This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 193.180.206.253

This content was downloaded on 01/04/2014 at 08:03

Please note that terms and conditions apply.

Hysteresis effects in the formation of a neutralizing beam plasma at low ion energy

View the table of contents for this issue, or go to the journal homepage for more

2013 EPL 104 35004

(http://iopscience.iop.org/0295-5075/104/3/35004)

Home Search Collections Journals About Contact us My IOPscience

November 2013

EPL, 104 (2013) 35004 www.epljournal.org

doi: 10.1209/0295-5075/104/35004

Hysteresis effects in the formation of a neutralizing beamplasma at low ion energy

Dmytro Rafalskyi(a) and Ane Aanesland

Laboratoire de Physique des Plasmas, CNRS-Ecole Polytechnique - route de Saclay, 91128 Palaiseau, France

received 18 September 2013; accepted in final form 4 November 2013published online 29 November 2013

PACS 52.40.Mj – Particle beam interactions in plasmasPACS 52.75.-d – Plasma devicesPACS 41.75.Ak – Positive-ion beams

Abstract – In this paper, the PEGASES II thruster prototype is used as an ion source generatinglow-energy (< 300 eV) positive Ar ion beam, extracted without an external neutralizer. The ionsare extracted and accelerated from the source using a two-grid system. The extracted positive ionbeam current is measured on a large beam target that can be translated along the accelerationaxis. The ion beam current shows a stepwise transition from a low-current to a high-currentextraction regime with hysteresis. The hysteresis region depends strongly upon the beam targetposition. Langmuir probe measurements in the plume show high plasma potentials and low plasmadensities in the low-current mode, while the plasma potential drops and the density increases in thehigh-current mode. The ion energy distribution functions of the beam are measured for differentregimes of ion extraction. The ion beam extracted in the high-current mode is indicated by thepresence of an additional low-energy peak corresponding to ions from an ion-beam plasma createdin the downstream chamber, as well as 10–20 times higher intensity of the primary ion beam peak.The hysteresis behavior is explained by the formation of a downstream neutralizing beam plasma,that depends on the target position and pressure in agreement with a Paschen-like breakdown bysecondary electrons. The obtained results are of high relevance for further development of thePEGASES thruster, as well as for improving existing neutralizer-free concepts of the broad-beamion sources.

Copyright c© EPLA, 2013

Introduction. – Broad-beam ion sources (IS) arewidely used as thrusters in space missions [1] and in dif-ferent technologies for manipulating surfaces [2], particu-larly in rare gas and reactive ion beam etching of GaAsopto-electronic devices and microwave integrated circuits,micro-machining, thin-film magnetic heads and nanostruc-ture fabrication [3,4]. Inductively coupled plasmas (ICP)are widely used in IS, due to the high and uniform plasmadensity (> 1018 m−3) and excellent reactive gas compati-bility [5]. Ion extraction and acceleration is provided bya multi- or single-grid ion-optical system which usuallyconsists of relatively thin (1mm or less) perforated elec-trodes (grids) spaced by several millimeters with appro-priate voltages applied between the grids [6–10].

Using IS in space thruster applications or for ion beamprocessing of insulating surfaces requires the neutral-ization of both the ion positive space charge and theion current. Usually, for this purpose electron injection

(a)E-mail: dmytro.rafalskyi@lpp.polytechnique.fr

from a neutralizer is used [1]. Currently, the lifetime,power/weight requirements and reliability of the neutral-izer system is a drawback for the gridded ion sources inspace [10].

The PEGASES thruster, acronym for Plasma Propul-sion with Electronegative GASES, accelerates alternatelypositive and negative ions for thrust [11]. In this way, thespace charge and current are compensated within the ISsystem without the need for additional electrons. Thisis obtained by using an ion-ion plasma formed in a low-pressure plasma source [12]. The alternate ion extractionis provided by a set of grids placed in the ion-ion plasma re-gion and biased with bi-polar voltage waveforms [13]. Animportant issue in the operation of PEGASES thruster isthe ion-ion beam transport and space charge compensa-tion downstream of the acceleration stage.

A detailed investigation of the downstream space re-quires large vacuum facilities to eliminate both pressureand wall effects. For economical reasons, initial experi-mental investigation for the intention of developing this

35004-p1

Dmytro Rafalskyi and Ane Aanesland

thruster is naturally limited to standard laboratory vac-uum facilities prior to a first proof-of-concept. A largethrust facility would need to be developed for the dedi-cated gases intended for the formation of ion-ion plasmas(electronegative gases such as iodine).

In this paper we therefore focus on the characteriza-tion of the positive ion extraction and transport from thePEGASES II prototype in order to eliminate pressure,chamber and diagnostic effects for the future experimentswith alternate beams. The investigations are carried outfor positive ion beam acceleration and transport withoutthe use of a downstream electron neutralization source.Neutralization of the ion beam space charge is provided bya secondary ion-beam plasma (also called plasma plume orbeam-produced plasma), which is normally formed in theion beam transport chamber at pressures above 0.01mtorrdue to secondary processes [9,14,15]. We show in this pa-per that breakdown and extinction of the ion-beam plasmadepends on different parameters and plays an importantrole in the low-energy ion beam transport.

Experimental setup. – The experimental setup isshown in fig. 1. The PEGASES II prototype has beendetailed previously [16,17]. Briefly, it is rectangular witha cross-section of 8 cm by 12 cm and 12 cm long. Theplasma is produced by a purely inductively coupled radiofrequency (RF) coil, powered at 4MHz via a transformercoupled impedance matching network. The coil is sep-arated from the plasma by a 2mm thick dielectric win-dow. A ferrite encapsulates the coil for better couplingefficiency [18]. The RF power from the generator is mea-sured by a standard reflectometer. The power transferefficiency is here 0.7 with a discharge power of 300W.

In order to isolate the plasma from the conductivethruster body a parallelepiped of PYREX is placed in-side the thruster with a cross-section of 116 × 76mm. Aset of holes is placed on both sides of the chamber bodyto inject the gas uniformly in volume. Argon was used asthe working gas in all experiments described in this pa-per. A transversal magnetic field is created by permanentmagnets such that the maximal magnetic field inside thesource is 245G on axis. The result using Ar gas is a re-duced electron temperature and density in front of theextraction grids [16].

The ion extraction and acceleration is provided by adouble-grid extraction system. The two grids are identicaland made of stainless steel, the apertures are circular witha diameter of 2.5mm, and the total grid transparency is0.6. The intergrid distance is 2mm and the grid thicknessis 1mm. The working area of the grids is a rectangle withdimensions 65 × 105mm. The high DC voltage source isconnected to the first grid in contact with the plasma,while the second grid is grounded. Note, that as in anygridded RF ion source with dielectric walls, the currentclosure when extracting positive ions from the source isensured by electrons flowing to the first grid in contactwith the plasma [2].

Fig. 1: Illustration of the experimental setup.

The PEGASES prototype is attached to a much largercylindrical beam transport chamber having length 70 cmand diameter 60 cm. The chamber is pumped by a turbo-molecular pump with a 2500 l/s pumping speed providinga residual pressure in the chamber of 2 · 10−6 torr. Thegas pressure is directly measured in the beam transportchamber, while pressure inside the PEGASES source isestimated from the gas flow rate knowing the grids ge-ometry. The extracted ion beam propagates through thischamber equipped with different diagnostics for analyzingthe downstream processes.

For a complete understanding of the extraction pro-cesses, both inductively coupled plasma (ICP) and ionbeam-produced plasma (IBP) are analyzed. Two Lang-muir probes are therefore placed in the ICP and IBPregions, and are, respectively, called the LP-ICP andLP-IBP probe. The LP-ICP probe is inserted in the pri-mary plasma through one extraction aperture and theLP-IBP probe is mounted on the target, as shown in fig. 1.The probe tips are identical and made from a Pt-Ir wirewith a diameter 0.05mm and length 7mm. The LP-IBPprobe measurements are carried out with the ALP sys-tem from Impedance Inc. [19], while the LP-ICP probemeasurements are obtained using a specially developedfloating ramp system “RelRamp” which uses the first gridpotential as reference. The ion flux from the primary ICPto the extraction grids is measured by a planar probe (PP)placed on the grid surface, also shown in fig. 1. The PPis a disc of copper with an area of 1 cm2, and biased by afloating low-leakage DC source at −25V vs. the potentialof the first grid.

The ion energy distribution functions (IEDFs) in thedownstream IBP chamber are measured with a simplesingle-grid retarding-field energy analyzer (RFEA) [8].The analyzer, embedded into the target, is made using astainless-steel mesh with 0.15mm apertures placed 2mmfrom a collector. The total grid transparency is 0.32.The grid is grounded while the collector is scanned us-ing the Semion acquisition system from Impedance Inc.The IEDFs are obtained by differentiation of the collectorI-V curves with Savitzky-Golay pre-smoothing.

35004-p2

Hysteresis effects in the formation of a neutralizing beam plasma at low ion energy

Fig. 2: The ion current density to the extraction system as afunction of (a) full RF power at fixed pressure (1 mtorr) and (b)the Ar pressure in the prototype at fixed RF power (200 W).

The total ion beam current is measured by a conductivetarget placed in the beam transport chamber. The targetis a Cu disc, 150mm in diameter and 5mm thick and canbe grounded or DC biased depending on the experiments.The distance between the downstream grid (source out-put) and the target can be varied from 50mm to 170mmwith a precision of ∼1mm. As mentioned above, the tar-get serves also as a holder for the RFEA and the LP-IBP.

All the potentials in this work are measured with respectto the grounded beam transport chamber. The second gridand entrance grid of the RFEA are also grounded.

Experimental results. – In order to define the rangeof operation parameters for the PEGASES II prototypeused as a positive ion source here, the ion flux to the ex-traction system was experimentally measured using theplanar probe placed on the first grid surface. Figure 2(a)shows the measured ion current density as a function ofthe RF power from the generator. The ion current den-sity increases close to linearly with the RF power. Theobserved shift from 0 to 100W for the initial point canbe explained by the RF power losses in the matchbox andinduction coil including the ferrite core.

Figure 2(b) shows the ion current density as a functionof the argon gas pressure inside the prototype at fixedRF power of 200W. The pressure range is near to thepoint of low-pressure discharge extinction since efficiention extraction and transport requires low pressure of op-eration [2,20]. It is seen from fig. 2(b) that there is a localmaximum of ion current density near to the 1mtorr pres-sure. Hence, this pressure was chosen for the followingexperiments. The corresponding Ar flow is 18 sccm andpressure in the beam transport chamber is 0.2mtorr. TheRF power of 200W from the generator is chosen, to al-low long-term operation of the thruster prototype withoutoverheating. This regime provides the generation of theion flow to the extraction system with a current densityof about 1mA/cm2.

Figure 3 shows a typical I-V curve measured using thebeam target at a constant acceleration voltage of 200V.The curve can be separated into two parts correspondingto positive and negative target biasing. At the positivetarget potentials the target reflects ions and collects elec-trons created in the beam transport chamber. When thetarget potential is negative, it collects all beam ions and

Fig. 3: The I-V curve of the target intercepting the ion beam,measured with 200 V of acceleration potential at x = 125 mm.The positive target current corresponds to the current of pos-itive particles.

Fig. 4: Dependences of the extracted positive ion current mea-sured by the negatively biased target Itgt on the accelerationvoltage ϕacc for two different positions x of the target: 125 mm(a) and 105 mm (b).

reflects electrons, with the presence of a saturation region(the target saturation current at a potential of −25V is inthe following referred to as “ion current to the target”).The hysteresis observed at the positive target biasing maybe understood by the processes in the IBP and will bediscussed below.

The extracted positive ion current as a function ofthe acceleration voltage was measured for two differentposition x of the target, and the results are shownin fig. 4(a), (b), where x is the distance between theprototype output and the target. The target is biased at−25V here. When the acceleration voltage is increasedthe ion beam current firstly decreases, which can beexplained by plasma expansion through the extractionsystem apertures at low voltages [14]. Then, after a slightincrease of the ion beam current, at the accelerationpotential of about 140V for the target position = 125mmand 280V for the position = 105mm, the ion currentjumps to a “high-current” mode with an increase of about5–10 times. This transition indicates that the ion-beamplasma is formed in the beam transport chamber andcompensates the beam space charge. Further increase ofthe acceleration voltage leads to the following increaseof the ion beam current up to a value of about 11mAat 300V. When the acceleration voltage is reduced, thesystem stays in the “high-current” mode of ion extractionuntil the voltage reaches about 120V for both targetpositions. Below this potential, a sudden transition to the

35004-p3

Dmytro Rafalskyi and Ane Aanesland

Fig. 5: (Colour on-line) Dependences of the plasma potentials(a) and densities (b) on the acceleration voltage. The targetposition is 125 mm. The filled triangles correspond to oscilla-tions of IBP and reflect the amplitude of the IBP potential (a)and density (b).

“low-current” mode occurs accompanied by a decrease ofmore than 10 times in the ion beam current.

In order to understand the physical reasons leading tothe observed behaviour of the extraction curves, we mea-sure the parameters of the ICP and IBP during differentphases of the ion extraction. The target position is herefixed at 125mm. The I-V curves of the Langmuir probesplaced in ICP and IBP regions are measured at differentacceleration voltages. The I-V curves are then analyzedfollowing the methods described in [5].

Figure 5 shows the deduced plasma potentials and den-sities as a function of the acceleration voltage. We cansee that the ICP potential follows the acceleration poten-tial and the ICP density remains mainly constant with avalue of about 2 · 1016 m−3. However, the IBP potentialand density dependences are non-linear having hystere-sis with transitions at the same potentials, as shown infig. 4(a). Initially, at voltages below 50V the IBP po-tential follows the ICP potential, meaning that plasmapenetrates through the extraction system [14]. At thesame time, the IBP density rapidly decreases after a slightinitial increase, indicating an intergrid sheath formation.At voltages between 60 and 110V there is no detectablesignal from the Langmuir probe, so the beam is appar-ently non-compensated by space charge and its currentis defined only by the Child-Langmuir limit [21] for thegiven distance between the target and the extraction sys-tem. At an acceleration voltage of about 120V the IBPis formed. In the range 120–140V beam plasma oscillates

Fig. 6: The IEDFs measured by the RFEA for “low-current”(a) and “high-current” (b) ion extraction modes. The targetposition is 125 mm.

with a frequency of about 200Hz. Figure 5 shows the am-plitude of plasma potentials and density oscillations (filledtriangles). The amplitude of the beam plasma potential isinitially close to the ICP potential and the density ampli-tude is about 1014 m−3. At the acceleration voltage cor-responding to the transition to the “high-current” mode(at about 140V), the IBP potential rapidly drops to ap-proximately 50V, while the IBP density slightly decreases.This transition is accompanied by the disappearance ofany IBP oscillations. After that, an increase of the ac-celeration voltage does not significantly change the IBPpotential and density. These results are in good agree-ment with previous works on the ion-beam plasma in asingle-grid ion-beam system [14,22].

Reverse scan of the acceleration voltage cause IBP ex-tinction at a voltage corresponding to a “high-current”to “low-current” transition. Thus, dependences of theion-beam plasma parameters on the acceleration voltagealso demonstrate hysteretic behavior, besides transitionsbetween extraction modes are accompanied by suddenchanges of the beam plasma potential and density.

In order to characterize the extracted ion beam we mea-sure the ion energy distribution functions with the RFEAdescribed in the experimental setup section. Figure 6shows the IEDFs obtained for different ion extractionmodes (“low current” and “high current”). For the ac-celeration voltages lower than 50V the amplitude of theion peaks decreases with increasing acceleration voltage,which is in agreement with the measured target current,see fig. 4(a). The measured ion energy corresponds to theICP potential. The IEDF measured for ϕacc = 20V hastwo peaks, which can be explained by the ion fluxes com-ing from regions before and after the magnetic filtering,indicating probably potential structures inside PEGASES,in agreement with previous works [17]. The ion peak forϕacc = 50V in fig. 6(a) apparently corresponds to the ionscreated in the IBP and accelerated in the sheath near tothe analyzer entrance, since IBP potential is a few voltshigher than the measured ICP potential (see fig. 5(a)).The increase of the acceleration voltage leads to the fullsuppression of the ion flow from the source, until the accel-eration voltage reaches about 120V, when IBP is formed(see fig. 6(a), (b)). Increasing the acceleration voltage inthe “high-current” mode cause a corresponding increaseof the positive ion energy (see fig. 6(b)). The mean width

35004-p4

Hysteresis effects in the formation of a neutralizing beam plasma at low ion energy

Fig. 7: The IEDFs measured at the same acceleration voltage of120 V for “low-current” and “high-current” extraction modes.

of the ion peaks in the “high-current” mode is less than20 eV, so the extracted beam is close to be mono-energetic.

The IEDFs measured at the same acceleration voltageof 120V for “low-current” and “high-current” extractionmodes are shown in comparison in the fig. 7. A significantdifference between the amplitudes of the high-energy ionpeaks is clearly seen from this figure, which is expectedfrom the previously measured target current at these twomodes. The additional low-energy peak on the “high-current” IEDF near to the IBP potential corresponds tothe slow ions created in the IBP. It is interesting to notethat in the “high-current” mode the flux of slow ions is atleast 20 times lower than the fast ion flux which resultsfrom quite low gas pressure in the beam transport chamber(about 0.2mtorr for this experiment) [20]. Note here also,that all the IEDFs observed in the “high-current” modecontain this low-energy ion peak, corresponding to the IBPpotential and therefore the ions created in the IBP.

The peak around zero energy appearing on all ion en-ergy distributions (see fig. 7, for example) is associatedwith secondary electrons created on the analyzer collec-tor surface due to ion-electron emission. The amplitudeis always about 15% of the detected ion current, and theposition is defined only by the grid potential (0V for allthe experiments described here). Previous works describ-ing the ion energy analysis with single-grid retarding-fieldanalyzers show similar results for this peak amplitude andposition [8,20,23].

Discussion. – The measured dependences of theplasma parameters, ion beam current, and ion energy dis-tribution functions show that the IBP plays an impor-tant role in the low-energy positive ion extraction fromPEGASES. The observed transitions and hysteresis on theextraction curves can be associated with the IBP forma-tion process that is confirmed by the Langmuir probe mea-surements (fig. 5). In order to understand the physicalreasons causing the observed phenomena let us considerthe general processes accompanying broad low-energy pos-itive ion beam transport without the presence of a neu-tralizer. As was previously investigated [9], the ion-beamplasma, which can be formed in this case, consists of at

least 4 kinds of charged particles: “fast” beam ions, “fast”electrons created due to secondary ion-electron emission,“slow” electrons created via ionization of residual gasatoms by “fast” electrons, and “slow” ions created dueto charge-exchange collisions and residual gas ionizationby “fast” electrons. The “slow” electrons are trapped inthe potential well of the positive ion beam providing spacecharge neutralization of the beam [9,24]. Thus, productionof the “fast” and “slow” electrons in the beam transportspace plays a key role in the IBP creation and mainte-nance, respectively [9]. Since the pd factor affects theionization rate (where p is the gas pressure, and d is thecharacteristic size of the system) [5], the IBP breakdownand extinction are consequently controlled by this factor.During the left-to-right scan of the acceleration voltage(i.e. increasing the voltage) after the intergrid accelerationsheath is formed (ϕacc > 50V) the IBP breakdown is diffi-cult due to the limited ion current to the target (or cham-ber wall) by space charge [21], and therefore controlledby a ‘pdx’ factor, where “dx” is responsible for both theionization rate and the space charge-limited ion current.Therefore, as seen by the experiments, the voltage of the“low”- to “high”-current mode transition is in very strongdependence upon the target position: a change of the tar-get position about 15% increases the IBP breakdown volt-age by almost the double (see fig. 4(a), (b)).

The voltage leading to a transition from the high to thelow-current mode is not strongly affected by the targetposition (see fig. 4), indicating the influence of anotherphysical mechanism on the IBP extinction. In fact, theIBP disappearing means that the creation of “slow” elec-trons does not compensate their losses to walls (especiallyto the dielectric surfaces, charged by the beam ions).The results indicate that there is some minimal ion beamcurrent for which the quantity of created “slow” electronsis enough for maintaining the ion-beam plasma for thegiven configuration of the beam transport chamber.Passing this threshold current leads to the IBP disappear-ance, and the value for the data presented here is about3–3.5mA of the ion beam current (see fig. 4). A slightdifference between these thresholds for two cases of targetpositioning can be explained by the beam divergence. Thehypothesis about the existence of a minimal ion currentrequired for IBP maintenance is also in accordance withthe experimentally measured target I-V curve (see fig. 3).

Note here, that the speculations above do not ex-plain the experimentally observed oscillations of IBP inthe acceleration voltage range 120–140V for the targetposition = 125mm. A complete understanding of thephysical picture here is difficult without numerical model-ing of the system.

Conclusions. – Thus, in the present paper, thePEGASES II thruster prototype is used for a low-energypositive Ar ion beam extraction. The ion current den-sity is measured as a function of the Ar pressure and RFpower. The ion current density to the extraction system

35004-p5

Dmytro Rafalskyi and Ane Aanesland

is about 1mA/cm2 at 300W of RF power at 1mtorr. De-pendences of the extracted positive ion beam current onthe acceleration voltage measured at different positionsof a large target intercepting the ion beam have shown ahysteresis transition from “low-current” to “high-current”regimes. This behavior is explained by different mecha-nisms of downstream plasma breakdown and extinction.The proposed explanation is confirmed by the Langmuirprobe measurements conducted in the primary plasma ofthe PEGASES thruster and in the ion-beam plasma in thebeam transport chamber. It is shown that the hystere-sis region dramatically depends on the distance from thethruster output to the beam target. The IEDFs of thebeam are measured for different regimes of ion extractionfrom the PEGASES II prototype. It is observed that theion beam extraction in the “high-current” mode is indi-cated by the presence of an additional low-energy peak cor-responding to the ions from the ion-beam plasma createdin the downstream chamber, as well as 10–20 times higherintensity of the primary ion beam peak. The mean widthof the ion peaks in the “high-current” mode is less than20 eV, so the extracted beam is close to be mono-energetic.

An important practical conclusion that can be madefrom the data presented in this paper is that an ion beam-produced plasma (IBP) can be formed in the PEGASESsystem operated at low pressures (∼ 0.1mtorr), providingfull space charge compensation of the ion beam. There-fore, alternate ion-ion extraction in technology-orientedapplications of PEGASES-based systems can be obtainedwith frequencies lower than those theoretically predictedby solving of pure electrostatic problem without takinginto account secondary processes [25]. In this case, chargecollection in the IBP will delay the beam suppressiondue to non-compensated charge. The IBP here can playa role of “capacitor” for the space charge, similarly toknown systems with quasi-simultaneous ion-electron ex-traction [8,20]. However, these issues require also the un-derstanding of downstream plasma formation during thenegative ion beam extraction, that is not yet much inves-tigated for the low-energy range (< 1 keV) in the knownliterature [26].

It can also be concluded that the observed hystereticbehaviour of the ion extraction curves is inherent for anybroad positive ion-beam system without the presence ofa neutralizer or additional source of the negative charges.This statement is in agreement with measurements con-ducted in the single-grid systems, where the hystereticbehaviour of the extraction curves has been explained bythe effects of plasma shaping near to the wires of theextraction grid [23].

∗ ∗ ∗This work was supported by the Marie Curie Inter-

national Incoming Fellowships within the 7th EuropeanCommunity Framework, by ANR (Agence Nationale de laRecherche) under Contract ANR-2011-BS09-40 (EPIC),and by EADS Astrium.

REFERENCES

[1] Goebel D. M. and Katz I., Fundamentals of ElectricPropulsion (Wiley, Hoboken) 2008.

[2] Brown I. G., The Physics and Technology of Ion Sources,2nd edition (Wiley-VCH, New York) 2004.

[3] Rius G., Llobet J., Esplandiu M. J., Sole L., Bor-

rise X. and Perez-Murano F., Microelectron. Eng., 86(2009) 892.

[4] Hsiao R., IBM J. Res. Dev., 43 (1999) 89.[5] Lieberman M. A. and Lichtenberg A. J., Principles of

Plasma Discharges and Materials Processing, 2nd edition(Wiley, New York) 2004.

[6] Kaufman H. R., J. Vac. Sci. Technol. A, 4 (1986) 764.[7] Bassner H., Killinger R., Leiter H., Muller J. and

Box P., Development Steps of the RF-Ion Thrusters RIT,in Proceedings of the 27th International Electric Propul-sion Conference, Pasadena, USA, 2001, IEPC-01-105,http://erps.spacegrant.org/uploads/images/images/

iepc articledownload 1988-2007/2001index/105 2.

pdf.[8] Dudin S. V. and Rafalskyi D. V., EPL, 88 (2009)

55002.[9] Dudin S. V., Zykov A. V. and Farenik V. I., Rev. Sci.

Instrum., 65 (1994) 1451.[10] Nishiyama K. and Kuninaka H., in Transactions of

The Japan Society for Aeronautical and Space Sci-ences, Aerospace Technology Japan, Vol. 10 (2012)pp. Tb 1Tb 8.

[11] Aanesland A., Meige A. and Chabert P., J. Phys.:Conf. Ser., 162 (2009) 012009.

[12] Amemiya H., Jpn. J. Appl. Phys., 30 (part 1) (1991)2601.

[13] Aanesland A., in Proceedings of the 64th Gaseous Elec-tronics Conference, Salt Lake City, Utah, USA, 2011,Bull. Am. Phys. Soc., 56, No. 15 (2011).

[14] Dudin S. V. and Rafalskyi D. V., Eur. Phys. J. D, 65(2011) 475.

[15] Huashun Zhang, Ion Sources (Science Press andSpringer-Verlag, Beijing) 1999.

[16] Aanesland A., Bredin J., Chabert P. and Godyak

V., Appl. Phys. Lett., 100 (2012) 044102.[17] Bredin J., Chabert P. and Aanesland A., Appl. Phys.

Lett., 102 (2013) 154107.[18] Godyak V. A., Plasma Sources Sci. Technol., 20 (2011)

025004.[19] Hopkins M. B., Graham W. G. and Griffin T. J.,

Rev. Sci. Instrum., 58 (1987) 475.[20] Dudin S. V., Rafalskyi D. V. and Zykov A. V., Rev.

Sci. Instrum., 81 (2010) 083302.[21] Langmuir I., Phys. Rev., 2 (1913) 450.[22] Dudin S. V. and Rafalskyi D. V., Rev. Sci. Instrum.,

83 (2012) 113302.[23] Rafalskyi D. V. and Dudin S. V., Appl. Phys. Lett.,

97 (2010) 051501.[24] Humphries J. S., Charged Particle Beams (John Wiley

& Sons, New York) 1990.[25] Aanesland A., Low temperature plasmas exposed to

external magnetic and electric fields, HDR (UniversitePierre et Marie Curie, Paris) 2013.

[26] Soloshenko I. A., Rev. Sci. Instrum., 75 (2004) 1694.

35004-p6