Optimized H − extraction in an argon–magnesium seeded magnetized sheet plasma

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Optimized H� extraction in an argon–magnesium seededmagnetized sheet plasma

Virginia R. Noguera a,*, Gene Q. Blantocas a,b, Henry J. Ramos a

a Plasma Physics Laboratory, National Institute of Physics, University of the Philippines, Diliman, Quezon City 1101, Philippinesb West Visayas State University, Lapaz, Iloilo City 5000, Philippines

Received 24 August 2007; received in revised form 27 December 2007Available online 1 February 2008

Abstract

The enhancement and optimization of H� extraction through argon and magnesium seeding of hydrogen discharges in a magnetizedsheet plasma source are reported. The paper first presents the modification of the production chamber into a hexapole multicusp con-figuration resulting in decreased power requirements, improved plasma confinement and longer filament lifetime. By this, a wider choiceof discharge currents for sustained quiescent plasmas is made possible. Second, the method of adding argon to the hydrogen plasmasimilar to the scheme in Abate and Ramos [Y. Abate, H. Ramos, Rev. Sci. Instr. 71 (10) (2000) 3689] was performed to find the optimumconditions for H� formation and extraction. Using an E � B probe, H� yields were investigated at varied argon–hydrogen admixtures,different discharge currents and spatial points relative to the core plasma. The optimum H� current density extracted at 3.0 cm from theplasma core using 3.0 A plasma current with 10% argon seeding increased by a factor of 2.42 (0.63 A/m2) compared to the measurementof Abate and Ramos [Y. Abate, H. Ramos, Rev. Sci. Instr. 71 (10) (2000) 3689]. Third, the argon–hydrogen plasma at the extractionchamber is seeded with magnesium. Mg disk with an effective area of 22 cm2 is placed at the extraction region’s anode biased 175 V withrespect to the cathode. With Mg seeding, the optimum H� current density at the same site and discharge conditions increased by 4.9times (3.09 A/m2). The enhancement effects were analyzed vis-a-vis information gathered from the usual Langmuir probe (electron tem-perature and density), electron energy distribution function (EEDF) and the ensuing dissociative attachment (DA) reaction rates at dif-ferent spatial points for various plasma discharges and gas ratios. Investigations on the changes in the effective electron temperature andelectron density indicate that the enhancement is due to increased density of low-energy electrons in the volume, conducive for DA reac-tions. With Mg, the density of electrons with electron temperature of about 3 eV increased 3 orders of magnitude from 2.76 � 1012 m�3

to 2.90 � 1015 m�3.� 2008 Published by Elsevier B.V.

PACS: 52.25.Xz; 52.50.Dg; 29.25.Ni; 52.70.Nc; 52.20.Hv; 52.25.Jm

Keywords: Magnetized sheet plasma; Hydrogen negative ions; Rovibrational excitation; Dissociative attachment; Atomic processes; Molecular processes

1. Introduction

H� ions are precursors to neutral beam injection pro-cesses in fusion devices. They are often used in acceleratorsfor charge exchange injection, for fusion plasma heating,directed energy weapon research and even in semiconduc-

tor applications for etching and material surface treatment.A case in point for the latter is the use of H� beams in del-aminating thin silicon layers up to depths of severalmicrons. This has resulted in the commercialization of anew method producing thin single-crystal silicon layers atlow cost and the industrialization of the manufacture ofsolar cells with higher energy efficiency [2]. Unfortunately,H� ions are difficult to create, and the very characteristicthat makes them attractive, the ease with which electronis detached from the ion, means that it is difficult to create

0168-583X/$ - see front matter � 2008 Published by Elsevier B.V.

doi:10.1016/j.nimb.2008.01.037

* Corresponding author. Tel./fax: +63 29204475.E-mail address: [email protected] (V.R. Noguera).

www.elsevier.com/locate/nimb

Available online at www.sciencedirect.com

Nuclear Instruments and Methods in Physics Research B 266 (2008) 2627–2637

NIMBBeam Interactions

with Materials & Atoms


high concentrations of these species due largely to substan-tial collisional losses in the extraction and acceleration pro-cesses [3–5]. For these reasons, there is a need to constantlydevelop particle sources and schemes that could efficientlyproduce higher yield of H� ions. For the most part, H�

ions are produced by sources with tandem structureswhereby there is first, a driver region for the creation ofplasma and the rovibrationally excited molecules, and sec-ond, a region for low temperature plasma to form the neg-ative ions [6–9].

H� ions can be produced either by surface or by volumeproduction. In the former mode, positive hydrogen ions areattracted towards a negatively biased electrode known asthe converter electrode. Upon impact, energy associatedwith the converter voltage is transferred to the H+ ions.The energy gained allows these ions to combine withdetached electrons from the electrode surface forming H�

ions. In the latter mode, negative hydrogen ions are formedvia a two-step process. In the first step, neutral hydrogenmolecules in the volume plasma are brought to higher rota-tional and vibrational states by collisions with energeticelectrons. The second step is the dissociative attachmentof low-energy electrons to the rovibrationally excited mol-ecules. Here, the ion source usually has a tandem structure.

For the case of magnetically filtered multicusp ionsources, which is known for having versatile applicationsdue to its capability of producing large volumes of uni-form, quiescent and high-density plasmas with high gasand electrical efficiencies, the ion beam yield can beenhanced by optimizing the chamber length, plasmaparameters and wall materials [10,11]. The most exploredplasma parameter is the total gas filling pressure. Resultsof some studies suggest that the negative hydrogen ionyield increases with pressure until an optimum pressure isattained [12–14]. Optimization of filling pressure is neces-sary especially for electron-beam generated plasma becausethe fraction of negative ions could decrease at high pres-sures. Furthermore, increase in pressure has drawbacksfor the case of neutral beam injection application becausemuch of the generated ions can be destroyed by electronstripping due to collision with residual molecules in theaccelerator.

Other works include investigations on the effect of add-ing noble gases and alkali metals to the hydrogen dis-charge. The addition of noble gases has been found outto significantly decrease the electron temperature [15] orincrease the low-energy electron density [16,17] therebyincreasing the probability of the dissociative attachment(DA) process. Another work indicates that the enhance-ment is due to a resonant energy exchange between excitedAr atoms and H2 molecules [18].

The commonly used alkali metal is cesium in whichenhancement is observed on both surface and volume pro-duction type ion sources. The enhancement rate is higher inthe surface-production mode, which can be attributed tothe decrease in the work function of the cesium-coveredconverter and the chamber wall material. Shinto et al.

observed the exponential dependence of H� current onthe decrease of the work function [19]. From the first effortof producing intense H� beams by Dimov et al., [20] shortpulse ion sources that employ the technique of cesiation arenow being used in the charge exchange injection into accel-erators and charge exchange extraction from accelerators[21].

Various experiments were also carried out to observe theeffect of different wall materials in volume produced hydro-gen plasma by installing various metal liners. Leung et al.installed different metal liners on the chamber wall of amagnetically filtered multicusp source [10]. They foundout that aluminum and copper could generate the highestH� current while stainless steel produces the lowest. Thedifference in yield is attributed to the amount of secondaryelectrons emitted from the wall surfaces. The finding ofLeung was then confirmed by Fukumasa et al. [22].

Control on the plasma parameters is also achieved byinstalling a grid between the driver region and the extrac-tion region [23,24]. Takahashi et al. employed the directcurrent laser photodetachment method to investigate theeffects of the grid voltage bias on the electron temperature,electron density and the H� current density. Their resultsaccentuated the established fact that H� current densityshowed positive correlation with electron density and neg-ative correlation with electron temperature.

Leung, et al. installed a barium washer insert at theextraction aperture of the ion source and observed anincrease in the H� ion yield by a factor of 3 at the optimumbias potential. The enhancement is brought about by thesurface conversion of low-energy positive ions reboundingfrom the barium insert [25].

Recently a new technique to enhance H� productioncalled the streaming neutral gas injection (SNGI) is devel-oped by Mendenilla et al. The H� current density isincreased by injecting a highly directional neutral hydrogengas near the extraction aperture to create a region of local-ized high electron–neutral collision rate thus lowering theeffective electron temperature of the plasma [26].

Another work utilizes a dipole magnetic field due to apair of Sm–Co permanent magnets installed behind the tar-get cathode to reflect electrons to the extraction electrodeand reduce the extraction power load [27]. This has conse-quently increased the H� production by 10%.

Unlike in most volume sources, the magnetized sheetplasma negative ion source (MSPNIS) [1,28] employed inthis work dispenses with the physical requirement of a dri-ver region. Endemic in the machine is the distinctive sepa-ration of different thermal electrons which allows for thecombination of the two-step procedure namely: H2 excita-tion, plus electron dissociative attachment (DA) in onlyone reaction chamber, denoted here as the extractionchamber. In this work, numerical constructs of the electronenergy distribution function (EEDF) using the Druyves-teyn relation clearly shows the separation of these differentthermal electron species. This inherent configuration facil-itates the production of H� in a more physically uncompli-

2628 V.R. Noguera et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 2627–2637

https://www.researchgate.net/publication/222798531_Grid_bias_effect_upon_the_negative_ion_density_in_a_tandem_plasma_source?el=1_x_8&enrichId=rgreq-f5eeab25c270733d03e81d51ff86e825-XXX&enrichSource=Y292ZXJQYWdlOzIyMzUyNTk0ODtBUzo5OTgxMTc4MDk4ODkzMkAxNDAwODA4MzkxMjI0


cated mode, an obvious characteristic advantage over othertypes of ion sources, which at times require magnetic filtersand mesh grids to separate the plasma into the required hotand cold regions. This paper presents two schemes ofenhancing H� production rate. Firstly, magnetic field cuspswere incorporated in the production chamber for betterplasma confinement. Secondly, plasma at the extractionchamber was seeded with argon and magnesium. Theenhancement effects were analyzed vis-a-vis informationgathered from the usual Langmuir measurements (electrontemperature and density), electron energy distributionfunction (EEDF) and the ensuing DA reaction rates at dif-ferent spatial points for various plasma discharges and gasratios. It appears that the modification of the MSPNIS intoa hybrid magnetic multicusp source together with the com-bined seeding of argon and magnesium has increased H�

yield by as much as 11.88 times compared to an earlierwork employing pure argon seed only [1]. To date, therehas never been any attempt to employ argon and magne-sium mixture as seed in enhancing H� yield. This workexplicitly demonstrates the cooling mechanism inducedby a group II metal. The dynamics of collisional processesin relation to plasma parameters and enhancement of H�

extraction is also discussed. The considerations hereinshould contribute to the database of H� volume produc-tion in sheet plasmas. These are insights that may be rele-vant to fusion.

2. Experimental setup and methodology

A schematic diagram of the MSPNIS is shown in Fig. 1.The ion source is composed of two distinct regions: theplasma production region and the extraction region. Theproduction chamber houses the gas inlet and the feed-through for a hair-pin shaped tungsten filament cathode(0.4 mm in diameter). The gas mixture is fed into the pro-

duction chamber and plasma is produced by thermionicemission at V2 equal to 30–60 V. A base pressure of2 � 10�5 Torr is achieved by a combination of a 4-inchoil diffusion pump and a 35 m3/h – pumping speed rotarypump.

With the composite magnetic field produced by the com-bination of the limiter E1 (enclosing a circular ferrite mag-net with maximum B-field strength of 350 Gauss at thecenter), the limiter E2 (enclosing a coreless magnetic coil),the Helmholtz coils at the extraction region and the dipoleSm–Co magnets (about 1.5 kG on the surface), a quiescentsheet plasma (13 cm wide and a few millimeters in thick-ness) is extracted at V1 equal to 72–115 V.

Another set of permanent ferrite magnets of dimensions(3 � 4 � 50 mm3) with maximum B-field strength of 0.7 kGat the surface is installed at the production chamber in ahexapole multicusp configuration. The effect of this addi-tional magnetic confinement to the production and extrac-tion of quiescent plasma, and on the enhancement ofextracted H� current are discussed in the next section.

Before conducting any actual data gathering, in situ Ar-plasma discharge cleaning of up to 5 A was performed forabout 1 h to eliminate adsorbed molecules and water vaporfrom the chamber walls. Similar cleaning procedure wasalso done prior to Mg seeding. Since the vacuum back-ground attained is in the order of 10�5 Torr only, this stan-dard protocol is effective in minimizing outgas of moleculesharmful to the formed H� ions, and additionally removesoxide impurities from the Mg disk attached to the anode.

Using a set of mass flow controllers, H2 and Ar gas arefed at different filling ratios. The proportion of argon to thetotal gas feed is fixed at 10%, i.e. 10% Ar: 90% H2. Two gasflow rates were used. One low filling pressure (9 sccmH2 + 1 sccm Ar) for a total gas feed of 1.3 mTorr and theother, high filling pressure (18 sccm H2 + 2 sccm Ar) corre-sponding to 2.7 mTorr gas feed. For each flow rates,

Fig. 1. Schematic diagram of the MSPNIS.

V.R. Noguera et al. / Nucl. Instr. and Meth. in Phys. Res. B 266 (2008) 2627–2637 2629


plasma parameters, i.e. effective electron temperature (Teff)and density (Ne) as well as the corresponding H� flux den-sity are measured at varied plasma discharge currents (Id).A single Langmuir probe (0.5 mm in diameter and 5 mm inlength) is used to measure Teff and Ne. Prior to the mea-surement, the probe is cleaned by sputtering in order toavoid the influence of contaminants on the probe surface.The H� flux density is measured by an E � B probe shownin Fig. 2. The effective magnetic field along the axis of theextraction region is used as the external field which bal-ances the scanning electric field between the probe’s deflec-tor plates. The extractor plate is biased at 190 V. Thecollimating plate is biased at the same voltage to avoidthe acceleration of extracted ions. The shield of the probeis grounded. The effective area of the collecting Faradaycup is 1 mm2. An extensive discussion on how the probewas calibrated and optimized in relation to H� detectionis found in [28].

Both E � B and Langmuir probes are oriented perpen-dicular to the sheet at the side opposite the pumps andare coupled to the ion source using stainless steel bellowsso that they can be moved at several distances from thesheet for localized measurements. Automatic data acquisi-tion routines store the probes’ measurements for subse-quent data processing and analysis.

A typical E � B probe trace is shown in Fig. 3, taken at3 cm from the sheet plasma for Id equal to 3 A at high fill-

ing pressure. Ipeak is obtained from the absolute value ofthe difference of the most negative current and the averageof current values before the dipping starts. The correspond-ing H� current density is then obtained from the ratio ofIpeak and the effective area of the Faraday cup.

The electron energy distribution function (EEDF), f(E),is obtained from the second derivative of the I–V traceusing the Druyvesteyn relation [29]

f ðEÞ ¼ 4

APe2

ffiffiffiffiffiffiffimV2e

rd2I

dV 2; ð1Þ

where Ap is the surface area of the probe exposed to theplasma, e is electron charge, m is electron mass, E is elec-tron energy, V = Vsat � Vp is probe voltage referenced tothe saturation potential and Ip is current collected by theprobe.

To obtain the second derivative of probe measurements,probe characteristics were smoothed by fitting it with aneight-term Gaussian function using MATLAB 7.0’s curvefitting tool facility

IpðV pÞ ¼X8

i¼1

ai exp � V p � bi

ci

� �2" #

; ð2Þ

where ai, bi and ci are constants.MATLAB 7.0’s curve fitting tool facility also allows one

to obtain the second derivative of the fitted function. TheDruyvesteyn relation can thus be used to obtain the elec-tron density (Ne) and effective electron temperature (Teff)using Eqs. 3,4,5

N e ¼Z Emax

0

f ðEÞdE; ð3Þ

T eff ¼2

3hEi; ð4Þ

hEi ¼ 1

N e

Z Emax

0

Ef ðEÞdE; ð5Þ

where Emax is the maximum energy at which the secondderivative approaches zero and hEi is the electron meanenergy.

By convention, the effective temperature (Teff) is theaccepted terminology whenever the Druyvesteyn relationis employed.

In cases where the analyses demands DA reaction rates(K), they are estimated by the usual convolution procedure,given by [30]

KðEÞ ¼ ð2mÞ

12

Z 1

0

E12rDAðEÞf ðEÞdE; ð6Þ

where

rDAðEÞ ¼ rpeak exp �E � Eth

0:45

� �; ð7Þ

rpeak is the cross section at the specified threshold value Eth

and f(E) is the actual EEDF obtained from the Druvesteynrelation.

Fig. 2. The E � B probe.

Fig. 3. Typical E � B trace for H� current density measurements.



The threshold values, Eth, with the corresponding crosssections for various vibrational levels, v, of H2 are takenfrom Ref. [31].

3. Results and discussion

3.1. EEDF measurements

In the MSPNIS, energetic electrons are confined at thesheet core. Low energy electrons scatter to the peripheryby e–e Coulomb collision or by collisions with hydrogenatoms diffusing out across the confining magnetic field ofthe Helmholtz coil and the Sm–Co magnets. This allowsthe electron impact excitation of the hydrogen moleculeto occur at the center and the subsequent production ofnegative hydrogen ions to occur at the periphery whenrovibrationally excited H2 combine with the cold electrons.Since the energetic electrons are separated from the H�

ions, loss due to high-energy electron impact is lessened.Thus, the distribution of electron population plays a cen-tral role in charged particle chemical kinetics as it isthrough the electrons that electrical energy is principallycoupled to the plasma. Therefore, the plasma discharge isbest characterized by measuring the EEDF especially forthe case of H� production where the formation anddestruction rates are very much affected by the availableenergies of the electrons. The novelty in the design of theMSPNIS is seen in the changes of the calculated EEDF(shown in Fig. 4) as the probe is moved away from thesheet. At the center (0 cm), three distinct electron groupsare present. The peaks are observed at 23 eV, 9 eV and2 eV. The highest energy is at 40 eV. At 1 cm from thesheet, the highest energy is still at about 40 eV but anincrease in the population of the electrons occupying the

lower energy part of the spectrum increases (peaks areobserved at 20.5 eV, 10.5 eV and 4 eV). At this distancethe EEDF is also characterized with three distinct electrongroups. A shift towards the lower energy values is observedas the probe is positioned at 2 cm from the sheet. The high-est energy is now at about 30 eV and there are only two dis-tinct electron groups present (peaks at 16.5 eV and 4.5 eV).Furthermore, only one electron group is observed for alldistances. At 3, 4 and 5 cm, the highest energy is at about20 eV, 18 eV and 15 eV, respectively and the peaks arefound at 7.5 eV, 4.5 eV and 4.5 eV, respectively. The driverregion required by most volume sources appears unneces-sary in the MSPNIS due to this inherent separation of fastand cold electrons. The two-step sequence of forming thenegative ion can be initiated in a single chamber. In thisrespect, a sheet plasma device supersedes other volumesources.

However, with the present configuration of the MSP-NIS, there is limited number of cold electrons in the extrac-tion chamber. As a result, production of H� ions viadissociative attachment is limited. Hence, the source maynot perform at its full efficiency without any source of coldelectrons or without any mechanism to induce electroncooling. Previous work of Abate et al. [1] on the samesource shows that the addition of argon in the hydrogendischarge significantly decreased the electron temperatureat the periphery resulting in the increase of H� yield by1.73 times (from 0.15 A/m2 to 0.26 A/m2). The currentpaper, serving as sequel to Abate’s work, investigates theeffects on H� yield in response to the following inputchanges:

(i) addition of a multicusp magnetic confinement in theproduction region with argon as seed and

(ii) seeding with argon and magnesium mixture.

3.2. Effect of the multicusp magnetic confinement in the

production region

The installation of additional magnets around theproduction chamber to form the multicusp magnetic con-finement described in Section 2 has improved the perfor-mance of the source. Stable plasma can be extracted evenat low plasma production currents. Lower values are pre-ferred to minimize electrode heating and to lengthen fila-ment lifetime.

3.3. Effect of argon addition

Experiments show optimized H� extraction occurring atId = 3 A. Fig. 5 shows the variation of H� ion yield withrespect to distance for this optimum discharge current.Results show maximum H� current density at 1 cm awayfrom the sheet core for low filling pressure and 3 cm forhigh filling pressure. The trend shows optimum extractionof 0.63 A/m2 at 3 cm. Ion density falls sharply up to theFig. 4. Typical EEDF profiles at various distances from the sheet.



6 cm distance, nonetheless showing minimal recovery at theperiphery near the chamber wall (7 cm).

Fig. 6 presents the spatial variation of the plasmaparameters for Id = 3 A. The gradual tapering of Teff andNe is the familiar feature of sheet plasmas. The 3 cm dis-tance especially for the case of high filling pressure couldbe the ideal site for maximum H� extraction since it liesin the region sandwiched between the core (dominated byhot electrons), and the periphery (comprising primarily ofcold electrons). This intermediate region brings togetherthe conditions that initiate dissociative electron attachmentto rovibrationally excited H2 molecule exiting the sheetcore.

Fig. 7 shows H� ion density as a function of Id at theoptimum 3 cm distance. The figure shows a steep rise inH� production rate for high filling pressure until 3 A.Above this threshold, H� extraction begins to level out.The presence of hot electrons and bare H+ ions at high dis-charges may cause collisional detachments (CD):e + H�? H + 2e, and mutual neutralization reactions:H� + H+ ? 2H reducing H� output. The small yield forlow filling pressure is easily attributed to low Ne, possiblynot enough to initiate additional DA reactions.

Fig. 8 presents the variation of plasma parametersagainst discharge currents. At low filling pressure, Teff

Fig. 5. H� current density measurements of Ar + H2 discharge atoptimum discharge current (Id = 3 A).

Fig. 6. Spatial variation of plasma parameters for Ar + H2 dischargemeasured at Id = 3A.

Fig. 7. H� current density measurements versus discharge currents at3 cm distance from the sheet core. Id = 3 A gives optimum extraction.

Fig. 8. Variation of plasma parameters against discharge currents at 3 cmdistance from the sheet core.



remains almost constant but rises when discharge isincreased above 3 A. The same response is observed forelectron density. For the case of high filling pressure, Ne

and Teff are greatly affected by the increase in dischargecurrent. The increase in the influx of hot electrons fromthe production region together with the increased collisionrate due to decreased mean free path in the volume leads tomore noticeable increase in Teff. A surprising featurethough is seen at Id = 3 A where Teff and Ne dropped shar-ply. These significant dips are remarkable in that Id = 3 Aas confirmed earlier is actually the discharge condition giv-ing the highest negative ion yield (shown in Fig. 7). Expect-edly the yield should have decreased due to the drastic dropin electron density. The EEDF plots of different plasmadischarges shown in Fig. 9 reveal an interesting find. Thedistributions of various plasma discharges except Id = 3A exhibit multi-peak profiles. It would appear that thepresence of multiple electron species or groups with highthermal energies is detrimental to the yield, as the likeli-hood of electron stripping of H� increases. The mono-peakfeature of the 3 A discharge describes a unique plasmaenvironment whereby the electron population is composedonly of a single thermal group with a characteristic energyof 3 eV (i.e. the maximum probable energy of the distribu-tion is 3 eV since the EEDF profile peaks at this value).Electrons of this energy range present in sufficient amount(�1012 m�3) favors dissociative attachment reactions,hence the observed H� enhancement at 3 A despite thedrop in Ne. The existence of a single electron species maytherefore be considered as one of the essential conditionsfor efficient H� extraction in a sheet plasma device.

Figs. 10 and 11 show the effect of Ar:H2 ratio on the ionyield and plasma parameters, respectively for both low andhigh gas flow rates. Measurements are made at optimumconditions, i.e. Id = 3 A with Langmuir and E�B probespositioned 3 cm away from the sheet. Results indicateincreasing argon content in the mixture leads to increaseddetection of H�. For low filling pressure, the increase canbe attributed to the cooling effect of argon as the effectiveelectron temperature is significantly reduced from 13 eVto slightly above 3 eV. However, in the case of high filling

Fig. 9. The EEDF profiles of different plasma discharges at 3 cm distancefrom the sheet core. Only the EEDF for Id = 3 A shows a single peakprofile, the rest shows multiple peaks. This mono-peak feature reveals thesole presence of a group of electrons with characteristic energy range of3 eV favors dissociative attachment, hence the observed enhancementeffect at this discharge condition.

Fig. 10. Effect of Ar:H2 ratio on ion yield for both low and high gas flowrates at optimum conditions, i.e. Id = 3 A, 3 cm distance.

Fig. 11. Effect of Ar:H2 ratio on plasma parameters for both low and highgas flow rates at optimum conditions, i.e. Id = 3 A, 3 cm distance.



pressure, the increase in H� current density is attributed toincreased density of low temperature electrons. Works ofNishiura and Bacal have established that enhanced coldelectron density leads to a much higher probability of H�

formation [32].

3.4. H� enhancement with magnesium

Mg disk with effective surface area of 22 cm2 is placed atthe anode of the extraction region. Energetic electrons inci-dent on the anode sputter the disk. Very high sputteringrate starts at Id = 2.5 A indicated by the change of plasmacolor from pink to green shown in Fig. 12. In this phase ofthe study, the low gas flow rate was no longer considereddue to sub-optimal H� yield established earlier using pureargon seeding. Fig. 13 presents the spatial variation of H�

signal, when both magnesium and argon are employed,measured at different discharge conditions. The seedingwas performed at high flow rate (18 sccm H2:2 sccm Ar)at a total gas feed of 2.7 mTorr. Maximum extractionoccurs at 3 A discharge occurring at the same 3 cm dis-

tance. The difference being an enhancement factor ofalmost 5 from 0.63 A/m2 to 3.09 A/m2. Discharge currentshigher than 3 A with lower H� yields are not shown in thefigure. As previously discussed, the 3 cm distance is indeedthe region favorable for the two-step H� formation. At thisoptimum distance, gleaning through the plasma parametersfor pure Ar (Fig. 6) and the dual Ar–Mg seeding (Fig. 14),the latter shows a significant increase in Ne by three ordersof magnitude (from 2.76 � 1012 m�3 to 2.9 � 1015 m�3).The average Teff increased slightly from 2.35 eV to3.03 eV. To exact a better comparison on the efficacy ofboth seeding processes, their DA reaction rates (K) at opti-mum distance and discharge for all possible vibrationalstates (m = 1–9) are estimated. Not surprisingly, one seesin Fig. 15, the reaction rate for the dual Ar–Mg seed tobe higher by about three orders of magnitude against thesingle Ar seed. All these suggest Mg to be an excellentsource of cold electrons.

Fig. 12. Change in plasma color due to Mg sputtering.

Fig. 13. Spatial variation of H� signal for Ar–Mg seeded plasmameasured at different discharge conditions.

Fig. 14. Spatial variation of plasma parameters for Ar–Mg seeded plasmameasured at Id = 3A. Increase in cold electron density by 3 orders ofmagnitude due to the presence of Mg.



Another notable observation in the H� profile (Fig. 13)is the recovery of the signal at the periphery. This two-peakfeature is also seen in the earlier argon-seeded plasma(Fig. 5) although not as pronounced as when Mg is incor-porated. Hydrogen molecules colliding with the chamberwalls bounce off with higher vibrational states [33]. Thepresence of abundant cold electrons (�7 � 1015 m�3) atthe edge increases DA reactions with these excited H2 mol-ecules, hence the observed second hump.

By locking the E � B probe again at 3 cm from the sheetcore and measuring the signal density at various Id cur-rents, it is further substantiated that Id = 3 A gives opti-mum yield shown in Fig. 16. Correlation analysis withplasma parameters (Fig. 17) under this condition showsthat the 3A discharge offers the most conducive environ-ment for maximum H� output whereby there exist the req-

uisite high Ne and low Teff, averaging about 3 eV. Raisingdischarge further simply increases the influx of hot elec-trons from the production region. It is plausible that fac-tors such as the incursion of hot electrons for Id > 3 Aand low electron densities for Id < 3 A bring about lowH� yield. The EEDF profiles of 2.5 A, 3 A and 3.5 A dis-charges shown in Fig. 18 give credence to these possibili-ties. Although all discharge conditions have mono-peakEEDF profiles, that of 2.5 A has a much lower particledensity, while that of 3.5 A shows enhanced density ofhigher temperature electrons. Low electron densities for2.5 A (Id < 3 A) inhibits H� formation. The buildup of

Fig. 15. Effects of argon alone and magnesium–argon seeding on thereaction rates of dissociative attachment process. Reaction rate for thedual Ar–Mg seeded plasma is higher by about three orders of magnitudeagainst the single Ar seeded plasma.

Fig. 16. H� current density measurements versus discharge currents forAr–Mg seeded plasma at 3 cm distance from the sheet core. Id = 3 A givesoptimum extraction.

Fig. 17. Variation of plasma parameters for Ar–Mg seeded plasmameasured at 3 cm from the sheet.

Fig. 18. EEDF profile for Ar–Mg seeded plasma at Id = 2.5 A, 3 A and3.5 A, 3 cm distance. Only the EEDF profile for Id = 3 A shows an energycharacteristic of 3 eV conducive for DA reactions. The EEDF of 2.5 Ashows low particle density while that of 3.5 A shows increased density ofhigh temperature electrons.



higher temperature electrons for 3.5 A (Id > 3 A) adverselyaffects H� extraction due to increased CD reactions.

Unlike in the pure argon seeding where increased Ar:H2

ratios translated to gains in H� (Fig. 10), the Ar–Mg seed-ing has an optimal ratio for maximum extraction seen inFig. 19 to within 10% Ar admixture only. Fig. 20 showsthe variation of Teff and Ne with respect to argon ratio.An increase in the proportion of Ar appears to have a cool-ing effect as Teff reduces from 4 eV to 1.5 eV. Ne howeverremains relatively constant judging by the error bars ofthe data points.

It is believed here that Ar plays a dual role, supplyingcold electrons for the enhancement process as well as pro-viding the appropriate plasma environment for Mg sputter-ing. The Mg disk attached to the anode is biased at 175 Vwith respect to the cathode. Fig. 21 shows the optical emis-sion spectra (OES) of the Ar + Mg + H2 discharge. A por-tion of the spectra is magnified in the inset. Compared withthe OES of the Ar + H2 discharge in Fig. 22, Mg lines aremore intense than either those of Ar or H. The strong Mgemissions suggest Mg atoms are sputtered from the disk.The ionization of Mg [34] releases additional electrons inthe volume which in turn improves negative ion density.In this experiment, the 10% argon content represents theideal proportion for maximum H� extraction. Exceeding

Fig. 19. Effect of Ar:H2 ratio on ion yield for Ar–Mg seeded plasma atoptimum conditions, i.e. Id = 3 A, 3 cm distance. Maximum extractionoccurs at 10% Ar content.

Fig. 20. Effect of Ar:H2 ratio on plasma parameters for Ar–Mg seededplasma at optimum conditions, i.e. Id = 3 A, 3 cm distance. An increase inthe proportion of Ar has a cooling effect as Teff reduces from 4 eV to1.5 eV. Statistically, Ne shows no significant variation.

Fig. 21. Typical optical emission spectra of the Ar + Mg + H2 discharge. The inset indicates Mg lines to be more intense than either those of Ar or H.

Fig. 22. Typical optical emission spectra of the Ar + H2 discharge.



this limit leads to the destruction of H� due to increasedcollisions with argon neutrals.

4. Conclusions and recommendations

The incorporation of a multicusp magnetic confinementat the production region in a magnetized sheet plasma ionsource has decreased the required discharge power to gen-erate a quiescent plasma for H� production. The filamentlifetime is enhanced because of the lower discharge condi-tions. A wider choice of operational plasma currents ismade possible. The addition of argon at 10%Ar–90%H2

balance yielded an increase by 2.42 times to 0.63 A/m2

optimum H� current density extracted at 3.0 cm from theplasma core and 3.0 A plasma current. The introductionof magnesium has further increased the H� current densityyield at the same position, discharge current, and gas ratioby 4.9 times from 0.63 A/m2 to 3.09 A/m2. Numerical con-structs of the EEDF profiles seem to suggest that an essen-tial condition for high H� current density in the MSPNIS isa substantial amount of cold electrons in the order of1012 m�3 or higher, characterized by thermal energiesapproximating 3 eV. Argon admixture has a twofold func-tion in the seeding process, first supplying low temperatureelectrons for the initial enhancement and second, offering asuitable plasma environment for Mg sputtering. The ioni-zation of Mg liberates additional cold electrons which fur-ther boosts the negative ion yield. The addition ofmagnesium increased the density of cold electrons witheffective electron temperature of about 3 eV by three ordersof magnitude from 2.76 � 1012 m�3 to 2.90 � 1015 m�3.Formed H� ions are however destroyed upon increasingthe argon content from the optimum (10%Ar:90%H2 attotal gas feed of 2.4 mTorr) due to increased stripping withthe neutrals. Further, although there are substantial gainsattained in the seeding processes due essentially to the cool-ing effects of the seeds, the spatial disparity in yield isattributable more to destruction rather than to production.

It is recommended that in addition to the calculation ofreaction rates for DA, the reaction rates for CD and otherrelevant molecular and atomic processes be obtained tofully understand the mechanisms of H� formation anddestruction in a magnesium–argon seeded hydrogenplasma. Furthermore, the identification of dominant rota-tional and vibrational states of H2 at various regions inthe volume is necessary to obtain the actual reaction rates.

References

[1] Y. Abate, H. Ramos, Rev. Sci. Instr. 71 (10) (2000) 3689.[2] Japan Atomic Energy Research Institute (JAERI) News, Applying

Fusion Technology to Semiconductors, FF No. 46, 10, 1999.[3] M. Wada, Thin Solid Films 316 (1998) 128.

[4] P. Franzen, H. Falter, B. Heinemann, Ch. Martens, U. Fantz, M.Berger, S. Christ-Koch, M. Froschle, D. Holtum, W. Kraus, S. Leyer,P. McNeely, R. Riedl, R. Suss, S. Obermayer, E. Speth, D.Wunderlich, Fusion Eng. Des. 82 (2007) 407.

[5] M. Kuriyama, N. Akino, T. Aoyagi, N. Ebisawa, N. Isozaki, A.Honda, T. Inoue, T. Itoh, M. Kawai, M. Kazawa, J. Koizumi, K.Mogaki, Y. Ohara, T. Ohga, Y. Okumura, H. Oohara, K. Ohshima,F. Satoh, T. Takenouchi, Y. Yokokawa, K. Usui, K. Watanabe, M.Yamamoto, T. Yamazaki, C. Zhou, Fusion Eng. Des. 39–40 (1998)115.

[6] J.R. Hiskes, A.M.J. Karo, J. Appl. Phys. 56 (1982) 1927.[7] K.N. Leung, K.W. Ehlers, M. Bacal, Rev. Sci. Instr. 54 (1983) 56.[8] Y. Okumura, Y. Fujiwara, A. Honda, T. Inoue, M. Kuriyama, K.

Miyamoto, N. Miyamoto, K. Mogaki, A. Nagase, Y. Ohara, K. Usui,K. Watanabe, Rev. Sci. Instr. 67 (1996) 1018.

[9] M.A. Leitner, K.N. Leung, Nucl. Instr. and Meth. A 427 (1999)250.

[10] K.N. Leung, K.W. Ehlers, R.V. Pyle, Appl. Phys. Lett. 47 (3) (1985)227.

[11] M. Bacal, A.A. Ivanov Jr., C. Rouille, M. Nishiura, M. Sasao, in:Proceedings of the 30th EPS Conference on Contr. Fusion andPlasma Phys., St. Petersburg, July 7–11, 2003, ECA, Vol. 27A, p.2.32.

[12] Y. Okumura, Y. Fujiwara, T. Inoue, K. Miyamoto, A. Nagase, Y.Ohara, K. Watanabe, Rev. Sci. Instr. 67 (3) (1996) 1092.

[13] Y. Belchenko, Y. Oka, M. Hamabe, O. Kaneko, A. Krivenko, Y.Takeiri, K. Tsumori, M. Osakabe, K. Ikeda, E. Asano, T. Kawam-oto, Rev. Sci. Instr. 73 (2) (2002) 940.

[14] T. Takanashi, T. Takeiri, O. Kaneko, Y. Oka, K. Tsumori, T.Kuroda, Rev. Sci. Intr. 67 (3) (1996) 1024.

[15] M. Tanaka, K. Amemiya, Rev. Sci. Instr. 71 (2000) 1125.[16] M. Bacal, M. Nishiura, M. Sasao, M. Wada, M. Hamabe, H.

Yamaoka, Rev. Sci. Instr. 73 (2002) 903.[17] D. Boilson, H.P.L. de Esch, R. Hemsworth, A. Krylov, P. Mass-

mann, R. Rada, L. Svensson, Fusion Eng. Des. 66–68 (2003) 609.[18] N.P. Curran, M.B. Hopkins, D. Vender, B.W. James, Plasma Sources

Sci. Technol. 9 (2000) 169.[19] K. Shinto, Y. Okumura, T. Ando, M. Wada, Jpn. J. Appl. Phys. 35

(1996) 1894.[20] Y.I. Belchenko, G.I. Dimov, V.G. Dudnikov, Nucl. Fusion 14 (1974)

113.[21] G.I. Dimov, Rev. Sci. Instr. 67 (1996) 10.[22] O. Fukumasa, S.J. Saeki, J. Phys. D: Appl. Phys. 20 (1987) 237.[23] K. Sato, S. Iizuka, Y. Nakagawa, T. Tsukuda, Appl. Phys. Lett. 65

(1994) 816.[24] H. Takahashi, T. Kasuya, M. Wada, Thin Solid Films 506 (2006)

575.[25] K.N. Leung, C.F.A. van Os, W.B. Kunkel, Appl. Phys. Lett. 58 (14)

(1991) 1467.[26] A. Mendenilla, H. Takahashi, T. Kasuya, M. Wada, Thin Solid Films

506–507 (2006) 527.[27] B.S. Park, J.S. Lee, K.R. Kim, B.H. Choi, Nucl. Instr. and Meth. B

261 (2007) 49.[28] M. Arciaga, A. Mendenilla, H. Ramos, Rev. Sci. Instr. 74 (2) (2003)

951.[29] M.J. Druyvesteyn, Z. Phys. 64 (1930) 781.[30] J.M. Wahedra, Phys. Rev. A 29 (1) (1984) 106.[31] R. Celiberto, R.K. Janev, A. Laricchiuta, M. Capitelli, J.M.

Wadehra, D.E. Atems, Atom. Data Nucl. Data Tabl. 77 (2001) 2.[32] M. Nishiura, M. Sasao, Y. Matsumoto, M. Hamabe, M. Wada, H.

Yamaoka, M. Bacal, Rev. Sci. Instr. 23 (2) (2002) 973.[33] J.R. Hiskes, A.M. Karo, et al., J. Appl. Phys. 53 (1982) 3469.[34] H. Tawara, T. Kato, Atom. Data Nucl. Data Tabl. 36 (1987) 167.




Optimized H − extraction in an argon–magnesium seeded magnetized sheet plasma

Documents