Characterization Of A Neutralizer-Free Gridded Ion Thruster

Characterization Of A Neutralizer-Free Gridded Ion Thruster IEPC-2015-90256 /ISTS-2015-b-90256 Presented at Joint Conference of 30th International Symposium on Space Technology and Science 34th International Electric Propulsion Conference and 6th Nano-satellite Symposium, Hyogo-Kobe, Japan Dmytro Rafalskyi 1 and Ane Aanesland 2 Laboratoire de Physique des Plasmas (CNRS, Ecole Polytechnique, Sorbonne Universités, UPMC Univ Paris 06, Univ Paris-Sud), Ecole Polytechnique, Palaiseau, 91128, France Abstract: A gridded electric thruster that operates with only one RF power source providing propellant ionization, ion acceleration and beam neutralization is described here. Acceleration of both ions and electrons is achieved by a single set of extraction grids and the system is therefore neutralizer-free. The concept has a strong technology heritage from existing ion thrusters, where the main difference is due to using RF power for the ion acceleration. The acceleration grids terminating the plasma source are biased by an RF voltage across a blocking capacitor, such that a dc self-bias is formed externally to the plasma between the extraction grids. This self-bias accelerate ions, while electrons are coextracted in short instance between the RF cycles. It is shown here that the RF acceleration principle allows to almost double the space charge limited ion current through the extraction system. We present experiments that provide a first characterization and principal proof-ofconcept of the prototype called Neptune. The Neptune thruster is an ICP-based plasma source with a two-grid system where the RF power of 4 MHz from a single generator is distributed between the ICP inductor and the grids. A broad quasi-neutral ion-electron beam with the ion energy up to 800 ev is successfully generated without the use of a neutralizer. The beam transport is investigated by measuring time-averaged and timeresolved ion and electron energies, fluxes, densities and angular distributions using retarding field energy analyzers, collimated Faraday probe and other electrostatic probes. φ grd φ tgt V tgt V SB I tgt P acc PTE IEDF EEPF Γ Ar θ Nomenclature = instantaneous potential difference between two acceleration grids = instantaneous potential of the beam target = time-averaged potential difference between the target and ground = self-bias voltage; average value of the grid potential waveform = current of the beam target = RF power dissipated through the acceleration channel = RF power transfer efficiency = ion energy distribution function = electron energy probability function = mass flow rate of Argon = angle between the probe axis and a beam centerline 1 Post-doc, LPP, dmytro.rafalskyi@lpp.polytechnique.fr. 2 Researcher CR1, LPP, ane.aanesland@lpp.polytechnique.fr. 1

S I. Introduction imultaneous emission of oppositely charged particles from a single source 1-7 may provide several advantages in space. The additional neutralizer is then redundant, reducing the corresponding power and efficiency losses in the thruster and subsystems etc. Concepts using magnetic nozzles or double layer structures formed at the plasma source output allow the acceleration of a quasi-neutral plasma 6,7. Though these concepts provide in principle long-life neutralizer-free operation, the achievable performance such as specific impulse and thrust efficiency are generally lower than the performances provided by the traditional electric propulsion concepts (ionand Hall-thrusters) 6-8. Concepts based on simultaneous extraction of positive ions and electrons or positive and negative ions accelerated via biased grids may also provide a quasi-neutral beam 1-5. One advantage of these systems is the similarity with space proven gridded thrusters allowing technology heritage. In the Neptune concept, a so-called plasma self-bias effect is used for acceleration and co-extraction of electrons 1-3. This source provides a separate control of the ion energy and flux of the broad quasi-neutral beam. The idea is to apply a radio frequency (RF) bias voltage to a two-grid acceleration system via a blocking capacitor. In this case the space charge sheath in front of the grids oscillates with the RF frequency. The blocking capacitor charges up as a result of the effective area ratio between the first and second grid being larger than one. This allows rectification of the applied RF voltage by plasma and formation of a DC self-bias. In this system ions are continuously accelerated within the grids while electrons are emitted in brief instants within the RF period when the RF space charge sheath collapses. The cathode-neutralizer is therefore redundant. An inductively coupled plasma discharge powered by the same RF source can be used for the gas ionization, and in this case only one single RF power generator is needed for the system operation, providing three separate tasks: propellant ionization, ion acceleration and beam neutralization. (a) (b) Figure 1. Simplified drawing of the two-grid RF powered system and an equivalent RF electrical circuit. An RF voltage is applied between the grids, t 1 and t 2 corresponds to those on Figure 2 below. II. The Neptune thruster characterization A. The Neptune acceleration concept RF powered electrodes in asymmetric capacitive coupled plasma sources are widely used in the semiconductor industry to control independently the ion flux and ion energy impinging on the biased surface 9. The RF voltage drop across the sheath in front of the driven and grounded surfaces is inversely proportional to the sheath capacitance, C=ε 0 A/d, where A and d are the sheath surface and thickness, respectively, and ε 0 is the vacuum permittivity. Hence, the largest voltage drop is across the smaller electrode. When this driven electrode is connected to ground via a blocking capacitor (avoiding DC currents to ground 9 ), a DC bias is formed that rectify the RF voltage drop across the sheath. The self-bias formed across the low pressure sheath is typically V dc ~0.78V rf, where V rf is the amplitude of the RF voltage applied to the electrode 9. As a result, the ions are accelerated continuously across the combined DC-RF sheath, while electrons escape during the short RF period where RF voltage approaches the plasma potential and the sheath collapses 9. The average ion and electron currents to the electrode are equal due to the presence of the blocking capacitor in the circuit. Due to the RF oscillating sheath, the ion energy distribution impinging on the electrode surface tends to be bi-modal with an energy spread decreasing with increasing RF frequency 9. 2

In the Neptune concept we propose to apply the RF voltage to a set of grid in a classical gridded ion thruster or ion source. In the following we will elaborate how this may accelerate positive ions creating thrust, with co-extracted electrons provide space charge neutrality. Note, that we call this new concept Neptune after a dedicated research project acronym. Figure 1 shows a simplified drawing of the grid system and an equivalent RF electric circuit in line with the one of Coburn and Kay 10. The plasma is created using inductively coupled, microwave or other low pressure gas discharge. The RF voltage is applied between the first and second grid. The first grid is directly in contact with the plasma while the second grid is placed downstream and seen by the plasma only through the holes in the first grid. Oscillating space charge sheaths, represented by a capacitor and diode in parallel, are formed between the plasma and first grid, and between the plasma and second grid. The diodes in the circuit appears due to the asymmetry in the time response between the heavy ions and the much lighter electrons to the applied electric field, where the electron motion is controlled by the instantaneous electric field while the ion motion is mainly defined by the average electrical field. The sheath capacitance between the plasma and the first grid C 1 is higher than the sheath capacitance between the plasma and the second gird C 2 since the second grid is partially screened by the first one, i.e. the effective area ratio between the first and second grid is larger than one. As a result, the applied RF voltage V RF mainly drops across the sheath formed in front of each aperture of the second grid. The RF voltage is rectified to a DC self-bias voltage V DC due to the mentioned asymmetry in time response between ions and electrons and due to blocked DC current by blocking capacitor. Thus, in this case the asymmetry or difference in the DC self-bias voltage is formed between the extraction grids (external to the main plasma source). The DC component then accelerates ions continuously while electrons are expelled in moments when the oscillating plasma potential approaches zero and the inter-grid sheath collapses. Figure 2 shows the expected RF plasma potential waveform with respect to the grounded second grid. A φ cr is the plasma potential corresponding to the sheath collapse, and can be estimated using the Child s law for the given aperture dimensions and ion flux from plasma 2,9. As for RF biased electrodes, the ion energy distribution depends strongly on the applied RF frequency 9. For the self-bias to form, the applied RF frequency ω must be situated between the ion ω pi and electron plasma frequency ω pe defined as: pe e n, 2 0 e 0me pi e n, (1) 2 0 e 0M i here e 0 is the electron charge, n e is the plasma density, ε 0 is the vacuum permittivity and M i is the ion mass. Usually, this range includes RF signals from a few to hundreds of MHz, where the ion energy spread is inversely proportional to the applied frequency 9. Figure 2. The plasma potential waveform. The plasma potential is shown with respect to the grounded second grid. φ cr is the plasma potential corresponding to the sheath collapse, and can be estimated using the Child s law for the given aperture dimensions and ion flux from plasma 2,9. An interesting feature of the RF acceleration scheme follows from the periodical electron injection and as a result decreased space charge in the acceleration gap. According to Lieberman 9, the space charge-limited ion current for the RF sheath is 50/27 1.85 times larger for the same space charge sheath thickness and equivalent self-bias voltage comparing to the DC sheath. This should leads to almost double increase in the space charge limited ion current for the same extraction system comparing to the DC extraction case. B. The experimental setup The experiments presented in this work are performed using the second Neptune prototype, developed specially to investigate physical processes accompanying coincident ion-electron extraction and to map parameter space of thruster operation. The prototype is therefore not optimized in terms of total efficiency, thrust and specific impulse values and mass-dimensions characteristics. The prototype is a rectangular parallelepiped with the inner size 8x12x12 cm (see Fig. 3). An inductively coupled plasma (ICP) discharge is generated by a ferrite-enhanced 7-turns planar RF antenna separated from the plasma by a 2 mm thick ceramic window and powered by a 4 MHz radio frequency (RF) generator. The gas (Argon in this paper) is fed through the special port integrated into the discharge chamber. An RF distribution and matching system is used to allow both RF matching with the plasma load and to control the RF power distribution between the RF antenna and extraction/acceleration gridded system. The 3

extraction system is rectangular with dimensions 65x105 mm, formed by a set of two grids placed at the source exit. Both grids are made of stainless steel and the inter-grid distance is 2 mm. Two different sets of the extraction grids are used in the experiments. Both sets consist of two grids having 3312 extraction holes, with extraction apertures 1.5 mm diameter in both grids of the set 1; and 1.5 mm and 1.2 mm diameter for the first and second grids of the set 2. The holes in the grids are circular and aligned. In all cases the first grid is in direct contact with the plasma and is biased with RF voltages in the range 0-800 Vp-p via a blocking capacitor (C B ). The second grid is always grounded. Note that the ground potential in this work corresponds to the potential of a beam propagation tank (see below). Note also, that the grids are here called the first and second grid referring to the position from the plasma source; in the literature they may also be referred to as the screen and acceleration grids, respectively. The RF power is continuously measured for both the ICP and extraction channels separately, and the source is usually operated at constant ICP power while changing the RF voltage applied to the extraction system. The Neptune prototype is connected to a 1m long, 70 cm diameter vacuum tank (see Fig. 4). The vacuum tank chamber is equipped with the RF-compensated retarding field energy analyzer (RFEA), Langmuir probe (LP), collimated Faraday probe (collimator aspect ratio is 10) and the beam target that can be moved along the system axis. RFEA, LP and Faraday probe can be moved perpendicularly to the chamber axis, and rotated around the holder axis with an arbitrary angle. The standard distance between these probes and the prototype exit is 10 cm. The system is pumped by 2500 l/s turbo-pump providing residual gas pressure in the tank better than 10-7 Torr. The operating pressure in the tank is 0.08-0.3 mtorr and in the thruster prototype 0.8-3 mtorr. Figure 3. Schematic view on the Neptune thruster prototype (not in scale). C. Prototype operability The Neptune prototype uses ferrite-enhanced ICP, where the RF antenna is embedded into the ferrite core. The magnetic flux from antenna is guided directly to plasma, causing increase of RF coupling with plasma and decrease of RF losses in the conductors outside the plasma 2. The power transfer efficiency (PTE) to plasma in the Neptune prototype at a fixed ICP power as a function of the Ar flow is shown on the Fig. 5. It is seen that the coupling efficiency reaches value of about 80 % with the 15 sccm gas flow, and increase with increasing flow. Decreasing coupling efficiency at the lowest gas flows is one of the factors defining final range of operational parameters of the thruster prototype. The reason of the PTE decrease at low propellant flow is decrease of the gas pressure in the prototype, leading to increased electron temperature and increased antenna currents for the same ICP power applied. The ratio between the flow of propellant and pressure inside the Figure 4. Schematic view on the vacuum tank and beam / plume diagnostics. Figure 5. The power transfer efficiency (PTE) to plasma as a function of the Ar flow. 4

prototype can be controlled by the ion (or ion-electron) extraction system. This is illustrated on the Fig. 6 (a,b), where the time-averaged beam current collected by a metal target placed 10 cm downstream prototype is measured as a function of the Ar flow at fixed acceleration voltage (a) and as a function of the generated DC selfbias voltage (b), for two different sets of the extraction grids. The target is biased to -30 V for repelling the electrons. As follows from the Fig. 6 the set 2 allows to increase extracted ion current on about 10-20% for the 15-30 sccm flow range and fixed acceleration voltage, and to extend area of efficient beam extraction to the much lower flow values. The extracted ion current increases with increased self-bias acceleration voltage in all the range of available RF voltages applied to the grid (~700 V p-p, corresponds to ~300 V of DC selfbias), and the extraction system initially designed for the acceleration voltages about 1000 V cannot reach maximum of efficiency. One of the most important questions to the operability of the Neptune thruster concept is whether the self-bias effect can be efficiently formed in all the range of required plasma density, discharge power and acceleration voltages. The Fig. 7 shows waveforms of the grid voltage measured in the Neptune prototype when 200, 420 and 650 V p-p of RF voltage is applied and an ICP power is fixed at the constant value (150 W). It is clearly seen that in all the cases the self-bias effect leads to almost complete rectifying of the applied RF voltage, such that the grid voltage waveform is V=V dc +V rf where V dc ~0.8V rf. The RF power dissipation through the grid channel should be ideally controlled only by the losses to ion acceleration, since moments of the electron extraction correspond to the lowest values of the potential difference between both grids and plasma (moments when the space charge sheath collapses). However, in the real system the RF power from the generator is firstly split between two channels (ICP inductor and Figure 6. The time-averaged ion current to the target. The ion curren I tgt is measured as a function of the Ar flow at fixed acceleration voltage (a) and as a function of the generated DC self-bias voltage (b) for two different sets of the extraction grids. The target is placed 10 cm from the source exit and biased to -30 V. Figure 7. Waveforms of the first grid potential in the Neptune prototype. Figure 8. RF power dissipated by the grid channel as a function of the self-bias voltage at the fixed ICP power. 5

grids) and then delivered through the matching to the grids. The RF losses depend on the RF voltage applied to the grid, RF current, connection length, grid material and a ratio between the ICP and grid power. Additional power losses is possible by dissipation in the insulators which are in direct contact with both the high voltage RF and grounded parts (due to the tg δ ). In the Neptune power system the RF losses for voltages below 600 V are found to be negligibly small. This is illustrated on Fig. 8, where the RF power dissipated by the grid channel is measured as a function of the self-bias voltage for the fixed ICP power (150 W). The grid power is obtained by subtracting the known ICP power from a power required to reach corresponding self-bias voltage (the ICP power is kept constant by keeping the same RF current in the inductor). At the RF voltages lower than 200 V p-p (and corresponding DC self-bias voltages < 100 V), the power dissipated by acceleration channel is very small (less than 1 watt), and significantly increases up to the 20 W at the higher RF voltages. For comparison, figure 8 shows DC power delivered to the grid when the RF acceleration is not used such that the prototype operates similarly to the common ion thruster (the applied DC voltage is equivalent to the self-bias in the RF case). It is seen that the acceleration power is very similar for both DC and RF acceleration schemes. Figure 9. The beam IEDFs for the Neptune prototype operating at 150 W ICP power D. Beam performance As shown previously 2,3, the general performance of the Neptune thruster is very similar to a classical ion thruster with similar geometry and operating at similar conditions (i.e. with the DC acceleration and an external neutralizer). However, since ions are accelerated by an oscillating electrical field, the ion energy distribution function (IEDF) has a more complicated broadened shape that depends on the ion flux, acceleration voltage and RF frequency. The IEDFs of the Neptune beam operating at 150 W ICP power are shown on Fig. 9 for self-bias voltages of 100, 200, 250 and 300 V. The lowest acceleration voltage provides an almost single-peaked IEDF, where the peak Figure 10. position roughly corresponds to the value e(v sb +V pl ), prototype. View on the plume of the Neptune where V sb is the DC self-bias voltage and V pl is the DC plasma potential relatively to the plasma source walls (about 30 V). With increasing RF acceleration voltage the IEDF broadening increases, and the main peak on the ion energy distribution almost disappears at 300 V. In this case the IEDF is distributed between 100 ev and 780 ev. The maximum ion energy should corresponds to e(v peak +V pl ), where V peak is the peak value of the applied RF voltage. The multi-peaked structure of the IEDFs is due to the ions being accelerated in the RF-DC space charge sheath located between the grids of the thruster and these peaks are reproducible and not a result of noise. However, the exact physical description of these multi-peaked IEDFs is still missing. One of the important beam parameters for the Neptune thruster is the ion beam divergence, since RF Figure 11. The angular distribution of the ion ion acceleration can be expected to strongly affect the flux to the collimated Faraday probe. 6

ion trajectories in the extraction system compared to the DC case. Preliminary observations including simple optical measurements have shown that the beam divergence is quite small and apparently close to standard systems with DC ion acceleration 2 (see Fig. 10 for the Neptune plume image). In this work, the angular distribution of the ion flux has been measured by a collimated Faraday probe where the aspect ratio of the collimator is 10. The Faraday probe is placed at the beam centerline 10 cm downstream the source and rotates around its axe. The measured ion current as a function of the probe orientation relatively to the centerline is shown on the Fig. 11 for 3 cases of the self-bias voltage (100 V, 200 V and 300 V). The calculated half-angle divergences of the ion beam are 22, 19 and 14 degrees for acceleration with 100 V, 200 V and 300 V self-bias, respectively. Thus, the beam divergence for the Neptune thruster is similar to classical ion engines with DC ion acceleration and beam neutralization downstream 11. The plume produced by the Neptune thruster consists of the fast accelerated ions, slow ions created due to secondary processes (including charge-exchange) and fast and slow electrons. In addition, the ion and electron fluxes and densities oscillate due to the periodical electron injection each RF period, and due to the ion acceleration in the RF sheath. The plume characterization should therefore consider both time-averaged and time-resolved parameters. The time-averaged IV curve of the beam target is shown on the Fig. 12 for the 100 V and 200 V selfbias voltages and reflects the averaged ion and electron emission properties of the Neptune thruster. At the negative voltages, the target repels electrons and collects ions (similarly to the planar probe), and ion current saturation value corresponds to a total ion current (without taking into account charge-exchange collisions). Fig. 12 shows that the ion current to the target is saturated at negative voltages lower than 20 V for both the 100 V and 200 V self-bias voltage. The floating target potential corresponds to equal fluxes of ions and electrons arriving to the target (timeaveraged), and is less than 25 Volts in both cases. At the positive target voltages lower than the acceleration voltage, the target still collects ions, but the electron flux is not suppressed. It should be mentioned that at the target potentials higher than plume potential (10-20 V) the plume starts to be disturbed 12. The time-resolved ion and electron fluxes in the thruster plume are investigated by measuring timeresolved current to the target when it is biased negatively enough to repel all electrons ( 1 on the Fig. 12) or positively to collect both ions and electrons (( 2 on the Fig. 12). The real electron flux oscillation is deduced by subtracting the ion current waveform from the data measured in the positive bias case. The time-resolved ion and electron currents to the beam target are shown on Fig. 13 in comparison with the grid voltage waveform for the case of 200 V self-bias voltage. The target is located 10 cm downstream from the thruster exit in order to measure almost non-relaxed oscillations of the ion and electron fluxes. The ion flow is almost constant, except during a short period (< 20 ns) where the Figure 12. Time-averaged IV curves of the beam target. Figure 13. Time-resolved ion and electron currents in the plume. Figure 14. Time-averaged fit of EEPF measured in plume of the Neptune thruster. 7

current drops to almost zero at the minima of the grid voltage. The electron current waveform (see also Fig. 13) show that the electron flow has a certain base level, oscillating several times during the RF period with approximately 30% pulsations amplitude. At each minimum on the grid voltage waveform initiate new packet of the electron current oscillations. Thus, the experimental measurements show that both ion and electron fluxes pulses and might be one of the reasons for the complicated broadened multi-peaks of the IEDFs. The electron flow in the plume has been further investigated using the cylindrical RF-compensated Langmuir probe and rotating double-grid RF-compensated retarding field energy analyzer (RFEA). The low-energy part of the electron energy probability function (EEPF) is measured by a Langmuir probe, while the high-energy part of EEPF is obtained by the RFEA. The reason is that the Langmuir probe can give quite precise results for relatively low energy electron measurement, while the RFEA can measure only the electrons with energy high enough to overcome the potential barrier formed between the averagely positive plume and the RFEA entrance. The combined timeaveraged EEPF measured in the Neptune plume 10 cm from the thruster exit is shown on Fig. 14 for the case of 200 V self-bias. It is seen that the EEPF consists of an almost Maxwellian population with a corresponding electron temperature of 6.5 ev and a high-energy tail. We suggest that the high-energy population corresponds to the pulsed electrons extracted directly from the source, while the low-energy Maxwellian core is created in the plume through the secondary processes. The possible electron flow anisotropy can be a key advantage of the Neptune thruster. Both co-extracted and co-directed ions and electrons can produce a low diverging plume of low potential relatively to the spacecraft, with a reduced plasma density outside the plume in comparison with traditional neutralizer-based systems. In order to demonstrate the directionality of the electron flow, the EEPF tail has been measured in the plume formed at 200 V self-bias, by the rotating the RFEA facing the beam and oriented at 90 degrees to intercept only perpendicular electrons. Fig. 15 shows that the electron flow is highly anisotropic, decreasing by about 5 times perpendicularly to the beam direction. E. Space charge compensation The Neptune thruster is a self-neutralizing thruster emitting both accelerated ions and electrons of the same flux. The ion/electron extraction is quasi-simultaneous, meaning that the continuous ion flow is accompanied by a pulsed electron flow at the applied RF frequency. The DC current is blocked in the system by the blocking capacitor, thus the average fluxes of ions and electrons are absolutely equal, as shown previously 2,3. However, the pulsed nature of the electron emission leads to plume potential oscillations, and the amplitude of these oscillations is a very important parameter for a final thruster application. Fig. 16 shows the plume potential waveforms measured by a floating target at different self-bias voltages (100 V, 200 V and 300 V). As expected, the plume potential oscillates at the RF frequency used for ion acceleration and electron extraction (here 4 MHz). In the case of 100 V self-bias the plume potential oscillates at about 15 V p-p, and for 200 V and 300 V the oscillations are quite similar but with a peak-topeak value about 25 V. The shape of the plume potential waveforms can be explained as follows. Each minimum of the plasma potential in the source (relatively to the second grid) corresponds to the electron extraction phase, when the space charge sheath between the extraction grids collapses. In these moments, the electron injection almost immediately decreases the plume potential (seen on Fig. 16 around 210 ns, 460 ns and 710 ns). After a relatively short time, the space charge sheath is again formed and the electron flow is blocked. This is accompanied by a fast Figure 15. Tails of EPFs of electrons in parallel and perpendicular to the plume directions. Figure 16. Plume potential waveform. 8

potential increase up to some base value, and then the plume potential continues to increase with a relatively small slope. Hence, the fast potential increase corresponds to the electron flow closure. When electron extraction is stopped the ion space charge in the plume starts to build up, and experiment shows that this process leads to an almost linear increase of the plume potential with an approximate slope of 60 V/μs (for the self-bias voltages more than 100 V). It should noted here that the boundary conditions in the experiment, tank pressure and other facility effects can be considered as parameters strongly affecting plume oscillations and dynamics of the space charge compensation in the plume produced by the Neptune thruster. Therefore, further experiments in much bigger space simulation tank operating at lower gas pressures are required to completely investigate these issues. III. Conclusion In this paper we presents experimental results of a new thruster prototype called Neptune, based on novel concept of simultaneous ion-electron acceleration from a plasma using one single RF power source for plasma generation, ion acceleration and electron neutralization. We show that both ions and electrons are accelerated by the RF powered grids ensuring a complete space charge compensated beam downstream. The self-bias generation, that is crucial for this concept is found to appear in the entire range of operational parameters (pressure 0.8-10 mtorr inside the prototype, and RF voltage range 200-700 V p-p). The self-bias effect leads to an almost complete rectifying of the applied RF voltage, such that the grid voltage waveform is given by V=V dc +V rf where V dc ~V rf. It is shown that at 200 V p-p RF acceleration voltage the IEDF is almost single-peaked where the peak position roughly corresponds to the value e(v sb +V pl ), where V sb is the DC self-bias voltage (about 100 V for this case) and V pl is the DC plasma potential relatively to the plasma source walls (about 30 V). With increasing RF acceleration voltage the IEDF broadening increases and the main peak on the energy distribution almost disappears at 600 V p-p voltage. In this case the IEDF is distributed between 100 ev and about 780 ev. The half-angle divergences of the ion beam are found to be better than 15 degrees for the high RF voltages (> 600 V p-p) and therefore similar to classical ion engines with DC ion acceleration and beam neutralization downstream. The ion beam is found to slightly oscillate, with an almost constant flux during the largest part of each RF cycle with short (~20 ns) periods where the ion flux almost vanishes (at the minima of the grid voltage). The measured electron current waveforms show that the electron flow has a certain base level where each minimum on the grid voltage waveform initiate oscillating bursts on the electron current. It is found that the EEPF of the produced plume consists of an almost Maxwellian population with a corresponding electron temperature of 6.5 ev and a high-energy tail with energies up to 50 ev. The high-energy population corresponds most likely to the pulsed electrons extracted directly from the source, while the low-energy core is created in the plume through secondary processes. The measured plume potential oscillates with the driving frequency and is less than 25 V p-p for RF acceleration voltages lower than 600 V p-p. It is also found here that the electron flow is highly anisotropic; the flux is about 5 times less in the perpendicular direction compared to the axial one. The result presented here are very promising for space applications, since the directionality of both the ions and electrons can allow to reduce the density of plasma cloud surrounding the spacecraft and hence reduce the interaction with antennas, diagnostic equipment, solar panels etc. Acknowledgments This work was supported by a Marie Curie International Incoming Fellowships within the 7th European Community Framework (NEPTUNE PIIF-GA-2012-326054). References 1 D. Rafalskyi and A. Aanesland, French patent application for a Dispositif de formation d'un faisceau quasi-neutre de particule de charges opposees, No. 14 53469, filed 17 April 2014. 2 D Rafalskyi and A Aanesland 2014 J. Phys. D : Appl. Phys., 47, 495203. 3 D Rafalskyi and A Aanesland Neutralizer-free gridded ion thruster, paper AIAA-2014-3423 from 50th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, July 28-30, 2014, Cleveland, OH, USA. 4 Aanesland A, Meige A, and Chabert P 2009 J. Phys.: Conf. Ser. 162 012009. 5 Chabert P 2008 Electronegative plasma motor US Patent 2008/0271430 A1. 6 C Charles and R W Boswell 2004, Phys. Plasmas 11, 1706. 7 J. C. Sercel Electron-Cyclotron-Resonance (ECR) Plasma Acceleration, paper AIAA-87-1407 from AIAA 19th Fluid Dynamics, Plasma Dynamics and Lasers Conference. June 8-10, 1987 Honolulu, Hawaii. 9

8 Groh K H and Loebt H W 1991 Journal of Propulsion and Power 7 573. 9 Lieberman M A And Lichtenberg A J 2004 Principles of Plasma Discharges and Materials Processing, 2nd ed. (Wiley, New York) 10 Coburn J. W. and Kay E. 1972 J. Appl. Phys. 43 4965 11 Goebel D M and Katz I 2008 Fundamentals of Electric Propulsion (Wiley, Hoboken) 12 S V Dudin, D V Rafalskyi 2011 Eur. Phys. J. D, 65, 475 10