SOx uptake and release properties of TiO2/Al2O3 and BaO/TiO2/Al2O3 mixed oxide systems as NOx storage materials

(1)

ContentslistsavailableatSciVerseScienceDirect

Catalysis

Today

jo u r n al h om epa ge :w w w . e l s e v i e r . c o m / l oc a t e / c a t t o d

SO

x

uptake

and

release

properties

of

TiO

2

/Al

2

O

3

and

BaO/TiO

2

/Al

2

O

3

mixed

oxide

systems

as

NO

x

storage

materials

Göksu

S. S¸

entürk

a

_,

_Evgeny

_I.

_Vovk

a,b

_,

_Vladimir

_I.

_Zaikovskii

b

_,

_Zafer

_Say

a

_,

_Aslı

_M.

_Soylu

a

_,

Valerii

I. Bukhtiyarov

b

_,

_Emrah

_Ozensoy

a,∗

a_Chemistry_Department,_Bilkent_University,₀₆₈₀₀_Bilkent,_Ankara,_Turkey b_Boreskov_Institute_of_Catalysis,₆₃₀₀₉₀_Novosibirsk,_Russian_Federation

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received15September2011 Receivedinrevisedform 23November2011 Accepted1December2011 Available online 30 December 2011 Keywords: Al2O3 BaO TiO2 Anatase Sulfurpoisoning DeNOx NOx SOx Sulfation NSR LNT HDS ClausProcess

a

b

s

t

r

a

c

t

Titaniawas used as a promoter toobtain novel materials in the form of TiO2/Al2O3 (Ti/Al)and

BaO/TiO2/Al2O3(Ba/Ti/Al,containing8wt%or20wt%BaO)thatarerelevanttoNOxstoragereduction

(NSR)catalysis.Twodifferentprotocols(P1,P2)wereutilizedinthesynthesis.Ti/Al(P1)manifestsitselfas crystallitesofTiO2on␥-Al2O3,whileTi/Al(P2)revealsanamorphousAlxTiyOzmixedoxide.Thestructures

ofthesynthesizedmaterialswereinvestigatedviaTEM,EDX,BETanalysisandXPSwhilethecatalytic functionality/performanceofthesesupportmaterialsuponSOxandsubsequentNOxadsorptionwere

investigatedwithTPDandinsituFTIRspectroscopy.Ti/Al(P1,P2)revealedahighafﬁnitytowardsSOx.

OverallthermalstabilitiesoftheadsorbedSOxspeciesandthetotalSOxuptakeoftheBa-freesamples

increaseinthefollowingorder:TiO2(anatase)␥-Al2O3<Ti/Al(P1)<Ti/Al(P2).ThesuperiorSOxuptake

ofTi/Al(P1,P2)supportmaterialscanbetentativelyattributedtotheincreasingspeciﬁcsurfacearea uponTiO2promotionand/orthechangesinthesurfaceacidity.PromotionofBaO/Al2O3withTiO2leads

totheattenuationoftheSOxuptakeandasigniﬁcantdecreaseinthethermalstabilityoftheadsorbed

SOxspecies.TherelativeSOxadsorptioncapacitiesoftheinvestigatedmaterialscanberankedasfollows:

8Ba/Ti/Al(P1)<8Ba/Ti/Al(P2)<8Ba/Al∼20Ba/Ti/Al(P1)<20Ba/Al<20Ba/Ti/Al(P2).

1. Introduction

Controllingthesurfacechemistryofmixed-oxidesurfacesisof vitalimportancetodesignnovelcatalyticmaterialswith unprece-dentedfunctionalities.Sulfur poisoningonmetaloxidesurfaces isoneof thefrequentlyobservedsurface deactivation phenom-enainheterogeneouscatalysis.AccumulationofSOxspecies on

metaloxidesurfaceshavebeenverycommonlystudiedinthe liter-atureinrelevancetothree-waycatalysis(TWC),selectivecatalytic reduction(SCR),NOxstoragereduction(NSR),

hydrodesulfuriza-tion(HDS)andothercatalyticprocesses,whereAl2O3isutilizedas

themainsupportmaterial[1–7].Twomajordeactivation phenom-enaareoftenreportedfortheNSRcatalysts.Theﬁrstrouteinvolves thermaldegradationofthestructuralintegrityofthecatalyst mate-rialduetosolidstatereactionsbetweenthecatalystcomponents

∗ Correspondingauthor.

E-mailaddress:ozensoy@fen.bilkent.edu.tr(E.Ozensoy).

andsinteringwhilethesecondrouteisassociatedwiththesulfur poisoning [8,9].The latterdeactivationphenomenon is particu-larlyachallengingproblem.InNSRapplications,NOx(g)andSOx(g)

adsorbates,competeforthesamebasicadsorptionsitesonoxide surfaces.Consequently,maximizationoftheNOxstorage

capac-ity(NSC)whileminimizingtheirreversibleSOxuptakeoftheNOx

storagesitesimpliescarefuloptimizationofthesurfaceproperties ofthecatalyticsupportmaterialssuchasthesurfacecomposition, acidity,morphologyandspeciﬁcsurfacearea(SSA).

SulfurpoisoningofNSRcatalyststypicallyleadstothe forma-tionofalkalineearth/preciousmetalsulfates,sulﬁtesorsulﬁdes [8]. For a large number of oxide substrates, the stability of some of the common adsorbates increases in the following order:NO2−∼CO32−<NO3−<SO42− [10–15]. Thus, sulfur

effec-tively blocks the catalytic sites for NOx storage and gradually

reducestheoverallNOxstoragecapacityoftheNSRcatalysts[16].

Effortstowardsimprovingthecatalytictoleranceagainstsulfur poisoninganddesigninghighlyactiveandstablenovelcatalysts arevitalfortheglobalizationoftheNSRtechnology[14,15,17–20]. 0920-5861/$–seefrontmatter © 2011 Elsevier B.V. All rights reserved.

(2)

Scheme1.SynthesisofTiO2-promotedNOxstoragematerials:P1andP2.

Misonoand Inui[21],Fritzand Pitchon [22]and severalothers [12,13,23–27]reporteddetailedstudiesontheimprovementand thedurabilityenhancementofNSRcatalysts.Thecommercial cat-alystsarealsosuccessfullyusedinlimitedmarketssuchasJapan wherethesulfurcontentofthedieselfuelisrelativelylow(below 10ppm)[28].AccordingtoYamazakietal.,Fewasfoundtodecrease thesulfuruptakeandpromotethedecompositionofBaSO4 [29].

Fansonetal.observedthatFefacilitatedtheformationofabulk nitratespecieswhichweresuggestedtoberesilientagainstsulfur poisoning[30].Kayhanetal.reportedthattheinitialNOxuptake

mechanismandtheratiobetweenthesurfaceandbulkBaOsites weresigniﬁcantlyinﬂuencedbyFeloading[20].

TiO2 containingsupportmaterialswerealsoreportedto

sup-pressthesulfurpoisoning[31,32].Thiseffectwasassociatedwith thesurfaceacidityofTiO2,inhibitingtheadsorptionofacidicSOx

species[33].Decompositiontemperatureofthesulfatesonpure TiO2 isknowntobelowerthanthatof␥-Al2O3[34,35].

Further-more,activesitesonthesulfatedTiO2surfacecanalsobereadily

regeneratedunderreactionconditions[36].Recently,itwasalso reportedthatTiO2sitesonthesupportsurfacecanfunctionas

efﬁ-cientanchoringsitesforBaO,whichmaybeexploitedtoenhance thesurfacedispersionofBaOdomainsandﬁne-tunethesurface morphology[37].

Inordertocircumventsomeoftheunfavorablepropertiesof pureTiO2suchasthelimitedthermalstability,lowsurfacearea,

andpoormechanicalproperties,itscombinationwithsecondary oxidesareutilizedtodesignnovelsupportmaterialswithenhanced properties.Alongtheselines,␥-Al2O3[14,19,32,36,38–42]and/or

ZrO2[43–45]arepromisingchoicesforthesecondaryoxidesthat

can beused in combination withTiO2. It was reported that if

sulfur-poisonedBa/Pt/Al2O3catalystwasblendedwithnon

sulfur-poisonedPt/TiO2catalyst,sulfurdesorptionfromtheBa/Pt/Al2O3

catalystunderrich conditionsisimproved[14].Furthermore,it wassuggestedthat theinterfacebetweenAl2O3 andTiO2 plays

animportantroleinthesulfate decompositionandthe desorp-tionprocesses[32].Matsumotoandco-workers[14,46]reported thatacombinationofTiO2 andlithium-doped␥-Al2O3 presents

anoptimumsurfaceaciditytowardssulfurpoisoning.The macro-scopicgeometricalstructure ofthe catalytic monolithwasalso foundtobeeffectivein limitingthesizeofthesulfate particles andcontrollingthesulfatedecomposition/desorptiontemperature [47,48].Itwasshownthatsulfatedecompositionisfacilitatedon

thesmallersulfatedomains.Therealsoexistanumberofsurface sciencestudiesonplanarmodelNSRcatalysts[49–53]focusingon thestructureandtheoperationalprinciplesofNSRsystemsatthe molecularlevel.Despitetheestablishedsulfurresistingeffectof TiO2asanadditiveinthecompositionofthePt/Ba/␥-Al2O3

cata-lysts,anumberofcrucialaspectsregardingtheinﬂuenceofTiO2

ontheinteractionbetweentheNOxstoragecomponentandthe

supportmaterialhavenotyetbeenelucidated.

Thus,inthecurrentwork,wefocusourattentiononthe inter-action ofSOxwiththeTiO2/Al2O3 and BaO/TiO2/Al2O3 surfaces

aswellastheinﬂuenceofTiO2domainsontheNOxuptakeafter

deactivationbySO2(g)+O2(g).

2. Experimental

TiO2/Al2O3binaryoxidesupportmaterials(whichwillbe

here-afterreferredasTi/Alinthetext)werepreparedviatwodifferent syntheticprotocolsP1[23,37]andP2[37]whichweredescribed in detail in ourformer reports. These syntheticprotocolswere schematically described in Scheme 1. Brieﬂy, in the ﬁrst syn-theticprotocol(P1),␥-Al2O3(PURALOX,200m2/g,SASOLGmbH,

Germany)andTiCl4(Fluka,titanium(IV)chloridesolution∼0.1M

in20%hydrochloric acid)wereusedasstartingmaterials.TiCl4

wasdilutedincooleddeionizedwaterundercontinuousstirring atatemperaturebelow333K.Then,␥-Al2O3powderwasslowly

added to theprepared solution. Next, 30vol% NH3 was slowly

addedtothesolutionuntilpH≥9.0wasachievedandagelwas formed.Thiswhite gelwasagedfor 24hunderambient condi-tions,ﬁltered,washedwithdistilledwater,andcalcinedat873K for 2hin air. In thesecond syntheticprotocol (P2), thebinary Ti/Aloxidesupportmaterialwassynthesizedbyasolgelmethod. In this synthetic protocol (P2), titanium and aluminum alkox-ides were used as precursors. First, aluminum-tri-sec-butoxide (97%,Sigma–Aldrich)wasdissolvedinthemixtureof propan-2-ol(99.5+%,Sigma–Aldrich)andacetylacetone(99.3%,Fluka).Then, titanium(IV)isopropoxide(97%,Sigma–Aldrich)wasaddedtothe solutionatroomtemperature. Next,0.5MHNO3 wasgradually

addedtothesolutionin ordertoinitiatethegelformation.The resulting gel was aged for 10 days under ambient conditions, ground, and baked at873Kfor 2hin air. The molefractionof TiO2 (i.e. TiO2=nTiO2/(nTiO2+nAl2O3))intheTiO2/Al2O3binary mixture wasequal to 0.3 (based onthe amounts of Ti and Al

(3)

precursorsusedinthesynthesis)inbothofthesyntheticprotocols. BaO/TiO2/␥-Al2O3 (whichwillbehereafterreferred asBa/Ti/Al)

sampleswithdifferentBaloadings(8and20wt%BaO)were syn-thesizedbyconventionalincipientwetnessimpregnationofthe Ti/AlbinaryoxidesupportmaterialswithBa(NO3)2.In the

cur-renttext,sampleswith8and20wt%BaOloadingwillbedenoted as 8Ba/Ti/Al(P1, P2) and 20Ba/Ti/Al(P1, P2), respectively. For a comprehensivediscussionofthestructuralcharacterizationofthe synthesized materials via SEM, EDX, BET, Raman spectroscopy andXRD,theanalysisofthesurfaceacidityoftheTi/Al(P1)and 8/20Ba/Ti/Al(P1)samplesviapyridineadsorption,aswellasNOx

adsorption/releasepropertiesviaFTIRandTPD,readerisreferred toourpreviouslypublishedwork[23,37]whichwillalsobeused frequentlyinthediscussionofthecurrentresults.␥-Al2O3(Puralox,

200m2_/g,_SASOL_GmbH,_Germany)_used_in_(P1)_synthetic_protocol

wasalsoutilizedasareferencematerialinsomeexperiments. TEMimageswere obtainedwitha resolution of 0.14nm on a JEM-2010 (200keV) microscopeequipped withan EDX spec-trometerwitha Si (Li) detectorhavinganenergy resolution of 130eV. The analyzed area in a typical EDX measurement was about10×10nm2_._The_samples_for_the_TEM_analysis_were

pre-paredby dispersing thepowders in anultrasonic ethanol bath andsubsequentdepositionofthesuspensionupona“holey” car-bonﬁlmsupportedonacopperTEMgrid.XPSdatawererecorded using a SPECS spectrometer with a PHOIBOS-100 hemispheri-calenergyanalyzerandamonochromaticAlK_␣X-rayirradiation (h=1486.74eV, 200W). FTIRspectroscopic measurementswere carriedoutintransmissionmodeinabatch-typecatalytic reac-tor[20]coupledtoanFTIRspectrometer(BrukerTensor27)and a quadrupole mass spectrometer(QMS,Stanford Research Sys-tems,RGA200)usedforTPDexperiments.AllFTIRspectrawere acquiredat323K.IntheTPDexperiments,alineartemperature rampwithaheatingrateof12K/minwasutilizedtoheatthe sam-plewithin323–1023K.TheQMSsignalswithm/zequalto18(H2O),

28(N2/CO), 30(NO),34(H2S),32(O2), 44(N2O/CO2), 46(NO2)and

64(SO2)weremonitoredduringtheTPDmeasurements.BETSSA

andporesizedistributionmeasurementswereperformedusinga MicromeriticsTristar3000surfaceareaandporesizeanalyzerby low-temperatureisothermaladsorption–desorptionofN2.

Thesulfurexposureexperimentswereperformedthroughfour differentconsecutivesteps.In theﬁrstspectralacquisitionstep, the sample was exposed to 0.6Torr of SO2(g)+O2(g) mixture

(SO2:O2=1:10)for1hat323Kand theﬁrstFTIRspectrum was

obtainedinthepresenceoftheSO2(g)+O2(g)mixtureoverthe

sample.Inthesecondstep,thesamplewasannealedto473Kin thepresenceoftheSO2(g)+O2(g),andaftercoolingto323K,the

secondFTIRspectrumwasacquired.Inthethirdstep,beforethe spectrumacquisition,thesamplewasannealedto673K(inthe presenceofSO2(g)+O2(g)),thencooledto323Kandsuccessively

evacuatedat323Kfor20min(Preactor<1×10−4Torr).Inthefourth

step,thesamplewasﬂashedto673Kinvacuumandafter cool-ingthesampleto323K,aFTIRspectrumwasobtained.ForNO2

adsorptionexperiments,thefreshsampleswerepre-poisonedby anexposureof0.6TorrofSO2(g)+O2(g)(SO2:O2=1:10)mixture

at323Kandwerefurtherheatedinthegasmixtureat473Kfor 30min.Afterhavingpumpedoutthereactor,poisonedsamplewas exposedto8TorrofNO2(g)at323Kfor20mininordertosaturate

thesurfacewithNOx.ThisisfollowedbytheevacuationandFTIR

spectraacquisitionat323K.

3. Resultsanddiscussion

3.1. TEMcharacterizationofTi/Almaterials

Fig.1representstheTEMmicrographsoftheTi/Al(P1,P2) sam-plesshowingthemorphologyofthesynthesizedsupportmaterials.

EDXanalysisofarea1giveninFig.1a,whichisabundantindarker domains,revealsthepresenceofTiO2andAl2O3containingspecies

withaTi:Alatomicratioequalto70:30.EDXanalysisofthebright domainssuchasarea2inFig.1a,demonstratedthattheseareas exclusivelycontainAl2O3.However,theFourieranalysisofarea2

inFig.1adidnotrevealwell-deﬁnedspots,mostlikelyduetothe smallparticlesizesandthedefectivenatureofthesealumina crys-tallites.TheimageinFig.1bshowsahigh-resolutionTEM(HRTEM) micrographofsuchabrightdomain.

ItisvisibleinFig.1athatthedarkerdomainscontainaggregates withsizesofabout10nm.Fig.1cshowsaHRTEMimageofone ofthesedarkerdomains.InterplanarspacingmeasuredonHRTEM micrographsofthesedarkerdomainsareinverygoodagreement withtheanatasephaseofTiO2.FastFourierTransform(FFT)picture

obtainedfromthisimageisalsoshownasaninsetinFig.1cand isingoodagreementwiththereﬂexesofanatasephase(ﬁleno. 21-1272inPDF-2Database,JCPDS-ICDD).

Fig.1d–fshowsTEMimagesoftheTi/Al(P2)sample.Itisclearly visibleintheseimagesthatthissampledisplaysarather homoge-nousmorphologyexhibitingauniformsponge-likeﬁnestructure whichisduetothemesoporouscharacterofthemixedTixAlyOz

oxidephasethatisabundantinTiOx/AlOxinterfaces.Area1given

in Fig. 1e highlights a characteristic region where this rather homogenousandporousstructureisapparent.Ti:Alelementalratio obtainedfromtheEDXspectrumofarea1inFig.1discloseto 15:85which isinvery goodagreementwiththenominal com-positionexpectedfrom therelative concentrationsof TiandAl precursorsusedin thematerialsynthesis (20:80).On theother hand,EDX elementalanalysisof a lesscharacteristic (minority) regioninFig.1d(labeledasarea2)revealsthepresenceofalmost exclusivelyAl2O3 particles.TheTEMimagegivenin Fig.1eand

HRTEMimagegiveninFig.1fshowthecharacteristicsponge-like structureofTi/Al(P2).FFTimageoftheTEMimagegiveninFig.1f isalsopresentedasaninset.FuzzycharacteroftheFFTimageis inagreementwiththeamorphouscharacteroftheTixAlyOzoxide

phase withsmall anddisorderedAlOxand TiOx domains.

How-ever,the steepintensitychanges near0.45nm fromthecenter showsthat disorderedstructure isprobablyduetothealumina spinelstructure(thisdistancecanbeassociated withd-spacing between(111)closepackedoxygenplanesinthelatticeof␥-Al2O3

(ﬁleno.29-0063inPDF-2Database,JCPDS-ICDD).FFTﬁltered pic-ture(Fig.1g)supportsthelackofanorderedcrystallinestructure. TheseobservationsindicatethattheTi/Al(P2)structureismostly composedofa TixAlyOz mixedoxideexhibitinga large

concen-tration ofTiOx/AlOx interfacialsites,while theTi/Al(P1) reveals

aninhomogeneousdistributionoflargerAl2O3andTiO2

agglom-erates witha relativelysmaller concentrationof suchinterface sites.

3.2. Poresizedistribution

BET SSA analysis demonstrates that Ti/Al(P1) and (P2) sam-pleshavesurfaceareavaluesof167.0and393.2m2_/g,_respectively [37]. SSA values of Ti/Al(P1, P2) support materials as well as 8(20)Ba/Ti/Al(P1,P2) materialsafter thermal treatments within 623–1023Kcanbefoundinoneofourformerreports[37].Thepore sizedistributionsofthesesupportmaterialsmonitoredduringN2

desorptionarepresentedinFig.2.TheTi/Al(P1)sampleshowsa relativelybroadpeakwithanaverageporesizeabout80 ˚Awhichis probablyduetotheconvolutionoftheporesizedistributionsofthe discreteTiO2andAl2O3domains,whileTi/Al(P2)sample

demon-stratesarelativelynarrowdistributionwithanaverageporesizeof 30 ˚A,correspondingtotheTixAlyOzmixedoxidenetwork.Ascanbe

seeninFig.2,theporesizeanalysisisconsistentwiththedissimilar surfacestructuresoftheTi/Al(P1)andTi/Al(P2)supportmaterials.

(4)

Fig.1.(a–c)TEMimagesofTi/Al(P1),region1:TiO2-richdomains,region2:␥-Al2O3-richdomains,and(d–g)Ti/Al(P2),region1:sponge-likestructureoftheTiOx AlOx

mixedoxide,region2:(dark)Al2O3domains.EDXspectraobtainedfromselectedregionsarealsogivenasinsets.

3.3. SOxinteractionwithTi/AlandBa/Ti/Almaterials:FTIR

Itshouldbenotedthatthereisnotaclearconsensusonthe vibrationalspectroscopicassignmentsoftheSOxspeciesonmetal

oxides(particularlyonAl2O3)duetotheheavilyconvolutednature

oftheFTIRsignalsoftheadsorbedSOxspecies.Assignmentsofthe

FTIRsignalsforvariousSOxspeciesdiscussedintheliteratureare

summarizedinTable1.

WeperformedFTIRinvestigationsofSO2(g)adsorptionon

␥-Al2O3.Theresults(datanot shown)are inagreementwiththe

formerinvestigations,revealingtheformationofsurfacesulﬁtes (SO32−)(notethatbulkAl2(SO3)3doesnotexist[60]),withoutany

additionaloxidationtosulfates.

300 250 200 150 100 50 0 0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 Ti/Al (P1) Ti/Al (P2) pore size (Å) pore volume (cm 3/g)

Fig.2.PoresizedistributionplotsforTi/Al(P1)andTi/Al(P2)supportmaterials.

TheadsorptionofSO2(g)+O2(g)onthe␥-Al2O3surfaceat323K

(Fig. 3a, spectrum i) revealsa major broad band at 1073cm−1 aswellasweakerandpoorlydeﬁnedadditionalfeaturesaround 1350,1250,1180and1005cm−1.ThepresenceofSO32−species

areapparentdue totheintense characteristic sulﬁtefeatureat c.a.1073cm−1(3)andtheshouldersat1049cm−1(3)aswellas

1005cm−1(1)[58–60].Minorfeaturesat1350(3)and1180cm−1

(1)canbeattributedtochemisorbedSO2 onabasic(O2−)

sur-facesites[59,60]whilethefeaturelocatedat1250cm−1 canbe associatedwiththeSO2adsorbedonAl3+Lewisacidsitesand/or

bidentatesulfates[59].AveryminorcontributionfromtheSO2(g)

speciesmayalsobepresentat1130(3),1150(1)and2499cm−1

(1+3,notshown)[60].

As the adsorption temperature increases to 473K (Fig. 3a, spectrumii)the1368,1102signalsbecomemore visiblewhich correspondtothe3 and1 modesofsurfacesulfatesonAl2O3, [60],respectively.Furthermore,theshoulderlocatedat1170cm−1 whichcanbeassociatedwithbulkAl2(SO4)3[56]startstobemore

discernible.Inaddition,arelativelyminorcontributionfrom sul-fatespeciesthatareinteractingwiththesurface–OHgroupswhich revealtypicalsignalsat1290and1080cm−1cannotbeexcluded [58].Increasingthetemperatureto673K(Fig.3a,spectraiiiandiv) resultsinthegrowthofthesurfaceandbulkaluminasulfatesin expenseofthesurfacesulﬁteandchemisorbedSO2species.

Theseresultsindicatethatsulﬁtesareinitiallyformedonthe aluminasurfaceduringtheSO2+O2 adsorptionat323Kwhileat

elevatedtemperatures(473–673K),thesespeciesare converted intorelativelystablesurfacesulfates(SO42−/Al2O3)aswellasbulk

Al2(SO4)3.Theformationofsulfatesonaluminasurfacewas

con-sideredinnumerousformerstudies[56,57,60,68].Typically,SO2

bindstotheacidicsites(i.e.coordinatelyunsaturatedaluminum cations,Al3+₎_forming_physisorbed_SO

2.TheadsorptionofSO2on

thebasicadsorptionsitesisfollowedbyacleavageofanAl Obond onthesurface(primarilyatOHsitesoratexposedoxygenatoms,

(5)

Table1

TypicalFTIRsignalsassociatedwithcommonSOxspecies[7,54–67].

Species Symmetry v1(cm−1) v3(cm−1) References

SO2(gas) C2v 1151 1362 [54] SO32−(freeion) C3v 961 1010 [54,55] SO42−(freeion) Td 1104 [54] S O O Al3+ (physisorbed) C2v 1135–1150 1300–1370 [7,55–58] S O O Al

(chemisorbed) C2v 1255(typeI),1189(typeII) [59]

S O O OH Al Cs 1148–1150 1330–1334 [55,59] O S O O Al Al C2v 1140 1320–1326 [55,59,60] S O O Al O C3v 1050–1065 1135 [56,59] S O O O O Al Al Al C3v 1045–1130 1380 [58,61] S O O O O Ti Ti Ti C3v 1005,1045 1370 [61] S O O O O C C (organicsulfates) C2v 1230–1150 1440–1350 [62,63] S O O S O O C2v ∼910 960–1000 S O O Fe O O C2v 1180,968 1375, 1025 [64] (NH4)2SO4 C2v 1090 1390 [65] BaSO4(surface) 1060 1120 [66] Ce(SO4)2(surface) 980 1340–1400 [67] Al2(SO4)3(bulk) 1190 [56] BaSO4(bulk) 1155,1248 [66] Ce(SO4)2(bulk) 1145–1240 [67]

O2−₎_leading_to_the_formation_of_chemisorbed_SO₃2−_._The_oxidation

ofadsorbedSO32−/SO2inoxygenatrelativelyhightemperatures

(673–773K)leadstotheformationofsurfacesulfatespecieswhich arecoordinatedtothemetalcationsoftheoxidesurfacethrough threeoxygenatoms[60].

In order to examine the sulfur accumulation on the TiO2

(anatase) surface, SO2+O2 adsorption experiments were also

performed on TiO2 (Fig. 3b). At 323K (Fig. 3b, spectrum i)

acomplexandaconvolutedgroupofsignalswereobservedwithin 1150–950cm−1whicharelikelytobeassociatedwithSO32−and/or

HSO3− species[60].At473K,thefeatureat1137cm−1 startsto

growtogetherwithanintensefeatureat1350cm−1.In analogy withthesimilarbehaviorobserved fortheAl2O3 surface, these

twobandsareassignedtoweaklyadsorbedmolecularSO2species.

This assignment is also consistent with the attenuationof the 1137cm−1 signalafterevacuationat673K(Fig.3b,spectrumiv).

(6)

1000 1100 1200 1300 1400 1049 1005 1350 1250

absorbance (arb.u.)

γ_-Al2O3 ii 1368 1170 ₉₉₅ 1102 1073 0.05 (a)

i

iii

iv

900 1000 1100 1200 1300 1137 0.05

ii

i

iii

iv

TiO2 (anatase) (b) 1002 1046 1083 1160 1287 1356 1366

1000

1100

1200

1300

1400

(c) Ti/Al (P1)

wavenumber(cm

-1

₎

1353 1173 1032 1235 1384

i

ii

iii

iv

0.05

900 1000

1100

1200

1300

1060 (d) Ti/Al (P2) 1360 1080 1160 1350 1260 0.05

i

ii

iii

iv

Fig.3. FTIRspectrafor0.6TorrSO2(g)+O2(g)(SO2:O2=1:10)co-adsorptionon(a)␥-Al2O3,(b)TiO2(anatase),(c)Ti/Al(P1),(d)Ti/Al(P2).(i)After1hexposuretoSO2(g)+O2(g)

at323K(spectrumwasobtainedinthepresenceofthegasmixture),(ii)aftersubsequentﬂashingofthesampleto473KinSO2(g)+O2(g)mixtureandcoolingto323K

(spec-trumwasobtainedinthepresenceofthegasmixture),(iii)aftersubsequentﬂashingto673KinSO2(g)+O2(g)mixtureandfurtherevacuationat323K(Preactor<1×10−4Torr),

(iv)aftersubsequentﬂashingthesampleto673Kinvacuumandcoolingto323K.

Notethatitisdifﬁculttofollowthefateofthe1350cm−1at ele-vatedtemperaturessincenewbandsstartstogrowinthesame spectralregionat673K.Thefeaturelocatedat1287cm−1together withthe1083cm−1featurecanbeattributedtobidentatesulfates interactingwiththeadsorbedwaterat473–673K[58]and ortho-chelatingbidentatesulfateswhichrevealvibrationalfeaturesina similarrange[64,69].Aftertheevacuationat673K,relatively well-resolvedfeaturesareobservedintheFTIRdata(Fig.3b,spectrum iv).Thusthebandsat1366–1356,1160,1046and1002cm−1are assignedtosurfacesulfatesontheTiO2surfacewhichareformed

aftertheoxidationoftheadsorbedSOxspeciesat673K[61,70]. Fig.3canddshowstheIRspectraoftheTi/Al(P1)andTi/Al(P2) supportmaterialsaftertreatmentinSO2+O2atdifferent

temper-atures(323–673K),followedbyevacuation.TheIRspectradueto

theSO2+O2adsorptiononTi/Al(P1,P2)at323and473K(spectra

iandiiinFig.3candd)exhibitastrongresemblancewherethe mostprominentfeatureisthebroadIRsignalat∼1050cm−1 cor-respondingtosulﬁtespecies.Themaindifferencebetweenthese twospectraisthepresenceoftheshoulderat1130cm−1forthe Ti/Al(P1) sample which maybe likely due toweakly adsorbed SO2species,similartotheonesthatwereobservedforbothpure

anataseand ␥-alumina surfaces.Increasing the temperatureto 673KinthepresenceofSO2+O2leadstotheformationof

addi-tionalfeaturesassociatedwithsurfacesulfatesat1350–1360and 1160–1175cm−1onbothTi/Al(P1,P2)samples.Hightemperature SO2+O2 exposure alsoseem to triggerthegrowth of the

con-volutedbandat 1230–1260cm−1 on both samples,which may indicatetheformationofbidentatesulfates.Thepresenceofthe

(7)

1160–1170cm−1signalinFig.3canddalsosuggeststhepossible formationofbulkAl2(SO4)3species.Ageneralcomparisonofthe

lineshapesoftheFTIRspectragiveninFig.3canddrevealsthat sulfatecontainingsurfacedomainsontheTi/Al(P1)sampleare rela-tivelymoreorderedandwell-crystallizedwithrespecttothatofthe Ti/Al(P2)sample,whichisconsistentwiththesharperandbetter resolvedsulfatebandsvisibleinFig.3c.

Analogous experiments were also performed with Ba con-taining samples. In these studies, conventional 8(20)Ba/Al benchmark samples that do not contain TiO2, were compared

with 8(20)Ba/Ti/Al(P1,P2) samples containing TiO2. After the

introductionofa SO2+O2 mixtureonto the8Ba/Al and 20Ba/Al

samplesat323K(Fig.4aandc,spectrai)amajorbroadbandat 1080cm−1,whichcanbeattributedtosulﬁtespecies,isobserved. Aftersubsequentheating at473K (intheSO2+O2 gasmixture)

no substantial changes are observed in the case of the 8Ba/Al sample(Fig.3a, spectrumii),while in thecaseof 20Ba/Al,two differentfeaturesat1250and1160cm−1becomeapparent.These featurescanbeassignedtobulkBaSO4[66].Subsequentheating

at673KinSO2+O2mixture(Fig.4a,spectrumiii)andinvacuum

(Fig.4a,spectrumiv)leadstotheappearanceofthebulkBaSO4

bandsforthe8Ba/Alsample.Theadditionalsignalat1350cm−1 is associated with surface sulfates on alumina domains. This bandis morepronouncedforthe8Ba/Alsample, incomparison to the 20Ba/Al surface due to the larger number of exposed alumina sites in the former case as a result of the lower BaO loading.

Thesulfationexperimentsof8Ba/Ti/Al(P1)and20Ba/Ti/Al(P1) samplesarepresentedinFig.4bandd. AfterSO2+O2 exposure

at 323K, a set of broad and overlapping bands appear within 1100–980cm−1indicatingtheformationofvarioussulﬁtespecies. Interestingly,onlyinsigniﬁcantlyminorchangesareobservedfor the8Ba/Ti/Al(P1)sampleevenafterhightemperaturetreatment (Fig. 4b, spectra ii–iv) indicating that the formation of surface aluminum sulfate and bulkBaSO4 species is rather suppressed.

The changes observed for the 20Ba/Ti/Al(P1) sample are more pronounced (Fig. 4d): bulk BaSO4 related features at 1260 and

1160cm−1appearafterheatingat473Kandthesebandsbecome slightlymoreintenseaftersubsequentheatingat673K.Itisclear thatsuppressionofbulk BaSO4 formation byTiO2 promotionis

effectiveforthe8Ba/Ti/Al(P1)samplewhileitislessefﬁcientfor 20Ba/Ti/Al(P1)sample.Wehavereportedinourpreviousstudies thatTiO2domainsfunctionasanchoringsitesforBaOsitesandlimit

thesurfacediffusionofBaOclusters[23,37].Thus,itislikelythat forlowBaOloadings,mostoftheBaOdomainsonthe8Ba/Ti/Al(P1) samplecaneffectivelybindtoTiO2domains,andpreventthe

sin-teringofBaOdomains.Thus,forlowBaOloadings,TiO2domains

canefﬁcientlysuppresstheformationoflargeBaOclusterswhere thermallystablebulkBaSO4canform.Ontheotherhand,forhigher

BaOloadings,asinthecaseof20Ba/Ti/Al(P1),alargerfractionof theBaOdomainsarelocatedonthealuminasurfacewheretheycan diffusefasterandformlarger3DBaOclusterswhichcanenablethe formationofbulkBaSO4 asinthecaseof20Ba/Al(Fig.4c). Itis

worthmentioningthatsimilarexperimentswerealsoperformed on8(20)Ba/Ti/Al(P2)samples(datanotshown)andaqualitatively similarbehavior wasobservedindicatingthesigniﬁcanceofthe Ba/TiratiointhepromotionaleffectofTiinSOxuptake.

Inordertoquantitatively comparetheamountofSOxstored

ontheinvestigated surfaces,XPSanalysisofthepoisoned sam-pleswasperformed.Becauseoftherelativelylowsensitivityofthe XPStechniquetowardssulfur,poisoningprocedurewasmodiﬁed toobtainabettersignaltonoiseratioandmoreaccurateatomic ratiovalues.Thus,beforetheatomicratiodeterminationbyXPS,the sampleswerepoisonedin10TorrofSOx(SO2:O2=1:10)at673K

for30min.TheshapeoftheFTIRspectrarecordedafterthis poi-soningprocedure(datanotshown)wereinverygoodagreement

withthecorrespondingspectrapresentedinFigs.3and4.Binding energy(BE)ofS2p(168–170eV)XPSsignalindicatedthepresence ofmainlysulfates(i.e.S6+_states)_on_all_samples._It_is_worth

men-tioningthatalthoughinsituFTIRdatagiveninFig.4bsuggestthe presenceofmostlysulﬁtespeciesonthe8Ba/Ti/Al(P1)sample,XPS dataindicatesthepresenceofpredominantlysulfatespecies.This maybeassociatedwiththeexposureofthe8Ba/Ti/Al(P1)sample toatmospherebeforetheXPSanalysisresultingintheoxidation ofsulﬁtesintosulfates.Fig.5demonstratesthatunderidentical poisoningconditions,Ti/Al(P1,P2)samplesaccumulateupto3–5 timesmoresulfurthan␥-Al2O3.Thisbehaviormightbe

originat-ingfromthelargerSSAof theTi/Almaterialsincomparison to ␥-Al2O3aswellasthedecreasingnumberoftotalLewisacidsites

inmixedTi/Aloxides.ThelessacidicTi/Almixedoxidesurfaces candemonstrateahigherafﬁnitytowardsacidicSOx.Theacidityof

TiO2 Al2O3mixedoxidesaswellas␥-Al2O3hasbeeninvestigated

inpreviousstudies[23,71]bypyridineadsorption.Itwassuggested thatadditionofTiO2toAl2O3introducessomemediumstrength

Lewisacidsites,[23]howeveritwasdemonstratedinanotherstudy thatthetotalamountofLewisacidsitesisdecreasedby30%for TiO2 Al2O3incomparisontopurealumina[71].

Analysis of Fig. 5 also reveals that for an identical mass of eachmaterial(i.e.20mg),8Ba/Ti/Al(P1,P2)samplesaccumulate alesserquantityofsulfurthanthe8Ba/Alsample,demonstrating thefavorablepromotionaleffectofTionlimitingsulfur accumula-tion.Furthermore,8Ba/Ti/Al(P2)sampleseemstostoremoreSOx

(perunitsampleweight)thanthe8Ba/Ti/Al(P1)samplewhichcan beexplainedbythelargerSSAoftheformermaterial(185cm2_/g

vs.150cm2_/g,_{respectively)}_[37]_._However,_it_is_important_to

men-tion that the promotion of 8Ba/Al sample with TiO2 favorably

decreasesthetotalSOxuptake(perunitsampleweight)

regard-lessofthepreparationmethod(i.e.P1orP2)(notethattheSSA ofthefresh8Ba/Al sampleis185cm2_/g_[20]_)._This_trend_is

par-tiallyreversedwhentheBaOloadingisincreasedto20wt%.Fig.5 revealsthatalthough20Ba/Ti/Al(P1)sample(SSA=79cm2_/g_[37]₎

storesalesserquantityofSOx(perunitsampleweight)thanthat

of20Ba/Alsample(SSA=126cm2_/g_[20]_);_{20Ba/Ti/Al(P2)}_sample

(SSA=173cm2_/g_[37]₎_stores_a_greater_total_amount_of_SO x.These

resultssuggestthatforhigherBaOloadings,thepromotionaleffect ofTiO2isweaker.BesidestheseSSAtrends,thisobservationmay

alsoarisefromthefactthatnotalloftheBaOdomainsarelocated ontheTiO2sitesforhighBaOloadingsandalargefractionofthe

BaOdomainsdirectlyinteractwiththeunderlyingaluminasitesas inthecaseof20Ba/Alsystem.

3.4. InﬂuenceofSOxpoisoningontheNOxadsorption

NOxadsorptiononfreshTi/Al(P1,P2)andfresh8(20)Ba/Ti/Al(P1,

P2)samplesviaFTIRtechniqueandthecorrespondingassignments oftheobservedvibrationalbandswerethoroughlydiscussedinone ofourrecentreports[37]andthuswillnotbereiteratedhere.In thelightofthesepreviousstudies,similarNOxadsorption

experi-mentswerealsoperformedonthesulfur-poisonedmaterialsand theresultswereanalyzed.

Fig.6a comparestheNOxuptake characteristicsof thefresh

and poisoned ␥-Al2O3 surface at 323K.After the saturationof

thefresh␥-Al2O3 surfacewithNO2(g),vibrationalfeatures

asso-ciatedwithdifferenttypesofnitratespecieswereobserved.These nitratespeciesadsorbedon␥-Al2O3wereintheformofbridged

(1258,1628cm−1),bidentate(1300,1604cm−1)andmonodentate nitrates(1300, 1564cm−1)[37,72].Additionally,theweakband locatedat∼1958cm−1 thatwasformedafterNOxadsorptionon

thepoisonedsurfaceisassociatedwiththeadsorbedNO+_and/or

weaklyadsorbedN2O3[73].Thepoisoningofthe␥-Al2O3surface

bySO2+O2decreasestheintensitiesofthenitratesignals,

(8)

1000 1100 1200 1300 1400 1043 1350 1250

absorbance (arb.u.)

8Ba/Al ii 1160 995 1080 0.05 (a)

i

iii

iv

900 1000 1100 1200 1300 1250 0.05

ii

i

iii

iv

8Ba/Ti/Al (P1) (b) 995 1046 1100 1160 1350 1000 1100 1200 1300 1400 1080 (c) 20Ba/Al

wavenumber (cm

-1

)

1350 1160 1040 1250

i

ii

iii

iv

0.05 900 1000 1100 1200 1300 980 1120 1046 (d) 20Ba/Ti/Al (P1) 1100 1160 1350 1260 0.05

i

ii

iii

iv

Fig.4.FTIRspectrafor0.6TorrSO2(g)+O2(g)(SO2:O2=1:10)co-adsorptionon(a)8Ba/Al,(b)8Ba/Ti/Al(P1),(c)20Ba/Al,(d)20Ba/Ti/Al(P1).(i)After1hexposureto

SO2(g)+O2(g)at323K(spectrumwasobtainedinthepresenceofthegasmixture),(ii)aftersubsequentﬂashingofthesampleto473KinSO2(g)+O2(g)mixtureand

coolingto323K(spectrumwasobtainedinthepresenceofthegasmixture),(iii)aftersubsequentﬂashingto673KinSO2(g)+O2(g)mixtureandfurtherevacuationat323K

(Preactor<1×10−4Torr),(iv)aftersubsequentﬂashingthesampleto673Kinvacuumandcoolingto323K.

inFig.6athatSOxpoisoningdecreasestherelativeratioof

mon-odentateandbidentatenitrates(1300cm−1)tothebridgednitrates (1258cm−1).Thisobservationsuggeststhatsulfatespeciesonthe ␥-Al2O3surfacesuppresstheformationofalltypesofnitrates,

par-ticularlythemonodentateandbidentatenitratesbyoccupyingthe correspondingadsorptionsites.

Theappearanceofthebandsat1375and1100cm−1inFig.6a forthepoisonedsampleindicatestheformationofsurfacesulfates. Althoughsulfateformationonthe␥-Al2O3surfaceintheabsence

ofNOxspecies(Fig.3a)requiresarelativelyhightemperature(i.e.

673K),sulfateformationcanreadilyoccurbyexposingthealumina surfacetoSO2(g)+O2(g)at473KfollowedbyNO2(g)exposureat

323K.Itisapparentthat,NOxspeciesfunctionasefﬁcientoxidizing

agentsintheoxidationoftheSO2 and SO32−tosulfatesonthe

␥-Al2O3surface.Thefollowingreactionpathwayscanbesuggested

fortheinteractionbetweensurfaceSOxspeciesandtheadsorbed

NOxspecieson␥-Al2O3:

I:

SO32−(ads)+NO2(ads,g)→ SO42−(ads)+NO(g) (1)

SO32−(ads)+2NO2(ads,g) →SO42−(ads)+N2O3(ads) (2)

N2O3(ads)→ NO2−(ads)+NO+(ads) (3)

II:

3NO2(g)+O2−(s)→2NO3−(ads)+NO(g) (4)

(9)

Fig.5. Relativesulfurcontent(i.e.percentileofsulfuratomsonthesurfacewith respecttoallotheratoms)oftheinvestigatedsamplesaftersulfurpoisoning(as describedinthetext)determinedbyexsituXPSanalysis.

NO2−(ads)+NO2(ads,g) →NO3−(ads)+NO(g) (6)

NO2−(ads)+2NO2(ads,g) →NO3−(ads)+N2O3(ads) (7)

PathwayIsuggeststhatthesulﬁtespeciesonthe_␥-Al2O3surface

thatareformedintheprocessofpoisoningcanbedirectlyoxidized tosulfateswiththehelpofNO2(g)orweaklyadsorbedmolecular

NO2.Ontheotherhand,pathwayIIproposesanalternativeroute

forthesulfateformation,wheresurfacenitratespeciesfacilitate theoxidationofsurfacesulﬁtes(orweaklyadsorbedmolecularSO2

species).Inthislatterroute,duringthesulfateformation,nitrates areconsumed at theexpense ofnitrite and NO(NO+₎_or _N

2O3

generation.Theobservationof1958cm−1featureintheFTIR spec-trumcorrespondingtothepoisonedsampleinFig.6a,which is notpresentinthecorrespondingspectrumforthefreshsample, supportstheformationofN2O3/NO+(reactions(2)and(7)).Itis

difﬁculttoassesswhichofthesetwopathwaysisfavoredonthe surfaceunderthecurrentexperimentalconditions.However,by takingintoaccountthefacileformationofnitratespeciesonthe ␥-Al2O3surface,itcanbearguedthatpathwayIImaypresumably

beoccurringmorereadilythanpathwayI.

Fig. 6b, presents analogous poisoning and subsequent NOx

uptakeexperimentsconductedonpureTiO2(anatase).TheLewis

acidsitesontheanatasesurfacerevealfourandﬁve-coordinated Ti4+_ions_(refereed_as_␣_and_␤_sites,_{respectively)}_where_␣-Lewis

acid sites (with two oxygen vacancies) favor bidentate (1578 andshoulder at∼1220cm−1)and bridge(1627 and1236cm−1) nitrateformationwhilethe␤-sites(Ti4+_with_one_oxygen_vacancy)

favortheformationofmonodentatenitrates(groupoffeaturesat 1550–1500cm−1andpeakat1282cm−1)[37,74–77].SimilarFTIR signals,thoughwithreducedintensitiesofallNOxvibrational

sig-nalsarealsoobservedforthepoisonedanatasesamplemarkedin redinFig.6b.Thepoisonedanatasesamplealsoshowssome addi-tionalbandsat1359and1030–1070cm−1thatareassociatedwith thesulﬁteandsulfatespecies.

SimilarexperimentswerealsoperformedonTi/Al(P1,P2) sur-facesasshowninFig.6candd.ItisreadilyseenthattheNOxuptake

issuppressedbySOxpoisoningonbothsurfaces.ItisvisibleinFig.6

thattheextentofthenitratesignalsuppressionismorepronounced forTi/Al(P1,P2)samplesincomparisonwithpure␥-Al2O3andTiO2.

ThisresultisinagreementwiththeXPSdatagiveninFig.5, indi-catingasigniﬁcantlyhigherafﬁnityofTi/Al(P1,P2)towardsSOx

thanthatofpure␥-Al2O3.Sucha behaviorcanbeexplainedby

anincrease intheSSAvalues and adecreasein thetotal num-berofLewisacidsitesuponTiO2promotion.Asimilarinteresting

behaviorwasalsoreportedintheliteratureforTiO2–ZrO2mixed

oxidesystems.Forinstance,itisknownthatpureTiO2(anatase)or

pureZrO2 (monoclinic/tetragonal)haveverylimitedNOxstorage

capacities[78].Ontheotherhand,whenthesetwodifferenttypes

ofoxidesarecombinedtoobtainanamorphousTiO2–ZrO2mixed

oxide,NOxstoragecapacitycanbeimprovedbymorethananorder

ofmagnitude,whichwasattributedtotheformationofnewLewis basicsitesattheTiO2–ZrO2hetero-junctions[78].Inanalogywith

theseobservations,Ti/Al(P1,P2)surfacesmayalsocontainnewand strongSOxadsorptionsites,whicharenotpresentoneitherpure

TiO2or␥-Al2O3surfaces.

WhentherelativepoisoningofTi/Al(P1)iscompared tothat ofTi/Al(P2) (Fig.6cand d),itis visiblethatnitratefeaturesare suppressedtoalesserextentonTi/Al(P2),althoughtheintensities ofsulfatefeatures(1360and1100cm−1)arestrongeronTi/Al(P2) (Fig.5d).Thisobservationsuggeststhattheporousanddisordered surfacemorphologyoftheTi/Al(P2)sampleresultsinalarger over-allSOxuptake,howeverduetoitslargesurfacearea(SSAoffresh

Ti/Al(P1)and Ti/Al(P2)are167and393m2_/g,_respectively _[37]_),

Ti/Al(P2)surfacestillpossessesalargernumberofNOxbindingsites

withrespecttoTi/Al(P1).Ontheotherhand,suchacomparison basedsolelyonIRintensitiesshouldbeconsideredwithcaution, astheIRabsorptioncross-sectionsofadsorbednitratespecieswith dissimilaradsorptionconﬁgurationscanbesigniﬁcantlydifferent, whichmayrenderadirectIRintensitycomparisonrelatively inac-curate.

Fig. 7 compares the NOx adsorption properties of

8(20)Ba/Ti/Al(P1, P2), samples before and after SOx

adsorp-tion at323K.The FTIRresultscorrespondingtotheadsorption of NO2 onthefreshsampleswerealready discussed elsewhere [37].Brieﬂy,uponNO2 adsorption,thefreshsamplesurfacesare

characterized by adsorption bands due to the presence of the bulk(ionic)Ba-nitrateslocatedat∼1320,∼1440and∼1480cm−1, surface (bidentate) Ba-nitrate features located at 1585, 1565, 1300cm−1andadditionalnitratebandsat1583and∼1630cm−1 thatareassociated withnitratesadsorbedontheTiO2 domains

havingbidentateandbridgedconﬁgurations.AfterSOxpoisoning

ofthesampleswithSO2+O2at473K,itisclearlyvisiblethatthe

NOx storage capacities decrease signiﬁcantly. Additionally, the

presenceofminorbandsat∼1150and∼1245cm−1 impliesthe existenceofsulfatespeciesonallofthesamples.Formationof sul-fatespeciesonallofthesamplesaftersubsequentNO2adsorption

at 323Kdemonstrates theefﬁcient oxidizingcapability of NO2

whichcanreadilyoxidizesulﬁtespeciesthatareformedduring the SO2+O2 exposure. It is worth mentioning that the similar

experimentsperformedon8(20)Ba/Alsamples(datanotshown) revealedqualitativelysimilarresultsindicatingthesuppressionof NOxuptakeuponsulfurpoisoning.

3.5. ThermalstabilityoftheadsorbedNOxspeciesonthe

sulfur-poisonedmaterials

In order to have a betterunderstanding of theinﬂuence of SO2+O2 treatmenton thethermal stability and thedesorption

propertiesoftheNOxspecies,TPDexperimentswerepreformed. Fig.8showstheNOxdesorptionproﬁlesfromthefreshandSOx

poisoned␥-Al2O3andTiO2(anatase)benchmarksamples.For

clar-ity,onlythem/z=30,m/z=32andm/z=46desorptionchannelsare shown.Forthefreshaluminasurface,twomajorNOandNO2

des-orptionfeaturesareobserved,thatisingoodagreementwiththe literaturedata[20,79].TheﬁrstNOxdesorptionfeatureat387K

is associated withthe desorption of monodentate nitrates and weaklyboundN2O3and/orNO+specieswhichdesorbintheform

ofNO2andNO(almostwithoutO2).Thesecondmajordesorption

bandhasitsmaximumat625Kandisassociatedwiththe desorp-tion/decompositionofbridgedandbidentatenitrateswhichyield agreaterNO2desorptionsignalalongwithNOandO2[20,79].

TheSO2-poisoning ofthealuminasurfacehasastrong

inﬂu-enceontheNOxdesorptionfeatures.AsitisseeninFig.8b,sulfur

(10)

1000 1200 1400 1600 1800 2000 2200

absorbance (arb.u.)

γ_-Al2O3 (a) 1958 1628 1604 1375 1300 1259 1102 10411003 Fresh sample 1564 0.2 Poisoned sample 1200 1400 1600 1800 2000 TiO2 (anatase) (b) Poisoned Sample Fresh Sample 1627 1578 1550 1359 1282 1236 1070 1028 0.2 1000 1200 1400 1600 1800 2000 2200 (c) Ti/Al (P1)

wavenumber (cm

-1

)

Fresh Sample Poisoned Sample 1631 1580 1294 1258 1040 0.2 1200 1400 1600 1800 2000 (d) Ti/Al (P2) Fresh Sample Poisoned Sample 1640 1584 1237 1212 1100 1360 1040 0.2

Fig.6. FTIRspectramonitoredafterNO2adsorptionat323Konfresh(blackspectra)andpoisoned(redspectra)(a)␥-Al2O3,(b)TiO2(anatase),(c)Ti/Al(P1),(d)Ti/Al(P2)

samples.Poisoningwasperformedbyexposingthesamplesto0.6TorrSO2(g)+O2(g)(SO2:O2=1:10)at323K,followedbyheatinginthegaseousmixtureat473Kfor30min

andaﬁnalevacuationat323K(Preactor<1×10−3Torr).NOxuptakewasperformedbyexposingthesamplesto8TorrofNO2(g)at323Kfor20min,followedbyevacuation

to<1×10−3_Torr._All_spectra_were_acquired_in_vacuum_at₃₂₃_K._(For_{interpretation}_of_the_references_to_color_in_this_ﬁgure_legend,_the_reader_is_referred_to_the_web_version

ofthisarticle.)

species.Forthepoisonedsample,themajorfeaturearound380K (associatedwiththemonodentatenitratesorweaklyboundN2O3

desorption)appearstoalargerextentthandesorptionof biden-tate/bridgednitrateat675K.Thisﬁndingisalsoconsistentwiththe FTIRresultsobtainedforNO2adsorptiononthesulfated␥-Al2O3

surface,yieldingthebandcorrespondingtoadsorbedNO+_/N 2O3

speciesat 1958cm−1 (Fig.6a).Relativelylower intensityofthe 1958cm−1 band(Fig.5a)incomparisontotheotherNOxbands

couldbeassociatedwithitslowIRabsorptioncross-section.Thus, SOx species seem tocompete withthe nitratespecies for

sur-faceadsorptionsites.Furthermore,inthepresenceofSOxspecies,

stronglyadsorbednitratesarepartiallyconsumedandconverted intoweaklyboundN2O3(reactions(2)and(7))orNO+ (reaction (3))whilesomeofthesulﬁtesandchemisorbedSO2speciesare

con-vertedintosulfatesthroughreactionssimilartotheonesproposed above(pathwayII).

Fig.8canddshowstheTPDproﬁlesforNOxdesorptionfrom

freshandSOxpoisonedTiO2(anatase)samples.FreshTiO2sample

presentsaweakshoulderat455Kandtwobroadfeatureslocated at610and900KinNOdesorptioncurve.Inthelightoftheformer studies[74–77]theshoulderat455Kcanberelatedto molecu-larlyboundNO2 speciesandmonodentatenitrates.Thebandat

(11)

1000 1200 1400 1600 1800 2000 2200 8Ba/Ti/Al (P2) 8Ba/Ti/Al (P1)

absorbance (arb.u.)

(a) 1630 1440 1310 1240 1150 1040 Fresh sample 1570 0.2 Poisoned sample 1200 1400 1600 1800 2000 1450 (b) Poisoned Sample Fresh Sample 1635 1580 1485 13051260 1150 1040 0.2 1000 1200 1400 1600 1800 2000 2200 1040 1440 20Ba/Ti/Al (P1) (c)

wavenumber (cm

-1

)

Fresh Sample Poisoned Sample 1630 1570 1320 1250 1150 0.2 1200 1400 1600 1800 2000 1320 1440 1570 1630 20Ba/Ti/Al (P2) (d) Fresh Sample Poisoned Sample 1250 1155 1040 0.2

Fig.7.FTIRspectramonitoredafterNO2adsorptionat323Konfresh(blackspectra)andpoisoned(redspectra)(a)8Ba/Ti/Al(P1),(b)8Ba/Ti/Al(P2),(c)20Ba/Ti/Al(P1),(d)

20Ba/Ti/Al(P2)samples.Poisoningwasperformedbyexposingto0.6TorrSO2(g)+O2(g)(SO2:O2=1:10)at323K,followedbyheatinginthegaseousmixtureat473Kfor

30minandaﬁnalevacuationat323K(Preactor<1×10−3Torr).NOxuptakewasperformedbyexposingthesamplesto8TorrofNO2(g)at323Kfor20min,followedby

evacuation.Allspectrawereacquiredinvacuumat323K.(Forinterpretationofthereferencestocolorinthisﬁgurelegend,thereaderisreferredtothewebversionofthis article.)

610Kisattributedtothedesorption/decompositionofthebridged andbidentatenitrates.Thehightemperaturedesorptionfeatureat 900Kcanbeattributedtothedesorptionofstronglyboundnitrates. Itisknown[23]thatphasetransitionfrombulkanatasetobulk rutilestartsatT>800K.ThereforethestrongNOxdesorption

sig-nalat900KinFig.8ccouldbeassociatedwiththedrasticdecrease inthesurfaceareaofTiO2asaresultofthephasetransitionfrom

anatasetotherutilephase.Notethatduringtheactivationofthe freshandpoisonedanatasesurfacesbeforetheTPDexperiments, TiO2sampleswerenotexposedtotemperatureshigherthan623K

inordertopreservetheanatasephasepuritythus,theactivation wasperformedviaO2(g)ratherthanNO2(g).

The poisoned TiO2 sample shows an additional minor

des-orption feature at 345K (Fig. 8d). We attribute this feature to thedesorptionofweaklyadsorbedN2O3/NO+ (desorptionoccurs

mainlyintheformofNO+NO2).Thedesorptionfeatureat450Kis

associatedwithweaklyadsorbedmolecularspeciesand/orwith monodentatenitrates. TPDproﬁleof thepoisoned TiO2 sample

(Fig.8d)revealsa decreasein the450K desorptionsignalwith respecttothatofthefreshsample(Fig.8c)inagreementwiththe FTIRdata(Fig.6b)indicatinga suppressionofthemonodentate nitratesignalsat1550and1282cm−1 uponpoisoning.Themain desorptionbandforthepoisonedTiO2sampleoccursat615K

(12)

1000 900 800 700 600 500 400

QMS intensity (arb.u.)

γ_{-Al2O3 fresh} _(a)

NO2 O2 NO 385 620

1x10-6

1000 900 800 700 600 500 400 γ_{-Al2O3 poisoned} (b) 675 630 NO NO2 O2

1x10-6

380 1000 900 800 700 600 500 400 x4

TiO2 (anatase) fresh (c)

Temperature (K)

NO2 O2 x4 NO

1x10-6

455 610 ₉₀₀ 900 800 700 600 500 400

TiO2 (anatase) poisoned (d)

x4 x4 NO2 NO O2 345 450 615 545 805

1x10-6

Fig.8.TPDproﬁlesobtainedfromfreshandpoisoned(aandb)␥-Al2O3,(candd)TiO2(anatase)samplesaftersaturationwith8TorrNO2(g)at323Kfor20min.Black,blue

andredcurvescorrespondtom/z=30(NO),m/z=32(O2)andm/z=46(NO2)signals,respectively.Poisoningwasperformedbyexposingthesamplesto0.6TorrSO2(g)+O2(g)

(SO2:O2=1:10)at323K,followedbyheatinginthisgaseousmixtureat473Kfor30minandaﬁnalevacuationat323K.(Forinterpretationofthereferencestocolorinthis

ﬁgurelegend,thereaderisreferredtothewebversionofthisarticle.)

theformofbridged/bidentatenitrates.Animportantaspectofthe poisonedanataseTPDproﬁleisthelackofastrongdesorption sig-nalatT>800KsuggestingtheblockingofthestrongNOxadsorption

sitesbysulfates.

Fig.9aandbpresentstheNOxdesorptionproﬁlesoffreshand

poisonedTi/Al(P1)samples.Takingintoconsiderationthesurface morphologyofthefreshTi/Al(P1)sample,whichiscomposedof rel-ativelyorderedandwell-crystallizedAl2O3andTiO2domains,the

lowtemperaturepeaksofNOandNO2desorptionat355and395K

canbeattributedtodesorptionofweaklyadsorbedN2O3 (NO+)

and/ormonodentatenitratesonthealuminadomains.Desorption ofbridgedandbidentatenitratesoccursintheformofNO2+O2

at625Kwhilethe780Kdesorptionsignalisobservedintheform ofNO+O2.Thedesorptionsignalat625KinFig.9amayoriginate

frombothaluminaandtitaniadomainsofTi/Al(P1)sincesimilar featuresexistintheTPDproﬁlesofpureTiO2andAl2O3samples

(Fig.8aandc).Ontheotherhand,themajordesorptionsignalin Fig.9aappearingat780Kwithashoulderat870Khasasigniﬁcantly higherdesorptiontemperaturethananyofthepureAl2O3

desorp-tionfeatures.Thus,the870KfeatureinFig.9acanbeattributedto thedecompositionofstronglyboundnitratesontheTiO2domains

orontheperipheraladsorptionsiteswhicharelocatedatthe inter-facebetweenTiO2andAl2O3domains.Itisimportanttomention

thattheoriginofthisdesorptionfeatureisprobablyquitedifferent thanthe900KdesorptionsignalinFig.8cwhichisobservedfor pureanatase.Ourprevious[23,37]XRD,Ramanspectroscopyand BETanalysisresultsindicatethatonTi/Al(P1,P2)surfaces,anatase torutileand␥-Al2O3→␣-Al2O3phasetransformationphenomena

(13)

1000 900 800 700 600 500 400

QMS intensity (arb.u.)

Ti/Al (P1) fresh (a)

NO NO2 O2 780 870 625 355 395

1x10-6

x4 x4 1000 900 800 700 600 500 400 Ti/Al (P1) poisoned (b) x4 x4 370 720 630 NO O2 NO2

1x10-6

1000 900 800 700 600 500 400 885 Ti/Al (P2) fresh (c)

Temperature (K)

x4 x4 NO O2 NO2 370 645 745 815 920

1x10-6

1000 900 800 700 600 500 400 NO O2 NO2 x4 x4 Ti/Al (P2) poisoned (d) 440 560 640 850

1x10-6

Fig.9. TPDproﬁlesobtainedfromfreshandpoisoned(aandb)Ti/Al(P1),(candd)Ti/Al(P2)samplessaturatedwith8TorrNO2(g)at323Kfor20min.Black,blueand

redcurvescorrespondtom/z=30(NO),m/z=32(O2)andm/z=46(NO2)signals,respectively.Poisoningwasperformedbyexposingthesamplesto0.6TorrSO2(g)+O2(g)

occuronlyattemperatureshigherthan1073K.Accordingly,onlya moderatedecreaseinthespeciﬁcsurfaceareavaluesareobserved atT≤900K.Thus,itisunlikelythatthe870KfeatureinFig.9aisdue toanatasetorutileor␥-Al2O3→␣-Al2O3phasetransitionsorany

otherassociateddrasticattenuationinthesurfaceareaofthe sam-ples.AnotherinterestingaspectoftheTPDdatagiveninFig.9isthe relativelylowintensityoftheO2desorptionsignalsincomparison

withthatoftheNOdesorptionsignals.Onepossibleexplanation forthisobservationisthat,duringthevacuumannealingand acti-vationoftheTi/Al(P1,P2)samplesinsidetheIR-TPDcell,someof thesurfaceTiO2domainsmaybepartiallyreduced.Thesereduced

TiO2surfacedomainscontainingoxygendefectsarehealedbythe

oxygenspeciesthataregeneratedduringthedecompositionofthe nitratespeciesinthecourseoftheTPDexperiments.

AfterpoisoningoftheTi/Al(P1)sample,desorptionoftheweakly adsorbed NOx species at < 400K proceedsmainly in the form

of NO2 (Fig. 9b). One can also see that NO2 desorption peak

at625–630Kissigniﬁcantlydiminished.Furthermore,themost prominentNOxdesorptionfeatureshiftstoalowertemperature

by60Kand islocatedat720Kfor thepoisonedTi/Al(P1) sam-ple.Thisis accompaniedbythealmostcomplete disappearance ofthehightemperatureNOdesorptionsignalat870K.Itisclear that thepresenceofsulﬁteand sulfate speciesontheTi/Al(P1) surfaceleadstotheblockingofthestrongNOxbindingsitesand

inhibitstheformationofstronglyboundnitrates.Inaddition,SOx

accumulationonthesurfacealsodestabilizesthenitratespecies whichareboundtointermediate-strengthNOxadsorptionsiteson

(14)

1000

900

800

700

600

500

400 QMS intensity (arb.u.)

8Ba/Ti/Al (P1) fresh (a)

NO NO2 O2 780 930 390

1x10-6

1000

900

800

700

600

500

400

8Ba/Ti/Al (P1) poisoned (b) 360 760 640 NO O2 NO2

1x10-6

1000

900

800

700

600

500

400

x4 20Ba/Ti/Al (P1) fresh (c)

Temperature (K)

NO O2 NO2 380 710 800 930

1x10-6

1000

900

800

700

600

500

400

370 770 20Ba/Ti/Al (P1) poisoned NO O2 NO2 (d) 490 640 690

1x10-6

Fig.10.TPDproﬁlesobtainedfromfreshandpoisoned(aandb)8Ba/Ti/Al(P1),(candd)20Ba/Ti/Al(P1)samplessaturatedwith8TorrNO2(g)at323Kfor20min.Black,blue

andredcurvescorrespondtom/z=30(NO),m/z=32(O2)andm/z=46(NO2)signals,respectively.Poisoningwasperformedbyexposingthesamplesto0.6TorrSO2(g)+O2(g)

TheTPDproﬁleoffreshTi/Al(P2)(Fig.9c)demonstratesminor NOandNO2desorptionpeaksat370KandsomebroadNO,NO2and

O2desorptionfeaturesthatappearbetween645–745K.These

fea-turescanbeattributedtotheNOxdesorptionfromAl2O3domains [37]oftheTi/Al(P2)surface.Thissamplealsorevealsa NO des-orptionsignalat900–920Kwhichcanbeassociatedwithstrongly boundbidentateorbridgingnitratesadsorbedonthedisordered anddefective bindingsites of theporousTixAlyOz mixedoxide

surface.

Afterpoisoning (Fig. 9d) the Ti/Al(P2) sample, the presence of sulfates inhibits the adsorption of NOx species particularly

on the weak (T<550K) and strong (T>870K) NOx binding

sites, while inﬂuencing the intermediate-strength NOx binding

sites (550K<T<870K) to a lesser extent. Thus, formation of weaklybound(N2O3/NO+/NO2)aswellasstronglybound

biden-tate/bridgingnitratesaresigniﬁcantlysuppressedafterpoisoning. Alongtheselines,themainNOandNO2 desorptionfeaturesare

alsoshiftedtolowertemperaturesafterSOxpoisoningindicating

thedestabilizationoftheadsorbednitrates.

SimilarTPDexperimentswerealsoperformedonBa contain-ingBa/Ti/Al systemsand therepresentativedatacorresponding tosomeoftheseexperimentsarepresentedbelow.TheNOx

des-orptionfeaturesfromthefresh 8Ba/Ti/Al(P1)sampleupon NO2

adsorptionat323KarepresentedinFig.10a.Thedominant fea-tureat 930Kis associated withthe decomposition of thebulk Ba-nitrates onthelargeBaO clusterswhich are locatedonthe

(15)

agglomeratedTiO2 particles[23].Thelow-temperaturefeatures

areattributedtotheconvolutionofsignalsoriginatingfromthe decompositionofnitratesonlargeTiO2crystallites(780K)andthe

decompositionofthesurface(bidentate)nitratesonBa-domains (650–700K)[23].

Thepoisoningof8Ba/Ti/Al(P1)samplewithSOxat323K,prior

to the NO2 adsorption increases the desorption signal of the

weaklyadsorbedspecies(Fig.10b)asobservedinthecaseofthe correspondingsupportmaterialTi/Al(P1).TheprominentNOx

des-orptionbandobservedforthepoisonedsampleislocatedat760K, witha shoulderat 640K.Thehightemperaturedesorption fea-tureat930Kisnotdetectedforthepoisoned8Ba/Ti/Al(P1)sample, suggestingtheinteractionoftheSOxspecieswiththelargerBaO

clustersandsuppressionoftheformationofthermallystablebulk nitratespeciesbyblockingthesurfacesitesandlimitingthe diffu-sionofnitratesintothesubsurfaceoftheBaOclusters.Unlikethe freshsample,poisonedsampleyieldsmorepronouncedNOx

des-orptionfeaturesat∼640KassociatedwithsurfaceBaOandTiO2

sitesratherthanbulkBaO.Furthermore,it isworthmentioning thatthedesorption/decompositionofthestoredNOxiscompleted

atca.900Kforthepoisoned8Ba/Ti/Al(P1)samplewhichclearly emphasizesthedestabilizationeffectofthesulfatesonthestored nitratespecies.

TPDdatagiven in Fig.10c andd for the20Ba/Ti/Al(P1)also demonstratethesuppressionofthehigh-temperaturedesorption signal located at 900–950K after sulfur poisoning. Concomi-tantly,medium-temperatureNOdesorptionsignalat600–800K which is associated with nitrates on well-dispersed titania domainsaswellasnitratesonaluminasitesis alsosuppressed to a certain extent. Suppression of the bulk Ba(NO3)2

forma-tion upon sulfur poisoning was also observed in similar TPD experiments(datanotshown)performedwith8(20)Ba/Ti/Al(P2) samples.

3.6. ThermalstabilityoftheadsorbedSOxspecies

DuringTPDexperimentsgiveninsection3.5,whichwere per-formedafterthesaturationofSOx-poisonedsurfaceswithNO2,

desorptionofSO2(m/z=64)wasalsomonitoredinaparallel

fash-iontotheNOxdesorptionchannels.ItisworthmentioningthatSO2

wastheonlysigniﬁcantSOxspeciesdesorbingfromthe

investi-gatedsurfacesinadditiontoaveryminoramountofH2S(m/z=34)

specieswhoseintensitywasatleast20-foldsmallerthanthatof SO2.TheSO2 andH2SdesorptionchannelsoftheseTPD

experi-mentsarepresentedinFig.11.

ThegeneralcharacteristicsoftheSOxdesorptionfromtheTiO2

(anatase)surface(Fig.11)revealrelativelyunstableSOxspecies

thatdesorbat452Kasabroadpeak(duetoSO32−and/orweakly

boundmolecularSO2).At760K,anintenseSO2desorptionsignal

isvisible.ThisdrasticSO2 evolutionataratherlowtemperature

(i.e.760K)couldbeassociatedwiththedecompositionofsulfates asaresultoftheanatasetorutilephase transitionandthe col-lapseoftheporousanatasestructuretoformthelow-surfacearea rutilepolymorph.For␥-Al2O3,twomajordesorptionsignalsare

observedintheSO2desorptiontrace.Theﬁrstdesorption

maxi-mumappearsat510KandisprobablyassociatedwithSO32−and/or

weaklyboundmolecularSO2.Thesecondandthemostprominent

desorptionsignalstartstoappearatT>900Kandrevealsalarge desorptionsignalwhosedesorptionmaximumisbeyondthe ulti-matetemperaturelimit(i.e.ca.1020K)thatcanbereachedinthe currentTPDexperimentalsetup.ThishightemperatureSO2

des-orptionsignalisattributedtothestronglyboundsurfaceand/or bulksulfatesonalumina[80].

TheSO2 desorptionproﬁleofthepoisonedTi/Al(P1)(Fig.11)

surfaceappearsastwoconvolutedbutdistinguishablelow temper-aturefeaturesat480and542Kaswellasahightemperaturefeature

whichbarelystartstoappearat900K(whosedesorptionmaximum islikelylocatedatT1020K).Comparisonofthelow-temperature bands arising from theTi/Al(P1) surface withthe ␥-Al2O3 and

TiO2referencematerials,indicatesthatthe480Kfeatureismost

likely associatedwiththedesorptionfromTiO2 domains,while

the542Kfeature isassociated withthedesorption fromAl2O3

domains.Inthelowtemperatureregion,SO2desorptionprobably

takesplacefromboth␥-Al2O3 andTiO2 domainswhilethehigh

temperaturedesorptionfeatureappearingatT>1000Kcouldbe relatedtothesulfatedesorptionfromthe␥-Al2O3domainsorfrom

theTiO2 Al2O3hetero-junction(interface)sites.Themost

impor-tantaspectoftheSO2desorptionsignalfromtheTi/Al(P1)sample

is thatthestrongly boundsulfatespecies desorbata tempera-turehigherthanthatofthe␥-Al2O3surface.TheSO2desorption

proﬁleoftheTi/Al(P2)sample(Fig.11d)demonstratesabroad des-orptionsignalwithin400–800Kwithamaximumat590K.This broadpeakcanbeenvisionedasaconvolutionofallofthe low-temperatureSO2 desorptionfeaturesthatareobservedforTiO2,

Al2O3andTi/Al(P1)whichisinlinewiththepoorlydeﬁnedandthe

amorphoussurfacestructureoftheTi/Al(P2)samples.Itis impor-tanttomentionthattheSO2desorptionfeaturesforTi/Al(P2)in

theintermediatetemperatureregion(400–800K)appearhigher than that ofTi/Al(P1). Thisobservationindicatesthat the ther-malstabilityofSOxspeciesontheTi/Al(P2)surfaceishigherthan

thatofTi/Al(P1).ApparentlyTi/Al(P2)samplealsopresentsa high-temperatureSO2 signalwhich startstoevolveat T>970Kwith

adesorptionmaximumlocatedatT1020K.Initialtake-offfor this high-temperaturedesorption featurealsostartsat ahigher temperaturethanthecorrespondingfeatureforTi/Al(P1) consis-tent with the higher stability of SOx species on Ti/Al(P2). The

SOxdesorptionproﬁlesofalloftheBa-containingsampleswere

rathersimilartoeachother(Fig.11),presentingalow-temperature SO2desorptionsignalwithabroadlineshapewithin400–700K.

Thisisfollowedbyahigh-temperatureSO2 desorptionsignalat

T>1000K,whichwasabovethetemperaturewindowaccessible withthecurrentlyusedTPDexperimentalsetup.Thus,although thecurrentTPDexperimentscannotdiscriminatetherelative ther-malstabilitiesofthestronglyboundSOxspecies,FTIRtechnique

canprovide valuableinsights regardingthis point, asdescribed below.

Fig.12showsaninterestingsetofFTIRdatademonstratingthe thermalregeneration capabilitiesoftheTiO2-promotedsamples

aftersulfurpoisoning.TheseseriesofFTIRspectrawereobtained onsampleswhichareinitiallypoisonedwithSO2+O2at323Kand

thensuccessivelysaturatedwithNO2at323Kandﬁnally

evacu-atedinvacuumat1023K.Afterthehigh-temperatureevacuation, all ofthe adsorbedNOx species werefoundto desorbfrom all

oftheinvestigatedsurfaces.Howeversomeofthethermally sta-bleadsorbedSOxspeciesstillexistonthesamplesurfaces,even

afterthethermal regenerationtreatmentat1023K.Considering thedatacorrespondingtothe8Ba/Al and 20Ba/Alconventional benchmarksamplesinFig.12,itisapparentthatthese conven-tional NOx storage materials still possess a signiﬁcant amount

of stable surface aluminum sulfates(1360 and 1125cm−1)and bulkBaSO4(1258and1160cm−1)evenafterthermalregeneration

treatmentat1023K.Inotherwords,theeffectofsulfur poison-ingisrelativelysevereandirreversibleonconventionalTiO2-free

materials, leading to extremely stable SOx species that cannot

bereadilyremoved evenafter evacuationatelevated tempera-tures.

On theother hand,TiO2-promotedsamples given in Fig.12

show a much better thermal regeneration behavior, revealing signiﬁcantlysmallerintensitiescorrespondingtotheSOx

vibra-tional features after evacuation at 1023K. This is particularly validfor8Ba/Ti/Al(P1,P2)and20Ba/Ti/Al(P1)samples.Combining theseresultswiththerelativetotalSOxuptakedatapresentedin

(16)

1000 900 800 700 600 500 400 γ_{-Al2O3 poisoned}

QMS intensity (arb.u.)

(a) x20 SO2 H2S 510 690

1x10-7

1000 900 800 700 600 500 400

TiO2 (anatase) poisoned (b)

x20 H2S SO2 760 775 450

1x10-7

1000 900 800 700 600 500 400 Ti/Al (P1) poisoned (c)

Temperature (K)

x20 480 540 720 SO2 H2S

1x10-7

1000 900 800 700 600 500 400 Ti/Al (P2) poisoned (d) x20 580 850 SO2 H2S 760

1x10-7

Fig.11.SO2desorptionproﬁlesinTPDexperimentsfor␥-Al2O3,TiO2,Ti/Al(P1andP2),sampleswhichwereinitiallypoisonedwith0.6TorrSO2(g)+O2(g)(SO2:O2=1:10)at

473Kfor30minandthenevacuatedandsuccessivelytreatedwith8TorrNO2(g)at323Kfor20min(seetextfordetails).

Fig.5,it isclearthat 8Ba/Ti/Al(P1,P2)and 20Ba/Ti/Al(P1) sam-plesnotonlyaccumulate asmalleramountof SOxspecies than

theconventional8(20)Ba/Almaterialsuponsulfurpoisoning,but theyalso signiﬁcantlydestabilizetheadsorbedSOx species and

facilitatetheireffectivethermalregeneration/removalviasimple annealingat 1023Kwithout anadditional reducing agentsuch asCO,H2 orhydrocarbons.Itis alsoimportant tomentionthat

thethermalregenerationtemperaturethatisusedinthecurrent work (i.e. 1023K) is close to temperatures used in such ther-malSOxremovaltreatmentsinrealisticexhaustemissionsystems

(e.g. 750◦C). It is worth mentioning that the current ﬁndings arein verygood agreementwiththerecent transientresponse measurement(TRM)studies[81]performedonnano-structured Pt/BaO/TiO2/Al2O3 NSR catalystsreporting thatTiO2 promotion

facilitates both the SOx removal and the catalyst regeneration

underoperationalconditionswhileenhancingthetotalNOx

stor-agecapacity. Ascanbeseen inFig.5,20Ba/Ti/Al(P2)sample is theonly sample which wasfound tostore a larger amountof SOxincomparisontoitsTiO2-freecounterpart(i.e.20Ba/Al).

How-everthe thermal regeneration ofthis sample (Fig.12b) reveals a slightly better SOx removal behavior than the 20Ba/Al case,

indicatingthepositiveinﬂuenceofTiO2-promotiononthe

destabi-lizationoftheSOxspecies.Itisworthmentioningthatwehave

alsoattemptedtoquantitativelyanalyzetheremaining amount ofsulfurspeciesafterthethermalregenerationstepviaXPSand EDX.HoweverthesulfursignalsinXPSandEDXafterthe regen-eration step were below the analytical detection limits for all samples.

(17)

1000 1100 1200 1300 1400 1500 20BaAl 20BaTiAl P1

absorbance( a.u.)

1362 1308 1241 1164 1127 1037 0.1 1184 1117 1084 1044 1248 1125 1088 1042 20BaTiAl P2 (b) 1000 1100 1200 1300 1400 1500 1216 8BaTiAlP2 8BaTiAlP1 x2 8BaAl

wavenumber (cm

-1

₎

_{wavenumber (cm}

-1

₎

absorbance( a.u.)

1360 1258 1160 1125 1045 x2 0.1 1307 1380 (a)

Fig.12.FTIRspectrademonstratingthethermalregenerationcapabilitiesof(a)8Ba/Al,8Ba/Ti/Al(P1),8Ba/Ti/Al(P2)and(b)20Ba/Al,20Ba/Ti/Al(P1),20Ba/Ti/Al(P2)samples aftersulfurpoisoning.Spectrawereobtainedat323Kinvacuumafterinitialsulfurpoisoningwith0.6TorrSO2(g)+O2(g)(SO2:O2=1:10)at323K,followedbyheatinginthis

gaseousmixtureat473Kfor30minandasubsequentevacuationat323K;followedby8TorrNO2(g)adsorptionat323Kfor20minandaﬁnalevacuationstepinvacuum

at1023K.

4. Conclusions

Inthecurrentwork,wehaveinvestigatedthestructural prop-ertiesandchemicalbehaviorofTiO2-promotedbinaryandternary

mixedoxidesystemsthatareparticularlyrelevanttoNSRcatalytic applications.Asystematic approach wasfollowed in whichthe structuralandchemicalbehavioroftheTi/Al(P1,P2)binaryoxide supportmaterialswereinitiallyanalyzedinacomparative fash-ionwithsimpleconventionalsupportmaterialssuchas_␥-Al2O3

andTiO2(anatase).Inthelightoftheseinitialstudies,more

com-plex8(20)Ba/Ti/Al(P1,P2)ternaryoxideNOxstoragematerialswere

investigated.Someoftheimportantﬁndingsofthecurrentstudies canbesummarizedasfollows:

(a)SurfacestructureandthemorphologyoftheTiO2 Al2O3mixed

oxide systems strongly depend on the synthesis protocols. Ti/Al(P1)systemiscomprisedofrelativelyorderedand inho-mogeneouslydistributedTiO2(anatase)crystalliteson␥-Al2O3

domainswhileTi/Al(P2)systemcontainsanamorphousanda highlyporousTixAlyOzstructure.

(b) SOxadsorptiononalloftheinvestigatedsamplesleadstoa

decreaseintheNOxstoragecapacityduetotheblockingofthe

NOxadsorptionsitesbysulﬁteandsulfatespecies.

(c) Ti/Al(P1, P2) systems present a higher SOx storage

capac-ity(per unit sample weight) and a higherthermal stability fortheadsorbedSOx speciesthan thatof both␥-Al2O3 and

TiO2 (anatase)surfaces,whileTi/Al(P2)systemdemonstrates

higher afﬁnity towardsSOx than Ti/Al(P1) system. Thiscan

beexplainedbytherelativelyhigherSSAoftheTi/Al(P1,P2) systems(betterdispersionofTiO2 domainson␥-Al2O3 with

respecttobulkanatase)aswellasthedecreaseinthetotal num-berofLewisacidsitesandthepresenceofnewadsorptionsites duetoTiO2promotion.

(d)Typically, TiO2-promoted NOx storage materials (e.g.

8(20)Ba/Ti/Al(P1)and8Ba/Ti/Al(P2))accumulatelessSOx(per

unit sampleweight) than TiO2-freeconventional 8(20)Ba/Al

samples under identical poisoning conditions. Furthermore the(P1)samplesaccumulatelessSOxthan(P2)sampleswhich

canbeassociatedwithbothlowerSSAofthe(P1)samplesand theirdifferentsurfacemorphology.

(e)TiO2-promoted 8(20)Ba/Ti/Al(P1, P2) NOx storage materials

werefoundtoexhibit superiorthermalregeneration behav-ior after sulfur poisoning in comparison to conventional 8(20)Ba/Almaterials.AdsorbedSOxspecieswereobservedto

besigniﬁcantlydestabilizeduponTiO2promotionandthus

cor-respondingsulfateandsulﬁtespeciescanreadilyberemoved fromthe surfaceby simple annealing in vacuumat 1023K, in theabsence of an additional reducing agent suchas H2,

COorhydrocarbons.Thisbehaviorcanbeattributedto well-dispersedBaOunitsonTiO2domainsandthesuppressionof

theformationofbulkBaSO4clusters.Theseresultssuggestthat

TiO2-promotionmightbeapromisingstrategyforenhancing

thesulfurtoleranceofNSRsystemswhichresultsinnotonlythe suppressionoftheSOxaccumulationonthecatalystsurface,but

alsoimprovementofthethermalregenerationcapabilityofthe sulfur-poisonedcatalystandpreventionofirreversiblesulfur uptakeunderoperationalconditions.

Acknowledgement

Authorsgratefullyacknowledgethefinancialsupportfromthe Scientific and Technical Research Council of Turkey (TUBITAK) (ProjectCode: 108M379). E.O.also acknowledgessupport from TurkishAcademyofSciences(TUBA)forthe“OutstandingYoung Investigator” grant.E.V.,V.Z.and V.B.acknowledgeRFBR (Rus-sia)#09-03-91225-CTаand#10-03-00596-а,forfinancialsupport. AuthorsalsothankMargaritaKantchevaforfruitfuldiscussionsand ErmanBenguforhishelpwiththeXPSexperimentsandTEMdata analysis.