ChineseJournalofCatalysis36(2015)181–187 

a v a i l a b l e   a t  www. s c i e n c ed i r e c t . c om  

j o u r n a l   h omep a g e :  www. e l s e v i e r. c om / l o c a t e / c h n j c

Article   

Afirstprinciplesstudyoftheenergeticsandcorelevelshiftsofanion‐dopedTiO2photocatalysts

WuchenDing,WeixueLi*StateKeyLaboratoryofCatalysis,DalianInstituteofChemicalPhysics,ChineseAcademyofSciences,Dalian110623,Liaoning,China

A R T I C L E I N F O  

A B S T R A C T

Articlehistory:Received16May2014Accepted9June2014Published20February2015

  We present a comprehensive and improved density functional theory (DFT) calculation of ani‐on‐doped(anion=B,C,N,F,P,S)anataseandrutileTiO2.Thefirstpartisafirstprinciplescalcula‐tionofthecorelevelshifts(CLS)forvariousaniondopantsinbothanataseandrutileTiO2.TheCLSresultsrevealed that interstitialNhadahigherN1sbindingenergythansubstitutionalN,whichagreedwellwithexperimentalresults.ThecalculationalsoshowedthatforB‐,C‐,S‐,andP‐dopedTiO2,theinterstitialdopanthadanenergythatishigherthanthatofasubstitutionaldopant,whichis similar toN‐dopedTiO2.However, forF‐dopedTiO2, theenergyof the substitutionaldopant ishigher,andthisisirrespectiveoftheTiO2crystallography.Wealsocalculatedtheenthalpyofdopingand found that the substitutional dopant had a higher enthalpy than the interstitial dopant. Theresultsrevealedthatsubstitutionaldopingrequiredsevereexperimentalconditions,whereasinter‐stitialdopingonlyrequiresmodestwetchemistryconditions.

©2015,DalianInstituteofChemicalPhysics,ChineseAcademyofSciences.PublishedbyElsevierB.V.Allrightsreserved.

Keywords:TitaniaCorelevelshiftsX‐rayphotoelectronspectroscopyAnionSubtitutionaldopingInterstitialdopingDensityfunctiontheoryThermodynamics

 

 

1. Introduction

In 1972, Fujishima et al. [1] discovered photoelectronchemical splitting of water on TiO2 electrodes,which led toextensiveinvestigationsofTiO2asaphotocatalyst.TiO2isoneof the most promising photocatalysts for the photocatalyticdegradationof organicpollutantsandphotocatalyticdissocia‐tionofwater.However,thewidebandgapofanataseTiO2(3.2eV)limitsitsphotocatalyticapplicationstotheultraviolet(UV)lightrange.Meanwhile,therelativelyhighrateofelectron‐holerecombination results in a low quantum yield and poor effi‐ciency in promoting photocatalytic reactions. These funda‐mental problems prevent the use of anataseTiO2 in practicalapplications.Therefore,oneof theendeavoursto improvetheperformance of TiO2 is to increase its optical efficiency by

shiftingtheonsetofitsresponsefromtheUVtothevisiblere‐gion[2–5].OnestrategyistodopeTiO2withananionsuchasB,C,N,F,P,orStodecreaseitsbandgap.

Comparedtootheraniondopants,Nhasbeenproventobeoneofthemostefficient forvisible light‐responsiveTiO2pho‐tocatalysts[3,5],andthissystemhasbeenstudiedextensivelybyexperiments.Variousmethodshavebeendevelopedtopre‐pareN‐dopedTiO2photocatalysts(powdersandfilms),suchassputtering[3,6–8], ionimplantation[9–12],controlledhydrol‐ysis or sol‐gel [13–24], and chemical treatment of TiO2[3,9,12,25–30]. Although these N‐doped TiO2 photocatalystsare visible light‐active, there are debates on their structures,particularlyonthelocationoftheactiveNatomsandtheoriginofthebandgapnarrowinginN‐dopedTiO2.Asahietal.[3]in‐dicatedthatNatomssubstitutedforOatomsinN‐dopedTiO2

*Correspondingauthor.Tel/Fax:+86‐411‐84379996;E‐mail:wxli@dicp.ac.cn ThisworkwassupportedbytheNationalNaturalScienceFoundationofChina(21173210,21225315,21321002),theNationalBasicResearchPro‐gramofChina(973Program,2013CB834603),andtheStrategicPriorityResearchProgramoftheChineseAcademyofSciences(XDA09030000).DOI:10.1016/S1872‐2067(14)60165‐0|http://www.sciencedirect.com/science/journal/18722067|Chin.J.Catal.,Vol.36,No.2,February2014

182 WuchenDingetal./ChineseJournalofCatalysis36(2015)181–187

andthedecreaseofthebandgaporiginatedfromthemixingoftheO2pandsubstitutionalN2p states.Diwaldetal. [9] con‐cluded thatNatomscanalsobe locatedat interstitial sites inN‐doped TiO2, and they further confirmed that this nitrogenstatewasresponsiblefortheobservedshiftofthephotochem‐icalthresholdofrutileTiO2(110)downto2.4eV.Irieetal.[12]proposed that the isolatednarrowN2pband formedby sub‐stitutionalNinN‐dopedTiO2abovetheO2pvalencebandwasresponsibleforthevisiblelightresponse.Serpone[31]arguedthatthevisiblelightactivationofN‐dopedTiO2wasduetode‐fects associatedwithoxygenvacancies that gave rise to colorcenters.Recently,Livraghietal.[19]investigatedN‐dopedTiO2by a combined experimental and theoretical approach andsuggestedthatNbcenters(singleatomnitrogeninthebulkofTiO2)playedanessentialroleintheabsorptionofvisiblelight,in the promotion of electrons to the conduction band, and inphotoinducedelectrontransfertoreducibleadsorbates.Sofar,themostcommontechniquetodetectdopedNatomsinTiO2isby X‐ray photoelectron spectroscopy (XPS). N 1s XPS peakswith different binding energies have been reported forN‐dopedTiO2preparedbydifferentmethods,buttheirassign‐mentsarestillnotconclusive.ComparedwithNinTiN,theN1sXPS feature with its binding energy at 397.0 eV is generallyassigned to N2– anions that are substituted for O in the TiO2lattice.However,Chenetal.[32]attributedaN1speakwithitsbinding energy at 401.3 eV to substitutionalN inO–Ti–N.Di‐waldetal.[9]preparedN‐dopedrutileTiO2(110)byheatinginNH3andobservedtwoN1speakswithbindingenergyat396.7and399.6eV,whichwereassignedtosubstitutionalandinter‐stitialNatoms,respectively.Rodriguezetal.[33]reportedtheN1sbindingenergyofatomicNadsorbedonrutileTiO2(110)tobe399.0eV.Therefore, a clearunderstandingof thestruc‐tures of N‐doped TiO2photocatalysts is of great importance.Recently,othernonmetal‐dopedTiO2withdopantslikeB,C,orFhave also received growing interest fromboth experimentsandtheoreticalcalculations[19,34–46].

Intheoreticalstudies,DiValentinetal. [19,36,41–46]havepublishedanumberofpapersonnonmetal‐dopingofTiO2. Intheirwork,theyperformedalotofelectronicstateanalysisonthelocalizedstatesformedinthebandgapofTiO2.Inaddition,theyalsoperformedcorelevelshift(CLS)studiesfordifferentdoping positions. Their CLS value of different N locations inN‐dopedanataseTiO2is1.6eVwiththeinitialstateapproxima‐tion.Theyalsoshowedintheircalculationsthatthereexistdadifferenceof2.6eVbetweeninterstitialBandsubstitutionalBwith initial state relaxation effects considered [36]. However,their CLS resultsmaybemore convincing if theyhad consid‐eredfinalstaterelaxationintheircalculations.

The locations of anions and thermodynamic stability aretwoaspectsofmostconcerntoexperimentalists.Therefore,inthis work, we comprehensively studied the CLS and thermo‐dynamicstabilityfordifferentdopinganions(B,C,N,F,P,andS), differentdoping types (substitutional and interstitial), anddifferent TiO2 phases (rutile and anatase). In the energeticscalculations,especiallywithN‐doping,wehaveconsideredthedifferencebetweenthePBEfunctionalandhybridHSE06func‐tional.

In the firstpartof thework,wedeterminedtheoptimizedstructuresoftheanion‐dopedTiO2.Theninthesecondpart,weemployed standardandhybriddensity function theory (DFT)calculations for an energetics study of anion‐doped TiO2.Westudythethermodynamicsinanefforttofindthefundamentalreason for the differentN‐doped TiO2 types identified by thedifferentpreparationmethods.Althoughtherearelessexperi‐mentalstudiesthanonN‐dopedTiO2,someconclusionsforthesystemsofB‐,C‐,F‐,P‐,orS‐dopedTiO2canbegiven.Inthelastpart,DFTcalculationswithfinalstaterelaxationeffectsinclud‐edwere employed to provide comprehensive and precise in‐sightintotheCLSresultsofanion‐dopedTiO2.Inthispart,westudieddifferentdopinglocations,anions,andTiO2phases.WehopetheseCLScalculationshelpdistinguishtheclosepeaksintheXPSspectra.

2. Theoreticalapproachesandcomputationaldetails

Spin‐polarized total energy calculations were performedbased on the all‐electron projected augmented wave (PAW)method and DFTwithin the generalized gradient approxima‐tion (GGA‐PBE) using a HSE06 hybrid functional as imple‐mentedintheViennaAbInitioSimulation5.2Package(VASP)[47–54].Acutoffof400eVwasusedfortheplanewaveexpan‐sion.Throughoutthepresentwork,optimizedequilibriumlat‐ticeconstants(anataseTiO2:a=3.82Å,c=9.62Å;rutileTiO2:a=4.66Å,c=2.97Å),whichagreedwellwithpreviouscalcula‐tions[45]andexperiments(anataseTiO2:a=3.78Å,c=9.52Å;rutileTiO2:a=4.59Å,c=2.96Å)[5],wereusedunlessother‐wisestated.TostudyaniondopingintheTiO2bulk,oneanionatomwas included in the anatase (3×3×1) supercell with di‐mensions of 11.46 Å × 11.46 Å × 9.62 Å and rutile (2×2×3)supercellwith dimension of 9.32 Å × 9.32 Å × 8.91 Å,whichcorresponded to thedoping concentrationof 1.4%and2.1%,respectively.FortheBrillouinzoneintegration,weemployedaMonkhorst‐Pack(2×2×4)and(4×4×4)Γ‐centeredk‐pointgridfor the anatase (3×3×1) and rutile (2×2×3) supercell. All theions in thesupercellwererelaxeduntil theresidual forceoneachionwaslessthan0.01eV/Å.

CLSwere calculated as the energy difference between thecore level binding energy of the atom of interest (Eicl) and areferenceatom(Erefcl),i.e.,CLS=Eicl–Erefcl.Thecorelevelbind‐ingenergy (Ecl) canbe calculated inboth the initial state andfinalstateapproximations.Intheinitialstateapproximation,Eclis the negative eigenvalue (–ɛc) of the orbital fromwhich thecoreelectronhasbeenexcited,whereEcl=–ɛc,andɛcreferstotheFermilevelforthesolidandvacuumforthegas.Inthefinalstateapproximation,Eclwascalculatedbythetotalenergydif‐ferencebetweenthetwoconfigurations,whichwereanexcitedconfigurationinwhichanelectronwasremovedfromapartic‐ularcorestate,Etot(nc–1)andthegroundstateconfiguration,Etot(nc).Thus,Ecl=Etot(nc–1)–Etot(nc),inwhichthecore‐holewasassumedtoremainlocalizedintheexcitedatom.

In thiswork, for the final state calculation, a core electronwas removed from the core by generating the correspondingcore excited ionic PAW potential in the calculation, and theremainingcorestateswererelaxedintheself‐consistentcalcu‐

WuchenDingetal./ChineseJournalofCatalysis36(2015)181–187 183

lation.

3. Resultsanddiscussion

3.1. Optimizedstructuresofanion‐dopedTiO2

Inthisworkbothanionsubstitutionandinterstitialdopingwereconsidered.Asall theOatomsare in thesameposition,thereisonlyonepossibilityfortheanionsubstitutionlocation.However,fortheinterstitial‐dopedanion,therearethreepos‐sible equivalent positions in anatase TiO2 and two possiblepositionsinrutileTiO2,asshowninFig.1.Theoptimizedcoor‐dinateoftheinterstitialNatomsineachstructureandthetotalenergyoftheoptimizedstructurearesummarizedinFig.2.ForinterstitialN‐doped anatase, the A3 typeexhibited the lowesttotal energy, in agreementwith the previous result [45]. Forinterstitial N‐doped rutile, the R2 type ismore stable. There‐fore,thea2typeandtheR2typewerechosenasthemodelsforthe further theoretical investigations of interstitial N‐dopedanataseandrutile,respectively.Similarworkwasdoneontheotheranion‐dopedTiO2.TheresultsarelistedinTable1.Itcanbe seen that the location depended on the size of the atomicradius. Larger anions prefer the larger interspace. ComparedwithbulkTiO2, the atoms surrounding thedopedanionwereslightlyrelaxedintheanion‐dopedTiO2.

3.2. Energeticsstudyofanion‐dopedTiO2

Tocomparethereactionenthalpyof thereactionsproduc‐inginterstitialandsubstitutionalanion‐dopedTiO2,weconsid‐eredthereaction,whichisthatofoneanionwithTiO2topro‐duce anion‐doped TiO2. The source of the anion atomwouldoffsetintheenthalpycomparisonoftwotypesofanion‐dopedTiO2, thus for interstitial doping, there is only one process inwhichanionsgetintothefavorableinterstitialspaceoftheTiO2lattice. However, for substitutional doping, there are threeprocesses inwhich O atoms first are removed from the TiO2latticeleadingtotheformationofoxygenvacancies,andanionsfillthesevacanciesatthesametime.AsthesourcesoftheanionandfinalformoftheOatomarenotunique,thedopingreactioncanbedescribedasthefollowingforanataseTiO2onthebasisofthestructuralmodel:Interstitialanion‐dopedreaction

A + Ti36O72 = Ti36O72A + H1 (1)Substitutinganion‐dopedreaction

A+Ti36O72=Ti36O71A+O+H2 (2)Ti36O72 = Ti36O71 + O + H3 (3)

A + Ti36O71 = Ti36O71A + H4 (4)For rutile TiO2, the doping reaction equations are the fol‐

lowing:Interstitialanion‐dopedreaction

A + Ti24O48 = Ti24O48A + H1 (1)Substitutinganion‐dopedreaction

A+Ti24O48=Ti24O47A+O+H2 (2)Ti24O48=Ti24O47+O+H3 (3)

A+Ti24O47=Ti24O47A+H4 (4)H was calculated by the equation H = ΣE(products) −

ΣE(reactants),inwhichErepresentsthetotalenergy.Reaction(2)isnotthesameasreaction(1),whichhassta‐

bleproducts,sothefinalformoftheOatomshouldbeincludedin reaction (2). Inotherwords, there shouldbe an additionalenthalpyH5addedtoH2inthecaseofthecomparisonwithH1,whereH5istheenthalpyoftheprocesswheretheOatomchangedintothefinalform,whichisdeterminedbythepartic‐ularreactionroute.

Toidentifytheerrorfromusingdifferentmethods,boththePBE functional and HSE06 hybrid functional were used forN‐dopedTiO2,aslistedinTable2.Itcanbeseenthattheener‐

(a) (b)

A1

A2A3R2

R1

Fig.1.DifferentinterstitialsitesinanataseTiO2(a)andrutileTiO2(b).

-1

0

1

2

3

4

0.55

3.00

R2

R1

A3

A2

A1

E (

eV)

3.73

Fig.2.EnergydifferenceinN‐dopedTiO2aftercrystalrelaxation.

Table1 Interstitial doping anion sites used in the calculations andPauli elec‐tronegativityoftheatoms.

AtomAnataseTiO2

RutileTiO2

Atomicradius (pm)[55]

Pauli electronegativity[56]

B A1a R2 85 2.04C A2 R2 70 2.55N A2 R2 65 3.04F A2 R2 50 3.98P A1 R2 100 2.19S A2 R2 100 2.58O 60 3.44Ti 140 1.54aLocationoftheanion(fromFig.1).

184 WuchenDingetal./ChineseJournalofCatalysis36(2015)181–187

getics did not exhibit a strong dependence on the functionalusedinthisstudy.BecausetheHSE06functionalismoretimeconsumingthanthePBE,theotheranion‐dopedTiO2energet‐ics calculationswere performedwith the standard PBE func‐tional.TheresultsareshowninFig.3andFig.4.Astheanionsdonotappearinreaction(3),thevaluesofH3donotchangefor the same system,which are 7.11 and 7.20 eV for anataseand ruile TiO2, respectively. The substitutional anion‐dopedreactionwasstronglyendothermicforbothanataseandrutileTiO2,buttheinterstitialanion‐dopedreactioncanbeexother‐micorslightlyendothermic.Thereexistedalargeenergybar‐rierforsubstitutionalN‐dopedTiO2forbothanataseandrutileTiO2, and thevalueofH3was a quitepositivenumber com‐paredwithH1.Thelargerenergyvaluecanexplainwhysub‐stitutionalN‐dopedTiO2usuallyrequiresseverereactioncon‐ditions,suchasionimplantationandsputtering,ornitridizingTiO2 in N2 or NH3 at high temperatures, whereas interstitialN‐doped TiO2 can be prepared via mild wet chemistry reac‐tions. However, this comparison cannot provide insights into

whichproduct is formedatagivenseverereactioncondition,andtheprocessofhowtheOatomchangedintoitsfinalformtherefore shouldbe taken into account. In the followingpart,weusedO2asthefinalformoftheOatomtoaddressthisissue.ThetotalenergyofO2wasalsoincludedinourwork,withtheenergyof aO2E(O2)=–9.857eVanda singleOatomE(O)=–1.90 eV. Thus another reaction is O = 0.5O2+H5, with thevalueofH5=–3.02eV.H5shouldbeaddedtoH2,andthencomparedwithH1.Fig.3 showsthat forB‐orP‐dopedTiO2,interstitialdopingwasstillmorefavorablethansubstitutionaldoping,whichmeansthatundersevereconditionsandwiththefinalformofO,interstitialB‐orP‐dopedTiO2isstillmoreeasilyprepared than substitutional B‐ or P‐doped TiO2. N‐, C‐, orS‐dopedTiO2tendedtohaveasimilarenthalpychange,whileF‐dopedTiO2showedthereversed trend.As thevalueofH5dependedonthe final formof theOatoms,differentprepara‐tionsolutionscanleadtodifferentmixturesof interstitialandsubstitutional doped products. In experimental preparationswhen severe conditionswere provided for N doping of TiO2,therewerealways twoN1speaks in theXPSspectra[3,9,12]whereaswhen theconditionsusedweremild, therewasonlyone peak [23,24]. From the energetics study,we can identifythatwithamildenvironment,onlyinterstitialN‐dopedTiO2isproduced, and both types ofN‐dopedTiO2 existwhen severeconditionsareused.

3.3. CLSstudyofanion‐dopedTiO2

3.3.1. CLSstudyofinterstitialandsubstitutionaldopingForthecoreleveloftheaniondopedintoTiO2,thetwodif‐

ferent types of doping: substitutional and interstitial dopingshowed two closepeaks in theXPS spectra.These twopeakscannotbedistinguishedjustbytheexperimentaldata,andtheCLSbyaDFTstudycanbeusedto identify theseclosepeaks.Wehave found that themost stable location is the interstitialsite.Then,weperformedaCLSstudyofthetwodifferentdop‐ingtypeswithstandardDFTforB‐,C‐,N‐,F‐,P‐,andS‐dopedTiO2.TheresultsareshowninFig.5andTable3.

FromFig.5,wecanseethatforB‐,C‐,N‐,F‐,P‐,andS‐dopedTiO2,thereexistsacleardifferencebetweensubstitutionalandinterstitial doping.Most interstitial anion dopants type had ahighercorelevelbindingenergythanthatofthesubstitutionaldopant, except for F‐doped TiO2. For N‐doped TiO2, it was

-9

-6

-3

0

3

6

9

-0.830.06-0.08

-6.45

-0.18

1.59

4.60

1.14

7.11

3.102.32

SPFC N

H

(eV

)

H2 (substitutional) H1 (interstitial)

E(OV):H3 = 7.11

B

Anatase TiO2

5.48

Fig.3.Comparisonofenthalpychangeforanion‐dopedanataseTiO2.

-9

-6

-3

0

3

6

9

0.94

0.66

-1.02

-9.40

-1.72

1.97

4.80

1.16

6.70

2.913.44

SPFC N

H (

eV)

H2 (substitutional) H1 (interstitial)

E(OV):H3 = 7.20

B

Rutile TiO2

7.20

Fig.4.Comparisonofenthalpychangeforanion‐dopedrutileTiO2.

Table2 Energetics comparison of N‐doped TiO2 obtained with the PBE andhybridHSE06functionals.

TiO2PBE/eV HSE06/eV

ΔH1 ΔH2 ΔH3 ΔH4 ΔH1 ΔH2 ΔH3 ΔH4

AnataseTiO2(3×3×1)

–0.82 3.11 7.11 4.00 –0.08 3.73 7.26 3.53

RutileTiO2(2×2×3)

0.94 2.90 7.20 4.30 2.22 3.30 5.86 2.56

-2

0

2

4

6

8 C 1s

A.

CL

S (

eV)

substitutional interstitial

R.A.R. A.R.A.R. A.R.A.R.

S 2pP 2pF 1sB 1s N 1s

Fig.5.CLSresultsofanion‐dopedanataseTiO2andrutileTiO2.

WuchenDingetal./ChineseJournalofCatalysis36(2015)181–187 185

foundthattheN1scorelevelbindingenergyoftheinterstitialNspecieswashigherthanthatof thesubstitutionalNspeciesforbothN‐dopedanataseTiO2andrutileTiO2.TheCLSswere2.52and2.17eVforN‐dopedanataseandrutileTiO2,respec‐tively. Experimentally, N 1s XPS peakswith different bindingenergies have been observed by XPS for N‐doped TiO2 pre‐paredbydifferentmethods.Asahietal. [3] assigned theN1speak with a binding energy at 396 eV to substitutional N inN‐dopedanataseTiO2preparedbysputteringaTiO2targetinaN2atomsphere.Chenetal.[32]attributedtheN1speakwithabindingenergyat401.3eVtosubstitutionalNinN‐dopedana‐taseTiO2preparedbytreatingTiO2withanexcessoftriethyl‐amine.Diwaldetal.[9]preparedN‐dopedrutileTiO2(110)byheatinginNH3andobservedtwoN1speakswithbindingen‐ergiesat396.7and399.6eV,whichwereassignedtosubstitu‐tional and interstitial N atoms, respectively. Fang et al. [23]assigned theN 1s peakwith a binding energy at 399.6 eV tointerstitialNinN‐dopedanataseTiO2preparedbyawetchem‐istrymethod.ComparingtheNinTiNwithaN1sbindingen‐ergyat397eV[57],ithasbeengenerallyacceptedthattheN1speakwithabindingenergyat396–397eVisfromsubstitution‐alNinN‐dopedTiO2.However,theevidencefortheassignmentoftheN1speakwithabindingenergyat399.6eVtointerstitialNinN‐dopedTiO2is lacking.Ourtheoreticalcalculationsfullysupport the assignments: substitutional N in N‐doped TiO2exhibitsN1sbindingenergybetween396and397eVandin‐terstitialNinN‐dopedTiO2exhibitstheN1sbindingenergyat399.6eV.OurresultsalsosuggestedtheassignmentoftheN1speakwithabindingenergyat401.3eV to substitutionalN inN‐dopedanataseTiO2alreadymadebyChenetal.[32].Mean‐while,onbasisofthecalculatedenthalpyfortheN‐dopingre‐action,itisalsoreasonabletoconcludethattreatingTiO2withan excess of triethylamine that was employed by Chen et al.[32]shouldnotbeabletoformthesubstitutionalNspecies.

3.3.2. CLSstudyofthedifferentanionsWe summarized the CLS results of different anion‐doped

TiO2inFig.5andTable3.OurcalculationsshowedthattheECLSvalueofanion‐dopedTiO2waspositiveexceptforF‐dopedTiO2.Thismeans that in the XPS spectra of anion‐doped TiO2, theinterstitialdopingpeakislocatedhigherthanthatofsubstitu‐tionaldoping.TheCLSresults(Fig.5)showedthattheFatomgaveareversetrendcomparedwithotheranions.ThiscanbeunderstoodthroughthePaulielectronegativityoftheseatoms[56] (Table1). Theelectronegativity valueofO is larger thanthoseofB,C,N,P,andS,whilesmallerthanthatofF.AfteranO

atom is substitutedby an anion, theTi atomsare closer thanwiththeOatoms,whileforinterstitialdoping,theOatomsarecloserthantheanion.Thedifferenceinthestructurebetweensubstitutional and interstitial doping is that more nearby Oatoms surround the anion in interstitial doping. According tothesequenceof theelectronegativity [56], there ismoreelec‐tron transfer fromO to F in interstitial doping than substitu‐tionaldoping;whileforB‐,C‐,N‐,P‐,orS‐dopingofTiO2,thereis the reverse electron transfer direction, and substitutionaldoping gets more electrons than interstitial doping. As weknow,whenmoreelectronsaremovedaway, thebindingen‐ergywouldbelarger.Thus,theB,C,N,P,andSatomsininter‐stitialdopingsiteshaveahighercorelevelenergythanthoseinsubstitutional doping sites; and F‐doped TiO2 has a reversetrend.When this conclusion is applied to theXPS spectra, forB‐,C‐,N‐,P‐,andS‐dopedTiO2,thepeakfrominterstitialdopingishigherthanthatfromsubstitutionaldoping;andforF‐dopedTiO2,thatfromsubstitutionaldopingishigherthaninterstitialdoping.

3.3.3. CLSstudybetweenthedifferentphasesofTiO2WehavestudiedtheanionCLSforboththerutileandana‐

tase,asshowninFig.5andTable3.Theresultsrevealedthatbothanataseandrutileshowedthesamesequenceofthesub‐stitutionalandinterstitialdopantpeaks.However,forthesamedopant,thecorelevelbindingenergyoftheanioninanataseisa littlehigherthanthat inrutile.Theprobablereasonmaybethat thedensityof rutile is larger than thatofanatase.As thedistancebetweentwoneighboringatomsisless,theinteractionbetweenTiandtheanionisstronger,thus,theanioncanacceptmoreelectronsinrutilethananatase.Asdiscussedabove,whenmoreelectronsarelost,thecorelevelbindingenergyishigher.Therefore,thebindingenergyoftheanioninanataseishigherthaninrutile.

4. Conclusions

WehaveperformedcomprehensiveandimprovedDFTcal‐culationsofanion‐dopedanataseandrutileTiO2.TheN1scorelevel binding energy calculations suggested that interstitial NspeciesexhibitahigherN1sbindingenergythansubstitutionalN species inN‐dopedTiO2. The calculated core level shift be‐tweeninterstitialandsubstitutionalNspeciesagreedwellwithexperimentaldata,whichstronglysupportstheassignmentsofthe experimental XPS results. Our calculations provide a fun‐damental understanding of N‐doped TiO2 photocatalysts. WealsoperformedCLScalculations forB‐,C‐,F‐,P‐, andS‐dopedTiO2andshowthatinF‐dopedTiO2,theinterstitialdopantpeakislowerthanthatofthesubstitutionaldopant.However,forB‐,C‐,N‐,P‐,andS‐dopedTiO2,thepeakoftheinterstitialdopantishigher than that of the substitutional dopant. The enthalpycalculationsdemonstratedthatthesubstitutionalanion‐dopedreactionisstronglyexothermicforbothanataseandrutileTiO2due to the intermediate production of oxygen vacancies,whereas the interstitial anion‐doped reaction ismodestly en‐dothermic.OurcalculationssuggestedthatforB,C,N,F,P,andS anion‐doped TiO2, severe experimental conditions are re‐

Table3 CalculatedB,C,N,andF1sandPandS2pCLSsinanion‐dopedanataseandrutileTiO2.Theenergyreferencesaresubstitutionalrutile(2×2×3)forB,C,N,P,andSandinterstitialrutile(2×2×3)forF.

TiO2CLS(eV)

B1s C1s N1s P2p S2p F1sAnataseTiO2(3×3×1)

interstitial 4.26 3.10 3.38 6.41 2.79 1.47substitutional 1.05 0.88 0.86 0.77 0.92 2.98

RutileTiO2(2×2×3)

interstitial 2.94 2.24 2.17 6.03 1.73 0substitutional 0 0 0 0 0 2.06

186 WuchenDingetal./ChineseJournalofCatalysis36(2015)181–187

quired for substitutional doping,while for interstitial doping,wetchemistrymethodswouldbeenough.

References

[1] FujishimaA,HondaK.Nature,1972,238:37[2] FujishimaA,RaoTN,TrykDA.JPhotochemPhotobiolC,2000,1:1[3] AsahiR,MorikawaT,OhwakiT,AokiK,TagaY.Science,2001,293:

269[4] KhanSUM,Al‐ShahryM,InglerWBJr.Science,2002,297:2243[5] ChenXB,MaoSS.ChemRev,2007,107:2891[6] Lindgren T, Mwabora J M, Avendano E, Jonsson J, Hoel A,

GranqvistCG,LindquistSE.JPhysChemB,2003,107:5709[7] Mwabora JM,LindgrenT,AvendanoE, JaramilloTF,Lu J,Lind‐

quistSE,GranqvistCG.JPhysChemB,2004,108:20193[8] NakanoY,MorikawaT,OhwakiT,TagaY.ApplPhysLett,2005,86:

132104[9] DiwaldO,ThompsonTL,GoralskiEG,WalckSD,Yates JT Jr. J

PhysChemB,2004,108:52[10] BatzillM,MoralesEH,DieboldU.PhysRevLett,2006,96:026103[11] Ghicov A,Macak JM, Tsuchiya H, Kunze J, Haeublein V, Frey L,

SchmukiP.NanoLett,2006,6:1080[12] IrieH,WatanabeY,HashimotoK.JPhysChemB,2003,107:5483[13] YuJG,YuJC,LeungMKP,HoW,ChengB,ZhaoXJ,ZhaoJC.J

Catal,2003,217:69[14] BurdaC,LouYB,ChenXB,SamiaACS,StoutJ,GoleJL.NanoLett,

2003,3:1049[15] SakthivelS,JanczarekM,KischH.JPhysChemB,2004,108:19384[16] GoleJL,StoutJD,BurdaC,LouYB,ChenXB.JPhysChemB,2004,

108:1230[17] WangZP,CaiWM,HongXT,ZhaoXL,XuF,CaiCG.ApplCatalB,

2005,57:223[18] LiHX,LiJX,HuoYN.JPhysChemB,2006,110:1559[19] LivraghiS,PaganiniMC,GiamelloE,SelloniA,DiValentinC,Pac‐

chioniG.JAmChemSoc,2006,128:15666[20] KischH,SakthivelS,JanczarekM,MitorajD.JPhysChemC,2007,

111:11445[21] CongY,ZhangJL,ChenF,AnpoM.JPhysChemC,2007,111:6976[22] WangJW,ZhuW,ZhangYQ,LiuSX. JPhysChemC,2007,111:

1010[23] FangJ,WangF,QianK,BaoHZ,JiangZQ,HuangWX.JPhysChem

C,2008,112:18150[24] HuoYN,BianZF,ZhangXY, JinY,ZhuJ,LiHX. JPhysChemC,

2008,112:6546

[25] JoungSK,AmemiyaT,MurabayashiM,ItohK.ApplCatalA,2006,312:20

[26] LiuG,LiF,ChenZG,LuGQ,ChengHM.JSolidStateChem,2006,179:331

[27] HeF,MaXF,LiT,LiGX.ChinJCatal,2013,34:2263[28] MohamedRM,AazamE.ChinJCatal,2013,34:1267[29] ZhouWQ,YuCL,FanQZ,WeiLF,ChenJC,YuJC.ChinJCatal,

2013,34:1250[30] LiuEQ,GuoXL,QinL,ShenGD,WangXD.ChinJCatal,2012,33:

1665[31] SerponeN.JPhysChemB,2006,110:24287[32] ChenXB,BurdaC.JPhysChemB,2004,108:15446[33] RodriguezJA,JirsakT,LiuG,HrbekJ,DvorakJ,MaitiA.JAmChem

Soc,2001,123:9597[34] LuN,QuanX,LiJY,ChenS,YuHT,ChenGH.JPhysChemC,2007,

111:11836[35] ZaleskaA,SobczakJW,GrabowskaE,HupkaJ.ApplCatalB,2008,

78:92[36] FinazziE,DiValentinC,PacchioniG.JPhysChemC,2009,113:220[37] Bettinelli M, Dallacasa V, Falcomer D, Fornasiero P, Gombac V,

MontiniT,RomanoL,SpeghiniA.JHazardMater,2007,146:529[38] ChenDM,YangD,WangQ,JiangZY.IndEngChemRes,2006,45:

4110[39] ZhaoW,MaWH,ChenCC,Zhao JC, ShuaiZG. JAmChemSoc,

2004,126:4782[40] MoonSC,MametsukaH,TabataS,SuzukiE.CatalToday,2000,58:

125[41] CzoskaAM,Livraghi S,PaganiniMC,GiamelloE,DiValentinC,

PacchioniG.PhysChemPhysChem,2011,13:136[42] DiValentinC,PacchioniG,OnishiH,KudoA.ChemPhysLett,2009,

469:166[43] DiValentinC,FinazziE,PacchioniG,SelloniA,LivraghiS,CzoskaA

M,PaganiniMC,GiamelloE.ChemMater,2008,20:3706[44] DiValentinC,FinazziE,PacchioniG,SelloniA,LivraghiS,Paganini

MC,GiamelloE.ChemPhys,2007,339:44[45] DiValentinC,PacchioniG,SelloniA,LivraghiS,GiamelloE.JPhys

ChemB,2005,109:11414[46] DiValentinC,PacchioniG,SelloniA.ChemMater,2005,17:6656[47] HeydJ,ScuseriaGE,ErnzerhofM.JChemPhys,2006,124:219906[48] HeydJ,ScuseriaGE.JChemPhys,2004,121:1187[49] HeydJ,ScuseriaGE,ErnzerhofM.JChemPhys,2003,118:8207[50] KresseG,HafnerJ.PhysRevB,1993,48:13115[51] KresseG,HafnerJ.PhysRevB,1994,49:14251[52] KresseG,FurthmullerJ.PhysRevB,1996,54:11169

GraphicalAbstract

Chin.J.Catal.,2015,36:181–187 doi:10.1016/S1872‐2067(14)60165‐0

Afirstprinciplesstudyoftheenergeticsandcorelevelshiftsofanion‐dopedTiO2photocatalysts

WuchenDing,WeixueLi*DalianInstituteofChemicalPhysics,ChineseAcademyofSciences

CorelevelshiftsinB‐,C‐,N‐,S‐,orP‐dopedTiO2aresuchthattheenergyofaninterstitialdopantishigherthanthatofasubstitu‐tionaldopant,butforF‐dopedTiO2,theenergyofthesubstitu‐tionaldopantishigher.

WuchenDingetal./ChineseJournalofCatalysis36(2015)181–187 187

[53] KresseG,FurthmullerJ.ComputMaterSci,1996,6:15[54] KrukauAV,VydrovOA,IzmaylovAF,ScuseriaGE.JChemPhys,

2006,125:224106

[55] SlaterJC.JChemPhys,1964,41:3199[56] AllredAL.JInorgNuclChem,1961,17:215[57] SahaNC,TompkinsHG.JApplPhys,1992,72:3072