Yi-Ming
Chen
a,
Tai-Yuan
Chang
aaDepartmentofElectronicsEngineering,NationalChiaoTungUniversity,No.1001,UniversityRoad,Hsinchu300,Taiwan,ROC bDepartmentofElectronicsEngineering,ChungHuaUniversity,Hsinchu300,Taiwan,ROC
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received18August2013 Receivedinrevisedform 10November2013 Accepted12November2013 Available online 19 November 2013 Keywords: SiGe Nanowire Plasma Fluorine Nitrogen
a
b
s
t
r
a
c
t
TheimprovementstoelectricalpropertiesofSiGenanowiresbysurfaceplasmatreatmentwere investi-gated.Variousdurationsofpre-oxidationwithfluorine;ambientsforpost-nitridationplasmatreatment, andannealingtemperatureafterplasmatreatment,800–950◦C,wereapplied.Pre-oxidationtreatment usingfluorineplasma;improvedtheconductanceofSiGenanowiresbecausetheSi–Fbindingenergy cre-atedamorestableinterfacestatethanbarenanowireonthesurfaceofSiGe.N2plasmaincorporatedmore
NthandoesinNH3plasma,andNH3hasthedrawbackofintroducingelectrontraps,causingSi–Hbonds
tobreakinthesubsequentannealingprocess.Sincethereparationofsurfacedefectsbyplasmatreatment isvalid,thehighpost-annealingtemperaturetoreducedefectbyre-crystallizingcanbereduced.Hence, Gediffusionatlowpost-annealingtemperaturedidnotreducethehighconcentrationofGeattheSiGe nanowiresurfaces.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
The modification of the band structure of strain Si metal–oxide–semiconductor filed-effect transistors by in-plane tensilestrainis knowntoimproveelectronand holemobilities
[1,2]. However, the hole mobility enhancement is only slight, becauseinSi1−xGexmaterialcontainsonlyasmallfractionofGe
[3,4].Thevalueofxshouldbeaslargeas0.3toenhancemobility significantly[5].TheGecondensationprocesshasthebenefitof increasingtheGefractionbyoxidation,allowingamuchlowerGe fractiontobeusedinthestartingmaterial.Alow-Gefractionis expectedtoinhibittheformationofdislocationsinSiGe,because itassociatedwithalowlatticemismatchbetweentheinitialSiGe andSOIlayer[6–9].Thefreesurfaceofasemiconductoris well knowntohaveahighsurfacestatebecauseithasmanydangling bonds.AfreeSisurfacegenerallyhasasurfacedensityof approx-imately1015cm−2eV−1 [10],buttheinterfacedensityatSiO2/Si
canbemadelessthan1011cm−2eV−1.Thus,surfacepassivationis
veryimportantinsuppressingthedegradationofcarriermobility in SiGefilms. Apassivated SiO2 layer,thermally grown[11] or
deposited byPECVD[12], effectively reducesdensityof surface stateswhenSiGeundergoesoxidationprocess.Therefore,fluorine
∗ Correspondingauthor.Tel.:+88635731887;fax:+88635731887. E-mailaddress:[email protected](C.-F.Chen).
and nitrogenhave beenincorporated intointerfacial layers(IL) anddielectriclayerforhigh-materialhasbeenreported[13–16]. Hence,themaingoalofthisinvestigationistoelucidatetheeffect of fluorine and nitrogen incorporation on the surface of SiGe nanowirebyplasmatreatment.
2. Experiment
Theside-wallspacermethod[17]isthecheapestandeasiest methodforfabricatingananowiresensor,anditwasutilizedheein tofabricatethenanowiresamples.Ap-type(borondoped)Si sub-strate(100)witharesistanceof8–10cm−1wasused.Following aninitialstandardRCAcleaning,the500nm-thickoxidewasgrown ontheSisubstratebywetoxidation.Then,theoxidelayerwas pat-ternedbyopticallithographyanddryetching.Theremainingoxide thicknesswasapproximately300nm.This300nm-highoxidelayer servedasthebottomoxideinside-wallspaceretching.A200 ˚A-thickamorphousSi(␣-Si)layerwasdepositedbyLPCVDat650◦C anda2000 ˚A-thickSi0.86Ge0.14layerwasdepositedbyUHV-CVDat
655◦CafterRCAcleaningofthewafer.Self-alignedetchingwas car-riedouttoformside-wall-spacernanowires.Fig.1schematically depictsthreeSiGenanowiresamplesthathadundergonesurface treatments.Fig.1(a)presentsabareSiGenanowire,withoutany surfacetreatment;Fig.1(b)showstheSiGenanowirethathadbeen pre-treatedwithCF4plasma.Aftertheformationofthe␣-Si/SiGe
layerandthepre-plasmatreatment,thesamplesweretreatedin
0169-4332/$–seefrontmatter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.apsusc.2013.11.047
Fig.1.TheschematicsofSiGenanowiresurfacetreatmentprocess(a)the“bare”SiGenanowirewithoutplasmatreatment,(b)theSiGenanowirewithCF4plasmatreatment beforeGecondensationbyoxidation,(c)theSiGenanowirewithN2orNH3post-plasmatreatment,and(d)theSiGenanowirewasannealedat800–950◦Candthethermal
oxidewasremoved.
adilutedO2gasambientenvironmentfor3min.AftertheGe
con-densationbyoxidation,thepre-treatedSiGenanowirewastreated withN2 orNH3 plasmatreatmentthatisdisplayedin Fig.1(c).
Fig.1(d)showsaSiGenanowirethatwasannealedat800–950◦Cfor 180sinN2gasbeforethethermaloxidewasremoved.FollowingGe
condensationandannealing,thesenanowireswereimplantedwith 3×1015ions/cm2ofBF
2toformp-typenanowires.Thefinal
metal-lizationprocessinvolvedthedepositionofa500nm-thicklayerof aluminumfollowedbyAlsinteringat400◦C.Finally,theelectrodes weredefinedbythemaskprocess.AHewlettPackardHP4156A instrumentwasutilizedtomeasuretheelectricalcharacteristics ofthenano-wiresensor.Thedrainvoltage(VD)wasvariedfrom −10Vto10Vinstepsof500mV,andthebackgatevoltagewas fixedat0V.Theelectricalcharacteristicsweremeasuredateach stageofthesurfacemodification,andthemeanconductancewas thenextractedfromtheID–VDcharacteristicswithVD=3–6V.
3. Resultsanddiscussion
Fig. 2(a) plots the I–V characteristic of Si0.86G0.14 nanowire
with/withoutthe plasma treatment.The plasma conditionwas
CF4/N2=10s/90sandthepostannealingtemperaturewas950◦C.
Thecurrentinplasma-treatedsampleandtheslopeofitsI–Vcurve ofthesamplewerehigherthanthoseoftheun-treatednanowire sample.Fig.2(b)plotstheconductanceoftheSiGenanowire sam-plesthatwereshowninFig.2(a).Theincreaseinconductanceof theplasma-treatedsampleiscausedbythereductionofthe den-sityofsurfacedefectsbyGecondensation.Hanrathetal.foundthat
[18];Genanowireexhibitedp-typebehavior;andthatGetended toaccumulateholesatitssurfaceowingtoatrappednegative sur-facecharges.Accordingly,theI–VcurveofthebareSiGenanowire sampleshouldhave hada higherslopethan thoseoftheother SiGenanowiresamples.Thisdifferencewascausedbyenhanced surface doping; therefore, the scattering of carriers by ionized impuritiesandsurfacescatteringaremajorconcernsinrelationto conductancevariation[11,19].Fig.3plotstheeffectofpost anneal-ingtemperatureonconductanceafterplasmatreatment.Therate ofincrease of theconductancewithtemperaturedeclined with increasingtemperature,sincetheGeatomsdidnotdiffusefrom theSiGenanowiresurfacetotheLTSi-bufferlayer.Gecondensation causedGetopile-upatsurfaceoftheSiGenanowire,sothe con-ductanceandsensingsurfaceareaincreasedwiththeGefraction.
Fig.2.(a)I–VcurvesoftheSiGenanowireswith/withoutplasmatreatment,and(b) thecharacteristicofconductivityofSiGenanowireswith/withoutplasmatreatment.
Fig.3.Thecharacteristicofconductivityofplasma-treatedSiGenanowiresarein differentpostannealingcondition.
Therefore,hightemperatureannealingresultsinGere-distribution anddiffusionfromthesensingsurface.Fig.4plotsthevariationof theconductanceofSiGenanowireswithperiodoftreatmentby CF4plasmaatpostannealingtemperatureof800◦C.Atreatment
timeof20syieldedapeakvalueconductance.Whenthetreatment periodwas10s,itassumedthattheFatomswereafewtorepair thesurfacedefectsonSiGenanowire;treatmentperiodwas30s, theconductancewasdegradedbyplasma-induceddamage[20,21].
Fig.5(a)plotstheI–VcharacteristicsofSiGenanowirethat under-wentpost-plasmatreatmentinvariousambient.Thesamplethat
Fig.4. Thecharacteristicofconductivityofplasma-treatedSiGenanowiresarein differentCF4treatmenttimeperiod.
Fig.5. (a)I–VcurvesoftheSiGenanowireswithN2orNH3plasmapost-treatment,
and(b)thecharacteristicofconductivityofSiGenanowireswithN2orNH3plasma
post-treatment.
underwentN2plasmatreatmenthadagreaterconductioncurrent
thansamplethatwastreatedwithNH3plasma.Fig.5(b)compares
theconductanceofthenanowiresthataredescribedinFig.5(a).The N2-treated samplehasthehigherconductancebecauseit
incor-poratesmoreNatomsthantheNH3-treatedsample.WeakSi–H
bondsareeasilybrokeninthefollowingannealingprocessathigh temperature,casingthedanglingbondstoreappear.IntheNH3
Fig.6.ThecharacteristicofsensitivityofnanowireswhenAPTMSwasdrippedontoitssurface(a)thecomparisonofsensitivityisbetweenpoly-Sinanowire;unplasma-treated SiGenanowireare,andplasma-treatedSiGenanowire,(b)thecomparisonofsensitivityindifferentpre-plasmaandpost-plasmatreatmentcondition.
Fig.7.(a)TheTEMimageofSiGenanowiresafteroxidationwhichoxidelayeronSiGenanowirewasstripped,and(b)theEDAXatsurfaceofSiGenanowireafteroxidation.
aretrappedinn-typetraps[22–25],reducingitsconductance.The conductancevariedwiththeplasmatreatmentconditionsand sen-sitivityoftheSiGenanowiremayalsohavevariedinamanner similartotheconductance.Thenanowiressampleswere function-alizedusing3-aminopropyltri-ethoxysilance(APTMS) tomodify thesilicon oxidesurfacesthat surroundedthenanowires. Each
hydroxylfunctionalgroupontheoxidesurfacewasreplacedby methoxygroupsofAPTMSmodules,whilethenanowiresurfaces weresimultaneouslyterminatedby aminegroups.The conduc-tanceofthenanowiredecreasedwhenAPTMSwasdrippedonto itssurface. APTMS waspronetobecome positivelycharged,as reflectedbythedecreaseinconductancewhenholecarrierswere
sotheSiGenanowirewithlargerconductancehasgreater sensi-tivity.Fig.6(b)plotsthevariationofsensitivitywithbothbythe post-plasmatreatment ambientspecies and thepost-annealing temperature.Thesamplethatunderwent N2 post-plasma
treat-ment and post annealed at 800◦C had greater percentage of sensitivitythan theothernanowiresamples.Thisresultis con-sistent withthecharacteristicsplotted fromFigs. 3–5.Fig.7(a) displaysthe TEMcross-section of SiGenanowires. The average widthofthenanowiresafteroxidationis80nm.Fig.7(b)presents theEDAXatthesurfaceofaSiGenanowirefollowingoxidation. ThefractionofGeintheoxidizednanowireis18.25%,which is approximately4.73%higherthanthat13.52%,intheun-oxidized SiGenanowire.
4. Conclusions
Thisstudyexaminedvariousmethodsforincreasingthe con-ductance of SiGe nanowires in the Ge condensation process. Fluorineplasmatreatmentefficientlyimprovedtheconductance andenhancedthesensitivityofSiGenanowiresbecausethehigh Si–Fbindingenergycreatesamorestableinterfacestatethanbare nanowiresampleonthesurfaceofSiGe.Fatomscaneliminatethe danglingbondsintheGecondensationprocessandimprovethe conductance ofthe nanowires.Theperiod of CF4 plasma
treat-ment canbe controlledtopreventthe reappearance ofdefects by plasma damage and consequent reducing of the conductiv-ityof theSiGenanowire.Post-plasma treatmentinN2 ambient
isbetterthanthat inNH3 becausetheHatomsaremoreeasily
dissolvedbythesubsequenthightemperatureannealing.Weak Si–Hbondsareeasilybrokeninthefollowingannealingprocess athightemperature,casingthedanglingbondstoreappear.The temperatureofpost-annealingcanalsobecontrolledtoimprove
[1]D.K.Nayak,J.C.S.Woo,J.S.Park,K.L.Wang,K.P.MacWilliams,Appl.Phys.Lett. 62(1993)2853.
[2]J.Welser,J.L.Hoyt,S.Takagi,J.F.Gibbons,Tech.Dig.Int.ElectronDevicesMeet 947(1994)373.
[3]T.Mizuno,S.Takagi,N.Sugiyama,J.Koga,T.Tezuka,K.Usuda,T.Hatakeyama, A.Kurobe,A.Toriumi,Tech.Dig.Int.ElectronDevicesMeet.943(1999)934. [4]T.Mizuno,S.Takagi,N.Sugiyama,H.Satake,A.Kurobe,A.Toriumi,IEEEElectron
DeviceLett.21(2000)230.
[5]R.Oberhuber,G.Zandler,P.Vogl,Phys.Rev.B:Condens.Matter58(1998)9941. [6]A.R.Powell,S.S.Iyer,F.K.LeGoues,Appl.Phys.Lett.64(1994)1856. [7]K.Brunner,H.Dobler,G.Abstreiter,H.Schäfer,B.Lustig,ThinSolidFilms321
(1998)245.
[8]F.Y.Huang,M.A.Chu,M.O.Tanner,K.L.Wang,G.D.U’Ren,M.S.Goorsky,Appl. Phys.Lett.76(2000)2680.
[9]F.Y.Huang,Phys.Rev.Lett.85(2000)784.
[10]P.Zhang,E.Tevaarwerk,B.N.Park,D.E.Savage,G.K.Celler,I.Knezevic,P.G. Evans,M.A.Eriksson,M.G.Lagally,Nature(London)439(2006)703. [11]H.Yang,D.Wang,H.Nakashima,H.Gao,K.Hirayma,K.-i.Ikeda,S.Hata,H.
Nakashima,Appl.Phys.Lett.93(2008)072104.
[12]C.H.Lai,K.M.Chang,C.F.Chen,C.T.Hsieh,C.N.Wu,Y.B.Wang,C.H.Liu,Micro NanoLett.7(2012)729.
[13]G.Shang,P.W.Peacock,J.Robertson,Appl.Phys.Lett.84(2004)106. [14]K.Tse,J.Robertson,Appl.Phys.Lett.89(2006)142914.
[15]C.S.Lai,W.C.Wu,K.M.Fang,J.C.Wang,S.J.Lin,J.Appl.Phys.44(2005)2307. [16]K.M.Change,B.N.Chen,S.M.Huang,Appl.SurfaceSci.254(2008)6116. [17]Y.-K.Choi,T.-J.King,C.Hu,IEEETrans.ElectronDevices49(2002)436. [18]T.Hanrath,B.A.Korgel,J.Phys.Chem.B109(2005)5518.
[19]S.Zhang,E.R.Hemesath,D.E.Perea,E.Wijaya,J.L.Lensch-Falk,L.J.Lauhon,Nano Lett.9(2009)3268.
[20]V.S.Chang,C.C.Chen,C.L.Wu,D.Y.Lee,T.L.Lee,S.C.Chen,M.S.Liang,Int.Symp. PlasmaProcessInducedDamage130(2003).
[21]C.C.Chen,H.C.Lin,C.Y.Chang,C.C.Huang,C.H.Chien,T.Y.Huang,M.S.Liang, Int.Symp.PlasmaProcessInducedDamage121(2005).
[22]M.L.Green,E.P.Gusev,R.Degraeve,E.L.Garfunkel,J.Appl.Phys.95(2001)2057. [23]T.Hori,GateDielectricsandMOSULSI,Springer,Berlin,1997.
[24]L.D.Thanh,P.Balk,J.Electrochem.Soc.135(1998)1797.
[25]P.Rohana,I.Akihiro,H.Reiji,K.Yukinori,InternationalConferenceonProperties andApplicationsofDielectricMaterials,Nagoya,Japan,2003,p.1084. [26]K.M. Chang, J.M. Kuo, W.C. Chao, C.J.Liang, H.H. Wu, W.H. Tzeng, T.L.
Wu,IEEEInternationalNanoelectronicsConference,Shanghai,China,2008, p.811.
