mm and a height of Z0 = 60 mm (Fig. 1). The boundaries of the region, except for the part of the cylinder
base (z = 0, r < 10 мм), adjacent to the plasma source were considered to be remote. Based on these boundaries, the
condition of free plasma fluid flow was formulated and the plasma temperature was fixed at a low level of T0 = 600 K
(which is by an order of magnitude lower than the plasma source temperature). The plasma source with ignition across

501
the dielectric surface was simulated by a thin ring 2 mm in diameter (Fig. 1) at a height of 0.5 mm above the surface of
the base.
The formulation of initial conditions was a separate challenge due to a specific character of the relations
describing plasma properties. The initial condition of an absolute vacuum is inacceptable, therefore in the initial point of
time t = 0 the computational cell was thought to be filled with cold plasma at an even pressure of 1 Pa at a temperature
of 600 K. In the first stage, (0 < t < 50 µs) the initial fields of plasma concentration, velocity and temperature were
formed, and the plasma source temperature was set at a level of 8000 K, which corresponds to the critical temperature
of the explosive-emission centers. From the point of time t1 ~ 50 µs the plasma source was initiated at a pressure level
of tens of kilopascals, and then the second, major stage of computations began. Considering the finite time of plasma
cloud expansion, a magnetic field pulse was delivered with a delay t of a few tens of microseconds with respect to the
plasma source pulse front. The magnetic field geometrical configuration, the delay and duration of the magnetic field
pulse front were subject to optimization during the computations (see below).
The thrust was characterized by the value of force FT, with which the plasma flow was supposed to act along
the axis of the source onto the surface surrounding the computational cell. Considering the data reported in [10],

FT   VVz dS , (15)
S1

where S1 is the upper part of the boundary surface (r  R0, z = Z0) ⋃ (r = R0, Z1  z  Z0) and Vz is the flow velocity
projection on the z-axis.
The plane with z = Z1 = 30 mm determined a conventional geometrical boundary of the accelerating device
within which the winding of the electromagnet was localized, therefore, the region with (z < Z1) was not included into
the thrust calculation. Determination of the thrust as a function of time was performed in the course of the entire
calculations; as a result we managed to calculate the recoil pulse and performed an estimative optimization analysis of
various parameters of the problem, including the pulse parameters and spatial distribution of the magnetic field.

SIMULATION RESULTS

The isolines in Fig. 2 show the structure of the fields of the absolute stream velocity (arrows indicate the
velocity direction field) and pressure in the course of plasma expansion at the point of time 60 µs.
An optimal magnetic pulse delay t has to be commensurable with the time of propagation of the plasma
boundary to the region of the magnetic winding localization (in fact, the time during which steady-state concentration
and velocity profiles were achieved in the computational region). A larger delay time reduces the gain in the thrust,
since the onset of the cycle is associated with an inefficient, close to isotropic, plasma yield without any inductive
correction of its angular expansion pattern. Note that a generation of longer magnetic field pulse is not reasonable, since
the inductive acceleration effect is determined by the steepness of the magnetic flux density increase dB/dt, and when
the magnetic induction reaches its peak value, the effect disappears and the structure of the plasma concentration and
velocity fields becomes the same as those without any magnetic field. It is critical to switch off the plasma source
before the magnetic field disappears, as it is shown, e.g., in Fig. 3. Indeed, removal of the magnetic field under the
conditions of a continuously operating plasma source results in the formation of reverse inductive currents with respect
to those appearing during the pulse front. As a result, part of the thrust pulse, “gained” in the field increase time, would
be compensated, and the effect on the whole would be close to zero. These considerations are validated by the
calculations of thrust (16) as a function of time Fт (t) in the course of numerical simulations.
To give an example, let us discuss a magnetic field source in the form of a planar circular current loop localized
on the bottom part of the forming cylinder (r = R0, z < Z1). The circular current imitates a winding whose turns are
distributed in accordance with the following expression:

I = I0exp[–(z/z0)2], (16)

502
z, m Р, Pa z, m V, km/s
0.06 0.06
5
4
0.055 0.055
4.5
0.05 3.5 0.05

4
0.045 0.045
3

0.04 0.04 3.5


10 2.5
0.035 0.035 3
2
0.03 0.03
2.5
10
0.025 1.5 0.025
2 2
0.02 0.02
1
1.5
0.015 0.015 3.5
10
0.5 1
0.01 0.01
2
10 0 3 0.5
0.005 0.005
3
10
0 0 0
–0.5
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
r, m r, m

a b

Fig. 2. Spatial characteristics of plasma in the course of its expansion inside the device:
a – pressure, Pa, b – velocity field, km/s; arrows show the field direction.

1
20 kPa
0.8
106 A/m
arb. units

0.6

0.4
t
0.2

0 100 200 300


t, s

Fig. 3. Normalized pulses of the plasma and magnetic field sources. The thrust
parameter was optimized with respect to the indicated peak values.

where I = iN is the linear current density, A/m, i is the strength of current in the winding, A, N is the winding density,
m–1, I0 is the maximum current density value in a point with (z = 0, r = R0), z0 is the optimization parameter determining
the characteristic scale of inhomogeneity of the winding density of the source along the z-axis (see Fig. 1).
Figure 4 presents the Fт (t) curves at different current amplitudes in the coil with the parameter z0 = 5 mm
(actually localized on the “bottom” of the device). The first sharp peak in the curves of dependence is caused by the
crossing of the computational cell boundary by the plasma condensation front, which is an ‘aftermath’ of the

503
1
Fт, N
0.8

4.8
0.6

0.4 4.4

0.2
2.6

0
0 50 100 150 200 250
t, s

Fig. 4. Thrust versus time at different values of the total current (indicated
in kA) in a short coil z0 = 0.5 mm.

microexplosion evolution (see also Fig. 2). Following this, a stationary level of the thrust was set at 80 mN. For the
plasma source operation time of 200 µs, this value corresponds to a pulse of 1610–6 Ns.
For the current amplitude in the magnetic field source of about 103 ampere-turns, the magnetic field effect
becomes appreciable. In the case of a stationary thrust, a bell-shape pulse is formed with a maximum shifted by 20 µs
towards longer times relative to the dI/dt maximum. At the current amplitude of i = 4.4103 ampere-turns (in which case
I0 = 1106 A/m), the effect becomes quite pronounced and transfers into an instability and an abnormal termination of
the computation. Near the threshold with i = 4.8103 ampere-turns (I0 = 1.1106 A/m), the shape of the thrust pulse
becomes complicated: the major maximum is added with one more maximum shifted in time by 40 µs relative to dI/dt.
The reason for such an ambiguous magnetic field influence on the thrust parameters can be understood by
analyzing the plasma characteristics at the point of time where the dI/dt maximum is attained (150 µs). Figure 5
presents the velocity and concentration fields (calculated on the basis of an average atomic mass of an In–Ga eutectic of
M = 9.8710–26 kg = 59 AM and an average charge number Z = 2) at the point of time 160 µs for the maximum current
density of I0 = 1.1106 A/m. As it follows from Fig. 5c, the plasma flow is sharply decelerated along the radial
component and, as a result, the periphery part of the plasma becomes denser (Fig. 5d), preserving the velocity
component along the system’s axis. Thus, the first maximum in the thrust characteristics is to a larger extent due to the
correction of the angular plasma spray pattern. When a critically high value of dB/dt is approached, in addition to the
deceleration and the related correction of the plasma flow direction, an focusing effect appears, which results in the
formation of a protuberance-like dense plasma sheet bordering a local velocity maximum. A certain current density
domain is formed, which propagates towards the upper boundary, reaching it within 20 µs, where the ‘pressure
detectors’ record one more mechanical pulse overshoot. Because of the local maximum in the velocity field, this
component of the thrust could be related to the inductive plasma acceleration along the z direction. It follows from
Fig. 4 that the efficiency of thrust enhancement due to the inductive acceleration along z could be commensurate with
the efficiency of its improvement by a deceleration along the r direction. The distribution of the induction eddy currents
in the plasma at the point of time 160 µs is shown in Fig. 6.
According to the calculations, when the coil of the magnetic field ‘becomes longer’, the inductive acceleration
of the plasma along z influences the Fт (t) curve increasingly weaker – the second maximum is smeared and overlapped
with the first maximum. As early as at the coil inhomogeneity parameter of z0 = 20 mm, the magnetic field effect could
be treated as a hybrid correction-focusing effect.
In addition to the magnetic field source, in this study we discussed the electromagnetic system configuration in
the form of a surface current circulating in the cylinder base within the (r >10 mm, z = 0) limits and a current distributed
over the volume of the device in the (r > 10 мм, z < Z1) limits. For all of the configurations discussed, the thrust pulse
maximum was achieved at a limiting contraction of the coil turns into an angular point of the computational cell (r = R0,
z = 0).

504
z, m V, km/s z, m ni, m–3
0.06 0.06 24
5
0.055 0.055
4.5
0.05 0.05 23
4
0.045 0.045
4 3·10
19
0.04 3.5 0.04 22

0.035 3 0.035

0.03 0.03
2.5 21
20
0.025 0.025 10
2
0.02 0.02
20
1.5 20
0.015 3.5 0.015 3·10
1 0.01
0.01
19
3 10
21
0.005 0.5 0.005

0
0 0
18
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
r, m r, m
a b

z, m V, km/s z, m ni, m–3


0.06 0.06 24
5 19
0.055
3·10
0.055
4.5
0.05 0.05 23
4
0.045 0.045
20
0.04 3.5 0.04
3·10
22

0.035 3 0.035

20
0.03 2.5
0.03 10
21
4
0.025 3.5 0.025
2
0.02 0.02 20
20
10 20
1.5 3·10
0.015 0.015

1
0.01 0.01
3 19
21
0.5 0.005
10
0.005

0 0 0
18
0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.005 0.01 0.015 0.02 0.025 0.03
r, m r, m
c d

Fig. 5. Fields of the plasma ion velocity (a, c) and concentration (b, d) at the point of
time 160 µs: without any magnetic field (a, b) and during the current rise through the
coil up to an amplitude of 4.8/N kA (c, d). Digits show the velocities in km/s and
concentrations in m–3.

505
z, m x104
0.06 3

0.055

0.05 2.5

0.045

0.04
2

0.035

0.03
1.5
5
0.025

0.02
10 1
0.015
15
0.01
20 0.5
0.005

0
0
0 0.005 0.01 0.015 0.02 0.025 0.03
r, m

Fig. 6. Distribution of the induction current in the plasma in the course of the current
rise through the short coil at the point of time 160 µs. Digits show the induction current
density in kA/m2.

CONCLUSIONS

The calculations have shown a possibility in principle to achieve a high-efficiency magnetic-inductive


correction of the high-gradient plasma expansion from a localized arc source and a respective severalfold enhancement
of the thrust pulse under the conditions of a pulsed magnetic field impact characterized by an induction amplitude on
the order of fractions of a Tesla and the steepness of the induction rise about 104 T/s. Note that this requires a pulse-
periodic mode of operation with a current pulse front delay in the magnetic field source with respect to the triggering
voltage pulse in the plasma source, which is determined by the plasma cloud expansion velocity within the device. The
principal mechanism of the magnetic-inductive correction of the plasma velocity field of an arc discharge is probably
the deceleration of its expansion in the radial direction. The effect of inductive acceleration is less stable and can be
therefore controlled by a strict regulation of the recoil pulse (at least within the limits of the magnetic field
configuration discussed in this study). The highest gain in the thrust is generated by the magnetic field source, whose
turns are most densely wound on the level of the arc source along the maximum possible radius determined by the
accelerating device design. On the other hand, an inclusion into consideration of the real anode geometry, the functional
tasks of the device, and the requirements to reduce the energy loss, heat scattering regime, etc., might result in
somewhat different estimates of the optimal magnetic field configurations in the electric thrust devices based on short
arcs.
This investigation has been supported by the Russian Science Foundation (Project No. 18-19-00270).

506
REFERENCES

1. M. Keidar, T. Zhuang, A. Shashurin, et al., Plasma Phys. Control. Fusion, 57, 014005 (2015).
2. A. K. Martin, J. Phys. D: Appl. Phys., 49, 025201 (2016).
3. S. A. Popov, A. N. Panchenko, A. V. Batrakov, et al., IEEE Trans. Plasma Sci., 39, No. 6, Part 1, 1412–1417
(2011).
4. S. A. Popov, Дубровская E. L. Dubrovskaya, and A. V. Batrakov, Russ. Phys. J., 61, No. 11, 2079–2084
(2019).
5. A. K. Hallock, A. K. Martin, K. A. Polzin, et al., IEEE Trans. Plasma Sci., 43, No. 1, 433–443 (2015).
6. S. I. Braginskii, Transport processes in a plasma, in Reviews of Plasma Physics, ed. by M.A. Leontovich,
Consultants Bureau, New York, 1, 205–311 (1965).
7. E. V. Nefedtsev and A. V. Batrakov, JETP, 121, No. 4, 706–716 (2015).
8. E. V. Nefedtsev and A. V. Batrakov, JETP, 126, No. 4, 541–549 (2018).
9. E. V. Nefedtsev and A. V. Batrakov, IEEE Conf. Pub.: Proc. 28th ISDEIV, Germany, 1, 85–88 (2018).
10. L. D. Landau and E. M. Lifshitz, Fluid mechanics, Course of Theoretical Physics, 2d ed., vol. 6, Pergamon
Press (1987).

507

You might also like