N. A. Gatsonistt Applied Physics Laboratory, The Johns Hopkins University Laurel, Maryland 20723

IEPC-93-142 1318 MODELLING OF ION THRUSTER PLUME CONTAMINATION R. I. Samanta Roy* and D. E. Hastingst Space Power and Propulsion Laboratory, Department of Aeronautics and Astronautics Massachusetts Institute of Technology Cambridge, Massachusetts 02139 N. A. Gatsonistt Applied Physics Laboratory, The Johns Hopkins University Laurel, Maryland 20723 Abstract jbi beam ion current density, A/m 2 k Boltzmann's constant (MKS) While the potential problems of spacecraft me,i electron, ion mass, kg contamination by the effluents of electric nbi beam ion density, m- 3 propulsion thrusters have been known for nie.n total ion, electron, neutral density, m- 3 some time, the limitations of ground (a further subscript "o" denotes experiments, and until recently, computational reference) power, have prevented accurate assessments Pe electron pressure, N/m q/m charge to mass ratio of 2 ions and predictions of spacecraft contamination. Re electron collisional drag term, N We are developing a hybrid plasma particle-in- r thruster beam radius, m cell (PIC) code to model the plume of an ion rr thruster radius, m thruster and the production of slow charge- Te electron temperature, *K exchange (CEX) ions in the plume and their Tw thermal wall temperature of neutrals, *K transport in the region exterior to the beam. vbi, e beam ion, electron velocity, m/s These CEX ions have the potential to be 0 Hall parameter transported into the backflow region and Ylp propellant utilization efficiency present a contamination hazard for the beam divergence angle, radians spacecraft. We present preliminary 2-D 9 electric potential, V Ve electron collision frequency, s - 1 axisymmetric results of the plume flow acex CEX cross-section, m 2 structure and clearly demonstrate the ion density enhancement around the spacecraft due to the slow CEX ions. I. Introduction Nomenclature For future space missions, advanced B fixed ambient magnetic field, T spacecraft propulsion systems such as various C neutral thermal velocity, m/s types of electric propulsion thrusters are being CEX charge exchange (ions) earnestly considered. However, many electric E electric field, V/m propulsion devices have low thrust e electron charge, C characteristics and hence must operate for Ib thruster beam ion current, A extended periods of time to achieve the necessary velocity changes. These long thrusting times can introduce a problem with Graduate Research Assistant, Student Member spacecraft contamination that may become t AA quite critical. While the backflow efflux of the t Professor Senior Member AIAA thruster may be small, protracted thruster SPostdoctoral Fellow, Member AIAA operation may aggravate any contamination IEPC 93-142 situation. 1

1319 IEPC-93-142 The potential problems of spacecraft experiments. One problem associated with ion contamination by the effluents of electric thrusters is that complete ionization can not be propulsion thrusters have been known for achieved with reasonable levels of power, and some time. However, due to the limitations of hence, neutral gas is emitted at thermal speeds. ground test vacuum facilities, accurate We are interested in these slow neutrals measurements of contamination levels are because they charge-exchange (CEX) with the difficult to obtain. Simple analytical models fast beam ions producing fast neutrals and can only provide rough estimates with large slow ions which can be influenced by local uncertainties, and until recently, accurate electric fields in the plume. The electric field computational models have been non-existent, structure in the plume, as seen in experiments 2 Furthermore, almost all ground tests have and in computational models 3, is radial, and involved thrusters of sizes and power far hence the slow ions are pushed out of the beam smaller than what are envisioned today for and move back towards the spacecraft. In this tomorrow's ambitious space missions. In paper, we present preliminary results of our addition, many previous tests employed modelling efforts of the plume for propellants like mercury, that are not being contamination assessment. Our work is considered for future space mission scenarios, complementing the modelling program for the Due to the quality of present data, to scale up International Topaz Test Program (ITTP), present tests to provide accurate predictions of formerly NEPSTP 4. Previous studies contaminants from large-scale multi-thruster regarding contamination have examined the configurations is rather problematic. A clear role of CEX ions in the sputtering of ion and fundamental understanding of both the thruster grids 5. However, these numerical plume backflow and the interactions between simulations were on the length scale of the size the exhaust products of a thruster and its host of a grid hole. The plume contamination spacecraft are necessary. Possible interactions problem requires orders of magnitude include sputtering and effluent deposition that increases in domain size to encompass the will affect such aspects of the spacecraft as plume and a large part, if not the whole, solar arrays, thermal control surfaces, optical spacecraft. sensors, communications, science instrument- In Section II of this paper, we formulate ation, general structural properties of materials, our approach to the problem and describe our and spacecraft charging as depicted in Figure 1. model. Numerical methods are discussed in Recentlyl, the state of the art in the present Section III, and selected results are presented understanding of spacecraft contamination due and discussed in Section IV. Lastly, to electric propulsion devices was reviewed, conclusions and future work are offered in and a general strategy employing modern numerical techniques was outlined. The Section V. general problem at hand is that of a thruster II. Physical Model emitting a plume of ionized and neutral gas. In addition, various components of the thruster The plumes of ion thrusters contain can sputter and erode, leading to the presence several major components: 1) fast (>10 km/s) of heavy metal species in the plume. The propellant beam ions, 2) neutral propellant, 3) transport of these species, which dynamically slow (initially thermal) propellant ions created interact, from the plume back onto the by charge-exchange processes, and 4) nonspacecraft comprises the backflow which is of propellant efflux, i.e. eroded grid and discharge primary concern. The essential question is: chamber material of which some is neutral, and how much of the plume will come back onto some is charged due to CEX reactions with the spacecraft? We have been studying the ion beam ions. We consider each of these species, thruster due to its maturity and the existence of along with neutralizing electrons below. a large database of ground and space Currently, our model is formulated in 2

IEPC-93-142 1320 cylindrical coordinates (r-z). Beam Ions: neutral propellant through the grids. CEX propellant ions The current density of the collimated beam Slow propellant ions are created inside the ions can be approximated by a parabolic beam due to charge-exchange reactions of the axisymmetric profile given by, following type between the fast beam ions and 2 I b1 _ 2 the slow thermal neutrals: J(r) = r I1 (1) Xet + Xelo-> Xelw + Xe which is subject to the normalization that at any downstream location in the beam, The result is a fast neutral that travels in a line of sight manner, and a slow ion that is Ib = jbi r dr do (2) easily effected by the local radial electric fields o Jo in the beam. The volumetric production rate of where Ib is the ion current being emitted from these CEX ions is given by, the thruster. The beam has a constant icex(r,z) = nf(r,z)nbi(r,z)vbioccx (6) divergence angle, 4, which is usually 15-20, and thus the beam radius is: rb = rr + z tan, where the relative collision velocity is taken to where rr is the thruster radius. The beam the beamon veloc Co sons of volumetric production with ground current is assumed to be predominantly axial, volumec rates d data with the beam velocity remaining approx- are good 4. imately constant over the length scale of interest of several meters, and hence the beamro la x Efflux NPE ion density is: S(rd) The presence of sputtered grid and nbi(r,z)= (3) discharge chamber metals in the plume e Vi presents a serious contamination hazard due to Neutral Efflux: these species' low vapor pressures. The production of these species is highly thruster dependent, and experimental data of sputter The unionized propellant that diffuses out yields is needed. However, estimates of NPE from the discharge chamber, exits in free- CEX production ratesl are orders of magnitude molecular flow. We use a simple point source less than those of the propellant CEX ions, and model "hidden" inside the thruster that thus their perturbative effect on the selfcompares reasonably well with Direct consistent potential structures in the plume will Simulation Monte Carlo (DSMC) be almost negligible. Hence, one can use the calculations 6 : potential fields computed self-consistently n()no (r) = n (z + rt) [(z + rt)2 + rt] 3 / 2 from the beam and propellant CEX ions, and then track the NPE ions in this field. The flux of neutrals is the Knudsen efflux, Currently, our model does not include these species. nnoc/4, where C= V8kT/rJmi. The neutral density at the thruster exit is controlled by the An important consideration for the beam current and the propellant utilization transport of the slow ions is the ambient and efficiency by the relation, thruster-induced magnetic fields. Table 1 4 I 1 - TP shows the gyroradii for thermal and beam ions nno= -An (5) in various magnetic field strengths e C corresponding to a range of orbital altitudes. where An is the flow-through area of the The thermal speed of the CEX ions is the 3

1321 IEPC-93-142 minimum speed, and represents ions that have continuity equation, not been accelerated through the potential drop of the beam. For the length scales that we are ane t + V. (nev) = 0 (8) interested in currently, (<2-3 m), the ions can For ion time scale behaviors, the electrons can be considered unmagnetized. However, be considered to be massless, and thus the depending on the type of thruster, strong momentum equation reduces to Ohm's law. thruster-induced fields must be taken into The electron drift velocity can then be solved consideration. directly, and expressed as, Table 1 Gyroradii for Xe ions vei =Fe m LEO GEO (B=0.2G) (B=0.001G). 1 F. p I bxfl Thermal CEX 15 m 3 km v e = m ion (T=500K). Beam ion > 680 m > 136 km where the force terms (parallel and V>10 km/s perpendicular to a magnetic field) include the electric field, the pressure gradient, and collisional drag between the beam and CEX Electons ions, and the neutrals: Electrons play a vital role in ion thruster operation in neutralizing the ion beam. A very F = -ee Pe + R n e ) important issue that remains to be resolved is The magnetic field will play an important the role of neutralizer electrons that the role in electron behavior in terms of plume spacecraft emits, and the effect of the ambient expansion. However, since our model is electrons. A rigorous formulation of the currently axisymmetric to capture the essential electron density would involve solving the physics of the CEX ion propagation, an electron continuity, momentum, and energy ambient magnetic field has been neglected. For equations and including the physics of a typical orbit raising mission scenarios where a neutralizer, which we will include in the future. nuclear powered spacecraft will spiral outward However, in our initial model, we treat beyond geostationary orbital altitudes, the electrons as an isothermal neutralizing fluid ambient magnetic fields are quite weak. Future with no drift velocity. The appropriate des- work will concentrate on developing a fully cription then is a Boltzmann distribution: three dimensional model to incorporate a ( ep\ magnetic field that will play a dominant role in ne = ne expk- (7) electron transport in LEO orbits. Since the plume cools as it expands, an Note that in this model, the electron density is a energy equation for the electron temperature is specified background density when the important to include. The temperature field potential reaches zero, or the reference space will be at a value of 1-5 ev that is typical in potential far from the beam. The Boltzmann thruster plumes, and will fall off to ambient relationship, often referred to as the temperatures in the far-field. In the future, we "barometric equation", has been experimentally will model this behavior by incorporating, verified, but only in local regions of the plume. V(3 V 3 Currently, we are assuming an isothermal tipe) + V(vPee) + pvvc = -V.cQe situation, but it is expected that the expanding (11) plasma will cool making the barometric where the electron heating term is due to ohmic relationship with a single temperature invalid, dissipation and collisional transfer. In the future, we will incorporate a rigorous electron fluid model comprising of the 4

IEPC-93-142 1322 I. Numerical Method production rate in the beam. The bulk of CEX ions are produced within 2-3 beam radii To model the expansion of an ion thruster downstream. plume, we employ the hybrid electrostatic Our model currently is two-dimensional plasma particle-in-cell (PIC) method. In the (r-z). Figure 2 shows a representative electrostatic PIC technique, ions and electrons computational grid which is nonuniform to in a plasma are treated as macro-particles, more efficiently handle the highly nonuniform where each macro-particle represents many density distribution in the plume. Since the actual particles. The charge of the simulation grid cell size should be on the order of the particles is deposited onto a grid and a charge Debye length, we have stretched the grids in density is computed. From this density, the r-direction to follow the increase in Debye Poisson's equation for the electrostatic potential length away from the centerline due to the is solved, and the particles are moved under the density decrease. influence of this self-consistent electric field. A With the Boltzmann distribution for the major shortcoming of explicit fully kinetic PIC electron density, the Poisson equation for the codes where electrons are treated as particles, is electric potential becomes nonlinear. This the very small time step that is required to equation is solved with a Newton-Raphson resolve the electron motion. Since we are Successive-Over-Relaxation (SOR) scheme. interested in the ion motion, we adopt the For large meshes, grid relaxation techniques hybrid approach where the ions are treated as are the methods of choice 7. Fixed potentials particles, but the electrons are treated as a fluid. are imposed on the spacecraft surfaces, and In this manner, the time step is now on the ion Neumann boundary conditions are held on all time scale, which for Xe ions, is about 490 exterior boundaries. times larger than the electron time scale. The equation of motion of each ion macro- IV. Selected Results and Discussion particle is integrated: dv i =q) We have performed a sample calculation d-= ()IE + vixb] (12) for a 15-cm Xe ion thruster operating with a beam current of 0.4 A, a propellant utilization where, in the electrostatic approximation, fraction of 0.84, and an accelerating potential of E=-Vtp, and the potential is determined from 1500 V. A beam divergence angle of 210 was Poisson's equation: used, as well as an electron temperature of 1 V2= e (nc - ni) (13) ev. A background ion density of 1010 m- 3 o speces was imposed. The following preliminary Note that the summation over the ion species results were run almost to steady-state. Figure allows different species such as propellant and 3 shows the potential contours of the beam, non-propellant ions. In our simulation model, and Figure 4 shows the CEX ion density the slow CEX ions are treated as particles, with contours in the plume. The propagation of the the real to macro-particle ratio around 1-2 CEX ions into the backflow region is clearly million. Particles are created each time step in seen, as well as the fact that the ion density each grid cell based on the volumetric CEX enhancement alters noticeably the beam production rate given by Eqn. 6. The velocities potential structure. Even though the CEX ion are those of a Maxwellian distribution with a density is at least two orders of magnitude less temperature corresponding to the wall of the than that of the beam ions, it must be selfdischarge chamber (usually around 500 0 K). consistently accounted for in Poisson's Particles that reach the simulation boundaries equation, and thus, particle tracking of slow and spacecraft surfaces are removed, and ions in a fixed potential field is not adequate. It steady-state is reached when the loss of is interesting to note how the CEX ions leaving particles at the boundaries balances the the beam form a "wing" structure. The sharp 5

1323 IEPC-93-14 2 potential drop at the beam edge which is the scale problems that will completely harness the accelerating mechanism for the CEX ions can power of massively parallel computers. We also be clearly seen. will also conduct parametric studies of Figure 5 shows the CEX ion current backflow fluxes for various thruster operating density vector field in the backflow region conditions such as beam current, specific behind the thruster plane. If, for instance, a impulse, and propellant utilization efficiency. highly biased solar array panel was located on These improvements in our model will result the spacecraft in this region, the backstreaming in an accurate numerical model of an ion ions would constitute a detrimental current thruster plume that can be used to accurately drain, provide estimates of contaminating fluxes so A number of representative CEX ion that spacecraft designers can treat the problem trajectories are shown in Figure 6. The CEX of integration with a much higher level of ions that are formed within the beam, leave the confidence. beam at angles almost normal to the beam edge and are accelerated to a speed corresponding to References the beam voltage drop, a speed that is greater 1) Samanta Roy, R.I. and Hastings, D.E., Electric than the Bohm velocity which is only a Propulsion Contamination, AIAA 92-3560, 28th Joint minimum velocity needed for a stable sheath. Propulsion Conference, Nashville, TN, July 1992. 2) Carruth, M.R., A Review of Studies on Ion As an example, for Te=leV, the Bohm Thruster Beam and Charge-Exchange Plasmas, AIAA velocity is about 860 m/s. However, the 82-1944, 16th International Electric Propulsion velocity achieved falling down a potential drop Conference, New Orleans, LA, Nov. 1982.. of roughly 11 V, is around 4000 m/s. Figures 3) Samanta Roy, R.I. and Hastings, D.E., Modelling 7 and 8 show phase-space plots of the CEX of Ion Thruster Plume Contamination, AIAA 93-2531,.te l v r y v s 29th Joint Propulsion Conference, Monterey, CA, ions. Figure 7 shows the radial velocity versus June 1993. radial position. At a radial distance of around 4) Gatsonis, N.A., et al, Modelling Induced 10 cm, we see a sharp acceleration up to nearly Environments and Spacecraft Interactions for the 4000 m/s. This is due to the large potential Nuclear Electric Propulsion Space Test Program drop at the beam edge. (NEPSTP), AIAA 93-2533, 29th Joint Propulsion Figure 8 shows a radial and axial velocity Conference, 5) Peng, X., Monterey, et al, Plasma CA, June Particle 1993. Simulation of plot which clearly shows two populations of Electrostatic Ion Thrusters, AIAA 90-2647, 21st ions. A low energetic population that is International Electric Propulsion Conference, formed inside the beam, and a more energetic Orlando, FL, July 1990. population that possesses a high radial velocity 6) T. Bartel, Private communication, Sandia component, as well as a backstreaming axial National Laboratory. component. A small number of ions that 7) Hockney, R.W. and Eastwood, J.W., Computer component. Simulation Using Particles. Adam Hilger, Bristol, expand around the top of the spacecraft and are 98gg. drawn to the spacecraft can also be seen. Acknowledgments IV. Conclusions This material is based upon work supported under a National Science Foundation Graduate Fellowship. We have developed a 2-D axisymmetric Any opinions, findings, conclusions or recommenda- Boltzmann electron hybrid PIC code with a tions expressed in this publication are those of the model the plume of an ion thruster for authors and do not necessarily reflect the views of the National Science Foundation. The authors would contamination purposes. We can see sharp like to acknowledge useful discussions with Professor potential structure in the beam that expels the M. Martinez-Sanchez of MIT, and Dr. Barry Mauk of CEX ions radially outward, as well as the CEX APL. We would also like to acknowledge the ion current density in the backflow region, support of the Office of Technology of the Ballistic Future work will be devoted to developing an Missile Defense Organization. electron fluid model that includes a neutralizer, and extending the simulation to 3-D on large- 6

S Charge-Exchange Plasma IEPC-93-142 132/ Solar Array Interactions Plume Expansion Communications. Interactions.i" S.Thruster Modelling 1.33... y (im).67i..67 Figure 2 Computational Grid.00,, Figure 2 Computational Grid 7

2-D PLASMA PLUME SIMULATION 1325 IEPC-93-142 2.40 POTENTIAL (Volts) POT L (Vo Figure 3 Plume Potential Contour Plot 11.0 10.0 9.0 1.60 8.0 y () 7.0 6.0 5.0.80 4.0 3.0 2.0 1.0.0.00.00 1.00 2.00 3.00 x ml 2-D PLASMA PLUME SIMULATION CEX ION DENSITY (m-3).400e+13.100e+13.500e+12 y (m) 1.60.100E+12.500E+11.100E+11.80.500E+10.100E+10.500E+09.100E+09.00.00 1.00 2.00 3.00 x (m) Figure 4 CEX Ion Density Contour Plot 8

2.00 2D PLUME SIMULATION CEX Ion Current Density IEPC-93-142 1326 1.33 y(m).67 * x) (m) 1.00 Figure 5 CEX Ion Current Density Plot 2-D PLUME SIMULATION CEX ION TRAJECTORIES.00.50 1.00 1.50 2.00 x(m) Figure CEX Ion CEX Iourrent Density Plotries.33.00.00.25.50.75 1.00 x (m) Figure 6 CEX Ion Trajectories 9

10000. 2D PLUME SIMULATION Vr - r Phase Plot 1327 IEPC-93-142 r (m/s) 4 5000. 2D PLUME SIMULATION Vz - Vr Phase Plot r (m) + ++ Figure 7 Vr-r phase plot 3000. + Vr (m/s) + ++ -1000. 600. -2000. 2000. 6000. 10000. Vz (m/s) Figure 8 Vr-Vz phase plot 10 10