and Rh with nanoparticles Au modified Photo-induced the reactions CO -methane in system on titanatenanotubes Applied Catalysis B: Environmental

Contents lists available atScienceDirect

Applied Catalysis B: Environmental

j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / a p c a t b

Photo-induced reactions in the CO ₂ -methane system on titanate nanotubes modified with Au and Rh nanoparticles

Balázs László

, Kornélia Baán

, Erika Varga

, Albert Oszkó

, András Erd ˝ohelyi

, Zoltán Kónya

, János Kiss

aDepartmentofPhysicalChemistryandMaterialsScience,UniversityofSzeged,Aradivértanúktere1.,SzegedH-6720,Hungary

bDepartmentofAppliedandEnvironmentalChemistry,UniversityofSzeged,RerrichBélatér1.,SzegedH-6720,Hungary

cMTA-SZTEReactionKineticsandSurfaceChemistryResearchGroup,RerrichBélatér1.,SzegedH-6720,Hungary

a r t i c l e i n f o

Articlehistory:

Received25March2016

Receivedinrevisedform20June2016 Accepted23June2016

Availableonline24June2016

Keywords:

Carbondioxidephotocatalysis Methanephotocatalysis Titanatenanotubes Rhodiumnanoparticles Goldnanoclusters

a b s t r a c t

Thephotocatalytictransformationofthemethane-carbondioxidesystemwasinvestigatedbyin-situ methodsinthepresentstudy.Titanatenanotube(TNT)supportedgoldandrhodiumcatalystswereused inthecatalytictests.Ourmaingoalwastheanalysisoftheroleofthecatalystsinthedifferentpartsofthe reactionmechanism.ThecatalystswerecharacterizedbyX-rayphotoelectronspectroscopy(XPS),high resolutiontransmissionelectronmicroscopy(HRTEM)anddiffusereflectanceUV–visspectroscopy(DR- UV–vis).Photocatalytictestswereperformedinacontinuousflowquartzreactorequippedwithmass spectrometerdetectorandmercury-arcUVsource.Diffusereflectanceinfraredspectroscopy(DRIFTS) wasusedtoanalyzethesurfaceofthecatalystduringphotoreaction.Post-catalytictestswerealsocarried outonthecatalystsincludingXPS,temperatureprogrammedreduction(TPR)andRamanspectroscopy methodsinordertofollowthechangesofthematerials.Titanatenanotubecanstabilizeeventhesmallest, molecular-likeAuclusterswhichshowedthehighestactivityinthereactions.Approximately3%methane conversionwasreachedinthebestcaseswhilethecarbondioxideconversionwasnottraceable.Itwas revealedthatwaterhasaveryimportantroleintheoxidationreaction.Themaindiscoveredreaction routesaremethanedehydrogenationandoxidation,themethylcouplingandtheformingofstructured carbondepositsonthecatalystsurface.ThesourceofthesurplusCOcanbemostlythereductionof carbondioxide.DuringthereductionprocessphotoelectronsandhydrogenionsbringsabouttheCO2

reductionviaCO2•−radicalanion.

©2016ElsevierB.V.Allrightsreserved.

1. Introduction

Inthepastyearsgreateffortsweremadeinthedryreforming ofmethanewithcarbondioxidetosyngas.Greenhousegases,pri- marilycarbondioxideandmethane,emittedbyhumanactivities contributetotheglobalwarming[1].Fromenvironmentalpoint ofview,themainadvantageofthisprocessistheutilizationand conversionofthetwomostdangerousgreenhousegases,CH4and CO2,intomorevaluablecompounds[2–9].BothCO2andCH4are stablemolecules,whicharenoteasytotransformintootherchem- icalsundermildreactionconditions.Theuseofphototechnology wouldbreakthethermodynamicbarrierofendothermicreactions buttheassistanceofsomeheatcanbestillnecessarytoworkout

∗ Correspondingauthorsat:MTA-SZTEReactionKineticsandSurfaceChemistry ResearchGroup,RerrichBélatér1.,SzegedH-6720,Hungary.

E-mailaddress:jkiss@chem.u-szeged.hu(J.Kiss).

thephotoreductionofCO₂[10].Generally,themainproductsofthe CO2+CH4reactionareCOandH2[10]buttheformationofacetone wasalsoreported[11].Recently,modifiedTiO₂ nanocomposites wereusedinphotocatalyticCO₂reductionbyCH₄[12–14].

Amongvarious semiconductors,titaniumdioxide(TiO2)asa photocatalysthasbeenresearchedexcessivelydue toitsadvan- tagessuchasrelativecheapness,availabilityinexcess,chemically andbiologicallystablecharacterandpossessionofhigheroxida- tivepotentials. UV-irradiationis abletogenerate electronsand holesinTiO₂,whicharegoodreductantsandpowerfuloxidants for redoxreactions[15–21].Duetoitsfavorable electronicand optoelectronicproperties,ithasbeenwidelyappliedtosolarcells andphoto-catalysts.However,improvedpropertiesarenecessary tomeet highdemandandcomplex requirements. Theprosper- ousdevelopmentoftitaniumdioxidenanomaterialshasthrived theinvestigationofaclassofTiO₂-basednanostructures;layered titanatematerials[22–24].Layeredtitanatematerialshaveattrac-

http://dx.doi.org/10.1016/j.apcatb.2016.06.057 0926-3373/©2016ElsevierB.V.Allrightsreserved.

tivefeaturesoftheirown,includingextremelylargeion-exchange capacity,fastiondiffusionandintercalation.

OnthebasisofthepioneeringworkofKasugaetal.[25]research effortsontitanateswereatfirstconcentratedonthehydrothermal synthesisandstructure elucidationoftitanatenanotubes(TNT).

Titanatenanotubesareopen-endedhollowtubularobjectsmea- suring7–10nminouterdiameterand50–170nminlength.They featureacharacteristicspiralcrosssectioncomposedof4–6wall layers.Thetypicaldiameteroftheirinnerchannelis5nm[25–27].

TitanateshaveageneralformulaasHxNa_2-xTi₃O₇·nH₂Oandtheir sodiumcontentcanbeloweredbyacidtreatments.Currentlyour interestisinthesodium-freeH2Ti3O7·nH2OformofTNTs.Titanate nanostructuresareofgreatinterestforcatalyticapplications,since theirhighsurfaceareaandcationexchangecapacityprovidethe possibilityofachievingahighmetal(e.g.Co,Cu,Ni,AgandAu)dis- persion[28–32].Rhinsmallsizescanbealsostabilizedintitanate nanotubesand,similarlytoAu,initiatesthetransformationfrom tubestructuretoanatasephase[33,34].

Numerousthermal-andphoto-inducedcatalyticreactionswere discovered on titanate supported metal catalysts up to now [22–24].The location of metal ions onthe nanocrystal surface mayproveimportantinmediatingelectrontransferreactionsthat haverelevanceinphotocatalysisorpowerstorage.Gold-containing titanatenanotubeswerefoundtodisplayhigheractivitythanthe DegussaP-25catalystinthephoto-oxidationofacetaldehyde[35], inthephotocatalyticdegradationofformic acid[36].Moreover, titanate-relatednanofibersdecoratedeitherwithPtorPdnanopar- ticlesshowsignificantphotocatalytic behavioras demonstrated bythedecompositionoforganicdyesinwater,thedegradation of organic stains on the surface of flexible freestanding cellu- lose/catalystcompositefilmandthegenerationofhydrogenfrom ethanolusingbothsuspendedandimmobilizedcatalysts.Theper- formanceofthenanofiber-basedcatalystmaterialscompeteswith theirconventionalnanoparticle-basedcounterparts[37–39].

Inthepresentstudyweinvestigatethephotocatalyticconver- sionof CO₂ andCH₄ over Auand Rhdopedtitanatenanotubes.

Wepayattentiontothesurfacestructureandopticalproperties ofnanoparticlesonnanotubes.DuringUVirradiationtheproduct distributionisdeterminedbymassspectrometryandthesurface intermediatesformedinphoto-inducedreactionsaredetermined byDRIFTS. We trytofindcorrelationbetween thestructure of nanoparticlesandthephotocatalyticactivity.

2. Experimental

2.1. Synthesisofthecatalyst

Thetitanatenanotubesweresynthesizedbyanalkalihydrother- mal method described previously [24,34,39,40]. The specific surfaceareaoftitanatenanotubesisapproximately185m²g⁻¹.

ForthesynthesisofgoldnanoparticledecoratedH₂Ti₃O7nano- tubes1goftheas-preparednanotubeswassuspendedin100ml distilled water by applyingultrasound irradiation for 1h. Then 5.2ml ofHAuCl₄ solutionwithanappropriate concentrationto provide∼1wt%goldloadingwasaddedtothewellhomogenized nanotubesuspension.After10minofstirring50mgofNaBH₄(sep- aratelydissolvedin5mlofdistilledwater)wasaddedrapidlyto achieve theinstantaneousformation of gold nanoparticles.The suspensionwaskeptstirredfor further20minthenwasrinsed withdistilledwaterthoroughly.Theas-purifiedsamplewasdried overnightin a temperatureprogrammedelectricovenat350K.

Usingthislow-temperatureAuloadingmethod,weescapedthe undesiredphasetransformationofnanotubestoanataseinitiated bygoldatelevatedtemperature(450–473K)[31].

Rh/TNT nanocomposite was produced by impregnating the titanatenanotubeswithRhCl3·3H2Osolutiontoyield1wt%metal content[32,33].Theimpregnatedpowderwasdriedinairat383K for3h.InordertogetmetallicRhthecatalystpassedoverfurther treatment(pre-treatment)justbeforethephotocatalyticmeasure- ments.Thepre-treatment consistedof 4sections: Annealingin oxygenflowfor1hat473K,flushingtheoxygenwithargonatthe sametemperature,reductioninhydrogenflowfor1hat523Kand finallyflushingthehydrogenwithargonfor1hat523K.

Au/TiO₂wasproducedbyimpregnatingTiO₂hombikatUV-100 (anatase phase) powderwith HAuCl4 solution.The preparation methodwasthesameasfortheRh/TNTcatalyst.Finally1wt%gold contentwasreached.Au/TNTandRh/TNTwith2.5%metalcontent werealsopreparedfortheUV–vismeasurementstoinvestigatethe concentrationdependenceofthebandgap.TheAu/TNTandRh/TNT notationsreferstothe1%metalcontentvariantshenceforward.

Itisimportanttoemphasize thatnocarboncontainingcom- poundwasusedatallduringthesynthesisofthecatalystsinorder toavoidanykindofincidentalcarboncontaminationinfiltratesinto thestructure.Thesekindsofcarboncontaminationscanresultsin surplusproductswhicharenotoriginatedfromthereactantshence resultsinmisleadingconversionsregardingtothecarbonbased reactants.

Thesurfaceareasofthecatalystsweremeasuredwitha‘BELCAT A’instrumentwithsinglepointBETmethod.Thesurfaceareasare 181,171,168and300m²/gforTNT,Rh/TNT,Au/TNTandAu/TiO₂, respectively.

Thephotocatalyticactivityofboththecompositesandthepure supporthad beeninvestigatedunder exactlythesamereaction parameters.The pre-treatmentprocess wasuniformfor allcat- alystsinordertogetbettercomparability:Themethodusedat Rh/TNTwasappliedinallcases.

2.2. Materials

Thepurityofthegasesusedforpretreatmentandfortheprepa- ration of the reactant mixtures were 99.5%, 99.995%, 99.995%, 99.996%and 99.999%for O2,CH4, CO2,Ar and H2 respectively.

Inthecaseofargonfurtherpurificationwasappliedwithanin- lineadsorptiontrapcontainingsilicageland5Azeoliteinorderto removewaterandcarbon-dioxidecontamination.

2.3. Characterizationofthecatalysts

XPspectraweretakenwithaSPECSinstrumentequippedwith aPHOIBOS150MCD9hemisphericalanalyzer.Theanalyzerwas operatedintheFATmodewith20eVpassenergy.TheAlK␣radi- ation(h=1486.6eV)ofadualanodeX-raygunwasusedasan excitationsource.Thegunwasoperatedatthepowerof210W (14kV,15mA).Typicallyfive scanswere summedtogeta sin- glehigh-resolutionspectrum. TheTi2p_3/2 maximum(458.9eV) wasusedasbindingenergyreference.Self-supportingpelletswere usedinXPSmeasurements.Forspectrumacquisitionandevalu- ationbothmanufacturer’s(SpecsLab2)andcommercial(CasaXPS, Origin)softwarepackageswereused.

The morphology of metal-modified titanate nanotubes was characterizedbyhighresolutiontransmissionelectronmicroscopy (FEITecnaiG²20X-Twin;200kVoperationvoltage,×180000mag- nification,125pm/pixelresolution).X-raydiffractometry(Rigaku MiniFlexII;CuK␣)andelectrondiffractionwereusedtodetermine thecrystallinityandthestructure.Themetalparticlesizedistri- butionwasdeterminedbyimageanalysisoftheHRTEMpictures usingthe‘ImageJ’software.Atleastfiverepresentativeimagesof equalmagnification,takenatdifferentspotsoftheTEMgridwere firstsubjectedtorollingballbackgroundsubtractionandcontrast enhancement,andthenthediameterofthemetalnanoparticlesin

theimagewasmanuallymeasuredagainstthecalibratedTEMscale bar.Thediameterdistributionhistogramwasconstructedfrom200 individualnanoparticlediametermeasurements.

DiffusereflectanceUV–vis spectroscopywasusedtoinvesti- gatethebandgapenergyofthecatalysts.Spectrawerecollected withahome-madefiberopticsystemconsistingofaMicropack HPX-2000lightsourceandanOceanOpticsUSB2000detector.The detectorhas2048pixelresolutioninthe200–1100nmwavelength range.Thefinalspectrumwasobtainedbyaveraging40scans.The spectrawereconvertedfromAbsorbance f_A(␭)typetoKubelka- Munk fKM(E)whereE=hinelectronvolts.Thenthespectrawere smoothedwiththeweightedmovingaveragemethod(51points, symmetric).Spectraldeconvolutionwasappliedinallcasesinorder togetclearpeakataround3.5eV.Simplexmethodwasusedtofit thespectrumwithgaussianfunctions.Themainrequirementfor thefittingwastogetatleast0.999forthevalueofR²whilekeepthe peaknumberaslowaspossible.Thebandgapwascalculatedfrom thegaussianpeakcenteredataround3.6eV.Thepeakwastrans- formedtoTauc-plotaccordingtothefollowingequation:Tauc-plot f_T(E)=(f_KM(E)*E)^1/nwheren=2forsemiconductorswithallowed indirectbandgaplikeTiO2[41].Thenalinearfunctionwasfittedto theleftinflectionpointofthef_T(E)curveanditsx-interceptyields theenergyoftheopticalbandgap(Eg).

2.4. In-situphotocatalyticmeasurements

The photocatalyticreactions wereperformed in a flow-type quartz reactor which consisted of cylindrical quartz and glass tubes.Animmersion-typemercury-arc lampplaced inthecen- terofthereactorwasusedforirradiation.Aheatabsorbingwater layerwasintroducedbetweenthefirstandsecondquartztubes tocutofftheinfraredradiationandtocooldownthemetalparts and thesealsofthereactor.Thefine catalystpowder wassus- pendedindeionizedwaterthenthemixturehadbeendriedonto theinnersurfaceofthethird(glass)tubeinsuchawaytocover approximately 430cm² area. Typically 0.5g catalyst was used.

Thevolumeofthereactorwas476cm³.Thereactantgaseswith controlledflow rates had beenintroduced betweenthesecond and third tubes. In this arrangement the catalyst surface faced towardsthelight sourcecancontactwiththegasesduringthe irradiation.Argonwasusedascarriergas.Theoverallflowrate was30cm³/mininallcases.Thereactantmixtureswereprepared pendingflowbymassflowcontrollers:methanewith0.9cm³/min flowratewasintroducedinto29.1cm³/minargonstreamtoget methane-argonmixture.0.9cm³/min methaneand 0.9cm³/min carbondioxidewereintroducedinto28.2cm³/minargonstream togetCH4 CO2 ArmixturewithCH4:CO2=1:1molarratio.Inthe caseofblankexperiments29.1cm³/min argonbubbledthrough waterat25^◦C toget0.9cm³/minpluswatervapor.Considering theoverall flowrateandthevolumeofthereactortheaverage retentiontimeofthegasmixturesisgenerally16min.Thetem- peraturecontrolofthecatalystwhichisnecessaryforthecorrect pre-treatmentisachievedwithanouterheaterbuiltfromaglass tube,someheaterwireandafeedbackthermocouple.Fig.1shows theschematicdrawingofthereactorandthesamplingsystem.

TheUVsourcewasan‘undopedTQ-718’highpressuremercury- arclamp(UV-ConsultingPeschl)operatedat500Wcontrolledby a‘P-EVG-10’ powersupply. Theirradiancereferstothecatalyst supportingsurfaceandwasmeasuredbya‘GentecUP19K-50L-H5- D0’powerdetectorwithaspectralrangeof0.19–20␮m.Asoda limeglasscutofffilterwasappliedtosplitthemeasurementrange.

Themeasuredirradianceswere0.143(±16%)and0.199(±6.5%) W/cm²inthe190–350and350–2000nmrange.Photonfluxwas calculatedforeachbandgapvalueinmolh⁻¹cm⁻².Thefirststepof thecalculationwasthenormalizationoftheemissionspectrumof thelampusingthemeasuredirradiancevalueofthe190–350nm

Fig.1. Theschematicofthephotoreactorwithsamplingpoints.

region.Afternormalizationthespectrumwasintegratedfromthe bandgapvalue(E_g)to6.2eV(200nm)togetthephotonflux.The photonfluxwastransformedfrommolh⁻¹cm⁻² tomolh⁻¹g⁻¹ usingthecatalystquantityandthesizeofthecoveredareawhich isslightlydifferentineachexperiment.Theoverallphotonconver- sionefficiency(␩)wascalculatedbythedivisionoftheformation rateofthemainproductwiththerespectivephotonflux.

Theproductsformedduringthephotocatalyticreactionswere analyzedwitha‘HidenHPR-70’gasanalysissystem.Itisequipped withanautomaticallycontrolled8-waybatchinletsamplingsub- system and a ‘HAL3F-RC’ quadrupole mass spectrometer with standard electron ionizersource.Separation technique wasnot usedinordertoachievehighsensitivitywiththemassspectrom- eter.Theinstrument’sdetectionlimitis500ppbtohydrogenand 100ppbtomethane.Samplesweretakenfromthegasflowinturns priorandafterthephotoreactorduringthereaction.The‘Multi- pleIonDetection’(MID)modewasappliedwiththefollowingm/z valuesselected:2,15,16,18,26,27,28,29,30,31,43,44,45.Thedif- ferenceofthem/zsignalsoriginatesfromthetwosamplingpoints wasusedhenceforwardinthecalculationsinordertominimize thenoiselevel.Thefragmentationpatternsandtheconcentrations werepreviouslycalibratedtotheexpectedproducts.Asmallvac- uumchamberequippedwithacapacitivegaugeandaleakvalve wasusedtopreparethedesiredconcentrationofagasneedtobe calibrated.Finalpressurewassettoatmosphericwithargon.Sam- plesweretakenfromthisstaticvolumebyathirdbatchinletport attachedtothechamber.Calibrationwasmadeatthemagnitude oftheexpectedconcentrationsforH₂,CH₄,N₂,O₂,CO,CO₂,C₂H₆ andmethanolseparately.One-pointcalibrationwasapplied.

Thephotocatalyticmeasurementsequenceconsistedofthefol- lowingsteps:pretreatmentofthecatalyst[Section2.2.],a6–9h baselinesection,3-hirradiation,3-hdarksection,thenrepeating theirradiationanddarksectionstwotimes.Theinsertionofthe darksectionswasnecessarytofollowtheadsorption-desorption processofthereactantsandtoeasethequalitativeanalysisofthe products.Thecoolantwaterlayerwasunabletoeliminatetheheat effectofthelampcompletelyatroomtemperature.Thetempera- tureofthecatalystwasapproximately403Kduringtheirradiation.

To minimize the temperature fluctuationbetween the UV and darksessionsthecatalystwaskeptat403Kinthedarksections too.Threetypesofreactionsweretestedphotocatalitically:The methanetransformation,theCH₄+CO₂reactionandablankexper- imentwithonlywaterpresentonargon.Theblankexperimentwas necessarytoverifythattheproductsdonotoriginatesfromsur- facecontaminations.Averageandmaximalformationrates(¯rand r)werecalculatedfromthemeasuredconcentrationsandknown parametersin␮molh⁻¹g⁻¹unitsforthemainproductsandreac- tantsregardingto9hirradiation.Thesignoftheratesispositive fortheformingproductsandnegativeforthewaningreactants.

Infraredspectroscopymeasurementswerecarriedoutwithan

‘AgilentCary-670’FTIRspectrometerequippedwith‘HarrickPray- ingMantis’diffusereflectanceattachment.Thesampleholderhad twoBaF₂windowsintheinfraredpathandaquartzwindowinthe UV-path.Afocusedmercuryshortarclamp(Osram,HBO100W/2) wasusedforUVirradiation.Thespectrometerwaspurgedwith drynitrogen.Typically16scanswererecordedataspectralresolu- tionof2cm⁻¹.Thespectrumofthepretreatedcatalystwasusedas background.Thesameexperimentalconditionswereusedasinthe photocatalyticmeasurements.TheUVirradiationwasintermitted duringthespectrumrecording.Thereactantswereflushedoutfrom thediffusereflectancecellwithheliumafteronehourirradiation.

Spectrawerecollectedafter30minflushingtoo.

2.5. Analysisoftheusedcatalyst

Post-catalyticmeasurementswereperformedinordertoinves- tigatethechangesoccurredinthecatalystduringthereaction.The usedcatalystwasremovedfromthereactorthenitwasanalyzed withfourdifferentmethods:Thequantityofthesurfacecarbon wasdeterminedwithtemperatureprogrammedreduction(TPR).

Ramanspectroscopymeasurementswerecarriedoutinorderto investigatethestructureofsurfacedeposits.X-rayphotoelectron spectroscopy(XPS)wasusedtoinvestigatetheoxidationstateof rhodiumandcarbononthesurfaceoftheusedcatalyst.

TheTPRmeasurementswerecarriedoutinthefollowingman- ner:Theusedcatalystwasplacedintoa10centimeterlongquartz tubeand heated up fromroom temperatureto 1173Klinearly at15K/minratein40ml/minhydrogenflow.Theproductswere analyzed with an ‘Agilent 7890’ gas chromatograph equipped with‘HPCarbonplot’capillarycolumn.Thermalconductivityand methanizer-sensitizedflameionizationdetectorswereused.

TheRamanspectraofthesamplesweremeasuredat532nm laserexcitationwith5mWpowerusinga‘ThermoScientificDXR RamanMicroscope’. Typically10 scansweremadewith2cm⁻¹ resolutionintherangeof100–1800cm⁻¹.

3. ResultsandDiscussion

3.1. CharacterizationofAuandRhnanoparticlessupportedon titanatenanotubes

Protonatedtitanatenanotubesdecoratedwithgoldnanoparti- cleswerecharacterizedbyXPS.TheXP-spectrumtakeninthegold 4fbindingenergyrangeispresentedinFig.2.Thefigureaddition- allyshowsthespectrumofaclean goldfilm(thickness:50nm) preparedonaglassplateforcomparison.Symmetric4f_5/2and4f_7/2 emissionswereobservedat87.7and84.0eVinbothcaseswhich isgeneralformetallicgold.Furthermore,ahigherbindingenergy peakappearedonAu/TNTwithAu4f_7/2at85.9eV.Itisimportantto mentionthatwhengoldwasdispersedonTiO₂filmonlyone4f_7/2 emissionappeared[42].

Twodifferentexplanationscanbeofferedfortheappearanceof thisunusuallyhighbindingenergygoldstateaswediscussedprevi- ously[25].Corelevelshiftsduetoparticlesizemustbeconsidered firstintheinterpretationofthespectraofnanoparticles[43–46].

ThesecondpossibleexplanationisthatAumayhaveundergonean ionexchangeprocess.ThisisnotpossibleonTiO₂becauseofthe lackofcationscompensatingtheframeworkcharge,however,itis quitelikelytohappenontitanateswhicharewell-knownfortheir ion-exchangeability[47].

TEMimageonFig.3/Ademonstratesthetubularmorphologyof theas-synthesizedtitanatenanotubeswithadiameterof∼7nm andlengthupto80nm.Theacidicwashingprocessresultedina milddestructionoftheinnerandouterwallsofthenanotubes.The

Fig.2. XP-spectrafromthegold4fregiontakenontitanatenanotubes(A)andona cleanAufilmpreparedonaglassplate(B).

sizeofAunanoparticleswasbetween2.0and8.0nmontheH-form titanate nanotubes (Fig. 3/B). The Au particle sizes was deter- minedbyXRD,too.Theaveragesizewas5.3nmcalculatedfrom theScherrerequation.TheTEMimageofRhdecoratednanotubes and nanowiresshowthepresence ofhomogeneously dispersed nanoparticlesonthesurfaceoftitanatenanostructures(Fig.3/C).

TheparticlesizedistributionofRhonTNTwascalculatedfromTEM andresultedin2.8nmasthemostabundantparticlediameter.As smallas1nmsizedmetalparticlesweredetectedtoointhissam- ple.Particleswithdiametersbiggerthan5nmwerenotobserved.

UnfortunatelywecouldnotobservepeaksforRhcrystalsintheXRD spectraprobablyduetolowconcentrationofparticles.Wemayalso assumethatcertainpartofRhunderwentionexchangeprocess [34].InourpreviousstudiesweobservedthatAuandalsoRhcatal- ysesthetransformationoftubestructuretonanoanataseabove 473Kand573K,respectively[32,34].Inpresentcasesthetemper- atureofpreparationandthephotocatalytictestexperimentsare muchlower.

TheRh3d_5/2peakat309.3eVat1%Rhcontentandat308.3eVat 2%metalcontentclearlysuggesttheexistenceofanoxidationstate ormorphologythatisdifferentfromthebulkbecausethebind- ingenergyoftheRh3d_5/2electronsisabout307.1eVformetallic Rh.Thehigherbindingenergystatesmaycorrespondtoverysmall clustersstabilizedin thestructure ofnanowiresandnanotubes.

ThestabilizationofRhclustersinsmallsizeandtheinfluenceof Rhnanoparticlesonthetransformationoftitanatestructurescan beexplainedalsobytheelectronicinteractionbetweenRh and titanate,whichwasobservedin severalcasesbetweenreduced titaniaandmetals,includingRh[48–51].

3.2. OpticalpropertiesofAuandRhdopedtitanatenanotubes Fig.4shows theabsorptionspectraofsix differentsamples, includingAuandRhloadedtitanatenanotubes.

Thepuretitanatenanotubeshowedstrongabsorptionat3.53eV (351nmwavelength)intheseexperiments.Thecalculatedbandgap energyfromtheTaucplotis3.07eV.Thisvaluewas3.03eVfor pureanatase[Fig.4].ThereductionofAuCl₄withNaBH₄yields1 or2.5wt%Auonthesurface.Thebandgapof1wt%Au/TNTslightly decreased(3.03eV). Nofurthersignificantchangewasobserved at2.5wt%Aucontent.Thischangewaslessthanin thecase of titanatenanowiressupportedAuproducedinasimilarway[32].

ThespectrumoftheAu/TNTshowsa strongabsorptionbandat 2.31eV(534nm).Thisisthecharacteristicabsorptionofthesur- faceplasmonofgoldnanoparticles(d>3nm)andarisesasaresult

Fig.3.TEMimagesofprotonatedTNT(A),Au/TNTpreparedbyNaBH4reduction(B)andRh/TNT(C),andtheparticlesizedistributionofRhonTNTcalculatedfromTEM imageanalysis(D).

ofthecollectivemodesofoscillationofthefreeconductionband electronsinducedbyaninteractingelectromagneticfield[52–54].

Interestingly,thespectrumshowssomeunresolvedpeaksathigher energies.Afterdeconvolutionwecanidentifythreeabsorptionsat 2.68,2.93,and3.19eV.Weshouldemphasizeforthesakeofidenti- ficationthatsmallgoldnanoparticles(d<3nm)losetheirbulk-like electronicproperties;forexample,theynolongershowtheplas- monexcitationcharacteristicsofrelativelylargegoldnanocrystals [53,54].It hasbeendemonstratedbyXPSand HRTEMmeasure- mentsthatourtitanatenanotubesamplescontaingoldinsmall sizes(d<3nm),too[Figs.2and3].Recentlyamultiplemolecular- liketransitionofathiol-protectedAu25clusterwasobserved.At leastthreewell-definedbandsat1.8,2.75,and3.1eVweredetected byUV–visspectroscopy[53].VeryrecentlyasimilarAu₂₅cluster wasidentifiedonCeO2 rodcatalyst.ItwasconsideredthatCeO2

rodshavealargeamountofdefectsites[55–57]andifloadingof Au₂₅(SR)₁₈nanoclustersisverylow,onecanreasonablyexpectthat therodsupportmaybehelpfultoanchorthegoldnanoclusters [58].Asitwaspointedoutinpreviousworks[22,25,27,32,34,40]

thetitanatenanotubesalsocontainahugeamountofdefectsand thesmallclustersandparticlesofgoldcangrowontheoutershell andintheinsideofthetubes.Furthermore,theclustercoalescence couldbepreventedbecausethedefectsintitanatenanotubeswere foundhelpfulforstrongbondingwithmetalnanoparticles[32].For comparisonwepreparedAunanoparticlesonanataseTiO2(Hom- bikatUV-100).Inthiscasetheintensityofplasmoniccharacterwas lessandthemolecular-likefeaturewashardlyseenafterdeconvo-

lutingtheUV–visspectrum[Fig.4].Theratioofthepeakareasof molecular-likebandsandtheplasmonicbandis0.28inthecaseof anatasesupportandis0.36fornanotubesupport.Fromthiscom- parisonwemayconcludethattitanatenanotubeshavetheability tostabilizethesmallparticleseveninclustersizeformation,similar toCeO₂rodcatalyst.

After impregnating the titanate nanotubes with rhodium- chloridesolutiontwonewveryweakbandsappearedat2.52and 3.07eV(492and404nm,respectively).TheRh(III)salthasaneffect ofslightlydecreasingthebandgappossiblyduetotheinfiltration ofrhodiumionsintothetitanatestructure.Thebandgapisslightly decreasedinthecaseof1%Rh/TNTto3.04eVcomparedtopure nanotube(3.07eV)butdoesnotchangedat2.5%rhodiumcontent (notshown).Astrongabsorptioninthevisibleregionemergeddue toreduction[Fig.4].Onthereducedsamplesthebandgapdoes notdecreasedatall(3.08eVfor1%Rhand3.16eVof2.5%Rh).The deconvolutionhasbiggeruncertaintyinthiscasecomparedtothe clean,protonatedtitanatenanotubesandtheRh(III)saltcontaining materialsduetothehighoverlapping,hencethebandgapenergy hasbiggererror,too.Thenewbroadbandhasamaximumat3.06eV (405nm)[Fig.4].

Of thenoble metals(Pt, Pd, Ru and Rh),a theoretical study foundthatonlyRhhasastrongUVplasmonicresponse[59],itwas supportedexperimentally,too[60].Consideringonlytherepresen- tativenanoparticlesizeandshapemodelwithrandomorientations, thetheoreticallypredictedpeakforthedipolarmodeinthetripod planenear3.3eV(375nm)isingoodagreementwiththeexper-

Fig.4.TheDR-UV–visspectrawiththecalculatedbandgapenergiesofsixdifferentsamples.Theoriginalspectraareshownbythethick,greycurves.Bandgapenergieswere calculatedfromfittedgaussfunctionswithTauc’smethodinallcases.

imentaldataobtainedonsiliconsubstrate[61].Weassumethat thelocalsurfaceplasmonresonance(LSPR)stronglydependson thenatureofsubstrateandtheresonantenergyincreasinglyred- shiftedwithincreasingsize.Theobservedbroadbandcenteredat 3.06eVcontainstheplasmoniccharacterofRhontitanatenano- tubes.

3.3. Photocatalytictests

Photocatalyticmeasurementsrevealedthatmethaneisactive towardsphoto-oxidationinallcasesevenifnootherreactantsare presentinthefeedmixture.Table1showstheaverageandmaximal formationratesoftheidentifiedproductsandthemethaneconver- sionvalues.Bothtitanatesupportedcatalystsexhibitedoneorder ofmagnitudehigheractivityinmethaneconversionthanpristine nanotubes.TheAu/TiO₂showedsmalleractivityinmethanetrans- formationthanthenanotubesupportedvariant.

Generallythemethanetransformstohydrogen,ethane,andoxi- dizestocarbon dioxide,carbon monoxideandmethanol. Fig.5 showstheconversionofmethaneandtheformationofproducts asafunctionofirradiationtime.

Ascanbeseenthemolarfractionoftheproductsareincreasing whilethequantityofmethanedecreasesduringtheUV-activesec- tions.Decreasinginthemolarfractionswithtimecanbeobserved inthecaseofRh/TNT.ThisdropisrestrictedtotheUVactiveperi- odsandistheconsequenceofactivityloss.InthecaseofAu/TNTno activitylosswasobservedintheexperiment.Wecanstatethatthe supportedmetalcatalystsareextremelyactiveinhydrogengener- ation.Thecontributionofethanetothehydrogenratesissmallin thesecasessoourmainprocessshouldbesomekindofmethane decompositionwherethecarbonhighlyoxidizesoritremainson thesurface.Itisimportanttoemphasizethattheformationofwater wasnotdetected,ormoreprecisely,therateofwaterformationwas

Fig.5.Conversionofmethaneandformationofproductsasafunctionofirradiation timeonRh/TNTandAu/TNTcatalystsinthemethanetransformationreaction.

underthedetectionlimit.Thecarbonbalance(_C)wascalculated fromtheaverageformationratesbyEq.(1).

C=

−rCH₄−2rC₂H₆−rCO₂−rCO−rCH₃OH

_CispositiveandlargeinthecasesofTNTsupportedmetals whichmeansthatsomecarbonismissingfromtheproductstream.

Itsreasoncanbeanundetectedproductorsomekindofsurface deposit.Itisimportanttonotethatthenanosizedgoldcatalyst (Au/TNT)hashigheractivityinethanegenerationbecausetherates areincreasedbyoneorderofmagnitudecomparedtotheothercat- alysts.Theintroductionofcarbondioxideintothereactantstream didnot resultsignificanteffects ontherates.Theconversionof

Table1

Averageandmaximalformationratesoftheidentifiedproductsandthecalculatedmethaneconversionsinthedifferentexperimentsetups.Thecarbondeficitandtheoverall photoconversionefficiency(␩)regardingtohydrogenformationarealsoshown.

reactants catalyst rateofformation(␮molh⁻¹g⁻¹) KCH4(%) carbondeficit(␮molg⁻¹) ␩H2

CH4 C2H6 H2 CO2 CO CH3OH

CH4 TNT −8.75 1.17 1.40,3.44 3.04^a 0.23^b 0.0837^a 0.23 28^b 1.4×10⁻⁶

Rh/TNT −49.6 1.95 115,235 11.2^a 5.44^a 0.138^a 1.41 260^b 1.2×10⁻⁴

Au/TNT −70.2 12.0 116,127 18.1^a 11.3^a 1.01^a 1.64 140^c 9.7×10⁻⁵

Au/TiO2 b.d.l. 1.50 48.0,57.3 9.28^a 1.70^a 0.108^a b.d.l. b.d.l. 6.4×10⁻⁵

CH4+CO2 TNT −6.05 0.721 0.746^a,1.61 b.d.l. 0.427^a b.d.l. 0.16 37.6^a 8.2×10⁻⁷

Rh/TNT −68.0 1.72 107,246 53.6 11.4^a 0.142^a 2.03 b.d.l. 1.2×10⁻⁴

Au/TNT −70.5 11.4 104,117 b.d.l. 11.9^a 0.954^a 1.66 310^a 8.8×10⁻⁵

Au/TiO2 −21.8 1.86 50.7,63.2 b.d.l. 5.76^a 0.144^a 0.73 110^b 6.2×10⁻⁵

H2O Rh/TNT b.d.l. b.d.l. b.d.l. 0.34^a b.d.l. b.d.l. – – –

Italicnumbersmeansthemaximalformationrates.

b.d.l.:belowdetectionlimit.

aEstimateddeviationis>10%but≤25%.

bEstimateddeviationis>25%but≤50%.

c Estimateddeviationisbiggerthan50%.

CO₂wasunderourdetectionlimitexceptonecasewhereCO₂was formeddespiteitshighbasicconcentration.Wecoulddeducethat CO₂isratherformsthandiminishesinthesecases.

Inthephoto-inducedCH₄decompositionprocessCO₂andCO (alsoC)appearedasproductssothereshouldbeanoxygensource inthesystem.Inthecaseoftitanatenanotubeswaterisaplausi- blereactionpartnerwhichservestheoxygenreactingwithmethyl radical.Itwasalreadyestablishedthattitanatenanotubescontain alargeamountofH₂O[28,34,62].Wehaveidentifiedadsorbedand latticewaterinourXPS,DRIFTSandDTG-MSexperiments.TheOH andH₂Ostretchingvibrationsbetween3000and3750cm⁻¹could bedetectedupto673Kontitanatenanotubes.TheOHandH2O deformationsignalat1618–1648cm⁻¹waspresentupto600–700 K.Interestingly,averyweekasymmetricinfraredsignalattributed toH2Oaround3730cm⁻¹ wasdetectedevenat773Konnano- tubes.An“OH”likephotoemissionemergedat532.6–532.8eVin theO1sXPspectrum.Thispeakdisappearedat573Konnanowires whileonnanotubesthisemissiondiminishedonlyabove673K.In agreementwiththeIRandDTG-MSresultsthepeakcorresponding towaterdecreasedsharplybetween293and573K.Ourhypoth- esisthatwater actsasanoxygensourceformethaneoxidation gainedstrengthwhenweintroducedwaterinsteadofcarbondiox- ideintothereactantstream:Notonlythemethaneconsumption butalsotheformationratesofH2,CO2andCOandthequantityof themissingcarbonwereincreased.

Oneblankexperimentwasconductedinordertomakesurethat theproductsarenotoriginatesfromsurfacecontaminations.When onlywaterispresentonargon,nomethane,ethaneorhydrogen formationwereobserved.OnlytracesamountofCO₂wasevolved.

Thismeansthattheproductsdetectedontheotherexperiments originatesfromthereactants.

3.4. In-situinfraredspectroscopymeasurements

Fig. 6 shows the infrared spectra registered after one hour irradiationinthemethaneconversionandCH₄+CO₂reactionsper- formedoverTNTandRh/TNTcatalysts.Peaksevolvedpartlydue totheadsorptionofreactantsorproductsandpartlyduetothe irradiationofthesample.Theadsorptionofwaterasreactantor contaminationresultsintheappearanceofapeakat1638cm⁻¹ andabroadbandbetween2700and3700cm⁻¹.Thecarbondiox- ideadsorptionresultedin strongpeaksat1558and 1375cm⁻¹ whichcanbeattributedtobidentatecarbonateswhichbindtothe surfaceoftitanatenanotubes[63,64].Thecarbonatepeakshave higher intensitywhen CO₂ is present in thefeed.The remain- ingsmall peaks arethe resultsof UVirradiation. Thebands at

2968and2885cm⁻¹areattributedtothesymmetricandasym- metricstretchingvibrationsofmethylgroupsonRhsurface[65].

ThemethylgroupmaybondtothetitanateviaanO-bondform- ingmethoxybutthes(C O)vibrationmodewhichshouldappear ataround 1050cm⁻¹ onmetal-oxides [64] wasnotdetectedin ourcase. The deformationmodeof methylvibration ataround 1350cm⁻¹ ispossiblyhiddenbyoverlapping signalsinourcase [65].Physisorbedcarbondioxidecanbeidentifiedat2337cm⁻¹. The Rh-bonded linear carbon-monoxide resulted in a peak at 2100cm⁻¹onpartiallyoxidizedRh,whereasbridgedCOappeared at1924cm⁻¹[66–68].Thepeakat2141cm⁻¹representsthethe- oreticalvibrationenergyofgasphaseCOwhichisphysisorbedon thesurfacelikethecarbondioxideinourcase.Thephysisorption ofCOandCO2wasnotobservedpreviouslyonthefreshcatalysts sowecanassumethatsomechangesareoccurredonthesurface duringthephoto-inducedreaction.

Duringphotoilluminationashoulderappearedat1664cm⁻¹ intheCH₄ decompositionandintheCH₄+CO₂reactiononpris- tineTNT.Thisbandcanbeattributedtoadsorbedformyl group [69] which forms inthe reaction.Monodentateformate can be alsoidentifiedfromthebandsat1585and1384cm⁻¹[66,70].The absorptionbandofformylismissingwhenmetalispresent.Very probablythisintermediateishighlyinstableandthemetalcataly- sesitsfurtherreactiontoformCO.Monodentateformateispresent inallcaseswhichmeanthatitsfurtherreactionisslow.

3.5. Ramanspectroscopyresults

WeperformedRamanspectroscopymeasurementsinorderto getinformation aboutthestructureofthesurface depositsdis- cussedintheprevioussection.TheRamanspectraareplottedin Fig.7.

Absorptionbandsappearedonlyinthe1800–100cm⁻¹region.

Theabsorptionbandswerelocatedatthesamewavenumbersin bothcases.Onlytheareaofthebandsdiffers.Thebandsat1598 and1335cm⁻¹correspondtotheGandDbandsofstructuredcar- bonlayerssuchasgraphene[71].TheDandDbandsataround 1620and1100cm⁻¹werenotobserved.TheintensityoftheDband ishighrelativetotheGbandwhichmeanshighdefectdensityin thegrapheneplane.Thispeakratioiscommoninthecaseofmulti- walledcarbonnanotubes.Wecanconcludethatstructuredcarbon withhighdefectdensityformedonthesurfaceduringmethane conversion.TheremainingbandsintheRamanspectrabelongto thetitanatenanotube[34,72].

Fig.6.DRIFTspectraofthecatalystscollectedafter1hUVirradiationintwodifferentreactionsperformedontheTNTbasedcatalysts.Thegasphasereactantswereflushed outwithheliumbeforecollectingspectra.

Fig.7.RamanspectraoftheRh/TNTcatalystsusedintheCH4+H2Oandinthe CH4+CO2reactions.

3.6. XPSmeasurements

HighresolutionXPspectrawerecollectedinthebindingenergy rangeofcarbon1sandrhodium3dorbitalsinordertoinvestigate theoxidationstatesoftheseelements(Fig.8).TheRh3d_5/2-3d_3/2 doublethasashoulder athigherbindingenergies. The307.3eV peakfor3d_5/2canbeidentifiedasRh⁰whereasthehighenergy shoulderat309.4eVischaracteristicforRh³⁺.Thisassignationis inaccordancewiththeinfraredspectroscopyresultsmentioned previouslybecausetheinfrared absorptionat2100cm⁻¹ corre- spondsto carbon monoxide bonded to oxidized rhodium. The colourchangeofthecatalystfromdarkgreytobrownishgreywas experiencedunderirradiationwhichisthesignofre-oxidationtoo.

ThehigherbindingenergystateofRh3delectronsmaycorrespond tosmallerRh-clustersizesontheotherhandandisaconsequence ofthefinalstateeffectwhichismoredominantinthecaseofcat- alystswithlessthan2%metalcontent[28].

Fig.8. TheXPspectraoftheRh/TNTcatalystusedintheCH4decompositionreaction.

AdditionalC1sspectrum(lower)showsthecarbonregionbeforeuse,justafter reduction.

Additional C 1s spectrum (lower) shows the carbon region beforeuse,justafterreduction.

O1sat530.4eVrepresentsthelatticeoxygenofTNT.Thepho- toemissionpeakat532.7eVinvolvesCO,methoxyandcarbonate likespeciesformedduringphotoreaction.AdsorbedH₂Oappears at534.9eV[34].

Wecouldidentify3peaksinthecarbon1sregionwhichcor- respondtodifferentoxidationstates:Thepeakat286.7eVisthe characteristicbindingenergyfortheC OandC Ocarbons.Itcan originatefromthecarbon-monoxidechemisorbedbyRhparticles whichisalready revealedbyinfraredspectroscopy.Thepeakat 284.4eVcorresponds to sp² hybrid state carbon (C, CH and CH₂)whichconfirmsthepresenceofstructuralcarbonconcluded fromRamanresults.Thepeakat282.2eVbelongstoreducedcar- bonwhichmeansmetal-carbonbonds.Itmeansthatthecarbon depositsaresittingonthesurfaceofRhparticles[73].Anadditional C1sspectrumisplottedinFig.8fromafreshlyreducedRh/TNTcat- alysthadnotbeenusedinphotocatalyticreactionyet.Itcanbeseen

Fig.9. Formationrateofmethaneduringthetemperatureprogrammedreduction experiments.

thatthecarboncontentismuchlowerinthiscasethanafterreac- tion.sp² carbonat284.5eVandoxidizedcarbonoriginatesfrom carbonatesat289.7eVcanbeidentified.

3.7. Temperatureprogrammedreductionmeasurements

Temperatureprogrammedreductionexperimentswerecarried outtoinvestigatethequantityandthereactivityofthesurfacecar- bonassumedtobeformedduringthephotocatalyticreactions.Two experimentswerecarriedoutontheRh/TNTusedintheCH4+H2O andintheCH4+CO2reactions.Twoblankexperimentswerecar- riedoutadditionallytomakesurefromthesourceofthesurface carbon.

Nomethaneformationwasobservedatallinthefirstblank experimentwherefreshRh/TNTwasheatedupinhydrogenflow.

Onlytraceamountofcarbondioxidewasdetected.Thismeansthat thefreshlypretreatedsampleisfreeofreduciblesurfacecarbon.

Ontheotherhandsmallamountofmethaneformedinthesec- ondblankexperimentwherethecatalystwastreatedinmethane flowfor1hat403KbeforetheTPRexperiment.Theformationofa largeamountofmethanewasdetectedduringthemeasurements madeontheusedcatalysts.Casually,theformationofsomeethane andthedesorptionofcarbon-dioxidewasobservedinthesecases.

Theappearanceofmethaneandethaneistheresultofthein-situ reductionofcarboncontainingsurfacedepositsformedduringthe photocatalyticreactions. Thesourceofthecarbondioxideisthe surfacecarbonateorhydrogencarbonatespecieswhich decom- posewithoutreductionasthetemperaturerises.Theformation rateofmethanewascalculatedinallcasesandwasplottedagainst temperatureinFig.9.

Themaximumofthemethaneformationrateoccurredatthe sametemperaturemeaningthatthereactivityofthesurfacecar- bonequalsinallcases.Themethane-waterreactionresultedin moresurfacecarbonwhichcorrespondswithlargerphotocatalytic methaneconversionsobservedinthiscase(notshown).Thetime integralof theTPRcurves givesustheoverall formedmethane relatedtocatalystquantitywhichis354␮mol/gand528␮mol/g for the used catalysts respectively. The 354␮mol/g value is in goodagreementwiththecalculatedcarbondeficitalreadyshowed inTable1. becausethematching iswithinthemarginoferror.

38␮mol/gmethaneformedintheblankexperimentwhichisone orderofmagnitudesmallerthanintheothercases.Wecancon- cludethatthemissingcarbonisonthesurfaceintheformofsome kindofeasilyreduciblecarboncontainingdeposit.

3.8. Presumedreactionmechanism

Wecanestablishthemechanismofmethanetransformationon thebasisofthepreviousconclusions.Electron-holepairsweregen- eratedonmetal(goldorrhodium)promotedtitanatenanotubes uponabsorptionofUV-lightirradiation[Eq.(2)].

TNT→^hTNT

e⁻,h⁺

(2) TNT

e⁻,h⁺

TNT

e⁻

+TNT

h⁺

(3) Afterthedissociation oftheexciton describedbyEq.(3) the electronandtheholestartstomigratetotheenergeticallyfavorable positions.Electronshavehigherpossibilitytobefoundonthemetal particles[Eq.(4).]duetoFermi-levelequilibrationwhichappears betweenthemetalandtheoxide[74].

TNT

e⁻

M

e⁻

(4) Inthecaseofcontinuousirradiationthis processresultsina potentialdifferencethatbuildsupbetweenthemetalandtheoxide henceanadditionaldriftcurrentstartsinthereversedirection.The chargeseparationreachesanequilibriumcontrolledbythediffu- sionanddriftcurrentsandstronglydependsontheratesofcharge carriergenerationandrecombination.

Surfacewatermoleculescancatchtheholeandproducereactive OH-radicalandH⁺whichdelocalisesonthenearbywatermolecules [Eq.(5)].Hydroxideradicalcanformfromhydroxideion,too,bythe sameprocess[Eq.(6)].

TNT

h⁺

+H2O_(TNT)→TNT+OH^•_(TNT)+H⁺ (5) TNT

h⁺

+OH⁻_(TNT)→TNT+OH^•_(TNT) (6)

Theasgenerated hydroxideradicals areveryaggressiveoxi- dantsandstarttooxidizemethaneinaradical-typereaction[Eq.

(7)].Theformedmethylradicaladsorbsonthemetalsurface.

OH^•_(TNT)+CH4(g)

→MH2O_(TNT)+CH^•_3(M) (7)

Wecannotexcludethatmethanedirectlyreactingwithholes resultsinmethylradicalsonthetitanatesurface[Eq.(8)].

TNT

h⁺

+CH_4(g)→TNT+CH^•_3(TNT)+H⁺ (8)

Thefurtherreactionrouteisdeterminedbythenatureofthe metal.Itisgenerallyacceptedthatthecouplingofmethylradicalsis favouredongold[75][Eq.(9)],whilethedehydrogenationprocess ratheroccursonrhodium[63][Eq.(10)].

2CH^•_3(M)→^MC2H6(M)→C2H6(g) (9) CH^•_3(M)→^MCH^•_2(M)+H^•_(M)→→C_(M)+3

2H_2(g) (10)

Gasphasehydrogenformsinthefollowingcouplingreaction [eq.(11)]:

2H^•_(M)→^MH_2(M)→H_2(g) (11)

Therecombinationofamethylandhydrogenradicalalsohasto beconsidered[eq.(12)]

CH^•_3(M)+H^•_(M)→^MCH4(M)→CH4(g) (12) HydrogenradicalcanformfromH⁺,too,bythecaptureofa photoelectronatthemetal-supportinterface:

M

e⁻

+H⁺→^MM+H^•_(M) (13) Thesourceof thesurplusCO canbemostlythereductionof carbondioxide.Duringthereductionprocessphotoelectronsand

hydrogenionsbringsabouttheCO₂ reductionviaCO₂^•−radical anion:

M

e⁻

+CO_2(g)→M+CO^•_2(M)⁻ (14) TNT

e⁻

+CO_2(TNT)→TNT+CO^•_2(TNT)⁻ (15)

TNT

e⁻

+CO^•_2(TNT)⁻ +H⁺→TNT+CO_(TNT)+OH⁻_(TNT) (16) ConsequentlybothCH₄andCO₂arefirstadsorbedoverthecat- alystsurfaceandthenconvertedtoCH3•andCO2^•−species.TiO2

[19]andmodifiedTiO₂nanocomposites[12]arealsoactiveinCO₂ photoinducedactivationbutthepresenceofmetaladatoms(Auor Rh)significantlyincreasesthespeedoftheactivationprocesses.

Monodentateformatecanformfromcarboxylateradicalanion andhydrogenion[Eq.(17)]:

CO^•_2(TNT)⁻ +H⁺→HCOO^•_(TNT) (17)

The formation of oxygen containing compounds can be explainedbythefollowing equations:Methanolisformedby a couplingreaction[Eq.(18)]ontheperimeterofthemetal-support interfacethenitrecombineswithaholetooxidizefurther[Eq.(19)].

CH^•_3(M)+OH^•_(TNT)→^MCH3OH_(TNT)→

slowCH3OH_(g) (18)

TNT

h⁺

+CH3OH_(TNT)→TNT+CH3O^•_(TNT)+H⁺ (19) Formaldehydemostprobablyformonthesupportasconcluded frominfraredresults[Eq.(20)]butmetalisneededforitsfurther reactions[Eqs.(21)–(22).]

CH₃O^•_(TNT)+OH^•_(TNT)→CH₂O_(TNT)+H₂O_(TNT) (20) CH2O_(TNT)+OH^•_(TNT)→CHO^•_(M)+H2O_(TNT) (21)

CHO^•_(M)→^MCO_(M)+H^•_(M) (22)

Itisremarkablethattherearemanyroutesinwhichhydrogenis formed.Asaconsequencethemainproductofthephoto-induced methanetransformationisthehydrogen.

3.9. Surface-modifiedtitanatenanotubesphotocatalysis

Thepristinetitanatenanotubesshowedmeasurablephotocat- alyticactivitysincethetimescaleofelectron-holerecombinationis commensuratewiththeredoxreaction.Inthisprocesstheincrease amountofdefects(Ti³⁺andoxygendeficient)intitanatesplayalso asignificantrole[34,76,77].Modificationoftheopticalandelec- tronicpropertiesofTiO2 resultsinnotonlythereductionofthe bandwidthviatheincorporationofadditionalenergylevelsbut increasedlifetimeofthephotogeneratedelectronsandholesvia effectivechargecarrierseparationandsupressonofelectron-hole recombination[20].Thisisvalidfortitanatenanotubes,too.UV irradiationinducesFermilevelequilibrationbetweenTiO₂andAu viachargedistributionand thereby Fermilevel shiftbyaround

−22mV[74,78].Such Fermilevel shiftincreasesthenumber of morereductiveelectronsonthemetalandpromotesefficientpho- tocatalyticreaction.Aswediscussedabovethismechanismcould playasignificantroleinourphoto-inducedreactionsintheCO2- methanesystemontitanatenanotubesmodifiedwithAuandRh nanoparticles.

ThenanoparticlesofAu(andsomeothermetals)arecoupled toTiO₂(includingtitanatenanorods)toutilizetheirpropertyof localized surface plasmonic resonance (LSPR) in photocatalysis [42,79–82].LSPRisthecollectivefreeelectronchargeoscillation inthemetallicnanoparticlesthatareexcitedbylight[83].This phenomenonusuallyoccursinnanoparticles(>3nm),andstrongly dependsontheparticlesize,shapeandlocaldielectricenvironment

[80].Duringlightirradiation,theelectrontransferfromthephoto- excitedAunanoparticlestotheTiO2conductionbandmayoccur.

Theotherscenarioisalsoplausible,namely;electronsexcitedfrom metaltransfertoreactantmolecules.SuchkindofAumediated reductionofC60wasdemonstrated[74].

We believe thatsimilar Auand Rh mediated photo-assisted reactionoccursinCO₂ activation[Eq.(14)]intheCH₄+H₂Oand CH4+CO2reactionsontitanatenanotubes.Asitisdemonstratedin Table1thegoldcontainingtitanatenanotubesexhibitsignificantly higherphotocatalyticactivitythanAu/TiO₂(anatase),thoughthe puretitanatenanotubesalonedonotshowhighactivity.Theinten- sityofplasmonabsorption(at2.31eV)washigheronAu/TNTthan onAu/TiO₂ (Fig.4).Consequently,theelectrontransferfromthe metaltothereactantsismorefavorable.Itshouldbenotedthat theLSPR-inducedphotoeffectsaresignificantlyinfluencedbythe propertiesofTiO₂(size,shape,surfacearea,crystallinity)[80].We believethatthetitanatenanostructureshaveapositivefeatureto localizethemetalnanoparticlesinthispointofview.

Theotherimportantobservationisthattitanatenanotubescan stabilizegold(andalsoRh)insmallsizes,below3nm(Figs.2and3).

Insuchdimensiontheplasmonicfeature(LSPR)doesnotoperate.

Atthesametimemultiplemolecular-liketransitionsofthegold clusterwasobservedbyUV–visspectroscopy(Fig.4).Theintensi- tiesofthesetypesoftransitionswerealsosignificantlyhigheron titanatenanotubes.Aswehavealreadydiscussedabove,thesmall metalclusterscanstronglybondtothedefectsitesintitanatenano- tubes.Theseclustersmaybedirectlyinvolvedinthephoto-induced reactions.Themolecular-likeclustersmayformcomplexeswith thereactantswheretheelectrontransferdirectlyoccursduringUV irradiation.

In the light of the possible photocatalytic mechanisms it seemsthatthetypesofinteractionbetweenmetalandsubstrate (titanates)playimportantrole.Fromthisrespectthelong-range andshortrange interactions shouldbetaken intoaccount[84].

Whilethereisnosignificantbandgapdecreaseduetothemetal adatoms,theAuandRhchangesthebandgappopulationandshift- ingoftheFermilevel.SuchFermilevelshiftincreasesthenumberof electronsonthemetalandpromotesefficientphotocatalyticreac- tion.Thisshiftisa consequenceof chargetransferbetweenthe metalandsupport[48,84].Besidesofthislong-rangeinteraction, short-rangeinteraction,affectingtheatomsoratomclusters(for exampleAu₂₅inpresentcase)atthegas-metal-supportinterface, couldbemoreimportant.Theshort-rangeinteractioncanbecon- sideredasa consequenceofthestrongelectricfields,whichare presentattheinterface.

Finally wecalculated photo-conversion efficienciesfromthe amount of hydrogen formed in each photo-induced reactions (Table.1).Thecalculatedvaluesareratherlow.Itiswell-known thatthecomplicated charge-carrier dynamicsand surfacereac- tionkinetics mainlylead tothe low quantum efficiency inthe multi-step processes of heterogeneousphotocatalysis [85]. The suitablethermodynamicproperties(includingbandgapsandCB/CV levels)donotguaranteegoodphotocatalyticefficiency.Itiscom- monly accepted that the mechanism governing heterogeneous photocatalysisconsistsoffourconsecutivetandemsteps;(1)light harvesting,(2)chargeexcitation/separation,(3)chargemigration, transportandrecombination,and(4)chargeutilization[86].There- fore,theoverallphotocatalysisefficiencyisstronglydependenton thecumulativeeffectsofthesefourconsecutivesteps.

4. Conclusions

Itwasdemonstratedinthepresentstudythattitanatenano- tubeshavenumerousadvantageouspropertiesthatplayimportant role in theinvestigated reactions. It is well knownthat larger