Silicon recovery from silicon-iron alloys by electrorefining in molten fluorides

(1)

O

pen

A

rchive

T

OULOUSE

A

rchive

O

uverte (

OATAO

)

OATAO is an open access repository that collects the work of Toulouse researchers and

makes it freely available over the web where possible.

This is an author-deposited version published in :

http://oatao.univ-toulouse.fr/

Eprints ID : 8620

To link to this article : DOI: 10. 1016/j.electacta.2013.02.065

URL : http://dx.doi.org/10.1016/j.electacta.2013.02.065

To cite this version : Massot, Laurent and Bieber, Anne-laure and

Gibilaro, Mathieu and Cassayre, Laurent and Taxil, Pierre and

Chamelot, Pierre Silicon recovery from silicon-iron alloys by

electrorefining in molten fluorides. (2013) Electrochimica Acta,

vol. 96 . pp. 97-102. ISSN 0013-4686

Any correspondence concerning this service should be sent to the repository

(2)

Silicon

recovery

from

silicon–iron

alloys

by

electrorefining

in

molten

fluorides

L. Massot

∗

_{, A.L.}

_{Bieber, M.}

_{Gibilaro, L.}

_{Cassayre, P.}

_{Taxil, P.}

_Chamelot

UniversitédeToulouse;UPS,INP,CNRS;LaboratoiredeGénieChimique;DépartementProcédésElectrochimiques;118RoutedeNarbonne,31062Toulouse,France

Keywords: Moltenfluorides Silicon Electrorefining Si–Fealloys Voltammetry

a

b

s

t

r

a

c

t

Electrorefiningofasilicon-ironmaterial(Si–4.7at%Fe)inmoltenNaF-KFat850◦_C_was_investigated_in

viewofrecoveringpureSi,usingelectrochemicaltechniques,ScanningElectronMicroscopycoupled withEnergyDispersiveSpectroscopy(SEM-EDS)andInductivelyCoupledPlasmawithatomicemission spectroscopy(ICP-AES)analyses.TheselectiveelectrochemicaldissolutionofSiwasevidenced. Elec-trorefiningrunsledtoamaximumrecoveryof80%ofSiinitiallycontainedinthematerial,intheform ofadensedepositatthecathode,withveryhighcurrentefficiencies.TheSipuritywasexaminedandno FewasdetectedbyICP-AESanalysis:therecoveredSipuritywasassumedtobehigherthan99.99%.

1. Introduction

Photovoltaictechnologyisaveryimportantrenewablesource

ofelectricalenergyproductionwiththehighestannualgrowthrate

[1,2]andthemostcommonbasematerialforphotovoltaiccellis

SolarGradeSilicon(SoG-Si),witharequiredpurityof99.9999%.Due

toagrowingdemandforSoG-Si,thedevelopmentofnewprocesses

allowingtheproductionofcheaperSoG-Siisamainissueforthe

solarenergyindustry[3].Inthe1980’s,siliconelectrodepositionin

moltensaltshasbeenconsideredasanattractiveoptionforSoG-Si

production[4].However,theelectrolyticSiproductionprocesshas

notfoundanycommercialapplicationuptonow.Intheframeof

morerecentconcernsaboutcarbon-freeenergyproduction,such

kindofprocesshasneverthelessbeenreassessed.Thus,basic

inves-tigationsonSi(IV)reductionmechanismhavebeencarriedoutin

moltenfluorides[5–14]and more recently,thesaltsproperties

andsilicondepositionstrategieswereexaminedinourlaboratory

[15,16].TheseresultsconcludedthatSi(IV)ionsarereducedinto

Siinaone-stepprocessexchangingfourelectronscontrolledby

diffusioninthesolution.

Thefinalobjectiveofthestudyistoevaluatethefeasibilityof

usingelectrorefining processin moltenfluoridesas preliminary

stepof purificationofmetallurgicalgradeSi(MG-Si)withgood

quality.

Table1comparesthecompositionofMG-Siandthemaximum

admissibleimpuritycontentinSoG-Si.Thesedatashowfirstthat

∗ Correspondingauthor.Tel.:+33561558194;fax:+33561556139. E-mailaddress:[email protected](L.Massot).

oxygenandironarethemostimportantpollutantsofMG-Siand

then,that thehighesttolerance inSoG-Siis forcarbon, oxygen

andiron[17–19].Inthiswork,onlymetallicimpurities(Mi)were

takenintoaccount.Morespecifically,thepresentworkisfocused

ontheseparationofsiliconandironbyelectrolysisatlaboratory

scale,inamoltenfluoridemixture,throughthedissolutionofa

Si–Feanode,withtheoperatingconditionsdefinedinreferences

[15,16].Thestudywasdividedintothreeparts:(i)determinationof

theSi–Fedissolutionpathwaybyelectrochemicaltechniques

cou-pledwithSEMobservationsandEDSanalysis,(ii)silicondeposits

purityexaminationwithEDSandICP-AESanalysesand(iii)anodic

andcathodiccurrentefficienciesevaluationbyweightingthe

elec-trodes.

2. Experimental

2.1. Cell

Thecellwasavitreouscarboncrucibleplacedinacylindrical

vesselmadeofrefractorysteelandclosedbyastainlesssteellid

cooledbycirculatingwater.Thedescriptionofthiscellhasbeen

detailedinpreviouswork[20].Theexperimentswereperformed

underaninertargonatmosphere(Linde).Thecellwasheatedusing

aprogrammablefurnaceandthetemperaturewasmeasuredusing

a chromel-alumelthermocouple. Theelectrolyticbathconsisted

oftheNaF-KF(40–60mol%)eutecticmixture(CarloErba99.99%).

Themixturewasinitiallydehydratedbyheatingunder vacuum

(6×10−2_mbar)_from_ambient_temperature_up_to_its_melting_point

(3)

Table1

ComparisonbetweenimpuritiescontentinMG-Siandmaximumadmissiblecontent (ppm)inSoG-Si.

Impurities MG-Si SoG-Si(Maximumcontent)

B 20–50 <1 P 25–50 <1 O 3000 <10 C 10–250 <10 Fe 1000–3000 <10 Al,Ca 100–1500 <2

Othermetals(Ti,Cr, Mn,Ni,Cu,Zr...)

5–150 <1

theformofsodiumhexafluorosilicateNa2SiF6(AlfaAesar99.99%)

powder.

2.2. Electrodes

Iron, copper, nickel, titanium, silicon plate

(4mm×8mm×11mm) (Goodfellow 99.99%), molybdenum

(Goodfellow 99.95%), chromium, manganese and zirconium

(Goodfellow99.5%)wires(1mmdiameter)and95.3at%Si–4.7at%

Fealloy(8mmdiameter)wereusedaselectrodes.Thesurfacearea

oftheworkingelectrodewasdeterminedaftereach experiment

bymeasuringtheimmersiondepthinthebath.Theauxiliary

elec-trodes were silicon plate(4mm×8mm×11mm)and graphite

carbonrods(3mmdiameter–Mersenspectroscopicquality)with

a largesurfacearea(3cm2_)._A_platinum _wire_immersed _in_the

moltenelectrolytewasusedasaquasi-referenceelectrode.

2.3. Techniques

Theelectrochemicaltechniqueusedfortheanodicbehaviour

investigationofeachmaterialwaslinearvoltammetry,atascan

rateof10mVs−1_,_while_{electrorefining}_runs_were_carried_out_by

impositionofaconstantcathodiccurrent.Alltheelectrochemical

polarizationswereperformedwithanAutolabPGSTAT30

poten-tiostat/galvanostatcontrolledbya computerusingtheresearch

softwareGPES4.9.

The electrodessurface and cross-section werecharacterized

byScanningElectronMicroscope(SEM)coupledwithEnergy

Dis-persiveSpectroscopy (EDS)probe,afterresidualsalt removalin

HCl–AlCl3–H2Omixtureat50◦Candultrasonicwaves.

SiandFecontentofbothsaltanddepositsampleswere

mea-sured,afterdissolution,usingInductivelyCoupledPlasma-Atomic

EmissionSpectroscopy(ICP-AES,HORIBAUltima2).

3. Resultsanddiscussion

3.1. Operatingconditionsforsiliconelectrorefining

Electrorefiningisamethodappliedforpurifyingmetalsusingan

electrolysiscell.Itconsistsinananodicdissolutionofthematerial

tobepurifiedandadepositionofthetargetmetalwithout

impu-ritiesonthecathode. Threemainissueshavetobeconsidered:

theanodicbehaviourofthematerial,thestabilityofthedissolved

speciesinthemoltensaltandtheelectrodepositionconditionsat

thecathode.In thespecificcaseof Sielectrorefining, twosteps

(Si(IV)stabilityin thesalt andSideposition) havealreadybeen

investigatedinourlaboratory[15,16],asdetailedbelow,whilethe

Siselectivedissolutionremainstobedemonstrated.

ConcerningtheSi(IV)stability,ithasbeenshownthatinmolten

fluoridesalts,Si(IV)ionsaresolvatedasSiF4+xx−inequilibriumwith

gaseousSiF4,accordingtotheequilibrium(1).

SiF4+xx−= SiF_4(g)+xF− (1) -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 -1.5 -1.0 -0.5 0.0 0.5 1.0 j/ A cm -2 E vs.Pt/V Fe MnMn FeCr Ni Mo Zr Si Ti Cu

Fig.1.Comparisonoflinearvoltammograms(anodicpart)onvariousmetalsat 10mVs−1_in_NaF-KF_at₈₅₀◦

C;Aux.El.:vitreouscarbon;RefEl.:Pt.

Bieberetal.showsthatSiF4+xx−stabilityishighlyinfluenced

by thefluoroacidity (free F− _ions _content) _and _that _SiF

4+xx− is

more stable in basic molten baths: themost suitable one was

proventobetheeutectic NaF-KF(40–60mol%)mixture[15].In

ordertominimizeSilossesin thegaseous phase,thissalt

mix-ture(Tmelting=718◦C)hasthusbeenselectedfortheelectrorefining

runs.

IthasbeenstatedthattheelectrochemicalreductionofSi(IV)

ionsis aone-stepprocessexchanging4electrons,whateverthe

fluoridesaltcomposition,asindicatedinreaction(2).

Si(IV)+4e−

= Si (2)

Adetailednucleationandelectrodepositionstudyhasbeen

per-formed[16]andledtothefollowingconclusions:

-anoperatingtemperatureof850◦_C,_about₁₃₀◦_C_above_the

sol-ventmeltingpoint,allowstoenhanceSielectricalconductivity,

-graphiteorSoG-Sicathodicsubstratesensurethebestadherence

toSideposits,

-aSi(IV)concentrationhigherthan 0.6molkg−1 _inhibits

potas-siuminsertionintothedepositatacathodiccurrentdensityof

−50mAcm−2_.

TheseexperimentalconditionshavethusbeenappliedtotheSi

electrorefininginmoltenfluorides.

3.2. Impactofimpuritiesontheelectrorefiningselectivity

3.2.1. Preliminarydiscussion

Theselectivityoftheelectrorefiningprocessdependsonthe

dis-solutionpotentials(Ediss)ofthemetalstobetreated.Intheframe

ofelectrorefiningofSibasematerialscontaininganimpurityMi,

threecaseshavetobeconsidered:

-case1:EMi

diss>EdissSi .Theimpurityisnotdissolvedduringanodic

dissolutionofSi:theselectivityoftheprocessisthusgoverned

bytheanodepotential.

-case2:EMi

diss<EdissSi .TheimpurityisdissolvedtogetherwithSi

butisnotreducedatthecathodeduetoalowerreduction

poten-tial:theselectivityoftheprocessisensuredbythecontrolofthe

cathodepotential.

-case3:EMi

diss∼E Si

diss.Theimpurity is dissolvedtogetherwithSi:

completeelectrochemicalreductionbehaviourhastobe

stud-iedinordertoevaluatethepossibilityofitsco-depositionwith

(4)

Fig.2.SEMmicrographsofthecrosssectionoftheSi–4.7at%Feanode:(a)bulk;(b)externalsurface(lightgrey:FeSi2;darkgrey:Si).

Fig.3.Fe–Siphasediagram[9].

Theelectrochemicalbehaviourofseveralmetalsconsideredas

commonpollutant in SoG-Sihasbeen studiedby linearsweep

voltammetryinNaF-KFat850◦_C._Linear_{voltammograms}_recorded

at10Mvs−1_are_plotted_in_Fig.₁_.

Thethreecasespreviouslymentionedarehighlightedinthis

figure:Cu,Mo,Ni,CrandFecorrespondtocase1,Zrtocase2,Ti

andMntocase3.

Iron(case1)wasselectedasa“modelimpurity”duetoits

rel-ativelycloseoxidationpotentialcomparedtoSi(1E∼450mV).In

ordertodemonstratethefeasibilityofSi–Feseparationandto

eval-uatecurrentefficiencies,acommercialSi–4.7at%Fealloywitha

ratherhighFecontentwasused.

3.2.2. CharacterizationoftheSi–4.7at%Feanode

Themicrostructureoftheanodematerial,asillustratedinFig.2

bySEMmicrographsofitscross-section,iscomposedoftwophases:

asiliconmatrixandanintermetalliccompound,FeSi2,in

agree-mentwiththeSi–Fephasediagram(Fig.3)[21].Basedonavailable

densitydata((Si)=2.3gmL−1_and_(FeSi

2)=4.7gmL−1[22]),the

volumeoftheintermetallicphasewithinthematrixisabout11%.

-0.05 0 0.05 0.1 0.15 0.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 I/ A E vs. Pt/V Si Si-4.7at%Fe Fe

Fig.4. ComparisonoflinearvoltammogramsofSi,FeandSi–4.7at%Feat10mVs−1

inNaF-KFat850◦

(5)

Fig.5. SEMmicrographsoftheSi–4.7at%Feanodesurface:(a)beforeelectrolysisand(b)afterelectrolysisinNaF-KF-Na2SiF6(C=0.73molkg−1)during25hat50mAcm−2

at850◦

C(Q=8659C).

Fig.4exhibitstheanodicbehaviourofSi,FeandSi–4.7at%Fe

recordedbylinearsweepvoltammetryintheNaF-KFsystemat

850◦_C_and₁₀_mV_s−1_._This_figure_verifies_that_the_dissolution_of_Fe

occursatamorepositivepotentialthanSi.ConcerningtheSi–Fe

alloy,theshapeofthelinearvoltammogramcouldbeexplainedby

thetwo-phasestructureoftheanode:SiandFeSi2.Asobservedin

Fig.2,thesampleiscomposedofpureSimatrixwithFeSi2phase.As

theSioxidationpotentialislowerthanFeandaccordingtoNernst

law,onlythepureSiphase isfirstoxidized,explainingwhythe

equilibriumpotentialisthesamethanSisample.Then,aftershort

polarizationduration,SiisoxidizedfrombothpureSiandFeSi2

phases;andaccordingtoNernstlawforSiactivitylowerthan1,

theSioxidationrateislowered.

3.3. Sielectrorefiningprocess

TheoptimizedexperimentalconditionsforSi electrorefining,

remindedinSection3.1,havebeenappliedtotheSi–4.7at%Fealloy.

3.3.1. Si–Feanodedissolutionpathway

Theselectiveanodicdissolutionwasdemonstratedfirst.

Elec-trolysisrunswereperformedinNaF-KF-Na2SiF6 (0.73molkg−1)

system,at850◦_C_and_a_current_density_of₅₀_mA_cm−2_during₂₅_h,

andFig.5showsSEMmicrographsoftheSi–Feanodesurfacebefore

(a)andafter(b)electrolysis.Beforeelectrolysis,bothSiandFeSi2

phasesareobservedwhereas,afterelectrolysis,theresidual

struc-turehasa“sponge-like”shapeandiscomposedofaround94.2at%

ofFe.

Magnificationsof the anode cross sections are presented in

Fig.6,forshort(5h)andlonger(25h)electrolysisdurations.They

indicate the selective dissolution of silicon: for short duration

(Fig.6a),thepuresiliconphaseisremovedwhiletheFeSi2phaseis

depletedinSiattheanode-bathinterface,formingFeSiandaphase

containing27.5at%ofSi.Forlongerelectrolysis(Fig.6b),the

sili-conmatrixismoredeeplydissolved(∼600mm),whileSi–Fephases

remaininthisarea.

Thecompositionprofileoftheanodeafteranelectrorefining

runof 25hwasinvestigated (seeFig.7).This profileevidences

thedecreaseoftheSicontentinintermetallicphasesfrom66.5

to27.5at%,fromthecoretotheelectrodesurface,exhibitingthe

followingdissolutionbehaviour:

-thepureSiphasewasdissolvedupto600mm.

- SiintheFeSi2phasewasselectivelyremovedtotheresidualvalue

of5.8at%,throughtheformationofFe-richcompounds

previ-ouslymentioned:FeSiandFe–27.5at%Si,inagreementwiththe

dissolutionofasolidsolution[23].

3.3.2. Sidepositspurity

Fig. 8 shows silicon coatings obtained after electrorefining

experimentsontwocathodes: SoG-Si(a)andgraphite(b),after

residual salt removal. TheSi deposits were adherentand their

roughnesswasrelativelyhigh.

EDSanalysisindicatesthatthepurityofsilicondepositishigher

than99.9%,correspondingtothedetectionlimitoftheapparatus

forelementstitration(seeEDSspectrumofFig.9).Formore

accu-ratequantificationofFe,theSicoatingwasdissolvedintoKOH

15molL−1 _and _Fe_was_analysed_by _ICP-AES_in_this _solution:_no

Fewassignificantlydetected,meaningthattheFeconcentration

islessthanthedetectionlimitwhichisaround10mgL−1.TheSi

purityisthenbetterthan99.99%.Ironconcentrationinthesaltwas

Fig.6. SEMmicrographsoftheSi–4.7at%FeanodecrosssectionafterelectrolysisinNaF-KF-Na2SiF6(C=0.73molkg−1)at50mAcm−2and850◦C:(a)duration:5h(Q=2448C)

(6)

Fig.7.ConcentrationprofileofSiintheSi–FeintermetallicphaseafterelectrolysisinNaF-KF-Na2SiF6(C=0.73molkg−1)at50mAcm−2and850◦C;duration25h;inset:SEM

observationoftheanodeafterelectrolysisshowingthelocationofSianalysis.

Fig. 8.Photographs of cathodes after electrolysis runs in NaF-KF-Na2SiF6

(C=0.73molkg−1₎_at₅₀_mA_cm−2 _at ₈₅₀◦

C:(a)SoG-Si cathodeafter 39h;(b) graphitecathodeafter25h.

alsotitratedaftertheelectrolysisandwasnotdetected,proving

thattheSidissolutionisselective.

3.3.3. Currentefficiencies

Anodicandcathodiccurrentefficiencies,anodicdissolutionand

SirecoveryprogresseswereestimatedbycomparingthefinalSi

Fig. 9. EDS analysis of cathodes after electrolysis runs in NaF-KF-Na2SiF6

(C=0.73molkg−1₎_at₅₀_mA_cm−2 _at ₈₅₀◦

C:(a)SoG-Si cathodeafter 39h;(b) graphitecathodeafter25h.

massofanodeandcathodetotheinitialmassofSiinSi–4.7at%Fe

anodes.Thesevalueswerecalculatedaccordingto:

Anodicdissolutionprogress= 1mdissolved

mSi into Si–Feanode

(3)

1mdissolved (g)isthemassofdissolvedSievaluatedfromthe

massbalanceoftheanode,andmSiintoSi–Feanode (g)isthemass

ofsiliconinitiallypresentintheanode.Thismasswascalculated

bymeasuringtheinitialanodemassandcorrecting itbytheSi

composition(90wt%Si).

Sirecoveryprogress= mSi deposited

mSi intoSi–Feanode

(4) m_Si_depositedisthesiliconmassdeposited.

Table2gathersthevaluesobtainedinseveralelectrorefining

runs,leadingtothefollowingconclusions:

Table2

AnodicSidissolutionandcathodicSirecoveryprogressesandanodicandcathodicfaradicefficienciesofelectrorefiningrunsperformedinNaF-KF-Na2SiF6(C=0.73molkg−1)

at50mAcm−2_and₈₅₀◦ C. Q(C) Anode:Siinitial mass(g) Anodedissolved mass(g) Dissolution progress Anodicfaradic efficiency Cathode:Si depositedmass(g) Sirecovery progress Cathodicfaradic efficiency 1 2448 1.851 0.178 10% 89% 0.205 11% 100% 2 5480 1.028 0.360 35% 98% 0.399 39% 92% 3 8659 1.095 0.548 50% 90% 0.648 59% 90% 4 10,391 1.076 0.646 60% 90% 0.716 67% 94% 5 13,622 1.111 0.878 79% 94% 0.989 89% 99% 6 22,248 1.755 1.457 83% – 1.619 92% 100% 7 23,425 1.818 1.636 90% – 1.809 99% 100%

(7)

-the faradic efficiencies are close to 100% and the difference

tounityislikely attributedtomateriallossduringfrozensalt

removalfromelectrodes,

-forexperiments6and7,theanodefeltintothebathbeforethe

endoftherun.Thisindicatesthatthelimitofdissolutionprogress

isaround80%,duetotheanodestructure(11vol%Fe).Athigher

dissolutionrate,theanodesbecometoobrittletoensurecohesive

structure,

-thecathodicefficiency(Sirecovery)reaches99%.

4. Conclusion

Thisworkdemonstratedthatelectrorefininginabasicmolten

fluoridesolventisanoptiontobeconsideredforpuresilicon

recov-eryfromsiliconmaterialscontainingFe.

Previous studiesstated onthe bestoperating conditions for

electrodeposition of Si: solvent (eutectic NaF-KF), temperature

(850◦_C),_Si(IV)_{concentration}_(>0.6_mol_kg−1₎_and_current_density

(50mAcm−2_).

ElectrorefiningrunsevidencedthatelectrodissolutionofSifrom

aSi–Feanodeinvolveda selectiverelease ofSiwithinthe

elec-trolyte,aremainingFerich“skeleton”structureanda complete

recoveryofpuresiliconatthecathode.Itislikelythat,inthecase

ofmaterialswithlowerinitialFecontents,theremainingFerich

phasewillnotformasolidstructurebutratherfalloffandsediment

inthebottomofthecell.

Moreover,theanodicandcathodiccurrentefficienciesareclose

to100%andtheSidepositspurityisassumedtobehigherthan

99.99%,asnoFewasdetectedbyICP-AESanalysisofthedeposit.

Theelectrorefiningmethod,demonstratedinthisworkinthe

specificcase ofFe, is mostprobably suited for thepurification

ofSimaterialscontainingTi,Mn,Cr,Ni,MoandCusinceitwas

shownthat intheNaF-KF mixturetheirdissolution potentialis

moreanodicthanFe.

References

[1]M.A.Green,Recentdevelopmentsinphotovoltaics,SolarEnergyMaterialsand SolarCells76(2004)3.

[2]A.Müllera,M.Ghosha,R.Sonnenscheinb,P.Woditscha,Siliconforphotovoltaic applications,MaterialsScienceandEngineering:B134(2006)257. [3]S.Pizzini,Towardssolargradesilicon:challengesandbenefitsforlowcost

photovoltaics,SolarEnergyMaterialsandSolarCells94(2010)1528. [4]D.Elwell,R.S.Feigelson,Electrodepositionofsolarsilicon,SolarEnergy

Mate-rialsandSolarCells6(1982)123.

[5]K.H.Stern,S.M.McCollum,Electrodepositionofrefractorymetalsfrommolten salts,ThinSolidFilms124(1985)129.

[6]U.Cohen,R.A.Huggins,Siliconepitaxialgrowthbyelectrodepositionfrom moltenfluorides,JournaloftheElectrochemicalSociety123(1976)381. [7]G.M.Rao,D.Elwell,R.S.Feigelson,ElectrowinningofsiliconfromK2SiF6–

moltenfluoridesystems,JournaloftheElectrochemicalSociety127(1980) 1940.

[8] G.M.Rao,D.Elwell,R.S.Feigelson,Electrodepositionofsiliconontographite, JournaloftheElectrochemicalSociety128(1981)1708.

[9] I.G.Sharma,T.K.Mukherjee,Astudyonpurificationofmetallurgicalgrade sil-iconbymoltensaltelectrorefining,MetallurgicalTransactionsB17B(1986) 395.

[10]J.DeLepinay,J.Bouteillon,S.Traore,D.Renaud,M.J.Barbier,Electroplating sili-conandtitaniuminmoltenfluoridemedia,JournalofAppliedElectrochemistry 17(1987)294.

[11]D.Elwell,G.M.Rao,MechanismofelectrodepositionofsiliconfromK2SiF6

-flinak,ElectrochimicaActa27(1982)673.

[12]K.L.Carleton,J.M.Olson,A.Kibbler,Electrochemicalnucleationandgrowthof siliconinmoltenfluorides,JournaloftheElectrochemicalSociety130(1983) 782.

[13]D.Elwell,Electrowinningofsiliconfromsolutionsofsilicainalkalimetal fluo-ride/alkalineearthfluorideeutectics,SolarEnergyMaterialsandSolarCells5 (1981)205.

[14]R.Boen,J.Bouteillon,Theelectrodepositionofsiliconinfluoridemelts,Journal ofAppliedElectrochemistry13(1983)277.

[15] A.L.Bieber,L.Massot,M.Gibilaro,L.Cassayre,P.Chamelot,P.Taxil,Fluoroacidity evaluationinmoltensalts,ElectrochimicaActa56(2011)5022.

[16]A.L.Bieber,L.Massot,M.Gibilaro,L.Cassayre,P.Chamelot,P.Taxil,Silicon electrodepositioninmoltenfluorides,ElectrochimicaActa62(2012)282. [17] J.R.Davis,A.Rohatgi,R.H.Hopkins,P.D.Blais,P.Rai-Choudhury,J.R.Mccormick,

H.C.Mollenkopf,Impuritiesinsiliconsolarcells,IEEETransactionsonElectron Devices27(1980)677.

[18]A.Luque,S.Hegedus,HandbookofPhotovoltaicScienceandEngineering,J. WileyandSonsLtded.,Chichester,2003,Chap.V.

[19]F.Schmid,D.B.Joyce,Recentadvancesinupgradingmetallurgicalgradesilicon forsolargradesiliconfeedstock,in:Proceedingof24thEUPVSEC,Hamburg, German,2009.

[20] L.Massot,P.Chamelot,P.Taxil,Cathodicbehaviourofsamarium(III)inLiF–CaF2

mediaonmolybdenumandnickelelectrodes,ElectrochimicaActa50(2005) 5510.

[21]O. Kubaschewski,Iron-BinaryPhaseDiagrams,Springer-Verlag,NewYork, 1982.

[22] FACTPuresubstancedatabase,http://www.factsage.com/

[23]L.Cassayre,P.Chamelot,L.Arurault,L.Massot,P.Palau,P.Taxil,Electrochemical oxidationofbinarycopper–nickelalloysincryolitemelts,CorrosionScience49 (2007)3610.