• No results found

A structural investigation of the interaction of oxalic acid with Cu(110)

N/A
N/A
Protected

Academic year: 2020

Share "A structural investigation of the interaction of oxalic acid with Cu(110)"

Copied!
11
0
0

Loading.... (view fulltext now)

Full text

(1)

warwick.ac.uk/lib-publications

Original citation:

White, T. W., Duncan, D. A., Fortuna, S., Wang, Y.-L., Moreton, Ben, Lee, T.-L., Blowey, P.,

Costantini, G. and Woodruff, D. P.. (2017) A structural investigation of the interaction of

oxalic acid with Cu(110). Surface Science.

Permanent WRAP URL:

http://wrap.warwick.ac.uk/94274

Copyright and reuse:

The Warwick Research Archive Portal (WRAP) makes this work of researchers of the

University of Warwick available open access under the following conditions.

This article is made available under the Creative Commons Attribution 4.0 International

license (CC BY 4.0) and may be reused according to the conditions of the license. For more

details see:

http://creativecommons.org/licenses/by/4.0/

A note on versions:

The version presented in WRAP is the published version, or, version of record, and may be

cited as it appears here.

(2)

Surface Science 668 (2018) 134–143

ContentslistsavailableatScienceDirect

Surface

Science

journalhomepage:www.elsevier.com/locate/susc

A

structural

investigation

of

the

interaction

of

oxalic

acid

with

Cu(110)

T.W.

White

a

,

D.A.

Duncan

b,c

,

S.

Fortuna

d

,

Y.-L.

Wang

e

,

B.

Moreton

a

,

T.-L.

Lee

c

,

P.

Blowey

f

,

G.

Costantini

a

,

D.P.

Woodruff

f,∗

aChemistryDepartment,UniversityofWarwick,Coventry,CV47AL,UK

bTechnischeUniversitätMünchen,PhysikDepartmentE20,Garching,D-85748,Germany cDiamondLightSource,Didcot,OX110DE,UK

dSISSA,ViaBonomea265,Trieste,Italy

eInstituteofPhysics&UniversityofChineseAcademyofSciences,ChineseAcademyofSciences,Beijing100190,China fPhysicsDepartment,UniversityofWarwick,Coventry,CV47AL,UK

a

b

s

t

r

a

c

t

TheinteractionofoxalicacidwiththeCu(110)surfacehasbeeninvestigatedbyacombinationofscanningtunnellingmicroscopy(STM),lowenergyelectron diffraction(LEED),softX-rayphotoelectronspectroscopy(SXPS),near-edgeX-rayabsorptionfinestructure(NEXAFS)andscanned-energymodephotoelectron diffraction(PhD),anddensityfunctionaltheory(DFT).O1sSXPSandOK-edgeNEXAFSshowthatathighcoveragesasinglydeprotonatedmonooxalateisformed withitsmolecularplaneperpendiculartothesurfaceandlyinginthe[110]azimuth,whileatlowcoverageadoubly-deprotonateddioxalateisformedwithits molecularplaneparalleltothesurface.STM,LEEDandSXPSshowthedioxalatetoforma(3×2)orderedphasewithacoverageof1/6ML.O1sPhDmodulation spectraforthemonooxalatephasearefoundtobesimulatedbyageometryinwhichthecarboxylateOatomsoccupynear-atopsitesonnearest-neighboursurfaceCu atomsin[110]rows,withaCu–Obondlengthof2.00±0.04Å.STMimagesofthe(3×2)phaseshowsomecentredmoleculesattributedtoadsorptiononsecond-layer Cuatomsbelowmissing[001]rowsofsurfaceCuatoms,whileDFTcalculationsshowadsorptionona(3×2)missingrowsurface(witheverythird[001]Cusurface rowremoved)isfavouredoveradsorptionontheunreconstructedsurface.O1sPhDdatafromdioxalateisbestfittedbyastructuresimilartothatfoundbyDFTto havethelowestenergy,althoughtherearesomesignificantdifferencesinintramolecularbondlengths.

© 2017TheAuthors.PublishedbyElsevierB.V. ThisisanopenaccessarticleundertheCCBYlicense.(http://creativecommons.org/licenses/by/4.0/ )

1. Introduction

It is well-establishedthat carboxylic acidsdeprotonate when ad-sorbedon many metalsurfaces,andnotably onCu(110), whichhas provedtobethemodelsurfaceusedinmanyofthesestudies.Indeed, moregenerally,carboxylate-substratebondingiswidelyusedtotether arangeofmoleculestosurfaces(e.g.[1]).Thepresenceoftheformate speciesHCOO,thedeprotonatedformofthesimplestcarboxylicacid, formicacid(HCOOH,FoA– seeFig.1),onCu(110)wasfirstinferred byWachsandMadixintemperature-programmedreactionspectroscopy (TPRS)fromapre-oxidisedCu(110)surfaceexposedtomethanol[2]. SubsequentlyYingandMadix[3] deducedfromTPRSthatsurface for-mateisformedbyexposureofCu(110)toformicacid.Spectroscopic ob-servationoftheformatespeciesonCu(110)wassubsequentlyachieved withbothultravioletandX-rayphotoelectronspectroscopy(UPSand XPS)[4],high-resolutionelectronenergylossspectroscopy(HREELS)

[5]andreflection-absorptioninfraredspectroscopy(RAIRS)[6].While thesespectroscopicstudiesclearlyindicatedthatformateisbidentate bondedtothesurfacethroughthetwo(equivalent)Oatoms,withthe molecularplaneperpendiculartothesurface,itwasultimately scanned-energymodephotoelectrondiffraction (PhD[7])thatprovideda de-tailed quantitative description of the local bonding geometry [8,9].

Correspondingauthor.

E-mailaddress:d.p.woodruff@warwick.ac.uk (D.P.Woodruff).

Specifically,themoleculeisadsorbedinashortbridgesiteonCu(110), themolecularplanelyingina[110]close-packeddirectiononthe sur-face,withthetwoOatomsalmostexactlyatoptwonearest-neighbour surfaceCuatoms(Fig.2).

Subsequently,asurfaceacetatespecies,formedbydeprotonationof aceticacid(AA,Fig.1)exposedtotheCu(110)surface,wasidentified spectroscopically[10–12]andshowntohaveitscarboxylateplanelying ina⟨110⟩azimuthwithitsC–Caxisperpendiculartothesurface,while aPhDinvestigation[13]identifiedthelocalbondingsitetobethesame asfortheformatespecies.Athirdsimplecarboxylate,namelybenzoate formedbythedeprotonationofbenzoicacid(BA,Fig.1)onCu(110), wascharacterisedbyHREELS,RAIRS,lowenergyelectrondiffraction (LEED) andscanningtunnellingmicroscopy (STM) [14–16] showing thatatlowcoveragesthephenylringofthebenzoateisessentially par-alleltothesurface,butathighcoveragethemoleculestandsup.APhD investigationofthisstanding-upphase[17]foundthelocalbonding ge-ometryonCu(110)tobeidenticaltothatofformateandacetate.The molecularorientationwas alsoconfirmedbyelectron-stimulated des-orptionion-angulardistributions(ESDIAD)[18].STMstudiesindicate thatCuadatomsmaybeassociatedwiththebenzoatespecies,atleastat low-coverages[19–21],formingbenzoatedimerswithanintermediate CuadatomorclustersoffourbenzoatemoleculesaroundapairofCu

https://doi.org/10.1016/j.susc.2017.10.025

Received19September2017;Receivedinrevisedform23October2017;Accepted26October2017 Availableonline28October2017

(3)

Fig.1.Schematicdiagramsofthecarboxylicacidsdiscussedinthetext,namely:formicacid(FoA),aceticacid(AA),benzoicacid(BA),oxalicacid(OA),succinicacid(SA),malicacid (MalA),tartaricacid(TA),fumaricacid(FuA),maleicacid(MaleA)andterephthalicacid(TPA).

adatoms.However,atleastforthehigh-coveragephases,thelocal bond-inggeometriesofallthesesimpledeprotonatedsingle-carboxylicacid groupmoleculesareidenticalandshownoevidenceforanyinfluence ofCuadatoms.

Interestingquestionsarise,however,regardingthebehaviourof di-carboxylicacids.Aretheysinglyordoublydeprotonated,andwhatit isthelocalbondinggeometry?Hereweseektoanswerthesequestions forthesimplestdicarboxylicacid,oxalicacid.Interestingly,therehave alreadybeenanumberofinvestigationsofseveralmorecomplex di-carboxylicacids,shownin Fig.1,motivatedin largepartbyinterest intheinteractionofchiralmolecules(andrelatednon-chiralspecies) onsurfaces[22].Forseveralofthesemolecules,namelytartaricacid, TA[23,24],succinicacid,SA[25],andmalicacid,MalA[26],the re-sultsofmainlySTMandRAIRSstudiesindicatethatthemoleculesare doubly-deprotonatedandadoptalying-downconfigurationatlow cov-erage,butaresinglydeprotonatedandstandup athighercoverages. Moreover,lowcoverageSTMandXPSstudiesofterephthalicacid,TPA, onCu(110)clearlyindicateadoubly-deprotonatedlying-downspecies

inseveraldifferentorderedphases[27],whileaRAIRSstudyindicates thatathighercoverageasingly-deprotonatedstanding-upphaseoccurs

(4)

T.W.Whiteetal. SurfaceScience668(2018)134–143

Fig.2. Schematicdiagramsshowingthelocaladsorptiongeometriespreviouslyreported fortheformate,monotartrateandbitartratespeciesonCu(110)basedonPhD investiga-tions.NotethatthePhDtechniqueisinsensitivetothelocationofHatoms,sotheseare omittedfromthediagrams.OHspeciesarerepresentedbytheassociatedOatomswith adifferentcolouring.(Forinterpretationofthereferencestocolourinthisfigurelegend, thereaderisreferredtothewebversionofthisarticle.)

havingbeenseenforthesimpledeprotonatedaminoacidsglycinateand alanate,inwhichbondingtothesurfaceisnotonlybythecarboxylate OatomsbutalsotheaminoNatom[30,31].

Ratherdifferentbehaviourhasbeenreportedfortheinteractionof maleicacid(MaleA)andfumaricacid(FuA),withtheapplication of STM, LEEDandRAIRS together withC 1s XPS, leading tothe con-clusionthat maleic acid is always doubly deprotonatedon Cu(110)

[32] whereasfumaricacidis onlydetected(withRAIRS) asa singly deprotonatedspecieson thissurface[33].These twoacidmolecules arethecis andtrans formofbutenedioicacid,andwhiletheconclusion regardingmaleicacidmaybe attributabletothe cis conformation of themoleculeitisnotclearwhythe trans conformationoffumaricacid shouldprecludethepossibilityofitbeingdoublydeprotonated.

TheonlypreviousstudyofoxalicacidonCu(110)appearstobethat ofMartin,ColeandHaq[34]whoconcluded,onthebasisofRAIRSdata, thatthemoleculestandsupandissinglydeprotonated(monooxalate). Atlowexposuresnoabsorptionbandscouldbedetected;however,this couldbeconsistentwiththepresenceofalying-down(presumably dou-blydeprotonated-dioxalate)speciesinwhichthemolecularvibrational modeswouldnotbedipoleactiveandthuscannotbedetectedinRAIRS. AninvestigationofoxalicacidinteractionwithCu(111),usingSTMand high-resolutionsoft(synchrotronradiation)XPS(SXPS),indicatedthat bothmono-anddi-oxalatespeciesoccurred,primarilyatlowandhigh coveragerespectively[35].

Here wereport the results ofexperiments conducted toobtaina rathercompleteunderstandingofthestructuralandchemical proper-tiesofthesurfacespeciesresultingfromexposureofCu(110)tooxalic acid.Basiccharacterisationofthesystemwasachievedthroughtheuse ofSTM,LEED,andSXPS,particularlyfromtheO1sstate.Quantitative structuralinformationhasbeenobtainedfromO1sPhDcombinedwith OK-edgeNEXAFS(near-edgeX-rayabsorptionfinestructure).Wefind thatoxalicacidfollowsthepatternofmostother(morecomplex) di-carboxylicacidsinadoptingadoublydeprotonatedlying-downoxalate atlowcoveragebutasinglydeprotonatedstanding-upspeciesathigh coverage.ThePhDdataprovideacleardeterminationofthelocal ad-sorptiongeometryforthemonooxalate,butidentifyingtheexact geom-etryofthedioxalateprovedmorechallenging.However,STMimagesof the(3×2)phaseledindirectlytothepossibilitythatthisphaseinvolves [001]missingrows,andDFTcalculationsconfirmthatthedioxalateis

morestronglyadsorbedonsuchareconstructedsurface.O1sPhDdata areconsistentwiththequalitativemodelfavouredbytheDFT calcula-tions.

2. Experimentaldetails

InitialcharacterisationexperimentswereperformedinaUHV sur-facesciencechamberbasedattheUniversityofWarwick,fittedwitha LEEDoptics,alow-temperatureSTM,andtheusualfacilitiesfor sam-plehandlingandcleaning.TheCu(110)samplewaspreparedinsituby cyclesof1keVAr+ionbombardmentandannealingto870Kfor5min untilacleanwell-orderedsurfacewasobtainedasindicatedbyLEED andSTM.Oxalicaciddepositionwasbyroomtemperatureevaporation fromakovarglasstube attachedtotheUHVsystem;duringvacuum bakeofthemetalliccomponents(toonly373K)thistubewaskeptclose toroomtemperaturetoavoidexcessmateriallossduetothehighvapour pressure.Thetubewasthenoutgassedwithaheatgun,purgedintothe pumpinglineofthesampleloadlock.

SXPS,NEXAFSandPhDmeasurementsweremadeintheUHV sur-facescienceend-stationofbeamlineI09atDiamondLightSource.This beamlineisfittedwithapairofcantedundulators,eachfeedinga dif-ferentmonochromator,butbothbringingthefocussedmonochromated beamstothesamespotonthesample.Thelower-energyundulator radi-ationbranchisfittedwithagrazing-incidenceplane-grating monochro-matorcapableofprovidingphotonsatthesampleintheenergyrange 100–1100eV.Thehigher-energybranch,usingadouble-crystalSi(111) monochromator,wasnotusedintheexperimentsreportedhere.Sample cleaninganddosingusedthesamemethodsasintheWarwick cham-ber,althoughsampleorderingandcleannessinthiscasewereassessed usingalow-currentLEEDopticsandSXPspectraobtainedusingaVG ScientaEW4000electronspectrometerwitha±30° acceptanceanglein theplaneofincidenceandpolarisationoftheradiation,mountedwithin thispolarisationplaneatanangleof60° betweentheincidentradiation andthedirection fromthesampletothecentreofthedetector.This analyserwasalsousedtocollectthephotoemissiondataforthePhD experimentsandtheAugerelectronemissionintheNEXAFSstudies.

3. Surfacecharacterisationresults

3.1. STM and LEED

InitialcharacterisationofthesystemusingSTMandLEEDshowed thatanominalexposureatroomtemperatureofoxalicacidof5×10−6 mbars,asmeasuredbyaniongaugewithinthemainchamber,ledtoa disorderedsurfacewithnoorderedmolecularimagingdetectablewith STM(Fig.3(a)).However,afterannealingto398KbothSTMandLEED indicatedthepresenceofa(3×2)orderedphase,asshowninFig.3(b andd).Thiswastheonlyorderedphaseseen byeithertechniqueat anysurfacecoverage.WhiletheprotrusionsseeninSTMimageofthe (3×2)phaselackanyvisibleinternalstructurethesearetentatively as-sumedtobeduetoindividualadsorbedmolecules.Thisattributionis supportedbySXPSdataobtainedlater(seebelow),whichprovideda coverageestimateofaphasegivinga(3×2)LEEDpatternof approxi-mately0.2ML,broadlyconsistentwithonemoleculeper(3×2)surface unitmesh(0.17ML).Furthersupportforthisinterpretationisprovided bytheSTMimageofFig.3(c),obtainedafterfurtherannealingto423K, whichledtoalossofmoleculesfromthesurfaceandproduceda dis-orderedsurfacethatdoesshowindividualad-species.Inparticular,a featureinthecentreofthesuperimposedgreencircleappearstobean isolatedadsorbedmoleculeinanotherwisecleanareaofthesurface,and indeedtheshapeofthisfeaturecouldbeconsistentwithalying-down fullydeprotonateddioxalatemoleculethatSXPSandNEXAFSdata(see below)indicatetobepresentonthesurfaceinthe(3×2)phase.

TheSTMimageofthe(3×2)phase(Fig.3(b))showsaparticularly surprisingsetoflocaldefects,namelyprotrusionsatthecentreofthe (3×2)orderedfeaturesthatappeartobequitesimilartothoseofthe

(5)

2 nm

[001] _ [110]

(0,1) (1,0)

2 nm

b

c

5 nm

a

_ [110] [001]

d

Fig.3. (a)STMimageofthedisorderedunannealedsurface.(b)STMimageand(d) LEEDpatternrecordedat148eVfromtheordered(3×2)phaseformedbyoxalicacidon Cu(110)followingannealingofahigher-coveragedisorderedphaseto398K.(c)shows anSTMimageobtainedafterfurtherheatingto423K.Tunnellingconditions:(a)120pA,

−1.1V;(b)180pA,−1.1V;(c)100pA,1.1V.

2 nm

+1.1 V

x

(nm)

1 2

60

40

20

0

z

(pm)

2 nm

-1.1 V

a

b

c

Fig.4. (a)and(b)showSTMimagesofthe(defected)(3×2)phaseobtainedatopposite biasvoltages(butthesametunnellingcurrentof180pA).(c)showstheline-scanprofiles alongtheequivalentlocationsinthetwoimages,colour-codedingreenandblue.(For interpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothe webversionofthisarticle.)

un-defectedregions.Atfirstsightthesecouldbeinterpretedas addi-tionalidenticalmolecules,butthecentreofa(3×2)mesh(oranymesh involvinganodd-numberedperiodicityinonedirection)cannot corre-spondtoanidenticallocalsite.Infacttheapparentheightofthecentred defectprotrusions,relativetothoseoftheregular(3×2)protrusions, doesdependonthetunnellingconditions(seeFig.4),withtheratioof theseheightsbeingcloseto1.0inpositivesamplebias(emptysurface

Fig.5. SXPspectraaroundtheO1sandC1semissionpeaksrecordedfromthetwo differentcoveragesofoxalicacidspeciesonCu(110).

stateimaging),butcloseto1.25innegativesamplebias(occupied sur-facestateimaging).Thisobservationisconsistentwiththeviewthatthe moleculesgivingrisetothesetwofeaturesdodiffer.

NoticethattheSTMimageofFig.3(b)alsoshowsanotherdefect structure,namelyanantiphasedomainboundarycloseto,andrunning parallelto,theright-handedgeoftheimage,markedwithan arrow-head.Images(notshown)fromlesswell-orderedpreparationsshowa muchhigherdensityoftheseantiphasedomainboundariestoproducea ‘striped’ structurethatledtoLEEDpatternswithspotsplittingof(n ×2) character where n issignificantly largerthan3.Notice,though,that thewidthoftheantiphasedomainboundariesappearstocorrespondto twicetheCu–Cuatomspacingalong[110]soallmoleculesonbothsides oftheseboundariesoccupyidenticallocalsites.

3.2. SXPS and NEXAFS

SamplepreparationfortheexperimentsperformedatDiamond fol-lowedessentiallythesamerecipesusedinWarwick,butcomparisonof theO1sandCu2pSXPspectraprovidedestimatesoftheassociated cov-erages.Fortheinitially-depositedhigh-coveragephaseandthe lower-coverageannealedphaseSXPSledtocoverageestimatesof0.35MLand 0.20ML,respectively.Asremarked above,thelowercoveragephase yieldeda(3×2)LEEDpatternthat,foronemoleculepersurfaceunit mesh(asimpliedbytheSTMimages)wouldcorrespondtoacoverage of0.17ML,consistentwiththeSXPSestimate.NoorderedphaseLEED patternwasobtainedfromthehigher-coveragepreparation;published densityfunctionaltheory(DFT)calculationsforamono-oxalatephase werebasedon(2×2)orderingwithacoverageof0.25ML[36] buta c(2×2)orderingwithacoverageof0.5MLwouldalsoappeartobe possible (andwas reportedbyMartinetal.[34]).TheSXPS-derived coverageliesbetweenthesetwovalues.

(6)

T.W.Whiteetal. SurfaceScience668(2018)134–143

bindingenergy,maybeattributedtoemissionfromacarboxylicmoiety thatisstillprotonated;wemaythereforeinferthatthisspectrumisfrom amonooxalatespecies.Suchaspeciesmustgiverisetothree chemically-distinctO1scomponentscorrespondingtotheOatomsofdeprotonated COOandtheOHandC=Ocomponentsoftheintactacidgroup.Aunique unconstrainedthree-componentfittothisspectrumisnotpossible,and thefitshowninFig.5 (usingDoniach-Sunjiclineshapes)is basedon theassumptionthattheCOOcomponenthasexactlythesamebinding energyandspectralwidthasthatseeninthespectrumfromthe low-coveragedi-oxalate spectrum,while theothertwocomponents,each haveanintegratedintensitythatis approximatelyonehalfofthatof theCOOcomponent.Theseconstraintswouldbeconsistentwitha sur-facebeingcoveredbyastanding-upmonooxalatespecies.Noticethat withtheseconstraintsappliedtheresultingfitshowninFig.5 hasthe OHandC=OGaussianpeakwidthsbeingapproximatelytwicethatof theCOOpeak;theoriginofthiseffectisunclearalthoughtheOatoms intheOHandC=Ocomponentsdohaveaverydifferentelectronic en-vironmentawayfromthesurfacetothatoftheOatomsintheCOOthat arebondedtothesurface.

Asafurtheraidtotheidentificationofthethreecomponents,DFT calculationsofthechemicalshiftstobeexpectedfromthismono-oxalate specieswereperformedusingtheCASTEPcode[37],assumingac(2×2) orderingofthe(structurally-optimised)mono-oxalatespecies,butwith asinglecoreholeinac(6×6)unitmeshtominimisehole-hole inter-actionsinthecalculation.Thesecalculationsledtoestimatedchemical shiftsrelativetotheCOOcomponentof0.02eVand1.9eVfortheC=O andOHcomponents,respectively.Thisclearlyindicatesthatthe compo-nentwiththehighestbindingenergyisfromtheOHoxygenatom,while theintermediatecomponent,withonlyasmallenergyshiftfromthe COOcomponent,correspondstotheC=Ocomponent.The correspond-ingchemicalshiftvalues(fortheC=OandOHcomponents)usedinthe experimentalfitofFig.5 are0.6eVand2.1eVwithrelativeintensities ofCOO:C=O:OHof1:0.58:0.64.However,fitstothisspectrumproved particularlyinsensitivetotherelativebindingenergyoftheC=O com-ponentandreasonablefitscouldalsobeobtainedwithamuchsmaller chemicalshiftoftheC=Ocomponentof0.2eV,muchclosertothe DFT-predictedvalue.Aconsistenttrendofallofthesefitswasthatthe rela-tiveintensityoftheCOOcomponentwaslowerthanthattobeexpected fromthemolecularstoichiometry;thisisconsistentwithattenuationof thiscomponentduetoinelasticscatteringofthephotoelectronspassing throughtheupperpartofthemolecularlayer.

Thisapparentlystraight-forwardinterpretationoftheO1sspectrain termsofmonooxalateanddioxalatespeciesathighandlowcoverages wouldleadonetoexpectasingleC1scomponentfromthelow cover-agephase,whilefromthehighcoveragephasetheremaybetwo com-ponentswithasplittingduetodifferencesbetweentheCOOandCOOH carbonatomenvironments(informateandformicacidonCu(110)a differenceof∼2eVhasbeenreportedwithindividualvaluesof287.5– 288.0eVand289.6–290.0eV,respectively[38,39]).Infactthe high-coveragephaseshowstwodominantpeaksatphotoelectronbinding en-ergiesof284.1eVand288.5eV,whilethelow-coveragespectrumhas threecomponentsatenergiesof284.1eV,287.7eVand289.9eV.This behaviourwithsimple‘clean’ O1sspectraandcomplexmulti-peakC1s spectrahasalsobeenreportedforoxalicacidinteractionwithCu(111) byFaraggietal.[35]. Thepeakat284.1eVlabelled#1inFig.5 is consistentwiththepresenceofgraphiticcarbon.Inotherexperiments wehaveconductedofoxalicacidonCu(111)thispeakcandominate thespectrumwithagraphiticcarboncoverageestimatesignificantlyin excessof1ML,onlyconsistent(inthepresenceofcoadsorbedoxalate peaks)withthepresenceofthree-dimensionalislandsofgraphitic ma-terial.Asthispeakonlyappearedafteroxalicaciddepositionitmust clearlyresultfrom some kindofdecomposition reaction on the sur-face.However,perhapssurprisingly,wefoundnoevidenceforthispeak toincreasewithcontinuedexposuretotheincidentsynchrotron radia-tion,indicatingthatitisnotduetobeam-induceddamage.Thepeaks at287.7eVand288.5eV(#2and#2′)mustbeattributedtoCatoms

Fig.6. OK-edgeNEXAFSspectrarecordedfromthelowandhighcoveragephasesformed byoxalicreactionwithCu(110),recordedforincidenceintwoorthogonalazimuthal di-rections,andattwodifferentincidenceangles,definedastheanglebetweenthesurface normalandtheA-vectoroftheincidentradiation.Normalincidenceisthuslabelled90°.

withintheadsorbedoxalatespecies.Thisenergydifferencebetweenthe twophasescannotbeattributedtothelowcoveragepeakbeingdue toonlyCOOwiththehigh-coveragepeakbeingduetobothCOOand COOHspecies,becausetheenergyshiftisofoppositesigntothatforthe HCOOandHCOOHspeciesreportedabove.Suchanenergyshiftshould alsobeaccompaniedbyabroadeningofthepeakfromthehigh cover-agephaseduetotheunderlyingdoublet,butthewidthsofthetwofitted peaksinFig.5areidentical.Theonlyplausibleexplanationforthe dif-ferenceseemstobetheverydifferentelectronicscreeningenvironment ofCatomsinalying-downandstanding-upmolecule.Theoriginofthe C1speak#3at289.9eVisevenlessclear.Faraggietal.intheirstudy ofoxalicacidonCu(111)havesuggestedthatthispeakmaybedueto somekindofCO2 orCO3 speciesbuttheseseemunlikelytobestableat roomtemperatureandevenlesslikelytosurvivethesampleannealing. Theenergyofthisfeaturecouldcorrespondtoanintactcarboxylicacid (asinthevaluereportedforformicacid),butthistoo,isunlikelytobe stableafterannealing,whiletheO1sspectrumshowsnohintofsucha speciesbeingpresentfromthelow-coveragesurface.Amoreprobable explanationisthatthisfeatureisashake-upsatelliteofthecarboxylate Cpeak#2;thisassignmentofasimilarfeatureinthecaseoftartaricacid onCu(110)wasfoundtoaccountforanotherwisenon-stoichiometryof themoleculeimpliedbytheC1sXPS[30].

Supportingevidencefortheassignmentofthelow-coveragephaseto alying-downmoleculeandofthehigh-coveragephasetoastanding-up moleculeisprovidedbytheOK-edgeNEXAFSdataofFig.6.These ori-entationsaredemonstratedbythedependenceoftherelativeintensity ofthesharppeakataphotonenergyofapproximately535eV,dueto

(7)

citationtoa𝜋-resonance,onthepolarisationdirectionoftheincident radiation.Thispeakisexpectedtohaveits maximumintensitywhen theA-vectoroftheincidentradiationisperpendiculartothemolecular plane.Thusinthelowcoveragephasethisintensityislargerat graz-ingincidencethanatnormalincidence,consistentwithalying-down configuration.Forthehighcoveragephasetheintensityofthisfeature isgreatestwhentheA-vectorliesinthesurfaceplaneinthe[001] az-imuth,indicatingnot onlythat themoleculestandsup,butthatthe molecularplaneliesinthe[110]azimuth.Notice,though,thatthe𝜋 -resonancepeakdoesnotcompletelyvanishatnormalincidencefrom thelowcoveragephase,noratnormalincidenceinthe[110]azimuth fromthehighcoveragephase,implyingasmallamountofdisorderor tiltingandtwisting.

4. Structuredetermination:PhDdata,simulations,andDFT calculations

4.1. Introduction

ThePhDtechniqueexploitsthecoherentinterferenceofthedirectly emittedcomponentof aphotoelectronwavefield fromanear-surface atomwiththosecomponentsscatteredbyatomsinthelocal environ-mentoftheemitter.Byvaryingtheincidentphotonenergy,the pho-toelectronkineticenergy(andhencethephotoelectronwavelength)is varied,causingthescatteredcomponentsofthephotoelectron wave-fieldtoswitchinandoutofphasewiththedirectlyemittedcomponent. Theresultingmodulationsinthephotoemissionintensityinaspecific direction,asafunctionofphotonenergy,providestructural informa-tiononthelocalenvironmentoftheemitter.InthepresentcasePhD modulationspectrawereobtainedbymeasuringphotoelectronenergy distributioncurves(EDCs)oftheO1speak(s),at4eVstepsinphoton energy,overthephotoelectronkineticenergyrangeof50–300eV,fora rangeofdifferentpolaremissionanglesinthe[110]azimuth.The2-D detectorfittedtothelargeacceptance-angleelectronspectrometer al-lowedseparatespectratobeextractedat5° intervalsovera50° polar emissionanglerangeforanyspecificsampleorientation.Thesespectra werethenprocessedfollowingourgeneralPhDmethodology(e.g.[6]) inwhichtheindividualEDCsarefittedbyoneormoreGaussianpeaks, aGausserrorfunction(step),andatemplatebackgroundobtainedfrom thewingsoftheEDCs.Theintegratedareasofeachoftheindividual peakswerethenplottedasafunctionofphotoelectronkineticenergy,

I (E ),andusedtodefineastiff spline,I 0(E ),throughI (E ),thatrepresents thenon-diffractive intensityandinstrumentalfactors.Thespline was thensubtractedfrom,andusedtonormalise,theintegratedareas,to providethefinalPhDmodulationspectrum,𝜒(E )=(I (E )− I 0(E ))/I 0(E ). Theresults of DFTcalculationsfor theadsorbedmonooxalate on Cu(110)havealreadybeenpublished[37].Thesecalculations,assuming thatthePBEexchangecorrelationsufficestodescribethechemisorbed configurations,haveshownthatverticallystandingmoleculesare ener-geticallyfavouredallowingforhighpackingdensities.

However,forflatlyinggeometriesthedispersionforcesshouldalso betakenintoaccountforanaccurateestimationoftheadsorption en-ergyandthecomputationalapproachusedforthenewcalculations re-portedherediffersfromthatreportedinref.[37]ashereLondonforces arealsoaccountedfor[40].Specificallycalculationswereperformed withthe Perdew–Burke–Ernzerhofgeneralised gradient-corrected ap-proximation(PBE-GGA)fortheexchangeandcorrelationenergy func-tional[41]withaddedsemi-empiricalGrimme’sDFT-D2vanderWaals interactioncorrection.Thespin-polarizedKohn-Shamequationswere solvedintheplanewavepseudopotentialframework,asimplemented inthePWscfcodeoftheQuantumESPRESSOdistribution[42,43] The surfacewasmodelledwithaslabconsistingoffiveatomiclayers, sepa-ratedinthezdirectionbymorethan13Å,withthelatticeparameterset totheequilibriumvaluecalculatedforthebulk(3.66Å).Thelower-most twolayersconstrainedtotheirbulk-likecoordinates,andenergieswere

Fig.7. ExperimentalPhDmodulationspectraatdifferentpolaremissionanglesinthe

[110]azimuthrecordedfromthetwofittedcomponentsoftheO1semissionfromthe high-coveragephaseproducedbyoxalicacidreactionwithCu(110).

calculatedusingaregulargridofk-pointscomparabletothe12×14×1 gridusedforthecleanCu(111)(1×1)surface.

Ingeneral,DFTcalculationsshowthatflat-lyingfullydeprotonated configurations,likelytodecomposeintoCO2 onthecleancopper sur-face,canbestabilisedbythepresenceofadatomsorsurface reconstruc-tions.TheformationenergyΔE associatedwiththeprocessofadsorbing

m oxalicacidmoleculesonacleancoppersurfacewiththereleaseofn

hydrogenmoleculesandtheremovalof k Cuatomscanbecalculated as:

Δ𝐸=𝐸tot+𝑛𝐸H2+𝑘𝐸Cu−𝑚𝐸COOH2−𝐸surf

where E tot istheenergyoftheassembledsystem,𝐸H2, E Cu ,𝐸COOH2are theenergiesofanisolatedmolecularhydrogen,abulkcopperatom,and anisolatedoxalicacidrespectively,whileE surf istheenergyoftheclean unreconstructedsurface.

4.2. High-coverage phase

AsdiscussedinSection3.2,theO1sSXPspectrafromthehigh cover-agephaseactuallycomprisethreechemically-shiftedcomponents,but there issome uncertaintyintheexactvalueofthechemical shiftof theC=OcomponentrelativetothatfromtheCOOoxygenatoms, al-though itis clearlysmall. Inevitably,thesignal-to-noiseratiosof the angle-resolvedspectra, extractedby averagingover only a5° range, areworsethanthoseofthewide-angleintegratedspectraofFig.5,so itprovedunrealistictoseparateallthreecomponentsinameaningful fashion,andfitstothesePhDEDCswereconductedwithonlytwo com-ponents,ahighbindingenergypeakcorrespondingtoemissionfromthe OHoxygenatom,andasinglelowbindingenergypeak,corresponding toemissionfromallthreeoftheCOOandC=Ooxygenatoms.Asetof theresultingPhDmodulationspectrarecordedat10° intervalsin emis-sionangleisshowninFig.7.

(8)

T.W.Whiteetal. SurfaceScience668(2018)134–143

Fig.8. Comparisonofexperimental1sPhDspectrafromthelowbindingenergy(COO+C=O)peakfromthehighcoveragephaseformedbyoxalicacidreactionwithCu(110)withthe resultsofmultiplescatteringsimulationsforthebest-fitstructure.Thisstructureisshownschematicallyontheright(includinganarbitrarily-placedHatom)togetherwiththestructural parametervalues,includingsmallperpendicularandparallelshiftsoftheCuatoms,bondedtothecarboxylate,relativetotheideally-locatedCu(110)surfaceatoms.

oftheCOOandC=Ooxygenatomsarealsoratherweak,butparticularly nearnormalemissionthereisclearlyanunderlyinglong-period modula-tion.Thislongperiodindicatesthatthescatteringpath-lengthdifference isreasonablyshort,whilethefactthatthemodulationsarestrongestat normalemissionsuggeststhattheOatomsgivingrisetothese modu-lationslieclosetoatopCusurfaceatoms,thusgivingrisetoscattering inthegenerally favoured180° backscatteringangle. However,while theCOOoxygenatomscontributetoaclearlong-periodmodulation, theemissionfromtheC=Ooxygenatomisexpectedtocontributeonly weakmodulations,likethatobservedfromtheOHspecies.The conse-quentialdampingofthePhDmodulationswill,atleastinpart,account fortherelativelyweakamplitudeofthemodulations(lessthan±10%) observedevenatnormalemission.

Thisinterpretationisconfirmedandquantifiedbytheresultsof mul-tiplescatteringsimulationsforarangeofpossibleadsorptionsites,Cu–O bond-lengths,moleculartwistsandtilts,localsurfacerelaxations,and vibrationalamplitudesoftheemitterandscattereratoms.Identifying thebest-fitstructureoversuchasearchisaidedbytheuseofan objec-tivereliability-or R -factordefinedasanormalisedsumofthesquared differencesbetweenthetheoreticalandexperimentalmodulation am-plitudesR =Σ(𝜒th𝜒ex)2 /Σ(𝜒th2 +𝜒ex2 ).Thestructuralmodelyielding thelowestvalueofR isidentifiedasthebest-fitstructure.Fig.8showsa comparisonofasubsetoftheexperimentalCOO+C=OO1sPhDspectra (takenfromthecompletesetshowninFig.7,mostlyforsmallpolar an-glesforwhichthemodulationsarestrongest)withtheresultsofmultiple scatteringsimulationsforthebest-fitstructuralmodel,withan associ-atedvalueof R of0.40.Thisrelativelyhighvalue(R valuesof0.2or lesscanbeachievedforsomeadsorbatesystems)reflectsthefactthat allmodulationsareratherweakandsotherelativelevelofthenoise isquitehigh.ThisstructureisalsoshownschematicallyinFig.8 to-getherwiththebest-fitstructuralparametervalues(theOHhydrogen atomisshownforcompletenessbutitslocationcannotbeobtainedfrom aPhDstudyduetoitsweakscatteringcross-section).Thelocal bond-inggeometryisequivalenttothatoftheformate,acetate,benzoateand mono-tartratespeciesontheCu(110)surface;themoleculebridgestwo nearest-neighbourCuatomsalongthe[001]azimuthwiththebonding Oatomsslightlyoff atopsites.Thismodelisalsoconsistentwiththe re-sultsofDFTcalculationsthatfindstanding-upmonooxalatespeciestobe

energeticallyfavouredoverawiderangeofcoverages[36].TheCu–O nearest-neighbourbondlengthofthebest-fitstructureshowninFig.8is 2.00±0.04Å,tobecomparedwiththevaluefoundintheDFT-D2 cal-culationsof2.00Å.Moreprecisely,whenintermolecularH-bondingis presenttheCu–OdistanceofanOalsoengagedintheH-bondis2.01Å andshortensto1.99Å fortheotherO.Thesecalculationsindicatethat theheightdifferenceofthecarboxylateOandCatomsis0.59Å;O1s PhDisrelativelyinsensitivetothisparameter,sothediscrepancywith thetheoreticalvalueisnotunreasonable.

4.3. Low-coverage phase

As the SXPS from the low-coverage lying-down dioxalate phase showsonlyasingleO1s peak,extractingtheassociatedPhDspectra wasstraightforward.Unfortunately,identifyingastructuralmodelfor whichmultiplescatteringsimulationsgaveasatisfactoryfittothesedata provedtobeconsiderablymorechallengingdespiterunningavast num-berofcalculationsbasedonthree-dimensionalgridsearchesarounda numberofplausiblestartingmodels,butalsorunningautomatedsearch routinesfromdifferentstartingpointsusingtheparticle-swarm optimi-sation(PSO)method [44].These proceduresidentifystructureswith thelowest R -factorsinamultidimensionalparameterhyperspace. Un-fortunately,althoughthisprocessdidyieldsomestructuralmodelswith valuesof R downto∼0.40,thesemodelsinvolvedstructural parame-tervaluesthatwerephysicallyimplausible-forexampleCu–Onearest neighbourdistancesof∼2.30Å ormore,orof1.6Å orless.

Ofcourse,theinformationfromNEXAFSandSXPSprovidessome clearconstraintsonthelikelystructure.ThesingleO1speaksinSXPS clearlyindicatesthespeciesisfullydeprotonatedandNEXAFSshowsthe molecularplaneisnear-paralleltothesurface,soitseemslikelythatall fourOatomsbondtotheCusurface.Ontheotherhand,theO1sPhD modulationsatnormalemissionareextremelyweak(seeFig.9),anddo notshowanycleardominantlong-periodmodulations.Thisseemsto in-dicatethattheOatomsare not directlyatopsurfaceCuatoms,asinthe monooxalatespecies,andindeedasinallothercarboxylatespreviously studiedbyPhDonCu(110).Ofcourse,theabilityofadicarboxylateto haveallOatomsclosetoatoponCu(110)dependsonthematchingof theO–OdistancesinthemoleculeandtheCu–Cudistancesonthe

(9)

Fig.9.Comparisonof4individualO1sPhDspectrarecordedfromthelowcoverage oxalatephaseonCu(110)withtherepresentativespectruminthesamegeometryfrom thehighcoveragephase(asshowninFigs. 7 and8 ).Theindividuallow-coveragespectra arerecordedfromdifferentpreparationsandindicatethedegreeofreproducibilityofthe weakmodulations.

face.Inthisregard,dioxalatelackstheflexibilityofboththeditartrate speciesandtheaminoacids.

WeakmodulationsmayinpartarisefromthefactthatthePhD mod-ulationscomefromanincoherentsumofthemodulationsfromfour differentOatomsinthedioxalate;thesefourOatomscouldhave dif-ferentlocaladsorptionsites,althoughthetwo-foldsymmetry ofboth themoleculeandthesubstratemightreasonablybeexpectedtoleadto mostOatomsoccupyingsymmetrically-equivalentsites.Nevertheless, findingamodelstructurethatshowssuchextremelyweakmodulations atnormalemissionproveddifficult,whilethegenerallyweak modula-tionsatallangles(andthushigherlevelsofnoiserelativetothe mod-ulations)necessarilyleadstohighervaluesofthe R -factorevenforthe bestmodels.Thisalsoresultsinweakvariationsof R inmanyregions ofstructuralparameterhyperspace,renderingsearchesbasedonlyon identifyingthelowest R valueslesseffective.

Thereis,however,onefurtherpieceofinformationemergingfrom theSTMdatathatmayhelptoidentifyoneimportantingredientofthe correctstructuralmodel.Specifically,theSTMimagesofFigs.3 and

4provideevidenceforapossiblesurfacereconstructionassociatedwith thelow-coverage(3×2)phase.Theseimagesappeartoshowthatsome ofthe(3×2) unitmesheshave moleculesat theircentresaswellas attheircorners, despitethefactthatthecentreandcornersitesare inequivalentwithrespecttoCu(110)surfacestructure.Inviewofthisit isunsurprisingthattheSTMlinescansatdifferentbiaspotentialsalso indicatethatthesecentredmoleculesdifferinsomewayfromthoseat thecorners.Fig.10 showsamodelthatcouldaccountforthiseffect. Specifically,ifthe(3×2)phasehaseverythird[001]outermostlayer Curowmissing,astandingupmonooxalatespeciescouldbondtothe exposedsecondlayerCuatomsinasitethatliesatthecentreofthemesh formedbythelying-downdioxalatespeciesoccupyinghigh-symmetry sitesonthedouble-Cuatomrowsthatarenotmissing.Ofcourse,the O1sSXPspectrafromthelow-coveragephaseindicatethepresenceof verylittle,ifany,standing-upmonooxalateinthesurfacepreparations fromwhichPhDdataweremeasured.However,theSTMimagesofthe defected(3×2)phasemaybetakentoprovideuswithevidencethat thesemissing rowsareacharacteristic ofa perfectlyordered (3×2) phase.

Toexplorethisideaofamissingrowmodelofthe(3×2)dioxalate phase,furthersearchesof possible structuresbasedon thissubstrate model,startingfromdifferentlocaladsorption sitesandorientations, wereundertakenusingthePSOroutine.AsforthePSOsearchesof mod-elsontheunreconstructedsurface,severaloptimisedstructureswithlow

R -factorswerefoundwithCu–Obondlengthsthatwerephysically

unrea-Fig.10. Schematicmodelofamissing-row(3×2)phaseformedbylying-downdioxalate species,showingthatadditionaladsorptionofstanding-upmonooxalatespecieswould occupysitesatthecentreofthismesh,potentiallyaccountingforthe’defect’sitesseenin theSTMimagesofFigs. 3 and4 .ForclaritytheoutermostCulayeratomsareshownin adifferentcolourfromthoseoftheunderlyingbulk.(Forinterpretationofthereferences tocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

sonablylongorshort,butonesolutionemergedthathadamore plausi-bleCu–Onearest-neighbourbondingdistanceof1.86Å.Thismodel,and acomparisonofasubsetoftheexperimentalO1sPhDspectrawith sim-ulationsbasedonthesamemodel,areshowninFig.11.Noticethatthe subsetofspectrachosenforthesecalculationsaremostlythosewiththe strongestmodulations,althoughthenormalemissionspectrumisalso included;clearlythecorrectmodelshouldatleastreproducethelackof strongmodulations(±5%orless)inthishigh-symmetrydirection.The valueofthe R -factorobtainedfromthestructureshowninFig.11 is 0.50,whichiscertainlyratherhigh.Notsurprisinglyinviewofthis, vi-sualinspectionofthespectrashowsclearlythatthesimulatedspectra includeafewmodulationsthataremuchlargerthanthoseofthe experi-mentaldata.Ontheotherhand,mostoftheexperimentalpeakenergies arereproducedquitewell,andthesimulationsforthenormalemission spectradoshowextremelyweakmodulationsovermostoftheenergy range,akeyfeatureoftheexperimentaldata.

Ofcourse,anothersourceofpossiblequantitativestructuralmodels isenergyminimisationcalculationsusingDFT.InfactthepublishedDFT studyofoxalicacidderivativesonCu(110)[37] didconsiderpossible structuresformedbyaflat-lyingdioxalateonthissurfacebutconcluded thattheadsorptionenergywasverysignificantlylowerthanthatofthe standing-upmonooxalateandindeedthatmostpossibleadsorptionsites didnotleadtostableadsorptionbutratherdissociationintotwoCO2 molecules.

NewDFT-D2calculationswereperformedforthemissingrow struc-tureleadingtotwopossiblestableadsorptionstructuresfordioxalate, oneofwhichhasasignificantlylowerenergythananyadsorption struc-tureontheunreconstructedsurface.TheseareshowninFig.12 along withthevaluesoftheassociatedadsorptionenergy.Specifically,with themoleculelyingabovethedoublerowsbetweenthemissingrows, butcentredaboveasecondlayerCuatomwiththeC–Cmolecularaxis alignedalongthe[110]azimuth,theenergywasfoundtobe171meV lowerthantheequivalentsiteontheunreconstructuredsurface.

(10)

vari-T.W.Whiteetal. SurfaceScience668(2018)134–143

Fig.11. ComparisonofexperimentalandsimulatedO1sPhDspectrarecordedfromtheCu(110)(3×2)dioxalatephaseforthestructuralmodelillustratedontheright.

Fig.12. Schematictopviewsofthetwostabledioxalateadsorptiongeometriesfoundin DFTcalculationsforthe(3×2)missingrowstructureofCu(110).Forclaritytheoutermost Culayeratomsareshowninadifferentcolourfromthoseoftheunderlyingbulk.Also indicatedarethecalculatedadsorptionenergiespermoleculeΔE.(Forinterpretationof thereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionof thisarticle.)

ationofthe R -factorwithchangestoeachstructuralparameter[6],we findthevaluesofinteratomicdistancestobe:Cu–O,1.86±0.06Å;C–C, 1.15±0.10Å;C–O,1.12(+0.21/−0.09)Å.Forcomparisonthevaluesof theseparametersfoundintheDFTcalculationsare:Cu–O,1.94Å;C–C, 1.58Å;C–O,1.27Å.Evenbearinginmindthelargeerrorestimatesin theprecisionofthesePhD-derivedvalues,therearesignificant quanti-tativedifferencesbetweentheoryandexperiment,despitethesimilarity inthequalitativestructureintermsoflocaladsorbate-substrateregistry andazimuthalorientation.Perhapsthemostobviousproblemwiththe experimentally-determinedstructuralparametervaluesistheC–C dis-tancethatisunphysicallyshort,eventakingaccountofthelimited

pre-cision.Ofcourse,theO1sPhDmeasurementsareprimarilysensitiveto therelativelocationofthenearest-neighbourCuatoms,whicharemore stronglyscatteringthanCatomsandarealsolocatedatmorefavourable locationsforbackscattering.InthisregarditisnotablethattheDFT cal-culationsindicatesignificantlateralrelaxationsoftheoutermostlayer Cuatoms,notonlyinthe[110]azimuth(intothespaceleftbythe miss-ingrowby0.08Å),butalsoalong[001]byalargervalueof0.20Å. TestsimulationsofthePhDspectraincorporatingthislarge[001]shift (thatwasnottestedintheoriginalPSOsearch)bymovingtheOatoms toretainthesameO–Cunearest-neighbourregistryledtoanincreasein theR -factor,andthusalesssatisfactoryagreementwiththePhD exper-imentalspectra

WethereforeconcludethatthePhDbest-fitstructureofFig.11 does representaclear(local) R -factorminimuminparameterspacethatis qualitativelysimilartothelowestenergystructure foundin theDFT calculationsbutdoesinvolvesomesignificantlydifferentstructural pa-rametervalues,atleastoneofwhich(theC–Cdistance)isunphysical. TherearealmostcertainlyotherlocalR -factorminimaatother param-etervalues,butwehavebeenunabletofindonewithrealisticvaluesof theparametertowhichtheOPhDismostsensitive,namelytheCu–O bondlength.DespitetheunphysicalC–Cbondlengthvalue,itseems ex-tremelylikelythatthisisthequalitativelycorrectstructuralmodel.The reasonsunderlyingthefailureofthePhDtechniquetoprovideatotally convincingfinalquantitativestructuralsolutiontothisparticular prob-lemareunclear.Theweakmodulationsandthuspoorersignal-to-noise ratioarecertainlyafactor, andcanbe attributedtothefactthatall theOatomsareinlowsymmetrysites,displacedbyalmost1Å from theidealatopsitesrelativetosurfaceCuatoms.Thiscontrastswiththe situationforthestanding-upphasesofothercarboxylicacidsinwhich theOatomsaredisplacedfromatopsitesbyonly∼0.1Å.Notethatin searching foralternativestructuresthatmightprovide abetterfitto thePhDdataarangeofCuadatomstructureswerealsoinvestigated; Cuadatomshavebeensuggestedtoplayaroleinanumberof molecu-laradsorbatephasesoncoppersurfacesincludingbenzoateonCu(110)

[45].Althoughsomeofthesemodelswerefoundtohavetotalenergies in DFTcalculationscomparabletothatofthemissing-row structure, noneofthesemodelsledtoanimprovedfittotheexperimentaldata.

Ofcourse,wecannotexcludethepossibilitythattwoormore co-existentadsorption geometriesarepresent.OurSTMimagesindicate thatcoadsorptionofstanding-upmono-oxalatespeciesadsorbedonto thesecondlayeratthemissingrowlocations,canoccur,butPhD

(11)

ulationsbasedonsuchaco-occupationmodelledtothebestfitbeing foundwithzerooccupationofthisspecies.Moreover,theO1sXPSdata excludeanysignificantco-occupationofthisstanding-upspecies.Itis alsonotablethattheDFTcalculationsstronglysuggestalternative ad-sorptionsitesofthedioxalatewouldbeunstable.

5. Conclusions

Thecombination of(particularly O1s)SXPSandOK-edge NEX-AFShaveprovidedclearevidencefortwodifferent adsorbedspecies onCu(110)resultingfromexposure tooxalic acid.Atlow coverages adoubly-deprotonateddioxalatespecieslieswithitsmolecularplane approximatelyparalleltothesurface,whileathighcoveragesthe ad-sorbateisasingly-deprotonatedmonooxalatewithitsmolecularplane perpendiculartothesurfacelyinginthe[110]azimuth.STMandLEED showthatthehighcoveragephaseformsanordered(3×2)structure, theimpliedcoverageof1/6MLbeingconsistentwithSXPSdata.STM imagesalsoshowthepresenceof additionalfeatures,assumed tobe extraoxalicacid-derivedmolecules,thatoccupysitesofthecentreof the(3×2)unitmeshformedbythedioxalatespecies.Symmetry argu-mentsclearlyexcludethepossibilitythattheseadditionalmoleculesare dioxalatespeciesatequivalentlocalsites,buttheimagescouldbe con-sistentwiththepresenceofmissing[001]outermostlayerCurowswith monooxalatespeciesbondedtosecondlayerCuatomsinthelocation ofthemissingrows.

O1sPhDdatafromthehigh-coveragephasewasfoundtobe con-sistentwiththemonooxalatespeciesbondedthroughthecarboxylateO atomsinnear-atopsitesrelativetonearest-neighboursurfaceatomsin the[110]rows.Thislocalbondinggeometryisequivalenttothoseof for-mate,acetate,benzoateandsingly-deprotonatedtartaricacid.Finding amodelstructureforthedioxalateontheunreconstructedCu(110) sur-facefromtheO1sPhDdataprovedtobemorechallenging,atleastin partduetotheweakmodulations,particularlynearnormalemission. However,DFTcalculations,based onareconstructed(3×2) surface, witheverythird[001]Cusurfacelayerrowmissing,assuggestedby thepresenceof thecentredunit meshSTMimages, showedthatthe dioxalatebonds significantlymorestronglytothissurface.O1s PhD simulationsbasedonadsorptiononthis reconstructedsurface identi-fieda best-fitstructural modelqualitativelysimilar tothatfoundto havethelowestenergyinDFTcalculations.Specifically,themolecule isfoundtooccupyahollowsiteontheCu[001]doublerows,directly aboveasecondlayerCu atom,withtheC–C axisalongthe[110] az-imuth.However,somesignificantdifferencesinstructural parameter valuesarefoundbetween theDFT-andPhD-derivedmodels,withat leastoneintramolecularbondlength(C–C)beingunreasonablyshortin thePhD-derivedmodel.

All data presented in this paper are available at http://wrap. warwick.ac.uk/94089.

Acknowledgements

Theauthors thank Diamond Light Sourcefor accesstobeamline I09(proposalnumberSI8436)thatcontributedtotheresultspresented here. DAD would like to acknowledge funding from the Alexander von Humboldt Foundation. S.F. acknowledges the CINECA Award N.HP10CJCRO0,2011fortheavailabilityofhighperformance comput-ingresourcesandsupport.G.C.acknowledgesfinancialsupportfromthe EUthroughtheEuropeanResearchCouncil(ERC)Grant“VISUAL-MS” andfromtheEPSRC (EP/G043647/1).

References

[1]R. Han , F. Blobner , J. Bauer , D.A. Duncan , J.V. Barth , P. Feulner , F. Allegretti , Chem. Comm. 52 (2016) 9805 .

[2]I.E. Wachs , R.J. Madix , J. Catal. 53 (1978) 208 . [3]D.S.Y. Ying , R.J. Madix , J. Catal. 61 (1980) 48 . [4]M. Bowker , R.J. Madix , Surf. Sci. 102 (1982) 542 . [5]B.A. Sexton , Surf. Sci. 88 (1979) 319 .

[6]B.E. Hayden , K. Prince , D.P. Woodruff, A.M. Bradshaw , Surf. Sci. 133 (1983) 589 . [7]D.P. Woodruff, Surf. Sci. Rep. 62 (2007) 1 .

[8]D.P. Woodruff, C.F. McConville , A.L.D. Kilcoyne , Th. Lindner , J. Somers , M. Surman , G. Paolucci , A.M. Bradshaw , Surf. Sci. 201 (1988) 228 .

[9]D. Kreikemeyer-Lorenzo , W. Unterberger , D.A. Duncan , M.K. Bradley , T.J. Lerotholi , J. Robinson , D.P. Woodruff, Phys. Rev. Lett. 107 (2011) 046102 .

[10]B.A. Sexton , Chem. Phys. Lett. 65 (1979) 469 . [11]M. Bowker , R.J. Madix , Appl. Surf. Sci. 8 (1981) 299 . [12]S. Bao , G. Liu , D.P. Woodruff, Surf. Sci. 203 (1988) 89 .

[13]K-U. Weiss , R. Dippel , K-M. Schindler , P. Gardner , V. Fritzsche , A.M. Bradshaw , A.L.D. Kilcoyne , D.P. Woodruff, Phys. Rev. Lett. 69 (1992) 3196 .

[14]B.G. Frederick , M.R. Ashton , N.V. Richardson , T.S. Jones , Surf. Sci. 292 (1993) 33 . [15]B.G. Frederick , F.M. Leibsle , S. Haq , N.V. Richardson , Surf. Rev. Lett. 3 (1996) 1523 . [16]B.G. Frederick , Q. Chen , F.M. Leibsle , M.B. Lee , K.J. Kitching , N.V. Richardson , Surf.

Sci. 394 (1997) 1 .

[17]M. Pascal , C.L.A. Lamont , M. Kittel , J.T. Hoeft , R. Terborg , M. Polcik , J.H. Kang , R. Toomes , D.P. Woodruff, Surf. Sci. 492 (2001) 285 .

[18]J. Lee , O. Kuzmych , J.T. Yates Jr. , Surf. Sci. 582 (2005) 117 .

[19]C.C. Perry , S. Haq , B.G. Frederick , N.V. Richardson , Surf. Sci. 409 (1998) 512 . [20]Q. Chen , C.C. Perry , B.G. Frederick , P.W. Murray , S. Haq , N.V. Richardson , Surf. Sci.

446 (2000) 63 .

[21]D.B. Dougherty , P. Maksymovych , J.T. Yates Jr. , Surf. Sci. 600 (2006) 4484 . [22]S.M. Barlow , R. Raval , Surf. Sci. Rep. 50 (2003) 201 .

[23]M. Ortega Lorenzo , S. Haq , T. Bertrams , P. Murray , R. Raval , C.J. Baddeley , J. Phys. Chem. B 103 (1999) 10661 .

[24]B. Behzadi , S. Romer , R. Faseland , K-H. Ernst , J. Am. Chem. Soc. 126 (2004) 9176 . [25]V. Humblot , M.O. Lorenzo , C.J. Baddeley , S. Haq , R. Raval , J. Am. Chem. Soc. 126

(2004) 6460 .

[26]C. Roth , D. Passerone , L. Merz , M. Parschau , K-H. Ernst , J. Phys. Chem. C. 115 (2011) 1240 .

[27]Y. Wang , S. Fabris , T.W. White , F. Pagliuca , P. Moras , M. Papagno , D. Topwal , P. Sheverdyaeva , C. Carbone , M. Lingenfelder , T. Classen , K. Kern , G. Costantini , Chem. Commun. 48 (2012) 534 .

[28]D.S. Martin , R.J. Cole , S. Haq , Phys. Rev. B 66 (2002) 155427 .

[29]D.A. Duncan , W. Unterberger , D.C. Jackson , M.J. Knight , E.A. Kröger , K.A. Hogan , C.L.A. Lamont , T.J. Lerotholi , D.P. Woodruff, Surf. Sci. 606 (2012) 1435 . [30]J.-H. Kang , R.L. Toomes , M. Polcik , M. Kittel , J.-T Hoeft , V. Efstathiou ,

D.P. Woodruff, A.M. Bradshaw , J. Chem. Phys. 118 (2003) 6059 .

[31]D.I. Sayago , M. Polcik , G. Nisbet , C.L.A. Lamont , D.P. Woodruff, Surf. Sci. 590 (2005) 76 .

[32]C. Karageorgaki , K-H. Ernst , Chem. Comm. 50 (2014) 1814 . [33]C. Karageorgaki , D. Passerone , K-H. Ernst , Surf. Sci. 629 (2014) 75 . [34]D.S. Martin , R.J. Cole , S. Haq , Surf. Sci. 539 (2003) 171 .

[35]M.N. Faraggi , C. Rogero , A. Arnau , M. Trelka , D. Écija , C. Isvoranu , J. Schnacht , C. Marti-Gastaldo , E. Coronado , J.M. Gallego , R. Otero , R. Miranda , J. Phys. Chem. C 115 (2011) 21177 .

[36]S. Fortuna , Surf. Sci. 653 (2016) 41 .

[37]S.J. Clark , M.D. Segall , C.J. Pickard , P.J. Hasnip , M.J. Probert , K. Refson , M.C. Payne , Z. Krist. 220 (2005) 567 .

[38]Y. Yao , F. Zaera , Surf. Sci. 646 (2016) 37 .

[39]A.F. Carley , P.R. Davies , G.G. Mariotti , Surf. Sci. 401 (1998) 400 . [40]S. Grimme , J. Comput. Chem. 27 (2006) 1787 .

[41]J. Perdew , K. Burke , Y. Wang , Phys. Rev. B 54 (1996) 16533 .

[42]S. Scandolo , P. Giannozzi , C. Cavazzoni , S. de Gironcoli , A. Pasquarello , S. Baroni , Z. Krist. 220 (2005) 574 .

[43]P. Giannozzi , S. Baroni , N. Bonini , M. Calandra , R. Car , C. Cavazzoni , D. Ceresoli , G.L. Chiarotti , M. Cococcioni , I. Dabo , A. Dal Corso , S. de Gironcoli , S. Fabris , G. Fratesi , R. Gebauer , U. Gerstmann , C. Gougoussis , A. Kokalj , M. Lazzeri , L. Mart- in-Samos , N. Marzari , F. Mauri , R. Mazzarello , S. Paolini , A. Pasquarello , L. Paulatto , C. Sbraccia , S. Scandolo , G. Sclauzero , A.P. Seitsonen , A. Smogunov , P. Umari , R.M. Wentzcovitch , J. Phys. Cond. Mat. 21 (2009) 395502 .

[44]D.A. Duncan , J.I.J. Choi , D.P. Woodruff, Surf. Sci. 606 (2012) 278 .

References

Related documents

Studying numerical results of wear test determined that, in micro-structured hard-metal, weight loss was further than nano-structured hard-metal but mean friction coefficient

sulla base della nostra esperien- za di cinque anni la presenza di un medico (spesso come co-conduttore) è molto utile per discutere ad esempio dei sintomi

Proposed approach is compared with hierarchical patterns (to discover patterns) and several function approximation techniques. The Miner hybrid framework optimizes a fuzzy

.) The procedure then se- lects an unreduced syntactic redex and adds a de- scription of its reduct at a disjoint position. Then the solve procedure is applied to resolve the

Sowohl IgG- als auch IgM-AK, die hauptsächlich gegen das LPS von Leptospiren gerichtet sind, können nach einer überstandenen schweren klinischen Leptospirose noch

Among the LAMAs, umeclidinium showed statistically sig- nificant improvement in trough FEV 1 at week 12 compared to tiotropium and glycopyrronium, but the results were not

As one cools the sample for the first time to 10 K one observes that hydrogen is trapped at octahedral sites associated with two substitutional vanadium

Six steel- string acoustic guitars were built for this study to the same design and material specifications except for the back/side plates which were made of woods varying widely