Effect of gadolinium on the catalytic properties of iron oxides for WGSR

(1)

ContentslistsavailableatSciVerseScienceDirect

Catalysis

Today

j o ur na l h o me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / c a t t o d

Effect

of

gadolinium

on

the

catalytic

properties

of

iron

oxides

for

WGSR

Caio

Luis

Santos

Silva

a

_,

_Sérgio

_Gustavo

_Marchetti

b

_,

_Arnaldo

_da

_Costa

_Faro

_Júnior

c

_,

Tatiana

de

Freitas

Silva

d

_,

_José

_Mansur

_Assaf

d

_,

_Maria

_do

_Carmo

_Rangel

a,∗

a_GECCAT_Grupo_de_Estudos_em_Cinética_e_Catálise,_Instituto_de_Química,_Universidade_Federal_da_Bahia,_Campus_{Universitário}_de_Ondina,_Federac¸_ão,₄₀

170-280Salvador,Bahia,Brazil

b_CINDECA,_Facultad_de_Ciencias_Exactas,_Universidad_Nacional_de_La_Plata,_calle₄₇_y_115,₁₉₀₀_La_Plata,_Argentina c_Universidade_Federal_do_Rio_de_Janeiro,_Ilha_do_Fundão,_21941-909_Rio_de_Janeiro,_Rio_de_Janeiro,_Brazil d_Universidade_Federal_de_São_Carlos,_Rod._Washington_Luiz,_Km_235,_13565-905_São_Carlos,_São_Paulo,_Brazil

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received19December2012

Receivedinrevisedform18February2013 Accepted20February2013

Available online 11 May 2013

Keywords: Hydrogen WGSR Gadoliniumferrite Hematite Magnetite Ironcarbide

a

b

s

t

r

a

c

t

Duetotheneedforenergysupplythroughcleanerandmoreefficienttechnologies,theinterestfor thewatergasshiftreaction(WGSR)hasincreasedespeciallyduetoitsroleinpurifyinghydrogen-rich streams.Inordertofindalternativecatalystsforthisreaction,theeffectofgadoliniumanditsamounton thepropertiesofironoxide-basedcatalystswasstudiedinthiswork.Sampleswithdifferentgadolinium toironmolarratio(0.05;0.1and0.15)werepreparedbysol–gelmethodandcharacterizedbychemical analysis,thermogravimetry,differentialscanningcalorimetry,infraredspectroscopy,X-raydiffraction, Mössbauerspectroscopy,specificsurfaceareameasurementsandthermoprogrammedreduction.The catalystswereevaluatedinWGSRat1atmintherangeof250–400◦_C._Hematite_and_gadolinium_ferrite

weredetectedforallfreshcatalystsbasedonironandgadoliniumwhilemagnetiteandironcarbidesand gadoliniumoxidewerefoundforthespentones.Thespeciﬁcsurfaceareaincreasedduetogadolinium, relatedtoitsroleasspacer.GadoliniummadethereductionofFe3+_and_Fe2+_species_more_difﬁcult_for_all

catalystsandtheninhibitedtheproductionofironcarbidesduringreaction,increasingtheactivity.The catalystwithGd/Fe=0.10showedthehighestactivitythatwasassignedtoitshighestspeciﬁcsurface area,whichexposedmoreactivesites.Nomethaneorethanewasfoundindicatingthattheironcarbides wereinactivetoFischer–Tropschsynthesisunderreactionconditions.

1. Introduction

Nowadays,hydrogenisprimarilyproducedfromhydrocarbons

oralcoholsthroughreformingprocesses. Regardlessthekindof

reforming(steam reforming, partialoxidation, dryreforming or

autothermal reforming), the reactor efﬂuent typically contains

mixturesofhydrogen,carbonmonoxide,carbondioxideandsteam

besidesunreactedhydrocarbon[1].Theremovalofcarbon

monox-idefromthishydrogen-richstreamisessentialformanyprocesses

sinceitcanirreversiblypoisonmostofmetalliccatalystsusedin

chemicalandpetrochemicalprocesses[2,3].Carbonmonoxidecan

alsopoisonplatinum electrocatalystsinprotonexchange

mem-branefuelcell(PEMFC),oneofthemostpromisingtechnologies

fortheproductionofcleanenergy[4].Anefﬁcientandwidelyused

waytopurifyhydrogen-richstreamsisprovidedbywatergasshift

reaction(WGSR)[2].Thisreactionhasaconsiderabletechnological

importancebecauseofitshighefﬁciencyinreducingtheamount

∗ Correspondingauthor.Tel.:+557133790922;fax:+557132374117. E-mailaddresses:mcarmov@ufba.br,mcarmov@gmail.com(M.d.C.Rangel).

ofcarbonmonoxidegeneratedinthereformingofnaturalgasor

hydrocarbons.Inaddition,itallowstheadjustmentofsynthesisgas

composition,amixtureofcarbonmonoxideandhydrogenlargely

usedinmanyindustrialprocessessuchasmethanolsynthesisand

Fischer–Tropschreaction[2,5].

Thereactionisreversibleandexothermic,favoredbylow

tem-peraturesandlargeamountsofsteam.However,inordertoobtain

highconversionsforeconomicalpurposes,theWGSRisusually

per-formedintwostagesinindustrialprocesses.Theﬁrstoneoccursin

therangeof350–420◦Cunderfavorablekineticconditionsandis

knownasHTS(hightemperatureshift)whiletheotherstep,called

LTS(lowtemperatureshift),isoftenperformedattemperatures

around200–250◦C.Forcommercialuse,iron-based(e.g.,Fe/Cr/Cu)

andcopper-based(e.g.,Cu/Zn/Al)catalystshaveusuallybeenused

forHTSandLTSreactions,respectively[2,5,6].TheHTScatalysts

arebasedonironoxidepromotedwithchromiumoxideandshow

severaladvantagessuchaslowcost,thermalstabilityand

resis-tanceagainstpoisons[2,5,7].Theyarecommercializedashematite

(␣-Fe2O3)andarereducedinsitutoproducemagnetite(Fe3O4)

whichistheactivephase.Thelatestgenerationofsuchcatalysts

alsocontainscopper,whichactsasanelectronicpromoterleading

(2)

toanincreaseinthecatalyticactivity[8,9].Severalstudies[2,5,10]

haveshowedthatchromiumoxide(Cr2O3)delaysmagnetite

sinter-ing,avoidingthelossofspeciﬁcsurfaceareaofthecatalystduring

operation.However,duetothecarcinogenicnatureofchromium,

thereis a needtosearch foralternative dopantsfor such

cata-lysts.Therefore,severalstudieshavebeencarriedoutaimingto

replacechromiumbyotherdopants[3,8,11–15]aswellastoﬁnd

alternativecatalystsforWGSR[16–21].

Withthisgoalinmind,gadolinium-containingironoxideswere

studiedinthiswork,inordertoﬁndalternativecatalystsforWGSR.

Thebeneﬁceffectofgadoliniumregardingtheactivityofiron

com-poundswaspreviouslydemonstratedbyTsagaroyannisetal.[18].

Theyhavefoundthatgadoliniumferriteswerepromisingcatalysts

forWGSR,beingmoreactivethansomeindustrialones,their

activ-itiesdependingonthepreparationmethod.Inthepresentwork,

wehaveinvestigatedtheeffectofdifferentamountsofgadolinium

concerningironoxideactivity(preparedbyadifferentmethod)in

WGSRusingastreamcompositionclosetotheindustrialone.

2. Experimental

2.1. Catalystpreparation

The starting materials used for the preparation of the

cat-alysts were as following: gadolinium (III) nitrate hexahydrate,

Gd(NO3)3·6H2O (Aldrich, 99.9%), iron (III) nitrate nonahydrate,

Fe(NO3)3·9H2O(VETEC,≥98%),andammoniumhydroxide,NH4OH

(Aldrich,ACSGrade).Thesecompoundswereusedwithoutfurther

puriﬁcation.

Samples were prepared by precipitation by mixing

simul-taneously iron nitrate (Fe(NO3)3·9H2O), gadolinium nitrate

(Gd(NO3)3·6H2O) and ammoniumhydroxide solutions at 25◦C.

SolidswithGd/Femolarratiosof0.05,0.10and0.15wereobtained

besidespureironoxideandgadoliniumoxide.Forpreparingthe

gadolinium-poorestsample(Gd/Fe=0.05),250mLofagadolinium

nitratesolution(0.05molL−1)were added simultaneouslywith

250mLofanironnitratesolution(1.00molL−1)and250mLofan

ammoniumhydroxidesolution(25%)toabeakercontainingwater,

understirring.Afterthecompletemixtureofthereactants,thepH

wasadjustedo11andthesolwaskeptunderstirringfor30min.It

wasthencentrifuged(2500rpm,5min),thegelwashedwithwater

toremoveammoniumandnitrateionsfromthesolidandthen

cen-trifugedagain,thesestepswererepeatedﬁvetimes.Thegelwas

driedinanovenat120◦C,for24h.Afterdrying,thesolid

(cata-lystprecursor)wasgroundandsievedin100mesh.Thesolidwas

thenheated(10◦Cmin−1)undernitrogenﬂow(100mLmin−1)at

800◦C,for2h.Forpreparingthesampleswithdifferentgadolinium

contents(Gd/Fe=0.10and0.15)thesameexperimentalprocedure

wasfollowed butvaryingthesolutionconcentrations(0.10 and

0.15molL−1ofgadoliniumnitrate,respectively).Thesame

exper-imentalprocedurewasfollowedtopreparepureironoxideand

gadoliniumoxidetobeusedasreferences.

2.2. Catalystcharacterization

Thecatalystswerecharacterizedbychemicalanalysis,

thermo-gravimetry(TG),differentialscanningcalorimetry(DSC),Fourier

transforminfrared spectroscopy (FTIR),X-raydiffraction (XRD),

speciﬁcsurfaceareameasurements,Mössbauerspectroscopyand

temperatureprogrammedreduction(TPR).

IronandgadoliniumcontentsweredeterminedbyX-ray

ﬂuo-rescence(XRF)inaBrukerS2PICOFOXspectrometerconsistingof

amolybdenumanodemetal-ceramicX-raytube(50kV,600_␮A),a

multilayermonochromator,a30mm2_silicon_drift_detector_and_an

automaticsamplechanger.Theexperimentsofthermogravimetry

(TG)anddifferentialscanningcalorimetry(DSC)wereperformed

onthecatalystprecursorsinaTAInstrumentsSDT2670.The

sam-ple(0.010g)washeatedat10◦Cmin−1intherangeof35–900◦C,

under nitrogen ﬂow (30mLmin−1). Fourier transform infrared

spectroscopy(FTIR)analyseswereperformedin aPerkinElmer

modelSpectrumOne,usingsamplesdilutedinpotassiumbromide

discswitha1:100ratio.Thespectrawererecordedintheregion

between4000and400cm−1,using32scansandaresolutionof

4cm−1.

X-raydiffractionmeasurementswereperformedatroom

tem-peratureinaShimadzuXRD-6000modelequipmentwithCuK␣

radiation(=1.54059 ˚A)generatedat40kVand30mAwithanickel

ﬁlter.Thediffractogramswereobtainedwithascanningvelocityof

2◦min−1intherangeof10–80◦(2).Thespeciﬁcsurfaceareaswere

measuredbynitrogenadsorptionat−196◦_C_using_a_{Micromeritics}

ASAP2010instrumentandcalculatedusingtheBETmodel.Before

analysis,thesample(0.5g)washeatedupto200◦Cundervacuum

inordertoremovevolatilesfromthesolids.

Mössbauerspectrawereobtainedinaspectrometerwith512

channelswithconstant accelerationand geometryof

transmis-sion.Asourceof57_Co_in_Rh_matrix_of_nominally₅₀_mCi_was_used.

Velocitycalibrationwasperformedagainsta12_␮m-thick_␣-Fefoil.

Allisomershifts(␦)mentionedinthispaperarereferredtothis

standard.Foreachcomponentofthespectra,Lorentzianslinesof

thesamewidthwereused.The spectrawereobtainedatroom

temperatureandfoldedtominimizegeometriceffects,being

eval-uatedusingacommercialcomputerﬁttingprogramnamedRecoil.

Thetemperature-programmedreductionanalyseswerecarriedin

aZetonAltamarAMI90 apparatusequippedwitha TCD

detec-tor.Beforeanalyses,thesamples(0.3g)wereheatedupto400◦C

underargonﬂow(30mLmin−1),inordertoremovewaterandthen

cooledto35◦C.Duringanalyses,thesampleswereheatedfrom35

to1000◦Cunderaﬂow(30mLmin−1)ofa10%H2/Armixture.

2.3. Catalystevaluation

ThecatalystswereevaluatedinWGSRatatmosphericpressure

intherangeof250–400◦Cinatubularﬁxedbedmicroreactorusing

stream(135mLmin−1)withsteamtoprocessgasmolarratio(3.7%

CO,3.7%CO2,22.2%H2,70.4%N2)of0.6(steam/carbon

monox-idemolarratio16:1)and150mgofcatalyst.Thegaseousefﬂuent

wasanalyzedbyonlinegaschromatography,usingaVarianModel

3800instrumentequippedwith2TCDdetectors.Otherrunswere

carriedoutat400◦C,for6hoverallcatalysts.Afterreaction,the

catalystswerecharacterizedbyX-raydiffraction,Mössbauer

spec-troscopyandspeciﬁcsurfaceareameasurementsinordertofollow

thechangeswhichmighthaveoccurredduringreaction.

3. Resultsanddiscussion

Table1showstheresultsofthechemicalanalysisforthe

cat-alysts.ForallsamplestheexperimentalGd/Femolarratioswere

thesameastheexpectedvalues,indicatingthattheexperimental

Table1

Chemicalcompositionofthecatalysts.G:gadoliniumoxide;F:ironoxide;FG05, FG10andFG15:gadoliniumandironoxideswithGd/Fe(molar)=0.05;0.10and 0.15,respectively. Sample wt%Fe (±1.0) wt%Gd (±0.8) Gd/Fe(molar) (expected) Gd/Fe(molar)(±0.01) (experimental) F 68.9 – – – FG05 67.4 9.1 0.05 0.05 FG10 65.4 17.9 0.10 0.10 FG15 63.8 25.4 0.15 0.14 G – 74.5 – –

(3)

100 200 300 400 500 600 700 800 900 F FG05 FG10 G Temperature (°C) Weight (a.u.) FG15 dm /d T (a.u .)

(a)

100 200 300 400 500 600 700 800 900 G FG15 FG05 F H eat flo w ( W .g -1 ) Temperature (°C) FG10

(b)

Fig.1. (a)Thermogravimetry(solidlines)andderivativethermogravimetry(dashed lines)curvesand(b)differentialscanningcalorimetrycurvesforthecatalyst pre-cursors.G:gadoliniumoxide;F:ironoxide;FG05,FG10andFG15:gadoliniumand ironoxideswithGd/Fe(molar)=0.05;0.10and0.15,respectively.

conditionsusedwereappropriatefortheprecipitationofironand gadoliniumcompounds.

Thethermogravimetryandderivativethermogravimetrycurves ofthecatalystprecursors(ironand/orgadoliniumhydroxide)are showninFig.1a.Thegadolinium-freesolid(Fsample)presented

thelowesttotalweightloss(12.0%)whiletheFG15sampleshowed

thehighesttotalweightloss(26.1%).ThesolidswithGd/Fe=0.05

and0.10showedtotalweightlossof15.7and15.8%,respectively.

Forpureironhydroxidefourstepsofweightlosswerenoted.The

ﬁrstone,intherangebetween42and139◦Cisassociatedwith

theloss ofvolatiles includingadsorbed water [22]. Thesecond

andthirdpeaks,intherangebetween185and309◦C,arerelated

topartialdehydrationofironhydroxide[22,23].Theweightloss

above300◦Ccanbeattributedtocompletedehydrationof iron

hydroxide to form hematite[22]. It is observed that the

pres-enceofgadoliniuminsolidsshiftedthesecond,thirdandfourth

peakstohigher temperatures,indicating that gadoliniummade

thedecompositionofironhydroxidemoredifﬁcult,increasedby

gadoliniumcontent.Puregadoliniumoxideshowedweightlosses

attributedtodecompositionof gadoliniumof carbonatespecies

occurredsimultaneouslywithdehydration,attemperaturesabove

400◦C[24].For allsamplesnoweightlosswasobservedabove

600◦C.

Thedifferentialscanningcalorimetrycurvesofthecatalyst

pre-cursors(Fig.1b)showedthatexceptforpuregadoliniumhydroxide

anendothermicpeak centered in therange of55–66◦C can be

noted,attributedtothelossofvolatilesincludingwater[25],in

agreementwiththeweightlossnotedbythermogravimetryinthis

temperaturerange.Forallcurves,anendothermicpeakcenteredin

therangeof287–291◦Cwasfound,whichisassociatedtopartial

dehydration[22].Forgadolinium-containingsamplesthe

exother-mic peak,characteristic ofhematite(centeredat around350◦C

forpureironhydroxide),wasshiftedtohighertemperaturesand

anotherexothermicpeakcentered at760◦C,attributedto

crys-tallizationandtheformationofgadoliniumferrite(FeGdO3),was

noted[26].Thesephasechangeswerenotrelatedtoanyweightloss

asnotedbyTGcurves.Theseresultsindicatethatgadolinium

hin-dershematiteformationandfavorstheproductionofgadolinium

ferrite(FeGdO3),instead.Theiron-freesolidshowedendothermic

peaksat220,440and530◦C,assignedtothedehydrationprocess

ofgadoliniumhydroxide(Gd(OH)3)toproducegadoliniumoxide

(Gd2O3)[26]aswellastothedecompositionofcarbonatespecies

[24].Thecurvesdidnotpresentthermaleventsattemperatures

above800◦C,indicatingthestabilityofthephasesformed.

FTIRspectrafortheprecursorsandforthecatalystsareshown

in Fig.2. Onecansee bandsat 3400and 1630cm−1 attributed

tostretchinganddeformationvibrationsofOHwatermolecules

bondsandironandgadolliniumhydroxides[27].Forallspectra,

bandsintherangeof1380–1600cm−1 canbenoted,whichare

assigned tocarbonategroups adsorbedonironand gadollinum

(oxyhydr)oxides[24,28,29].For alliron-basedsamplesthereare

absorptionbandsat800,570and460cm−1,whichcanberelated

tostretchingvibrationsofFe-Obond[30].FTIRspectrumof

iron-freesampleshowedstrongabsorptionsat600–400cm−1relatedto

Gd-Ovibrationmodes[31].

FromtheX-raydiffractogramsforthefreshcatalysts(Fig.3a),

itcanbenotedthatthegadolinium-freesampleshoweda

typi-calpatternofhematite,_␣-Fe2O3,(JCPDS87-1166)inagreement

withpreviousstudies[8,10].Forgadoliniumandiron-containing

samples,aperovskitephaseofgadoliniumferrite,FeGdO3,(JCPDS

78-0451)wasalsofoundbesideshematite.Thisﬁndingindicates

that thecalcination temperaturewas suitable totheformation

of this phase, in accordance with previous work [26]. For the

gadolinium-richestsamplethecubicphaseofgadoliniumoxide,

Gd2O3,(JCPDS76-0155)wasalsodetected.Fortheotherironand

gadolinium-containingsolids,this phasecouldnotbeidentiﬁed

becausetheir peaksare coincident withsomepeaks related to

hematite.Theiron-freesampleshowedpeakscorrespondingtothe

cubicphaseofgadoliniumoxideinagreementwithprevious

stud-ies[24],only.AfterWGSR,theiron-basedsamplesshoweddifferent

proﬁles,asshowninFig.3b.Magnetite,Fe3O4,(JCPDS19-0629)was

foundforalliron-basedsamples,indicatingthathematitechanged

tomagnetiteduringreaction,inagreementwithpreviousworks

[8,10,12].Forthegadolinium-freesample,ironcarbidessuchas

␹-Fe5C2(JCPDS20-0509)and␪-Fe5C2(JCPDS76-1877)werealso

detected.Fortheironandgadolinium-containingsolids,␹-Fe5C2

phase wasfoundwhilethepresenceof␪-Fe5C2 phase,

gadolin-iumoxideandgadoliniumferritecouldnotbeconﬁrmedbecause

theirpeakswerecoincidentamongthemorwiththoserelatedto

magnetite.Onlyforthegadolinium-richestsamplethepresenceof

gadoliniumoxidewasconﬁrmed.

The Mössbauerspectra for the freshcatalysts are shown in

Fig.4 and the hyperﬁne parameters are shown in Table2. For

pure iron oxide, it can be noted a sextuplet whose hyperﬁne

parametersaretypicalofhematite(␣-Fe2O3)[32,33].Fortheiron

and gadolinium-containing samples, thespectra alsoshowed a

sextuplet, but the ﬁtting was not suitable witha single

inter-action, asshown intheinternal lateralofthepeaks(especially

(4)

4000 3500 3000 2500 2000 1500 1000 500 (Fe-O) (O-C-O) (H-O-H) G FG15 FG10 FG05 Transmittance (a.u.) Wavenumber

( cm

-1₎ F (O-H)

(a)

4000 3500 3000 2500 2000 1500 1000 500 G FG15 FG10 FG05 Transmittance (a.u) Wavenumber ( cm-1₎ F (Fe-O)

(b)

Fig.2.Fouriertransforminfraredspectrafor(a)theprecursorsandfor(b)the cat-alysts.G:gadoliniumoxide;F:ironoxide;FG05,FG10andFG15:gadoliniumand ironoxideswithGd/Fe(molar)=0.05;0.10and0.15,respectively.

necessarytoincludeasecondsextupletintheﬁtting.Thesextuplet

withthehighesthyperﬁne magnetic ﬁeld hasparameters

typi-calofhematite,similartogadolinium-freesample.Noappreciable

decreaseforthemagnetichyperﬁneﬁeldascomparedto“bulk”

hematitewasnoted.Theslightdecreaseofthemagnetichyperﬁne

ﬁeldinrelationto“bulk”hematiteallowstheestimativeof

aver-agecrystalsizesfordifferentcatalystsusingthemagneticcollective

excitationmodel[34].Avalueofabout150nmwasobtainedforall

samples. 10 20 30 40 50 60 70 80 F FG05 FG10 FG15 G In ten s ity (a .u. )

(a)

10 20 30 40 50 60 70 80 o o o o F-S FG05-S FG10-S FG15-S o o o o o o o o o o Intensity (a.u.) 2θ o G-S o o o o

(b)

(degrees) 2θ (degrees)

Fig.3. X-raydiffractogramsforthefresh(a)andspent(b)catalysts.G:gadolinium oxide;F:ironoxide;FG05,FG10andFG15:gadoliniumandironoxideswithGd/Fe (molar)=0.05;0.10and0.15,respectively.()␣-Fe2O3;()FeGdO3;(O)Gd2O3;(*)

Fe3O4;(␪)␪-Fe3C;(␹)␹-Fe5C2.

Inordertoassignthesecondsextuplet withlowermagnetic

hyperﬁne ﬁeld, the formation of an iron and gadolinium

com-poundwasconsidered.Amongthepossiblecompoundsthatcould

beformedintheexperimentalconditions,GdFeO3andGd3Fe5O12

should be considered [35]. The GdFeO3 compound is a cubic

perovskite with all Fe3+ _ions _located _in _equivalent _octahedral

sites surrounded by oxygen anions then the Mössbauer

spec-trumproducesasinglesextuplet[36,37].Ontheotherhand,the

Gd3Fe5O12compoundisagadoliniumirongarnetwhosegeneral

Table2

Mössbauerparametersforfreshcatalystsat25◦_C._F:_iron_oxide;_FG05,_FG10_and_FG15:_gadolinium_and_iron_oxides_with_Gd/Fe_(molar)₌_0.05;_0.10_and_0.15,_{respectively.}

Species Parameters F FG05 FG10 FG15 ␣-Fe2O3 H(T) 51.79±0.01 51.52±0.01 51.56±0.01 51.55±0.01 ı(mm/s) 0.37±0.01 0.37±0.01 0.37±0.01 0.37±0.01 2ε(mm/s) −0.22±0.01 −0.22±0.01 −0.22±0.01 −0.22±0.01 % 100 94±1 92±1 88±1 GdFeO3 H(T) – 48.5±0.1 49.5±0.1 49.4±0.1 ı(mm/s) – 0.36(*) 0.40±0.01 0.37±0.01 2ε(mm/s) – 0(*) 0(*) 0(*) % – 6±1 8±1 12±1

H:hyperfinemagneticfieldinTesla;␦:isomershift(alltheisomershiftsarereferredto␣-Feat25◦_C);_2␧:_quadrupole_shift;_:_quadrupole_splitting._(*)_Parameters_held fixedinfitting.

(5)

Transmiss ion (a. u. ) F FG05 -12 -8 -4 0 4 8 12 FG10 FG15 Velocity (mm/s) Transm iss ion ( a .u .) F-S FG05-S FG10-S -12 -8 -4 0 4 8 12 FG15-S Velocity (mm/s)

(a)

(b)

Fig.4. Mössbauerspectraat25◦_C_for_the_(a)_fresh_and_(b)_spent_catalysts._G:_gadolinium_oxide;_F:_iron_oxide;_FG05,_FG10_and_FG15:_gadolinium_and_iron_oxides_with_Gd/Fe (molar)=0.05;0.10and0.15,respectively.Redline:␣-Fe2O3;blueline:GdFeO3;greenlines:␪-Fe3Cand␹-Fe5C2andpurplelines:Fe3O4.(Forinterpretationofthereferences tocolorinthisﬁgurelegend,thereaderisreferredtothewebversionofthearticle.)

formulaisX3B2A3O12,whereGd3+ionsarelocatedinXsites

sur-rounded byeightoxygenanions,whereas Fe3+_ions_are_located

inbothoctahedral(B)andtetrahedral(A)siteswithapopulation ratioof2:3,respectively.Forthisreason,theGd3Fe5O12compound

producedaMössbauerspectrumwithtwoclearlydistinguishable sextuplets,especiallybecausethesignalcorrespondingtoFe3+_ions

locatedat A sites have a magnetic hyperfine field significantly smallerthanthoseatBsites(approximately40vs49Trespectively)

[37].Thepreviousdescriptionallowsustoruleouttheexistence

oftheGd3Fe5O12garnetandconﬁrmthepresenceoftheGdFeO3

perovskite.ItisinterestingtonotethattheamountofGdFeO3phase

waslowerthanthatexpectedtakingtheGd/Feratiointoaccount.

ThetheoreticalweightpercentagesexpectedfortheFG05,FG10

and FG15 sampleswould be14.7, 26.6and36.6%, respectively.

However,theGdFeO3 percentagesdetectedbyMössbauer

spec-troscopywere6_±1,8_±1and12+1,respectively.Therefore,the

excessgadoliniumshouldsegregate,probablyasgadoliniumoxide

(Gd2O3).ThisphasewasdetectedbyXRDonlyfortheFG15sample,

probablyduetothelowamountofthisphaseintheothersolids.

Therefore,itcanbeconcludedthatthecatalystsconsistofthree

segregatedphases:␣-Fe2O3,GdFeO3andGd2O3.

AfterWGSR,pureironoxide(F-Ssample)showedavery

com-plexMössbauerspectrumwithaboutnine peaksin thecentral

region(betweenvelocitiesofabout−3.9and4.0mm/s),twopeaks

in the negative velocities and two other peaks in the positive

velocitiesforironoxidestypicalvalues(Fig.4bandTable3).This

spectrumwasﬁttedusingsixsextuplets.Twoofthem,withthe

largest magnetic hyperﬁne ﬁelds, canbe assigned toFe3+ _ions

locatedin Asitesand Fe2.5+ _ions_located_in _B_magnetite _sites

(Fe3O4)[32,33].Theotherfoursextets,withlowvalues of

mag-netichyperﬁneﬁeldsand isomershifts,canbeassignedtoiron

carbides,suchascementite(␪-Fe3C)andtothreesites(I,IIand

III)ofHäggcarbide(␹-Fe5C2)[38].TheMössbauerspectraofspent

gadolinium-containingcatalysts(Fig.4bandTable3)changed

dra-matically ascompared topureiron oxide. Allof them showed

predominantsignalstypicalofironionslocatedinAandBsites

ofmagnetite.Inthecentralregionofthespectraonlysmallpeaks

relatedtoironcarbidesweredetected.Thespectrashowed

dif-ferencesdependingongadoliniumloadinginthesolids.For the

gadolinium-poorestsample(FG05-S),onlyanadditionalsextuplet,

assignableto␪-Fe3C,wasnecessarytocompletetheﬁttingprocess,

probablythethreesitesof␹-Fe5C2arepresentinverylow

concen-trations.Ontheotherhand,forFG10-Ssamplebothironcarbides

(␪-Fe3Cand␹-Fe5C2)withallsitesweredetected.Finally,forthe

gadolinium-richestsample(FG15-S)␪-Fe3CandtheIsiteof␹-Fe5C2

wereidentiﬁed. Table4displaystheamountofiron-containing

phaseformedineachcase.Fromtheseresults,itcanbeconcluded

thatthegadolinium-freecatalystisbasicallymadeofironcarbides

whiletheironandgadolinium-basedsolidsaremainlyformedby

magnetite.Therefore,gadoliniumpreventstheproductionofiron

carbides.BycomparingtheseresultswiththoseobtainedbyXRD,

onecanconcludethattheiron andgadolinium-containing

sam-plesaremadeofmagnetite,ironcarbidesandgadoliniumoxide,

sincegadoliniumferritehasnotbeendetectedbyMössbauer

spec-troscopy.

ThephasechangesnotedduringWGSRcanberelatedtoboth

reducing atmosphere and temperature which are able to

pro-ducemetalliciron(␣-Fe)fromhematite(␣-Fe2O3)duringreaction.

Fig.5 illustrated a schemeof thephase changes which is

sup-posedtooccurinreactionconditions.Hematitecanbetransformed

tomagnetite (Fe3O4)and metallic iron (␣-Fe) which is able to

dissociatecarbonmonoxide.Thecarbonatomscanthendiffuse

insidethe_␣-Felatticeproducingdifferenttypesofironcarbides.

For gadolinium-containingsolids,theGdFeO3 perovskiteis also

reducedbut gadolinium(asgadoliniumferriteorasgadolinium

(6)

Table3

Mössbauerparametersforspentcatalystsat25◦_C._F:_iron_oxide;_FG05,_FG10_and_FG15:_gadolinium_and_iron_oxides_with_Gd/Fe_(molar)₌_0.05;_0.10_and_0.15,_{respectively.}

Species Parameters F FG05 FG10 FG15 Fe3+_in_A_sites_of_Fe 3O4 H(T) 49.4±0.2 48.90±0.02 48.83±0.03 48.86±0.03 ı(mm/s) 0.28±0.03 0.28±0.01 0.27±0.01 0.27±0.01 2ε(mm/s) −0.03±0.05 −0.02±0.01 −0.04±0.01 −0.04±0.01 Fe2.5+ inBsitesofFe3O4 H(T) 46.0±0.2 45.84±0.03 45.86±0.03 45.88±0.04 ı(mm/s) 0.66±0.03 0.66±0.01 0.66±0.01 0.67±0.01 2ε(mm/s) 0.03±0.05 0.00±0.01 0.02±0.01 0.02±0.01 ␪ironcarbide H(T) 19.7±0.1 20.7±0.2 20.6±0.3 21.1±0.3 ı(mm/s) 0.25±0.01 0.18±0.02 0.20±0.02 0.22±0.03 2ε(mm/s) −0.04 ±0.02 0.06±0.04 0.09±0.03 0.11±0.05

␹ironcarbidesiteI H(T) 18.24±0.05 n.d. 17.9±0.7 18.4(*)

ı(mm/s) 0.18±0.01 n.d. 0.18±0.06 0.22(*)

2ε(mm/s) 0.05±0.01 n.d. −0.1±0.1 0(*)

␹ironcarbidesiteII H(T) 21.55±0.03 n.d. 21.7(*) n.d.

ı(mm/s) 0.24±0.01 n.d. 0.25(*) n.d.

2ε(mm/s) 0.11±0.01 n.d. −0.01(*) n.d.

␹ironcarbidesiteIII H(T) 11.34±0.05 n.d. 11.8(*) n.d.

ı(mm/s) 0.19±0.01 n.d. 0.22(*) n.d.

2ε(mm/s) 0.02±0.01 n.d. −0.09(*) n.d.

H:hyperfinemagneticfieldinTesla;␦:isomershift(referredto␣-Feat25◦_C);_{2␧:quadrupole}_shift;_(*)_Parameters_held_fixed_in_fitting;_n.d.:_not_detected._␪_iron_carbide: Fe3C(cementite);␹ironcarbide:Fe5C2(Häggcarbide).

Table4

AmountsofironphasesforspentcatalystsobtainedfromMössbauerspectraat 25◦_C._F:_iron_oxide;_FG05,_FG10_and_FG15:_gadolinium_and_iron_oxides_with_Gd/Fe (molar)=0.05;0.10and0.15,respectively.

Sample %Fe3O4 %Ironcarbides

F-S 8±1 92±3

FG05-S 93±1 7±1

FG10-S 83±3 17±6

FG15-S 92±1 8±1

onlyminorquantitiesofironcarbidesareproducedandthemain phaseismagnetite.TheproductionofironcarbidesduringWGSR doesnotagreewithpreviousworks[8,10].Thiscanbeassignedto

thehighcalcinationtemperature(800◦C)whichseemstofavorthe

productionofmetallicironintheconditionsofHTSreaction.

ThespeciﬁcsurfaceareasofthesolidsarepresentedinTable5.

Thevalues are typically low asa result of thehighcalcination

temperatureofthesolids.Asexpected[2,7,26]gadoliniumoxide

showedavaluehigherthanironoxide.Theadditionof

gadolin-iumtoiron-basedsolids generated a solid withhigher speciﬁc

surfaceareas ascompared to pureiron oxide,showingits role

astexturalpromoter.Thiseffectcanbeassignedtotheactionof

gadoliniumasaspacer,creatingabarrieramong theironoxide

Table5

Speciﬁcsurfaceareasofthecatalystsbefore(Sg)andafter(Sg*)WGSRperformedat 400◦_C_and_hydrogen_consumption_obtained_from_TPR_curves._G:_gadolinium_oxide;

F:ironoxide;FG05,FG10andFG15:gadoliniumandironoxideswithGd/Fe=0.05; 0.10and0.15.

Sample Sg(m2_g−1₎ _Sg*_(m2_g−1₎ _Hydrogen_consumption (␮molg−1₎ F 10 11 4129 FG05 11 11 2904 FG10 17 17 3831 FG15 12 9 1938 G 18 12 0

particles and preventing sintering, as noted previously for the ammoniasynthesis catalysts[39] andforaluminum-doped iron

oxidecatalystsforWGSR[8].DuetothelargerGd3+_ions₍₁₀₇_pm)_as

comparedtoFe3+_ions₍₆₉_pm),_they_are_not_expected_to_enter_into

hematiteormagnetitelatticebutrathersegregateasgadolinium

oxideorasgadoliniumferriteasconcludedbyMössbauer

spec-troscopyandXRD.Thesamplewiththeintermediateamountof

gadolinium(Gd/Fe=0.10)showedthehighestspeciﬁcsurfacearea

(17m2_g−1₎_while_the_{gadolinium-poorest}_sample_showed_the

low-estvalue.Duringreaction,thespeciﬁcsurfaceareas(Table5)did

(7)

100 200 300 400 500 600 700 800 900 Ti_Onset = 544 ºC Ti Onset = 517 ºC Ti_Onset = 516 ºC Ti_Onset = 515 ºC Tm_Onset = 379 ºC Tm_Onset = 371 ºC Tm Onset = 364 ºC FG15 FG10 F

Hydrogen consumption (a.u)

FG05 Tm_Onset = 403 ºC

Temperature (°C)

1000

Fig.6. TPRcurvesforthecatalysts.G:gadoliniumoxide;F:ironoxide;FG05,FG10 andFG15:gadoliniumandironoxideswithGd/Fe(molar)=0.05;0.10and0.15, respectively.

notchangesigniﬁcantly,exceptforthegadolinium-richestsample

(FG15),probablyduetothepresenceoflargeramountof

gadolin-iumoxide and/or gadolinium ferrite which show typically low

speciﬁcsurfaceareas.

Thecurvesoftemperatureprogrammedreductionforthe

cat-alystsareshowninFig.6andthehydrogenconsumptionduring

theTPRexperimentsareshowninTable5.Puregadoliniumoxide

(Gsample)showednoreductionpeakatthetemperaturerange

oftheexperiment. On theotherhand, pureiron oxide(F

sam-ple)showedareductionpeakbeginningat403◦Cattributedtothe

reductionofFe3+_to_Fe2+_species_and_the_peak_at_higher

temper-aturesbeginningat544◦CrelatedtothereductionofFe2+_to_Fe0

species[8,40].Itcanbenotedthatthegadoliniumadditionshifted

thepeakstolowertemperatures.Gadoliniumledtoadecreasein

hydrogenconsumption,thereducibilityofthesolidsincreasedin

theorderFG15<FG05<FG10<Fandthennoregulartrendcould

beobserved.Theseﬁndingscanberelatedtodifferentamountsof

hematite,gadoliniumferriteandgadoliniumoxidesaswellasto

theirdistributioninthedifferentsamples.Fromtheseresultsone

canconcludethatgadoliniummakesthereductionofFe3+_and_Fe2+

speciesdifﬁcult,probablyduetothediffusioneffectsofhydrogen

intheparticlesofdifferentsizes,aswellastodifferentamounts

ofthesephasesineachsample.Theseresultscanexplainwhythe

gadolinium-containingsamplesproducedlessironcarbidesfrom

metalliciron.

Fig. 7a shows the conversion of carbon monoxide over the

catalystsasafunctionofreactiontemperature.Asexpected,the

conversionincreasedwithtemperatureduetokineticfactors.At

hightemperatures(350–400◦C),thecatalyticactivityofironoxides

increasedduetogadoliniumandthiscanbeassignedtothe

abil-ityofthisdopantindelayingtheproductionofmetallicironand

thenironcarbides,asfoundbyTPRandMössbauerspectroscopy.

Therefore,moremagnetiteisproducedingadolinium-doped

sam-plesinsteadofironcarbides.Theactivityincreasedwiththedopant

amountuptothesamplewithGd/Fe=0.10(FG10),indicatingthat

thereisanoptimumamountofgadoliniumforitspromoteraction.

Asthissamplehasthesmallestamountofmagnetite,itssuperior

activitycanberelatedtoitshighestspeciﬁcsurfacearea,

allow-ingtheexpositionof moreactive sitesonthesurface.Previous

works[5,41]haveshownthattheHTSreactionoccursover

iron-based catalyststhrough the regenerative mechanism according

towhich thesurfacegoesonsuccessivecyclesofoxidation and

250 300 350 400 0 3 6 9 12 15 CO Con v er s ion (% )

(a)

0 30 60 90 120 150 180 210 240 270 300 330 360 0 3 6 9 12 15 18 21 CO Co nve rsi on (%)

(b)

Temperature (°C) Time (min)

Fig.7.Conversionofcarbonmonoxideoverthecatalystsasafunctionof(a)reaction temperatureandof(b)reactiontimeinWGSRperformedat400◦C.G(--)sample: gadoliniumoxide;(-–)Fsample:ironoxide;(-䊉-)FG05,(--)FG10and(--)FG15: gadoliniumandironoxideswithGd/Fe(molar)=0.05;0.10and0.15,respectively.

reduction bycarbonmonoxideand water,respectively, toform

hydrogenandcarbondioxide.Itiswell-knownthattheactivephase

ofhematite-basedcatalystsismagnetite(Fe3O4)[42,43],inwhich

latticethequantum jumpsbetweentheFe2+ _and _Fe3+_levels_in

theoctahedralsitesenablestheironoxidation andreductionby

waterandcarbonmonoxide,respectively,duringWGSR.Thelow

amountofmagnetiteexplainsthelowactivityoffree-gadolinium

sample[44].Ontheotherhand,theroleofgadoliniumferritein

catalyzingtheWGSRhasbeenpointoutearlierbyHaralambous

et al.[45,46],who assigned itsperformance tothe presenceof

pointdefectsinthelatticeassociatedtop-typesemiconductivity,

favoringthechemisorptionofcarbonmonoxidemoleculesontheir

surface[45,46].However,inthepresentworkgadoliniumferriteis

notsupposedtoplayarole,onceitisconvertedtomagnetiteand

ironcarbidesduringreaction.

Theresultsofconversionovercatalystsasafunctionofreaction

time,performedat400◦Cfor6h(Fig.7b),showedthatthe

cat-alystswerestableunderexperimentalconditions,indicatingthat

thephasechangesoccurredintheﬁrstminutesofreaction.Only

carbondioxideandhydrogenweredetectedandnomethaneor

ethanewasfoundforallcatalysts,indicatingthattheironcarbide

phaseswereinactivetoFischer–Tropschsynthesisinthereaction

(8)

4. Conclusions

The hydrolysis of iron nitrate and gadolinium nitrate with

ammoniahydroxide,followedbycalcinationat800◦C,producesa

mixtureofhematite,gadoliniumoxide(Gd2O3)andGdFeO3

perov-skite.Forthesesolids,gadoliniumleadstoanincreaseofspeciﬁc

surfaceareaandadecreaseinreducibility.Thesecatalystswere

activein water gasshiftreaction(WGSR) andselective onlyto

hydrogenandcarbondioxide.Duringreaction,theygoonphase

change,producingmagnetite andironcarbidesco-existingwith

gadoliniumoxide.Thepresenceofgadoliniuminhibitsthe

produc-tionofironcarbides,increasingtheactivityofthecatalysts.Iron

carbidesareinactivetoFischer–Tropschsynthesisandthendonot

affecttheselectivityofthecatalysts.Themostactivecatalystisthe

solidwithGd/Fe(molar)=0.1,afactthatwasrelatedtoitshighest

speciﬁcsurfaceareaallowingtheexpositionofmoreactivesiteson

thesurface.

Acknowledgments

CLSSthanksCNPqforhisscholarship.Theauthorsaregratefulto

Dr.S.T.deOlivaforhishelpintheX-rayﬂuorescenceanalyses.The

authorsacknowledgeCNPqandFINEPfortheﬁnancialsupport.

References

[1]D.Holladay,J.Hu,D.L.King,Y.Wang,CatalysisToday139(2009)244–260.

[2]D.S. Newsome, Catalysis Reviews-Science and Engineering 21 (1980) 275–318.

[3]J.L.R.Costa,S.G.Marchetti,M.C.Rangel,CatalysisToday77(2002)205–213.

[4]S.X.Wang,D.Gao,Z.Yuan,N.Liu,C.Zhang,InternationalJournalofHydrogen Energy33(2008)3710–3718.

[5]L.Lloyd,D.E.Ridler,M.V.Twigg,in:M.V.Twigg(Ed.),CatalysisHandbook,Wolfe ScientiﬁcBooks,London,1996,pp.283–339.

[6]E.M.Fuentes,A.C.F.Júnior,T.F.Silva,J.M.Assaf,M.C.Rangel,CatalysisToday 171(2011)290–296.

[7]M.C.Rangel,R.M.Sassaki,F.Galembeck,CatalysisLetters33(1995)237–254.

[8]G.C.Araujo,M.C.Rangel,CatalysisToday62(2000)201–207.

[9]E.B.Quadro,M.L.M.Dias,A.M.M.Amorim,M.C.Rangel,JournaloftheBrazilian ChemicalSociety10(1999)51–59.

[10]A.L.C.Pereira,G.J.P.Berrocal,S.G.Marchetti,A.Albornoz,A.O.deSouza,M.C. Rangel,JournalofMolecularCatalysisA:Chemical281(2008)66–72.

[11]I.LimaJr.,J.Millet,A.Mimoun,M.C.Rangel,AppliedCatalysisA-General283 (2005)91–98.

[12]A.L.C.Pereira,N.A.dosSantos,M.L.O.Ferreira,L.A.M.Albornoz,M.C.Rangel, StudiesinSurfaceScienceandCatalysis167(2007)225–230.

[13]J.Y.Lee,D.W.Lee,K.Y.Lee,Y.Wang,CatalysisToday146(2009)260–264.

[14]C.Martos,J.Dufour,A.Ruiz,InternationalJournalofHydrogenEnergy34(2009) 4475–4481.

[15]L.Zhang,J.M.M.Millet,U.S.Ozkana,JournalofMolecularCatalysisA:Chemical 309(2009)63–70.

[16]M.S.Santos,G.J.P.Berrocal,J.L.G.Fierro,M.C.Rangel,StudiesinSurfaceScience andCatalysis167(2007)493–498.

[17]P.S.Querino,J.R.C.Bispo,M.C.Rangel,CatalysisToday108(2005)920–925.

[18]J.Tsagaroyannis,K.J.Haralambous,Z.Loizos,G.Petroutsos,N.Spyrellis, Mate-rialsLetters28(1996)393–400.

[19]M.Laniecki,M.Ignacik,CatalysisToday116(2006)400–407.

[20]S.YuChoung,M.Ferrandon,T.Krause,CatalysisToday99(2005)257–262.

[21]C.Wheeler,A.Jhalani,E.J.Klein,S.Tummala,L.D.Schmidt,JournalofCatalysis 223(2004)191–199.

[22]A.M.Gadalla,T.W.Levingston,ThermochimicaActa145(1989)1–9.

[23]R.Furuichi,M.Hachiya,T.Ishi,ThermochimicaActa133(1988)101–106.

[24]I.Y.Park,D.Kim,J.Lee,S.H.Lee,K.J.Kim,MaterialsChemistryandPhysics106 (2007)149–157.

[25]R.G.Villacelos,L.Herman,J.Morales,J.Tirado,JournalofColloidandInterface Science101(1984)393–397.

[26]J.Tsagaroyannis,K.J.Haralambous,Z.Loizos,N.Spyrellis,MaterialsLetters14 (1992)214–221.

[27]R.A.Niquist,R.O.Kagel,InfraredSpectraofInorganicCompounds,Academic Press,Orlando,1971.

[28]J.R.Bargar,J.D.Kubicki,R.Reitmeyer,J.A.Davis,GeochimicaetCosmochimica Acta69(2005)1527–1542.

[29]G.A.M.Hussein,JournalofPhysicalChemistry98(1994)9657–9664.

[30]S.H.Yariv,E.Mendelovici,AppliedSpectroscopy33(1979)410–411.

[31]J.A.Goldsmith,S.D.Ross,SpectrochimicaActaPartA23(1967)1909.

[32]E.Murad,J.H.Johnston,in:G.J.Long(Ed.),MössbauerSpectroscopyAppliedto InorganicChemistry,vol.2,PlenumPublishingCorporation,NewYork,1987.

[33]R.E.Vandenberghe,E.DeGrave,C.Landuydt,L.H.Bowen,HyperﬁneInteractions 53(1990)175–196.

[34]S.Mørup,H.Topsøe,AppliedPhysics11(1976)63–66.

[35]F.Söderlind,L.Selegård,P.Nordblad,K.Uvdal,P.-O.Kall,JournalofSol-Gel ScienceandTechnology49(2009)253–259.

[36]M.Eibschütz,S.Shtrikman,D.Treves,PhysicalReview156(1965)562–577.

[37]S.C.Zanatta,L.F.Cotica,A.Paesano,S.N.Medeiros,J.B.M.Cunha,B.Hallouche, JournaloftheAmericanCeramicSociety88(2005)3316–3321.

[38]R.LCohen,ApplicationsofMossbauerSpectroscopy,vol.II,AcademicPress, NewYork,1980.

[39]A.Nielsen,H.Bohlbro,InvestigationonPromotedIronCatalystsforthe Syn-thesisofAmmonia,Ed.GjllerupForlag,Copenhagen,1968,pp.p.221.

[40]J.C.Gonzalez,M.C.Gonzalez,M.A.Laborde,N.Moreno,AppliedCatalysis20 (1986)3–8.

[41]J.K.Boreskov,T.M.Yureva,A.S.Sergeeva,KineticsandCatalysis11(1970) 1476–1478.

[42]C.H.Bartolomew,R.J.Farrauto(Eds.),FundamentalsofIndustrialCatalytic Pro-cesses,JohnWiley&Sons,Hoboken,NJ,2006.

[43]C.Lund,IndustrialandEngineeringChemistryResearch35(1996)2531–2534.

[44]C.Rhodes,G.J.Hutchings,A.M.Ward,CatalysisToday23(1995)43–48.

[45]K.J.Haralambous,Z.Loizos,N.Spyrellis,MaterialsLetters11(1991)133–138.