Sensitivity of photoelectron energy loss spectroscopy to surface reconstruction of microcrystalline diamond films

(1)

ContentslistsavailableatSciVerseScienceDirect

Applied

Surface

Science

jo u rn a l h om epa g e :w w w . e l s e v i e r . c o m / l o ca t e / a p s u s c

Sensitivity

of

photoelectron

energy

loss

spectroscopy

to

surface

reconstruction

of

microcrystalline

diamond

ﬁlms

Denis

G.F.

David

a

_,

_{Marie-Amandine}

_{Pinault-Thaury}

b

_,

_Dominique

_Ballutaud

b

_,

Christian

Godet

c,∗

a_Instituto_de_Física,_Universidade_Federal_da_Bahia,_Campus_{Universitário}_de_Ondina,_40.210-340_Salvador,_Bahia,_Brazil b_GEMaC,_UMR₈₆₃₅_–_CNRS,₁_place_Aristide_Briand,₉₂₁₉₅_Meudon_Cedex,_France

c_Physique_des_Surfaces_et_Interfaces,_Institut_de_Physique_de_Rennes_(CNRS_UMR_6251),_Université_Rennes_1,_Beaulieu_–_Bât._11E_–₃₅₀₄₂_Rennes,_France

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received12November2012

Receivedinrevisedform1February2013 Accepted19February2013

Available online 27 February 2013 Keywords:

XPS

Electronenergyloss Plasmon

Diamond

Surfacereconstruction

a

b

s

t

r

a

c

t

InX-rayPhotoelectron Spectroscopy(XPS),binding energiesandintensitiesofcorelevelpeaksare

commonlyusedforchemicalanalysisofsolidsurfaces,aftersubtractionofabackgroundsignal.This

backgroundduetophotoelectronenergylossestoelectronicexcitationsinthesolid(surfaceandbulk

plasmonexcitation,interbandtransitions)containsvaluableinformationrelatedtothenearsurface

dielectricfunctionε(ω).Inthiswork,thesensitivityofPhotoelectronEnergyLossSpectroscopy(PEELS)

isinvestigatedusingamodelsystem,namelythewell-controlledsurfacereconstructionofdiamond.

Boron-dopedmicrocrystallinethinﬁlmswithamixtureof(111)and(100)preferentialorientations

werecharacterizedintheas-grownstate,withapartiallyhydrogenatedsurface,andafterannealingat

1150◦Cinultrahighvacuum.Afterannealing,thebulk(␴+␲)plasmonofdiamondat34.5eVisweakly

attenuatedbutnoevidenceforsurfacegraphitizationisobservednear6eV,asconﬁrmedbyelectronic

properties.Unexpectedfeatureswhichappearat10±1eVand19±1eVintheenergylossdistribution

arewelldescribedbysimulationofsurfaceplasmonexcitationsingraphite-likematerials;alternatively,

theyalsocoincidewithexperimentalinterbandtransitionlossesinsomegraphenelayers.This

compar-ativestudyshowsthatthePEELStechniquegivesaclearsignatureofweakeffectsinthediamondsurface

reconstruction,evenintheabsenceofgraphitization.ItconﬁrmsthesensitivityofPEELSacquisitionwith

standardXPSequipmentasacomplementarytoolforsurfaceanalysis.

1. Introduction

InX-rayPhotoelectronSpectroscopy(XPS)studies,kinetic

ener-giesandintensitiesofcorelevelphotoelectronsemittedfromthe

differentatoms locatedinthe subsurfaceregion arecommonly

usedforchemicalanalysisofsolidsurfaces,aftersubtractionofa

backgroundsignalarisingfromenergylossestoelectronic

excita-tionsinthesolid[1–4].Duringtheirtransportandescapethrough

thesolidsurface,photoelectronsexperienceelasticandinelastic

interactionswhichoccurbothinthebulkandatthesurface;such

extrinsicenergylossesrelatedtointerbandtransitionsand

plas-monexcitationsinduce,respectively,atailingoftheprimarycore

levelpeakoverseveraleVand abroadenergydistributionover

severaltensofeV,onthelowkineticenergyside[1–6].Hencethe

energylossdistributioninPhotoelectronEnergyLossSpectroscopy

∗ Correspondingauthorat:EPSI–IPR(Bât.11E–Beaulieu),UniversitéRennes1, 35042RENNES,France.Tel.:+33223235706;fax:+33223236198.

E-mailaddress:christian.godet@univ-rennes1.fr(C.Godet).

(PEELS)containsvaluableinformationrelatedtothenearsurface

dielectricfunctionε(ω).

ItisemphasizedthatXPSandPEELSacquisitionsareperformed

inthesamerunandatthesamelocationofthesample(although

withpossiblydifferentenergyresolutionandsignal-to-noiseratio)

incontrastwithstudiescombiningXPSandreﬂectionEELS(REELS).

Inaddition,angularPEELSanalysiscanbereadilyperformedusing

conventionallaboratoryXPSspectrometers,whileforspeciﬁc

stud-ies,synchrotronlightsourcesprovidebetterspatialandspectral

resolutions.Thephysicsandsurfacesensitivityofplasmonlosses

inPEELSandREELSaresimilarinprinciple,exceptforthepresence

oftheelectron–holeinteractionandthelackofacollimatedbeam

inphotoelectronspectroscopy.However,theREELStechniqueuses

aprimaryelectronbeamgeneratedbyaﬁeldemissiongunwhich

maydamagefragilesamples.

Thisworkaims atassessing thesensitivity ofPEELS usinga

modelsystem,namelythewell-controlledsurfacereconstruction

ofdiamondﬁlmsuponannealinginultrahighvacuum.Diamondis

awidebandgapsemiconductorwhichhasbeenextensivelystudied

foritsoutstandingrobustnessandsurfaceelectronicproperties.A

truenegativeelectronafﬁnityisfoundforthehydrogenatedsurface

(2)

[7]makingdiamondanefﬁcientphotoemitterwhileafairlyhigh

p-typesurfaceconductivityisobservedafterairexposure[8].The

negativeelectronafﬁnityisrelatedtosurfaceC Hdipoleswhile

thehighsurfacedensityofholesisexplainedbyelectrontransfer

fromthediamondvalencebandtothedeeperchemicalpotential

intheadsorbedwaterlayercontainingprotons,screenedbywater

moleculesandtheircorrespondingcarbonateanions[8,9].

Photoelectron spectroscopy studies have been performed in

order to characterize the electron afﬁnity and work function

of hydrogen-terminated diamond surfaces, for both (111) and

(100) orientations[10]. XPS and UPS were used to study the

partialhydrogencoverageandthevarietyofoxygencontaining

chemical functionalitiesin the as-grownstate,after intentional

surfacehydrogenationandafterliquidphaseorgasphase

oxida-tion[11–14].Alternatively,HighResolutionEELShasbeenusedto

followchangesinsurfacechemicalbondsafterhydrogenationor

annealingsteps[15].

Energylossspectroscopyofphotoelectronsemittedfromthe

C1scorelevelhasbeenusedpreviouslyeitherforinvestigationsof

theroleofsurfacehydrogenandoxygenatomcoverage[11,12]as

wellasforthestudyofseedingandgrowthofnano-and

micro-crystallinediamondﬁlms[16–19].Fortheassessmentofthenon

diamondresidualphase,theratioofthebulk(␴+␲)plasmonloss

ofthediamondphasetothemainC1sline(characteristicofthe

near-surfaceovertypically5nm)wastentativelycorrelatedtothe

ratioofGandDlinesinRamanspectra(characteristicofabulk

depth>100nm)[17].

Surfacereconstructionofdiamondﬁlmshasbeenwidely

stud-ied,bothexperimentally[20,21]andtheoretically[22,23].Some

atomistic modelshave beenproposedto takeinto accountthe

incompleteremovalofoxygenatomsalongwiththesp3_to_sp2

car-bonatomconversion.Onboth(100)and(111)surfaces,whichare

thedominantorientationsinourmicrocrystallinediamondﬁlms,

removalofterminating Hatoms leadstoa 2×1reconstruction

(dimerisationoftwodanglingbondsonCatoms,formingaC C

bond)[23].

Previousstudiesshowthatahigherannealingtemperatureis

requiredformicrocrystallinediamondreconstruction[14]as

com-paredwithnanodiamondﬁlms [15,24]. In this work,annealing

conditionsat1150◦CinUltraHighVacuum(UHV)werechosen

toobtaincompletedesorptionofhydrogenatthediamond

sur-facewhichleadstosomesurfacereconstructionofcarbonatom

bonding,asshown previously[14];in theabsenceofextensive

graphitization,suchweaksurfacechangesare notdetectablein

bulkRamanspectraandmakethisreconstructionprocesssuitable

forassessingthesensitivityofPEELSanalysis.Interpretationofthe

plasmonlossfeaturesexperimentallyobservedinPEELSdifference

spectraistentativelyperformedbycomparisonwithsurfaceand

bulkplasmonlossfunctionscalculatedusingthetabulated

dielec-tricfunctionsofsinglecrystaldiamond[25]andgraphite[26].

2. Experimental

2.1. Diamondﬁlmssynthesisandcharacterization

Microcrystallineborondopeddiamondﬁlms([B]=1019_cm−3₎

were deposited on silicon substrates by hot ﬁlament chemical

vapordecompositionofhydrocarbons.Thesampleswerekeptin

air.Theas-grownsamplesweresubmittedtoannealunderUHV

at1150◦Cduring6h.Whileaonehourannealingat850◦C(under

10−9Torr)issufﬁcienttoremovemostofhydrogenatoms

cova-lentlybonded to the surface [7],annealing at1150◦C leadsto

completeeffusionofbothsurfaceandbulkhydrogen.

Thediamond structure ofthesamples wascharacterizedby

Ultraviolet (UV) Ramanspectroscopy (laserexcitation: 325nm)

beforeandafterannealing.TheXPSmeasurementswerecarried

outinaVG220iXLsystem,withabasepressureof5×10−10Torr,

usingamonochromatizedAlK␣(1486.5eV)X-raysourcewitha

passenergy of 20eV (analyzer resolution 0.2eV). Energylevels

werecalibratedwithaAusinglecrystal.Thespectrawereprocessed

usingtheVGEclipseDatasystemsoftware,usingVoigtproﬁlesanda

Shirleybackgroundcontribution[1]whichisincludedintheﬁtting

process.

2.2. PEELSanalysis

Incarbonbasedmaterials,theC1scorelevelpeakatabinding

energyEB≈285eVisfollowedbyastructuredbackground

extend-ing toward higher binding (lower kinetic)energies. This broad

lossspectrumcorrespondstoC1sphotoelectronsthat have

suf-feredenergylossesontheirwaytothesamplesurface(andacross

thesample surface)and itis thus characteristicfor thesample

underinvestigation.Plasmonlossesareadirectconsequenceofthe

dielectricresponseoftheﬁlmtotheexternalelectromagnetic

radi-ation.Collectiveexcitations(plasmons)runaslongitudinalcharge

densityoscillationsthroughthevolumeofthesolidandalongits

surface.

Thesensitivityofsurfaceandbulkplasmonexcitationto

sur-facereconstructionofdiamondthinﬁlmsisinvestigatedbyPEELS,

overtheenergyrange0–90eV.Thezerooftheenergylossscale

istakenattheenergyoftheC1speak(zero-losspeak)maximum

andthespectraarenormalized toa commonheightofthis

so-calledzero-losspeak.Fermilevelchangesarethuscompensated

bythePEELSanalysisprocedure.Aconstantbackground

subtrac-tionisperformedforsettingtozerothehighkineticenergyside

oftheC1speak.Sincethisstudyisperformedwitha

monochro-matized X-raysource, subtraction of the satellite signal is not

required.

Single(34.5eV)andmultiplebulkplasmonexcitationsof

dia-monddominateatlargelossenergies,whilesurfaceexcitations

andinterbandtransitionsareexpectedtodominatetheenergy

loss distribution at lower energies. Note that loss features at

very low energies are difﬁcult toobserve in PEELSdue tothe

broad C1s line resulting from a variety of chemical

environ-ments (ether, carbonyl) of C atoms. This is the case, e.g. for

transitionsfromvalencebandtounoccupiedsurfacedefect

lev-elswithinthebandgap,previously observedbyEELSnear2eV

[27].

Apracticalmethodhasbeenproposed[28]toderivethesingle

plasmonloss distributionIm[−1/ε(ω)]from XPSdata,

follow-ingthetechniquedeveloped byEgerton[6]and Werner[5]for

ElectronEnergyLossSpectroscopy(EELS).SuchanalysisofPEELS

experimentalspectraincludesdeconvolutionofmultipleplasmon

lossesandseparationofbulkvssurfaceplasmonexcitations.

Care-fulremovalofsingle-electronscattering(e.g.interbandtransitions)

atlowlossenergyisaprerequisitesteptoderivethesingleplasmon

lossdistribution.

Inrecentwork[28],severalmethodswerecomparedtoseparate

interbandtransitionsfrombulk orsurfaceplasmonsexcitation,

takingamorphoussiliconasawell-knownreferencematerial.In

diamondﬁlms,theanalysisismorecomplicatedbecauseinterband

transitionsoccurathigherenergies(ω<25eV)withapeakaround

12eVinthedielectricfunctioncalculatedfromtabulatedoptical

indexdata[25].Sinceitdeservesmoredetaileddevelopments,the

dielectricfunctionobtainedfromPEELSdatausingourinversion

algorithm willbereported elsewhere[29].Hence, this studyis

focusedonqualitativeinformationderivedfromthedifferencein

energylossfunctions,beforeandafterannealing,inthelossenergy

range(smallerthan25eV)wheremultiplescatteringeventscanbe

(3)

Fig.1.Scanningelectronmicroscopyimageofthemicrocrystallinediamondﬁlm grownonaSi(100)substrate,withamixtureof(111)and(100)preferential orientations(pyramidsandbiplanes,respectively).

3. Experimentalresults

Hydrogenation/thermal desorption experiments have been

repeatedseveraltimesinordertoperformelectrochemical

charac-terizationswithmicrocrystallinediamondelectrodes,withagood

reproducibilityoftheirsurfacechemistryandelectronicproperties

afterhydrogenationandannealing.ThePEELSresultspresented

here correspondto a typical microcrystalline diamond sample,

whichhasbeencharacterizedindetailinapreviouspublication

(Ref.[14]).ScanningElectronMicroscopy(SEM) imagesof2␮m

thickmicrocrystallinediamondﬁlmsshowanaveragegrainsize

of0.5␮mwithmainly(100)and(111)texture(Fig.1).Secondary

IonMassSpectrometry(SIMS)proﬁlesexhibituniformbulk

hydro-gencontent(1×1020_at._cm−3₎_which_is_removed_after_UHV_anneal

(1150◦C).

3.1. Raman

UV Ramanspectroscopy (laser excitation:325nm) doesnot

showobviouschangesinthebulkdiamondstructureafterUHV

annealingat1150◦C(Fig.2).Comparisonoftheintensityofthe

1332cm−1diamondlinewiththeintensityofthebroadDandG

1000 1500 2000 2500 3000 3500 0 1 2 3 4 5

G (1565 cm

-1

)

D (1350 cm

-1

)

Diamond

(1332 cm

-1

)

Raman UV (325 nm)

Raman scattering intensity (arb. units)

Wavenumber (cm

-1

)

As-grown Annealed

Fig.2.UVRamanspectraofmicrocrystallinediamond,as-grownandafterthermal annealing(1150◦_C)_in_UHV.

290

289

288

287

286

285

284

283

282

0

1

2

3

As G_Diamondrown

a)

XPS intensity (10

4

CPS)

Bind

ing

en

erg

y (eV)

C ex C env C1 C2 C-O-C COOH

290

289

288

287

286

285

284

283

282

0

1

2

3

_Annea_led

b)

1150°C

XPS intensity (10

4

CPS)

Bind

ing

en

erg

y (eV)

C ex C env C1 C2 C-O-C COOH

Fig.3.DecompositionoftheXPSC1sspectraofmicrocrystallinediamond,as-grown (a)andafterthermalannealing(1150◦C)inUHV(b).

linesat1350and1565cm−1showsthatcarbonsp2_bond

concen-trationinthebulkislessthan1%[30].Suchsp2_graphitic_defects

maybepresentatmicrocrystallinediamondinternalsurfaces,due

toahighdensityofgrainboundariesanddislocations.

3.2. XPS

Chemical modiﬁcations and reconstruction of the diamond

surface whichappear afterUHV annealing(Figs.3 and4)were

discussedpreviously[14]intermsofcarbonandoxygenbonding

andbandbending(holeaccumulationlayer)inducedbyadsorbed

water.

TheC1s XPSspectrumfortheas-grown sample(keptat the

ambientatmosphere)isreportedinFig.3a.Itisdecomposedinto

fourlinesatrespectively283.8eV(labeledC1),284.5eV(labeled

C2),286.0and288.7eV.Ithasbeensuggested[14]thatthese

dif-ferentlinesare theresultofa partlyhydrogenatedsurface.The

componentsat286.0and288.7eVareattributedtoC O Cether

bonds(286.0eV)andtoCOOHcarboxylfunctions(288.7eV).The

mainpeak(C2)correspondingtoC Csp3_is_found_at_284.5_eV_below

theFermilevel(beforeandafterUHVanneal)becausestrongboron

dopinginducesashiftoftheFermileveltowardthevalenceband.

AfterUHVannealing,componentC1(283.9eV)hasdisappeared

(Fig.3b);thisisexplainedbyavanishingoftheholeaccumulation

layer(electrontransferfromdiamondtothephysisorbedwater

layer)andlossofthechemisorbedsurfacehydrogenatoms(C H

(4)

538

537

536

535

534

533

532

531

530

529 -2

-1

0

1

2 microcrystalline

diamond

film

O1s

H

₂

O ads.

OH ads.

XPS signal (10

4

CPS)

Bind

ing en

erg

y (eV)

As-grown Annealed Difference

Fig.4. XPS O1s spectra of as-grown andannealed (1150◦_C)_{microcrystalline}

diamond;thedifferenceO1sspectrumshowsthatbothadsorbedH2OandOH

envi-ronmentsarepartiallydesorbedafterUHVannealing.

theCOOHcomponent (288.4eV) andanincrease oftheC O C

component(285.9eV)areobserved.

TheO1sXPSspectra,beforeandafterannealing,arereported

inFig.4.AlthoughtheO1scorelevelisweaklysensitivetooxygen

atombindingenvironment,weobserveatleasttwocomponents,

withpeak positions which coincide with adsorbed H2O/OH as

measuredrecentlyonoxidizedsilicon surfaces[31].The

differ-encespectrumisconsistentwiththeremovalofbothphysisorbed

waterandeitherphysisorbedorchemisorbedR OHhydrocarbons

(includingacidandalcoholfunctionalities).

3.3. PEELS

ThenormalizedPEELSspectraarereportedinFig.5,forthe

dia-mondﬁlmmeasuredatnormalemissionanglebefore(crosses)and

after(fulldots)UHVannealing.Theobservationofweakchanges,

ontheorderof1%ofthemainC1speakintensity,illustratesthe

requirementofaverygoodsignal-to-noiseratioforPEELSanalysis.

Theﬁrstplasmonlossofas-growndiamondappearsnear34eV,

alongwithcontributionsofmultiplebulkplasmonlossesuptothird

orderovertherange0–90eV.Usingadeconvolutionmethodgiven

-10 0 10 20 30 40 50 60 70 80 90 -0.2 -0.1 0.0 0.1

A

(x 4)

Bulk plasmon

α

= 0

°

Normalized XPS intensity

Loss e

nerg

y (eV)

As-grown Annealed _______ Difference

D

C

B

Fig.5. NormalizedenergylossdistributionofC1sphotoelectronsescapingthe diamondsurfaceatnormalemissionangle:as-grown(crosses)andafterthermal annealing(circles).Thedifferencespectrum(fullline)showsattenuationofthebulk (␴ +␲)plasmon(D)at34eVandenhancementofsurfacecomponentsat10±1eV (peakB)and19±1eV(peakC).NegativepeakAcorrespondstoadecreaseof O C OHfunctionalities. 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0.0 0.5 1.0 1.5 2.0

B

(

E

) Loss Function

As-grown microcrystalline diamond film

Palik PEELS

Energy (e

V)

Fig.6. Bulklossfunction(fulldots)derivedfromPEELSdatafortheas-grownstate ofmicrocrystallinediamond(Fig.5)ascomparedwithB(E)derivedfromoptical data(Palik,Ref.[25])(fullline).Thepeakinthesingle(␴ +␲)bulkplasmonloss distributionislocatedat34.5eV.

inRef.[5],thefunctionB(E)=Im[−1/ε(ω)],relatedtothesingle

bulk(␴+␲)plasmonlossdistribution,ispeakedat34.5±0.5eV

(Fig.6);abroadeningofthelossdistributionisobservedas

com-paredwithB(E)calculated fromopticaldata [25].In Fig.5,the

shoulderfoundnear21eVisattributedtothecorresponding

sur-faceplasmon;itispossiblyenhancedbyoff-normalphotoelectron

emissionatindividualcrystallitesurfaces(typically45◦–55◦).As

emphasizedbefore,determinationofthe(␴+␲)surfaceplasmon

lossdistribution,centeredat20.5±1eV,remainsquitesensitiveto

theaccurateremovalofinterbandtransitions.

Thedifferencespectrum(fulllineinFig.5)clearlyshows

vari-ouspositiveandnegativecontributions:(i)somecarboxyl(O C O)

moietiesareeliminatedattheﬁlmsurface(negativepeakAat4eV);

(ii)thebulk(␴+␲)plasmonofdiamond(negativepeakDat34eV)

is weaklyattenuated; (iii) characteristicfeatures near 10±1eV

(positivepeakB)and19±1eV(positivepeakC)appearintheloss

distributionafterannealing;(iv)incontrast,thelackofanyfeature

at6–7eV,takenasasignatureofgraphitizationin␲-bonded

sys-tems,isanimportantresultwhichrevealstheabsenceofspatially

extendedandorderedgraphiticstructures[15,32,33].

ThesequalitativeresultsshowthatthePEELStechniquegivesa

clearsignatureofweakeffectsinthediamondsurface

reconstruc-tion,evenintheabsenceofgraphitization.Thepossibleoriginof

theunexpectedpeaksBandCisdiscussedinthenextSection.

4. Discussion

Inthiswork,thesensitivityofPEELStoreconstructionof

micro-crystallinediamondﬁlmsurfacesisaddressedanddiscussedinthe

broadercontextofcarbon-basedmaterials.

4.1. Carbon-basedmaterials

Incarbon-basedmaterials,theenergydistributionofplasmon

excitationsis expected tobe sensitivetoa variety of chemical

modiﬁcations:(a)somesp3_–sp2 _conversion_in_the_carbon _atom

hybridizationmightaffectbothsurfaceandbulk(grainboundaries)

(␴+␲)plasmonexcitationduetoachangeintheatomdensity[34];

(b)spatialextensionandorderingingraphiticstructuresmayaffect

theintensityofthe␲–␲*transitionfeaturenear6eV[27];(c)the

surface(_␴+_␲)plasmonexcitationprobability(SEP)isdependent

ontheallotropicformofcarbon[35]anditcouldalsobesensitiveto

(5)

dipolecoverageortovariationsintheholeaccumulationlayerat

thediamondsurface.

SucheffectshaveindeedbeenobservedinpreviousPEELS

stud-iesofamorphouscarbon(a-C)ﬁlms:(i)severalworkshaveshown

thattheplasmonenergyincreasesmonotonouslyasafunctionof

theaveragesp3 _{hybridization}_[36–38]_,_(ii)_UHV_annealing_of_a-C

above600◦C producesa␲–␲*transitionfeatureat5.5eV

with-out affecting the bulk plasmon loss distribution (Fig. 3 in Ref.

[39]), (iii)angular PEELSanalysisofsp3_-rich _amorphous_carbon

ﬁlms, grown by pulsed laser deposition, hasgiven evidence of

asigniﬁcantincrease oftheSEPafterimmobilizationofa dense

molecularmonolayer,usingeitherperﬂuorinatedorester

function-alizedorganicmolecules,withoutaffectingthebulkplasmonloss

distribution[39].

Inthecontextofcarbon-basedmaterials,modelingoftheloss

distributionmaythusbehelpfulforanaccurateinterpretationof

thisbroadrangeofenergylosseffects.

4.2. Diamondsurfacereconstruction

Wehaveusedthediamondsurfacereconstructionprocessto

estimatethesensitivityofthePEELStechniqueforsurface

analy-sis.ThisprocesshasbeenwidelycharacterizedbyXPSbut,toour

bestknowledge,thisistheﬁrstqualitativeanalysisbyPEELS,which

limitscomparisonwithpreviouswork.

Inthisstudy,XPSresultsindicatethattheas-grown

microcrys-tallinediamondﬁlmispartlyhydrogenated.IntheC1scorelevel

(Fig.3),thelargeC1peakat283.9eVisattributedtoanupward

surfacebandbending(holeaccumulationconsistentwithprevious

studiesofair-exposedsamples[8,14]).PeakC2at284.6eVisdueto

sp3_C_atoms_below_this_{nanometer-thick}_region._Peak_C1_disappears

afterannealing,consistentwiththeobservedremovalofthep+

sur-facelayerandtheso-called“transferdopingmodel”proposedby

Maieretal.[8]toexplainthesurfaceconductivity.

ItisstressedthatXPS(Fig.3)showsnoevidenceforasurface

sp2_C1s_component_after_annealing,_which_sets_an_upper_limit_for

sp2 _C _smaller_than _one_monolayer._Independent_electrical

char-acterizationsafterannealingshowthat:(i) thep+ _surface _layer

isnolongerobservedinHallconductivityand(ii)thewide

elec-trochemicalwindowtypicalofpurediamondismaintained[40],

whichwouldnotbethecaseifagraphiticlayerwerepresentat

thesurface.BothXPSandelectricalresultsarethusconsistentwith

PEELSdatawhichgivenoevidenceofgraphitization,inthesenseof

someformationofa“conductivesurface”madeofnanometer-size

layer.

InordertoproposeaninterpretationtotheunexpectedpeaksB

andC(Fig.5),thedielectrictheorywasusedtocalculatebulkand

surfaceplasmondistributions,respectivelyproportionalto

B(ω)=Im[−1/ε(ω)] (1a)

and

S(ω)=Im[(ε(ω)−1)2/ε(ω)(1+ε(ω))]

=Im[(1/ε(ω))−(4/1+ε(ω))], (1b)

whereε(ω)isthedielectricfunction.Thesurfaceandbulkloss

functions(Eqs.(1a)and(1b))werecalculatedusingtabulated

opti-cal dielectric functionsof diamond[25] and graphite(ordinary

index)[26].TheresultsareshowninFig.7wherewehavealso

marked asgray areas the positions of peaks Band C obtained

experimentally(Fig.5).Inclusionofmultiplelosseswouldnot

sig-niﬁcantlyaffectthemaincalculatedfeaturesintheenergyrange

below30eVshowninFig.7.

Notethattheshouldernear21eVobservedpreviouslyin

exper-imentalenergylossdistributionofdiamondhasbeenattributed

eithertointerbandtransitions[41]ortoasurfaceplasmonmode

0 5 10 15 20 25 30 -3 -2 -1 0 1 2 3 4

Surface Loss function

Loss e

nerg

y (e

V)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

a)

Bulk Loss function

A

B

C

Diamond

0 5 10 15 20 25 30 -3 -2 -1 0 1 2 3 4

Surfa

c

e Loss function

Loss e

nerg

y (e

V)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

b)

Bulk Loss function

B

A

C

Graphite

Fig.7.EnergylossprobabilitiescorrespondingtothesurfaceS(ω)(dashedlines) andbulkB(ω)(fulllines)plasmonexcitations,calculatedusingEqs.(1a)and(1b) andtabulatedopticalindexvalues:(a)diamond[25],(b)graphite(ordinary refrac-tiveindex)[26].Experimentalpeaksinthedifferencespectrum(Fig.5)aredepicted bytheshadedareas;peaksBandCcoincidewithsurfaceplasmonexcitationdueto somesp2_phase.

[18,27,42];however,Fig.7showsthatthisfeaturealonecannotbe

takenasasignatureofthesurfaceplasmonbecausethecalculated

surfaceplasmonpeakofdiamond(at21.2eV)overlapstheshoulder

at22.5eVinthebulkplasmon(Fig.7a).

Fig. 7a also shows that the calculated surface plasmon of

diamond (21.2eV) matches the shoulder observed at 21eV in

experimentalspectraforas-grownandannealeddiamondﬁlms;

incontrastitislocatedatalargerlossenergyvalueascompared

withpeakC(19eV)inthedifferencespectrumofFig.5.Hence,we

believethatpeakCinthedifferencespectrumisnotthesignature

ofasurfaceplasmonofdiamond.

Incontrast,theexperimentallossfeaturesnear10±1eVand

19±1eVarewelldescribed bythesurfaceplasmonofgraphite

(Fig.7b).AlthoughpeakBisratherweak,consideredtogetherpeak

BandCareconsistentwiththecalculatedsurfaceplasmonusing

thegraphitedielectricfunction;interestingly,Fig.7bshowsthat

thesurfaceplasmoncontributionatthelocationofpeakBmustbe

smallerthanthatatpeakCposition.

Since thematching of PEELSdata withthesurface plasmon

ofgraphite apparentlycontradictstheabsenceofa “conductive

surface” made of nanometer-size graphitized layer, alternative

explanationscouldbeinteresting.Ontheonehand,someincrease

inthe␴–␴*interbandtransitionstrength,expectednear12–13eV

[43],seemstoohighinenergytoaccountforpeakB.Ontheother

hand,somelossfeaturesnear11eVand19eVwereobservedin

theelectron energyloss distributionofa graphene singlelayer

(6)

bythermalannealingat300◦C,whichisbelievedtoremovedefects

andmetastable sp3 _sites_[44]_._These_features _were _respectively

attributedto␲→␴*and␴→␲*singleelectrontransitions,which

mayrevealdistortedenvironmentsofcarbonatoms.

Furtherworkwithsinglecrystalratherthanmicrocrystalline

diamond may be helpful to address the PEELS signature and

toinvestigatelossdistributionbroadening andsurface plasmon

enhancementeffects.

5. Conclusion

ThesensitivityofPhotoelectronEnergyLossSpectroscopyhas

beeninvestigated using thewell-controlled surface

reconstruc-tionofmicrocrystallinediamondthinﬁlmsuponUHVannealing

at1150◦C.ThiscomparativestudyshowsthatPEELSdetects

sur-faceeffectswhicharenotseenbyUVRamanspectroscopy.After

annealing,thebulk(␴ +␲)plasmonofdiamondat34.5eVisweakly

attenuatedbutnoevidenceforsurfacegraphitizationisobserved

near6eV.Characteristicfeaturesat10±1eVand19±1eVinthe

energylossdistributionmatchthecalculatedopticalsurface

plas-monsingraphite-likematerials;alternativelytheyalsocoincide

withexperimentalplasmonlossesinsomegraphenelayers.This

workshowsthatPEELSacquisitionwithstandardXPSequipment

givesasensitivesignatureofweakeffectsinthediamondsurface

reconstruction,evenintheabsenceofgraphitization.Itconﬁrms

thatPEELSisasensitivetoolforsurfacedielectricfunctionanalysis,

complementarytoXPSchemicalanalysis.

Acknowledgment

Oneofus(C.G.)is gratefultotheCNPqagency(Brazil)for a

visitingresearchergrantintheCiênciaSemFronteirasprogramme.

References

[1] AugerandX-rayPhotoelectronSpectroscopy,in:D.Briggs,M.P.Seah(Eds.), PracticalSurfaceAnalysis,vol.1,seconded.,JohnWileyandSons,Chichester, UK,1990.

[2]S.Tougaard,PhysicalReviewB34(1986)6779.

[3] S.Tougaard,SurfaceandInterfaceAnalysis11(1988)453–472.

[4]A.CohenSimonsen,F.Yubero,S.Tougaard,PhysicalReviewB56(1997)1612. [5]W.S.M.Werner,SurfaceandInterfaceAnalysis31(2001)141–176.

[6]R.F.Egerton,ElectronEnergy-LossSpectroscopyintheElectronMicroscope, SecondEdition,PlenumPress,NewYork,1996.

[7]J.B.Cui,J.Ristein,L.Ley,PhysicalReviewLetters81(1998)429.

[8]F.Maier,M.Riedel,B.Mantel,J.Ristein,L.Ley,PhysicalReviewLetters85(2000) 3472.

[9]J.Ristein,F.Maier,M.Riedel,M.Stammer,L.Ley,DiamondandRelatedMaterials 10(2001)416.

[10]L.Diederich,O.M.Küttel,P.Aebi,L.Schlapbach,SurfaceScience418(1998)219. [11]P.K.Bachmann,W.Eberhardt,B.Kessler,H.Lade,K.Radermacher,D.U.Wiecher,

H.Wilson,DiamondandRelatedMaterials5(1996)1378.

[12]C.H.Goeting,F.Marken,A.Gutierrez-Sosa,R.G.Compton,J.S.Foord,Diamond andRelatedMaterials9(2000)390.

[13]J.I.B.Wilson,J.S.Walton,G.Beamson,JournalofElectronSpectroscopyand RelatedPhenomena121(2001)183.

[14]D.Ballutaud,N.Simon,H.Girard,E.Rezpka,B.Bouchet-Fabre,Diamondand RelatedMaterials15(2006)716.

[15]Sh. Michaelson, A. Hoffman, Diamond and Related Materials 15 (2006) 486–497.

[16]B.R.Stoner,G.H.M.Ma,S.D.Wolter,J.T.Glass,PhysicalReviewB45(1992) 11067.

[17]S.Haq,D.L.Tunnicliffe,S.Sails,J.A.Savage,AppliedPhysicsLetters68(1996) 469.

[18]M.S.Haque,H.A.Naseem,J.L.Shultz,W.D.Brown,S.Lal,S.Gangopadhyay, JournalofAppliedPhysics83(1998)4421.

[19]J.C.Arnault,S.Saada,M.Nesladek,O.A.Williams,K.Haenen,P.Bergonzo,E. Osawa,DiamondandRelatedMaterials17(2008)1143.

[20]R.Klauser,J.M.Cheng,T.J.Chuang,L.M.Chen,M.C.Shih,J.C.Lin,SurfaceScience 356(1996)L410–L416.

[21]F.Maier,R.Graupner,M.Hollering,L.Hammer,J.Ristein,L.Ley,SurfaceScience 443(1999)177–180.

[22] T.Frauenheim,U.Stephan,P.Blaudeck,D.Porezag,H.G.Busmann,W. Zimmer-mannedling,S.Lauer,PhysicalReviewB48(1993)18189–18202.

[23]G.Kern,J.Hafner,J.Furthmuller,G.Kresse,SurfaceScience352–354(1996) 745.

[24]T.Petit,J.C.Arnault,H.A.Girard,M.Sennour,P.Bergonzo,PhysicalReviewB84 (2011)233407.

[25]D.F.Edwards,H.R.Philipps,in:E.D.Palik(Ed.),HandbookofOpticalConstants ofSolids,Academic,SanDiego,1985,p.665.

[26]A.Borghesi,G.Guizzetti,in:E.D.Palik(Ed.),HandbookofOpticalConstantsof Solids,vol.2,Academic,NewYork,1991,pp.449.

[27]S.Waidmann,M.Knupfer,B.Arnold,J.Fink,A.Fleszar,W.Hanke,Physical ReviewB61(2000)10149.

[28]D.David,C.Godet,H.Sabbah,S.Ababou-Girard,F.Solal,V.Chu,J.P.Conde, JournalofNon-CrystallineSolids358(2012)2019–2022.

[29]D.David,C.Godet,unpublished.

[30] K.M.McNamara,K.K.Gleason,D.J.Vestyk,J.E.Butler,DiamondandRelated Materials1(1992)1145.

[31]A.B.Fadjie,S.Ababou-Girard,C.Godet,JournalofAppliedPhysics112(2012) 113701.

[32] S.Waidmann,M.Knupfer,J.Fink,B.Kleinsorge,J.Robertson,Diamondand RelatedMaterials9(2000)722–727.

[33]M.L.Theye,V.Paret,Carbon40(2002)1153.

[34]R.Haerle,E.Riedo,A.Pasquarello,A.Baldereschi,PhysicalReviewB65(2001) 045101.

[35] N. Pauly, M.Novak, S. Tougaard, Surface and Interface Analysis(2013), http://dx.doi.org/10.1002/sia.5167.

[36]A.C.Ferrari,A.Libassi,B.K.Tanner,V.Stolojan,J.Yuan,L.M.Brown,S.E.Rodil,B. Kleinsorge,J.Robertson,PhysicalReviewB62(2000)11089.

[37] P.Reinke,M.G.Garnier,P.Oelhafen,JournalofElectronSpectroscopyand RelatedPhenomena136(2004)239.

[38]A.Zebda,H.Sabbah,S.Ababou-Girard,F.Solal,C.Godet,AppliedSurfaceScience 254(2008)4980–4991.

[39] C.Godet,D.David,H.Sabbah,S.Ababou-Girard,F.Solal,AppliedSurfaceScience 255(2009)6598.

[40]N.Simon,H.Girard,D.Ballutaud,S.Ghodbane,A.Deneuville,M.Herlem,A. Etcheberry,DiamondandRelatedMaterials14(2005)1179.

[41]S.V.Pepper,SurfaceScience123(1982)47.

[42]Y.Wang,H.Chen,R.W.Hoffman,J.C.Angus,JournalofMaterialsResearch5 (1990)2378.

[43]L.Calliari,S.Fanchenko,M.Filippi,Carbon45(2007)1410–1418.

[44]A.Siokou,F.Ravani,S.Karakalos,O.Frank,M.Kalbac,C.Galiotis,AppliedSurface Science257(2011)9785.