UniversidaddeOviedo

DepartamentodeCienciadelosMateriales eIngenieraMetalrgica

Sntesis de catalizadore es basados en xidos de metales de transicin para la oxidacin preferencial de mon nxido de carbono en corrientes d de hidrgeno

TESISDOCTORAL IreneLpezSurez p de Materiales Poros sos Funcionales Grupo InstitutoNacionaldelCarbn n 2010

UNIVERSIDADDEOVIEDO
DepartamentodeCienciadelosMateriales eIngenieraMetalrgica

SNTESISDECATALIZADORESBASADOSENXIDOSDE
METALESDETRANSICINPARALAOXIDACIN PREFERENCIALDEMONXIDODECARBONOENCORRIENTES DEHIDRGENO

TESISDOCTORAL

IreneLpezSurez 2010

INSTITUTO NACIONAL DEL CARBN (INCAR)

GREGORIO MARBN CALZN, INVESTIGADOR CIENTFICO DEL INSTITUTO NACIONAL DEL CARBN, DEL CONSEJO SUPERIOR DE INVESTIGACIONES CIENTFICAS, CERTIFICA: Que autorizo la inclusin en la Memoria de Tesis Doctoral presentada por la Ingeniera Qumica Irene Lpez Surez Sntesis de catalizadores basados en xidos de metales de transicin para la oxidacin preferencial de monxido de carbono en corrientes de hidrgeno de los siguientes artculos de los que soy coautor:
The synthesis of high surface area cerium oxide and cerium oxide/silica nanocomposites by the silica aquagelconfined coprecipitation technique. Irene Lpez, Teresa Valds Sols,GregorioMarbn.MicroporousandMesoporousMaterials,127(2010)198204. An attempt to rank copperbased catalysts used in the COPROX reaction. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. International Journal of Hydrogen Energy, 33 (2008)197205. Preferential oxidation of CO by CuOx/CeO2 nanocatalysts prepared by SACOP. Mechanisms of deactivation under the reactant stream. Gregorio Marbn, Irene Lpez, TeresaValdsSols.AppliedCatalysisA:General,361(2009)160169. Copper manganite as a catalyst for the PROX reaction. Deactivation studies. Teresa ValdsSols, Irene Lpez, Gregorio Marbn. International Journal of Hydrogen Energy, 35(2010)18791887. Highly active cobalt based catalysts prepared by SACOP for the preferential oxidation of CO in excess hydrogen. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. Artculo enviadoaChemCatChem. Highly active structured catalyst made up of mesoporous Co3O4 nanowires supported on a metal wire mesh for the preferential oxidation of CO. Gregorio Marbn, Irene Lpez, Teresa ValdsSols, Antonio B. Fuertes. International Journal of Hydrogen Energy, 33 (2008)66876695.

INSTITUTO NACIONAL DEL CARBN (INCAR)

TERESA VALDSSOLS IGLESIAS, CIENTFICA TITULAR DEL INSTITUTO NACIONAL DELCARBN,DELCONSEJOSUPERIORDEINVESTIGACIONESCIENTFICAS, CERTIFICA: Que autorizo la inclusin en la Memoria de Tesis Doctoral presentada por la Ingeniera Qumica Irene Lpez Surez Sntesis de catalizadores basados en xidos de metales de transicin para la oxidacin preferencial de monxido de carbono en corrientes de hidrgeno de los siguientes artculos de los que soy coautora:
The synthesis of high surface area cerium oxide and cerium oxide/silica nanocomposites by the silica aquagelconfined coprecipitation technique. Irene Lpez, Teresa Valds Sols,GregorioMarbn.MicroporousandMesoporousMaterials,127(2010)198204. An attempt to rank copperbased catalysts used in the COPROX reaction. Irene Lpez, TeresaValdsSols,GregorioMarbn.Int.J.HydrogenEnergy,33(2008)197205. Preferential oxidation of CO by CuOx/CeO2 nanocatalysts prepared by SACOP. Mechanisms of deactivation under the reactant stream. Gregorio Marbn, Irene Lpez, TeresaValdsSols.AppliedCatalysisA:General,361(2009)160169. Copper manganite as a catalyst for the PROX reaction. Deactivation studies. Teresa ValdsSols, Irene Lpez, Gregorio Marbn. International Journal of Hydrogen Energy, 35(2010)18791887. Highly active cobalt based catalysts prepared by SACOP for the preferential oxidation of CO in excess hydrogen. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. Artculo enviadoaChemCatChem. Highly active structured catalyst made up of mesoporous Co3O4 nanowires supported on a metal wire mesh for the preferential oxidation of CO. Gregorio Marbn, Irene Lpez, Teresa ValdsSols, Antonio B. Fuertes. International Journal of Hydrogen Energy, 33 (2008)66876695.

ndicegeneral
Agradecimientos ....................................................................................................... VII Resumen .........................................................................................................................IX Abstract ...........................................................................................................................XI Introduccin ...................................................................................................................1
1. Economadelhidrgeno.........................................................................................1 2. Produccinypurificacindehidrgenoabordodelautomvil ............................6 3. OxidacinpreferencialdeCO(PROX).....................................................................9 4. Mtodosdepreparacindenanocatalizadores...................................................11 4.1. Nanoestructurasnosoportadas:mtodoSACOP.......................................11 4.2. Nanoestructuras soportadas: mtodo de evaporacin inducida de amoniaco ................................................................................................... 13

Objetivosyplanteamientodelamemoria ......................................................17
1. Objetivos...............................................................................................................17 2. OrganizacindelaMemoria.................................................................................18 3. Asistenciaacongresos..........................................................................................19

Experimental ................................................................................................................21
1. Materiales.............................................................................................................21 1.1. Nanoestructurasnosoportadas .................................................................21 a)MtodoSACOP...........................................................................................21 b)Impregnacindeslicecomercial...............................................................23 1.2. Nanoestructurassoportadas ......................................................................23 2. Caracterizacindelosmateriales.........................................................................25 2.1. DifraccinderayosX ..................................................................................25 2.2. IsotermasdeadsorcindeN2 .................................................................... 25

II

ndicegeneral 2.3. Microscopa electrnica de barrido (SEM) y anlisis por energa dispersivaderayosX(EDX).........................................................................26 2.4. Microscopaelectrnicadetransmisin(TEM) ..........................................26 2.5. EspectroscopainfrarrojaportransformadadeFourier(FTIR) ..................26 2.6. EspectroscopafotoelectrnicaderayosX(XPS) .......................................26 2.7. Reduccinatemperaturaprogramada(TPR).............................................27 2.8. Termogravimetra(TG) ...............................................................................27 3. Anlisisdeactividadcataltica..............................................................................27 3.1. Reactores ....................................................................................................27 a)Reactordecuarzo ......................................................................................27 b)Reactordeacero........................................................................................28 c)Reactormulticanal .....................................................................................28 3.2. Procedimientoexperimentalytratamientodedatos ................................33

Artculos .........................................................................................................................37
ResumenArtculoI......................................................................................................37 Artculo I: The synthesis of high surface area cerium oxide and cerium oxide/silica nanocomposites by the silica aquagelconfined coprecipitation technique, Irene Lpez, Teresa ValdsSols, Gregorio Marbn. Microporous and MesoporousMaterials,127(2010)198204. ............................................................ 41 ResumenArtculoII.................................................................................................... 57 Artculo II. An attempt to rank copperbased catalysts used in the COPROX reaction. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. International JournalofHydrogenEnergy,33(2008)197205 ....................................................... 61 ResumenArtculoIII................................................................................................... 81 ArtculoIII.PreferentialoxidationofCObyCuOx/CeO2nanocatalystspreparedby SACOP. Mechanisms of deactivation under the reactant stream. Gregorio Marbn, Irene Lpez, Teresa ValdsSols. Applied Catalysis A: General, 361 (2009)160169........................................................................................................... 85 ArtculoIII.InformacinSuplementaria................................................................... 111

III
ResumenArtculoIV................................................................................................. 121 Artculo IV. Copper manganite as a catalyst for the PROX reaction. Deactivation studies.TeresaValdsSols,IreneLpez,GregorioMarbn.InternationalJournal ofHydrogenEnergy,35(2010)18791887 .............................................................. 123 ResumenArtculoV.................................................................................................. 143 Artculo V. Highly active cobalt oxide catalysts prepared by SACOP for the preferential oxidation of CO in excess hydrogen. Irene Lpez, Teresa Valds Sols,GregorioMarbn,enviadoaChemCatChem(cctc.201000351) ..................... 147 ResumenArtculoVI................................................................................................. 165 Artculo VI. Highly active structured catalyst made up of mesoporous Co3O4 nanowires supported on a metal wire mesh for the preferential oxidation of CO. Gregorio Marbn, Irene Lpez, Teresa ValdsSols, Antonio B. Fuertes. InternationalJournalofHydrogenEnergy,33(2008)66876695 ........................... 167

Conclusionesgenerales .........................................................................................185 Bibliografa ..................................................................................................................187

ndicedeFiguras

Figura1.Esquemadeunapiladecombustiblealimentadaporhidrgenoyoxgeno. ..... 2 Figura 2. Esquema de obtencin de electricidad a partir de biomasa en la futura economadelhidrgeno..................................................................................................... 7 Figura3.Produccindehidrgenoapartirdebioalcoholes.............................................. 8 Figura4.EsquemademtodoSACOP. ............................................................................. 22 Figura 5. (a) Aspecto de las mallas catalticas en dos momentos distintos de su obtencin; (b) microfotografa SEM de la malla recubierta por las nanoestructuras deCo3O4.Enelinsertosepuedeobservarlamallametlicasinrecubrir. ...................... 24 Figura6.Microrreactordeacerorellenodelamallacataltica........................................ 28 Figura 7. (a) Fotografa del sistema de reaccin multicanal; (b) detalle de la parte superior de los seis reactores; (c) captura de pantalla del programa de control del sistemamulticanal............................................................................................................ 29 Figura8.Esquemadeunreactordeacerodelsistemamulticanal.................................. 31 Figura 9. Despiece del tapn de doble muelle de un reactor de acero del sistema multicanal. ........................................................................................................................ 31 Figura10.Esquemadelsistemadereaccinmulticanal.................................................. 32 Figura11.ProcedimientoexperimentalparalasreaccionesPROX.................................. 33

ndicedeTablas
Tabla 1. Objetivos del DOE para el almacenamiento de hidrgeno en vehculos, incluida laactualizacinrealizadaen2009. ......................................................................................4

VII

Agradecimientos
Quiero expresar mi ms sincero agradecimiento a los doctores D. Gregorio Marbn Calzn y Da. Teresa ValdsSols Iglesias, bajo cuya direccin ha sido realizada esta Memoria, por su dedicacin y continuo asesoramiento durante el desarrollodelostrabajosrealizados. Al Ministerio de Ciencia e Innovacin por la concesin de una beca FPI para la realizacindeestatesisdoctoral. Al Consejo Superior de Investigaciones Cientficas por permitir la realizacin de este trabajo en el Instituto Nacional del Carbn, y especialmente al director del centro,D.CarlosGutirrezBlanco. A los doctores D. Narcs Homs y Da. Pilar Ramrez de la Piscina por permitirme realizarunaestanciabreveenlaUniversidaddeBarcelona. AldoctorD.AntonioBenitoFuertesporlaayudaproporcionada. A Da. Ana M Lpez Iglesias del Departamento de Ciencia de los Materiales e IngenieraMetalrgicaporsuamabilidadybuenhacer. A mis compaeras de despacho del INCAR: Aida, Elena, Marta, Patri, Patricia y Raquel. A Goyo por estar conmigo en el laboratorio y por las estupendas hojas Excel con macros. A Teresa por ayudarme en todo momento, incluso en la distancia. A los dosporquesiemprequehenecesitadovuestraayudahabisestadoah. AmiscompaerosdeldepartamentodeQumicaInorgnicadelaUniversidadde Barcelona, Raquel, Marcos, Luis, Valerio y Sonia, por hacerme ms agradable la estancialejosdecasa. A mis amigos, y en concreto a Azucena, Patri Parolo y Bea por las quedadas de losviernestarde.

VIII

Agradecimientos

En ltimo lugar a mis padres, mi hermano Salvi, mi sobrino Diego y especialmenteaRoberto;graciasatodosporapoyarmeestosaos.

IX

Resumen

En este trabajo se han estudiado distintos catalizadores activos para la reaccin de oxidacin preferencial de monxido de carbono en presencia de hidrgeno (preferential oxidation,PROX). Esteprocesoes enla actualidadel mssencillo yeficaz para disminuir la concentracin de CO residual de las corrientes de hidrgeno que alimentan las pilas de combustible de membrana polimrica, evitando as el envenenamiento del catalizador que se encuentra en el nodo de dicha pila. Al tratarse de una reaccin heterognea es conveniente que los catalizadores empleados presenten reas superficiales elevadas; de este modo se obtiene una mayor relacin superficie/volumen paralainteraccingasslidoy,porconsiguiente,mayoresactividadescatalticas. Se han preparado dos tipos de nanocatalizadores de elevada rea superficial para su posterior anlisis en la reaccin PROX: no soportados y soportados en una malla metlica.Paralaobtencindenanoestructurasnosoportadasseutilizcomotcnicade sntesis la coprecipitacin confinada en hidrogeles de slice (silica aquagel confined co precipitation, SACOP), desarrollada por el Grupo de Materiales Porosos Funcionales del INCAR. Este mtodo permite obtener nanoestructuras de elevada rea superficial debido al colapso de la slice sobre los hidrxidos metlicos durante la etapa de precipitacin. El mtodo SACOP se optimiz para la obtencin de nanoestructuras de CeO2, lo cual permiti profundizar en el conocimiento de los mecanismos especficos de la tcnica. Posteriormente el mtodo SACOP se emple para preparar, entre otros, xidos basados en cobre (CuOx/CeO2 y CuMn2O4) tras un riguroso anlisis comparativo de los catalizadores de cobre empleados en la reaccin PROX a lo largo de los aos. Estos catalizadores muestran una elevada actividad en la reaccin PROX, si bien estn sujetos a cierto grado de desactivacin, cuyos mecanismos se estudiaron en profundidad.Laformacindecarbonatossuperficialeseslacausaprincipaldelaprdida de actividad de los nanocatalizadores CuOx/CeO2, mientras que la desactivacin de la espinela de cobre y manganeso puede ser atribuida a la progresiva reduccin del cobre sobrelasuperficiedelcatalizadorporaccindelmonxidodecarbono. Mediante el mtodo SACOP se obtuvieron los catalizadores basados en cobalto (Co3O4/CoOOH) ms activos para la reaccin de oxidacin preferencial de CO. Esta tcnicapermitiobtenernanocatalizadoresconconcentracionesvariablesdeCoOOHde

Resumen

alta superficie en funcin del mtodo de precipitacin empleado. Este compuesto, presente en las muestras frescas, es el responsable de las excelentes prestaciones de estos catalizadores, puesto que en las condiciones reductoras de la reaccin PROX se transformagradualmenteenCo3O4deelevadaactividadcataltica. Por ltimo se sintetizaron nanoalambres de Co3O4 mesoporosos soportados en una malla metlica de acero inoxidable mediante el mtodo de evaporacin inducida de amoniaco. Estos catalizadores estructurados unen a sus conocidas ventajas sobre los lechos catalticos de partculas (mayor manejabilidad, baja cada de presin, excelentes coeficientes de transmisin de calor, etc.) el hecho de presentar actividades catalticas muy elevadas y, lo que es ms remarcable, una notable estabilidad en los ensayos de largaduracinentodoelintervalodetemperaturasanalizado.

XI

Abstract

The aim of this work is to study different active catalysts for the preferential oxidation (PROX) of CO in excess hydrogen. Nowadays this is the easiest and most efficient process to reduce residual concentrations of CO in the streams fed into polymeric fuel cells.ItiswellknownthathydrogenstreamsneedtobealmostentirelyCOfreeinorder to avoid platinum poisoning. As the reaction is heterogeneous, high surface area catalysts are preferred, since this involves higher surface/volume ratios for the gassolid interactionand,consequently,highercatalyticactivitiescanbeachieved. Two types of high surface area nanocatalysts were prepared and analyzed in the PROX reaction: unsupported nanostructures and metal wire supported nanostructures. The silica confined coprecipitation (SACOP) technique, developed by the Functional Porous Materials Group at INCAR, was used as the reference method to obtain unsupported nanostructures. This is a modified silicatemplate route for obtaining high surface area nanostructures due to the collapse of the silica surrounding the metallic hydroxides during the precipitation stage. The SACOP procedure was optimized for the synthesis of cerium oxide, which contributed to our knowledge of the mechanisms involved. The procedure was then employed to prepare copperbased catalysts (CuOx/CeO2 and CuMn2O4), after a rigorous comparison of the activities of similar catalysts found in the literature tested for the PROX reaction. The catalysts used in this study exhibit a high activity for the PROX reaction but gradually undergo a slow deactivation during the reaction. The deactivation of CuOx/CeO2 nanocatalysts is caused mainly by carbonates formation, whereas for CuMn2O4 it seems to be produced by the progressive reduction ofcopperviatheactionofcarbonmonoxide. The SACOP method has produced the most active cobaltbased catalyst for the PROX reaction ever designed. The resulting nanocatalysts had a variable content in high surface area CoOOH, depending on the type of precipitation used. This species which was present in all the fresh samples and was the main contributor to catalyst performance,wasgraduallyreducedtohighlyactiveCo3O4underPROXconditions. Furthermore, supported Co3O4 mesoporous nanostructures were obtained by means of amild templatefreeprocedureknownas theammoniaevaporationinduced method.A

XII

Abstract

stainless steel wire mesh was employed as the metallic support. Apart from the well known advantages of employing these monoliths over catalytic beds of particles (easy handling, low pressure drop and excellent heat transmission coefficients), these structured catalysts exhibited a good catalytic activity and a remarkable stability over theentiretemperaturerangestudied.

Introduccin

1.

ECONOMADELHIDRGENO

El modelo energtico actual basado en el empleo de combustibles fsiles (carbn, gas natural y petrleo) est en entredicho. Los problemas medioambientales asociados al uso de dichos combustibles, esto es, la emisin de CO2 y otros gases de efecto invernadero y sus consecuencias en el denominado cambio climtico, han llevado a la clase poltica a tomar conciencia de la necesidad de avanzar hacia otros modelos energticos.Almargendelosgravesproblemasmedioambientales,cabedestacarqueel sistema energtico actual est sometido a numerosas incertidumbres. As, unos pocos pases, por lo general inestables polticamente, controlan el mercado del petrleo y el gas natural, provocando continuas fluctuaciones en los precios. Si adems se tiene en cuenta la inclusin en la demanda energtica de las economas emergentes, unido al hechodequeloscombustiblesfsilessonunbienescaso,seconcluyequeestasituacin noresultasostenibleamedioplazo. Es por ello que se estn dando los primeros pasos hacia lo que se conoce como Economa del Hidrgeno; un proyecto a largo plazo que podra definirse como un esfuerzo decambioen el panoramaenergticohaciaunsistema limpioysostenibleque pretende combinar la limpieza del hidrgeno como vector energtico con la eficacia de laspilasdecombustiblecomosistematransformadordesuenergaenelectricidad,calor y movimiento [1]. Este concepto implicara que el hidrgeno como fuente de energa tendra que obtenerse a partir de recursos renovables o bien a partir de combustibles fsiles como sucede en la actualidad, pero asociados a sistemas de captura y almacenamientodeCO2. Cabe preguntarse por qu basar el futuro energtico en el hidrgeno. La energa msica producida en la combustin del hidrgeno es la mayor existente entre todos los combustibles, con un poder calorfico inferior 2.4, 2.8 y 4 veces superior al del metano, gasolina y carbn, respectivamente; aunque como combustible slo se usa en los sistemasdepropulsinde vehculos espaciales.Laprincipalventajaderivadadeluso del H2 como combustible es la ausencia de emisiones de CO2, as como de otro tipo de contaminantescuandoseutilizaenpilasdecombustibledebajatemperatura[1].

Introduccin

Una pila de combustible es un dispositivo electroqumico que convierte la energa qumica de una reaccin directamente en energa elctrica, siendo alimentados los reactivos de forma continua. Esta transformacin electroqumica no est limitada por el rendimiento del ciclo de Carnot, lo que permite conseguir eficacias relativamente altas (50% aunque tericamente podran ser superiores). Existen diversos tipos de pilas de combustible alimentadas por hidrgeno, siendo las ms habituales las de membrana polimrica (polymer electrolyte membrane fuel cell, PEMFC). En ellas, la temperatura de funcionamiento est limitada por la estabilidad del polmero empleado, generalmente Nafin, en torno a 70100C. A temperaturas tan bajas slo el platino posee la actividad cataltica necesariaparaimpulsar lasreaccioneselectroqumicas delapila.En generalel hidrgeno (combustible) se alimenta al nodo y el oxgeno (oxidante) al ctodo. El hidrgeno suministrado a la pila se difunde a travs del nodo poroso y es activado por un catalizador (Pt) reaccionando sobre la superficie del ctodo con los iones OH, formandoaguayliberandoelectronessegnlasiguientereaccin[2]: H2+2OH2H2O+2e (R1)

En la Figura 1 se presenta un esquema del funcionamiento de una pila de combustible PEM.

H2 H+

+
O2
H2 2H+ +2e O2 +4H+ +4e 2H2O

H2O
nodo Ctodo Electrolito
Figura1.Esquemadeunapiladecombustiblealimentadaporhidrgenoyoxgeno. Evidentemente la implementacin de este nuevo sistema energtico habr de producirse paulatinamente a lo largo de las prximas dcadas, exigiendo un gran esfuerzo en investigacin para solventar los distintos problemas que presenta este modelo. As, el hidrgeno, como vector energtico, ser tan limpio como lo sea su proceso de obtencin. En este sentido tienen un especial inters los sistemas de

3 produccin basados en el empleo de alcoholes que se obtienen a partir de recursos renovables (biomasa y residuos orgnicos) pero tambin a partir de fuentes fsiles en caso de ser necesario (gas natural, principalmente), ya que permitiran una transicin suave desde un sistema energtico basado en combustibles fsiles a la economa verde deloscombustiblesrenovablesobiocombustibles. El mayor reto en el campo de la produccin de hidrgeno surge en las aplicaciones mviles, y ms en concreto en el sector del transporte. En el ao 2005 y en el marco de laUninEuropea,el27%delasemisionesdeCO2deorigenantropognicosedebierona este sector. El problema medioambiental del sector transporte reside en que los vehculos son focos altamente dispersos de emisiones de CO2, en los que no pueden implementarse sistemas de captura y almacenamiento, como ocurre en los focos estacionarios. Las opciones para utilizar hidrgeno en vehculos son fundamentalmente dos: (1) su almacenamiento a bordo y (2) la produccin en el propio vehculo del hidrgeno justo antes de alimentarlo a la pila de combustible. El almacenamiento puede realizarse en forma lquida, gaseosa, combinado qumicamente o adsorbido en slidos porosos, mientrasquelaproduccinenelvehculopuedellevarseacabomediantereformadode un combustible, renovable o no, como un alcohol rico en hidrgeno (p. ej. metanol o etanol). El almacenamiento de hidrgeno a bordo del automvil es un proceso que suscita un gran inters investigador con el fin de limitar tanto los costes como el peso y volumen deldepsitodehidrgeno,yaqueladensidadvolumtricadeenergaenelhidrgenoes muy baja. La densidad de la gasolina es 0.7kg/L, mientras que la del H2 es 0.03, 0.06 y 0.07 kg/L a 350 atm, 700 atm y licuado (20 K), respectivamente, con lo que se necesitan grandes volmenes de hidrgeno para satisfacer los requisitos energticos [1].El Departamento de Energa de los Estados Unidos (DOE, www.energy.gov) ha fijado unos objetivos a corto plazo que se deberan cumplir en el almacenamiento de hidrgeno para su uso en vehculos y que se toman como referencia para determinar la viabilidad de los sistemas de almacenamiento de hidrgeno. Estos objetivos fueron revisados en el ao 2009 para tener en cuenta expectativas ms realistas de almacenamiento. Los objetivosdelDOEsemuestranenlaTabla1.

Introduccin

Tabla 1. Objetivos del DOE para el almacenamiento de hidrgeno en vehculos, incluida laactualizacinrealizadaen2009. Objetivos2010 Objetivos2010 revisados Objetivos2015 Objetivos2015 revisados Densidad gravimtrica 0.06kgH2/kg 0.045kgH2/kg 0.09kgH2/kg 0.055kgH2/kg Densidad energtica 1.5kWh/L (45g/L) 0.9kWh/L (27g/L) 2.7kWh/L (81g/L) 1.3kWh/L (39g/L) Coste $4/kWh porconvenir $2/kWh porconvenir

Existen diversas alternativas para almacenar hidrgeno de forma ms compacta que pugnan por lograr los objetivos del DOE: i) gas comprimido, ii) hidrgeno lquido, iii) hidrurosmetlicosyiv)adsorcinenmaterialesconelevadaporosidad,comomateriales carbonosos. El almacenamiento de hidrgeno comprimido requiere el empleo de tanques presurizados a 300 atm, si bien se est investigando el uso de tanques de fibra decarbonorecubiertosdematerialinertequesoportanpresionesdehasta700atm.Sin embargo estos tanques son muy caros y presentan riesgos, siendo de hecho la compresin del gas en s misma la parte ms complicada y peligrosa del proceso, adems de tener un coste muy elevado [3]. Por otro lado, el almacenamiento de H2 licuado(251.8C)precisadetanquescriognicos,muybienaislados,quesuelenresultar caros y pesados. Ambas opciones, hidrgeno comprimido o licuado, estn penalizadas energticamente,estoes,serequiereunapartedelaenergaquecontieneelhidrgeno parapresurizarloolicuarlo(un20%yun40%respectivamente[4]). La adsorcin en slidos de reas superficiales elevadas, como las nanoestructuras de carbono, es una alternativa en principio atractiva por resultar una forma sencilla y ms segura de manejar el hidrgeno. La capacidad de almacenamiento de H2 de los materiales carbonosos se asocia generalmente a su rea superficial y microporosidad, aunque estudios ms recientes apuntan al hecho de que la microporosidad estrecha (menor de 0.7 nm) es el factor determinante que gobierna dicha capacidad de almacenamiento [5]. Sin embargo, los mejores resultados de almacenamiento en estos materiales (6.9% en peso) [6] se alcanzan a196C y 20 atm, por lo que de nuevo hay queenfrentarsealapenalizacinenergticaqueconllevaesteproceso.

5 Por ltimo, el almacenamiento en hidruros metlicos recargables est atrayendo poderosamentelaatencindelacomunidadcientfica,ypareceapriorielsistemamejor posicionado para alcanzar los objetivos del DOE. Se fundamenta en la capacidad de muchos metales y aleaciones de quimisorber reversiblemente hidrgeno, mediante una reaccin qumica que da lugar al correspondiente hidruro. Cuando se necesita, el hidrgenoesliberado,loquerequiereunaumentodelatemperaturayunadisminucin delapresin[7].CompuestoscomoelLiBH4handemostradotenerunamayordensidad de energa que el hidrgeno lquido. Sin embargo, los valores de presin y temperatura empleados para la quimisorcin y desorcin del hidrgeno no estn dentro de los intervalos de aplicabilidad en dispositivos mviles, que son 010 atm y 0100C, respectivamente. Adems, la ciclabilidad de estos materiales puede verse afectada por los cambios de volumen que experimentan durante la adsorcin y desorcin del hidrgeno, favoreciendo la disgregacin de los hidruros. El almacenamiento en hidruros metlicos requiere mucha ms investigacin para conocer mejor los procesos fsicos y qumicosquegobiernantantolaquimisorcincomolaliberacindelhidrgeno[4]. Evidentemente se est avanzando en la bsqueda de medios novedosos de almacenamiento, pero la opcin de producir el hidrgeno a bordo mediante reformado de biocombustiblesdisfruta,en todo caso,deun desarrollotecnolgicomselevado.La generacin a bordo de hidrgeno tambin presenta inconvenientes, y precisa de mejoras,comoeldesarrollodeconvertidoresporttilesmseficientesybaratos,perose eludenlosproblemasactualesasociadosalalmacenamientodehidrgenoyloselevados costes de su distribucin hacia las estaciones de servicio. El mtodo ms interesante para la produccin de hidrgeno en vehculos es el reformado de un alcohol. El reformadoconsisteenlareaccindelalcoholconvapordeagua(conosinoxgeno)para producir H2 y CO2, adems de una cierta cantidad de CO (subproducto no deseado). Posteriormente el CO ha de eliminarse para evitar la contaminacin del nodo de Pt de la pila de combustible polimrica, aplicndose usualmente una etapa de reaccin de desplazamiento de vapor de agua (water gas shift, WGS) seguida de otra etapa de oxidacin cataltica del CO residual a CO2 (preferential oxidation, PROX) para lograr corrientesdepuradasdeH2. Esta tesis forma parte del conjunto de investigaciones que se llevan a cabo en los laboratorios del Grupo de Materiales Funcionales Porosos del INCAR, encaminados a la puesta a punto de un sistema de produccin de hidrgeno a partir de bioalcoholes a bordo del automvil. En concreto el presente trabajo tiene como objetivo el estudio de distintos aspectos de la etapa de oxidacin preferencial de CO empleando para ello

Introduccin

nanocatalizadores no soportados sintetizados por un mtodo propio de nanomoldeo y nanocatalizadores soportados en mallas metlicas sintetizados por el mtodo de evaporacininducidadeamoniaco.

2.

PRODUCCIN Y PURIFICACIN DE HIDRGENO A BORDO DEL AUTOMVIL

Actualmente, se producen alrededor de 0.1 Gton de H2 por ao, el 98% a partir de combustibles fsiles por procesos de reformado [1]. El H2 se utiliza principalmente en el refino del petrleo y la produccin de amoniaco y metanol. En un futuro sostenible el hidrgeno habr de obtenerse de otras fuentes de energa, en procesos en los que, al menosalargoplazo,lasemisionesdeCO2tendrnqueserminimizadas. En este sentido, la biomasa constituye uno de los recursos renovables con perspectivas msfavorablesydemayorintersparalaobtencindecombustiblesparaeltransporte. A partir de la biomasa se pueden obtener combustibles lquidos, con lo que se aprovechara la red de distribucin de combustibles fsiles lquidos ya existente y el depsito de combustible no sufrira modificaciones significativas. Adems, si se evita la utilizacindecombustiblesfsilestantoduranteelprocesodeobtencindelbioalcohol, comoeneldesutransformacinenhidrgenoseconsigueunbalancenetonulodeCO2. Si durante la etapa de obtencin de hidrgeno se aplican adicionalmente sistemas de captura y almacenamiento de CO2 el balance global puede ser negativo, es decir el procesoglobalconstituiraunsumiderodeCO2. Clculos recientes desarrollados en nuestro laboratorio muestran que la disponibilidad estimada de biomasa en el ao 2050 ser suficiente para asegurar la produccin de hidrgeno para el transporte mediante el proceso biomasa biometanol hidrgeno, representado en la Figura 2 [1]. Con otros biocombustibles, ms populares hoy en da, como el bioetanol de caa o maz o el biodisel de aceites vegetales, habra un claro dficitdebiomasaenergtica[1].Enestesentido,nuestrogrupodetrabajohaapostado por el empleo del metanol frente al etanol por varios motivos adems de su disponibilidad: tiene una elevada densidad energtica, es fcilmente biodegradable y requiere menores temperaturas de reformado que el etanol. Adems, al tratarse de un compuesto ms sencillo que el etanol, el nmero de reacciones que pueden tener lugar es menor y por tanto se disminuye la formacin de compuestos intermedios no

7 deseados. Pese a su toxicidad, el metanol es ms seguro en su manejo que la propia gasolina.

CiclodelCO2 (emisionesnetascero)

Produccin centralizada Bioalcohol (lquido) Gasificacin(alta eficacia)

Distribuciny suministro enfase lquida

Procesosen localizacindeuso Reformado Insitu Produccin

WGS+PROX PEMFC

Purificacin

Utilizacin

Biomasa
Cultivos,residuos orgnicos

Electricidad
Hogar,transportes

Figura 2. Esquema de obtencin de electricidad a partir de biomasa en la futura economadelhidrgeno. La produccin de hidrgeno a partir de alcoholes se lleva a cabo mediante procesos de reformado cataltico. Como se ha indicado en el apartado anterior, el reformado consiste en la reaccin de un alcohol (metanol (R2) o etanol, principalmente) con vapor de agua (reaccin endotrmica) para producir H2 y CO2. Mediante la adicin de O2 a la reaccin es posible alcanzar condiciones prcticamente autotrmicas (R3). A continuacinsedetallanlasreaccionesparaelmetanol: Reformado (R2)

0 1 CH3OH+H2O CO2+3H2;H 298K=49.5kJmol

Reformadooxidativo

0 1 CH3OH+(12a)H2O+aO2 CO2+(32a)H2;a=0.125H 298K=11.0kJmol (R3)

Adems de estas reacciones, otros dos procesos catalticos producen hidrgeno a partir demetanol,comosonlaoxidacinparcialyladescomposicin.

8 Oxidacinparcial

Introduccin

0 1 CH3OH+0.3O2 0.6CO2+2H2+0.4CO;H 298K=79.2kJmol

(R4)

Descomposicin (R5)

0 1 CH3OH CO+2H2;H 298K=90.2kJmol

Las reacciones de reformado de metanol suelen llevarse a cabo empleando catalizadores de cobre soportados sobre distintos materiales (ZnO, Cr2O3, CeO2) y empleando diversos aditivos (Al2O3, grafito, etc). La presencia de CO en los gases de salidahadesercontroladaestrictamentedadoquesuconcentracinhadesermuybaja (<1ppm)paraevitarlapotencialdesactivacindel catalizadorde lapiladecombustible. Por esta razn, la corriente rica en hidrgeno que sale del reformador ha de ser posteriormente purificada. De forma global, el proceso de produccin de hidrgeno a partirdeunbioalcoholsemuestraenlaFigura3.

H 2O Bioalcohol O2

Reformado oxidativo

Reformado Oxidacin Q parcial

CO WGSAltaT CO2 H2 WGSBajaT

300500C 70250C 70 100C

PROX

PEMFC

200300C CO + O 2 CO 2 CO + H CO+ H2 O 2 2

Figura3.Produccindehidrgenoapartirdebioalcoholes. El proceso de purificacin de la corriente de hidrgeno suele consistir habitualmente en la reaccin cataltica de desplazamiento de vapor de agua (WGS), para elevadas concentracionesdeCO(R6),seguidadeunaetapadeoxidacinpreferencial(oselectiva) del CO residual (PROX) (R7). En la etapa PROX el monxido de carbono residual se oxida catalticamenteadixidodecarbono.Lasreaccionessemuestranacontinuacin: Reaccindedesplazamientodevapordeagua (R6)

0 1 H2O+CO CO2+H2;H 298K=41.1kJmol

OxidacinpreferencialdeCO (R7)

CO+O2CO2;H0298K=283kJmol1

9 La reaccin de desplazamiento de vapor de agua es un proceso bien conocido desde los aos 40. Se trata de una etapa intermedia en la fabricacin de amoniaco, adems de emplearse en la produccin de otros compuestos que se obtienen a partir del gas de sntesis(mezcladeCOeH2).Enelcasodelaobtencindehidrgenoconpurezaelevada la reaccin de desplazamiento de vapor de agua se lleva a cabo en dos etapas: la primera a temperaturas altas (300500C) con el empleo de catalizadores de FeCr (8 12% de Cr2O3) y la segunda a temperaturas ms bajas (200300C) utilizando catalizadores de CuZnAl [8]. No obstante, en los ltimos aos se estn investigando diversas alternativas, como la incorporacin de una membrana en el reactor WGS, combinandoaselprocesodereaccinconeldeseparacindelH2[9].

3.

OXIDACINPREFERENCIALDECO(PROX)

La reaccin de oxidacin preferencial de CO consiste en la oxidacin cataltica del CO residual para producir CO2 (R7), en presencia de una gran cantidad de hidrgeno, procurando minimizar la oxidacin simultnea del hidrgeno a agua (R8). Esta reaccin suele llevarse a cabo a temperaturas comprendidas entre 70 y 250C [10], las cuales vienen determinadas por la temperatura a la salida de la unidad de WGS de baja temperatura (250C) y la temperatura de operacin de la pila PEM (70100C). La reaccin es irreversible, exotrmica y entra en competencia con la oxidacin del hidrgenoparadaragua. H2+O2 H2O;H0298K=242kJmol1 (R8)

Los catalizadores slidos empleados para la reaccin de oxidacin preferencial de CO (R7) han de cumplir los siguientes requisitos: (i)elevada actividad cataltica en el intervalo de temperatura convenido; (ii) elevada selectividad hacia la oxidacin de CO frentealaformacindeH2O;(iii)toleranciaalapresenciadeCO2yH2Oy(iv)conversin estableconeltiempo. Para lograr actividades catalticas elevadas se emplean nanocatalizadores, es decir, compuestos con tamaos de partcula en la escala nanomtrica, por debajo de 100 nm. Estos materiales tienen la ventaja de poseer una mayor relacin superficie/volumen y portantopresentan una mayorsuperficieactiva paraelacceso delgasyasfavorecerla reaccin. Los catalizadores inicialmente investigados para la reaccin de oxidacin preferencial de CO estaban basados en metales nobles, en semejanza a los empleados en la purificacin de CO de la sntesis de amoniaco [1114]. Sin embargo, dado el

10

Introduccin

elevado precio y escasa disponibilidad de los metales nobles, en los ltimos aos ha surgido un gran inters en el desarrollo de nanocatalizadores basados en metales no nobles. As, el xido de cobre ha recibido una amplia atencin en los ltimos aos porque presenta una actividad y selectividad elevadas [15]. Ms recientemente el xido mixto de cobre y cerio, CuOx/CeO2, que se emplea en otras reacciones catalticas, como la reaccin de desplazamiento de vapor de agua y el reformado de alcoholes, ha sido propuesto como catalizador para la reaccin PROX, dando lugar a numerosas publicaciones en la literatura [1620]. Dada la repercusin de los catalizadores basados en cobre, en el presente trabajo se ha efectuado una exhaustiva bsqueda bibliogrfica y se ha desarrollado un mtodo que permite comparar la actividad cataltica de los diferentes catalizadores con el propsito de conocer los factores que afectan a su actividadcataltica(ArtculoII). Adems de los xidos de cobre y cerio se han utilizado otros xidos metlicos para la oxidacin preferencial de CO, tales como CuMn2O4 y el Co3O4. La espinela de cobre y manganeso (CuMn2O4), usada comnmente en la oxidacin de CO para proteccin respiratoria, se ha analizado recientemente tanto en el reformado cataltico de metanol [21] como, de forma lateral, en la oxidacin de CO en exceso de hidrgeno [22], encontrndose para este proceso actividades moderadas y elevadas selectividades a temperaturassuperioresa100C.Enestatesissehaanalizadolaactividadenlareaccin de oxidacin preferencial de CO de la espinela de cobre y manganeso sintetizada con la tcnicaSACOP(ArtculoIV). Los xidos de cobalto son compuestos de elevado inters industrial. La espinela de cobalto, Co3O4, se emplea en bateras recargables, como material magntico [23] y en sensores de CO [24]. Adems, el Co3O4 se ha utilizado en la oxidacin de CO a baja temperatura [25, 26], y ms recientemente en la oxidacin preferencial de CO [2729]. Porsuparte,eloxihidrxidodecobalto,CoOOH,tambinseempleacomosensordeCO [30] o como aditivo en los electrodos de nquel para bateras recargables [31]. En la presente Memoriasehan sintetizadoxidosdecobaltomediantelatcnica SACOPpara su posterior anlisis de actividad cataltica en la reaccin de oxidacin preferencial de CO(ArtculoV). El principal problema que presentan los catalizadores es su desactivacin, es decir, la prdida de actividad cataltica con el tiempo de uso. En el caso de aplicaciones no estacionarias, donde la vida del catalizador ha de ser similar a la del propio vehculo en el que se instale, es clave disponer de catalizadores que no sufran un alto grado de desactivacin. Dicha prdida de actividad puede deberse a distintas causas, como la

11 sinterizacin, el envenenamiento o la acumulacin de especies carbonosas. La sinterizacin se produce por la prdida de superficie activa debido al crecimiento y aglomeracin de los cristales. Se habla de envenenamiento cuando determinadas impurezas provenientes de la corriente de alimentacin bloquean de forma irreversible los centros activos del catalizador. Por otro lado, el bloqueo de dichos centros se puede producir por la acumulacin de especies carbonosas (por ejemplo carbonatos) que se forman durante la reaccin [8]. Debido a la importancia de esta problemtica, en el presente trabajo se han analizado las causas de la desactivacin en dos tipos de nanocatalizadores: CuOx/CeO2 (Artculo III) y CuMn2O4 (Artculo IV), ambos sintetizados mediantelatcnicaSACOP.

4.

MTODOSDEPREPARACINDENANOCATALIZADORES

4.1.Nanoestructurasnosoportadas:mtodoSACOP
Enelmbitodelacatlisis,laobtencindenanopartculasonanoestructurasdeelevada superficie especfica es uno de los retos a los que se enfrenta actualmente la investigacin cientfica. As, los procesos catalticos que emplean catalizadores slidos de tamao nanomtrico estn ampliamente extendidos y resultan indispensables para llevar a cabo de forma econmica la mayora de las reacciones industriales [32]. En la bibliografa se encuentra un gran nmero de trabajos que describen diversas tcnicas para sintetizar estos materiales con un elevado control de sus caractersticas texturales, empleando para ello procedimientos de tipo solgel [33], microemulsin [34] o las llamadastcnicasdenanomoldeo[35]. Los mtodos de nanomoldeo han tenido una amplia repercusin en los ltimos aos. Se caracterizan por utilizar un material poroso, ya sean slices mesoporosas amorfas u ordenadas, o un carbn poroso como nanomolde en cuyos poros tiene lugar la sntesis controlada del material. Al llevarse a cabo la sntesis en un espacio confinado el crecimiento de las nanopartculas que tiene lugar durante el tratamiento trmico se ve limitado por las paredes de los poros del molde. As, tras la eliminacin del molde, es posible obtener materiales nanoparticulados o nanoestructurados. En el caso de las nanopartculassutamaonopodrsermayorqueeltamaodeporodelnanomoldede partida, mientras que en el caso de las nanoestructuras el tamao de poro de las mismas estar relacionado con el tamao de las paredes entre poros del nanomolde original. A pesar de su simplicidad se trata de un procedimiento laborioso que requiere mltiples etapas de impregnacin del nanomolde para obtener rendimientos

12

Introduccin

significativos y que exige el control exhaustivo de una serie de parmetros: (i) grado de llenado de los poros, (ii) homogeneidad de la impregnacin y (iii) control de la etapa de secadoparaevitarlamigracincapilardelosprecursoresintroducidosenlosporos. Para intentar solventar los inconvenientes que presentan los mtodos de nanomoldeo convencionales se ha desarrollado un mtodo, basado en las tcnicas de nanomoldeo y solgel, denominado coprecipitacin confinada en hidrogel de slice o SACOP [36] (silica aquagelconfined coprecipitation) que permite obtener materiales de superficies ms elevadas que las de los preparados por el mtodo convencional. El mtodo SACOP permitealcanzarrendimientosdexidosmsaltossinnecesidaddemltiplesetapasde impregnacincalcinacin y con un elevado grado de homogeneidad. Adems, no es necesario llevar a cabo un control tan exhaustivo del secado puesto que los precursores metlicos se encuentran durante esa etapa como precipitados de hidrxidos metlicos [36]. El mtodo SACOP parte de un procedimiento para la sntesis de xerogel de slice de porosidadcontroladaapartirdesilicatosdico[37].Elprecursormetlicosedisuelveen la disolucin cida utilizada para sintetizar la slice de forma que se obtiene un hidrogel en el que los cationes metlicos forman parte de la estructura, y por tanto existe un contacto ntimo entre el molde de slice y el precursor del metal. Los iones metlicos confinados en el hidrogel se someten a un proceso de precipitacin durante el cual son transformados en hidrxidos para prevenir la prdida de metal durante las etapas de lavado y secado del gel. Una vez precipitados los hidrxidos del metal correspondiente, lamuestrasecasecalcinaparaqueseformenlosxidosdelosmetalesconfinadosenla microestructura del gel y por ltimo se elimina el nanomolde de slice mediante disolucin con NaOH, obtenindose finalmente los xidos nanoparticulados. Sin embargo, todas las muestras preparadas por SACOP presentan cierta cantidad de slice residual (<10% peso). Este problema es inherente a todas las tcnicas de moldeo con slice y se debe principalmente a la formacin de silicatos de difcil solubilidad. Afecta a metales como cerio, hierro o lantano, mientras que apenas tiene influencia en el caso del cobre, manganeso o cobalto. Por este motivo la etapa de eliminacin de la slice fue optimizadaparalapreparacindexidodecerio. En la literatura reciente existen mtodos similares al mtodo SACOP, conocidos como onepot nanocasting techniques, en los cuales el precursor del xido se mezcla con el precursor del molde de slice con el propsito de producir un hidrogel o alcogel en los que el metal est homogneamente distribuido [38, 39]. Sin embargo estos mtodos han sido desarrollados para la obtencin de nanocomposites y no llevan a cabo etapas

13 de precipitacin ni de disolucin de la slice. Adems, una caracterstica comn de estos trabajoseselusodeprecursoresdeslicedeelevado coste,como eltetraetilortosilicato (TEOS),queencareceelprocesodesntesis. En esta tesis han sido sintetizados diversos compuestos mediante el mtodo SACOP. Se ha revelado como un mtodo muy efectivo para producir xidos de elevada superficie especfica en el caso de la espinela CuMn2O4 [40]. Se han preparado tambin otros xidos como Fe2O3, CuOx/CeO2 y CoFe2O4, presentando en todos los casos reas comprendidasentrelos230y300m2g1yestandoformadosporagregadosdepartculas de entre 2 y 8nm [36]. En este trabajo se ha optimizado el mtodo SACOP para la obtencin de xido de cerio adems de estudiar los mecanismos implicados en las distintas etapas de sntesis (Artculo I). El mtodo SACOP se aplica como mtodo de referencia para la sntesis de los catalizadores empleados en la realizacin de este trabajo.

4.2. Nanoestructuras soportadas: mtodo de evaporacin inducida de amoniaco

La inmovilizacin de las nanopartculas o nanoestructuras en diversos soportes para su posterior uso en reactores catalticos presenta una serie de ventajas respecto a la utilizacin directa de nanopartculas o nanoestructuras, por lo que la investigacin en este campo se est viendo impulsada. Los catalizadores estructurados son fcilmente recuperados del medio de reaccin, provocan cadas de presin prcticamente nulas y en algunos casos pueden favorecer la transmisin de calor a lo largo del reactor. En la literatura se describen distintos soportes, como placas metlicas [41], espumas metlicas [42] o monolitos cermicos [43]. Los mtodos que se suelen utilizar para la sntesis de catalizadores soportados en dichos materiales requieren numerosas etapas parasupreparacin. En la presente memoria se han obtenido nanoestructuras mesoporosas de Co3O4, soportadas en una malla metlica de acero inoxidable, que fueron utilizadas en la reaccindeoxidacinpreferencialdeCO (ArtculoVI).Paraellosetomcomo puntode partida el mtodo de evaporacin inducida de amoniaco utilizado por Li y col. [44, 45] ensustrabajos,queconstadeunaetapadesntesisseguidadeunaetapadecalcinacin en aire del sustrato recubierto. Este sencillo procedimiento experimental es muy interesante porque tanto la sntesis del catalizador como la inmovilizacin del mismo tienen lugar en un solo paso. Este mtodo permiti a sus autores el crecimiento de

14

Introduccin

nanoalambres huecos de Co3O4 sobre distintos sustratos, como lminas de titanio o silicio para su uso en bateras de Liin. Las nanopartculas obtenidas por estos autores presentabanreassuperficialesrelativamentealtas(74m2g1).DeacuerdoconLiycol. laformacin deCo3O4 esconsecuencia deunareorganizacin topotcticaque ocurreen presenciadeoxgenodurantelatransformacindeCo(OH)2enCo3O4mesoporoso. Sin embargo, entre el trabajo de Li y col. y el realizado en nuestro laboratorio existen algunasdiferenciassignificativas.Ennuestrocasohemosempleadocomosoportemallas deaceroinoxidable muy delgadas.Esto permiteelfcilarrollamientodelasmismasuna vezdepositadalafaseactiva,locualasegurasuuso comorellenocatalticoen reactores tubularesconbajascadasdepresinyelevadoscoeficientesdetransmisindecalor.Al emplear como sustrato una malla metlica de acero inoxidable, el crecimiento de las nanoestructuras de Co3O4 tuvo lugar en forma de nanoalambres slidos, de lo que se deducequeelsustratoempleadoinfluyedemanerasignificativasobrelageometrafinal delosnanoalambres.Adems,elrendimientoobtenidofuede7mgdeCo3O4cadacm2 de malla. En cambio Li y col. obtuvieron un rendimiento de 23 mgcm2 empleando comosustratoslminasdetitanio. En un trabajo reciente de nuestro grupo el mtodo de preparacin de la espinela soportada fue optimizado [46], obtenindose resultados que contradicen parcialmente los mostrados por Li y col. en sus artculos. Para estos autores, la formacin de la espinela mesoporosa de alta superficie tiene lugar durante la etapa de sntesis y no durante la calcinacin posterior. Sin embargo, en nuestro trabajo observamos que una mayorpresenciadeCo(OH)2enlamuestrasincalcinarproducamuestrascalcinadascon mayores superficies especficas, siendo la abundancia de Co(OH)2 inversamente proporcional al tiempo empleado en la sntesis. Este comportamiento se debe a que para largos tiempos de sntesis el Co(OH)2 se va transformando gradualmente en Co3O4 de baja superficie especfica (tamao de cristal de 26 nm), mientras que el Co3O4 producidodurantelacalcinacindelCo(OH)2tienemenorestamaosdecristal(~10nm) y es por tanto el responsable de una mayor rea superficial en las muestras calcinadas. Por esta razn es conveniente obtener la mayor cantidad de Co(OH)2 durante la etapa desntesis. Para analizar la actividad cataltica en la reaccin PROX de estos catalizadores se construy un microrreactor tubular de acero de 1/4 de dimetro externo, descrito en detalle en el apartado Experimental. En este tubo se introdujo enrollada una malla de 5cm 3cm recubierta de nanoestructuras de espinela de cobalto, para una altura final del monolito metlico de 3 cm. Dada la cantidad media de fase activa en el monolito

15 (100mg de Co3O4) y el volumen del mismo, se obtuvieron relaciones masa de catalizador/volumen de reactor un orden de magnitud superiores a las obtenidas en otrossistemasdereaccincomoalgunosmicrorreactoresacanalados. Adicionalmente, las mallas metlicas recubiertas con Co(OH)2/Co3O4 han sido analizadas por nuestro grupo en la reaccin de descomposicin de metanol [46] para producir H2 y CO (R5). Para mejorar la actividad cataltica de este catalizador se emple cobre como agente dopante, obtenindose los catalizadores basados en metales no nobles ms activos, selectivos y estables para la descomposicin de metanol recogidos hasta el momentoenlabibliografa.

17

Objetivosyplanteamientodelamemoria
1. OBJETIVOS
Como se ha puesto de manifiesto en la Introduccin, los procesos de produccin de hidrgeno a partir de bioalcoholes suponen una alternativa razonable para realizar una transicin progresiva hacia la llamada Economa del Hidrgeno. La purificacin de la corrientedehidrgenoesunaetapafundamentalalahoradeevitarelenvenenamiento delnododeplatinopertenecientealapiladecombustiblepolimrica. Por todo ello y conforme a lo anteriormente expuesto, la presente memoria tiene como objetivo general la obtencin de nanocatalizadores basados en xidos de metales de transicin que sean activos, selectivos y estables en el proceso de purificacin de corrientes de hidrgeno mediante la oxidacin preferencial de monxido de carbono. Para cumplir este objetivo general se han planteado distintos objetivos especficos, que sedetallanacontinuacin: 1. Estudio de la tcnica de coprecipitacin confinada de hidrogeles de slice, SACOP, para la obtencin de catalizadores de alta superficie especfica. Optimizacin de la tcnicaparalaobtencindexidodecerio. 2. Obtencin de nanocatalizadores de xido de cobre y xido de cerio por el mtodo SACOP.Estudiodesuactividad,selectividadyestabilidadcatalticaenlareaccinPROX. 3. Sntesis de manganita de cobre nanomtrica por el mtodo SACOP. Estudio de su actividad,selectividadyestabilidadcatalticaenlareaccinPROX. 4.PreparacindenanoestructurasdexidodecobaltoporelmtodoSACOP.Estudiode suactividad,selectividadyestabilidadcatalticaenlareaccinPROX. 5. Obtencin de nanocatalizadores mesoporosos de Co3O4 soportado sobre mallas metlicas. Estudio de su actividad, selectividad y estabilidad cataltica en la reaccin PROX.

18

Objetivos

2.

ORGANIZACINDELAMEMORIA

Laconsecucindelosobjetivosespecficosmencionadosanteriormentehatenidocomo resultado la publicacin de cinco artculos ms uno recientemente enviado para su publicacinenChemCatChem.PorestaraznlapresenteMemoriasepresentacomoun compendio de publicaciones. Adems de la Introduccin, esta Memoria consta de un captulo Experimental en el que se describen tanto los procedimientos de sntesis como las tcnicas de caracterizacin y los equipos empleados para analizar la actividad cataltica de las muestras. La siguiente seccin est compuesta por el compendio de artculos, cada uno de los cuales va precedido de un breve resumen donde se recogen los aspectos fundamentales de los mismos. Finalmente se presentan las principales Conclusionesderivadasdelarealizacindeestetrabajo. LosartculosincluidosenestaMemoriasonlossiguientes: I. The synthesis of high surface area cerium oxide and cerium oxide/silica nanocompositesbythesilicaaquagelconfinedcoprecipitationtechnique. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. Microporous and Mesoporous Materials,127(2010)198204. II.AnattempttorankcopperbasedcatalystsusedintheCOPROXreaction. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. International Journal of Hydrogen Energy,33(2008)197205. III. Preferential oxidation of CO by CuOx/CeO2 nanocatalysts prepared by SACOP. Mechanismsofdeactivationunderthereactantstream. Gregorio Marbn, Irene Lpez, Teresa ValdsSols. Applied Catalysis A: General, 361 (2009)160169. IV.CoppermanganiteasacatalystforthePROXreaction.Deactivationstudies. Teresa ValdsSols, Irene Lpez, Gregorio Marbn. International Journal of Hydrogen Energy,35(2010)18791887. V. Highly active cobalt based catalysts prepared by SACOP for the preferential oxidation ofCOinexcesshydrogen. Irene Lpez, Teresa ValdsSols, Gregorio Marbn. Artculo enviado a la revista ChemCatChem.

19 VI. Highly active structured catalyst made up of mesoporous Co3O4 nanowires supported onametalwiremeshforthepreferentialoxidationofCO. Gregorio Marbn, Irene Lpez, Teresa ValdsSols, Antonio B. Fuertes. International JournalofHydrogenEnergy,33(2008)66876695. Adems, otro artculo ha sido publicado durante la ejecucin de esta Memoria, aunque noformapartedeellaporestarfueradelobjetivogeneraldelamisma. A highly active, selective and stable copper/cobaltstructured nanocatalyst for methanol decomposition. Gregorio Marbn, Alba Lpez, Irene Lpez, Teresa ValdsSols. Applied CatalysisB:Environmental,99(2010)257264.

3.

ASISTENCIAACONGRESOS

De forma complementaria a las publicaciones indicadas anteriormente han sido aceptadosparasucomunicacinencongresoslossiguientestrabajos: Desactivacin de catalizadores de cobre en la oxidacin preferencial de CO. I. Lpez, G. Marbn, T. ValdsSols, A. B. Fuertes. IX Reunin del Grupo Espaol delCarbn,Teruel2224Octubre2007(pster). Reformado cataltico de metanol sobre catalizadores de CuMn2O4 preparados mediante coprecipitacin confinada en gel de slice. T. ValdsSols, I. Lpez, G. Marbn, A.B. Fuertes. IX Reunin del Grupo Espaol del Carbn, Teruel 2224 Octubre2007(pster). Sntesishidrotermaldecatalizadoresparaelreformadocatalticodemetanol.G. Marbn,I.Lpez,T.ValdsSols,A.B.Fuertes.IXReunindelGrupoEspaoldel Carbn,Teruel2224Octubre2007(pster). Activity and causes of deactivation in COPROX reaction of CuOx/CeO2 nanocatalysts prepared by SACOP. G. Marbn, I. Lpez, T. ValdsSols. Hyceltec 2008 I Simposium Ibrico de Hidrgeno, Pilas de Combustible y Bateras Avanzadas,Bilbao,14Julio(comunicacinoral). Co3O4 catalyst prepared by inverse SACOP for the preferential oxidation of CO. I. Lpez, G. Marbn, T. ValdsSols. Hyceltec 2009 II Iberian Symposium on Hydrogen, Fuel Cells and Advanced Batteries, Vila Real, Portugal, 1317 Septiembre(comunicacinoral).

21

Experimental

1. SNTESISDECATALIZADORES
1.1. Nanoestructurasnosoportadas
a)MtodoSACOP
El mtodo SACOP [36] (silica aquagel confined coprecipitation), esquematizado en la Figura 4, parte del procedimiento descrito por Fuertes [37] para la preparacin de xerogel de slice utilizando silicato sdico como precursor. Para obtener el hidrogel de slice con los iones metlicos incorporados en su fase acuosa, se disolvi el precursor (el nitrato del metal correspondiente) en una disolucin 4 M de HCl o HNO3 con proporciones molares SiO2/H+/H2O =1/6.54/193.9 y Si/Metal=4 o 1.5. La formacin del hidrogel con los metales dispersos tiene lugar en dos etapas, como ocurre con el hidrogel de slice. En la primera etapa, de hidrlisis, la mezcla se agit a temperatura ambiente durante 24 h para obtener el cido silcico a partir del silicato de sodio. En la segunda etapa la mezcla hidrolizada se someti a una etapa de condensacin que tuvo lugara100Cdurante24hencondicioneshidrotermales(autoclavecerrado). Tras la etapa de condensacin se obtuvo un gel en cuya fase acuosa se encontraban disueltoslosionesmetlicos.Paraevitarellixiviadodeloscationesduranteellavadodel gel, se precipitaron previamente en forma de hidrxidos. Para ello fue necesario aumentar el pH del hidrogel mediante el uso de un agente bsico. Dependiendo del catin metlico se emple amoniaco en fase vapor o NaOH (directamente en polvo o bienendisolucin).Enel casodelxidodeceriose utilizamoniacoenfasevapor,para lo cual se borbote el agente a travs de un tubo en U en el que se encontraba el gel, hasta la precipitacin del hidrxido de cerio. Para el xido mixto de cobre y cerio, CuOx/CeO2, y la manganita, CuMn2O4 se emple NaOH en polvo, para lo cual se fue vertiendo el agente bsico poco a poco sobre el gel controlando la subida del pH. En el caso de las muestras conteniendo Co3O4/CoOOH se analizaron dos tipos de precipitacin: (i) por vertido del gel sobre una disolucin 1.25 M de NaOH y (ii) por adicindeunadisolucin3.5MdeNaOHgotaagotasobreelhidrogel.

22
HCloHNO3 +nitratosde metal

Experimental

NaOHSiO2 H2O

Loshidrxidosdelmetalestn confinadosenlaestructuradelaslice

Hidrlisisy condensacin (recipiente cerrado) Compuestosinorgnicos nanomtricoso nanoestructurados

Adicindeun agentebsico Hidrxidosno solubles(lavado, secadoy calcinacin) Eliminacindela slice Nanocomposite

Figura4.EsquemadelmtodoSACOP. Una vez llevada a cabo la precipitacin de los cationes metlicos en forma de hidrxido, la pasta resultante se someti a tres etapas de lavado con agua y centrifugado (7500rpm, 3 min), para eliminar las especies solubles. El agua que forma parte del hidrogel se intercambi posteriormente con acetona en un procedimiento en dos etapas. En la primera de ellas se mantuvo el gel disperso en acetona durante una hora. Posteriormenteelprecipitadosefiltrysevolviaverterenacetonadondesemantuvo durante el tiempo necesario para asegurar el intercambio completo (de 24 a 48 h). Este procedimiento asegur la completa sustitucin del agua estructural por acetona, lo que permiti llevar a cabo el secado del gel a temperatura ambiente durante 24h. La slice con el hidrxido metlico embebido en su matriz se calcin en una mufla a la temperatura correspondiente durante 4 horas (15Cmin1), obtenindose el composite xidometlico/slice. La ltima etapa del procedimiento consisti en la eliminacin del molde de slice para recuperar las nanoestructuras del xido metlico. Como procedimiento estndar el composite de slice/xido metlico(~12 g) se dispers en 50 mL de una disolucin de NaOH(2M)bajoagitacinmecnicaa30Cdurante~48h,cambiandoladisolucinalas 24 horas y finalizando con un lavado en agua destilada. Posteriormente el procedimientodelavadoseoptimizmediante elusodeNaOH3.5Men unvasoaislado trmicamente,bajoagitacinmecnicaydurante6h.

23

b)Impregnacindeslicecomercial
Adicionalmente se sintetizaron nanocatalizadores de idnticas composiciones a las obtenidas por SACOP, empleando para ello un mtodo de nanomoldeo convencional (hard template) con una slice mesoporosa comercial. Las sales hidratadas de los metales (nitratos) se disolvieron en etanol. Con esta solucin se impregn hasta humedad incipiente una slice comercial (Aldrich, SBET=340m2g1, Vp=0.89cm3g1, KJS=12.4 nm). La muestra impregnada se sec en estufa a 100C durante 2 h. El ciclo de impregnacinsecado se repiti un nmero de veces (generalmente 56) hasta que toda la disolucin precursora fue impregnada en la slice. La muestra impregnada final se calcin en aire en una mufla en las condiciones adecuadas de tiempo y temperatura para obtener los xidos embebidos en la matriz de slice. La recuperacin de las nanoestructuras se llev a cabo por eliminacin del molde mediante disolucin de la sliceconunaetapaestndardelavadoenNaOH(2M).

1.2. Nanoestructurassoportadas

El mtodo de sntesis de la espinela de cobalto (Co3O4) soportada en una malla metlica se bas en el utilizado por Li y col [45] para la obtencin de este xido en soportes de titanio, slice, etc. La tcnica, denominada evaporacin inducida de amoniaco (Ammoniaevaporationinduced method), consisti en la mezcla de 35 g de agua y 15 g de amoniaco al 30% en peso en un vaso de tefln en el que previamente se haban introducido 10 mmol de Co(NO3)2 y 5 mmol de NH4NO3. La mezcla se agit magnticamente durante media hora al aire y una vez retirado el imn se introdujo en su seno la malla de acero inoxidable limpia, de dimensiones 5 cmx3 cm (dimetro de alambre=29 m y 45 m de luz de malla). El recipiente se cubri con un vidrioderelojy la malla permaneci en la disolucin a 90C durante 17 horas. Despus la malla se someti a un lavado con abundante agua destilada y se sec a 50C en una estufa de vaco.Porltimo,secalcinenairea250Cdurante2horas.EnlaFigura5asemuestran dos mallas, la de la derecha es la malla limpia sin recubrir y a la izquierda se muestra el aspecto final de la malla recubierta de nanoestructuras de Co3O4. Una microfotografa deestasnanoestructurassereproduceenlaFigura5b.

24

Experimental

200 m

Figura 5. (a) Aspecto de las mallas catalticas en dos momentos distintos de su obtencin; (b) microfotografa SEM de la malla recubierta por las nanoestructuras de Co3O4.Enelinsertosepuedeobservarlamallametlicasinrecubrir. Para el anlisis de la actividad cataltica de las mallas, stas se enrollaron y se introdujeron en reactores tubulares de acero de 1/4 de dimetro externo. Durante esta operacin result inevitable el desprendimiento de algunas partculas. Para minimizar estas prdidas, la malla hubo de ser previamente recubierta con una capa de alcohol polivinlico (polyvinyl alcohol, PVA). Para ello, se sumergi en una disolucin acuosadePVA(900010000gmol1)al2%enpeso,despusseeliminelexcesodePVA soplando suavemente con aire y por ltimo la malla se sec a vaco a 40C durante 20min. Una vez insertada en el reactor, el conjunto se sumergi nuevamente en un bao de agua durante algunas horas para eliminar la capa protectora por disolucin del PVA.Parafinalizarelconjuntoreactor/mallasesecavacoa50C.

25

2.CARACTERIZACINDELOSMATERIALES
2.1.DifraccinderayosX
La determinacin de las fases activas de los catalizadores se llev a cabo mediante difraccin de rayos X. Para obtener los difractogramas de las muestras preparadas se utiliz un equipo Siemens D5000, provisto de un nodo de cobre con radiacin CuK (=0.15406nm).Elequipotrabajaa40kVy20mAydisponedelosprogramasDiffracAt y Eva para interpretar los espectros. El ngulo de barrido 2 se vari entre 10 y 90. Las muestras se prepararon por el mtodo de polvo soportado en un portamuestras de polimetilmetacrilato. El tamao de cristal se obtuvo aplicando la ecuacin de Scherrer a picos seleccionados delosespectrosdedifraccin[47]:

d XRD =

k cos

(E1)

Donde dXRD es el dimetro de cristal en nm, es la longitud de onda de la radiacin, 0.15406 nm, es el ngulo de difraccin del pico analizado, k es la constante de Scherrer, que toma un valor medio de 0.89, y es la anchura a media altura del pico expresadaenradianes.

2.2.IsotermasdeadsorcindeN2
Las isotermas de adsorcin fsica de N2 a 196C se obtuvieron en dos equipos Micromeritics ASAP 2010 y 2020. El rea superficial (SBET) se dedujo del anlisis de la isotermaenelintervalodepresinrelativade0.04a0.20.Elvolumentotaldeporos(Vp) se calcul a partir de la cantidad de N2 adsorbido a una presin relativa de 0.99. El dimetro efectivo de partcula (dBET) se calcul a partir del valor del rea superficial y considerandopartculasesfricasdeigualtamao,segnlasiguienteexpresin:

dBET =

6 xido S BET

(E2)

Donde xidoes la densidad real del xido metlico. La distribucin de tamaos de poro (pore size distribution, PSD) se determin aplicando el mtodo de KrukJaroniecSayari (KJS) a la rama de adsorcin [48], y se consider como tamao medio de poro (KJS) el correspondientealmximodelacurvaPSD.

26

Experimental

2.3. Microscopa electrnica de barrido (SEM) y anlisis por energa dispersivaderayosX(EDX)

La microscopa electrnica de barrido (SEM) permite la observacin y caracterizacin superficial de materiales slidos, tanto inorgnicos como orgnicos. En este trabajo se emple un microscopio Zeiss DSM 942 equipado con un analizador Oxford LinkIsis de energa dispersiva de rayos X (EDX) para establecer la composicin superficial de la muestraanalizada.

2.4.

Microscopaelectrnicadetransmisin(TEM)

Lamicroscopaelectrnicadetransmisinpermiteobservarlaaglomeracinyeltamao efectivo y la forma de las nanopartculas y de los materiales nanoestructurados obtenidos. Las micrografas se obtuvieron en un microscopio electrnico de transmisin JEOL2000EXII,operandoa160kVyconunpoderderesolucinde3.4entrelneas.

2.5.

EspectroscopainfrarrojaportransformadadeFourier(FTIR)

La espectroscopa infrarroja es una tcnica muy extendida en la caracterizacin cualitativa y cuantitativa de la superficie de materiales slidos. Se utiliz un espectrofotmetroNicoletMagnaIRSystem560paraobtenerlosespectrosIR.Elslido fuedispersadoenunamatrizdeKBrconunaproporcinde1mgdematerialen100mg deKBr.Seprepararontrespastillaspormuestra.

2.6.

EspectroscopafotoelectrnicaderayosX(XPS)

La espectroscopa fotoelectrnica de rayos X es una tcnica analtica de superficies ampliamente utilizada en la actualidad. Los espectros XPS se obtuvieron en un equipo Specs Phoibos 100 que emplea una fuente de radiacin de rayos X con doble nodo de MgK (1253.6 eV) trabajando a 50W. Los espectros obtenidos fueron corregidos empleando la energa de enlace del carbono adventicio (284.6 eV) en la regin C1s. Todas las regiones analizadas fueron sometidas a desconvoluciones de tipo mixto GaussianaLorentziana (90:10). Para los anlisis cuantitativos se emplearon los factores desensibilidaddisponiblesenlabasededatosdelprogramaCasaXPS(v2.3.12Dev6).

27

2.7.

Reduccinatemperaturaprogramada(TPR)

La reduccin a temperatura programada de las muestras se llev a cabo en un analizadorAutoChemIIequipadoconundetectorTCDyunespectrmetrodemasas.En un experimento estndar se emplearonaproximadamente25 mg demuestra porla que se hizo pasar un flujo de 50 mLmin1 de una corriente compuesta por un 10% de hidrgenodiluidoenAr.Lasmuestrassecalentaronhasta600u850Ca5Cmin1.

2.8.

Termogravimetra(TG)

Los experimentos de termogravimetra permiten determinar la variacin de masa que experimenta una muestra durante su calentamiento en atmsfera controlada. Los anlisisse llevaronacabo enuna termobalanzaCIElectronics, tratndose lamuestra en un flujo continuo de N2 de 100 mLmin1 con una velocidad de calentamiento de 5Cmin1hasta700C.

3. ANLISISDEACTIVIDADCATALTICA
3.1. Reactores
Los anlisis de actividad PROX se llevaron a cabo en tres dispositivos experimentales diferentes,quesedetallanacontinuacin.

a) Reactordecuarzo
Se emple un reactor de cuarzo de 7 mm de dimetro interno insertado en un horno elctrico vertical de temperatura programada. La muestra (1530 mg) se diluy con carburodesilicio(285275mg)ysecolocentredoscapasdefibradevidrio,paraevitar gradientes de temperatura en el lecho y trabajar en condiciones de flujo de pistn. La temperatura de la muestra se midi mediante un termopar situado encima del lecho cataltico. Elanlisis de los gases de salida del reactor se realiz mediante un cromatgrafodegasesHP6890equipadoconundetectorTC.

28

Experimental

b) Microrreactordeacero
ParaanalizarlaactividadcatalticadelasnanopartculasdeCo3O4soportadasenlamalla metlicaseconstruyunreactordeaceroconundimetrode1/4y3cmdealtura,tal y como se muestra en la Figura 6. La malla metlica se enroll para introducirla dentro del tubo y una vez colocada se procedi a retirar la capa protectora de alcohol polivinlicomedianteunlavadoconaguadestilada,comoseindicenelapartado1.2de esta seccin Experimental. Despus de esta operacin se unieron sendas tuberas de acero de 1/8 por ambos extremos del reactor para la entrada y salida de gases y el conjuntoseinsertenelhornoverticaldescritoenelapartadoanterior.Latemperatura delamuestrasemidipormediodeuntermoparsituadoenelmicrorreactor.Lamezcla de gases de reaccin se obtuvo a partir de botellas patrn empleando controladores de flujo msico. El anlisis de los gases de salida del reactor se realiz mediante un cromatgrafodegasesHP6890equipadoconundetectorTC. Mallaenrollada Mallaenrrollada

Tubo de1/4 Tubo de3/8 con con racores racores

Figura6.Microrreactordeacerorellenodelamallacataltica.

c) Reactormulticanal
Porltimo,parallevaracabolasreaccionesPROXenlaetapafinaldelatesis,seemple unsistemadereaccinmulticanaldiseadoennuestrolaboratorio.Elequipoconstade: (a) un sistema de entrada de gases; (b) un sistema de direccionamiento y divisin de las corrientes de fluidos, consistente en una zona de vlvulas y controladores de presin automatizados para regular la presin y entrada de flujo hacia los distintos reactores; (c)un sistema de 6 reactores insertos en un horno elctrico de temperatura programable y (d) un sistema de anlisis que comprende una vlvula de multiposicin

29 para seleccionar el reactor cuyo gas de salida va a ser analizado y un espectrmetro de masas. En la Figura 7a se observa una fotografa del sistema de reaccin multicanal, con la estufaenlapartesuperioryelsistemadedireccionamientodeflujoenlaparteinferior. Enlaimagen7bse muestraen detallelapartesuperiordelosseisreactoresmetlicosy la localizacin del termopar de control, mientras que en la Figura 7c se ofrece una capturadepantalladelprogramadecontroldelsistemadereaccinmulticanal.
Figura 7. (a) Fotografa del sistema de reaccin multicanal; (b) detalle de la parte superior de los seis reactores; (c) captura de pantalladelprogramadecontroldelsistema multicanal.
R-6 R-1

T1control

Lasprincipalescaractersticasdeestesistemaseenumeranacontinuacin: Permiteanalizarsimultneamentelaactividadcatalticadehasta6muestras. Permite operar a temperaturas de hasta 400C, con control programado de mltiples etapas de calentamiento por medio de una resistencia elctrica. Un termopar insertado endicharesistenciaelctricacontrolalatemperaturadereaccin(Fig.7b),mientrasque dos termopares adicionales situados cada dos reactores permiten conocer las diferenciasdetemperaturaentrelosreactores.

30 Sepuedenemplearpresionesdetrabajodehasta16atmsferas.

Experimental

Se pueden alimentar dos corrientes de fluidos de forma automatizada: el fluido test para el anlisis de actividadde los catalizadoresyel fluidoinerte (He) para llevara cabo pretratamientos y para mantener la presin del sistema. La composicin del fluido test se regula por medio de controladores de flujo, mientras que el flujo de la corriente que entra a cada reactor se controla de forma individualizada por medio de un conjunto de controladoresanalgicosdeflujomsico. Todo el sistema de alimentacin y reaccin se encuentra calefactado, bien con hilo trmico o en estufa, para poder introducir y analizar tanto corrientes gaseosas como de vapores. El anlisis de la concentracin de los fluidos de entrada y salida del reactor se lleva a cabo por medio de un espectrmetro de masas (Modelo Omnistar 300O de Baltzers) operandoenlnea. Laseleccindelreactorcuyacorrientedesalidahadeseranalizadasellevaacabopor mediodeunavlvulademultiposicinautomatizada. El control de las vlvulas de todo o nada, de la vlvula de multiposicin, de los controladores de presin, del calentamiento/enfriamiento de los reactores y del espectrmetro de masas se lleva cabo por medio de un programa de control (en entornoWindows,Fig.7c),locualpermitelaoperacindesatendidadelreactordurante largosperodosdetiempo. Una parte fundamental de este equipo es el conjunto de reactores de acero. Cada uno tiene un dimetro interno de 3.86 mm y el catalizador se sita entre dos zonas de fibra devidrio,comoseindicaenlaFigura8.Lafibradevidriodelazonainferiorseapoyaen un tubo de 1/8 situado dentro del reactor. El sistema de cierre de los reactores de acero, por medio de juntas tricas de vitn, permite la insercin y extraccin de los mismos mediante tapones de doble muelle que se enroscan y desenroscan de forma manual, asegurando la completa estanqueidad del sistema a la mayor presin de operacin (16atm). En la Figura9 se muestranlasdistintaspiezasdeuntapn dedoble muelle (uno de los muelles est situado en el interior del cuerpo cilndrico que aparece enelcentrodelafigura).

31

Entradade gases Fibradevidrio Muestra Tubode1/8 Reactor1/4

Salidade gases

Figura8.Esquemadeunreactordeacerodelsistemamulticanal.

Figura 9. Despiece del tapn de doble muelle de un reactor de acero del sistema multicanal. En cada ensayo estndar con nanoestructuras sin soportar se emplearon 5.5 mg de catalizador diluidos en 75 mg de carburo de silicio, y se hizo pasar un flujo de 35mLmin1 de mezcla gaseosa por cada uno de los 6 reactores, obtenindose velocidadesespacialesmuyelevadas,deaproximadamente39000h1. EnlaFigura10semuestraunesquemageneraldelsistemadereaccinmulticanal.

32

Experimental

Espectrmetro demasas (Omnistar 300O) VlvulaFPC Manometro Capilar

U12
Hasta ~16atm

Venteo

Vlvula multiposicin
v29

U3 U14
U28

v25 v26
U27

v27

28

v30

MF

MF

MF

MF

MF

MF

Sistema de reactores R1 R2 R3 R4 R5 R6

U10

Controladores U4

U5
U16

U6
U18

U7
U20

U8
U22

U9
U24

LEYENDA
U26

Controladores de flujo msico (U4-11) Vlvulas on-off (v1-30) U12-28 Vlvulas antirretorno
(rosa = 1/3 psi, blanco = 2 psi)

v1

v2

v5

v6

v9

v10

v13 v14

v17 v18

v21 v22

U15

U17

U19

U21

U23

U25

Venteo

v3

v4

v7

v8

v11

v12

v15

v16

v19

v20

v23

v24

MF

Microfiltros U1-3 Controladores de presin

BPC

U1 BPC

U2

Fluido inerte
200500mL/min, 1atm

Fluido test
200500mL/min, 1atm

Elementos en estufa Lneas calefactadas

Entrada degasesexterna (sistema de6controladores deflujo)

Figura10.Esquemadelsistemadereaccinmulticanal

U13
(0-500) (0-100)

U11

(0-100)

(0-100)

(0-100)

(0-100)

(0-100)

(0-100)

33

3.2.

Procedimientoexperimentalyanlisisdedatos

La metodologa seguida en la determinacin de la actividad cataltica de los catalizadores preparados fue independiente del reactor utilizado. En la Figura 11 se representa esquemticamente el modo de operacin. Durante el pretratamiento de las muestras se realiz el calibrado del equipo detector (cromatgrafo de gases o espectrmetro de masas) haciendo pasar el gas patrn o la mezcla de gases patrones directamentealanalizadorporelcircuitoalternativoalreactor(bypass).
Calibrado Reacci n Pre tratamiento Pre tratamiento Calibrado

Figura11.ProcedimientoexperimentalparalasreaccionesPROX.

En funcin del analizador utilizado se modific el procedimiento para hacer los calibrados. Con el cromatgrafo de gases el calibrado se realiz pasando hacia el analizadorlamezcladereaccinmientrasqueenelcasodelespectrmetrodemasasse hizo pasar primero cada gas por separado y despus la mezcla de reaccin. Posteriormente, durante la etapa de reaccin, se hizo pasar la mezcla de gases por el reactor y se tomaron los datos en continuo. Tras las etapas de reaccin se finaliz el experimento con un nuevo calibrado. Para considerar la deriva de las seales obtenidas pormediodelespectrmetrodemasasenexperimentosdelargaduracin,sellevarona cabo calibrados intermedios durante las etapas de enfriamiento de dichos experimentos. Tanto al emplear el cromatgrafo de gases como el espectrmetro de masas se inici la etapa de reaccin slo tras haber obtenido una seal estable de la mezcladegasesenlaetapadecalibrado. El pretratamiento de las muestras se llev a cabo siempre con una corriente de He, a temperaturas en torno a 200C. En los reactores de cuarzo y acero conectados al cromatgrafo de gases se utilizaron dos tipos de mezcla durante la etapa de reaccin PROX;debajayaltaconcentracin.Lamezcladebajaconcentracin(lowconcentration, LC) estuvo compuesta por 300 ppm CO, 300 ppm O2 y 10000 ppm H2 (resto helio), mientrasqueladealtaconcentracin(highconcentration,HC)estuvoformadapor5000 ppm CO, 5000 ppm O2 y 25% H2 (resto helio). En el caso del sistema de reaccin

34

Experimental

multicanal se utiliz la siguiente mezcla gaseosa: 4000 ppm CO, 4000 ppm O2 y 15% H2 enhelio. LaconversindeCOobtenidadurantelareaccinPROXseevaluapartirdelosdatos decomposicindegasesofrecidosporlosanalizadorescomo: [CO]out X CO = 1 (E3) [CO]in Donde [CO]in y [CO]out son las concentraciones de CO a la entrada y a la salida del reactor. Un parmetro importante en la reaccin PROX es la selectividad, S, es decir, la proporcin de oxgeno que oxida el CO a CO2 frente al que se consume en la indeseada oxidacin del hidrgeno para formar agua. Este parmetro se evalu mediante la siguienteexpresin: S= 0.5 ([CO]in [CO]out ) [O2 ]in [O2 ]out (E4)

[O2]iny[O2]outsonlasconcentracionesdeoxgenoalaentradayalasalidadelreactor. La cantidad de oxgeno estequiomtrica necesaria para oxidar todo el CO a CO2 es la mitad en moles del CO alimentado. Sin embargo, dado que la reaccin simultnea de oxidacin del hidrgeno se produce en cierta extensin, conviene alimentar un exceso de oxgeno para poder obtener conversiones totales de CO. El exceso de oxgeno respectodelestequiomtricovienedefinidoporelparmetrodeoperacin: = 2

[O 2 ]in [CO]out

(E5)

En la mayora de los experimentos llevados a cabo para la realizacin de este trabajo, el parmetro adopt el valor de 2, es decir, se aliment el doble de la cantidad estequiomtricarequerida. Por otra parte, con el propsito de comparar los catalizadores preparados en nuestro laboratorio con los publicados en la literatura, se calcul su actividad cataltica asumiendo un mecanismo tipo LangmuirHinshelwood para la oxidacin de CO sobre catalizadores de cobre y cerio descrito por Liu y FlytzaniStephanopoulos [15], cuyas expresionesmatemticassemuestranacontinuacin:

35
m k LKLPCOPO 2

rCO =

1 + KLPCO

(E6) (E7) (E8)

k L = A L exp( Ea,L / RT )
K L = BL exp(Q / RT )

Donde PCO y PO2 son las presiones parciales de CO y O2; KL es la constante de equilibrio de adsorcin de CO y kL es la velocidad de reaccin superficial de CO, que se asume como la actividad cataltica intrnseca. Para el clculo de kL (molCOgcat1s1barm) se tomaron en consideracin los valores de m y KL determinados por Sedmak y col. [49] para catalizadores CuOx/CeO2 nanoestructurados (m=0.15; KL=7.53 exp[8700/RT] [=] bar1; R [=] Jmol1K1). Asumiendo reactor integral y flujo de pistn, se obtuvo la siguienteexpresinparakL:

F kL =

0 CO

w cat
m

0 0 K L PCO PO 2

X dX CO 0 P0 1 0.5 CO (1 X CO ) 0 PO 2

1 m 0 0 K L PO P 2 1 1 0.5 CO X CO (E9) + 0 0.5(1 m) P O 2

En esta ecuacin los superndices 0 indican las propiedades de la corriente de gas a la entrada del reactor; XCO es la conversin de CO a CO2; FCO es el flujo molar de CO (mols1)ywcateslamasadecatalizador(g). De modo anlogo, para comparar la selectividad de los catalizadores se defini el parmetro TS=0.8 como la temperatura a la cual la selectividad hacia la formacin de CO2 esmayoroigualque0.8.

37

ResumenartculoI. THESYNTHESISOFHIGHSURFACEAREACERIUMOXIDEANDCERIUM OXIDE/SILICANANOCOMPOSITESBYTHESILICAAQUAGELCONFINED COPRECIPITATIONTECHNIQUE

Sntesisdexidodecerioynanocompositesxidodecerio/slicede elevadasuperficieespecficamediantelatcnicadecoprecipitacin confinadaenhidrogeldeslice.
La obtencin de nanopartculas o nanoestructuras de elevada superficie especfica es objeto de una extensa investigacin. En el mbito de la catlisis es bien conocida la relacinentretamaodepartculayactividadcataltica,porloqueresultaindispensable sintetizar catalizadores con una elevada rea superficial activa. Con la intencin de obtener xidos metlicos de elevada rea superficial, en nuestro laboratorio se ha desarrollado recientemente la tcnica de coprecipitacin confinada en hidrogeles de slice (SACOP, silica aquagel confined coprecipitation), tcnica basada en la precipitacin de hidrxidos altamente dispersos en un hidrogel de slice, descrita en detalle en la parte Experimental. En el presente trabajo se han optimizado diversos parmetros del mtodo SACOP para la obtencin de xido de cerio con elevada superficie especfica. El CeO2 es un compuesto con un gran inters industrial debido a sus propiedades redox, su capacidad de almacenamiento de oxgeno y a que mejora la dispersin de los metales cuando se emplea como soporte de catalizadores. Debido a estas caractersticas en los ltimos aos ha sido extensamente investigado en mezclas conxidodecobreparalaoxidacinpreferencialdeCO. Dado que el mtodo SACOP es un procedimiento reciente resulta indispensable optimizar las distintas variables de sntesis, teniendo en cuenta los mecanismos de interaccin de los cationes metlicos con el hidrogel de slice. Estos mecanismos se han analizado comparando las caractersticas del xerogel de slice, los nanocomposites SiO2CeO2 y las nanoestructuras de CeO2. Se ha estudiado la influencia de los distintos precursores as como de las condiciones de proceso sobre la formacin de xido de cerio: (i) tipo de cido (HCl o HNO3) y relacin molar slice/cido (Si/H+); (ii) concentracin de cerio en la solucin precursora (Ce/Si); (iii) tiempo de envejecimiento del hidrogel y (iv) procedimiento de eliminacin del molde de slice. Las principalesconclusionesdeesteestudiohansidolassiguientes:

38

ArtculoI

Tipo de cido y relacin molar Si/H+: los xerogeles de slice preparados con HNO3 presentan reas superficiales ligeramente superiores a los preparados con HCl, aumentando dichas reas con el incremento de la relacin molar Si/H+. Las correspondientes nanoestructuras de CeO2 presentan elevadas reas superficiales (140200 m2g1) y tamaos de cristal de 23 nm. Sin embargo, las caractersticas texturalesdelaspartculasdeCeO2noreflejanlaestructuradesusxerogeleshomlogos debido a que la estructura de la slice colapsa durante el proceso de precipitacin. Para las siguientes etapas de la optimizacin del mtodo se escogi la relacin Si/H+=0.15 porquepermitelaobtencindelxerogelconelvolumendeporomayor,locualfavorece el empleo de mayores cargas de cerio durante la sntesis. Asimismo se opt por el uso deHNO3paraevitarposiblescontaminacionesportrazasdeproductosclorados. Relacin molar Ce/Si: el aumento de la cantidad de cerio en los composites da lugar a una disminucin del rea superficial y del volumen de poros de los mismos, disminuyendoasimismosutamao mediodeporo. Enel casodelasnanoestructurasde CeO2, el aumento de la cantidad de cerio desde una relacin molar Ce/Si de 0.25 hasta 1.00dalugaraunincrementodel tamaoefectivo departcula de4.6a6.4nm,aunque siemprepordebajodeltamaomediodeporodelasliceoriginal(11.4nm). Losvalores de superficie especfica de las muestras preparadas con una elevada concentracin de cerio siguen siendo muy altos (130180 m2g1). En este trabajo se propone un mecanismo de formacin de las nanopartculas en el que el proceso de precipitacin produce el colapso de la microestructura interna del hidrogel de slice sobre los hidrxidos metlicos, siendo este fenmeno el responsable de las variaciones texturales observadas. Tiempo de envejecimiento del hidrogel: el tiempo de envejecimiento tiene una escasa influenciasobreelreasuperficialdelosxerogelesdeslice.Encambiosqueseobserva cierta tendencia tanto en el volumen de poros como en el tamao medio de poro, que experimentan una disminucin a los 5 das para despus aumentar a los 20 das de tratamiento. Por lo observado, el envejecimiento tiene lugar de forma anloga a la descrita por Okkerse (Physical and chemical aspects of adsorbents and catalysts. Academic Press, Londres, 1970), en dos etapas: (1) reforzamiento de la estructura y (2) disolucinreprecipitacindelaspartculasmspequeasdeslicesobreotraspartculas deslice(mecanismoOstwaldripening). Los nanocomposites presentan valores del rea superficial muy similares para los distintos tiempos de envejecimiento (334360 m2g1). Se observa en dichos materiales ciertatendenciaporlacualelvolumenytamaomediodeporosaumentanligeramente

39 hasta llegar a un mximo para tiempos de envejecimiento en torno a 510 das, a partir del cual dichos parmetros comienzan a disminuir. Las tendencias observadas son consecuencia de complejas relaciones entre la estructura de gel envejecido antes de llevarseacabolaprecipitacinyladelcompositeunavezproducidadichaprecipitacin. Elresultadodeestasinteraccionesseexplicamedianteunacombinacindelmecanismo de envejecimiento del hidrogel de slice con el de disolucinreprecipitacin de la slice durante el proceso de precipitacin. Sin embargo, ninguno de estos mecanismos se ve reflejado en el tamao de cristal obtenido para el CeO2, lo que indica que las caractersticas finales de las nanoestructuras obtenidas estn principalmente influenciadasporlacargadecerioenlasolucinprecursorayelcolapsodelaestructura delaslicedurantelaprecipitacin. Eliminacin del molde de slice: todas las muestras de CeO2 presentan una cierta cantidad de slice residual (410% en peso). Durante la eliminacin de slice de los nanocomposites CeO2/SiO2 se encontr que el empleo de NaOH 3.5 M, en vez de 2 M o 5 M, mejoraba la eficacia del proceso. Por otro lado, se observ una proporcionalidad directa entre la cantidad de slice residual y el tamao de partcula. Las partculas ms pequeas (obtenidas con relaciones bajas Ce/Si) presentan una mayor relacin de superficie externa a volumen de partcula y por lo tanto la interaccin superficial de dichas partculas con la slice es mayor, dificultndose as su eliminacin. Se comprob que una relacin elevada entre el volumen de disolucin de NaOH y la masa de nanocomposite, as como un nmero adecuado de etapas de lavado permita alcanzar bajoscontenidosensliceresidual(4%enpeso). En conclusin, las nanoestructuras de CeO2 obtenidas mediante el mtodo SACOP presentan reas superficiales en el intervalo 140200 m2g1, con tamaos de partcula queoscilanentrelos46nm,dependiendodelascondicionesdepreparacin.Eltamao de partcula de CeO2 se puede controlar modificando la concentracin de cerio en la solucin precursora. Las elevadas reas superficiales obtenidas mediante el mtodo SACOPsedebenalcolapsodelaslicesobreloshidrxidosmetlicosdurantelaetapade precipitacin. Asimismo, el envejecimiento del hidrogel no produce variaciones significativaseneltamaofinaldelaspartculasdexidodecerio.

41

ArtculoI.

1.Introduction The preparation of high surface area metal oxide is currently the subject of extensive research due to their multiple applications in different areas (elec tronic devices, sensing, biomedicine, catalysis, etc.). In the field of catalysis, the wellknown relationship between particle size and catalytic activity [1] is evidence of the importance of develop ingtechniquestofabricatenanoparticles with tailored sizes. Furthermore, when conveniently immobilised in the appro priate supports [2, 3], catalyst nanopar ticles present clear advantages over conventional catalysts in terms of en

hanced active phase distribution [4] and catalyticactivity[5]. High surface area metal oxide can be obtained by various methods, including precipitation [69], hydrothermal syn thesis [1012], solvothermal synthesis [13], solgel [1416], microemulsion [17 22], template procedures [2326], pyro lysis [27], etc. These methods are espe cially suitable for the preparation of nanometric metal oxide particles (below ~100nm). Recently we developed a novel template procedure (silica aquagel confined co precipitation (SACOP)) to synthesise (mixed) metal oxides with surface area values that are

42 considerably higher than those obtained by most of the other techniques re ferred to above [28]. SACOP is a modi fied silicatemplate route that is based on the forced precipitation of metal hydroxides (MOH) in a silica aquagel medium. Subsequent thermal treatment of the dried composite (MOHSiO2) causes metal oxide nanoparticles to form in the silica mesopores. This method results in a better metal distri bution on the silica matrix than that achieved by the conventional hard template route and allows higher metal oxidetosilicamassratiostobeachieved in one single step. The final removal of the silica matrix produces either metal oxide nanoparticles or nanostructures, depending on the type of oxide used [28].AsSACOPisanovel procedure,it is necessarytoperformasystematicinves tigationofthesynthesisparametersthat influence the final surface area (or parti cle size) of the oxides. This is the aim of this work. This information may also serve to clarify the mechanisms that cause SACOP to produce nanostructures with a lower effective particle size than those obtained by the conventional sil icatemplateroute[29]. Out of the numerous metal oxides that could have been employed to perform this investigation, we selected cerium oxide due its unique redox and oxygen storage properties [30], which make it highly suitable for a variety of catalytic

ArtculoI applications. For example, ceria is an additive in automobile exhaust catalysts (three way catalysts) and is also com monly used as a support in a large num ber of different catalysts because of its abilitytofacilitatethedispersionofbase metals[31]. In this work we have analysed the effect of different synthesis parameters on the surfacearea(oreffectiveparticlesize)of cerium oxide nanostructures prepared by SACOP, and we have tried to clarify the synthesis mechanisms of the nanos tructures obtained. To this end we fo cused our attention on the following parameters: (i) the type of acid, (ii) the silica/acid molar ratio (Si/H+), (iii) the cerium concentration in the synthesis suspension (Ce/Si), (iv)the aquagel age ing time and (v) the silica removal pro cedure. 2.Experimental 2.1.Preparationofmaterials Cerium oxide was prepared by the silica aquagelconfined coprecipitation method as reported elsewhere [28]. Cerium nitrate hexahydrate was dis solved in an acid solution (HCl or HNO3 in water, 4 M) which was then added dropwise to the sodium silicate suspen sion under stirring (final Si/H2O molar ratio=0.005).Hydrolysiswascarriedout at RT for 24 h in an autoclave, and then

Lpezetal.Microp.Mesop.Mat127(2010)198204

43

the solution was heated in an oven at 100C for 24 hours (condensation step). After cooling, the metallic cations in the gel were subjected to precipitation with NH3 vapour. During this process pH in creases from very acid values (pH~0) to pH=8. The so formed cerium hydroxide is not dissolved during the subsequent washing steps. The slurry was repeat edlywashedwithwateranddispersedin acetone. The sample was then filtered, dried at RT and calcined at 550C (1Cmin1). To analyse the effect of the synthesis process on the final properties of the nanostructures, the following parameters were taken into account: (i)the acidity of the medium: Si/H+ mo lar ratio = 1.00, 0.33 and 0.15; (ii) the amount of cerium: Ce/Si molar ratio = 0.25, 0.50 and 1.00; (iii) the aquagel ageing time (030 days) and (iv) theprocedure used to remove the silica matrix. The initial procedure for remov ingthesilicamatrixwasbydissolutionin aqueous NaOH (2 M) at ~30C for two days. The effect of the metallic ions on the gel was indirectly evaluated by comparing the nanocomposites (SiO2CeO2) and nanoparticles(CeO2)withthesamplesof ceriumfree silica xerogels (SiO2). These samples were prepared like the nano composites, except that there was no cerium nitrate in the precursor solution, which eliminated the need for the pre cipitationstep.

2.2.Characterisationofmaterials The nitrogen adsorption and desorption isothermswereevaluatedat196Cona Micromeritics ASAP 2010 volumetric adsorption system. The BET surface area was inferred from the isotherm analysis in the relative pressure range of 0.04 to 0.20. The Pore Size Distributions (PSD) were calculated by means of the Kruk JaroniecSayari method [32], and char acterised by the average pore size (KJS). An effective particle size for the result ing particles was evaluated from the values of the BET surface area and the true density of the oxides by assuming spherical geometry (dBET). The metal distributions and silica contents of the samples were studied by means of a Scanning Electron Microscope (Mod. DSM 942, Zeiss) and an attached Energy Dispersive X Ray Detector (EDX). Xray diffraction(XRD)patternswereobtained inthewideanglerange(2=1090)ona Siemens D5000 instrument operating at 40 kV and 20 mA, using Cu K radiation (=0.15406 nm). The crystal size values were estimated from the XRD patterns byapplyingScherrersequation(dXRD).In the case of the XRD spectra all the re ported peak areas, positions and widths were the result of GaussianLorentzian fittings. The structure of the prepared oxides was characterised by Transmis sion Electron Microscopy (TEMJEOL 1010).

44 3.Resultsanddiscussion 3.1. Acid type and concentration in the precursorsuspension Samples of ceriumfree silica xerogels and cerium oxide were obtained by us ingtwodifferentacids,HClandHNO3,at different concentrations in the synthesis suspension (Si/H+ = 1.00, 0.33 and 0.15). The structural characteristics of these samplesaresummarizedinTable1. The porous structure of the silica xe rogels is clearly influenced by the Si/H+ molar ratio used during their synthesis [33] as is shown in Table 1. It can be seen that the xerogels prepared with

ArtculoI HNO3 have slightly higher BET surface areas than those obtained with HCl. This demonstrates the feasibility of synthe sizing the gel with HNO3 and avoids pos sible chlorine contamination arising from the use of HCl. As already reported for similar materials [33], the BET sur face area values of the ceriumfree xe rogels increase with the Si/H+ molar ratio. For all the silica samples these values are higher than 500m2g1. How ever, although the CeO2 oxide samples present elevated surface areas (140 200m2g1) their dimensions do not re flect the pore structure of their cerium freesilicacounterparts.

Table 1. Structural characteristics of calcined ceriumfree silica samples and CeO2 sam ples(Ce/Simolarratio=0.25). Acid Si/H+ Ceriumfreesilicaxerogel CeO2

SBET Vp (m2/g) (cm3/g) 1.00 707 0.60 HNO3 0.33 650 0.93 0.15 517 1.10 1.00 617 0.60 HCl 0.33 606 0.88 0.15 507 1.11 Fig. 1 shows a TEM image of a SiO2CeO2 nanocomposite, with agglomerates of CeO2 nanoparticles of <10 nm, embed ded in a silica frame. An XRD spectrum of a sample of CeO2 nanoparticles (HNO3, Si/H+=0.15, Ce/Si=0.25) is shown in Fig. 2. Applying Scherrers equation to

SBET dBET dXRD SiO2 KJS (m2/g) (nm) (nm) (wt.%) (nm) 3.0 173 4.9 2.0 7.6 6.6 158 5.3 2.4 8.2 11.4 181 4.6 2.5 6.8 2.8 140 6.0 2.5 6.5 5.9 170 4.9 2.5 4.7 10.1 198 4.2 2.8 7.3 this spectrum yields a particle size of 23nm. For all the samples described in Table 1, the particle size values evalu ated from the BET surface areas (dBET = 46 nm) were in all cases higher than those estimated by XRD analysis (dXRD = 23nm).

Lpezetal.Microp.Mesop.Mat127(2010)198204

45

20 nm
Fig.1.TEMimageofaSiO2CeO2nano composite(HNO3,Si/H+=0.15, Ce/Si=0.25). This finding suggests that the CeO2 sam ples are formed by aggregates of 2 6crystals. As already mentioned, the crystal/particle size of the CeO2 samples does not seem to be influenced by the structural characteristics of the equiva

lent(insynthesisconditions)ceriumfree silica xerogel, as can be seen in Table1. Thus, when the suspension acidity is increased the average pore size of the xerogelgrowsfrom3to11nm,whereas the dBET values for the cerium oxide samples are more or less randomly dis tributed in the 46 nm range. This may be attributed to one of two causes: (1) the silica structure collapsed onto the cerium hydroxide nanoparticles during precipitation; (2) the cerium hydroxide only filled the smaller silica pores (theo retically, for a Ce/Si molar ratio of 0.25 such as that used for the samples de scribed in Table1, the cerium oxide should occupy between 9% and 17% of the total pore volume of silica, depend ing on the Si/H+ ratio). As will be shown in next section, the first assessment must be taken as being the correct one.

(111) (220)

Intensity (a.u.)

(311) (200)

10

20

30

40 2

50

60

70

Fig.2.XRDpatternofaCeO2sample(HNO3,Si/H+=0.15,Ce/Si=0.25).

46 A Si/H+ molar ratio of 0.15 produces samples of ceriumfree silica xerogel with the highest pore volume values. This silica/acid ratio was therefore cho sen for the next stage of the optimisa tion of the SACOP method. During this stage the effect of increasing the cerium load on the final properties of the ce rium oxide was analysed. HNO3 was used rather than HCl in order to avoid potentialcontaminationbychlorides. 3.2. Cerium concentration in the precur sormixture(Ce/Simolarratio) The samples were prepared using differ ent Ce/Si molar ratios for the precursor mixture (0.25, 0.50 and 1.00) in order to analyse the influence of this parameter on the structural characteristics of the nanocomposites and cerium oxide sam ples. The N2 adsorption isotherms of the SiO2CeO2 nanocomposites are shown in Fig. 3 and the values of their structural characteristicsarepresentedinTable2. The nanocomposites display a type IV isotherm, corresponding to meso porous materials with quite a uniform pore size in the 1216 nm range (Ta ble2). As was expected, the samples obtained by increasing the Ce/Si molar ratios show decreasing values for sur face area, pore volume and average pore size. The average pore sizes of the three nanocomposites are higher than those of the equivalent ceriumfree sil ica xerogel (12.515.2 nm for SiO2CeO2

ArtculoI nanocomposites versus 11.4 nm for the silica xerogel). This apparently contra dictory result can be explained by the schematic tentative mechanism pro posedinFig.4.
1.00 0.75 0.50 0.25 0.00 1
400 Adsorbed volume (cm STP/g) 300 200 100 0 0.0
Ce/Si=1.00 Ce/Si=0.50 Ce/Si=0.25

dV/dlogD (cm /g)

Ce/Si=1.00 Ce/Si=0.50 Ce/Si=0.25

10

100

1000

Pore size (nm)

0.2

0.4

0.6

0.8

1.0

Relative pressure (p/p0)

Fig.3. (a) N2 adsorption isotherms and (b) pore size distribution plots of the SiO2CeO2 nanocomposites, obtained for different cerium loads (HNO3, Si/H+=0.15). It was observed that the precipitation process caused the collapse of the mac roscopic structure of the aquagel, which turned into a slurry. In the mechanism

Lpezetal.Microp.Mesop.Mat127(2010)198204

47

proposed,theporousmicrostructures of the slurry resemble those of the original aquagel. The smallest silica particles of these microstructures (represented by the different images in Fig. 4) tend to dissolve in the basic medium, as the pH is increased during the precipitation process, and then reprecipitate onto the cerium hydroxide particles. This is at tributed to flocculation of the reprecipi tated silica aggregates caused by the ceriumcationsbeingadsorbedonsilanol groups,inaccordancewithamechanism in which these cations neutralise the negative charges on the surface of the silica aggregates [34]. The collapse of the internal silica microstructure on the surface of the cerium hydroxide parti cles helps to explain two former obser vations: (a)the average pore size of the nanocomposites is higher than that of the ceriumfree silica xerogel (as can be inferred from Fig. 4) and (b) the values

of surface area obtained by the SACOP method are higher than those of the conventional silicabased hardtemplate methods, where silica collapse does not occur. Needless to say the effective par ticle size of CeO2 increases with the in crease in cerium load as a result of which the average nanocomposite pore size decreases, as illustrated in Fig.4. Thus, a higher cerium concentration in the precursor mixture produces an in crease in the final crystal (particle) size of CeO2, from 2.5 (4.6) nm for Ce/Si=0.25 to 4.4 (6.4) nm for Ce/Si=1.00, and a concomitant decrease in the specific surface area, from 181m2g1 to 132 m2g1. Fig. 5 shows a TEM image of the resulting CeO2 aggre gates (Ce/Si=1.00) after the silica is re moved. It can be seen that the remain ing sample is composed of nanoparticles ofaround5nm.

Table 2. Structural characteristics of synthesised SiO2CeO2 nanocomposites and CeO2 samplespreparedwithdifferentCe/Siratios(Si/H+=0.15). Ce/Si (molar) 0.25 0.50 1.00 SiO2CeO2nanocomposites SBET SiO2 Vp KJS (m2/g) (cm3/g) (nm) (wt.%) 350 1.01 15.2 63.6(a) 58.2(b) 285 0.58 13.9 43.1(a) 42.6(b) 209 0.40 12.5 28.4(a) 23.9(b) CeO2 SBET dBET dXRD SiO2(b) (m2/g) (nm) (nm) (wt.%) 181 4.6 2.5 6.8 160 5.3 3.5 6.4 132 6.4 4.4 4.3

(a)TheoreticalvalueobtainedfromtheinitialCe/Simolarratio. (b)DatafromEDXanalyses.

48

ArtculoI

Fig.4. Proposed mechanism for the precipitation process of nanocomposites for differ entCe/Simolarratios.

3.3.Aquagelageingtime In the case of the ceriumfree silica aquagels, the ageing time is the period between the end of the condensation step at 100C and the point at in which the gel is washed with acetone. This period varied between 0 and 20 days. The variation of this parameter did not have any influence on the BET surface area of the ceriumfree xerogels, as can be appreciated from the data summa rized in Table 3. However it did have a noticeable effect on the values of the pore volume and average pore size; i.e. a decrease after 5 days of ageing time

20 nm
Fig. 5. TEMimageofaCeO2samplewith a residual SiO2 content of 4.3wt.% (HNO3,Si/H+=0.15,Ce/Si=1.00).

Lpezetal.Microp.Mesop.Mat127(2010)198204

49

and subsequent increase after 20 days. This trend apparently differs from what is reported in the literature [3537]; i.e. the silica pore volume increases in the first hours of ageing and then decreases or becomes stable. We attribute this discrepancy to the less reactive condi tions under which the gel is aged in this work (a very acid mother liquor and room temperature) as compared to the conditionsusedin theliterature(alcohol orammoniaand/orhighertemperatures are employed), which may have the effect of accelerating the dissolution precipitation processes. A tentative age ing mechanism for the ceriumfree silica aquagels is schematized in Fig. 6a. Here ageing takes place in a two step process similar to that described by Okkerse [38]. During the first 5 days, the gel structure is reinforced through a reac tion between the smallest (more reac tive) silica aggregates in the silica micro structures (Fig. 6a), as a result of which the gel structure contracts (the average pore size and total pore volume de crease). After this period, the gradual increase in the amount of silica in the solution (mainly as a result of the disso lution of the smallest silica particles) ends with the silica reprecipitating onto the larger aggregates [39], via Ostwald

ripening. This process allows the ap proximate recovery of the original struc tureofthesilicaaquagel(Fig.6a). In the case of the SiO2CeO2 aquagel, ageing is the room temperature stage between the end of the condensation step at 100C and the precipitation of the cerium hydroxide. The ageing time variesbetween0and30days.Thestruc tural characteristics of the resulting SiO2CeO2nanocompositesand theCeO2 samples are presented in Table 3. It can be observed that the nanocomposites showlargesurfaceareaslikethecerium free silica xerogels, are not significantly affected by the ageing process. On the other hand, the average pore size of the nanocomposites increases after 5 10days and decreases after longer peri ods. This behaviour is the opposite to that observed for the ceriumfree silica xerogels (Table 3 and Fig. 6a) and is ten tatively attributed to a combination of mechanisms occurring during the ageing (Fig. 6a) and precipitation steps (Fig.4). Thus, ceriumcontaining aquagels with microstructures such as those depicted in Fig. 6a (for ceriumfree aquagels) can be expected to yield precipitated nano composites with microstructures similar tothoseshowninFig.6b.

50

ArtculoI

Fig.6. (a) Proposed mechanism of the ageing process for ceriumfree silica aquagels. (b)Tentative structure of the nanocomposites after the precipitation of cerium containingsilicaaquagelsagedfordifferentperiodsoftime. In the proposed combined mechanism, the ceriumcontaining silica aquagel after 5 days of ageing has a reinforced structure due to the formation of silox ane bonds between the smallest and most reactive silica particles. The inner wallsarethereforethinnerthanthoseof the nonaged microstructure and the aquagelafter20daysofageing,atwhich point the walls thicken due to the Ost wald ripening phenomenon. Conse quently, the inner walls of the 5dayold aquagelaremoreeasilydissolvedduring precipitationduetoselectiveetching.As a result the silica is reorganised (repre cipitation) around the cerium hydroxide nanoparticles. This gives rise to a higher average pore size for the precipitated nanocomposite than in the less and more aged samples (Fig. 6b). However none of these behaviours is reflected in the final CeO2 particle size, which is quite similar for all the samples (Table 3), thus confirming that the particle size obtained by SACOP is mainly influenced by the cerium load and silica collapse duringprecipitation.

Lpezetal.Microp.Mesop.Mat127(2010)198204

51

Table 3. Textural characteristics of the ceriumfree silica xerogels (not calcined) and structural characteristics of the nanocomposites and CeO2 samples prepared for differ entageingtimes(Ce/Si=0.25,Si/H+=0.15). Ceriumfreesilicaxerogel SBET Vp 2 (m /g) (cm3/g) 604 1.37 604 1.14 598 1.45 SiO2CeO2nanocomposites Ageingperiod SBET Vp (days) (m2/g) (cm3/g) 0 350 1.01 1 339 0.92 5 339 1.13 10 334 1.09 20 339 0.87 30 360 0.90 CeO2 Ageingperiod SBET dBET (days) (m2/g) (nm) 0 181 4.6 1 194 4.3 5 176 4.8 10 181 4.6 20 173 4.9 30 177 4.8 (a)Valuesobtainedfrommassbalances. (b)DatafromEDXanalyses. Ageingperiod (days) 0 5 20 3.4.Silicamatrixremoval A certain amount of residual silica (5 10wt.%) remains in the ceria samples after the removal of the silica (NaOH, 2M, ~30C, 2days, Tables 13). This is a detrimental factor for the application of silica template techniques. It was found that a NaOH concentration of 3.5 M was more effective than lower or higher NaOH concentrations for removing silica (Table 4). From the same table it can also be appreciated that the extent of silica removal appears to be related to thesizeof theceriaparticle.Thesmaller particles (obtained by using lower Ce/Si ratios) present higher Sext/Vpaspect ra tios so that the surface interaction be tween the CeO2 nanoparticles and the KJS (nm) 12.5 9.0 15.7 KJS (nm) 15.2 15.5 21.1 21.2 15.4 15.2 dXRD (nm) 2.5 2.4 2.8 2.6 2.3 2.1 SiO2(a) (wt.%) 63.6 63.7 64.1 59.7 60.4 59.3 SiO2(b) (wt.%) 6.8 8.3 7.9 9.2 7.7 8.8

52 silicaisstronger,makingitmoredifficult toremovethesilica. Table 4. Structural characteristics of the samples washed with different NaOH solutions(Si/H+=0.15). Ce/Si 0.25 0.25 0.25 0.50 1.00 SBET SiO2 dXRD NaOH 2 (M) (m /g) (wt.%) (nm) 5 217 8.3 2.5 3.5 216 7.8 2.8 2 192 14.1 2.3 2 160 7.7 3.5 2 132 5.5 4.4

ArtculoI 3.5M. These treatments are summa rized in Fig. 7, where the residual SiO2 content is plotted against the W0/V ra tio, W0 being the initial weight of the nanocomposite (g) and V the volume of NaOH dissolution (mL). As can be ob served from this figure, neither stirring nor temperature has any significant influence on the final SiO2 content. The figure shows that the main controlling factors in the stage of silica removal are the W0/V ratio and the number of wash ing steps. Very low SiO2 contents (~4 %) can be achieved by using low W0/V ra tios and three short washing steps in NaOH3.5M.

Modifications to temperature, stirring and the number and duration of the washing steps were tested for washing the samples (Ce/Si=0.25) using NaOH
20

S-84C-6h

Residual SiO2 content (wt%)

Residual SiO 2 content (wt%)

16
S-84C-18h 1 washing

12

WS-100C16.5h

S-20C-18h S-20C-2h

8
S-84C-6h 3 washings (2 hours each)

4
S-20C-2h

S: stirring WS: without stirring

0 0.000

0.004

0.008

0.012

0.016

0.020

Wo/V (g/mL)

Fig. 7. Residual SiO2 content (wt.%) versus W0/V ratio for the different silica removal treatments(sampleswithCe/Si=0.25).

Lpezetal.Microp.Mesop.Mat127(2010)198204

53

4. Conclusions The silica aquagel confined co precipitation (SACOP) technique is able to produce CeO2 nanostructures with crystal(particle)sizesinthe25(46)nm range, and very high surface area values (140200 m2g1). The effective particle size of CeO2 can be controlled by modi fying the cerium load in the synthesis suspension, and is independent of the textural properties of the equivalent ceriumfree silica xerogel. It was found thattheSACOPtechniqueproducesvery high surface areas because of a silica flocculation mechanism around the ce rium hydroxide nanoparticles during the precipitation stage. Ageing of the silica

aquagel does not produce any signifi cant variation in the final ceria particle size. Very low residual silica contents can be achieved by reducing the ratio of nano composite weight to NaOH solution (3.5M)volume. Acknowledgments Funding through the Spanish National Project MAT200800407 and the FICYT Regional Project (IB05001) is acknowl edged. IL thanks the Ministerio de Cien cia e Innovacin of Spain for her FPI grant and TVS thanks the Junta General del Principado de Asturias for the award ofaClarnpostdoctoralcontract.

References [1] G. Evans, I. V. Kozhevnikov, E. F. Kozhevnikova, J. B. Claridge, R. Vaidhyanathan, C. Dickinson, C. D. Wood, A. I. Cooper, M. J. Rosseinsky, J. Mater. Chem. 18 (2008)5518. [2] T.Sanders,P.Papas,G.t.Veser,Chem.Eng.J.142(2008)122. [3] H.Koga,T.Kitaoka,H.Wariishi,Chem.Comm.(2008)5616. [4] R. Rinaldi, A. d. M. Porcari, T. C. R. Rocha, W. H. Cassinelli, R. U. Ribeiro, J. M. C. Bueno,D.Zanchet,J.Mol.Catal.AChem.301(2009)11. [5] J.M.Nadgeri,M.M.Telkar,C.V.Rode,Cat.Comm.9(2008)441. [6] C. Delacourt, P. Poizot, M. Morcrette, J. M. Tarascon, C. Masquelier, Chem. Ma ter.16(2004)93. [7] Y.I.Kim,D.Kim,C.S.Lee,PhysicaB337(2003)42. [8] R. Takahashi, S. Sato, T. Sodesawa, N. Nakamura, S. Tomiyama, T. Kosugi, S. Yo shida,J.Nanosci.Nanotechno.1(2001)169. [9] M.S.Tsai,Mat.Sci.Eng.B110(2004)132.

[10] M.Zawadzki,J.AlloyCompd.454(2008)347.

54 [11] Q.Liu,H.Liu,Y.Liang,Z.Xu,G.Yin,Mater.Res.Bull.41(2006)697.

ArtculoI

[12] D. H. Chen, D. R. Chen, X. L. Jiao, Y. T. Zhao, M. S. He, Powder Technol. 133 (2003)247. [13] D.S.Bae,B.Lim,B.I.Kim,K.S.Han,Mater.Lett.56(2002)610. [14] B.Ksapabutr,E.Gulari,S.Wongkasemjit,Mater.Chem.Phys.99(2006)318. [15] X.F.Dong,H.B.Zou,W.M.Lin,Int.J.HydrogenEnergy31(2006)2337. [16] X.Zheng,X.Zhang,Z.Fang,X.Wang,S.Wang,S.Wu,Cat.Comm.7(2006)701. [17] J.S.Lee,J.S.Lee,S.C.Choi,Mater.Lett.59(2005)395. [18] J.Zhang,X.Ju,Z.Y.Wu,T.Liu,T.D.Hu,Y.N.Xie,Chem.Mater13(2001)4192. [19] N. Laosiripojana, W. Sutthisripok, S. Assabumrungrat, Chem. Eng. J. 127 (2007) 31. [20] J.Rebellato,M.M.Natile,A.Glisenti,Appl.Catal.A:Gen.339(2008)108. [21] I. Ritzkopf,S. Vukojevic,C. Weidenthaler,J.D. Grunwaldt,F.Schuth,Appl.Catal. A:Gen.302(2006)215. [22] R.D.K.Misra,A.Kale,R.S.Srivastava,O.N.Senkov,Mater.Sci.Techn.19(2003) 826. [23] S.C.Laha,R.Ryoo,Chem.Comm.(2003)2138. [24] E. Rossinyol, J. Arbiol, F. Peiro, A. Cornet, J. R. Morante, B. Tian, T. Bo, D. Zhao, SensorActuat.BChem.109(2005)57. [25] W. H. Shen, X. P. Dong, Y. F. Zhu, H. R. Chen, J. L. Shi, Microp. Mesop. Mat. 85 (2005)157. [26] J. Roggenbuck, H. Schafer, T. Tsoncheva, C. Minchev, J. Hanss, M. Tiemann, Mi crop.Mesop.Mat.101(2007)335. [27] J.D.Hu,Y.X.Li,X.Z.Zhou,M.X.Cai,Mater.Lett.61(2007)4989. [28] G.Marbn,A.B.Fuertes,T.ValdsSols,Microp.Mesop.Mat.112(2008)291. [29] T.ValdsSols,G.Marban,A.B.Fuertes,Chem.Mater.17(2005)1919. [30] B. M. Reddy, G. Thrimurthulu, P. Saikia, P. Bharali, J. Mol. Catal. AChem. 275 (2007)167. [31] X.C.Zheng,X.L.Zhang,X. Y.Wang, S.H.Wu,React. Kinet.Catal. Lett.92(2007) 195. [32] M.Kruk,M.Jaroniec,A.Sayari,Langmuir13(1997)6267. [33] A.B.Fuertes,Chem.Mater.16(2004)449. [34] R.K.Iler,Thechemistryofsilica.JohnWileyandsons,NewYork,1979,p.622.

Lpezetal.Microp.Mesop.Mat127(2010)198204

55

[35] A.Fidalgo,L.M.Ilharco,Microp.Mesop.Mat.84(2005)229. [36] S. Smitha, P. Shajesh, P. R. Aravind, S. Rajesh Kumar, P. Krishna Pillai, K. G. K. Warrier,Microp.Mesop.Mat.91(2006)286. [37] J. Estella, J. C. Echeverria, M. Laguna, J. J. Garrido, Microp. Mesop. Mat. 102 (2007)274. [38] C.Okkerse, Physical and chemical aspects of adsorbents and catalysts. Academic Press,Londres,1970. [39] C.J.Brinker, G.W.Scherer, Solgel science: the physics and chemistry of solgel processing.AcademicPress,Boston,1990.

57

ResumenartculoII. ANATTEMPTTORANKCOPPERBASEDCATALYSTSUSEDINTHECO PROXREACTION

Unintentodeclasificacindecatalizadoresdecobreutilizadosenla reaccinCOPROX
En la literatura se describen un gran nmero de catalizadores para la oxidacin preferencialdeCOquesonactivosenunampliointervalodetemperaturas(100250C). Estos catalizadores pueden clasificarse en cuatro grupos: (i) metales nobles (p. ej. Au, Pd, Pt), (ii) perovskitas (p. ej. LaFeO3), (iii) espinelas (p. ej. CuMn2O4), y (iv) dispersiones slidas/soluciones de xidos de metales de transicin (CuOx/CeO2). De todos ellos, los catalizadores que incluyen cobre en su formulacin se revelan como los de mayor inters por su elevada actividad y bajo coste. Sin embargo, la disparidad de condiciones experimentales en las cuales se han analizado los catalizadores basados en cobre (distintas velocidades espaciales), hace muy difcil la comparacin de su actividad cataltica. Por esta razn, el objetivo del presente artculo consiste en recopilar la informacin sobre los catalizadores basados en cobre que han sido descritos en la bibliografa e intentar ordenarlos en funcin de su actividad cataltica en la reaccin PROX. A partir del mecanismo de tipo LangmuirHinshelwood, propuesto por Liu y FlytzaniStephanopoulos (J. Catal. 153 (1995) 317) para la oxidacin de CO sobre catalizadores de CuOx/CeO2, es posible deducir una ecuacin cintica que permite calcular la actividad cataltica, considerando que la constante kL100 representa la actividad cataltica intrnseca a 100C. Slo en unos pocos de los artculos considerados lafaltadedatosimpidicalcularlaconstantedeactividadcataltica. Adems de la actividad cataltica, se tabul la selectividad (S) de los catalizadores hacia la oxidacin de CO. Para ello se emple el parmetro TS=0.8, definido como la temperaturaalacuallaselectividadesmayorde0.8(S1.0). Los catalizadores analizados se dividieron en cuatro grupos segn su composicin: (i)cobrexidodecerio,(ii)cobrexidodeceriosoportadosenalmina,(iii)cobrexido de cerio con la adicin de modificadores y (iv) catalizadores basados en cobre sin xido decerio. Analizada la actividad cataltica de los cuatro grupos de catalizadores se puede concluir que tanto el empleo de soportes de almina como la adicin de modificadores (Zr, Sm,

58

ArtculoII

Sn, Co o Fe) o la eliminacin del xido de cerio de la formulacin del catalizador no favorecen su actuacin en la reaccin PROX, sino que ms bien disminuyen la actividad deloscatalizadores. Una vez comprobado que los mejores catalizadores eran los de cobrexido de cerio, se estudi el efecto de diversos parmetros que a priori deberan afectar a la actividad cataltica, como son el rea superficial especfica, el tamao de partcula y el contenido en cobre. Ninguno de estos tres parmetros muestra una correlacin con la actividad cataltica,comotampocopareceafectarlatemperaturadecalcinacindelcatalizador. El mejor catalizador encontrado en la bibliografa fue el preparado por Liu y col. (J. Mol. Catal. AChem. 267 (2007) 137) mediante el mtodo solgel con bromuro de cetiltrimetilamonio, seguido del sintetizado por Marbn y Fuertes (Appl. Catal. B: Environ. 57 (2005) 43) mediante citratos. Cabe destacar que Liu y col. (J. Mol. Catal. A Chem. 255 (2006) 103) obtienen un catalizador prcticamente igual al mejor con el mtodo de coprecipitacin. Por lo tanto, no se pueden extraer conclusiones generales sobrelainfluenciadelmtododeobtencinenlaactividadcataltica. Al analizar la selectividad de estos catalizadores hacia la oxidacin de CO se obtienen resultados similares, sin que se observen tendencias claras. El catalizador ms selectivo de los analizados, en el mayor intervalo de temperaturas, es el descrito por Avgouropoulosycol.(Catal.Today75(2002)157). Puesto que el modo de preparacin y las caractersticas de los catalizadores no se correlacionaron con el grado de actividad cataltica, sta ha de estar influida por las condiciones experimentales. Los datos disponibles no siempre son suficientes para determinar la velocidad espacial en el sistema. Es por ello que en la mayora de los artculos es imposible discernir si los valores de actividad cataltica estn siendo enmascarados por la resistencia a la difusin externa o por la existencia de dispersin axial, para lo que necesitaramos conocer al menos el tamao de partcula del catalizador,lasdimensionesdelreactorylapresencia/ausenciadediluyentesinertes. Uncatalizador,ademsdeactivoyselectivo,hadeserestableduranteeltiempodeuso. En la mayora de los artculos analizados no se incluyeron ensayos de estabilidad. En los trabajos restantes, los experimentos de estabilidad se realizaron en un intervalo de temperaturascomprendidoentre80175Cyconduracionesentre12y200h.Conviene destacar el hecho de que estos ensayos de estabilidad fueron llevados a cabo en condiciones experimentales (masa de catalizador, temperatura, flujo total de gases y presiones parciales) que permitieron en casi todos los casos alcanzar conversiones

59 superiores al 90%. Al realizar los experimentos en condiciones de conversin total, la desactivacin puede no verse reflejada en los valores de conversin durante un largo periodo de tiempo pese a estar sucediendo. De este modo los catalizadores probados podrannosertanestablescomoseconcluyeenlostrabajos. Enconclusin,delos catalizadoresbasadosen cobreanalizados eneste trabajo,losms activosresultaronserlosdecobreyxidodecerio.Laadicindemodificadoresoeluso de soportes en estos materiales no supusieron una mejora en la actividad cataltica. Los mejores catalizadores se obtuvieron mediante el empleo de agentes quelantes como cidoctricoobromurodecetiltrimetilamonio.

61

ArtculoII.

In future vehicles powered by low tem perature H2fed fuelcells, hydrogen must be supplied with less than 100 ppm of CO, due to its deactivating effect on the fuel cell electrocatalysts [1]. Two major solutions for feeding the H2 fuel are envisaged: (i) Onboard storage of ultrapure H2 and (ii)Onboard H2 produc tion from hydrocarbons. The second option offers several advantages from thepointofviewoffueldistributionand supply, safety and cost, especially when biomethanol is the source of H2 via an onboard steam reforming process [2]. However, the production of H2 from

hydrocarbons results in the undesired generation of CO as a byproduct and consequentlyapurificationstepmustbe introduced prior to the fuel cell stage. The purification process usually consists of a watergas shift reaction (required only for a high CO concentration) fol lowed by preferential CO oxidation (PROX) [3], a process by which the re sidual CO is catalytically oxidised to CO2 while the simultaneous oxidation of H2 to water is minimised. There are basi cally two approaches to PROX catalysts. Initially, noble metal catalysts (Au, Ag, Pt, Ru, etc.) were employed as in the

62 case of the catalysts developed for CO purification in the ammonia synthesis process [4]. Later on it was observed that cheaper copperbased catalysts appeared to be equally active and much more selective [5] given the high tem peratures required for the onboard process between the H2 production stage (250C for biomethanol steam reforming) and the fuel cell stage (70100C). Different compositions of copperbased catalysts have been pre pared and tested for producing the PROX reaction in the 100250C tem perature range, and have been reported in a considerable number of papers. However, other testing conditions re flect a certain degree of diversity be tween different research works, espe cially with respect to gas flow rate and amount of catalyst (spatial velocity), thus making it difficult to compare the reported catalytic activities (usually ex pressed in standard CO conversion val ues). Consequently there is only a lim itednumberofarticlesthatprovidesuch a comparison [6, 7]. In the majority of works the catalytic activities of a num ber of different catalysts prepared by the same authors and tested under identical conditions are compared [8 10]. The main reason for the lack of comparative works might be the diffi culty in finding a common basis of com parison regarding the activity, selectivity and stability of the copperbased cata

ArtculoII lysts. In addition, in many cases it is dif ficult to perform a comparison from the literature in which the experimental data provided are clearly insufficient (lack of data such as flow velocity, cata lyst mass, etc). Finally, the stability of the catalyst, a very important parame ter, apart from being rarely reported, is usually tested at conditions in which complete conversion is achieved. As a consequence the reaction rate and its real variation trend cannot be extracted fromtheinformationprovided. In this work we present a critical com parison of the performance of the ma jority of those copperbased catalysts whose activities in the PROX process have been reported up to date in an attempt to situate each catalyst in its proper place. To achieve this the first task is to find a common basis of com parison. Theactivityofthecatalystscanbecalcu lated by means of the Langmuir Hinshelwood mechanism used by Liu and FlytzaniStephanopoulos [11] (LF mechanism) for CO oxidation over cop perceria catalysts. This is described by thefollowingreactionrateexpression:

rCO =

m k L K L PCOPO 2

1 + K L PCO

(R1)
(R2) (R3)

k L = A L exp( E a,L / RT )

K L = B L exp(Q / RT )

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

63

where PCO and PO2 are CO and O2 partial pressures and KL and kL can be taken as the CO adsorption equilibrium and sur face reaction rate constants, respec tively. In this work it is assumed that the constant kL (molCOg1cats1barm) repre sents the intrinsic catalytic activity. To calculate this constant we considered the values of m and KL reported by Sed mak et al. [12] for a nanostructured CuOx/CeO2 catalyst (m=0.15; KL=7.53 exp[8700/RT][=]bar1;R[=]Jmol1K1).In the work by Sedmak et al. [12], a rate expression derived from a Marsvan Krevelen (MVK) mechanism was also analysed and was found to fit the reac tion rate values for the CuOx/CeO2 cata lysts equally well. However, when we applied this expression in the present work we found many inconsistencies in
0 FCO

the evaluated reaction rates for some of the reviewed catalysts, including nega tive values. Added to this, in a recent study by Vannice [13] the original ex pression of the MVK mechanism for heterogeneous oxidation reactions was found to be wrong. Therefore we se lected the LF expression (R1) in order to estimate the normalised values of the reaction rates (kL). Although the expres sion R1 was originally used to fit reac tion rate data obtained at differential conditions[11,12],itwasalsoprovento be valid at high conversion values (inte gralreactor)[11]. The surface reaction rate kL can be then calculated by assuming the integral re actor and plug flow behaviour (an as sumption taken for granted in all pub lishedworks)asfollows:

kL =

w cat
m

0 0 K L PCO PO 2

X dX 0 0 PCO ( 1 X ) 1 0 . 5 0 PO 2

0 1 1 0.5 PCO 0 0.5(1 m) PO 2

0 K L PO 2

1m X

(R4)

In this equation superscript 0 represents the properties of the gas stream enter ingthereactor,XsignifiesCOconversion to CO2, FCO is the molar flow of CO (mol/s) and wcat is the catalyst mass (g). We explored the current literature on copperbased catalysts for PROX, and selected the papers that afford kinetic

data to solve equation R4. Around 40 catalysts from a similar number of pa pers are compared in this work. In some caseskLcouldnotbeevaluatedbyequa tion R4 due to lack of data, such as cata lyst mass or total flow (used to calculate FCO) [1416]. In those works where dif ferent catalysts were tested, only the

64 most active catalysts were generally considered. In the PROX reaction, CO competes with H2fortheavailableoxygen.Theselectiv ity (S) of the catalysts towards CO oxida tion is an important parameter, but no kinetic data of H2 oxidation in the pres ence of CO are available, and therefore comparison of the selectivity values must be performed in a semi quantitative way. A review of the PROX literature reveals that in most catalysts the selectivity towards CO oxidation is 1 or close to 1 (0S1) at low tempera tures and starts to decrease at a given temperature, usually over 100C. In the present work a new parameter is intro duced, TS=0.8, which represents the tem perature for which S=0.8 (S>0.8 for lowertemperaturesandS<0.8forhigher temperatures). Only in six reports [17 22] was the selectivity of the catalyst found to be below 0.8 for all the tested temperatures. In these cases the tem perature at the highest selectivity was usedforthecomparison. In the reviewed works the conversion and selectivity values areusually offered in graph format, with temperature on the X axis. To extract the numerical val ues of X and S from the plots a Visual Basic program fed with bmpformatted scanned images of the graphs was used [6].

ArtculoII
Rankingofcatalyticactivities

The catalytic activities of the different catalystscan berankedin termsofkL,as evaluated by equation R4. Fig. 1 offers the Arrhenius plots for selected copper ceria catalysts, including the most (A) and least (G) active catalysts, in terms of kL, found in the literature. References to the original works and preparation con ditions for catalysts A to G can be found in Table 1. The catalytic activity rate constants for the most and least active copperbased catalysts differ by three orders of magnitude, although this range is significantly reduced if catalysts F and G are ignored. These two catalysts were tested in the presence of water (see Table 1), whereas the rest of the catalysts were analysed only with CO/H2 mixtures carried by an inert gas. A common feature to many of the Ar rhenius plots in Fig. 1 is the change in slope (activation energy) at high tem peratures, as a result of a sharp de creaseinselectivity.Forthezoneofhigh selectivity (low temperature zone) the slopes have similar values for most cata lysts, indicating comparable values of activation energies. Only catalyst F has a very different slope, which is caused by the inhibiting effect of water at low temperatures. For this catalyst, a much higher activity can be expected at low temperaturesintheabsenceofwater.

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

65

1.E-02

T>100C
1.E-03 (D) 1.E-04 kL [mol/(gsbar )]
0.15

T<100C
kL (A) (B)
100

(C)

1.E-05 (F) (G) 1.E-06 (E)

1.E-07

1.E-08 0.0017

0.0020

0.0023 1/T (K )
-1

0.0026

0.0029

0.0032

Fig. 1. Arrhenius plots for kL constants evaluated for selected copperceria catalysts: A [6], B [24], C (this work), D [7], E [10], F [29] and G [19] (descriptions of these catalysts canbefoundinTable1).

The values of kL at 100C (kL100), situated in the zone of high selectivity for all catalysts, were evaluated by inter / ex trapolation (Fig. 1) and used to rank their absolute catalytic activities. The results are listed in Tables 1 to 4, to gether with the references to the origi nal works, preparation and testing con ditions and selectivity values (TS=0.8). The catalysts have been grouped according to their composition: (i) copperceria catalysts (Table 1), (ii)alumina sup ported copperceria catalysts (Table 2), (iii) modified copperceria catalysts (Ta ble 3) and (iv)nonceria based copper

catalysts(Table4).Withineachtablethe catalysts are ranked in descending order of catalytic activity (kL100 value), allowing the most active catalysts to be recog nisedataglance. For this work our group tested the same catalyst at an initial CO partial pressure of 5103 bar, within the applicability range of the LF equation, and at the highest spatial velocity (FCO0/wcat) ever used for any of the reported catalysts (Table1).AdecreaseinthekL100valueby ~3.5 times was observed although the catalyst was still one of the most active, only surpassed in activity by the cata

66 lysts prepared by Liu et al. [23, 24] (Fig. 1andTable1). There appears to be agreement be tween members of the scientific com munitythatactivityinthePROXreaction is favoured by a high interdispersion of Cu(II) on Ce(IV), as this permits the for mation of oxygen vacancies in the cop perceria boundaries and improves the reducibility of copper [25]. For this rea son segregation of copper oxide into clusters is considered to be detrimental for the activity of the catalysts. This seg regation, detected by XRD analysis, usu ally takes place for copper contents of over ~56 wt% [6, 25]. Thus, there must be a relationship between catalytic ac tivity and parameters such as specific surfacearea,coppercontentanddisper sion (preparation method), at least for the copperceria catalysts (Table 1). However, trying to extract information from Tables 1 to 4 to help identify pa rameters which may be considered key to preparing an active catalyst is a dis couraging task. Copper content, specific surface area and calcination conditions are parameters whose values are ran domly distributed over the entire range of catalysts, when plotted against their catalytic activity values. Supporting the copperceria catalysts on alumina does not produce any significant catalytic improvement (Table 2). Neither does adding to the copperceria mixtures specific modifiers, such as Zr, Sm, Sn, Co

ArtculoII or Fe (Table 3). Even worse results are obtained when dealing with copper based catalysts where ceria does not form part of the composition (Table 4). Considering the previously mentioned catalytic synergy between ceria and copper, this result is far from surprising. On some occasions the catalysts were thermally pretreated under oxidising or reducing conditions prior to their testing (see symbols included in the calcination column in Tables 14). As can be ob served in the tables, it seems that these treatments may have a negative effect on the resulting catalytic activity values, since they are generally associated to the least active catalysts (Table 1). Oxi dative treatments, which are the most abundant together with inert or non performed pretreatments, may accel erate catalyst deactivation, as has been reported to occur with copperbased methanol reforming catalysts [26], al though a closer evaluation of this possi ble contribution is required before a finalconclusioncanbedrawn.Thus,asa counterbalance to our former observa tion, the least active catalyst reported in Table 1 was that subjected to a reducing pretreatment. Chelatingmethodssuchascitrates[6]or solgel with cetyltrimethylammonium bromide [24], which are known to pro vokeanexcellentdispersionofcopperin ceria, seem to be the most effective methods for fabricating active catalysts

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

67

(Table1).Inlightoftheabovediscussion on citrate catalysts [6], the catalyst pre pared by Liu et al. [24] must be consid eredthemostactivecatalystreportedin this review. However, an apparently simple procedure i.e. the co precipitation of the metal precursors in a solution of K2CO3/KOH, as reported by Liu et al. [23], produces a catalyst which is almost identical in activity to the most active catalyst obtained by the same authors [24] using the solgel technique (Table 1). According to these authors, theadditionofKOHtothebasicsolution inhibits the growth of the CuOCeO2 particles and yields nanostructured catalysts (dXRD=58 nm) which are highly active in the PROX process. Yet, crystal size does not seem to be the most im portantparameteraffectingthecatalytic activity of the reported materials. The catalysts with the smallest crystallite sizes are those reported by Luo et al. [27] (slightly over 3 nm) and Avgouro poulos et al. [10] (4nm), but none of these catalysts are ranked among the mostactivecatalysts(Table1).The cata lyst by Luo et al. [27] waseven prepared by a solgel method so that the disper sion of copper ought to be optimal, add ing further mystery to what is already a puzzlingproblem. The parameter of selectivity selected in thiswork(TS=0.8)hasanaveragevaluefor all the catalysts of 142 23C. For the unmodified copperceria catalysts, in

cluding the aluminasupported ones, this value is 148 26C. Again this pa rameter seems to be unrelated to any characteristic property of the catalysts, although in this case it is clearly de pendent on the testing conditions, which may alter the relative kinetics of the CO and H2 oxidation reactions. The catalyst which is selective towards CO oxidation in the widest temperature range is that prepared by Avgouropou los[9],withavalueofTS=0.8=182C. Experimental artifacts might be the rea son for the apparent absence of trends between the preparation parameters and catalytic activities. Most of the re viewed works employed microreactors with internal diameters in the 415 mm range, though reactor dimensions are not reported in about half of the re viewed papers. In these reactors, sam ples with weights in the range of 15 500mg were tested. Only in a couple works were sample weights over 1 g used [19, 28]. The majority also em ployed total flow rates in the 30 300mL/min range. With the available data, it is difficult to determine whether it is resistance to external diffusion and/or the existence of axial dispersion that are masking the reported values of thereactionrates.Inordertodetermine this,valuesforparameterssuchasparti cle size, reactor dimensions or weight of inert particles (when dilution is per formed) are also needed. Applying stan

68 dard assumptions for the bed porosity and apparent density of the catalysts and inert materials, only in the case of seven works [12, 19, 24, 27, 2931] was it possible to evaluate the existence or absence of diffusion restrictions by standard correlations [32]. External dif fusion resistances were absent in all the analysed works (Carberry number below 0.05 [32]), whereas axial dispersion, evaluated by means of the Peclet num ber [32], was found to be a potentially occurring factor in four [12, 19, 24, 27] outofthesevenanalysedworks. In addition, dilution of the catalyst sam ple in a bed of inert particles was not always performed. This procedure al lows the heat produced during the reac tion to dissipate more easily, thus avoid ing the generation of hot spots in the

ArtculoII bed. Therefore, catalysts in which dilu tion was not performed might be ex pectedtoreflectthedifferencebetween the actual temperature of the bed and the temperature measured by the ther mocouple, which is always placed close to but outside the bed. In Tables 1 to 4 the works in which catalyst dilution was performed are identified by a symbol in the column of spatial velocity (FCO0/wcat). As can be observed, fewer than half of the catalysts were tested under dilution conditions. Nevertheless, this argument must be regarded as merely speculative, with no specific weight unless meas urements of the actual differences be tween the control and bed tempera turesarealsoreported.


Support None 40 50 4.110 1.4105 1.00 1.00 1.00 1.00 1.25 1.25 1102 1102 1.4105 80 23 6.8106 1102 1102 1.25 1.25 1.00 1.25 50 50 50 50 50 50 50 25 50 50 1.4105 2.3105 1.4105 7.4105 1.4105 1.5105 7.4105 1102 1102 1102 5103 1102 1102 310 99 138 40 n.i. 107 20 39 n.i.
6

Table 1. Catalyic activity of unsupported or CeO2supported CuOCeO2 catalysts (metal contents refer to the catalyst in cludingthesupport).
Calcination (air) 550C4h 1.00 1.00 33 CIT SGCBr CP(K2CO3/ KOH) CIT CP(NaOH) CeO2:ME CuO:IW SGPR UNC CeO2:ME CuO:IW CeO2:P CuO:IW SGPR 33 550C4h 500C2h 500C3h 550C4h 500C5h 500C2h 650C4h 550C1h 500C2h 300C5h 650C1h

Ref.a Method CIT 4.1106 3104

Cu (wt%)

SBET FCO0/wcat 2 (m /g) (mol/gs)

PCO0 (bar)

PO20/ PCO0

PH20/ PCO0

kL100 106 181.0 165.0 103.0 97.0 51.1 44.4 42.6 38.0 25.2 33.0 29.9 28.2

TS=0.8 (C) 194 188 107 122 140 143 131 182 171
S=0.7 100C

Marbn[6](A) None None None None None CeO2 None None CeO2 CeO2 None

10.66

Marbn[6]

6.02

Liu[24](B)

5.00

Liu[23]

5.00

Thiswork(C)

10.66

Kim[44]

8.27

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

Gamarra[25]

1.00

Avgouropoulos [9]

1.90

Avgouropoulos [45]

6.02

MartnezArias [21]

1.00

Manzoli[7]b(D)

3.99

n.i. 135

Sedmak[12]

3.90

69


Support

70

Table1(cont.)
SBET
(m /g)
2

Ref.a Calcination (air) 300C 128 92 80 50 25 90 100 148 150 39 n.i. 22 2.8106 2.8106 2.810
6

Cu (wt%) None CeO2 CeO2 1.00 1.00 0.50 1.00 1.00

2

FCO0/wcat (mol/gs) Method CP(NaOH) CeO2:ME CuO:IW CeO2:P CuO:IW CeO2:com CuO:EI CeO2:com CuO:IW 35 50 80 1.00 110
2

PCO0 (bar) 1102 1.00 1.25 50 100 50 10 110 1102 5103 2102 1102 1102
2

PO20/ PCO0 27.8c 27.1 26.8 23.2 17.9 Urea gelationCP CPNaOH 50 1.25 50 SGCBrP UNC 16.0 13.9 1.5 12.4

PH20/ PCO0

kL100 106

TS=0.8 (C) 106

S=0.7 100C

Ko[8] 500C2h 7.410 1.4105 6.8106 1.4105 4.5106 6.8106 1.410

5 6

8.27

6.8106

MartnezArias [22](B) 300C5h 450C2h 450C4h 650C4h 100C 400C4h 550C1h 500C1h 400C2h

1.00

Manzoli[7]b CeO2 CeO2 None None None None None CeO2 110

3.99

n.i. 130 188 161 100 158 146

Caputo[33]b

4.31

Mario[37]

1.00

Liu[36]

3.90

Ko[8]

8.27

Luo[27]

4.70

Avgoroupoulos [40]

6.02

Papavasiliou [41]

6.02

1102 1102

1.25 1.25

50 50

UNC CeO2:P CuO:DP (K2CO3)

12.3 9.4

146 139

Avgoroupoulos [10])(E) None 700C

2.40

ArtculoII

Jung[20]

5.10

n.i.

3.4106

8103

0.95

90

CP(Na2CO3)

3.4d

S=0.7 150C

Table1(cont.)
Support
(m /g)
2

Ref.a Calcination (air) 500C4h n.i. 109 124 9.3108 1.25 151 5103 1.4105 1.00 71 1102 CP(NaOH) CP(KOH) 1.50 65 CP(Na2CO3) 500C5h 500C5h 2.3106 1102 2.3 0.4e 0.2f SBET Method None None None

Cu (wt%)

FCO0/wcat (mol/gs)

PCO0 (bar)

PO20/ PCO0

PH20/ PCO0

kL100 106

TS=0.8 (C) 135 174 S=0.4 150C

Zou[46]

7.99

Bae[29](F

8.27

Ratnasamy[19] (G)

4.00

n.i.: not indicated; P: Precipitation; IW: Incipient wetness; Com: Commercial; EI: Excess impregnation; CP: coprecipitation; UNC: urea ni trate combustion; DP: depositionprecipitation. CIT: citrates; SG: Solgel (CBr: cetyltrimethylammonium bromide, PR: Peroxoroute); ME: Microemulsion.

ThecatalystwassubjectedtooxidationtreatmentwithO2/inertgaspriortotheactivitytest.

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

:Catalystdilutedininertparticlesforactivitytests. ThecatalystwassubjectetoareductiontreatmentwithH2/inertgaspriortotheactivitytest

LettersinbracketsrefertoplotsinFig.1. TheseauthorsalsopreparedCuOx/CeO2sampleswithzirconiaandfoundnobeneficialeffectoncatalyticactivity.

Catalyticactivitytestperformedinthepresenceof0.02barH2O. Catalyticactivitytestperformedinthepresenceof0.0235barCO2. e Catalyticactivitytestperformedinthepresenceof0.14barH2O+0.135barCO2.

Catalyticactivitytestperformedinthepresenceof0.26barH2O+0.23barCO2.

71

72

Table 2. Catalytic activity of aluminasupported CuOx/CeO2 catalysts (metal contents refer to the catalyst including the sup port).
Support
(m /g)
2

Ref.

Cu

(wt%)

Ce (wt%) Al2O3/ CeO2 1.00 Support:SG CuO:EI CeO2:EI UNC 50 Alfoam 500C1h 1.25 50 6.8106 1.00 60 1102 n.i. 8.1107 1102 500C3h 200 1.2106 6103 4.7

Calcination (air)

SBET

FCO0/wcat (mol/gs) Method

PCO0 (bar)

PO20/ PCO0

PH20/ PCO0

kL100 106

TS=0.8 (C) S=0.8 150C

Moretti [18]

5.00

15.36

Papava siliou[41] Al2O3 500C4h n.i.

1.72

21.49

4.2

140

Park[30, 31,47]

4.00

16.00

EI

3.5

175

n.i.:notindicated;EI:Excessimpregnation;UNC:ureanitratecombustion;SG:Solgel. ThecatalystwassubjectedtooxidationtreatmentwithO2/inertgaspriortotheactivitytest.

:Catalystdilutedininertparticlesforactivitytests.

ArtculoII

Table3.CatalyticactivityofmodifiedCuOx/CeO2catalysts(metalcontentsrefertothecatalystincludingthesupport).
Support Calcination SBET (air) (m2/g) 250C1.5h+ 500C3.5h 94 1.7105 2.00 97 1102 SupportCP (NH4OH) CuO:EI SupportCP (NH4OH) CuO:IW 50 35 1.00 1102 1102 1.00 1.00 60 40 65 FCO0/wcat (mol/gs) Method 23.6 PCO0 (bar) PO20/ PCO0 PH20/ PCO0 kL100 106 TS=0.8 (C) 127

Ref.

Cu (wt%) SDCa

Modifier (wt%)

Wang[39]

4.76

Sm(8.20)

Chen[43] Ce0.9Zr0.1O2 78 43 n.i. 34 n.i. 3.4105 2.3105 6.8106 1102 1.4105 2102 0.50 1.00 450C4h 500C4h 550C4h 500C4h 650C4h 1.1106 1102

5.59 Ce0.9Sn0.1O2 84 1.00 50

Sn(6.49)

650C4h

1.1106

1102

18.3 15.5 Supportcom CuO:IW EI UNC CP(Na2CO3) 13.6 10.5 4.6 0.5b

125 123 112 171 141 157

Chen[42]

5.59

Zr(5.07)

Mario[37] Al2O3 None None

1.00

Zr(21.64) Ce0.63Zr0.37O2

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

Park[30,31, 47]

4.00

Co(0.20)&

Sirichaip rasert[38]

8.20

Fe(23.09)

Dong[34]

3.63

Zr(5.21)

ThecatalystwassubjectedtooxidationtreatmentwithO2/inertgaspriortotheactivitytest. Catalystdilutedininertparticlesforactivitytests & 15.8wt%Ce. n.i.:notindicated;CP:coprecipitation;EI:excessimpregnation;IW:incipientwetness;Com:commercial;UNC:ureanitratecombustion. a Samariadopedceria. b Catalyticactivitytestperformedinthepresenceof0.25barCO2.

73

74

Table4.CatalyticactivityofnonceriabasedCuOxcatalysts(metalcontentsrefertothecatalystincludingthesupport).
Support
(m /g)
2

Ref. SBET 118 30

Cu (wt%) None 500C3h CP(NH4OH) 7.3107 1.3102 1.00 2.9a n.i

Modifier

(wt%)

Calcination (air)

FCO0/wcat (mol/gs) Method TS=0.8 (C)

PCO0 (bar)

PO20/ PCO0

PH20/ PCO0

kL100 106

Tanaka [28] None 400C5h 160 0.98 74 SGEG 3.4106 1102 2.3b

30.00

Zn (24.10) Al(21.17)

Kramer [35] SBA15 500C4h 569 1.00 45 2.3106 1102

16.08

Mn (55.59)

120

Tu[17]

4.70

None

PGAPTS

0.6

S=0.3 80C

ThecatalystwassubjectedtooxidationtreatmentwithO2/inertgaspriortotheactivitytest.

ThecatalystwassubjectedtoreductiontreatmentwithH2/inertgaspriortotheactivitytest.

Apackingdensityof1.25g/cm3wasassumedforthecatalyst.

n.i.notindicated;CP:coprecipitation;SGEG:solgelwithethylenglycol;PGAPTS:postgraftingwithsilaneAPTS.

Catalyticactivitytestperformedinthepresenceof0.25barH2O+0.125barCO2. Catalyticactivitytestperformedinthepresenceof0.2barCO2.

ArtculoII

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

75

Catalyststability

The stability of the catalysts is a prop erty of great importance affecting their applicability. Deactivation during the PROX reaction of copperceria catalysts is not usually attributed to coke deposi tion but to the accumulation of hydroxyl species on active interfacial sites and/or to copper sintering [21]. The articles reviewed here show quite different re sults for stability. In most of the cases stability tests were not performed [7, 8, 17, 2022, 25, 27, 29, 3339]. In the re maining works, stability tests were con ducted for time durations ranging from 12 to 200 hours, at temperatures in the 80175C range. However, in most pub lished works the experiments were per formed under conditions (catalyst mass, temperature, total flow of reactants and partial pressures) that allowed the com plete or almost complete conversion of CO (X>0.9 [9, 14, 18, 19, 23, 24, 4043]; 0.8<X<0.9 [15, 28]). Under complete conversion conditions the catalysts may experience deactivation so that the con version value may not be affected for a long period of time. This would affect the assessment of the real effect of timeonstream on catalytic activity. In

addition, the value of the kinetic con stant is subject to greater uncertainty when it is evaluated for conversion val uescloseto1.Insomecasestheauthors consider the deactivation produced by CO2 and H2O simultaneously [9, 31, 40]. Only a few authors [6, 12, 16, 44] have analysed the deactivation of the PROX catalysts in experimental conditions in whichX<0.6orunderdrasticexperimen tal conditions (high gas hourly space velocities). The experimental conditions of these works are summarised in Table 5. The last two rows include the deacti vation data reported by Tanaka and coworkers [28] under experimental con ditions that in principle cannot be con sidered as very appropriate for perform ing a deactivation analysis for the PROX process, such as low spatial velocity (FCO0/wcat), high temperature and high initial conversion (X=0.89). In these less severe conditions, the effect of the method of preparation can be clearly seen, since two catalysts of identical composition and tested under the same experimental conditions exhibit a totally different behaviour: one of the catalysts undergoes almost complete deactiva tion,whiletheotherremainsstable.

76

Table5.Deactivationexperimentsfordifferentcatalysts.
SBET
(m /g)
2

Ref.(Catalyst) n.i. 1102 1.00 1.25 SgPR CIT CD EI 7.3107 18 n.i. 28 100 150 150 4.1106 42 174 1450.0 (0.93) n.a. (0.25) 4.4(0.89) 4.4(0.89) 1.00 0.50 30 30 23 33 50 6.8106 55 80 10.0 (0.40) 50 504 170 48.9 (0.60) 23 1102 40 3104 n.i. 2102 118 1.3102 1.00 118 1.3102 1.00 CP(NaOH) 1.4105

Cu Method

(wt%)

Calcina tion(air)

PCO0 (bar) TOS (h) T (C)

PO20/ PCO0

PH20/ PCO0

FCO0/wcat (mol/gs)

kL100106 (X)t=0a

Kim[44] (CuO/CeO2)

8.27 500C5h

kL100106 (X) t=TOSb 22.9 (0.30) 7.2 (0.31) 836.0 (0.74) n.a. (0.20) 0.2 (0.07) 4.4 (0.89)

Sed,al[12] (CuO/CeO2)

3.90

650C1h

Marbn[6] (CuO/CeO2)

10.66

550C4h

Usachev[16] (CuO/CeO2)

3.99

300C2h

Tanaka[28]

5.00

(CuO/ZnO/Al2O3)

500C3h

Tanaka[28]

30.0

CP(NH4OH) 7.3107 17

(CuO/ZnO/Al2O3)

500C3h

ThecatalystwassubjectedtooxidationtreatmentwithO2/inertgaspriortotheactivitytest.

ThecatalystwassubjectedtoreductiontreatmentwithH2/inertgaspriortotheactivitytest.

Apackingdensityof1.25g/cm3wasassumedforthecatalyst.

:Catalystdilutedininertparticlesforactivitytest. n.i.notindicated;n.a.notavailable;CIT:citrates;CP:coprecipitation;SG:solgel;SolPR:solgelperoxoroute;CD:combineddecompo

sotionofnitratesfromsolution;EI:excessimpregnation. Dataatt=0.

ArtculoII

Dataatt=finalTOS.

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

77

In the rest of the works referred to in Table 5 deactivation always takes place. The occurrence of deactivation adds moreuncertaintytotheactivitydata(kL) reported in Tables 14. The assessment of CO conversion values is always per formed by means of experiments at increasing or decreasing temperatures, whichlastforseveralhours.Evenifcata lyst stability is not analysed, deactiva tion may take place anyway, which might affect the conversion values ob tained at the end of the activity tests. For instance, the catalyst prepared via excess impregnation by Tanaka et al. [28] undergoes almost complete deacti vation at 150C in a period as short as 2hours. This phenomenon can be also observed in the hysteresis loop of con version that often occurs when the test ing experiments are performed at in creasing and then decreasing tempera tures [6]. In conclusion, for the PROX experimentsitisveryimportanttospec ify the timeonstream at which the con version values are obtained, and stabil ity tests should be always performed andreported.

In summary, the catalytic activities of the different catalysts reported in the literature, which the LF equation places put on a common basis, seem to display a certain degree of uncertainty, caused either by the dubious selection of condi tions during some of the reported activ ity tests or by the occurrence of catalyst deactivation during such tests. In any case, it is clear that the addition of modifiers does not improve the catalytic activity of the copperceria catalysts, andthatchelatingmethodsarethemost appropriate procedures for the prepara tion of copperbased catalysts for CO PROX.
Acknowledgments

Funding by the Spanish National Project MAT200500262 and the FICYT Regional Project (IB05001) is acknowledged. Irene Lpez thanks the Ministerio de Educacin y Ciencia of Spain for her FPI grant.

References

[1] H.F.Oetjen,V.M.Schmidt,U.Stimming,F.Trila,J.Electrochem.Soc.143(1996) 3838. [2] G.Marbn,T.ValdsSols,Int.J.HydrogenEnergy32(2007)1625. [3] D.L.Trimm,Appl.Catal.A:Gen.296(2005)1.

78 [4] M.J.Kahlich,H.A.Gasteiger,R.J.Behm,J.Catal.171(1997)93.

ArtculoII

[5] G.Avgouropoulos,T.Ioannides,H.K.Matralis,J.Batista,S.Hocevar,Cat.Lett.73 (2001)33. [6] G.Marbn,A.B.Fuertes,Appl.Catal.B:Environ.57(2005)43. [7] M. Manzoli, R. Di Monte, F. Boccuzzi, S. Coluccia, J. Kaspar, Appl. Catal. B: Envi ron.61(2005)192. [8] E. Y. Ko, E. D. Park, K. W. Seo, H. C. Lee, D. Lee, S. Kim, Catal. Today 116 (2006) 377. [9] G. Avgouropoulos, T. Ioannides, C. Papadopoulou, J. Batista, S. Hocevar, H. K. Matralis,Catal.Today75(2002)157.

[10] G. Avgouropoulos, J. Papavasiliou, T. Tabakova, V. Idakiev, T. Ioannides, Chem. Eng.J.124(2006)41. [11] W.Liu,M.FlytzaniStephanopoulos,J.Catal.153(1995)317. [12] G.Sedmak,S.Hocevar,J.Levec,J.Catal.213(2003)135. [13] M.A.Vannice,Catal.Today123(2007)18. [14] P.K.Cheekatamarla,W.S.Epling,A.M.Lane,J.PowerSources147(2005)178. [15] T.Utaka,T.Takeguchi,R.Kikuchi,K.Eguchi,Appl.Catal.A:Gen.246(2003)117. [16] N. Y. Usachev, I. A. Gorevaya, E. P. Belanova, A. V. Kazakov, O. K. Atal'yan, V. V. Kharlamov,Russ.Chem.Bull.53(2004)538. [17] C.H.Tu,A.Q.Wang,M.Y.Zheng,X.D. Wang,T.Zhang,Appl.Catal.A:Gen.297 (2006)40. [18] E. Moretti, M. Lenarda, L. Storaro, A. Talon, R. Frattini, S. Polizzi, E. Rodriguez Castellon,A.JimenezLopez,Appl.Catal.B:Environ.72(2007)149. [19] P. Ratnasamy, D. Srinivas, C. V. V. Satyanarayana, P. Manikandan, R. S. S. Kuma ran,M.Sachin,V.N.Shetti,J.Catal.221(2004)455. [20] C. R. Jung, J. Han, S. W. Nam, T.H. Lim, S.A. Hong, H.I. Lee, Catal. Today 9395 (2004)183. [21] A. MartinezArias, A. B. Hungria, G. Munuera, D. Gamarra, Appl. Catal. B: Envi ron.65(2006)207. [22] A.MartinezArias,A.B.Hungria,M.FernandezGarcia,J.C.Conesa,G.Munuera, J.PowerSources151(2005)32. [23] Z.Liu,R.Zhou,X.Zheng,J.Mol.Catal.AChem.255(2006)103. [24] Z.Liu,R.Zhou,X.Zheng,J.Mol.Catal.AChem.267(2007)137.

Lpezetal.Int.J.Hydrog.Energy33(2008)197205

79

[25] D. Gamarra, A. Horns, Zs. Koppany, Z. Schay, G. Munuera, J. Soria, A. Martnez Arias,J.PowerSources169(2007)110. [26] P.H.Matter,D.J.Braden,U.S.Ozkan,J.Catal.223(2004)340. [27] M.F.Luo,J.M.Ma,J.Q.Lu,Y.P.Song,Y.J.Wang,J.Catal.246(2007)52. [28] Y. Tanaka, T. Utaka, R. Kikuchi, K. Sasaki, K. Eguchi, Appl. Catal. A: Gen. 238 (2003)11. [29] C.M.Bae,J.B.Ko,D.H.Kim,Cat.Comm.6(2005)507. [30] J. W. Park, J. H. Jeong, W. L. Yoon, H. Jung, H. T. Lee, D. K. Lee, Y. K. Park, Y. W. Rhee,Appl.Catal.A:Gen.274(2004)25. [31] J.W.Park,J.H.Jeong,W.L.Yoon,Y.W.Rhee,J.PowerSources132(2004)18. [32] J.A.Moulijn,A.Tarfaoui,F.Kapteijn,Catal.Today11(1991)1. [33] T.Caputo,R.Pirone,G.Russo,KineticsandCatalysis47(2006)756. [34] X.F.Dong,H.B.Zou,W.M.Lin,Int.J.HydrogenEnergy31(2006)2337. [35] M. Kramer, T. Schmidt, K. Stowe, W. F. Maier, Appl. Catal. A: Gen. 302 (2006) 257. [36] Y.Liu,Q.Fu,M.F.Stephanopoulos,Catal.Today9395(2004)241. [37] F.Mario,C.Descorme,D.Duprez,Appl.Catal.B:Environ.58(2005)175. [38] K. Sirichaiprasert, A. Luengnaruemitchai, S. Pongstabodee, Int. J. Hydrogen En ergy32(2007)915. [39] J.B.Wang,S.Ch.Lin,T.J.Huang,Appl.Catal.A:Gen.232(2002)107. [40] G.Avgouropoulos,T.Ioannides,Appl.Catal.A:Gen.244(2003)155. [41] J. Papavasiliou, G. Avgouropoulos, T. Ioannides, Appl. Catal. B: Environ. 66 (2006)168. [42] Y.Z.Chen,B.J.Liaw,H.C.Chen,Int.J.HydrogenEnergy31(2006)427. [43] Y.Z.Chen,B.J.Liaw,C.W.Huang,Appl.Catal.A:Gen.302(2006)168. [44] D.H.Kim,J.A.Cha,Cat.Lett.86(2003)107. [45] G. Avgouropoulos, T. Ioannides, H. K. Matralis, Appl. Catal.B: Environ. 56 (2005) 87. [46] H.B.Zou,X.F.Dong,W.M.Lin,Appl.Surf.Sci.253(2006)2893. [47] J.W.Park,J.H.Jeong,W.L.Yoon,C.S.Kim,D.K.Lee,Y.K.Park,Y.W.Rhee,Int. J.HydrogenEnergy30(2005)209.

81

ResumenartculoIII. PREFERENTIALOXIDATIONOFCOBYCUOX/CEO2NANOCATALYSTS PREPAREDBYSACOP.MECHANISMSOFDEACTIVATIONUNDERTHE REACTANTSTREAM

OxidacinpreferencialdeCOpornanocatalizadoresCuOx/CeO2 preparadosporSACOP.Mecanismosdedesactivacinenlacorriente dereaccin.
Los catalizadores de xido de cobre y xido de cerio destacan por su alta actividad y selectividad enlareaccin deoxidacin preferencial deCO.Sinembargo,esbiensabido queestosmaterialessedesactivandeformagradualdurantelareaccin,loquelimitasu aplicacin prctica. Este trabajo tuvo como objetivo el estudio de las causas de dicha desactivacin, para lo cual se emple un catalizador CuOx/CeO2 sintetizado mediante coprecipitacinconfinadaenhidrogelesdeslice(SACOP). El anlisis de actividad de este catalizador se realiz bajo dos condiciones de reaccin distintas, alta y baja concentracin (HC: 5000 ppm CO, 5000 ppm O2 y 25% H2 y LC: 300ppm CO, 300 ppm O2 y 10000 ppm H2), durante largos periodos de tiempo para analizar su estabilidad. En ambos casos las velocidades espaciales fueron muy elevadas, de 0.4105 y 2.3105 molCOgcat1s1, respectivamente. Estos ensayos de estabilidad demostraron que las condiciones de alta concentracin eran ms desactivantes que las debajaconcentracin. Deformaadicionalsellevaronacaboexperimentosderespuestaaescalncondistintas mezclas de gases (CO/He, CO+H2/He y CO2+H2/He) con el propsito de determinar el potencial desactivante de los distintos gases implicados en la reaccin PROX. Estos ensayos consistieron en una primera etapa PROX en condiciones de baja concentracin durante 1 hora, tras la cual se introdujo un escaln de CO/He, CO+H2/He o CO2+H2/He por un tiempo de 18 horas. Finalmente, en una tercera etapa, se volvi a someter al catalizadoracondicionesPROXparaevaluarelefectodelescalndelaetapaanterior. En los ensayos de respuesta a escaln se comprob que la adicin de CO/He, CO+H2/He o CO2/He produca slo un pequeo cambio en la actividad del catalizador, cifrado en una disminucin media de la conversin de CO de 10%. De hecho, las condiciones

82

ArtculoIII

PROX ejercieron el mayor efecto desactivante en el catalizador, con una disminucin en la conversin en el mismo intervalo de 30%, lo cual se atribuy a la presencia de oxgeno, ausente en el resto de escalones. Analizando la diferencia entre el consumo de CO y la formacin de CO2 fue posible deducir que las condiciones PROX tenan un claro efecto promotor en la formacin de especies carbonosas superficiales. En cualquier caso, estas no parecieron afectar de forma significativa a la actividad de los nanocatalizadoresenlascondicionesdebajaconcentracin. Se llevaron a cabo anlisis XPS y FTIR ex situ sobre muestras sometidas a reaccin PROX a 175C a distintos tiempos (1 y 20 h) y sobre la muestra de partida. Combinando los resultados obtenidos de los anlisis XPS y FTIR se postul un mecanismo de desactivacin a temperaturas inferiores a 250C compuesto de dos submecanismos distintos,laformacindecarbonatossuperficialesylaagregacindeCu2+disperso,cuya importancia relativa es funcin del tiempo de reaccin, la temperatura y la concentracindeCO. Los resultados de XPS sugirieron la formacin de depsitos de coque, lo que provocara la disminucin de la concentracin superficial de cerio. Sin embargo, creemos que esto no debera afectar a la desactivacin puesto que de ser as, se detectara de forma ms patente en los experimentos de larga duracin en condiciones de alta concentracin de CO. Adems, la concentracin de carbono graftico en la primera hora de reaccin no varisustancialmente,siendoesteperiodoeldemayordesactivacin. Se demostr que el catalizador fresco contena carbonatos superficiales lbiles producidos por exposicin atmosfrica. A temperaturas bajas y condiciones de baja concentracin, durante los primeros instantes de la reaccin PROX los carbonatos iniciales se eliminaron mediante descomposicin oxidativa, mientras que simultneamente se formaban de manera progresiva carbonatos asociados a los iones Ce3+. En condiciones de alta concentracin y baja temperatura (175C) la formacin de dichoscarbonatosfuemuchomsrpidaqueencondicionesdebajaconcentracin,sin afectarenningunodelosdoscasosalasespeciesCu2+. Paratemperaturasmselevadas(200C)yconcentracionesbajasdeCOlasinterizacin delCu2+dispersofuemspatente,peroseviocompensadaporunamenorformacinde carbonatos, lo que provoc que la velocidad de desactivacin fuera menor que a 175C. EncondicionesHCseformaronmscarbonatossuperficialesa200Cqueencondiciones LC,peroenmenormedidaquea175C.

83 En la Tabla 1 se resumen los dos mecanismos de desactivacin que operan a temperaturas inferiores a 250C, as como el efecto que provoca el incremento de las variables de inters (temperatura, concentracin de CO y tiempo) en los distintos mecanismos.

Tabla1.Mecanismosdedesactivacinatemperaturasmenoresde250C.
Especie metlica involucrada Influenciadelincrementodelasvariables Temperatura Concentracin deCO Tiempo Ocurreenlas primeras15horas encondicionesHC ylentamenteen condicionesLC

Tipode desactivacin

Formacinde carbonatos (mecanismo mayoritariode desactivacin) Poragregacinde Cu2+dispersoen clsteresdeCuO (sinterizacin)

Ce3+formado durantela La reaccinPROX desactivacin (encontacto disminuye conclsteres deCuO) Cu2+disperso Aumenta.Es dudosamente significativaa 175C

Lavelocidadde desactivacin aumenta

Ocurrehasta Aparentemente desaparicinde noafecta Cu2+disperso

Laprincipalcausadedesactivacindelcatalizadorparatemperaturassuperioresa300C fuelasinterizacindelasespeciesdecobre,tantoparabajacomoaltaconcentracinde CO. Estos mecanismos explicaran adems el fenmeno de histresis en los ciclos PROX no isotrmicos que depende de las concentraciones de gases empleadas. En el ciclo en condicionesLC,laconversindeCOesmayoratemperaturascrecientesquelaobtenida a temperaturas decrecientes, lo contrario que ocurre cuando se utilizan condiciones de alta concentracin HC. En condiciones LC la desactivacin empieza a notarse a alta temperatura por sinterizacin de las especies de cobre, por lo que el tramo de temperaturas crecientes ofrece una mayor reactividad. Por el contrario, en condiciones HC la elevada concentracin de CO hace que la desactivacin por formacin de carbonatos superficiales sea conspicua a baja temperatura. Estos carbonatos desaparecen cuando se eleva la temperatura, por lo que en el tramo de temperaturas decrecientesseobservaunareactivacindelcatalizador.

85

ArtculoIII.

1.Introduction There is widespread interest in the de velopment of catalysts based on non noble metals for massscale processes such as those involved in the production and conversion of hydrogen in fuelcell powered cars [1, 2]. One of these proc esses is the purification of the hydrogen stream produced onboard by the steam reforming of a liquid fuel (e.g. metha nol), so that residual CO, which is an activepoisonforelectrocatalysts,canbe quantitatively eliminated. The preferred methodofeliminationisthepreferential (selective) catalytic oxidation of CO (PROX reaction). One of the most active and selective metal oxides in this proc ess is copper oxide, whose activity is significantly increased by the inclusion of cerium oxide in the catalyst composi tion. This mixture (CuO/CeO2) has been extensively studied in the PROX reaction at 100250C, as recently reviewed by Lpez et al. [3]. However, it is well known that these catalysts suffer pro

86 gressive deactivation during the PROX reaction [413] which limits their use under practical conditions although not much interest has been shown in the studyofthecausesofthisphenomenon, as pointed out in the above mentioned review [3]. In this study the PROX activ ity of CuOx/CeO2 catalysts prepared by silicaaquagel confined precipitation (SACOP[14,15])isdiscussedforthefirst time. However, the main objective of this study is to determine the causes of the slow deactivation of these catalysts during the reaction. As CuOx/CeO2 cata lysts are widely used in other catalytic processes such as water gas shift reac tion and steam reforming, in which the

ArtculoIII samereactantsappear,theobservations reported in this work might be of inter est also for people studying these reac tions. 1.1.Antecedents One of the few detailed mechanisms of CO oxidation on CuO/CeO2 catalysts was offered by MartnezArias et al. [16]. They proposed a reaction scheme ac cording to which Ce4+O2Cu2+ pairs are rapidly reduced by CO to produce CO2 (R1 reaction in scheme shown below). The subsequent reactions R2 to R5, adapted from MartnezArias mecha nism, take place in order to restore the oxidationstateofthecatalyst. (R1) (R2) (R3) (R4) (R5)

3+ + 2Ce4+O2Cu2++2CO 2Ce Cu +2CO2 3+ + 2Ce3+ Cu++2CO 2Ce Cu CO 4+ + Ce3+ Cu+CO+O2 Ce O2 Cu CO 4+ 2 2+ Ce4+O2Cu+CO Ce O2 Cu +CO 4+ 2 2+ Ce4+O22Cu2++Ce3+ Cu+CO 2Ce O Cu +CO

A similar MarsvanKrevelen (MvK) mechanism was proposed by Sedmak et al. [5]. According to these authors CO (and H2) molecules are adsorbed on the copperceria interfacial region of the catalyst and react with the lattice oxy gen to produce CO2 (and H2O), as a re sult of which the copper ion Cu2+ is re duced to Cu+. The cerianite lattice then supplies the lost oxygen anion to the

copperceria interface so that finally the vacancies are refilled by gaseous oxy gen. In a subsequent work by MartnezArias et al. [17], exhaustive analyses using OperandoDRIFT, XPS and EPR tech niques were carried out to establish the reaction mechanism and the causes of deactivation of a CuOx/CeO2 catalyst.

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

87

They found that, although H2 and CO oxidationtakesplaceonthesameactive centres (basically CuO/CeO2 pairs in interfacial positions), there is a higher selectivity towards CO oxidation at intermediate temperatures (<200C), which is due to a higher reducibility of CuO by CO than by H2 (confirmed in a subsequent work [18]). The deactivation observed was attributed to both an ac cumulation of hydroxyls in interfacial active centres and to a certain degree of sintering. Manzoli et al. [19] suggested that the reactivity of CuOx/CeO2 catalysts is de termined by the presence of surface CuO clusters in the vicinity of ceria sites, through a MvK mechanism similar to that represented by R1 (Cu2+O2Ce4+
+ 3+ Cu Ce ). These authors attribute

Sedmak et al. [5] observed a fast, small but irreversible deactivation which was attributed to the redistribution of cop perspeciesonthecatalystsurface. It seems clear that there is not much agreement regarding the causes of de activation, partly because numerous different causes are able to provoke similar results (e.g. both sintering and the accumulation of carbonates may cause a decrease in the copper surface) and also because of the inherent diffi culty in interpreting correctly the spec troscopic results (mainly XPS and FTIR spectra). In this work an attempt will be made to shed some light on all of these issues. 2.Experimental 2.1. Catalyst synthesis and characteriza tion The CuO/CeO2 catalysts were prepared by silica aquagel confined co precipitation (SACOP) as reported else where and thoroughly described in the Supporting Information file. The main characteristicofthissyntheticprocedure it that calcination of the active phase takes place inside the pores of the silica template so that sintering is restricted during the heating process. The synthe sis parameters, final composition and texturalpropertiesofthepreparedcata lysts are shown in Table 1. The composi tion of the catalyst was evaluated by SEMEDX analysis (DSM 942, Zeiss). The

deactivation to the reduction of the catalysts during nonisothermal treat ments. Kim and Cha [19] consider that catalyst deactivation is produced by the accumulation of surface carbonates during long duration reaction experi ments in the presence of CO2 and H2O, so that activity can be restored by ther mal treatment under different atmos pheres (10% O2, 50% H2 and He). These authors synthesized their CuO/CeO2 catalysts by means of coprecipitation followed by air calcination at 500C, so that presumably sintering did not occur during the reaction. On the other hand

88 crystal size (CeO2) of the catalysts was determined by applying Scherrer's equa tion to the XRD diffractograms obtained in a Siemens D5000 diffractometer op erating at 40 kV and 20 mA with CuK radiation. The BET surface area was evaluated by N2 physisorption at196C in an ASAP 2010 analyser (Micromerit ics). Exsitu Xray photoelectron spectros copy (XPS) was carried out by means of

ArtculoIII gion. The backgrounds were subtracted bymeansofShirleybaselines,exceptfor the Cu2p3/2 (linear) and Ce3d (3rd or derpolynomial)regions.Alltheanalyzed regions were deconvolved by mixed GaussianLorentzian functions (90:10). For the quantitative analyses the atomic sensitivity factors stored in the CasaXPS database(v2.3.12Dev6)wereused. Exsitu FTIR spectra of the materials compressed in discs with KBr were re corded on a Nicolet Magna IR560 spec trometer tted with a DTGS KBr absorb ance detector. The analyzed regions were deconvolved by mixed Gaussian Lorentzianfunctions(90:10).

a Specs spectrometer, using MgK (1253.6 eV) radiation emitted from a double anode at 50 W. The binding en ergies of the spectra obtained were cor rectedwiththebindingenergyofadven titious carbon (284.6 eV) in the C1s re Table1.Catalystsusedinthedeactivationstudies. Cu/(Cu+Ce) SiO2a Reference Acid Base a atomic (wt.%) HNO3 CuCe1a 0.130.01 13.32.1 HNO3 NaOH CuCe1b 0.090.03 8.12.5 CuCe2 HCl 0.090.02 8.82.8 a SEMEDX;bXRD(Scherrer'sequation)

Na2Oa (wt.%) 0.70.3 0

dXRDb (nm) 23

SBET (m2/g) 237 235 266

2.2. Catalytic activity tests and step responseexperiments Catalytic activity tests during the PROX reaction were performed by using a quartz reactor of 7 mm internal diame ter inserted in a vertical furnace. The catalyst sample (1530 mg) was diluted in SiC particles (285270 mg) and placed in the reactor between two zones of glasswool,thusensuringisothermaland plugflow conditions. Any water present in the exit gases was removed with Mg(ClO4)2. Analyses of H2, CO, CO2, O2 and N2 in the product gases were then performed using a gas chromatograph (HP 6890) equipped with a TC detector. Twodifferentsetsofreactionconditions were applied: (a) "LC" conditions (low concentration): 300 ppm CO + 300 ppm O2 + 1% H2 in He (total flow rate=

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

89

300mL/min, STP) and (b) "HC" condi tions (high concentration: 0.5 vol% CO + 0.5 vol% O2 + 25% H2 in He (total flow rate=200 mL/min, STP). In both cases the values of the spatial velocities were very high (LC: 0.4105 molCOgcat1s1, HC:2.3105 molCOgcat1s1). Before the reactions each sample was pretreated in a helium flow at 200C for 1 h. Long duration isothermal experiments were generally performed, though for specific samples nonisothermal experiments under decreasing temperatures were also carried out at pseudosteady state conditions (~12 h per temperature point). ConversionofCOwasevaluatedas:

CO+H2/He and CO2+H2/He) with the purpose of assessing the deactivating potential of the gases involved in the PROXreactionoverthedilutedcatalysts. XPS and FTIR analyses were carried out over catalyst samples labelled as: (1)"Fresh": fresh sample (as synthesised catalyst), (2)"PROX1h": catalyst sub jected to He pretreatment at 200C followed by PROX reaction at 175C for 1h (under LC and HC conditions) and (3)"PROX20h": the sample pretreated and subjected to the PROX reaction at 175C for 20h (HC). To avoid external influences between sample preparation andanalysis,thecatalystsafterthereac tion were cooled down in helium flow and quickly recovered from the glass wool. All samples were stored in a vac uum vessel until the XPS and FTIR analy seswereperformed.
3.Resultsanddiscussion 3.1.Catalyticactivity

X CO = 1

[CO]out [CO]in

(R6)

Where in and out refer to the inlet and outlet gaseous stream respectively, whereas the selectivity towards CO oxi dationwasestimatedas:

S=

0.5 ([CO]in [CO]out ) [O 2 ]in [O 2 ]out

(R7)

The reaction rate can be calculated by means of the LangmuirHinselwood mechanism proposed by Liu and Flyt zaniStephanopoulos [20] for CO oxida tion over copperceria catalyst, as thor oughlydescribedelsewhere[3]. Moreover, stepresponse experiments were performed in the quartz reactor using different gas mixtures (CO/He,

Fig. 1 shows the temporal variations of CO conversion and selectivity for two CuOx/CeO2 catalysts with different re sidualsilicacontents(8.1and13.3wt.%) at different experimental conditions (low/highconcentrationconditions,175 208C). In spite of their high activity, the slow deactivation of the catalysts is con spicuous, especially at the lowest tem perature. At 208C deactivation seems to produce a slight increase in selectiv ity. However, at lower temperatures

90 selectivity remains practically constant duringthereaction. At 175C the decay inactivityisfast dur ing the initial few hours of reaction and

ArtculoIII then it becomes gradually less marked and steadier. On the other hand, at 208C the initial decay in CO conversion is altogether absent and the subsequent deactivationislesspronounced.

1.0
208C, 8.1 wt% (LC)

CO conversion (XCO)

0.8

0.6

175C, 8.1 wt% (LC)

0.4

175C, 8.1 wt% (HC) 175C, 13.3 wt% (LC)

0.2

1.0
175C, 13.3 wt% (LC)

Selectivity (S)

0.8

175C, 8.1 wt% (LC) 175C, 8.1 wt% (HC)

208C, 8.1 wt% (LC)

0.6 0 5 10 15 Time (h) 20 25

Fig. 1.Catalyticactivity(XCO)andselectivity(S)valuesofCuOx/CeO2catalystswithdiffer ent amounts of residual silica, 8.1 wt.% (CuCe1b) y 13.3 wt.% (CuCe1a) evaluated at differenttemperaturesandconcentrationconditions(LCandHC).

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

91

It can also be seen from Fig. 1 that the residual silica content has a negative effectonthecatalyticactivity.Thiscould be attributed to silica that is difficult to dissolve [15]. The silica in this catalyst seems to be interacting with the active phase in such a way that it is difficult to removecompletely,becauseofthelarge amount of residual silica still present after two days of alkaline washing (>8wt.%). Techniques for obtaining a higherdegreeofsilicaremovalarebeing currentlystudied. A comparison of the slopes of conver sion curves shown in Fig. 1 proves that the initial deactivation rate at 175C is higher under HC conditions than under LC conditions. As can be seen in the figure, for short reaction times (12h) the difference in activity is already significant. The constant selec tivity values through the reaction at 175C (Fig. 1) is another argument in support of the fact that both CO oxida tion and H2 oxidation take place at the same active centres [17]. It seems clear that the lower selectivity at HC condi tionsisduetotheeffectofthegascom position on the oxidation kinetics of CO andH2. In the light of these results, the follow ing sections will focus on the causes of thedeactivationoftheCuCe1bcatalyst.
3.2.Deactivation

3.2.1.Stepresponseexperiments

Fig. 2 shows the results of the step response experiments carried out with the CuCe1b catalyst at 175C. After helium pretreatment at 200C, the first stage consists of a 1hlong PROX reac tion at LC conditions in order to deter mine the initial conversion degree reached by each sample and to assure that all the step response experiments arecarriedoutundersimilarexperimen talconditions.Thesecond stage consists of an 18hlong treatment with a gas mixture that may be the PROX stream, CO in He, CO+H2 in He or CO2+H2 in He. Finally, the third stage consists of a PROXreactionlasting23htodetermine the degree of deactivation after the different treatments of the previous stage. As can be observed in the figure all the nonPROX steps in the second stage produce only a slight decrease in the activity of the catalyst in stage 3, which corresponds to a ~10% reduction of the CO conversion obtained at the end of stage 1. The PROX step in stage 2 provokes a reduction in CO conversion of around 30%. This must be ascribed to the presence of oxygen in the reaction mixture, which is absent in the other steps of stage2. When CO2 is passed through the catalyst bed in stage 2, no consumption of this gas is observed so that CO2, by itself, is unable to oxidise the catalyst surface or to form surface carbon species. On the other hand,

92 4.7molCO/molcat (molcat = molCu + molCe) were consumed in step 2 when the mix ture of 303 ppm of CO in He was passed through the reactor (3.1molCO/molcat in the presence of H2) and were formed 4.3molCO2/molcat (2.8molCO2/molcatin the presenceof H2). If CO were reducing the oxidation state of the catalyst the maximum amount of CO2 formed should be below 2 molCO2/molcat, so that at least part of the CO2 detected must be originated by other reaction. Since the molar ratio of COconsumedperCO2formedis closeto one, the formation of the latter must be ascribedtoinsignificantandunavoidable oxygen inlets (<15 ppm) through the
Stage 1 (PROX) 1.0
PROX (300 ppm CO + 300 ppm O2 + 1%H2) 303 ppm CO

ArtculoIII siliconequartz soft joints in the reactor, as a result of which a small fraction of COisoxidized.However,thehighercon sumption of CO with respect to forma tion of CO2 suggests the potential occur rence of some coke deposits via the Boudouard equilibrium (2CO C+CO2) and/or the formation of surface carbon ates. The imbalance between CO con sumption and CO2 formation [(molCO molCO2)/molcat] in stage 2 can be quanti fied as: 0.24 mol/molcat (CO+H2), 0.39mol/molcat(CO)and1.42mol/molcat (PROX). It seems clear that the forma tion of carbon deposits as coke or car bonates is enhanced under PROX condi tions.

175C Stage 3 (PROX)

0.8

300 ppm CO + 1%H2 308 ppm CO2 + 1%H2

0.6 XCO

Stage 2

0.4

0.2

0.0 0 5 10 Time (h) 15 20

Fig.2.Stepresponseexperimentsat175CwiththeCuCe1bcatalyst.

Fig. 3 shows the reaction rate curves, at 200C (LC), of a catalyst (CuCe2) sub

jected to different initial and intermedi ate treatments. In the first group of pre

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

93

treatments (ac), carried out at T300C overdifferentsamplesofthesame cata lyst, the reaction rate curves are similar, with slight variations as a function of time, temperature and gas concentra tion during the pretreatment. In the second group of treatments the catalyst is first heated in air at 400C (d) and then subsequently subjected to increas ing temperature treatments (e)(g) un derH2.Theresultsclearlyshowthatpre treatment in air at 400C (d) has a nega tive effect on the reaction rate, as com pared to treatments at lower tempera
(a) 300C-5%H2-1h
1.6E-05

tures, that might be attributed to active phase sintering, this effect being even more noticeable in the final treatment (g) at 500C. It should be reminded that the catalysts were prepared by a silica template technique, so that sintering of the nanoparticles during the synthesis was prevented by the silica template, in spite of the relatively high calcination temperature (550C) employed. There fore, once the silica was removed the resultingnanoparticles becameproneto suffer sintering at temperatures lower thanthecalcinationtemperature.

(b) 200C-He-1h
1.4E-05 1.2E-05

(c) 200C-5%H2-6h

CuCe2 PROX at 200C (LC conditions)

rPROX, molg s

-1 -1

1.0E-05 8.0E-06 6.0E-06 4.0E-06 2.0E-06 0.0E+00 0 10 20 30 40 50 60 70 80 90

(d) 400C-Air-1h + (e) 300C-5%H2-1h + (f) 400C-5%H2-1h + (g) 500C-5%H2-1h

time (h)

Fig.3.Reactionrateat200CfortheCuCe2catalystafterithasbeensubjectedtodiffer enttreatments.

The treatments at 300C in 5% H2 for 1h (a,e) seem to produce a temporary en hancement of the reaction rate com pared to similar treatments at lower

(b,c) or higher (f) temperatures, al though for high reaction times the curves obtained after treatments at T300C tend to equalize, regardless of

94 the type of pretreatment (ac). The use of hydrogen at this specific temperature (300C) might either reduce the active phase to some extent or clean the sur face of detrimental species (e.g. carbon ates), without promoting the sintering that might be occurring at higher tem peratures. Discussion of the surface analysesofthesamplessubjectedtothe PROX reaction will enable us to reexamine Fig. 3 in the light of the new datapresentedbelow.
3.2.2.Surfaceanalysis(XPSandFTIR)

ArtculoIII the samples, peak assignments were performed as indicated in the figure and exhaustively discussed in the Supporting Information, based on literature data. Briefly,intheC1sregion,peaksat286.9 287.3 eV (peak maxima for all samples) are attributed to CO, COH and COC bonds [2123], whereas peaks at 288.3 288.9eVare associated toC=Obondsof varying strength [2325]. In the case of adventitiouscarbon(284.6eV),although different approaches can be found in literature (see Supplementary informa tion)webelievethatwhentheanalytical conditionsareidentical,thevariationsin peak areas between different samples should be ascribed to real differences in the samples rather than different de grees of contamination. In the O1s re gion, the peak at~529.5 eV is attributed to oxygen in copper or cerium oxides [2629], whereas the peak at 531.6 eV can be ascribed to SiO [30, 31], C=O [21, 32] and CO [21] bonds. The as signments of the Ce3d region are taken from Qiu et al. [33] and permit to calcu late the amount of Ce4+ in the sample as indicatedintheSupplementaryInforma tion. For the Cu2p3/2 region the peak at ~933 eV is associated to Cu+ species, whereas the peak at 933.6934.3 eV is attributed to Cu2+ species [20, 34]. Apeaklocatedat930eVisattributedto highly dispersed Cu2+ ions from surface Ce4+O2Cu2+pairs(seeSI).

Asdescribedintheexperimentalsection XPS and FTIR were performed exsitu. The samples to be analysed were pro tectedfrom externalinfluencesby keep ing them in a vacuum vessel between reaction and analysis. Exsitu FTIR per mits the analysis of the stable species formed on the surface of the catalysts (deactivating agents) when insitu tech niques, such as DRIFTS, are not avail able. The raw curves obtained from XPS and FTIR analyses were subjected to deconvolution procedures as described in the experimental section to deter mine the contribution of different spe cies to the global curve, as usually done for XPS. Fig. 4 shows an example of the deconvolution procedures performed with the XPS spectra (O1s spectral re gion for the PROX1h samples). In order to determinethe surface composition of

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

95

HC (O1s) HC (peaks) LC (O1s) Intensity (a.u.) LC (peaks) Metal oxides CO C=O SiO2 Charge effect

O1s PROX-1h

524

526

528

530 B.E. (eV)

532

534

536

Fig.4.XPSdeconvolutionresultsforO1sspectralregion(PROX1hsamples).

Basedonthedescribedstrategy,Table 2 shows the surface composition of CuCe1b catalysts after each reaction stage. The combined analysis of FTIR and XPS results will help to identify the causes of deactivation, that is suspected to be due to the formation of coke and other carbon compounds (analysed by means of surface carbon analysis) or active phase sintering and oxidation states modification (studied by the analysis of metallic phases and concen tration).
Surface carbon (combined XPS and FTIR results): As can be observed in Table 2 the surface concentration of adventi

tious carbon is very high in all the ana lyzed samples (a large proportion of this may be from contamination during the analysis) and follows a regular trend, clearly increasing after 20 h of PROX reaction at HC. The possible formation of coke deposits during the reaction has alreadybeensuggestedinthediscussion on the stepresponse experiments at LC conditions. It seems logical to assume that at HC conditions the formation of cokedepositsviatheBoudouardequilib rium is even more likely. This might ex plain the higher deactivation rate ob served at HC conditions(Fig.1)although this is not the most probable explana tionaswillbediscussedbelow.

96
Table2.SummaryofXPSsurfacecompositionofsamples(atom%). Species (B.E.,eV) CARBON(C1s) Sample Fresh

ArtculoIII

PROX1h PROX1h PROX (HC) (LC) 20h(HC)

AdventitiousCarbon Nonadvent.Carbon;CO Nonadvent.Carbon;C=O

COPPER(total)(Cu2p3/2) Cu DispersedCu2+ CuOclusters CERIUM(total)(Ce3d) Ce4+ Oxygentononadventitiouscarbon atomratio,O/C** Coppertometalsatomratio, Cu/[Ce+Cu]
+

30.43 1.43 0.72 2.28 0.56 0.36 1.36 7.71

30.97 1.38 0.69 1.79 0.09 0.49 1.19 7.85

29.32 0.26 0.13 2.19 0.00 0.50 1.65 8.26

36.40 2.50 1.30 1.45 0.05 0.16 1.21 4.74

6.24 6.32 7.90 3.33 (81.0%)* (80.4%)* (95.7%)* (70.3%)* 2.9 0.23 3.7 0.19 2.9 0.21 2.9 0.23

*%Ce4+intotalcerium **Oxygen concentration evaluated from peak at ~532 eV in O1s region, once corrected for oxygen in SiO2. Carbon concentration corresponds to non adventitious carbon (CO andC=O) The fresh sample also displays a small amount of non adventitious carbon in the form of CO and C=O type species, whose surface concentration clearly increases after 20h of PROX reaction at HC conditions but decreases after only 1h of PROX reaction at LC conditions. Such behaviour should be reflected in the FTIR spectra obtained for the differ ent samples (Fig. 5, HC conditions). Thus,themajorpeaksofFig.5shouldbe attributed to monodentate [3537] and bidentate carbonates [17, 37, 38] and carboxylates [3537] as summarised in Table 3, although an unambiguous iden tification of the peaks is difficult to ob tain since several compounds share peakswiththesamefrequencies.

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

97

PROX 1h

1511

1588

Intensity (a.u.) .

1301 1263

1384 1369

1453

1415

Intensity (a.u.)

Bidentate carbonates

Si-O

1200

1300
Carbonates

1400

1500
-1

1600

1700 PROX-20h PROX-1h Fresh

Wavenumber (cm )
SiO-H Formates

400

800

1200

1600

2000

2400
-1

2800

3200

3600

4000

Wavenumber (cm )

Fig. 5. Infrared spectra for CuCe1b after the different reaction stages at HC (Inset: de convolutionexample).

Copper carbonates (or hydroxides) were not detected by XPS on any of the ana lyzed samples (Table 2), therefore both carboxylates and/or carbonates must have been formed on the cerium oxide surface. The small peaks located at 1263cm1 and in the ~28502950 cm1 range (Fig. 5) are attributed to asym metric strecthing COC vibration [39, 40] and cerium formates (both biden tate and bridged) [41, 42] respectively. No metallic hydroxyls (isolated bands between3600and3700cm1),carbonyls (2100 cm1 [17]) or molecular water (in tense peak at 1640cm1 [43]) were de tectedonthesamples.

The evolution of the peaks associated to carbonates and carboxylates can be followed in Fig. 5 (a mathematical ap proach can also be found in the supple mentary information). The fresh sample exhibits peaks that are characteristic of monodentate and bidentate carbonates (850, 1370 cm1). A small fraction of carboxylates cannot be disregarded, although their most representative fre quencies (1305, 1510 cm1) are also dis played by carbonates and, furthermore, the O/C ratio for the fresh sample (see SI) takes the value of 2.9, close to that expected for carbonates (~3). After 1h under PROX at HC conditions, the areas of these peaks are slightly affected and

98 only small changes can be detected by FTIR (Fig. 5 and SI), although a clear in crease is observed after PROX20h. The originof carbonateson thefreshsample must be in the action of atmospheric gasesonthesurfaceofthecatalystafter calcination. In principle, they could sur viveboththepretreatmentinhelium at 200C and the PROX reaction at 175C, accordingtoseveralauthors[36,44].On the other hand, carboxylates do not survive under PROX conditions [36], which could explain the initial decrease of the areas of some peaks partly origi nated by this species (~1510 and ~1305cm1). The values of Table 2 show that the PROX1h sample obtained at

ArtculoIII low concentration conditions (LC) pre sentsverysmallsurfaceamountsofCO and C=O functionalities, as shown in Table 2 and Fig.4. This suggests that at the beginning of the PROX reaction a fraction of the carbonates and all car boxylates which are initially present on the surface of the fresh sample are re moved, at least under LC conditions. However, this is not in conflict with the postulates of the above mentioned au thors [36, 44], since they used HC condi tions in their works. Obviously the re moved carbonates are of a more labile nature than those formed during the reaction otherwise, they would have survivedthereactionconditions.

Table3.FTIRpeakassignment(placedincellsindescendingorderofintensity). Potentialassignment Ourvalues* Literaturevalues

Monodentatecarbonate

1370 1450 ~1510** 850 ~1305 1588

~1305 ~1510

13511367[35,37] 1464[36] 1517[35] 854[17],850[38] 1297[17] 1586[17] 1316[37] 1510[37][36]

Bidentatecarbonate

Carboxylate

* The values of the peaks in this work correspond to the values of the individual deconvolved curves,whiletheliteraturevaluescorrespondtotheapparentmaximuminthespectrum. **Initalicsthepeaksthatcanbesharedbydifferentspecies.

This seems to indicate that there is an equilibrium of formation and decompo sition of the different species involved (monodentateandbidentatecarbonates andcarboxylates).

To summarise, the surface of the fresh sample contains monodentate carbon ates, bidentate carbonates and carboxy lates. In view of the similar intensities of themostintensepeakscorrespondingto

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

99

thebidentate(1588cm1)andmonoden tate (1370 and 1510 cm1) carbonates, both types of species should be present on the ceria surface in similar propor tions.Eitherduringthepretreatmentor under the PROXconditions the carboxy lates are removed. The labile surface carbonates initially produced from at mospheric exposure, probably on iso lated ceria sites, that survive the helium pretreatment at 200C [36, 44], are eliminated at the beginning of the PROX reaction at LC conditions. Afterwards, at HC conditions new carbonates are formed on the catalyst surface. How ever, at LC conditions the low CO con centration in the stream limits the for mation of new carbonates, which there fore takes place more slowly than under HC conditions (Table 2). Consequently, the deactivating effect of carbonates at the beginning of reaction is less con spicuous than under HC conditions (Fig. 1). The carbonates formed during reac tion are more strongly held on the cata lyst surface than those produced by atmospheric exposure on the fresh sample,so thatthey arenotremovedas they are formed. As will be shown later these carbonates are produced from CO chemisorptiononinterfacialceriumsites in close contact with copper sites, this being the potential cause for their higher stability. The removal of the car bonates at the start of the reaction may takeplacebyoxidativedecomposition,a

common occurrence with cerium for mates and cerium monodentate car bonates at low temperature (180C), as described by Karpenko et al. [45]. These authors associated the presence of car bonates with that of Ce3+. The oxidative removal of the carbonates produced Ce4+ and CO2. The same association can beobservedintheCuCe1bcatalyst.Fig. 6 shows the relation between the con centration of surface carbonates (C=O species evaluated by XPS; Table 2) and the concentration of surface Ce+3, both relative to total cerium concentration. Ascanbeobserved,therelationisclear. The [C=O]/Ce3+ molar ratio for all the samples is in the 0.360.92 range, below the value which one might expect for pure cerium(III) carbonate (1.5) but rep resenting an important fraction of sur facecoverage. Asdescribedabove,theFTIRshoulderat 1263 cm1 (Fig. 5), corresponds to the COCvibration(Table3).Thispeakwas notdetectedintheFTIRspectrumofthe fresh sample, so its origin might be the oxidation of coke deposits generated during the reaction. The peaks associ atedtoceriumformates(bothbidentate and bridged, ~28502950 cm1) increase continuously during reaction. These compounds were identified as interme diates in the PROX reaction in former works, as they are products specific to the reaction between CO and surface hydroxyls[36].Theabsenceofhydroxyls

100 (bands in the 36003700 cm1 region) explains the insignificant presence of formates on the catalyst surface. Hy droxyls have been detected by MartnezArias et al. [17] via insitu DRIFT analyses so that they should be attributed to reaction intermediates rather than to deactivating species. Ad ditionally, a certain amount of water is present during the reaction (selectivity <1), which is known to slow down the formation of formates on the ceria sur face[46].
0.35 PROX-20h (HC) 0.30 0.25 PROX-1h (HC) / total Ce 0.20 Fresh 0.15 0.10 0.05 0.00 0.00

ArtculoIII significantly higher than the surface copper concentration of the PROX1h sample achieved under HC conditions (Table2).Inthiscase,thelossofsurface copper cannot be due to thermal sinter ing,asthiswouldalsoaffect the catalyst under LC conditions. The higher concen tration of surface cerium carbonates in the sample obtained at HC conditions would seem to be a more logical expla nation for its lower surface copper con centration.Furthermore,thisdiminution of surface copper must be associated to the CuO clusters and not to the highly dispersed Cu2+ species, whose concen tration is independent of the reaction conditions, as can be observed in Table 2 for both PROX1h samples. The pro moting effect of CuO on the formation of carbonates in interfacial CuO/CeO2 sites has already been suggested by MartnezAriasetcol.[47]. From Table 2 it can be seen that the fresh catalyst has a significant Cu+ con tent (~25% of total copper) and a smaller content of dispersed Cu2+ spe cies (~16%), the rest being small CuO clusters distributed over the ceria sur face.Fromprecedentdiscussionandthe results presented in Table 2 it can be deduced that during the pretreatment inheliumat200Cmostofthedispersed Cu2+ species are aggregated into small CuO clusters (partial sintering), whereas the Cu+ species remain stable (the re duced copper species are stable on the

Ce

3+

PROX-1h (LC)

0.10

0.20

0.30

surface C=O / total Ce

Fig. 6. Ratio of Ce3+ concentration (rela tive to total Ce) to concentration of C=O type surface species (relative to total Ce).

Metallic species. The surface copper concentrationinthefreshsampleisvery similar to that of the PROX1h sample obtained under LC conditions, although

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

101

ceria surface without forming solid solu tions [26]). Thus, aggregation of dis persed Cu2+ during pretreatment is inferred from the fact that it already takes place at a lower temperature (175C), as proven (a) by the diminution of dispersed Cu2+ concentration during the reaction and (b) by the increased concentration of CuO clusters in the PROX1h sample (LC), in which the ef fects of carbonate formation on the copper concentration are minimal. Tak ing into consideration the mechanisms reported by MartnezArias et al. [16] and Manzoli et al. [19], a redox equilib rium is established at the beginning of the reaction at 175C so that dispersed Cu+ is transformed by oxidation with lattice oxygen into dispersed Cu2+ (Ce4+O2Cu2+; reaction R5). In fact, the sum of the concentrations of both spe cies (dispersed Cu+ and Cu2+) in the PROX1h samples is practically equal to the Cu+ concentration in the fresh sam ple (0.500.56 vs. 0.57 atom%). Assum ing that after 1 h's reaction the concen trationsofsurfacespeciesareatequilib rium [19], then reaction R1 must be the slowest step in the mechanism, taking into account the low concentration of reducedcopperinthePROX1hsamples. Both highly dispersed Cu2+ species and CuO clusters in close contact with ce rium oxide participate in the equilibrium (R1). These data suggest that R1 reac tion is the controlling step of the PROX

mechanism. The decrease in reduced copper is more evident at LC conditions, under which Cu+ completely disappears after 1 hour's reaction (Table 2). This is in agreement with a lower CO concen tration in the gas stream and, therefore, alowerreactionrateforR1. Fig. 7 shows Wagner's diagram for cop per [48], which plots the binding energy of maximum intensity in the Cu2p3/2 region versus the kinetic energy of maximumintensityintheCuLMMregion (Auger spectrum). Briefly, this diagram allows us to visualize the degree of elec tronic interaction / dispersion of the copperspecieswith/onthecerialattice (which increases when we descend par allel to the dashed lines) and of the oxi dation state of the copper species (which increases when we move to the left and perpendicular to the dashed lines). Fig. 7 shows the data obtained in this work (black squares) and the data points taken from a work by Gamarra et al.[18],whichareusedasreferences.As can be observed in the diagram, the point corresponding to the fresh sample is situated in an intermediate zone be tween the lines representative of Cu2+ and Cu+, confirming the high content of reduced copper in this sample. On the other hand, the points for the PROX1h samples are situated right on the Cu2+ line, in agreement with their highly oxi dized state (Table 2). The rather eccen tric location of the point corresponding

102 to the PROX20h sample (it should be a priori on the Cu2+ line) might have been caused by a certain oxygen defect in the surface CuO clusters (CuO1x), and it is directly related to the shift of the bind

ArtculoIII ing energy for the Cu2+ peak in the Cu2p3/2 region to a lower value (933.6eV in the PROX20h sample with respect to 934.0934.3 eV in the PROX 1hsamples).
919
Cu(0) bulk 5%CuO/CeO2 (IW of CeO2 prepared by microemulsion, Gamarra et al., 2007) CuO bulk

918

PROX 20h (HC)

Cu2O bulk

917 E.K. (CuLMM ) (eV)

PROX-1h (LC) 916 PROX-1h (HC) Fresh 915

Cu2+ ZSM-5

914
Cu2+ line Cu+ ZSM-5 Cu+ line

936

935

934 B.E. (Cu2p3/2) (eV)

933

913 932

Fig. 7.Wagner'sdiagramshowingtheevolutionofcopperspeciesduringthePROXreac tion.CirclesrepresentdatatakenfromaworkbyGamarraetal.[18].

The point corresponding to the fresh sample is located in a zone of the dia gram considerably lower than the point

for bulk CuO. This implies that there is a high degree of electronic interaction of copper with the cerium oxide as a con

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

103

sequence of the high dispersion of spe cies (dispersed Cu+ + dispersed Cu2+ + small CuO clusters). The point for the PROX1h sample (HC) has a similar ordi nate to that of the fresh sample, con firming that the dispersed Cu+ in the fresh sample has been transformed into dispersed Cu2+. Thus these species must be responsible for the high degree of electronic interaction between copper and cerium oxide. The point for the PROX1h sample (LC) is in a somewhat higher position, as a consequence of the higher concentration of "visible" CuO clusters (unaffected by the interfacial carbonates),whichinteractmoreweakly than dispersed Cu2+ with cerium oxide. Finally the point for the PROX20h sam ple is located in a higher zone of the diagram,consistentwiththepresenceof CuO clusters and, especially, with the lower concentration of dispersed Cu2+. Even so, this point lies below those cor responding to the cerium oxide nanoparticles prepared by microemul sion that have been wet impregnated with copper salts [18]. This evidences the high dispersion of copper obtained bytheSACOPtechnique. Ceriumoxideundergoessomereduction during the reaction under HC conditions (the percentage of Ce4+ decreases from ~80% in the PROX1h sample to ~70% in

the PROX20h sample). As mentioned before the reduced cerium species are related to the formation of surface car bonates during the reaction. On the otherhand,thesurfaceconcentrationof cerium diminishes notably for 20 h in the PROX reaction (from 7.8 to 4.7atom%), which is coincident with a parallel increase in adventitious carbon in ~6atom%. It is thus evident that coke deposits have accumulated on the ceria surface during the reaction, since the decrease in surface copper concentra tion is less conspicuous and its causes have been clearly established in the preceding discussion. Besides, it has recentlybeenobservedthattheamount ofcokedepositedonCuO/CeO2catalysts during the steam reforming of methanol is inversely related to their copper con tent[49].
3.3.Deactivationmechanism

In light of the results obtained in this work we postulate a deactivation mechanism during PROX at tempera tures below 250C comprising two dif ferent submechanisms whose relative importance is a function of (1) the time on stream, (2) the temperature and (3) the reaction conditions (LC or HC). Table 4 lists the different mechanisms, and shows how they affected by these pa rameters.

104
Table4.DeactivationmechanismsatT<250CinCuCe1b. Deactivation mode

ArtculoIII

Metallic spe Temperature* Timeonstream* ciesinvolved

CO concen tration*

Ce3+ formed Carbonate forma during reac tion(maincauseof tion (in con Decreases deactivation) tact with CuO clusters) Aggregation of 2+ dispersed Cu to Dispersed CuO clusters (par Cu2+ tialsintering)

Takesplacein15 Increases h at HC and more deactiva slowlyatLC tionrate

Increases. It is Takes place until Appar hardly signifi disappearance of ently not cantat175C dispersedCu2+ affected

*Variationofextentofdeactivationwiththeincreaseineachvariable stantially vary (Table 2). It is well known 3.3.1.Cokeeffect that the transport of oxygen towards In our opinion, although the formation the interfacial centres of reaction of coke deposits revealed by XPS analy (Cu/Ce) occurs by a MvK mechanism sis (Table 2) provokes a decrease in the through the crystal lattice of cerium surface concentration of ceria, it does oxide [19], so that as long as there is not really contribute to a higher deacti some free surface ceria to link the gase vation of the catalyst. If this were the ous oxygen, its subsequent ionic trans case, it would affect more clearly the port through the lattice should not be shape of the CO conversion curve at HC seriously affected by the presence of conditions displayed in Fig.1. In this coke. experiment about half of the lost cata lytic activity took place in the first hour 3.3.2.Deactivationbycarbonates of reaction, during which the value of kL Carbonates are the main cause of deac decreased ~25% (from 1.94104 to tivation and they are also behind the 1.47104 molCOgcat1s1bar0.15), differences apparent in Fig. 1 between whereas after 20 h the value of kL was the CO conversion curves at different ~52% of the initial one concentration conditions (LC and HC) (1.01104 molCOgcat1s1bar0.15). How during the first hours of reaction. Ini ever, the surface concentration of coke tially the fresh sample holds some labile in the first hour of reaction did not sub carbonates which are associated to the

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

105

reduced cerium species (Ce3+) on the catalyst surface. During the first mo ments of reaction at 175C there is a rearrangement of the cerium oxide sur face,causedbytheoxidativeelimination of carbonates. Parallel to this fact, the PROX reaction starts to take place and, simultaneously, deactivation by newly formed carbonates helps to slow it down. Since the results of the step re sponse experiments indicate that both oxygen and CO are needed to form the interfacial carbonates, we postulate the following deactivation submechanism (R3b) to be added to the reaction scheme proposed by MartnezArias et al.[16]: Ce4+O2Cu+CO Ce3+CO3Cu2+ (R3b) Thus, at high CO concentration condi tions,equilibriumR4isshiftedtotheleft (formation of Ce4+O2Cu+CO), so that reaction R3b is favoured from the be ginning, whereas at low CO concentra tion conditions, reaction R4 is initially favouredoverR3b. Nevertheless, as R3b is irreversible, a sufficient reaction time mustlead to a similar amount of surface carbonates under both reaction condi tions (Fig. 1). Thus, carbonates prevent or delay the oxygen activation per formed by reactions R3R4, as already suggestedbyManzolietal.[19]. These carbonates provoke also a loss of surface CuO clusters, as well as a small loss of surface cerium [~0.4 mol% in the

first hour of reaction, estimated by the difference in the amount of surface ce rium between the PROX1h samples (LC and HC; Table 2)], though the main cause of diminution of surface cerium during the reaction continues to be the accumulationofcokedeposits. Obviously an increase in temperature produces a decrease in the formation of surfacecarbonates,notonlyduetotheir enhanced lability, but also because of the reduced availability of CO as a con sequence of the PROX reaction. This explains the lower deactivation slope of the CO conversion curve at ~200C (Fig.1).
3.3.3. Deactivation by aggregation of dispersedCu2+(partialsintering)

The slow sintering (aggregation) of highly dispersed Cu2+ to form CuO clus ters at 175C is evident in the Wagner's diagram (Fig. 7), and in the conspicuous decrease of its concentration during the reaction (Table 2), which, like the de crease in the surface concentration of CuO clusters, is unrelated to the parallel formation of carbonates. Extended sin tering of these clusters is possibly be hind the high degree of deactivation observed at temperatures over ~300C (Fig. 3). Thus, the highly dispersed Cu2+ ions must be the only protagonists of the potential deactivation ascribed to sintering at T<300C. However, it is doubtful that the sintering of this spe

106 cies influences the PROX reaction at 175C.Inthestepresponseexperiments at 175C it was concluded that the pres ence of oxygen was needed in order to provoke catalyst deactivation through the formation of surface carbonates. Consequently, the deactivation pro duced during stage 2 (Fig. 2) in the ab sence of oxygen must be mainly as cribed to the sintering of dispersed Cu2+ species (although the small leak of oxy genmightcontributetotheformationof some deactivating carbonates). Thus, the results in Fig. 2 show that under low concentration conditions the loss of conversion produced by deactivation in 19 h of reaction at 175C is 29% of the initial conversion value, to which sinter ing contributes no more than ~10% and carbonate formation no less than ~19%. On the other hand, under high concen tration conditions (Fig. 1) sintering causes the same degree of deactivation as under LC conditions, whereas the formation of carbonates causes a loss of CO conversion, after 19 h, of no less than 25% of the initial value of conver sion. Therefore, at 175C after 19 h of reaction, carbonate formation contrib utes to deactivation at least between 2 (LC) and 2.5 (HC) times more than the sinteringofdispersedCu2+species. At 200C the sintering of dispersed Cu2+ can be expected to be more obvious, though inany case it iscompensated for a decrease in the formation of carbon

ArtculoIII ates,whichresultsinalowernetdeacti vation rate than that observed at 175C (Fig. 1). It should also be taken into ac count that the highest catalytic burden is carried by the CuO clusters, and not by the dispersed Cu2+ ions, whose lower reducibility makes them a priori a less activespecies[18]. Another unexplained phenomenon is the hysteresis observed in non isothermal PROX cycles, whose trend varies as a function of the reaction con ditions. Thus, under low concentration conditions (LC),the curve oftheCO con version at increasing temperatures is formedbyhigherconversionvaluesthan that obtained at decreasing tempera tures[8].Exactlytheoppositeisthecase athighCOconcentrationconditions(HC) [19]. This opposite behaviour can be better visualised in Fig. 8. The cause of hysteresis under LC conditions reported previously [8] was of a kinetic nature, so that copper oxide reduction (the cause of deactivation according to Marbn et al. [8]) proceeded more rapidly at de creasing temperatures. Surprisingly, catalyst reduction was also the cause advanced by Manzoli et al. [19] to ex plain the opposite hysteresis in the PROX cycle at HC conditions, in which theincreaseandsubsequentdecreasein temperature led to higher CO conver sion values being observed during the coolingdown.

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

107

Manzoli et al. (2005)

atesisquitefast,theeffectofwhichwill be more apparent at the lowest tem peratures.

b

XCO

HC conditions
Marbn et al. (2005) XCO

LC conditions
Temperature

On reaching a temperature of around 250C (highest temperature in the PROX cycles analyzed by Manzoli et al. [19]) the surface carbonates decompose, favouring higher CO conversion values during the subsequent cooling down phase compared to those obtained dur ingtheheatinguppartofthecycle.

Fig. 8. Schematic hysteresis cycles of conversion found during PROX reaction underLC[8]andHC[19]conditions.

We believe that the deactivation mechanisms described in this work ex plain the different hysteresis trends reasonably well. The sintering of dis persed copper species is the most prob able cause of hysteresis in the PROX cycle at LC [8], since under these condi tions carbonate formation is much slower and therefore it would produce the opposite hysteresis only if each point of the reaction cycle were evalu ated after a sufficiently long reaction time (several hours), a precaution that was not taken in the previous work by Marbn et al. [8]. Thus, the sintering of part of the active phase, which occurs mainly at the highest temperatures, explainsthehysteresisin thePROX cycle at LC conditions. On the other hand, at HC conditions the formation of carbon

The combined deactivation mechanisms also contribute to an understanding of the influence that some of the treat ments have on the activity curves, at 200C, displayed in Fig. 3. Thus, treat ment (a) (300C5%H21h) causes a cer tain reduction of the catalyst which in turns leads to an enhancement of its initial reactivity (higher concentration of Cu+/Ce3+ pairs). However, the slow but continuous formation of carbonates during the first 40 h of reaction allows the activity curve to converge with that ofthecatalystsubjectedtothestandard pretreatment (b) (200CHe1h). In the caseofpretreatmentwithH2atalower temperature [(c) (200C5%H26h)] the reducing strength of this gas is minimal (unlike CO [17], this being the key for thehighselectivityofthesecatalysts),so that the lower initial reactivity with re spect to that observed after the stan dard pretreatment (b) occurs only be cause the copper is subjected to a

108 longer period of sintering during pre treatment.

4.Conclusions

ArtculoIII temperature, unlike the other deactiva tion mechanism, which is simply the aggregation of dispersed Cu2+ to CuO clusters (partial sintering). At tempera tures over 300C deactivation is proba bly produced by extended sintering of the active phase of the catalyst. These mechanismscanalsobeappliedtoother CuO/CeO2 systems and may help to ex plain the hysteresis commonly observed with conversion in nonisothermal activ ityexperiments.
Acknowledgements

We have analyzed the catalytic activity, in the PROX process, of CuOx/CeO2 cata lysts prepared by silica aquagel confined coprecipitation. At low concentration conditions their catalytic activity is simi lar to some of the most active catalysts described in the literature. However thesecatalystssufferacertaindegreeof deactivation when subjected to high concentration conditions. We have ana lyzed the surface composition of the catalysts and identified the different speciesonthemduringthereactionasa previous step to formulating a deactiva tion mechanism. Although coke deposi tion occurs during the reaction, this does not affect catalytic activity. Deacti vation at temperatures below 250C is produced mainly by carbonate forma tion, and is inversely dependent on

Funding through the Spanish National Project MAT200500262 and the FICYT Regional Project (IB05001) is acknowl edged. IL thanks the Ministerio de Cien cia e Innovacin of Spain for her FPI grant and TVS thanks the Junta General del Principado de Asturias for the award of a Clarn postdoctoral contract funded byPCTI20062009.

References

[1] G.Marbn,T.ValdsSols,Int.J.HydrogenEnergy32(2007)1625. [2] G.Marbn,T.ValdsSols,Int.J.HydrogenEnergy33(2008)927. [3] I.Lpez,T.ValdsSols,G.Marbn,Int.J.HydrogenEnergy33(2008)197. [4] D.H.Kim,J.A.Cha,Cat.Lett.86(2003)107. [5] G.Sedmak,S.Hocevar,J.Levec,J.Catal.213(2003)135. [6] N. Y. Usachev, I. A. Gorevaya, E. P. Belanova, A. V. Kazakov, O. K. Atalyan, V. V. Kharlamov,Russ.Chem.Bull.53(2004)538.

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

109

[7] Y. Tanaka, T. Utaka, R. Kikuchi, K. Sasaki, K. Eguchi, Appl. Catal. A: Gen. 238 (2003)11. [8] G.Marbn,A.B.Fuertes,Appl.Catal.B:Environ.57(2005)43. [9] Z.Liu,R.Zhou,X.Zheng,J.Mol.Catal.AChem.255(2006)103.

[10] Z.Liu,R.Zhou,X.Zheng,J.Mol.Catal.AChem.267(2006)137. [11] P.K.Cheekatamarla,W.S.Epling,A.M.Lane,J.PowerSources147(2005)178. [12] Y.Z.Chen,B.J.Liaw,H.C.Chen,Int.J.Hydrog.Energy31(2006)427. [13] G.Avgouropoulos,T.Ioannides,Appl.Catal.B:Environ.67(2006)1. [14] G.Marbn,T.ValdsSols,A.B.Fuertes,Cat.Lett.118(2007)8. [15] G.Marbn,A.B.Fuertes,T.ValdsSols,Microp.Mesop.Mat.112(2008)291. [16] A. MartinezArias, M. FernandezGarcia, O. Galvez, J. M. Coronado, J. A. Ander son,J.C.Conesa,J.Soria,G.Munuera,J.Catal.195(2000)207. [17] A. MartinezArias, A. B. Hungria, G. Munuera, D. Gamarra, Appl. Catal. B: Envi ron.65(2006)207. [18] D.Gamarra, A. Horns, Zs. Koppany,Z. Schay, G. Munuera, J. Soria, A.Martnez Arias,J.PowerSources169(2007)110. [19] M. Manzoli, R. Di Monte, F. Boccuzzi, S. Coluccia, J. Kaspar, Appl. Catal. B: Environ.61(2005)192. [20] W.Liu,M.FlytzaniStephanopoulos,J.Catal.153(1995)317. [21] H.Darmstadt,C.Roy,S.Kaliaguine,Carbon32(1994)1399. [22] L.Q.Chu,X.N.Zou,W.Knoll,R.Forch,Surf.Coat.Technol.202(2008)2047. [23] M. P. Casaletto, L. Lisi, G. Mattogno, P. Patrono, G. Ruoppolo, Appl. Catal. A: Gen.267(2004)157. [24] D. G. Georgiev, R. J. Baird, G. Newaz, G. Auner, R. Witte, H. Herfurth, Appl. Surf. Sci.236(2004)71. [25] W.Wang,J.Zhang,H.Huang,Z.Wu,Z.Zhang,Appl.Surf.Sci.253(2007)5393. [26] B.Skarman,D.Grandjean,R.E.Benfield,A.Hinz,A.Andersson,L.R.Wallenberg, J.Catal.211(2002)119. [27] X.W.Yu,G.Q.Li,J.AlloyCompd.364(2004)193. [28] Q.Yang,P.X.Yan,J.B.Chang,J.J.Feng,G.H.Yue,Phys.Lett.A361(2007)493. [29] E.Cano,C.L.Torres,J.M.Bastidas,Mater.Corros.52(2001)667. [30] M. Wang, B. Gong, X. Yao, Y. Wang, R. N. Lamb, Thin Solid Films 515 (2006) 2055.

110

ArtculoIII

[31] M.Annaka,C.Yahiro,K.Nagase,A.Kikuchi,T.Okano,Polymer48(2007)5713. [32] J.Kiss,K.Revesz,F.Solymosi,Surf.Sci.207(1988)36. [33] L.M.Qiu,F.Liu,L.Z.Zhao,Y.Ma,J.N.Yao,Appl.Surf.Sci.252(2006)4931. [34] E.S.Bickford,S.Velu,C.S.Song,Catal.Today99(2005)347. [35] O. S. Alexeev, S. Krishnamoorthy, C. Jensen, M. S. Ziebarth, G. Yaluris, T. G. Ro berie,M.D.Amiridis,Catal.Today127(2007)189. [36] O.Pozdnyakova,D.Teschner,A.Wootsch,J.Krohnert,B.Steinhauer,H.Sauer,L. Toth,F.C.Jentoft,A.KnopGericke,Z.Paal,R.Schlogl,J.Catal.237(2006)1. [37] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc. , Faraday Trans.1Phys.Chem.CondensedPhases85(1989)929. [38] Y. Denkwitz, A. Karpenko, V. Plzak, R. Leppelt, B. Schumacher, R. J. Behm, J. Ca tal.246(2007)74. [39] C.Zhang,J.Wyatt,D.H.Weinkauf,Polymer45(2004)7665. [40] J.Lichtenberger,M.D.Amiridis,Catal.Today98(2004)447. [41] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc. , Faraday Trans.1Phys.Chem.CondensedPhases85(1989)1451. [42] T.Shido,Y.Iwasawa,J.Catal.136(1992)493. [43] A.Fidalgo,L.M.Ilharco,J.NonCryst.Solids347(2004)128. [44] F.Gu,Z.Wang,D.Han,C.Shi,G.Guo,Mater.Sci.Eng.,B139(2007)62. [45] A.Karpenko,R.Leppelt,J.Cai,V.Plzak,A.Chuvilin,U.Kaiser,R.J.Behm,J.Catal. 250(2007)139. [46] K.G.Azzam,I.V.Babich,K.Seshan,L.Lefferts,J.Catal.251(2007)153. [47] A.MartinezArias,A.B.Hungria,M.FernandezGarcia,J.C.Conesa,G.Munuera, J.PowerSources151(2005)32. [48] G.Moretti,J.Electron.Spectrosc.Relat.Phenom.95(1998)95. [49] Y.Men,H. Gnaser,R.Zapf,V.Hessel,C.Ziegler,G.Kolb,Appl.Catal.A:Gen.277 (2004)83.

111

ArtculoIII.Informacinsuplementaria
Silica Aquagel Confined Co Precipitationtechnique(SACOP)
The silica source (sodium silicate, Al drich, 27% SiO2+14% NaOH) and deion ised water were poured under stirring into a Teflon autoclave. In a different vessel the metallic nitrates (15mol% Cu and85mol% Ce) weredissolvedina4 M HNO3 (or HCl) solution and the resulting mixture was added dropwise to the Tef lon autoclave containing the sodium silicate suspension under stirring. The amounts of each precursor were ad justedinordertoestablishthefollowing molar compositions in the synthesis suspension: SiO2/H+/H2O = 1/6.54/193.9 and Si/M = 4 (M = total moles of metals in the final metallic oxides). The closed autoclave was left under stirring for 24h at RT (hydrolysis step), and then placed in an oven at 100C for 24h in order to producethegel(condensationstep).Co precipitation of the metallic cations con fined in the water phase of the resulting gel was carried out with powdered NaOH. The slurry formed after the co precipitation stage was subsequently subjected to several water washing and centrifugation steps, and finally it was dispersed in acetone to replace the re maining water, vacuum filtered and left to dry at RT overnight. The resulting silicametal hydroxide composites were calcined in air at 550C for 4h. Finally, the silica matrix was removed by disso lution in NaOH (2M) in an autoclave at ~30Coveraperiodof12days.

Identification of peaks in the XPS spectra

Figure SI1 shows some examples of the deconvolution procedures performed with the different XPS spectral regions. In order to determine the surface com position of the samples peak assign ments were performed as indicated in thefigure. C1s region. All XPS peaks were success fully identified except those situated at 281283 eV (binding energies) in the C1s region. These peaks are commonly as signed to carbide species [13], and in deed were thus assigned by Men et al. [4] in a work on methanol reforming (SRM) by CuO/CeO2/Al2O3 catalysts. However, in our work these peaks (one or two) appeared in all the samples, at a constant ratio of about 19% of surface atoms.Some carbidespeciesmight have been formed during the PROX reaction, as Men et al. [4] suggested for SRM, but there is no sound explanation for their

112 presence in the fresh sample. Reassign ing these peaks to adventitious carbon (284.6) would cause an unacceptable shift in the binding energies of the other spectral regions of more than 23 eV withrespecttotheliteraturevalues.The presence of these peaks might have been originated by a charging effect [5], though this explanation is too often taken as valid without any experimental support.Inthissituationwedecidednot to consider these peaks, since their con stant contribution to the surface com position leaves the relative concentra tions of the rest of species practically unaffected. In this region the possible overlapping oftheCe4speakat ~290 eV should always be taken into account (Figure SI1). Otherwise the carbon spe cies might be overestimated, as might already have occurred in certain revised works [6, 7]. The adventitious carbon peak (reference peak at 248.6 eV) is often arbitrarily assumed to be pro duced by secondary causes such as con tamination from the oil pump (which does not form a part of our equipment) or by carbon deposits, either elemental or hydrocarbons, that were originally present on the samples. This is why not everybody agrees on its use as a refer ence [8]. We believe that when the ana lytical conditions are identical, the varia tions in peak areas between different samples should be ascribed to real dif

ArtculoIIISI ferences in the samples rather than dif ferent degrees of contamination. Peaks at 286.9287.3 eV (range of peak maxima for all samples) are attributed to CO, COH and COC bonds [911], whereas peaks at 288.3288.9 eV are associated to C=O bonds of varying strength[1113]. O1s region. The shoulder at ~526.5 eV (figure SI1) has a binding energy that is too low to be assigned a priori to any oxygen species. In the literature it has been attributed to a variety of causes: chargingeffect[14,15];asatelliteofthe X ray source [16]; speculative "active" oxygen atoms associated to ions in a high oxidation state [17]; oxygen ions in crystal lattices with a high electronic density [18] or subsurface oxygen in K2O multilayers [19]. The supporting evidence for attributing this peak to oxygenspeciesisveryweak,soweshare the most widely accepted view i.e. that of the charging effect. The peak at ~529.5 eV is attributed to oxygen linked to copper or cerium [2023], whereas the peak at 531.6 eV can be ascribed to SiO bonds [24, 25], C=O bonds (car bonates, C=O or COOR) [9, 26] and CeOH [21] or CuOH [27] bonds. CO bonds (COH, COC) appear at some what higher binding energies (around 533.5eV [9]), so that they must be in cluded in the wide peak at a high bind ingenergy(FigureSI1).

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

113

LC (O1s)

Charge effect

CC CH

C1s PROX-20h (HC)

Intensity (a.u.)

LC (peaks) HC (O1s) HC (Peaks)

O1s PROX-1h

Intensity (a.u.)

CO C=O Ce4s

Metal oxides

Charge effect

CO C=O SiO2

277

280

283

286 289 B.E. (eV)

292

295

524

526

528

530 B.E. (eV)

532

534

536

Satellites CuO Intensity (a.u.)

Dispersed Cu

2+

Cu2O

Unknown Cu species

Cu2p3/2 PROX-1h (HC)

928 932 936 B.E. (eV)
v0 u'1 v2 v1

924

940

944

948

Intensity (a.u.)

Ce3d5/2 (u0, u1, v0, v1,v2) Ce3d3/2 (u'0, u'1 v'0, v'1,v'2) v0 u1 v1 v2 u0

u0

%Ce

4+

= 100

A(v 2) 0.14 A total

Ce3 PROX-1h

876

881

886

891

896 901 B.E. (eV)

906

911

916

921

Figure SI1. Examples of XPS spectral regions analyzed to determine the surface com positionofthesamples.

114 TheO/Cratiocanbeestimatedfromthe XPS peaks placed at ~287 eV (CO in C1s), 288289 eV (C=O in C1s) and 531.6eV (O1s, after subtraction of the SiO2oxygen)
Ce3d region. Complete peak assign mentsintheCe3dregionaredepictedin Figure SI1 and can be found in a work by Qiu et al. [28]. In our analysis the most interesting peak is v2, situated at ~917 eV, because it is completely iso lated from the other peaks and there fore the evaluation of its area is unam biguous (Figure SI1). For pure CeO2 particles the value of this area amounts to 14% of the total area of the Ce3d region, whereas it is zero for pure Ce2O3[20, 29]. Thus, the percentage of Ce4+ in a sample (with respect to total cerium) can be estimatedasindicated in FigureSI1. Cu2p3/2 region. This region displays four main peaks and two satellites at high binding energies (Figure SI1). The peak at 932.4933.1 eV is assigned ei ther to Cu+ species [30, 31] or to ele mental copper [30]. According to some authors [32, 33] these oxidation states can be distinguished from each other by the modified Auger parameter (sum of binding and kinetic energies), whose value is 1850.8 eV for Cu and 1848.3 eV for Cu2O. For the CuCe1b catalyst, the modified Auger parameter has values in the 1848.11848.8 eV range, so that the

ArtculoIIISI peak at ~933 eV must be associated to Cu+species. The peak at 933.6934.3 eV is attributed to Cu2+ species [30, 34]. The small peak situated at 936.0937.3 eV has no spe cific assignment. It could be CuF2 [35], but this species is absent in our cata lysts.FurthermoretheCu(OH)2orCuCO3 species present peaks at lower values of B.E. (935 eV [36]). Since the proportion of this peak is insignificant we have de cidedtoleaveitunassigned.Information on the copperceria interaction can also be obtained via XPS. Thus the ratio of thesatellitepeakareastothemainpeak areas is Is/Im = 0.55 for pure CuO, whereas it is in the range of 0.350.40 for solid solutions of Cu2+ in ceria [34] (this is only applicable when the sample is free of Cu(0) or Cu+ species, whose presence also causes a diminution of Is/Im[20]). The peak at 930 eV is the cause of the main interpretation difficulty. According to the literature it could be produced either by a charging effect [14], related to the O1s peak at ~527 eV, or by the presence of Cu2+ species which are highly dispersed on the ceria surface [30].Inourspectrathepeaksat930and 527 eV do not seem to show any clear relation (peak area ratios [527 eV]/ [930eV] = ~13, ~2 and ~10 for the fresh, PROX1h and PROX20h samples, respectively). Nevertheless, assuming

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

115

that the charge effect gives rise to the peak at 930 eV, then the values of the Is/Im ratio would be 0.24 (for the fresh sample), 0.34 (for the PROX samples at HC) and 0.55 (for the PROX samples at LC). The low Is/Im value for the fresh sample, with respect to pure CuO, may beduetoitshighCu+content(TableS1; ~30% of total copper) [20], but what could be the reason for the low Is/Im ratio in the PROX samples (HC)? This might be due to the fact that some of the copper forms a solid solution with ceria, but this would also provoke a diminution of FWHM [34] (which is 3.6eV for the PROX samples, the same as that reported for pure CuO [34]). In

addition, considering that the Cu/(Cu+Ce) atom ratio in the fresh sam ple is higher when evaluated by XPS (~0.20; Table 2 in the paper, Table S1) than by EDX (0.09; Table 1 in the paper) it must be concluded that copper oxide is concentrated on the catalyst surface, which is normally the case for catalysts with Cu/(Ce+Cu) ratios below 0.25 [37]. Finally, the presence of surface Ce4+O2 Cu2+pairsfromthehighly dispersedCu2+ ions would lead to a similar decrease in the Is/Im value. It was decided therefore that the peak at 930 eV should be at tributed to highly dispersed Cu2+ ions andnottoanychargingeffect.

116
TableS1.XPSsurfacecompositionofsamples(atom%). Sample Species (B.E.,eV) CARBON(C1s) Fresh PROX1h (HC)

ArtculoIIISI

PROX1h (LC)

PROX 20h(HC)

AdventitiousCarbon Nonadventitiouscarbon CO C=O

OXYGEN(total)

30.43 1.43 0.72 46.54

30.97 1.38 0.69 48.82 2.5

29.32 0.26 0.13 50.68 3.0

36.40 2.50 1.30 43.59 2.0

Oxygen to metals atom ratio, (O1s, 529.5 eV) 1.9 O/[Ce+Cu] Oxygen to nonadventitious carbon atom ratio, (O1s, 2.9 531.62SiTotal)O/Cna SILICON(total)(Si2s) 10.90 COPPER(total)(Cu2p3/2) Cu
+ 2+

3.7 8.50 1.79 0.09 0.49 1.19 0.02 0.25 7.85 6.32 (80.4%)* 0.19 100

2.9 9.16 2.19 0.00 0.50 1.65 0.05 0.42 8.26 7.90 (95.7%)* 0.21 100

2.9 10.03 1.45 0.05 0.16 1.21 0.04 0.30 4.74 3.33 (70.3%)* 0.23 100

2.28 0.56 0.36 1.36 0.00 0.20 7.71 6.24 (81.0%)*

DispersedCu

CuOclusters Cucompound? Is/Im CERIUM(total)(Ce3d) Ce4+

Copper to metals atom ra 0.23 tio,Cu/[Ce+Cu] TOTAL(C+O+Si+Cu+Ce) 100

*%Ce intotalcerium

4+

Identification of peaks in the FTIR spec tra

An example of the deconvolution proce dure performed with the different FTIR peaks is given in Figure 5. The evolution of the peaks from the Fresh sample to

the samples subjected to PROX (1h and 20h) is given in Figure S2. Major peaks are attributed to carbonates (monoden tate and bidentate) as well as to car boxylates as indicated in Table S2. Mi nor peaks in the ~28002950 cm1 range

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

117

are attributed to bridged and bidentate formates. As the values of the peaks of this work are related to the individual peaks that form the curve there are
20 18 Convolved area (n.u.) 16 14 12 10 8 1370 1302 6 4 848 2 0 Minor peaks 0.6 Convolved area (n.u.) 1514 Major peaks

small discrepancies between the values of the apparent maximum (that is the peak usually employed for identifica tion)andthedeconvolvedpeak.
1509 1588
1649 1384

1369

1588

1511 1369

1588

1301

1308

1453

1452

849

850

1421

1627

1263

1608

1469
2925 2853 2962

1409

1384

1660

1263

1415

1628 1660

0.4

2925

2854

0.0

1263

Fresh

2960

1384

PROX-1h Sample

2853

2961

0.2

2925

PROX-20h

Figure SI2. Evolution of the peak area from FTIR of samples Fresh, PROX1h (HC) and PROX20h. The numbers in the figure indicate the position of the maximum of each peak.

118

ArtculoIIISI

TableS2.FTIRpeakassignment(placedincellsindescendingorderofintensity). Majorpeaks (cm1) Minorpeaks (cm1) Potentialassignment Peaksandref.

~1510
1588

Monodentatecarbonate Carboxylate Bidentatecarbonate Carboxylate Bidentatecarbonate Carboxylate Monodentatecarbonate Monodentatecarbonate Bidentatecarbonate

1517[38] 1514[7]*1510[39] 1586[40]* 1598[38] 1297[40]* 1316[7]*[39] 1464[7]* 1367[38],1358[7]* 854[40]*,850[41]* 1263[42],1260[43] 2920[44]* 2936[7]* 2933[41]* 2960[44]* 2950[7]* 2850 [44]*, 2848 [45]*,2856[46]* 2848[7]*, 2845[41]* 2844[7]*

~1625 ~1660

~1305
1450 1370 850

~1415 ~1384
1263

COCvibration Formate (CHvibration) Bidentateformate Bridgedformate Formate (CHvibration) Bridgedformate Formate (CHvibration)

2925

~2960

2853

Bidentateformate Bridgedformate

*Studiesoverceriumoxide
References

[1] I.Dontas,S.Kennou,DiamondRelat.Mater.10(2001)13. [2] A. C. Anacleto, N. Blasco, A. Pinchart, Y. Marot, C. Lachaud, Surf. Coat. Technol. 201(2007)9120. [3] M.Xiang,D.Li,H.Qi,W.Li,B.Zhong,Y.Sun,Fuel86(2007)1298.

Marbnetal.Appl.Catal.A:Gen.361(2009)160169

119

[4] Y.Men,H. Gnaser,R.Zapf,V.Hessel,C.Ziegler,G.Kolb,Appl.Catal.A:Gen.277 (2004)83. [5] M. M. Titirici, A. Thomas, S. H. Yu, J. O. Muller, M. Antonietti, Chem. Mater. 19 (2007)4205. [6] S.Patel,K.K.Pant,J.PowerSources159(2006)139. [7] O.Pozdnyakova,D.Teschner,A.Wootsch,J.Krohnert,B.Steinhauer,H.Sauer,L. Toth,F.C.Jentoft,A.KnopGericke,Z.Paal,R.Schlogl,J.Catal.237(2006)1. [8] P.Swift,Surf.InterfaceAnal.4(1982)47. [9] H.Darmstadt,C.Roy,S.Kaliaguine,Carbon32(1994)1399.

[10] L.Q.Chu,X.N.Zou,W.Knoll,R.Forch,Surf.Coat.Technol.202(2008)2047. [11] M. P. Casaletto, L. Lisi, G. Mattogno, P. Patrono, G. Ruoppolo, Appl. Catal. A: Gen.267(2004)157. [12] D. G. Georgiev, R. J. Baird, G. Newaz, G. Auner, R. Witte, H. Herfurth, Appl. Surf. Sci.236(2004)71. [13] W.Wang,J.Zhang,H.Huang,Z.Wu,Z.Zhang,Appl.Surf.Sci.253(2007)5393. [14] C. Lamonier, A. Bennani, A. DHuysser, A. Aboukais, G. Wrobel, J. Chem. Soc. , FaradayTrans.92(1996)131. [15] M. Yousuf, A. Mollah, T. R. Hess, Y. N. Tsai, D. L. Cocke, Cem. Concr. Res. 23 (1993)773. [16] B. E. Koel, G. Praline, H.I. Lee, J. M. White, R. L. Hance, J. Electron. Spectrosc. Relat.Phenom.21(1980)31. [17] F. Solymosi, J. Rasko, E. Papp, A. Oszko, T. Bansagi, Appl. Catal. A: Gen. 131 (1995)55. [18] J.M.Sohn,M.R.Kim,S.I.Woo,Catal.Today83(2003)289. [19] B. Lamontagne, F. Semond, D. Roy, J. Electron. Spectrosc. Relat. Phenom. 73 (1995)81. [20] B.Skarman,D.Grandjean,R.E.Benfield,A.Hinz,A.Andersson,L.R.Wallenberg, J.Catal.211(2002)119. [21] X.W.Yu,G.Q.Li,J.AlloyCompd.364(2004)193. [22] Q.Yang,P.X.Yan,J.B.Chang,J.J.Feng,G.H.Yue,Phys.Lett.A361(2007)493. [23] E.Cano,C.L.Torres,J.M.Bastidas,Mater.Corros.52(2001)667. [24] M. Wang, B. Gong, X. Yao, Y. Wang, R. N. Lamb, Thin Solid Films 515 (2006) 2055. [25] M.Annaka,C.Yahiro,K.Nagase,A.Kikuchi,T.Okano,Polymer48(2007)5713.

120 [26] J.Kiss,K.Revesz,F.Solymosi,Surf.Sci.207(1988)36.

ArtculoIIISI

[27] I.LouisRose,C.Methivier,J.C.Vedrine,C.M.Pradier,Appl.Catal.B:Environ.62 (2006)1. [28] L.M.Qiu,F.Liu,L.Z.Zhao,Y.Ma,J.N.Yao,Appl.Surf.Sci.252(2006)4931. [29] J. Z. Shyu, K. Otto, W. L. H. Watkins, G. W. Graham, R. K. Belitz, H. S. Gandhi, J. Catal.114(1988)23. [30] W.Liu,M.FlytzaniStephanopoulos,J.Catal.153(1995)317. [31] Y.Z.Chen,B.J.Liaw,C.W.Huang,Appl.Catal.A:Gen.302(2006)168. [32] X.M.Liu,G.Q.Lu,Z.F.Yan,Appl.Catal.A:Gen.279(2005)241. [33] W.L.Dai,Q.Sun,J.F.Deng,D.Wu,Y.H.Sun,Appl.Surf.Sci.177(2001)172. [34] E.S.Bickford,S.Velu,C.S.Song,Catal.Today99(2005)347. [35] C.D.Wagner, W.M.Riggs,L.E.Davis,J.F.Moulder,G.E.Muilenberg,HandbookofX Ray Photoelectron Spectroscopy. Perkin Elmer Coorporation, Eden Prairie, Minn.,1979. [36] B.Li,J.H.Qu,H.J.Liu,C.Z.Hu,Chemosphere68(2007)2153. [37] D. Gamarra, A.Horns,Zs. Koppany,Z. Schay,G. Munuera,J. Soria,A. Martnez Arias,J.PowerSources169(2007)110. [38] O. S. Alexeev, S. Krishnamoorthy, C. Jensen, M. S. Ziebarth, G. Yaluris, T. G. Roberie,M.D.Amiridis,Catal.Today127(2007)189. [39] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc. , Faraday Trans.1Phys.Chem.CondensedPhases85(1989)929. [40] A. MartinezArias, A. B. Hungria, G. Munuera, D. Gamarra, Appl. Catal. B: Envi ron.65(2006)207. [41] Y. Denkwitz, A. Karpenko, V. Plzak, R. Leppelt, B. Schumacher, R. J. Behm, J. Catal.246(2007)74. [42] J.Lichtenberger,M.D.Amiridis,Catal.Today98(2004)447. [43] C.Zhang,J.Wyatt,D.H.Weinkauf,Polymer45(2004)7665. [44] S. F. Maria, L. M. Russell, B. J. Turpin, R. J. Porcja, Atmos. Environ. 36 (2002) 5185. [45] T. Tabakova, F. B. Boccuzzi, M. Manzoli, D. Andreeva, Appl. Catal. A: Gen. 252 (2003)385. [46] K.G.Azzam,I.V.Babich,K.Seshan,L.Lefferts,J.Catal.251(2007)153.

121

ResumenartculoIV. COPPERMANGANITEASACATALYSTSFORTHEPROXREACTION. DEACTIVATIONSTUDIES

ManganitadecobrecomocatalizadorparalareaccinPROX. Estudiosdedesactivacin
Comosehamencionadoconanterioridad,laestabilidaddeuncatalizadorconeltiempo de reaccin es un parmetro de gran importancia a la hora de su aplicacin prctica. En estetrabajonosplanteamoselanlisisdelascausasdeladesactivacindelamanganita de cobre, CuMn2O4, sintetizada por el mtodo de coprecipitacin confinada en hidrogeles de slice, en la reaccin de oxidacin preferencial de CO. Para tal fin, adems de los experimentos de reaccin PROX de larga duracin, se llevaron a cabo ensayos de respuesta a escaln (CO/He, CO+H2/He, CO2+H2/He) para determinar el efecto desactivante de los distintos gases implicados en la reaccin PROX, de la misma manera enlaqueseprocediconelcatalizadorCuOx/CeO2. En condiciones de alta concentracin la espinela de cobre y manganeso mostr una mayor actividad en la reaccin PROX que el catalizador CuOx/CeO2. Para CuMn2O4 todos los tratamientos de respuesta a escaln dieron lugar a una desactivacin superior a la que se produjo en condiciones PROX, al contrario de lo que ocurra con el catalizador CuOx/CeO2. La secuencia de gases de mayor a menor poder desactivante result ser: CO/He > CO+H2/He > CO2+H2/He> PROX. Los balances de materia llevados a cabo con los datos de los experimentos de respuesta a escaln mostraron la formacin de especiescarbonosasenlasuperficiedelcatalizador,formacinqueseviofavorecidapor la presencia de oxgeno en la mezcla de gases. Sin embargo, la influencia de estas especiescarbonosasenlaactividaddelamanganitadecobrenofuesignificativa. Los resultados de FTIR mostraron que durante la reaccin PROX a 175C se produca la evolucin de los carbonatos desde formas menos estables a ms estables. El anlisis conjunto con los balances de CO2 indic que, adems de esta evolucin, se form una fraccin significativa de nuevos carbonatos. La formacin de carbonatos monodentados fue corroborada por los resultados de XPS, la relacin O/C se increment de valores 1 (valor esperado para los carbonatos bidentados) a 2.5 (valores esperados para carbonatosmonodentados).Aunqueestoscarbonatosnoafectaronaladesactivacin,s influyeron en la concentracin relativa de los distintos estados de oxidacin del

122

ArtculoIV

manganeso. Por otro lado, los anlisis XPS descartaron la presencia de carbonatos de cobre. Combinando todos los resultados obtenidos, se observ que al comienzo de la reaccin elmanganesosereducaenunamayorproporcinqueelcobre,sibienelmanganesose reoxidaba parcialmente a lo largo de la reaccin. La reduccin inicial no poda estar provocada por la corriente gaseosa, ya que el cobre, siendo ms susceptible de ser reducido que el manganeso, se reduca en menor grado. Este resultado sugiri que la reduccin inicial del manganeso estaba relacionada con la presencia de carbonatos bidentados y de tipo puente en la muestra fresca. Estos carbonatos casi desaparecieron despus de 1 h de reaccin en las condiciones de reaccin PROX, un proceso que sucedi paralelamente a la reduccin del manganeso. Por esta razn se propuso un mecanismo para la descomposicin reductiva de los carbonatos tipo puente y bidentadosdemanganesoqueexplicaraladisminucindeloscarbonatosylaevolucin delosestadosdeoxidacindelmanganeso. Por otro lado, considerando las concentraciones relativas de Mn2+, Mn3+ y Mn4+, Cu+/Cu2+ despus de 20 h de reaccin PROX, la desactivacin de la espinela de cobre y manganeso podra ser atribuida a la progresiva reduccin del cobre en la superficie del catalizadorporlaaccindelCO,comoyasugirieronotrosautoresconanterioridad.

123

ArtculoIV.

1.Introduction Hydrogen streams for fuelling PEM fuel cellsneedtobealmosttotallyCOfreein order to avoid platinum poisoning. For this reason purification processes, such as preferential CO oxidation, need to be included in the H2 production process in order to achieve CO concentrations be low 20 ppm. Copperbased catalysts have shown good catalytic activity and selectivity for CO oxidation in rich hy drogen streams (PROX reaction) [1, 2].

Recent works on the PROX reaction with copperbased catalysts show that the use of nanosized catalysts improves the activityandselectivityofthereaction[1, 3, 4]. CuMn2O4, technically known as hopcalite, is a well known oxidation catalyst [57] which is employed to re move air pollutants such as carbon monoxide and nitrous oxide from ex haust gas. In CO oxidation, its high activ ity is usually attributed to the resonance system Cu2+Mn3+/Cu+Mn4+ and the high

124 adsorptionofCOonCu2+/Mn4+andofO2 on Cu+/Mn3+ [8, 9]. Recently, CuMn2O4 has been applied in the catalytic steam reforming of methanol [10, 11] and pre ferential oxidation processes [12], with moderate activity and high selectivity at temperatures over 100C. Recently de veloped template methods for synthe sising high surface area nanocatalysts have been used to promote a significant increase in their catalytic activity com pared to that of their low surface area counterparts. In this area of research, we have demonstrated that the silica aquagel confined coprecipitation method (SACOP) is a useful tool for syn thesizing high surfacearea CuMn2O4 nanocatalysts providing improved cata lytic activity for the methanol steam reformingprocess[13]. In this work we analyse the activity and stability of nanosized CuMn2O4 synthe sised by SACOP for the PROX reaction. The stability of the catalyst is a parame ter that is often disregarded in PROX studies [1], and yet it is of paramount importance for longterm onboard ap plications. We have also conducted sur face studies to unravel the causes of the slow deactivation appreciated in the catalytictestsperformed. 2.Experimental 2.1.Catalystssynthesis CuMn2O4 nanocatalysts were synthe sised following the SACOP procedure

ArtculoIV [14]. Basically, Cu(NO3)23H2O and Mn(NO3)26H2O at a 1:2 molar ratio were dissolved in HNO3. This solution was added dropwise to a mixture of sodium silicate and water at a final con centrationof SiO2:H+:H2O= 1:6.54:193.9 andSi/(Cu+Mn)=4.Afterhydrolysis(24h at room temperature) and condensation (24 h at 100C) a silica aquagel contain ing widely dispersed metallic cations was formed. To prevent the cations from leaching during washing, precipita tion inside the gel was performed using powdered NaOH. The slurry was repeat edly washed and dispersed in acetone, then filtered, dried at RT and calcined at 550C(5C/min,4h).Ascalcinationtakes placeinsidetheporesofthesilicamatrix (confined process) growth of the nano materials is restricted and consequently high values of surface area can be achieved [10]. The silica matrix was fi nallyremovedby NaOH dissolution(2M) at~30Coveraperiodoftwodays. Forcomparisonreasons,CuO/CeO2cata lysts were synthesized following the same procedure. A detailed description of the characteristics and the behaviour of these catalysts under PROX reaction canbefoundelsewhere[4].

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

125

2.2.CatalystsCharacterization The composition of the catalyst was evaluated by SEMEDX analysis (DSM 942, Zeiss). XRD was performed in a Siemens D5000 diffractometer to obtain the phase composition. The crystal size of the catalysts was determined by ap plying Scherrers equation. The BET sur face area was evaluated by N2 phy sisorption at 196C in an ASAP 2010 analyser(Micromeritics). Exsitu Xray photoelectron spectros copy (XPS) was carried out by means of a Specs spectrometer, using MgK (1253.6 eV) radiation emitted from a double anode at 50W. The binding en ergies of the spectra obtained were cor rected using the binding energy of ad ventitious carbon (284.6 eV) in the C1s region. The backgrounds were sub tracted by means of Shirley baselines, except for the Cu2p3/2 region (linear baseline). All the analyzed regions were deconvolved by mixed Gaussian Lorentzian functions (90:10). For the quantitative analyses the atomic sensi tivity factors stored in the CasaXPS da tabase(v2.3.12Dev6)wereused. Exsitu FTIR spectra of the materials compressed in discs with KBr were re corded on a Nicolet Magna IR560 spec trometer tted with a DTGS KBr absorb ance detector. The analyzed regions were deconvolved by mixed Gaussian Lorentzianfunctions(90:10).

2.3. Catalytic activity tests and step responseexperiments Catalytic activity tests were performed during the PROX reaction by using a quartz reactor of 7 mm internal diame ter inserted inside a vertical furnace. The catalyst sample (1530 mg) was diluted in inert SiC particles (285 270mg) and placed in the reactor be tween two zones of glass wool, thus ensuring isothermal and plugflow con ditions. Any water present in the exit gases was removed with Mg(ClO4)2. AnalysesofH2,CO,CO2,O2andN2inthe product gases were then carried out using a gas chromatograph (HP 6890) equippedwithaTCdetector. Twodifferentsetsofreactionconditions were applied: (a) "LC" conditions (low concentration): 300 ppm CO + 300 ppm O2 + 1% H2 in He (total flow rate= 300mL/min, STP) and (b) "HC" condi tions (high concentration: 0.5 vol% CO + 0.5vol% O2 + 25% H2 in He (total flow rate=200 mL/min, STP). In both cases the values of the spatial velocities were very high (LC: 0.4105molCOgcat1s1, HC:2.3105 molCOgcat1s1). Before the reactions each sample was pretreated in a helium flow at 200C for 1h. Long duration isothermal experiments were performedforallsamples,butinspecific cases nonisothermal experiments at decreasing temperatures and pseudo

126 steady state conditions (~12 h per tem peraturepoint)werealsocarriedout. TheconversionofCOwasevaluatedas:

ArtculoIV

X CO = 1

[CO]out [CO]in

(R1)

where in and out refer to the inlet and outlet gaseous stream respectively. The selectivity towards CO oxidation was estimatedas:

S=

0.5 ([CO]in [CO]out ) [O2 ]in [O2 ]out

(R2)

The reaction rate was calculated by means of the LangmuirHinselwood mechanism proposed by Liu and Flyt zaniStepahopoulus [15] for CO oxida tion over a copperceria catalyst, as de scribedelsewhere[1]. Additionally, stepresponse experi ments (LC conditions) were performed in the quartz reactor using different gas mixtures(300ppmCO/He,300ppmCO+ 1% H2/He and 300 ppm CO2+ 1% H2/He) in order to assess the deactivating po tentialofthe gasesinvolvedinthePROX reactionoverthedilutedcatalysts. XPS and FTIR analyses were carried out on the catalyst samples which were la belled as: (1)"Fresh": the fresh sample (as synthesised catalyst), (2) Pre treated: the catalyst subjected to He pretreatment at 200C (3)"PROX1h": the catalyst subjected to He pre treatment at 200C followed by the

PROX reaction at 175C for 1h (HC con ditions) and (4)"PROX20h": the sample pretreated and subjected to the PROX reaction at 175C for 20h (HC). To avoid external influences between the prepa ration of each sample and its analysis, the catalysts were cooled down after thereactioninaheliumflowandquickly retrieved from the glass wool. All the samples were stored in a vacuum vessel until the XPS and FTIR analyses were performed. Exsitu techniques permit to determine the presence of stable spe cies formed on the surface of the cata lysts that are the cause of deactivation, although no information for intermedi ate reaction species is given. Exsitu techniques for the study of deactivation have been successfully employed in a previouswork[4].
3.Resultsanddiscussion

As mentioned in previous papers [13, 14], the SACOP technique is a simple procedure that employs lowcost pre cursorsforobtainingmetaloxideswitha high specific surface area. Under the synthesis conditions reported above, CuMn2O4 is the predominant phase as observed by XRD [13]. A crystallite size (dXRD) of 4 nm was obtained by applying Scherrers equation. The N2 isotherm gives a surface area value of SBET=296 m2/g. The EDX of this sample shows that the silica content is negligi ble. TEM images of the CuMn2O4 nano

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

127

materials can be found elsewhere [13]. These images show that the manganites exhibit a peculiar structure made up of thinsheets (ofthickness <3nm),someof whicharerolledup.Thecatalyticactivity of CuMn2O4 in conditions of high con centration (HC) and decreasing tem peratures is plotted in Fig. 1, and com pared to the activity of CuO/CeO2 which wasalsosynthesisedbySACOP.Itcanbe observed that CuMn2O4 is more active than CuO/CeO2 throughout the whole
1.0

temperature range. However, these catalysts suffer from a slow but clearly visible deactivation. In Fig. 2 the results of the long term PROX experiments at 175Candhighconcentrationconditions are plotted for both the CuMn2O4 and CuO/CeO2 catalysts. The deactivation of CuO/CeO2catalystsinthePROXreaction hasbeenthoroughlyanalysedelsewhere [4] and, at low temperature (175C), is mainly caused by the formation of car bonatesonthesurfaceofthecatalysts.
1

XC O

0.8

0.8

0.6

0.6 S

0.4 CuMn2O 4 0.2 CuO x /CeO 2

0.4

0.2

0.0 50 100 150 T, C T (C) 200

0 250

Fig. 1. Values of CO conversion (circles) and selectivity (squares) under HC conditions and decreasing temperatures for CuOx/CeO2 (empty symbols) and CuMn2O4 (solid sym bols)synthesisedbySACOP.

Stepresponse experiments were carried out over the CuMn2O4 catalysts at 175C in order to obtain more detailed

information about the deactivation mechanism. As indicated in the experi mental section, after a pretreatment

128 and the first PROX step (low concentra tion: 300 ppm CO+300ppm O2+1% H2) for 1h (stage 1) the catalysts were sub jected to different gas streams for a ~18hperiod(stage2).Finallyastandard PROX stream (LC) was passed through the catalysts to evaluate the activity of
1.0 0.8 0.6 XCO 0.4 0.2 0.0 0 10 20 TOS (h) 30 CuO x /CeO 2 CuMn2O 4

ArtculoIV thecatalystsafterstage2.Theresultsof these experiments are plotted in Fig. 3. It is immediately clear that all the treat ments employed in stage 2 are more deactivating than the PROX conditions, deactivation following the sequence: CO>CO+H2>CO2+H2>PROX.

40

Fig. 2. Longterm catalytic activity at 175C of CuOx/CeO2 (triangles) and CuMn2O4 (cir cles)catalystsathighconcentration(HC)conditions.

In the absence of oxygen, the presence ofCO hasa negativeeffectontheactiv ity of the catalysts. Although no oxygen was originally added to the gas streams we detected small leakage of oxygen entering through the junctions of the reactor. It is this oxygen that is partly responsible for the formation of CO2 during stage 2. Thus, the formation of

CO2 may proceed via any one of the followingreactions: CO+O2CO2 CO+MoxOCO2+Mred 2COC+CO2 (R3) (R4) (R5)

where Mox and Mred are oxidised and reduced forms of the metals present on thecatalystsurface.

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

129

1.0

1 PROX

0.8

3 0.6 XC O 0.4 0.2

2 CO CO + H2 CO 2 + H2 0 5 10 TOS (h) 15 20 25

0.0

Fig. 3. Step response experiments with CuMn2O4 nanocatalysts at 175C [stage 1: PROX; stage2:CO(squares),CO+H2(triangles),CO2+H2(stars);stage3:PROX]. Table 1 contains the mass balances to CO and CO2 for all the stepresponse experiments. For the experiments per formed in the absence of CO2, the vari able CO2 to balance represents the imbalance of CO2 due to the fact that all the CO consumed was used to produce CO2 in accordance with reactions (R3) and (R4). This imbalance is indicative of the formation of carbonaceous species (either elemental carbon, R5, or carbon ates) on the surface of the catalyst. The CO2 to balance observed during the PROX process is at least 10 times higher than that of the stepresponse experi ments. This indicates that the formation of surface carbonaceous species is fa voured when oxygen is present in the reaction stream, as in the case of the CuOx/CeO2 catalysts [4]. The main dif ference between the CuOx/CeO2 and CuMn2O4 nanocatalysts seems to be related to the deactivation caused by the formation of carbonaceous species on the surface. As stated before, the presence of carbonates in CuOx/CeO2

130 leads to a significant deactivation [4] whereas for CuMn2O4 the formation of carbonaceous species produces hardly anydeactivation.Thus,thereisnodoubt that during PROX the formation of car bonaceousspeciesispromoted(Table1) but deactivation occurs to a lesser ex tent than in the case of the other treat ments (Fig. 3). In short, the results in Table 1 and Fig. 3 lead to the conclusion that the deactivation of the CuMn2O4 nanocatalysts is due to the effect of CO, but not through the formation of carbo naceous species. In addition, the pres ence of hydrogen in stage 2 does not

ArtculoIV seem to increase deactivation. On the contrary it has a beneficial effect on catalytic activity (Fig. 3). From Table 1, it can be seen that, when H2 is introduced together with CO, a smaller amount of CO2 is formed than when only CO is in troduced. It must therefore be assumed that hydrogen competes with CO for adsorption on the active centers prone to be deactivated by CO (possibly via reaction R4). Indeed it is well known that the reduction ability of CO is greater than that of H2 at moderate temperatures[16].

Table 1. Mass balance of CO and CO2during the step response experiments (PROX, LC Conditions:300ppmCO,300ppmO2,1%H2).
Stage2 PROX(LC) 300ppmCO 300ppmCO+ 1%H2 300ppmCO2 +1%H2 COinlet (mol) COoutlet (mol) COcon sumed (mol) CO2inlet (mol) CO2outlet (mol) CO2to balance (mol)

9.41103 4.07103 3.73103 0

1.32103 3.58103 3.40103 0

8.09103 4.82104 3.28104 0

0 0 0 4.93103

7.35103 4.13104 3.15104 4.88103

7.4104* 6.9105* 1.3105* 5.0105

*AssumingthatalltheCOconsumedformsCO2inaccordancewithreactionR3 Fig. 4 shows the FTIR spectra of the CuMn2O4 nanocatalysts after the suc cessive treatments described in the experimental section. Here it can be seen that there is a broad band in the 28003600 cm1 range whose area de creases during the pretreatment and subsequent PROX reaction. This band can be attributed to nonstable hydrox yls (usually around 36003700 cm1 with associated bands in 30003800 cm1) [17]. The peaks in the 400550 cm1 range are attributed to MnO6 octahe dral units and those at ~600 cm1 to CuO4units [18]. None of the peaks can be attributed to carbonyls (2000 2100cm1).

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

131

Fresh

PROX-20h Fresh PROX-1h Pretreated Pre-treated Fresh

1100 1200 1300 1400 1500 1600 1700 1800

I (a.u.)

PROX-20h PROX-1h

400

800

1200

1600

2000

2400
-1

2800

3200

3600

4000

W avenumber (cm )

Fig. 4.FTIRspectraofCuMn2O4fresh, pretreatedunderHelium, after1hPROXandafter 20hPROX.

The main differences between the dif ferent spectra appear in the 1200 1800cm1 region, which is amplified in the inset of Fig. 4. As the FTIR measure ments were conducted under identical conditions, for an equal period of time, and on tablets with the same mass of catalyst, we may apply a semi quantitative analysis of the peaks in this spectral zone, by a deconvolution pro cedure of the type usually employed withXPS and FTIR[4]. Therelative varia tions of the FTIR peaks in the 1200 1800cm1rangeareplottedinFig.5.The peaks located at around 13401363, 14181421 and 14841488 cm1 increase

continuouslyduringthereactionandare to be attributed to monodentate car bonates [17, 19, 20]. The small peak at ~1384 cm1 can be assigned to bridged carbonates [17, 19], whereas the peaks located at 15501565 and 1632 1639cm1 which decrease during the reaction are usually attributed to biden tatecarbonates[1722].Thisassignment seems to be the most plausible one since the thermal stability of carbonates shows the following order: bridged car bonates < bidentate carbonates < monodentate carbonates [19], the latter being stable attheworking temperature used in these experiments [23, 24]. Ad

132 ditionally, the different gaseous compo sitions may also modify the stability of

ArtculoIV thesurfacecompound[25,26].

Fresh Pre-treated Peak area, relative units PROX-1h PROX-20h

1340-1363

1384

1418-1421

1484-1488

1550-1565

1632-1639

Main peaks position, cm -1

Fig.5.RelativeintensityevolutionofthedeconvolvedFTIRpeaks.

These assignments suggest that at the beginning of the reaction bidentate and bridge carbonates disappear (at least partially) while at the same time other species are formed as the reaction pro gresses (especially those located at ~1350 cm1) either as a result of the CO in the gaseous stream or as a result of the evolution of bridge or bidentate carbonates into new species. As can be

observed in Table 1 there is a high amount of carbonaceous species that are formed (CO2 to balance) on the catalyst surface during the PROX reac tion at 175C. This seems to indicate thatatleast afractionof thecarbonates formed are due to the effect of the gaseous stream on the surface and not to the conversion of the less stable car bonatespeciestomorestableones.

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

133

Table2.XPSsurfacecomposition(PROXexperimentsperformedunderHCconditions). Fresh Pretreated PROX1h PROX20h C1s %a %a %a %a PeakCO 21.7 22.6 12.8 13.1 (288.3288.9eV) O1s O(COb)/C(CO) 1.1 1.0 2.5 2.4 PeakMO/MOxc 1.1 1.1 1.0 1.0 (529.3529.6eV) Cu2p3/2 %d %d %d %d Peak1(Cu+Oh) 2.9 0.7 4.8 0.6 (929.5930.8eV) + Peak2(Cu Th) 23.7 72.3 (932.2eV) Peak3(Cu2+Oh) 97.1 99.3 71.4 27.1 (933.5934.4eV) Mn2p3/2 %d %d %d %d Peak1(Mn2+) 36.3 16.5 (640.3641eV) 3+ Peak2(Mn ) 82.8 82.8 51.1 69.1 (641.7642.3eV) Peak3(Mn4+) 17.2 17.2 12.6 14.5 (644.1644.5eV) Cu/(Cu+Mn) 0.28 0.25 0.21 0.22 a Atompercentageonsurface(adventitiouscarbonfree). b AfterthesubtractionofoxygenfromSiO2. c Ratio of oxygen evaluated in O1s (~529.5 eV) to stoichiometric oxygen in metals (Cu2p3/2andMn2p3/2).Usedtocertifyoxygenbalance. d Atompercentageonametalbasis(copperormanganese).

134
0E-01 0E-01 0E-01 0E-01 0E-01 0E-01

ArtculoIV
2.20E-01

Mn2p3/2
641.7

PROX 20h

2.00E-01 1.80E-01 1.60E-01 1.40E-01 1.20E-01 1.00E-01

932.2

Cu2p3/2

0E-02 0E-02 0E-02 0E-02

934.4

640.3 644.2

8.00E-02 6.00E-02 4.00E-02 2.00E-02

2.00E-01 0.00E+00 649 928 1.80E-01 1.60E-01 1.40E-01 1.20E-01 1.00E-01 8.00E-02 6.00E-02

929.5
931 934 937 940 943 946

0E-01 0E+00 637 0E-01 0E-01 0E-01 0E-01 0E-01 0E-02 0E-02 0E-02 0E-02
0E-01 0E+00

639

641

643

645

647

PROX 1h
642.3 641.0

933.9

932.2

644.5

4.00E-02 2.00E-02 0.00E+00 2.00E-01 649 928

1.80E-01 1.60E-01 1.40E-01 1.20E-01 1.00E-01 8.00E-02 6.00E-02

930.8
931 934 937 940 943 946

637
0E-01 0E-01 0E-01 0E-01 0E-01 0E-02 0E-02 0E-02 0E-02 0E+00 0E-01 637

639

641

643

645

Pretreated

647

933.5

641.8

644.2

4.00E-02 2.00E-02

930.4
931 934

639

641

643

645

647

0E-01 0E-01 0E-01 0E-01 0E-01 0E-02 0E-02 0E-02 0E-02 E+00

Fresh
641.8

2.00E-01 0.00E+00 649 928 1.80E-01 1.60E-01 1.40E-01 1.20E-01 1.00E-01 8.00E-02 6.00E-02

933.6

937

940

943

946

644.1

4.00E-02 2.00E-02

637

637

639

639

641

641

643 B.E. (eV)

643

645

645

647

647

649 928

0.00E+00 928 649

930.0

931

931

934

934

937

937

940

940

943

943

946

946

B.E. (eV)

B.E. (eV)

Fig. 6. Evolution of Mn2p 3/2 and Cu2p 3/2 XPS spectra of CuMn2O4 subjected to differ enttreatments(Fresh,pretreated,PROX1h,PROX20h).

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

135

The surface of the catalysts was also analysed by means of XPS in order to elucidate the surface composition of the metal oxides and to obtain a detailed picture of the chemical state of the cations and anions during the reaction. Fig. 6 shows the XPS spectra in the cop perandmanganese2p3/2regionsofthe catalysts after being subjected to differ ent treatments. From the areas of the peaks it was possible to evaluate the surfacecompositionsatdifferentstages. The results of this evaluation are dis playedinTable2. The formation of monodentate carbon atesonthesampleduringthereactionis corroborated by an increase in the O/C ratio, [O1s ~531 eV, peak CO after sub tracting the oxygen of the residual SiO2; C1s ~288 eV, peak CO] from values of around 1 (a value that might be ex pected for bidentate carbonates given the fraction of the oxygen in the car bonate associated to the metal) to val ues of about 2.5 which is similar to the values that might be expected for monodentatecarbonates. ThedeconvolutionoftheCu2p3/2signal reveals the states of the copper on the surface of the catalyst. The peak located at around 934 eV is attributed to Cu(II) [18, 27], which is generally assumed to be octahedrally coordinated. The peak locatedat932eVisusuallyassociatedto the presence of Cu(I) or Cu(0), species

that can be distinguished from each other by the modified Auger parameter [28, 29]. However, other authors [9, 11, 30, 31] consider that there is a negative shift in the binding energy and attribute the peaklocatedat ~930eVtoCu(I)in a tetrahedral position [9, 31] and the peaks at ~932 and ~934 eV to Cu(II) octahedrally and tetrahedrally coordi nated respectively [9]. This negative chemical shift has been attributed by some authors to differences in the Madelung potential [7, 8]. Additionally, other authors have associated the peaks at ~930 to the presence of Cu(I) in an octahedral environment [8, 32]. From the results of our stepresponse experi mentswededucethatitistheCOthatis reducing the catalyst. This assumption, together with the assignments provided by the majority of the works reported above, incline us to assign the peak lo cated at ~932 eV to Cu(I). As no Cu(0) wasdetectedintheAuger spectrum, we believe that the shoulder at ~930 eV is probably due to the presence of Cu(I) in an octahedral position [8, 32], whereas the peak at around 932 eV is due to Cu(I) in tetrahedral position. This would imply a negative shift of around 1 eV [27] which is much lower than the pre vious shift (~2 eV [9, 30]). Finally the peak at 934 eV is assigned to Cu(II) [18, 27]. No peaks related to copper carbon ates(~935eV[33])weredetected.

136 The deconvolution of the Mn2p3/2 sig nal reveals the manganese valences. According to the literature, the Mn2p3/2 signal is useful for distinguish ing between the states of Mn(II), Mn(III) and Mn(IV). Studies performed on pure manganese oxides provide binding en ergyvaluesofaround640.7640.9eVfor MnO [3436], 641.7641.9 eV for Mn2O3 [34, 35] and 642.4643.4 eV for MnO2 [34, 35]. Several authors however, have found a shift towards higher values for certain manganites, including nickel manganites [37], coppernickel manganites [30] and copperzinc manganites [38] for which three peaks located at 641.3641.6 eV; 642.4 642.7eV and 643.7644.9 eV can be found. Assignments of these peaks dif fer. In some cases, the first peak is at tributed to Mn(III) and the others to Mn(IV) depending on the coordination environments [38]. Other authors, how ever, consider that the first peak should be assigned to Mn(II) [30]. For CuMn2O4 synthesised in air most of the copper can be expected to be Cu(II) whereas mostof the manganesecanbeexpected to be Mn(III) [6]. In our fresh samples, we detected a preponderance of Cu(II) so we assume that the peak located at around 642 eV is associated to Mn(III) and the peak located at ~644eV to Mn(IV). During the first few moments of the reaction, a new peak appears at ~641 eV which we attribute to Mn(II). In

ArtculoIV the fresh and pretreated samples the amount of Mn(IV) detected could be ascribed to a deficiency of Cu on the surface of the catalysts (Cu/(Cu+Mn)< 0.33, Table 2). At the beginning of the PROX reaction an intense reduction of the catalysts seems to be occurring. During the reaction Mn(II) appears, which is probably tetrahedrally coordi nated since the most likely cation distri bution for CuMn2O4, according to the literature, is Cu+ and Mn2+ on tetrahe dral sites and Cu2+, Mn3+ and Mn4+ on octahedral sites, giving the following ionicconfiguration[18]: (Cu+Mn2+1)Th(Cu2+1Mn3+Mn4+)OhO24 (R6)

As can be seen from Table 2, during the reaction the peak located at ~933 eV decreases, while a peak located at ~932eV appears and progressively in creases. This indicates the progressive reduction of Cu2+ to Cu+. At the begin ning of the PROX reaction Mn3+ is par tiallyreducedtoMn2+,whileatthesame time Mn4+ is partially reduced to Mn3+. However, the amount of Mn2+ is slightly higher than that of Cu+ after 1h PROX, indicating an excessive reduction of manganese. For longer reaction times these nonsteady valences of manga nese evolve into a more oxidised state, and the reduction of Cu2+ to Cu+ be comesmoresignificant.

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

137

To the best of our knowledge, there is no mechanistic study on CO oxidation in thepresenceofH2incoppermanganese systems. However, in the absence of H2 it is generally assumed that the reaction proceedsasfollows[8]: CO+Mn4+CO+ads+Mn3+ O2+Cu+Oads+Cu2+ CO+ads+OadsCO2 Cu2++Mn3+ Cu++Mn4+ (R7) (R8) (R9) (R10)

As stated before, the presence of carbo naceous species does not seem to lead to deactivation. However, such species mayaffecttherelativeconcentrationsof the different oxidation states of manga nese. At the beginning of the reaction manganeseisreducedtoahigherextent
O C O 3+ Mn O O O + CO/H2 O O 2+ Mn O

than copper (Table 2), whereas during the reaction manganese is partially re oxidised. The initial reduction cannot have been produced by the gaseous stream since the copper, which is more prone to reduction [5], is less reduced than the manganese at this stage. In our search for an alternative explanation we suspect that the initial reduction of manganese might be related to the presence of bidentate/bridge carbon ates in the fresh sample. These have almostdisappearedafter1hunderPROX conditions,aprocessthatoccursparallel to the reduction of manganese. A possi ble mechanism for the reductive de composition of the bidentate/bridge carbonates is suggested by reactions R11andR12,whichmayalsoexplainthe concomitantreductionofmanganese:

2+ Mn O

+ CO2/H2O + CO2

3+ Mn

(R11)
2+ Mn O O 2+ Mn O

O C O O 3+ Mn O O 3+ Mn O + CO/H2

+ CO2/H2O + CO2

(R12) sponse experiments, the FTIR analyses and the evolution of the O/C ratio (Ta

During the reaction new carbonates are formed, as implied from the step re

138 ble2). As already indicated, the forma tion of these carbonates is enhanced by the presence of oxygen. Simultaneously the oxidation of Mn2+ to Mn3+ takes place (Table 2). A possible explanation

ArtculoIV for this could be that of reactions R13 (oxidation of Mn2+) and R14 (formation of new carbonates) in which CO+ads comesfromreactionR7[8]:
O22+ Mn 2+ Mn O O + O2 O 3+ Mn O 2+ Mn O

(R13)

O C O23+ Mn O O 2+ Mn O + CO+ads O O 3+ Mn O 2+ Mn O O

(R14)

Thus, considering the relative concen trations of Mn2+, Mn3+ and Mn4+ and Cu+/Cu2+ after 20 h under PROX condi tions (CO+H2+O2), the loss of activity should be attributed to the progressive reduction of copper on the surface of the catalyst through the action of CO. Thisiscorroboratedbytheresultsofthe stepresponse experiments (see above). It is not yet clear how the reduction of CuO can have a negative effect on the reaction mechanism (R7R10 for non preferential CO oxidation). In fact, the presence of Cu+ seems to be necessary for adsorbinggaseous oxygen, according to the standard CO oxidation mecha nism (R8). Veprek et al. [8] found a high

proportionofCu+onthecatalystsurface of deactivated hopcalite catalysts (catalysts subjected to heattreatment at 550C), although these authors did not offer any mechanistic explanation. Dollimor and Tonge [39] suggested that Cu+ forms stable CO omplexes that may block the surface, causing lower activity. However, we detected no such com plexeswithFTIR.TheexactroleofCu+in the deactivation process of these PROX catalysts must remain a subject for fu tureresearch.

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

139

4. Conclusions

We have shown that CuMn2O4 synthe sised by SACOP is an active catalyst for the preferential oxidation of CO at mod erate temperature, even at high spatial velocities. However, this catalyst under goes slow deactivation in longterm experiments. Although carbonates are References [1] I.Lpez,T.ValdsSols,G.Marbn,Int.J.HydrogenEnergy33(2008)197.

formed under reaction conditions in the manganese phase present on the cata lyst surface, they are not the cause of deactivation. In the analysis conditions employed in this work deactivation ap pears to be produced by the progressive reduction of Cu(II) to Cu(I) through the actionofCO.

[2] A. Gurbani, J. L. Ayastuy, M. P. GonzlezMarcos, J. E. Herrero, J. M. Guil, M. A. GutirrezOrtiz,Int.J.HydrogenEnergy34(2009)547. [3] G.Marbn,A.B.Fuertes,Appl.Catal.B:Environ.57(2005)43. [4] G.Marbn,I.Lpez,T.ValdsSols,Appl.Catal.A:Gen.361(2009)160. [5] F.C.Buciuman,F.Patcas,T.Hahn,Chem.Eng.Proc.38(1999)563. [6] P.Porta,G.Moretti,M.Musicanti,A.Nardella,Catal.Today9(1991)211. [7] C.Yoon,D.L.Cocke,J.Catal.113(1988)267. [8] V.Veprek,D.L.Cocke,S.Kehl,H.R.Oswald,J.Catal.100(1986)250. [9] G.Fortunato,H.R.Oswald,A.Reller,J.Mater.Chem.11(2001)905.

[10] T.ValdsSols,G.Marbn,A.B.Fuertes,Catal.Today116(2006)354. [11] J.Papavasiliou,G.Avgouropoulos,T.Ioannides,J.Catal.251(2007)7. [12] M. Kramer, T. Schmidt, K. Stowe, W. F. Maier, Appl. Catal. A: Gen. 302 (2006) 257. [13] G.Marbn,T.ValdsSols,A.B.Fuertes,Cat.Lett.118(2007)8. [14] G.Marbn,A.B.Fuertes,T.ValdsSols,Microp.Mesop.Mat.112(2008)291. [15] W.Liu,M.FlytzaniStephanopoulos,J.Catal.153(1995)317. [16] D.Gamarra, A. Horns, Zs. Koppany,Z. Schay, G. Munuera, J. Soria, A.Martnez Arias,J.PowerSources169(2007)110. [17] A. MartinezArias, A. B. Hungria, G. Munuera, D. Gamarra, Appl. Catal. B: Envi ron.65(2006)207. [18] B.Gillot,S.Buguet,E.Kester,J.Mater.Chem.7(1997)2513.

140

ArtculoIV

[19] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc. , Faraday Trans.1Phys.Chem.CondensedPhases85(1989)929. [20] M.Manzoli,A.Chiorino,F.Boccuzzi,Appl.Catal.B:Environ.52(2004)259. [21] O. Seiferth, K. Wolter, B. Dillmann, G. Klivenyi, H.J. Freund, D. Scarano, A. Zec china,SurfaceScience421(1999)176. [22] R. Xu, Z. y. Ma, C. Yang, W. Wei, W. h. Li, Y. H. Sun, J. Mol. Catal. AChem. 218 (2004)133. [23] W.M.Shaheen,M.M.Selim,Thermochim.Acta322(1998)117. [24] R.L.Frost,Z.Ding,J.T.Kloprogge,W.N.Martens,Thermochim.Acta390(2002) 133. [25] O.Pozdnyakova,D.Teschner,A.Wootsch,J.Krohnert,B.Steinhauer,H.Sauer,L. Toth,F.C.Jentoft,A.KnopGericke,Z.Paal,R.Schlogl,J.Catal.237(2006)17. [26] O.Pozdnyakova,D.Teschner,A.Wootsch,J.Krohnert,B.Steinhauer,H.Sauer,L. Toth,F.C.Jentoft,A.KnopGericke,Z.Paal,R.Schlogl,J.Catal.237(2006)1. [27] M. Lenglet, A. D'Huysser, J. Kasperek, J. P. Bonnelle, J. Durr, Mater. Res. Bull. 20 (1985)745. [28] X.M.Liu,G.Q.Lu,Z.F.Yan,Appl.Catal.A:Gen.279(2005)241. [29] W.L.Dai,Q.Sun,J.F.Deng,D.Wu,Y.H.Sun,Appl.Surf.Sci.177(2001)172. [30] J.Tpfer,A.Feltz,D.Grf,B.Hackl,L.Raupach,P.Weissbrodt,Phys.StatusSolidi A134(1992)405. [31] V.A.M.Brabers,F.VanSetten,J.Phys.DAppl.Phys.16(1983)L169L172. [32] A. D'Huysser, B. LereboursHannoyer, M. Lenglet, J. P. Bonnelle, J. Solid State Chem.39(1981)246. [33] C.D.Wagner,W.M.Riggs,L.E.Davis,J.F.Moulder,G.E.Muilenberg,HandbookofX Ray Photoelectron Spectroscopy. Perkin Elmer Coorporation, Eden Prairie, Minn.,1979. [34] G.C.Allen,S.J.Harris,J.A.Jutson,J.M.Dyke,Appl.Surf.Sci.37(1989)111. [35] L. M. Ioffe, P. Bosch, T. Viveros, H. Sanchez, Y. G. Borodko, Mater. Chem. Phys. 51(1997)269. [36] P. Z. Si, E. Bruck, Z. D. Zhang, O. Tegus, W. S. Zhang, K. H. J. Buschow, J. C. P. Klaasse,Mater.Res.Bull.40(2005)29. [37] J.Tpfer,A.Feltz,SolidStateIonics59(1993)249. [38] F. Li, L. Zhang, D. G. Evans, X. Duan, Colloid Surf. APhysicochem. Eng. Asp. 244 (2004)169.

ValdsSolsetal.Int.J.Hydrog.Energy35(2010)18791887

141

[39] D.Dollimor,K.H.Tonge,J.Chem.Soc.A(1970)17281731.

143

ResumenartculoV. HIGHLYACTIVECOBALTOXIDECATALYSTSPREPAREDBYSACOPFOR THEPREFERENTIALOXIDATIONOFCOINEXCESSHYDROGEN

CatalizadoresdexidodecobaltopreparadosporSACOPconelevada actividadenlaoxidacinpreferencialdeCOenexcesodehidrgeno
Losxidosdecobaltosoncompuestosdeelevadointersindustrialyrecientementehan sido empleados en la reaccin de oxidacin preferencial de CO en exceso de hidrgeno. El presente trabajo tiene como objeto la sntesis de xidos de cobalto mediante el mtodoSACOPysuposterioranlisisPROXenunsistemadereaccinmulticanal. Los xidos de cobalto fueron obtenidos utilizando dos mtodos de precipitacin, bien vertiendoelgelsobreunadisolucindeNaOH(mtodoA)obienechandogotaagotala disolucindeNaOHsobreelgel(mtodoB).Pormotivosdecomparacinademsdelos nanocatalizadores preparados por SACOP, se sintetizaron otros dos catalizadores de Co3O4, uno obtenida mediante impregnacin de slice comercial y otro por precipitacin conNaOH. El mtodo SACOP permiti la obtencin de nanocatalizadores de Co3O4/CoOOH con reas superficiales de hasta 120 m2g1. La proporcin de ambos xidos vari en funcin del mtodo de precipitacin, siendo el mtodo B el que produjo una mayor proporcin de CoOOH (80%). Mediante la precipitacin A se obtuvieron catalizadores con un contenido de CoOOH de aproximadamente el 50%. El anlisis textural de las muestras precipitadas por el mtodo A indujo a pensar que una parte de los iones de cobalto haban migrado fuera de las microestructuras de la slice durante la etapa de precipitacin,produciendoCo3O4debajasuperficieenlaetapadecalcinacinposterior. Esta migracin indeseable se minimiz mediante el empleo del mtodo B. El contacto entrelasespeciesdecobaltoylosgrupossilanolesdelasparedesdelasliceprodujoun silicato bsico de cobalto que en la etapa de eliminacin de la slice se transform en CoOOH, siendo ste el compuesto responsable de la elevada rea superficial de los catalizadoressintetizadosporSACOP. En los catalizadores sintetizados utilizando el mtodo de precipitacin A se estudi la influencia de la temperatura de calcinacin y de la concentracin de cobalto en la disolucin precursora. Las muestras calcinadas a 200C presentaron valores del rea superficial ms pequeos que los de los catalizadores calcinados a temperaturas ms

144

ArtculoV

altas. Este comportamiento es debido a la calcinacin incompleta de Co(OH)2 hacia CoOOH, ya que para su transformacin posterior a Co3O4 se necesitan temperaturas superiores a 200C. Por ello se puede asumir que el CoOOH proveniente de la calcinacin incompleta del Co(OH)2 presenta reas superficiales menores que el Co3O4 obtenido tras la calcinacin total. Por consiguiente, las muestras calcinadas a 200C estn compuestas por CoOOH de elevada rea superficial proveniente de las interacciones cobaltoslice, CoOOH de menor rea superficial producido a partir de la calcinacin incompleta de Co(OH)2 y diferentes fracciones de Co3O4. En las muestras calcinadas a temperaturas por encima de 200C se observ una ligera tendencia de disminucin delreasuperficialconelincremento delatemperatura.Estefenmenose puede explicar considerando la sinterizacin de las partculas de Co3O4 formadas fuera delamatrizdeslice. Por otro lado, bajas concentraciones de cobalto en la disolucin precursora produjeron nanocatalizadores con reas superficiales ligeramente ms altas, aunque las diferencias no fueron significativas, lo que puede ser debido a la presencia de Co3O4 formado fuera de la matriz de slice. Se estim que la cantidad de Co3O4 que se form dentro de la matrizdeslicerepresentabaentreel30yel45%deltotaldeCo3O4producido. Los espectros de difraccin indujeron a pensar que una gran parte de las muestras estaban formadas por material amorfo. Cuanto mayor era la fraccin amorfa, ms contenido en CoOOH tena la muestra y ms rea superficial presentaba, de lo que se dedujo que la especie CoOOH estaba en estado amorfo y era la responsable de la elevadasuperficieespecficadeestoscatalizadores. El sistema de reaccin multicanal se puso a punto para el anlisis de reactividad de los xidos de cobalto preparados en este trabajo. Este sistema permite analizar simultneamente la actividad cataltica de hasta seis catalizadores, utilizando aproximadamente 5.5 mg de catalizador en cada experimento y velocidades espaciales muyelevadas(GHSV39000h1).La mayoradeloscatalizadoresmostraronuna elevada estabilidad,conprdidasdeactividaden20hinferioresal15%.Todosellospresentaron selectividades por encima del 0.7. Las pequeas diferencias de actividad entre las muestrassepuedenatribuirapequeasdiferenciasensureducibilidad. El catalizador ms activo result ser el preparado mediante la precipitacin tipo B y calcinado a 400C (121 m2g1), con una actividad 1.5 veces superior al segundo (434.7107 y 281.4107 molCOgcat1s1, respectivamente). Adems, present una selectividad de 0.81 y una estabilidad muy elevada, perdiendo solo un 4.5% de su actividad desde las 5 h hasta las 20 h de reaccin. Su gran actividad es debida a la

145 presencia de CoOOH de elevada rea superficial que en condiciones PROX se reduce a Co3O4. Los nanocatalizadores Co3O4/CoOOH preparados por SACOP presentaron actividades catalticas ms elevadas que los catalizadores de Co3O4 ms activos descritos en la literatura. Este hecho demuestra que la tcnica SACOP es un mtodo vlido para la obtencin de catalizadores de cobalto con elevada actividad cataltica, selectividad y estabilidadparalareaccindeoxidacinpreferencialdeCO.

147

ArtculoV Highly active cobalt oxide catalysts prepared by SACOP for thepreferentialoxidationofCOinexcesshydrogen
IreneLpez,TeresaValdsSols,GregorioMarbn
Abstract Cobalt based catalysts synthesized by the silica aquagel confined coprecipitation method (SACOP)andanalysedathighspatialvelocities(39,000h1)forthe preferential oxidation of CO in the presence of hydrogen showed the highest values of catalytic activity of all the cobaltbased catalysts described in the literature together with good valuesofselectivityandstability.Co3O4/CoOOHnanocatalystshadspecificsurfaceareas of up to 120 m2g1 depending on the type of precipitation used. SACOP samples precipitatedbypouringaquagelintotheNaOHsolution(methodA)gaverisetosamples with less CoOOH than when the NaOH solution was poured over the aquagel (method B). This difference is ascribed to the migration of cobalt out of the aquagel microstructure during precipitation by method A. Precipitation type B produced the most active catalyst due the larger amount of amorphous CoOOH in the fresh sample, which was formed via the removal of silica from the cobalt silicate hydroxide. The CoOOH is gradually reduced to high surface area Co3O4 in the presence of H2 during the PROX reaction. The mechanisms of formation of CoOOH and Co3O4 via SACOP are also discussed. Keywords: PROX, hydrogen, CO, cobalt oxide, cobalt spinel, cobalt oxyhydroxide, catalyticactivity,stability

1.Introduction With the increasing social demand for a cleaner energy system there is mounting pressure for a gradual transition to hydrogen powered vehicles in order to mitigate CO2 emissions to the atmosphere. To achieve this goal a reliable hydrogen fuelling system needs to be developed. Taking into account the current problems associated with on board hydrogen storage (high cost, low mass density, the need for a generation/ distribution network, etc.), onboard hydrogen production via the steam reforming of bioalcohols seems to be a most reasonable and best developed option. The steam reformingofbioalcohols(e.g.methanol)producesahydrogenstreamwithavariableCO

148

ArtculoV

content which must be almost completely removed to prevent the platinum anode in the PEMFC from being poisoned. In order to purify the hydrogen stream, a water gas shift reaction (for high CO concentrations) followed by preferential oxidation of any residualCO(PROX)iscommonlyapplied. The PROX reaction consists of the catalytic oxidation of residual CO to CO2 in the 100250C temperature range, while at the same time avoiding the oxidation of hydrogen to water. From the economic point of view there is increasing interest in the development of nonnoble metal catalysts for this application. Several transition metal oxides are potential candidates for use in this process. CuOx/CeO2 is one of the most investigated catalysts for the PROX reaction [1]. However, this catalyst is prone to gradual deactivation [2]. In the search for more stable catalytic formulations, the cobalt spinelCo3O4 hasrecently beenproved tobea highlyactive,selectiveandstablecatalyst [35], although in some cases the methanation of CO does occur, even at temperatures below200C[4]. Theuseofnanoscaledcatalystsimprovescatalyticactivitybecauseoftheirveryhigh surface to volume ratio. In this work we employed the silica aquagel confined co precipitation (SACOP) method [6, 7] to synthesize cobalt oxide based nanostructures. The SACOP technique is based on the coagulation precipitation processes that occur when the metallic cations of salts dissolved in a medium of silica aquagel are made to precipitate by introducing basic reagents. This synthesis strategy gives rise to high surface area nanosized metallic oxides, and has been successfully applied in the preparation of CuOx/CeO2, CoFe2O4, CuMn2O4[6], and CeO2 [7]. In the present work the activity and stability of cobalt oxide catalysts prepared by the SACOP technique have been tested for the preferential oxidation of CO. Our findings demonstrate that the SACOP method endows the resulting catalysts with properties that make them more activethanthoseproducedbyapplyingothernanotechnologicalsyntheticprocedures. 2.Experimental Catalystsynthesis The cobalt based catalysts were prepared by the SACOP procedure [6]. First Co(NO3)26H2OwasdissolvedinHCl.Thenthissolutionwasaddeddropwisetoamixture of sodium silicate and water using molar compositions of SiO2/H+/H2O = 1/6.54/193.9 and Co/Si = 0.25 or 0.67. After hydrolysis (24h at RT) and condensation (24h at 100C) a silica aquagel containing widely dispersed cobalt cations was formed. Precipitation

Lpezetal.EnviadoaChemCatChem

149

insidethegelwasperformedviatwodifferentapproaches.InmethodAtheaquagelwas slowlypouredintoasolutionofNaOH(1.25M);inmethodBasolutionofNaOH(3.5M) was poured into the aquagel in order to cause the cobalt ions to precipitate. In both cases the nonpoured component was subjected to vigorous magnetic stirring. The slurry formed after the precipitation stage was repeatedly washed and dispersed in acetone, then vacuum filtered and left to dry at RT overnight. The resulting silicametal composites were calcined in air at different calcination temperatures (200500C) for 4h (1Cmin1). As calcination takes place inside the pores of the silica matrix (confined process) growth of the nanomaterials is restricted and consequently high values of surface area can be achieved (template technique). Finally, the silica matrix was removed by dissolution in NaOH (3.5 M) in an isolated stirred beaker at ~85C for 6hours.Takingintoaccountthenegativeeffectofresidualsilicaandsodiumoncatalytic activity (as reported in a previous paper [2] and confirmed experimentally in this work), all the samples were prepared with low amounts of silica and sodium (below 2.0and 0.3wt.%,respectively).Atthesevaluesnoeffectoncatalyticactivitywasevidenced. The samples synthesised by SACOP were labelled STCoP, where S indicates SACOP method, T the calcination temperature (200, 300, 400, 500C), Co the amount of cobalt in the synthesis mixture (L for low concentration, Co/Si=0.25; H for high concentration Co/Si=0.67) and P the precipitation method (A for the samples prepared by pouring the gel into the NaOH solution and B for the samples obtained by adding the NaOH solution to the gel). Thus, S300HB refers to the sample synthesised by SACOP at 300C and Co/Si=0.67,whereprecipitationwastriggeredbyaddingNaOHtothegel. For comparison purposes, Co3O4 nanoparticles were prepared by a hard silica template method, as reported elsewhere [8] and calcined at 400C (T400). Co3O4 particles were also obtained via the precipitation of Co(OH)2 with NaOH powders from a solution with the same composition as the aquagel precursor solution (except for the absence of sodium silicate) followed by calcination at 200 and 400C. These samples were named P200andP400,respectively. Catalystcharacterization Nitrogen adsorption isotherms were performed at 196C using a Micromeritics ASAP 2020 volumetric adsorption system. The BET surface area was deduced from an analysis oftheisothermsintherelativepressurerangeof0.04to0.20.ThePoreSizeDistribution (PSD) was calculated by the KrukJaroniecSayari method [9]. Transmission electron

150

ArtculoV

micrographs (TEM) were taken on a JEOL (JEM2000 EX II) microscope operating at 16kV.XRDspectrawereobtainedinaSiemensD5000diffractometeroperatingat40kV and 20 mA with CuK radiation. The degree of crystallinity (DOC) was calculated as the ratio between the height of the main Co3O4 peak (2 =36.7) (subtracting the baseline) and the total height including the background of the XRD diffractogram. TPR analyses were performedinachemisorptionanalyzer(Autochem II)equippedwitha TCDreactor and a Mass Spectrometer (OmniStar 300O). For each analysis approximately 25 mg of sample was treated with a 50 mLmin1 stream of 10% H2 in Argon from 100C to 850C at5Cmin1.TGexperimentswereperformedinaCIElectronicsdevicefromRTto700C ataheatingrateof5Cmin1underacontinuousflowofN2(100mLmin1).TGreduction experimentswerecarriedoutat175Cunderacontinuousflowof15%H2inN2. Catalyticactivitytests Catalytic activity tests on the cobaltbased nanoparticles were performed using a sixreactor system. A computercontrolled multiposition valve fitted to the mass spectrometer permitted simultaneous analysis of the outlet streams. In the standard PROXreaction5.5mgofnanocatalystwasdilutedin75mgofSiCparticlesandplacedin the stainlesssteel reactor between two layers of glass wool. This ensured plugflow and isothermal conditions within the bed. The samples were heated in a flow of He at 195C for 60 min. The reactant stream was then passed through the catalyst and its catalytic activity and selectivity were sequentially evaluated at decreasing temperatures (175, 150 and 125C) for long periods of time. The reactant gas feed contained 15 vol.% H2, 0.4vol.% CO and 0.4 vol.% O2 in He. The outlet streams were analyzed online by mass spectrometry (Omnistar 300O). O2, H2, CO, CO2 and CH4 were quantitatively analyzed with the help of prereaction and postreaction calibration steps. In all the PROX experiments only traces of methane (below 0.004%) were detected at the reaction conditionsusedinthiswork Theconversionwasevaluatedasfollows:

X CO = 1

[CO]out [CO]in

[E1]

where the in and out subscripts refer to the inlet and outlet gaseous streams, respectively.SelectivitytowardsCOoxidationwascalculatedasfollows:

Lpezetal.EnviadoaChemCatChem

151

S=

0.5 ( [CO]in [CO]out ) [O 2 ]in [O 2 ]out

[E2]

The reaction rate (molCOgcats1) was determined by means of the LangmuirHinselwood mechanism proposed by Liu and FlytzaniStepahopoulus [10] for CO oxidation over a copperceriacatalyst,asdescribedelsewhere[1]. Tocharacterizethestabilityofthecatalysts,thefollowingparameterwasemployed:

% r = 100

20h 5h r175 C r175C 5h r175 C

[E3]

wherertTisthe reactionrate at time t and temperatureT. Thisparametermeasuresthe variationofthePROXreactionratebetween5and20hoursofreactionat175C.Slightly negative, zero or positive values of %r indicate a high degree of stability in the PROX reactionrate.
3.Resultsanddiscussion 3.1.Structuralfeatures

By means of the SACOP procedure it is possible to obtain cobalt oxidebased nanostructures with high surface area values. The presence of cobalt spinel (Co3O4) and cobalt oxyhydroxide (CoOOH) in the catalysts was confirmed by XRD, as can be seen in Figure 1. The amount of CoOOH in selected samples was evaluated by means of thermogravimetric analysis in a nitrogen atmosphere [11]. Table 1 summarizes the structuralcharacteristicsandthecatalyticactivityvaluesofthepreparedsamples.


Typeofprecipitation Co/Si (molar) 0.3 200 300 400 500 200 300 400 500 400 400 400

152

Table1.Structuralcharacteristicsandcatalyticactivityofthecobaltnanocatalystspreparedinthiswork. Tcalc(C) CoOOH(a) SBET DOC r20h175C107(d) %r 20h 20h (f) S 175 C (wt.%) (m2g1) (c) (molCOgcats1) (e) n.e.(b) 77.0 27 25 37 37 34 40 33 39 14 27.5 0 24 97.0 86 48 195 208 245 202 195 206 435 8.3 268 263 92.8 88.8 72.6 67.1 84.3 82.2 83.2 121.3 281 n.e. n.e. n.e. 55.2 n.e. n.e. 43 80.3 14 12 16 12 14 19 7.5 3.8 4.5 6.3 7.4
1

Ref.

S200LA 0.3 0.3 AGeloverNaOH 0.7 0.7 0.7 0.7 BNaOHovergel NaOHoverdissolution 0.3 0.3 0.3

0.7 0.7 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.4 0.8

S300LA

S400LA

S500LA

S200HA

S300HA

S400HA

S500HA

S400LB

P400

T400

(a)TGAanalyses

(d)PROXreactionrateat175Cafter20h(molCOgcats )

(e)VariationofthePROXreactionrateat175Cbetween5 and20h(equationE3) (c)Degreeofcrystallinitycalculatedat2=36.7(peakcorrespondingtothe (f)Selectivityat175Cafter20hreaction(equationE2) Co3O4phase)asindicatedintheexperimentalsection

(b)Notevaluated

ArtculoV

Lpezetal.EnviadoaChemCatChem

153

a b c d
15 20 25 30

* Co3O4 CoOOH

Intensity (a.u.)

35

40 2

45

50

55

60

65

70

Figure1.XRDprofilesof(a)S400LB,(b)S200LA,(c)T400and(d)P400samples.

The isotherms of the SACOP samples, the main examples of which are displayed in Figure2,indicatethepresenceofatinyfractionofsmallmesopores(23nm)associated to a high fraction of structural porosity (~0.30.4 cm3g1). The average pore size of this porosity ranges from 55 to 80 nm for the samples prepared via method A and has a valueof17nmforthesamplepreparedviamethodB.Thesevaluesenableustoforman initial impression of the mode of formation of CoOOH and Co3O4 when applying the SACOP technique. The sample obtained by method B (S400LB), which has the highest specific surface area (Table1; 121 m2g1), displays a small average pore size for structural porosity (17nm), which is of a similar order to that of the silicaCeO2 nanocomposites obtained by SACOP in a previous work [7]. This suggests that the structural porosity in this case is made up of intraparticle porosity and the space occupied by the silica walls of the nanocomposite formed during the precipitation procedure. S400LB, therefore, must have been formed as a result of very close contact between the cobalt species and the silanol groups on the silica walls. As S400LB is mainlyformedbyCoOOH(Table1;80.3wt.%)itcanbeconcludedthatthisspeciesisthe productofinteractionbetweenthecobaltionsandthesilicawalls,andisresponsiblefor thehighsurfaceareaofthissample(hereafterthisCoOOHfractionwillbereferredtoas hCoOOH, meaning high surface area CoOOH). In contrast, the samples prepared by method A present higher average pore sizes in their structural porosity (5580 nm),

154

ArtculoV

lowervaluesofspecificsurfacearea,andarecomposedofsmalleramountsofhCoOOH (Table 1). These data are consistent with a picture in which part of the cobalt ions migrate out of the silica microstructure during the precipitation process and produce bulk Co3O4 in the subsequent calcination step. The structural porosity of these samples therefore receives an additional contribution from the interparticle voids (of high average size) in the Co3O4 nanoparticles resulting from the migration of cobalt ions (hereafter bulk Co3O4). This composite porosity can be appreciated in the pore size distribution curve of S400LA (inset of Figure 2), in the shoulder in the 1020 nm range (contributed by hCoOOH) and the maximum at 55 nm (contributed by the interparticle voids of bulk Co3O4). If this interpretation is correct, then method B appears to be the most suitable precipitation procedure, since it allows the migration of cobalt from the aquagelmediatobeminimized.However,thishighsurfaceareahCoOOHfractionisnot able to survive the calcination step in most of the samples (the TGA experiments showed that it was transformed into cobalt oxide during the 150500C temperature interval, the maximum rate of decomposition occurring in the 250300C range). This productmust,therefore,havebeenformedinthestepfollowingcalcination,namelythe silica removal step. Wide peaks corresponding to cobalt silicate hydroxide, Co3(OH)4Si2O5, were appreciated in the XRD spectrum performed on a CoSiO2 nanocomposite obtained by SACOP, where the wavy baseline of amorphous silica was also evident. This kind of silicate is known to be produced by the deposition precipitationofcobaltontoasilicamatrix[12],andisabletosurviveevenatthehighest calcination temperature (500C) [13]. It seems clear that the silica removal step transforms this silicate into high surface area CoOOH. A longrun TG experiment in a continuous flow of H2 (15%) demonstrated that the hCoOOH present in the S400LB nanocatalyst was gradually reduced to a Co3O4 at 175C, making this the predominant speciesunderPROXreactionconditionsforalltheanalysedcatalysts. There are no great differences in the BET surface areas of the SACOP catalysts precipitated by method A (6793 m2g1), which indicates that the calcination temperature has little influence on the properties of the samples. Surprisingly the samples calcined at 200C present lower values of specific surface area than those calcined at higher temperatures (Table 1). This is ascribed to the partial calcination of Co(OH)2 at 200C. It is known that bulk Co(OH)2 is converted to Co3O4 in two steps [14]: first it is transformed to CoOOH at temperatures below 200C; then the CoOOH is converted toCo3O4inthe temperaturerangeof200300C.Itcan thereforebe assumed that the CoOOH formed by the partial calcination of bulk Co(OH)2 has a lower surface area than the Co3O4 phase formed after complete calcination. This CoOOH phase is

Lpezetal.EnviadoaChemCatChem

155

therefore different to the hCoOOH phase and hereafter will be termed lCoOOH, signifyinglowsurfaceareaCoOOH.InconclusionlCoOOHisonlypresentinthesamples calcined at 200C,and accompanies the fractionof hCoOOHformedby thesilicacobalt interaction,aswellasthedifferentfractionsofCo3O4(seebelow).
. 250
3 -1 dV/dlog(D), cm g

Adsorbed volume, (cm 3STPg-1) 200

0.4 0.3 0.2 0.1 0.0 1

S400L-B S400L-A

150

100

10 100 1000 Pore size (D), nm

50

S400L-B S400L-A

0 0.0 0.2 0.4 0.6 0.8 1.0 Relative pressure (p/po)

Figure 2. N2 physisorption isotherms and pore size distribution curves for S400LA and S400LB.

After calcination at temperatures equal to or above 300C, when bulk Co(OH)2 is completelytransformedintobulkCo3O4,thespecificsurfaceareavaluesfollowaslightly decreasingtrendwith calcinationtemperature.Thismightbeduetothesinteringofthe bulk Co3O4 particles. In general, the samples prepared with a lower Co/Si ratio present slightlyhighervaluesofspecificsurfacearea,ascanbeseeninTable1.Thisbehaviouris consistent with the SACOP mechanisms reported in a previous work [7], although it is not as marked as might be expected due again to the formation of some bulk Co3O4 outsidethesilicamatrix. Certain specificfeatures of the XRDspectra (i.e. highbaselines, wide peaks, etc) suggest that a high fraction of the SACOP samples is formed by amorphous material. Therefore, for comparative purposes, we have related the amorphous material fraction in the SACOP samples to the degree of crystallinity (DOC) evaluated (vide supra Experimental

156

ArtculoV

section). The resultingvaluesaresummarizedinTable1.Figure3 showsthevariationof the CoOOH concentration (wt.%) and the BET surface area with DOC for different samples. As can be seen, there is a conspicuous inverse relation between the concentration of CoOOH and DOC, which means that this species occurs in an amorphousstateinthesesamples.Furthermore,CoOOH(whichisonlyhCoOOHforthe samples calcined at T>200C) is the main contributor to the specific surface area of the samples, as is deduced from the variation of BET surface area with DOC (Figure3). However,twosamplesdonotfullyconformtothistrend(S200HAandT400).
90 75 CoOOH (wt.%) 60 45 30 15 0 0 20 40 DOC 60 80
S200H-A P400 T400 S500H-A

S400L-B

140 CoOOH BET surface area 120 100 80 60 40 20 0 100 SBET (m g )

2 -1

Figure 3. Variation of CoOOH concentration (wt.%) and BET surface area with DOC for differentsamples.

Thelowerthanexpectedsurfacearea ofS200HAisduetothepresence,inthissample, ofacertainamountoflCoOOHproducedbythepartialcalcinationofbulkCo(OH)2.The higher than expected surface area for the T400 sample has a somewhat more complicated explanation. This sample was obtained by the hard silica templating method. In this method there are fewer silanol groups available for the formation of hCoOOH[6](andthereforetheDOCishigherthanthatofS400LB).HowevertheCo3O4 fraction is mainly formed inside the silica pores and contributes more to the total specificsurfaceareathanthebulkCo3O4fractionofthesamplespreparedviamethodA (i.e.S500HA).Thisfractioniseitherpartiallyortotallyformedbythemigrationofcobalt ionsawayfromthesilicaaquagel.InordertoclarifythispointTPRanalyseswerecarried out to determine whether any of the spinel phase had been formed inside the silica

Lpezetal.EnviadoaChemCatChem

157

1. CoOOH Co3O4 2. Co3O4(ex-CoOOH)CoO

3. Co3O4(in-SiO2)CoO 4. Co3O4(out-SiO2)CoO CoOCo 2

S400L-B

Normalized H2 consumption

2 S200H-A 1 3 4

S300L-A

CoOCo

Co3O4CoO P200 CoOOH Co3O4

50

150

250

350

450

550

Temperature (C)

Figure4.TPRprofilesofS400LB,S200HA,S300LAandP200catalysts.

158

ArtculoV

matrix during the precipitation procedure. As can be seen in Figure 4, two zones stand outintheTPRprofiles.Accordingtosomeliteraturereports[11,15,16]thezoneatlow temperature (<270C) should be attributed to the reduction of CoOOH and Co3O4 to Co2+,whereasthesecondzone,inthetemperaturerange270520C,shouldbeascribed tothereductionofCo2+toCo0. However, mass balance calculations for the samples whose composition was known (in terms of CoOOH and Co3O4 content; see Table 1) show that the reduction of the cobalt species to Co2+ in the first zone is incomplete. A small shoulder at the beginning of the second zone is situated in a region which is close to that where reduction of bulk Co3O4 toCoOoccursinP200.Wemay,therefore,assumethatthisshoulderisproducedbythe reductionof bulkCo3O4(outSiO2)toCoO. Hence,byconsideringthecompositionofthe catalysts, we can apply a deconvolution procedure that allows the first zone in the TPR profiles to be thoroughly characterized. In this way the amount of Co3O4 which was formed inside the silica matrix (inSiO2) can be estimated. This is shown in the TPR profiles of S400LB and S200HA, the CoOOH contents of which are known (around 80 and55wt.%,respectively).Thefirstpeak(#1)correspondstothereductionofCoOOHto Co3O4. The second peak (#2) represents the reduction of this Co3O4 (exCoOOH) to CoO. Peak #3 is ascribed to the reduction of the spinel phase formed inside the silica matrix (inSiO2) to CoO, whereas peak #4 represents, as already mentioned, the reduction of bulk Co3O4 (outSiO2) to CoO. The rest of the high temperature zone corresponds to the reduction of CoO to Co, as is indicated in the figure. This deconvolution procedure was performed for S200HA, S400LB (both shown in Figure 4), and S500HA. A small increase in the peak temperatures with the calcination temperature (1530C) was observed, while the percentage of total Co3O4 that formed inside the silica matrix (in SiO2) was observed to range between 30 and 45%, the rest being bulk Co3O4 reduced overthe temperaturerangeofpeak#4.ItisworthpointingoutthatP200showsasmall peak at low temperature which is ascribed to the reduction of lCoOOH. This peak is much wider than those obtained for the SACOP samples calcined at 300 or 400C (Figure4), but similar to that obtained for the SACOP sample calcined at 200C (S200HA), possibly due to the higher average particle size in P200 and S200HA. This proves that the CoOOH in the samples calcined at 200C is the sum of the hCoOOH fraction and a small fraction of lCoOOH resulting from the partial calcination of Co(OH)2. Close inspection of specific samples by TEM corroborates to some extent the above picture. Figure 5a shows a TEM image of the S200LA catalyst. As can be seen, the

Lpezetal.EnviadoaChemCatChem

159

sample is composed of amorphous agglomerates of nanoparticles (hCoOOH and Co3O4 formed inside the silica matrix) and more isolated needleshaped nanostructures of length ~100 nm and width 36 nm (lCoOOH and bulk Co3O4 produced by the migration ofcobaltionsoutofthesilicamatrix). Theincreaseinthecalcinationtemperatureappliedtothesamplespreparedbymethod A produced catalysts with the same morphology though some agglomeration of nano needles was also observed (Fig.5b). As expected, the amount of bulk Co3O4 nano needles diminished dramatically in thecaseof theSACOP Bsample(Fig.5c;S400LB). In thecaseofthesamplesynthesizedbydirectprecipitation(Fig.5c;P400)theamountand size of bulk Co3O4 nanoneedles increased, some to lengths of almost 200 nm. The formation of the nanoneedles therefore must have been the result of the free precipitation of Co(OH)2 into the alkaline solution, unlike the amorphous hCoOOH which was produced after direct interaction between the silica matrix and the cobalt ions.

100 nm

200 nm

40 nm

40 nm

100 nm

Figure5.TEMimagesofthecatalyst:(a):S200LA,(b):S400LA,(c):S400LB,(d):P400.

160
3.2.Catalyticactivitytests

ArtculoV

The evolution of the CO conversion corresponding to the three best catalysts at three different temperatures (100C, 150C and 175C) is plotted in Figure 6. After an initial loss of activity during the first 4 hours, the three catalysts show a remarkable stability. Equation E3 was used to derive a parameter (%r) to characterize the stability of the catalystsaftertheinitialdeclineinconversion.Thevaluesofthisparameterareincluded in Table 1, together with those of the reaction rate and selectivity after 20h of reaction at 175C. In most cases, the stability of the catalysts is quite high (with %r values exceeding15%).Itisalsoworthnotingthatthemostactivecatalyst(S400LB)displaysa %r value of 4.5%, a clear indication of its high stability, as is apparent in Fig. 6. The selectivity at 175C of all the catalysts synthesized by silica templating methods exceeded 0.7 (Table 1), which is much higher than the value obtained for the catalyst produced by solution precipitation (P400). The small differences between the catalysts can be ascribed to slight variations in their reducibility. Thus, as can be appreciated in Figure 7, the selectivity values for the SACOP catalysts prepared with a low Co/Si molar ratio follow a clear trend with the peak temperature corresponding to the reduction of CoOOHtoCo3O4.
1 0.8 CO conversion 0.6 0.4 0.2 0 0 4 8 12 Time on stream (h) 16 20 24

175C

150C S400L-B T400 S200L-A 100C

Figure 6. Longterm catalytic activity of S400LB, S200LA and T400, evaluated at differenttemperatures.

Lpezetal.EnviadoaChemCatChem

161

0.85 SACOP (A) SACOP (B) 0.80 Selectivity

0.75

0.70 175

180

185

190

195

200

205

Temperature of TPR peak #1 (C)

Figure7.VariationofselectivitywiththetemperatureoftheTPRpeak#1(seeFig.4)for theSACOPcatalystspreparedatCo/Si=0.25.

The SACOP samples prepared by method A presented PROX reaction rates between 2.0105 and 2.8105 molCOgcat1s1 (Table 1). The samples calcined at 200C exhibited the highest reaction rate values, even though their specific surface areas were significantly lower. As already mentioned, these samples include a small fraction of lCoOOH produced by the partial calcination of bulk Co(OH)2. However, under the reductive atmosphere of the PROX reaction, the lCoOOH fraction (Figure 4; peak #1 in P200 and part of peak #1 in S200HA) is readily transformed into higher surface area Co3O4. In view of the low degree of crystallinity of these samples (Table1), this final transformation may have increased the active surface area of these catalysts to values of over 90 m2g1 (see point corresponding to S200HA in Fig. 3) which would explain their high catalytic activity. If these catalysts are excluded, a clear relation between catalyticactivityandthespecificsurfaceareaofthefreshcatalystsisappreciated(Figure 8). The most active catalyst prepared in this work was that obtained by methodB and calcined at 400C (S400LB), its r20h175C value being 4.3105molCOgcat1s1. This is also thecatalystwiththehighestsurfacearea,which,asmentionedintheprecedingsection, is because of its high amorphous hCoOOH content resulting from the removal of silica fromcobaltsilicatehydroxide,acharacteristicfeatureoftheSACOPprocedure.

162

ArtculoV

r20h175C x107 (molcogcat-1s-1)

500 400 300 200 100 0 0 20 40 60

2

80
-1

100

120

140

SBET (m g )

Figure 8. Variation of r20h175C with SBET for all catalysts calcined at temperatures over

200C. Structured Co3O4 nanocatalysts recently prepared by hydrothermal synthesis [3] have proved to be the most active cobaltbased catalysts for the PROX reaction until now. Such catalysts have PROX reaction rates of around 1.5105 molCOgcat1s1 at 175C (spatialvelocity:33,000h1).However,thecobaltbasedSACOPsamplesdescribedinthis work have been shown to be more active in the PROX reaction, even though they were tested at a higher spatial velocity (~39,000 h1). There can be no doubt that SACOP is a reliable technique for synthesizing highly active, selective and stable cobalt oxide catalystsforthepreferentialoxidationofCO.
4.Conclusions

WehaveanalyzedthecatalyticactivityinthePROXreactionofaseriesofCo3O4/CoOOH catalysts, prepared by silica aquagel confined coprecipitation. The influence of the precipitation step was studied by using two different methods. Pouring the gel into the NaOH dissolution (method A) produced cobalt based nanostructures with BET surface areas between 70 and 90 m2g1, and a balanced mixture of Co3O4 and CoOOH phases. Slowly pouring the alkaline solution into the gel (method B) gave rise to a sample composed mainly of amorphous CoOOH with a high surface area (120m2g1). Most of theCo3O4formedinthefreshsamples viaSACOPislowsurfaceareaCo3O4producedby themigrationofcobaltionsoutsidethesilicamatrixduringprecipitation,aphenomenon that is therefore minimizedby applyingmethodB. Ontheotherhand,highsurfacearea amorphousCoOOHisproducedbyremovingthesilicafromthecobaltsilicatehydroxide

Lpezetal.EnviadoaChemCatChem

163

thatisformedduringprecipitation.UnderPROXconditions,CoOOHisgraduallyreduced to high surface area Co3O4, which is the main contributor to the catalytic activity of SACOP catalysts. The catalysts obtained showed excellent stability and selectivity in the PROXreactionandthehighestvaluesofcatalyticactivityofallthecobaltbasedcatalysts reportedinliteraturetodate.
Acknowledgements

Funding through the FICYT Regional Project IB08103 is acknowledged. IL thanks the MinisteriodeCienciaeInnovacinofSpainforherFPIgrant.

References

[1] I.Lpez,T.ValdsSols,G.Marbn,Int.J.HydrogenEnergy33(2008)197. [2] G.Marbn,I.Lpez,T.ValdsSols,Appl.Catal.A:Gen.361(2009)160. [3] G. Marbn, I. Lpez, T. ValdsSols, A. B. Fuertes, Int. J. Hydrogen Energy 33 (2008)6687. [4] E. Y. Ko, E. D. Park, K. W. Seo, H. C. Lee, D. Lee, S. Kim, Catal. Today 116 (2006) 377. [5] Z.Zhao,M.M.Yung,U.S.Ozkan,Cat.Comm.9(2008)1465. [6] G.Marbn,A.B.Fuertes,T.ValdsSols,Microp.Mesop.Mat.112(2008)291. [7] I.Lpez,T.ValdsSols,G.Marbn,Microp.Mesop.Mat.127(2010)198. [8] T.ValdsSols,G.Marban,A.B.Fuertes,Chem.Mater.17(2005)1919. [9] M.Kruk,M.Jaroniec,A.Sayari,Langmuir13(1997)6267.

[10] W.Liu,M.FlytzaniStephanopoulos,J.Catal.153(1995)317. [11] C.W.Tang,C.B.Wang,S.H.Chien,Thermochim.Acta473(2008)68. [12] T.Vrlstad,H.K.Magnusson,J.Sjblom,Microp.Mesop.Mat.106(2007)155. [13] B.C.Dunn,P.Cole,D.Covington,M.C.Webster,R.J.Pugmire,R.D.Ernst,E.M. Eyring,N.Shah,G.P.Huffman,Appl.Catal.A:Gen.278(2005)233. [14] T.N.Ramesh,J.SolidStateChem.183(2010)1433. [15] H.Wang,H.Zhu,Z.Qin,G.Wang,F.Liang,J.Wang,Cat.Comm.9(2008)1487. [16] L.Xue,C.Zhang,H.He,Y.Teraoka,Appl.Catal.B:Environ.75(2007)167.

165

ResumenartculoVI. HIGHLYACTIVESTRUCTUREDCATALYSTMADEUPOFMESOPOROUS CO3O4NANOWIRESSUPPORTEDONAMETALWIREMESHFORTHE PREFERENTIALOXIDATIONOFCO

Catalizadoresestructuradosconstituidospornanoalambresde Co3O4mesoporososoportadosenunamallametlicaconelevada actividadenlaoxidacinpreferencialdeCO
En este trabajo se analiz la actividad cataltica de xido de cobalto soportado en una malla metlica de acero inoxidable en la oxidacin preferencial de CO. La caracterstica ms sobresaliente de este material es que, al estar ya inmovilizadas las nanopartculas enunsoporte,semanejanconmsfacilidady,dadaslaspropiedadesdelamalla,suuso en un microrreactor produce una cada de presin casi nula adems de mejorarse la transmisindecalor. Para la obtencin de este catalizador se emple el mtodo de evaporacin inducida por amoniaco, que ya se explic en la seccin Experimental. Este procedimiento permite obtener tanto la espinela de cobalto soportada en la malla como agregados de partculas no adheridas a la superficie metlica. La espinela de cobalto soportada present un rea superficial de 71 m2g1 y el rendimiento logrado fue de 7 mg de espinela por cm2 de malla. Mediante microscopa electrnica de barrido se observ que laspartculassoportadascrecanenformaderamilletesdenanoalambresmesoporosos. Dichos alambres tienen una longitud de 30 m y una anchura de 500 nm. Las partculas no adheridas presentaron tamaos de cristal de 11 nm y un rea superficial de93.5m2g1. Para evitar la disgregacin de las partculas de espinela adheridas a la malla metlica durante su manipulacin, sta se recubri con una capa de alcohol polivinlico. Una vez protegida, se enroll y se introdujo en un reactor de acero de 1/4. El conjunto se lav con abundante agua para eliminar la capa protectora. Por ltimo se unieron sendas tuberas para la entrada y salida de gases. En el caso de las nanopartculas sueltas se utiliz un reactor de cuarzo de 7 mm de dimetro interno en el que se introdujeron 30mg de catalizador diluido en 270 mg de SiC, entre dos tapones de fibra de vidrio. Las condiciones experimentales empleadas fueron las mismas en ambos sistemas de

166

ArtculoVI

reaccin,aexcepcindelamasadecatalizador,queenelcasodelcatalizadorsoportado fuefuncindeltamao delamalla.Por estaraznlasvelocidades espacialesempleadas (F0CO/wcat) fueron 6.8106 molgcat1s1 para las mallas y 2.2106 molgcat1s1 para las partculassueltas. La actividad cataltica de la espinela de cobalto soportada en la malla metlica dobl la actividad obtenida para las partculas sueltas, adems de resultar muy estable en el intervalo de temperaturas analizado (100175C). La formacin de metano, que es una delasdesventajasdeloscatalizadoresbasadosencobalto,soloseobservenelcasode laespinelanosoportada.Laselectividadhacialaformacindemetanoconrespectoala produccindeCO2fuedel0.005%. La actividad cataltica de estos catalizadores se compar con la de nanocatalizadores de espinela de cobalto sintetizados mediante un mtodo estndar de nanomoldeo, observndose una mayor actividad cataltica en las partculas sintetizadas en este trabajo,tantosoportadascomoenformananoparticulada. La comparacin de la actividad de las muestras preparadas para este trabajo con otras descritas en la literatura, permiti establecer una correlacin entre la temperatura de calcinacin y la actividad de las muestras. A menor temperatura de calcinacin, ms reducible result ser la muestra y por tanto mayor fue su actividad cataltica en la reaccinPROX.

167

ArtculoVI

1.Introduction Hydrogen streams for fuelcell powered cars need to be almost completely free of CO poisoning [1, 2]. To achieve this requirement, hydrogen produced by onboardsteamreformingofaliquidfuel (e.g.methanol)mustbesubjectedtothe preferential (selective) catalytic oxida

tion of CO (PROX reaction). Among the transition metal oxides that are poten tialcandidatesforuseinthisprocessare cobalt oxide catalysts which are known to be active catalysts in the low tem perature oxidation of CO [36], although it is only recently that some authors have analyzed the performance of these

168 oxides in the PROX reaction, either alone [7], supported [8, 9] or accompa nied by other active phases, such as Au and CeO2 [10, 11]. Some of these cata lysts are described in Table 1. The lim ited number of studies on these cata lysts might be due to their apparently low activity compared to other more commonly studied catalysts such as CuO/CeO2 mixtures [11], or to the ten dency of these catalysts to produce the undesired methanation of CO, even at temperatures below 200C [7]. Recently Li et al. [12, 13] developed a very simple synthetic procedure to produce arrays of Co3O4 nanowires with excellent prop erties for use as anodes in Li ion batter ies. The nanoparticles obtained have a relatively high surface area (~74 m2/g) and a mesoporous texture with an aver ageporesizeof3.3nm.Theabovemen tioned authors fabricated the nanoparti cles on different holding media (Ti foils, Siwafers,etc)withsatisfactoryadhesion properties. In principle all these charac teristics make the nanowires suitable candidates for use in the PROX reaction. In the present work we employed this syntheticapproachtofabricatearraysof nanowires on a metal wire mesh for use as a catalytic filler in a structured micro reactor. In this kind of microreactor, which is based on similar commercial systems [14], the metal wire mesh pro vides a high geometric surface area for holding the Co3O4 nanowires with a neg

ArtculoVI ligible pressure drop and it also allows a good heat transfer through the reactor. The synthetic procedure simultaneously produces unsupported mesoporous Co3O4 particles with slightly different characteristicstothoseofthesupported system. Both systems (loose particles and structured microreactor) were tested in this work in the PROX reaction attemperaturesbelow200C. 2.Experimental 2.1. Catalyst synthesis and fabrication of thestructuredreactor Themethodusedtosynthesizethecata lysts is based on that reported by Li et al. [13]. It is a mild templatefree me thod known as Ammoniaevaporation induced method. During a standard preparation 35 g of H2O and 15g of 30wt.% ammonia solution were con secutively poured without agitation into a Teflon vessel containing 10 mmol Co(NO3)2 and 5 mmol NH4NO3. The mix ture was then magnetically stirred for halfanhourinair.Themagnetwasthen removed from the vessel and a rectan gular piece (53 cm) of stainless steel wire mesh [29 m wire diameter and 45m screen opening] that had been previously washed in HNO3 (4M) at 60C for several hours was introduced into the reaction solution. The vessel with the solution and the mesh was covered by a watch glass and heated in an oven at 90C for 17 hours in order that the

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

169

nanowires could grow. After the synthe sisthemetalwiremeshwastakenoutof the solution and the particles remaining in the solution were filtered and thor oughly washed with deionized water and vacuum dried at 50C. The wire mesh was also washed and dried in the same manner. Finally the samples were calcined in air at 250C for 2 hours. The acronyms for the loose (unsupported) Co3O4 particles and the metal wire meshsupported Co3O4 particles will be LPCo3O4andWMSCo3O4,respectively. For the construction of the microreactor the wire mesh must be first rolled and insertedinastainlesssteeltube.Apriori it seemed more appropriate to load the mesh into the tube before proceeding with the synthesis step. This procedure prevented the nanowire arrays from growing on the metallic wire mesh sur face during the synthesis step, probably due to mass transfer limitations within the tubes filled with the mesh. For this reason the unfolded wire mesh was first coated outside the tube during the syn thesis step and then inserted into the tube. In order to fill the stainless steel tube (" i.d., 3 cm height) with the coated metal wire mesh, it was rolled up tightly by hand. This operation dislodged many of the particles from the wire mesh. To preventthisfromoccurringtheunfolded mesh was previously immersed in a

2wt.% aqueous solution of poly(vinyl alcohol) (9,00010,000 g/mol, 80% hy drolyzed), then air blown to eliminate the excess of solution and vacuum dried at 40C for 20 minutes. This procedure formedaprotectivecoatingofpoly(vinyl alcohol) (PVA) on the supported parti cleswhichallowedthemeshtoberolled up without undermining the adhesion of the particles to the metal wires. Once insidethetube,themeshwasimmersed in a water bath for several hours in or der to remove the water soluble PVA coatingandvacuumdriedat50C. For the purpose of comparison a sample of Co3O4 nanoparticles was prepared by a silica template method as reported elsewhere [15]. These particles have a BET specific surface area of 97 m2/g, a particle size of 10.4 nm (obtained by applying Scherrer's equation to the XRD spectrum) and a very low residual silica content (2.2 wt.%). The acronym for this sampleisSTCo3O4. 2.2.Catalystcharacterization Nitrogen adsorption isotherms were performed at196C on a Micromeritics ASAP 2020 volumetric adsorption sys tem. The BET surface area was deduced from the isotherm analysis applied in the relative pressure range of 0.04 to 0.20. The Pore Size Distribution (PSD) was calculated by means of the Kruk JaroniecSayari method [16]. Micro scopic images of the samples were ob

170 tained using a Scanning Electron Micro scope (Mod. DSM 942, Zeiss). Xray dif fraction(XRD)analysiswascarriedoutin a Bruker instrument (D8 Advance) oper ating at 40 kV and 40 mA and using CuK radiation (=0.15406 nm). TPR analyses were performed in a chemi sorption analyzer (Autochem II) equippedwithaTCDreactorandaMass Spectrometer (OmniStar 300O). In a given analysis approximately 25 mg of sample was treated with a 50 mL/min stream of 10%H2 in Argon from 100C to 850Cat5C/min. 2.3.Catalyticactivitytests Catalytic activity tests were performed on the loose particles by using a quartz catalyst particles was diluted in 270 mg of SiC particles and placed in the reactor between two zones of glass wool. This ensured plugflow and isothermal condi tions reactor of 7 mm internal diameter insertedintoaverticalfurnace.30mgof withinthebed.Atthegasflowrateused in the experiments (200 mL/min) this configuration resulted in a high space velocityof~55,000h1.Thebedtempera ture was controlled by means of a ce ramicprotected thermocouple inserted into the reactor. The samples were heated under a He flow at 200C for

ArtculoVI 60min. The reactant stream was then passed through the catalyst and its cata lytic activity and selectivity were evalu ate at de creasing temperatures from 175C to 100C. The reactant gas feed contained 25 vol.% H2, 0.5 vol.% CO and 0.5 vol.% O2 in He. After removing the water with Mg(ClO4)2, the product gases were analyzed for H2, CO, CO2, CH4, O2 and N2 by means of a gas chromato graph (HP 6890) equipped with a TC detector. The conversion of CO was eva luated as X = 1[CO]out/[CO]in and the selectivity of oxygen to form CO2 was calculatedas: S=0.5([CO]in[CO]out)/([O2]in[O2]out). To test the PROX performance of the structured microreactor (a 3 cmhigh tube filled with the metal wire mesh supported Co3O4 nanowires) the quartz reactor was removed and replaced by the structured unit, which was attached to the system by means of Swagelock fittings. Except for the weight of the catalyst, which was a function of the wire mesh surface area, all testing con ditions were identical to those used to test the loose particles (total flow rate and gas composition). The conditions employed for both systems are summa rizedinTable1.


Support SiO2 Al2O4 176 59 38 7 87 n.i. 94 71e 1.9104 2.2105d 6.8106f 7.4106 1102 1102 5103 5103 1.0 80 0.5 93.5 1.0 1.0 1.5 50 50 50 TiO2 ZrO2 CeO2 SrCO3 none metalwire Co3O4(Air mesh 250C3h) Co3O4(Air 500C5h) 193 1.0 60 1.9106 1102 1.6 0.2 1.4 4.6 1.7 7.4c 4.6 7.0 14.6 14.5 Activephase SBET FCO0/wcat (calcination) (m2/g) [mol/(gs)] PCO0 PO20/ PH20/ rPROX (bar) PCO0 PCO 107a TS=0.8 (C)b 114 10 P (NaOH) EI NH3EIM CoO (H2200C) n.i. (Air450C) Co3O4(Air 250C3h) IW Smax=0.35 at50C 140 152 135 Smax=0.52 at140C 143 140 125 125 Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

Table1.PropertiesofcobaltbasedcatalystssynthesizedbydifferentmethodsandacomparisonoftheircatalyticactivityduringPROX.

Ref.

Cobaltcontent Method (wt.%)

Zhaoetal.[8]

Koetal.[7]

Omataetal.[9]

78.6 (pureoxide) 1.3

LPCo3O4

MonolithWMS Co3O4

73.4 (pureoxide) 18.4(Mesh= 75wt.%)

n.i.notindicated;P:precipitation;IW:incipientwetness;WI:excessimpregnation;NH3EIM:ammoniaevaporationinducedmethod. aPROXreactionrateat100C(molCOgcat1s1). bTemperatureabovewhichS>0.8[whenS<0.8overthewholetemperaturerangethenthemaximunvalueofselectivity(Smax)is shown]. cCatalyticactiviytestperformedinthepresenceof2102barH2O. dSpatialvelocity=55,000h1. eSurfaceoftheCo3O4particles(meshexcluded). fThecatalystweightisconsideredtobethatofCo3O4(meschexcluded).Spatialvelocity=33,000h1.

171

172 To test the PROX performance of the structured microreactor (a 3 cmhigh tube filled with the metal wire mesh supported Co3O4 nanowires) the quartz reactor was removed and replaced by the structured unit, which was attached to the system by means of Swagelock fittings. Except for the weight of the catalyst, which was a function of the wire mesh surface area, all testing con ditions were identical to those used to test the loose particles (total flow rate and gas composition). The conditions employed for both systems are summa rizedinTable1. 3.Resultsanddiscussion 3.1. Structural properties of the unsup portedmesoporousCo3O4particles The synthesis process produced unsup ported particles of a black color, typical of Co3O4 spinel, with 3640% yields of the cobalt that was originally present in the precursor solution. XRD analysis

ArtculoVI clearly confirms the formation of spinel, as observed in Fig.1. The application of Scherrer'sequationtothemainpeaksof the XRD spectrum yields a crystal size of ~11nm. The BET surface area of the particles was found to be 93.5 m2/g, higher than that obtained in the original work by Li et al. [13] (74m2/g). Calcula tion of the effective particle size from theBETsurfacearea,assumingspherical geometry, gives a value of 10.6 nm, which implies that the porosity of the microparticles is made up of single (nonagglomerated) crystals, thereby confirming previous results [12]. The porosity of the particles prepared in this work is made up of mesopores of a rela tively uniform size within the 27 nm range. This is illustrated in Fig. 2 which shows the nitrogen sorption isotherm and PSD (inset) of this material. The pore size distribution shows a peak at 4.1 nm which is a bit higher than the mean mesopore size obtained in the workbyLietal.[13](3.3nm).

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

173

(311) Intensity (a.u.)

(111)

(220) (511) (422)

(440)

(222) (400)

10

20

30

40 2 ()

50

60

70

Fig. 1.XRDspectrumoflooseCo3O4particles.Thehklindexesindicatedintheplotcorre spondtotheCo3O4spinelstructure.

80 dV/dlog(D), cm /g
3.4 nm

70 Adsorbed volume, (cm STP/g) 60 50 40 30 20 10 0 0.0

0.3 0.2 0.1 0.0 1

4.1 nm

10 100

Pore size (D), nm

LP-Co3O4 Co3O4 in WMS-Co3O4

0.2

0.4 0.6 Relative pressure (p/po)

0.8

1.0

Fig. 2. Nitrogen sorption isotherms and pore size distributions (inset) of loose Co3O4 particlesandmetalwiremeshsupportedCo3O4(onaspinelmassbasis). Fig. 3 shows SEM images of the me soporousCo3O4particles.Ascanbeseen in Fig. 3A and 3B, the particles have a size of around 100 m and are formed by a nucleus coated with the nanowires. In Fig. 3A and 3B most of the mass of

174 eachparticleseemstobeconcentratedinthenucleus.

ArtculoVI

Fig.3.SEMimagesofloose(unsupported)Co3O4particles. Li et al. [12] obtained a homogeneous pink solution when mixing the precur sors but they did not report the produc tion of loose particles at the end of the synthesis. In our case, as we mixed the precursors, we were careful not to agi tate the mixture. We thus obtained a reddish solution which readily produced ablacksolid oflowdensity.Someofthis floatedandwasclearlyvisibleduringthe magnetic stirring. We believe this pre cipitate to be the origin of the nuclei that provide additional free surface for thegrowthofthenanowires(Fig.3Cand 3D). The XRD results (Fig. 1) show that these nuclei have the same composition as the nanowires (Co3O4), and must therefore be responsible for the at tenuation of the intensity of peak (111), compared to that of the nanowires [12,

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

175

13]. What is more, the nuclei must be at least as porous as the nanowires, oth erwise the specific surface area of the composite material (nucleus + nano wires) would be much lower than that reported by Li et al. [13] for the individ ualnanowires.Theporosityinthenuclei might have been created during the thermal treatment at 90C via the same mechanism that produces the porosity in the nanowires [12]. This mechanism involves a solid reaction between the precipitated brucite (Co(OH)2) and the dissolved oxygen in the solution, giving rise to the Co3O4 spinel and the topotac tic reorganization of the solid material thatoriginatesthemesopores. The nanowires that have grown on the surface of the nuclei have diverse orien tations, from random to almost parallel to the nucleus axis (Fig. 3B and 3C). The nanowires have a length which varies from ~10 to ~40 m (some of the small est wires seem to have been produced by breakage during handling) and a di ameterof~500nm(Fig.3D).Thegrowth of the nanowires seems to be a result of the accumulation of brucite scales of size~300nm(Fig.3D).

3.2.Structuralcharacteristicsofmeso porousCo3O4nanowiressupportedona metalwiremesh As explained in the experimental sec tion, a clean stainless steel wire mesh was inserted into the precursor solution in order to produce simultaneously the loose Co3O4 particles (see former sec tion) and the metal wire mesh sup portedCo3O4 particles. Fig. 4 shows an image of the metal wire meshes before (Fig. 4A) and after (Fig. 4B) the synthesis process. As can be seen the coating of the metal wire mesh is very homogene ous. A mesh of 53 cm2 could hold 124mgofCo3O4particles, weighedafter calcination at 250C. After coating the meshwithPVA,insertingitintothetube and washing the PVA coating off, the final weight of the supported Co3O4 was found to be ~100 mg, with a yield of ~7mg/cm2, which compares favorably with the 23 mg/cm2 obtained on Ti foils by Li et al. [13], although in their case the geometric surface area was smaller than that of the metal wire mesh em ployedinthisstudy.

176

ArtculoVI

Fig. 4. Illustration of the process of fabrication of the catalytic microreactor. (A)Metal wire mesh; (B) Metal wire mesh coated with Co3O4 nanowires; C) Reactor scheme and image. 1) Synthesis of Co3O4 nanowires and calcination; 2) Rolling and insertion of the Co3O4nanowires/Metalwiremeshcompositeintoastainlesssteeltube. Fig. 2 shows the nitrogen sorption iso thermandthePSDoftheCo3O4particles in the WMSCo3O4 sample (support ex cluded). The specific surface area of the supported particles was found to be 71m2/g. As expected, this value is very similar to that obtained by Li et al. [13] (74 m2/g). The mean pore diameters for the WMSCo3O4 sample and the sample prepared by Li et al. are also almost equal (3.4 vs. 3.3 nm). Fig. 5 presents SEM images of both the clean metal wire mesh and the supported Co3O4 particles. As can be seen, the spinel par ticles are arranged in flowerlike nanowire arrays that are evenly spaced on the metal wire mesh (Figs. 5A and 5B) in the manner represented by the equivalent rhomboids shown in Fig.s 5A and 5B. Each array is composed of nanowires with a length of ~30 m (Fig.5C) and a width of ~500 nm (Fig. 5D). The nanowires prepared by Li et al. [12, 13] are hollow, unlike the nano wires grown on the metal wire mesh, which are solid (inset in Fig. 5D). This implies that the growing substrate must have a specific effect on the final ge ometryofthenanowires.Asobservedin Fig.s 3D and 5D, the nanowires grown on the metal surface are identical to thoseformedonthesurfaceoftheloose Co3O4particles.

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

177

Fig. 5. SEM images of the clean metal wire mesh (A) and the supported Co3O4 particles (BCD). NoPVAcoatingwasdetectedbySEMon the surface of the nanowire arrays, thoughitseffectwasclearlyestablished: passing ones finger gently over the un protectedmetalwiremeshcoveredwith the spinel had the effect of dislodging many black particles from its surface. After the mesh had been protected with a PVA coating, no particles were dis lodged from the mesh surface. We be lieve that the PVA tends to concentrate at the junctions of the metal wire mesh and the nanowire arrays, thus reinforc ing their adhesion, although being in a blind zone, the PVA remains undetected bySEM. 3.3.Catalyticactivitytests Fig. 6 shows the catalytic activity results for the unsupported Co3O4 particles prepared by the ammoniaevaporation induced method. The formation of methane was observed only at the high

178 est temperature (175C) and then only to an insignificant extent (the conver sion of CO to CO2 was 0.99 whereas the conversion of CO to CH4 was just 0.005). The selectivity was relatively low at 175C (S~0.5) but increased significantly at temperatures equal to or below 150C (S>0.7). However, at these tem peratures the stability of the catalyst was not as good as that achieved at
1.0 CO conversion, X 0.8 0.6 0.4 0.2 0.0 1.0 CO selectivity, S 0.8 0.6 0.4 0.2 175C 0.0 0 10 20 150C 175C 150C ST-Co3O4 LP-Co3O4

ArtculoVI 175C. Fig. 6 also shows the results ob tained with the catalyst prepared by a silica template method [15] (STCo3O4) at 175C. Although this catalyst has very similar structural properties to LPCo3O4 (97 m2/g vs. 93.5 m2/g and 10.4nm vs. 11nmcrystalsize)itscatalyticactivityat 175C is considerably lower, as can be observedinFig.6.

125C

ST-Co3O4 LP-Co3O4

125C 30 40

100C 50

time on stream (h)

Fig.6.EvolutionoftheCOconversionandselectivityoftheLPCo3O4catalystatdecreas ingtemperaturesandoftheSTCo3O4catalystat175Cforlongrunexperiments. TPR analyses were performed on both samples in order to determine the rea son for the difference in catalytic activ ity. Fig. 7 shows the H2TPR profiles for LPCo3O4 and STCo3O4. Two peaks are clearly observed in the TPR profile of LP Co3O4 ( and ). The same peaks are observed for the STCo3O4 sample, though they have shifted ~30C to high er temperatures and the spectrum now

100C

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

179

displays a broad feature from ~350 to ~500C (maximum at 420C). Peak is attributed to the Co3+ Co2+ transition,

whereas peak and the broad high temperature feature are attributed to thereductionofCo2+toCo0[10,17].

LP-Co3O4

295C ()
Peak temperature (C Peak temperature (C)

450

Wang et al., 2008 ST-Co3O4 LP-Co3O4 Peak

350 Ko et al., 250 2006

Peak

Normalized H2 consumption

232C ()

Polynomial fits 150 0 200 400 600 Calcinationtemperature temperature (C) Calcination (C)

ST-Co3O4

324C () 261C ()

420C

100

150

200

250

300

350

400

450

500

550

600

Temperature, C

Fig. 7. H2TPR profiles of the LPCo3O4 and STCo3O4 catalysts. In the inset, effect of the calcinationtemperatureonthetemperatureofthereductionpeaksfortheLPCo3O4and STCo3O4catalystsandtwoselectedCo3O4samplesdescribedintheliterature[7,10]. The same peaks were also observed by Ko et al. [7], but in their catalyst the low temperature peaks ( and ) were situ ated at 207 and 257C, well below those obtained with our catalyst (232 and 295C; Fig. 7). The Co3O4 particles syn thesized by Ko et al. [7] by precipitation with NaOH (1 M) at pH=8, were not cal cinedbutagedat80C(1h)anddriedat 100C (overnight). A similar profile to that of the upper plot in Fig. 7 was ob tained by Wang et al. [10] for Co3O4

180 particles prepared by solgel precipita tion followed by calcination at 500C (18.3 nm crystal size), but the two char acteristic peaks and have shifted to higher temperatures (315 and 425C). Thus, there seems to be a connection between the calcination temperature and the reducibility of the catalyst, which can be clearly observed in the insetinFig.7. Among the catalysts used to construct this plot, LTCo3O4, STCo3O4 and the
1.0 CO conversion, X 0.8 0.6 0.4
150C P0O2 P0CO = 1.0 P0O2 P0CO = 1.5

ArtculoVI catalyst of Ko et al. [7] were also tested in the PROX reaction, revealing the fol lowing order of activity: Ko et al. [7] > LTCo3O4 > STCo3O4 (Table 1 and Fig. 6). Althoughtheactivephaseofthecatalyst of Ko at al. [7] is CoO (the catalyst was prereduced in H2 at 200C) there is a clear relation between the catalytic ac tivity and the reducibility of the sample, which is in turn dependent on the calci nationtemperature.

125C

125C

175C

0.0 CO selectivity, S 0.8 0.6 0.4

175C

150C

100C

175C

100C

150C

125C

125C

175C

0.0 0

175C

150C

100C

175C

100C

10

20

30

40

50

60

16

24

time on stream (h)

time on stream (h)

175C

0.2

175C

0.2

32

Fig. 8. Evolution of the CO conversion and selectivity of the microreactor filled with the wire mesh supported Co3O4 particles during a long run experiment at decreasing tem peraturesandatP0O2/P0CO=1(left)andP0O2/P0CO=1.5(right). Fig. 4 shows an illustration of the fabri cation process of the catalytic microre actor (photograph in Fig. 4C). As men tioned above, the catalyst was fabri

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

181

cated by filling the " stainless steel tubewitharollofthecoatedmetalwire mesh (WMSCo3O4). Fig.8 shows the catalytic activity results obtained with the microreactor (WMSCo3O4). Com pared to the unsupported particles (Fig.
1.0E-04 X close to 1

6) the structured catalyst displays a lower activity at high temperatures but a higher performance at low tempera tures. This can be more easily appreci atedfromtheArrheniusplotinFig.9.

-1 rPROX (molCOgcat-1s )

X < 0.8 1.0E-05

1.0E-06

LP-Co3O4 (P0O2/P0CO=1.0) WMS-Co3O4 (P0O2/P0CO=1.0) WMS-Co3O4 (P0O2/P0CO=1.5)

1.0E-07 0.0018

0.002

0.0022

0.0024 1/T (K-1)

0.0026

0.0028

0.003

Fig. 9. Arrhenius plots of the PROX reaction rate (rPROX) evaluated for the LPCo3O4 and WMSCo3O4samples. The lower activity at high temperatures is partially explained by the limited availability of oxygen. In fact, at 175C and P0O2/P0CO = 1 there is not enough oxygen to burn all the CO, because of the high catalyst mass and its low selec tivity at this temperature (S~0.45). When the partial pressure of O2 was increased to P0O2/P0CO = 1.5 there was a clearincreaseinCOconversionat175C, althoughthisfailedtoreachthevalueof 1(Fig.8),possiblyduetoasmallbypass ingthroughthecentralholeoftherolled wiremesh.ThusthepointsofFig.9situ atedin thezoneofconversioncloseto1 cause an unnatural change in the slope of the Arrhenius curve because they are biased by the high catalyst mass em ployed in the analyses, whereas the points situated in the zone of X<0.8 ex press the intrinsic activity of the catalyst more accurately. Table 1 shows that the WMSCo3O4 sample is the most active catalystoutofallthoselisted.Itsactivity is, in general, one order of magnitude higher than that of the catalysts pre pared by Zhao et al. [8] and 23 times higher than that of the catalysts pre

182 pared by Ko et al. [7] and by Omata et al. [9]. However, the most remarkable feature of the structured catalyst is its stability for reaction periods of several hours under most of the conditions studied in this work, as can be clearly observed in Fig. 8. The catalyst was sub jected to the PROX reaction at decreas ing temperatures from 175C. After wards, when the temperature was in creased again to 175C, the initial con version was recovered. In conclusion, the structured catalyst demonstrated an unprecedented catalytic activity during the PROX reaction together with a high stability. 4.Conclusions A novel and simple synthetic procedure for preparing particulate and metal wire mesh supported mesoporous Co3O4 nanoparticles by a simple onestep pro cedure is presented. The unsupported mesoporous Co3O4particles are formed by a nucleus coated with solid nano wires. These particles exhibit a catalytic activity in the PROX reaction similar to that of the most active Co3O4 catalysts reported in the literature, although at

ArtculoVI temperatures below 175C they experi ence some deactivation. A comparison with other Co3O4 particles prepared at different calcination temperatures showedthat thereisaninverserelation ship between the calcination tempera ture and the reducibility of the spinel, which in turn seems to be proportional tothecatalyticactivityoftheparticles. Thestructuredcatalyticsystemmadeup of mesoporous Co3O4 nanowires sup ported over a metal wiremesh exhibits several important properties: a) a high heat exchange rate and a low pressure drop, b) a remarkable stability over the whole temperature range analyzed (100175C)andc)agoodcatalyticactiv ity which doubles that of the unsup portedCo3O4particles. Acknowledgements Funding through the Spanish National Project MAT200500262 and the FICYT Regional Project (IB05001) is acknowl edged. IL thanks the Ministerio de Cien cia e Innovacin of Spain for her FPI grant and TVS thanks the Junta General del Principado de Asturias for the award ofaClarnpostdoctoralcontract.

Reference [1] G.Marbn,T.ValdsSols,Int.J.HydrogenEnergy32(2007)1625. [2] G.Marbn,T.ValdsSols,Int.J.HydrogenEnergy33(2008)927. [3] D.Perti,R.L.Kabel,AIChEJ.31(1985)1420.

Marbnetal.Int.J.Hydrog.Energy33(2008)66876695

183

[4] E. Gulari, C. Gldr, S. Srivannavit, S. Osuwan, Appl. Catal. A: Gen. 182 (1999) 147. [5] J.Jansson,A.Palmqvist,E.Fridell,M.Skoglundh,L.sterlund,P.Thormhlen,V. Langer,J.Catal.211(2002)387. [6] J.Jansson,M.Skoglundh,E.Fridell,P.Thormhlen,Top.Catal.16/17(2001)385. [7] E. Y. Ko, E. D. Park, K. W. Seo, H. C. Lee, D. Lee, S. Kim, Catal. Today 116 (2006) 377. [8] Z.Zhao,M.M.Yung,U.S.Ozkan,Cat.Comm.9(2008)1465. [9] K.Omata,Y.Kobayashi,M.Yamada,Catal.Today117(2006)311.

[10] H.Wang,H.Zhu,Z.Qin,G.Wang,F.Liang,J.Wang,Cat.Comm.9(2008)1487. [11] G.Marbn,A.B.Fuertes,Appl.Catal.B:Environ.57(2005)43. [12] Y.Li,B.Tan,Y.Wu,J.Am.Chem.Soc.128(2006)14258. [13] Y.Li,B.Tan,Y.Wu,NanoLett.8(2008)265. [14] M.Lyubovsky,S.Roychoudhury,Appl.Catal.B:Environ.54(2004)203. [15] T.ValdsSols,G.Marban,A.B.Fuertes,Chem.Mater.17(2005)1919. [16] M.Kruk,M.Jaroniec,A.Sayari,Langmuir13(1997)6267. [17] L.Xue,C.Zhang,H.He,Y.Teraoka,Appl.Catal.B:Environ.75(2007)167.

185

Conclusionesgenerales

En la presente memoria se han sintetizado y analizado distintos catalizadores en la reaccin de oxidacin preferencial de CO en presencia de elevadas concentraciones de hidrgeno. A continuacin se muestran las principales conclusiones derivadas de este trabajo. Se ha analizado la actividad cataltica en la oxidacin preferencial de CO de catalizadores basados en xido de cobre descritos en la literatura. Se ha comprobado que los catalizadores ms activos son aquellos que incluyen xido de cerio en su formulacin. El mtodo ms efectivo para la sntesis de catalizadores activos se basa en el mtodo solgel con la adicin de agentes quelantes. El mtodo de coprecipitacin confinada en hidrogeles de slice o SACOP es una ruta sinttica que permite obtener nanoestructuras con elevadas reas superficialesdebidoalcolapsodelaslicesobreloshidrxidosmetlicosdurante laetapadeprecipitacin.Estemtodohasidoutilizadoconxitoparalasntesis tanto de nanoestructuras de CeO2 como de nanocatalizadores CuO/CeO2y CuMn2O4quemuestranunaelevadaactividadenlareaccinPROX,sibienestn sujetos a cierto grado de desactivacin cuyas causas han sido estudiadas en profundidad. Los nanocatalizadores CuOx/CeO2, a temperaturas inferiores a 250C se desactivan principalmente debido a la formacin de carbonatos superficiales. Adems de este mecanismo se produce la sinterizacin parcial de las especies Cu2+ disperso para formar clsteres de CuO. El aumento de la temperatura incrementa la importancia de la desactivacin producida por la sinterizacin de lasespeciesdecobre,alavezquedisminuyelaformacindecarbonatos. La desactivacin de la espinela de cobre y manganeso, CuMn2O4, puede ser atribuida a la progresiva reduccin del cobre en la superficie del catalizador por laaccindelCO.

186

Conclusiones El mtodo SACOP permite obtener los catalizadores basados en cobalto ms activos para la reaccin PROX, con una notable selectividad y gran estabilidad. Este hecho se debe a que el mtodo SACOP produce nanocatalizadores con una elevadaconcentracinde CoOOHdealtasuperficiequeencondicionesPROXse transformaenCo3O4degranactividadcataltica. Se han sintetizado nanocatalizadores de espinela de cobalto soportados en mallas de acero inoxidable mediante el mtodo de evaporacin inducida de amoniaco. La fase activa de estos nanocatalizadores posee una estructura de nanoalambres de espinela de cobalto (SBET=71 m2g1), con una longitud de 30 myunaanchurade500nm.Estoscatalizadoresestructuradospresentan actividadescatalticaselevadasyunanotableestabilidadenlosensayosdelarga duracinenlareaccinPROX.

187

Bibliografa
[1] G.Marbn, T.ValdsSols. Towards the Hydrogen Economy?, Int. J. Hydrogen Energy32(2007)16251637. [2] G.San Miguel, J.Dufour, J.A.Calles, J.A.Botas. La economa del hidrgeno una visin global sobre la revolucin energtica del siglo XXI. 2. Aplicaciones convencionales del hidrgeno y pilas de combustible, Acta Cientfica y Tecnolgica.RevistadelaAsociacinEspaoladeCientficos10(2006)2127. [3] L.Schlapbach, A.Zuttel. Hydrogenstorage materials for mobile applications, Nature414(2001)353358. [4] A.Sartbaeva, V.L.Kuznetsov, S.A.Wells, P.P.Edwards. Hydrogen nexus in a sustainableenergyfuture,Energy&EnvironmentalScience1(2008)7985. [5] M.Sevilla, N.Alam, R.Mokaya. Enhancement of Hydrogen Storage Capacity of ZeoliteTemplated Carbons by Chemical Activation, The Journal of Physical ChemistryC114(2010)1131411319. [6] Z.Yang, Y.Xia, R.Mokaya. Enhanced Hydrogen Storage Capacity of High Surface AreaZeolitelikeCarbonMaterials,J.Am.Chem.Soc.129(2007)16731679. [7] E.Tzimas, C.Filiou, S.D.Peteves, J.B.Veyret. Hydrogen storage: stateoftheart and future perspective, http://setris. jrc. ec. europa. eu/docs/files/H2 storage report.pdf(2010). [8] J.M.Thomas, W.J.Thomas. Principles and practice of heterogeneous catalysis, WileyVCH,1997. [9] O.Iyoha, R.Enick, R.Killmeyer, B.Howard, B.Morreale, M.Ciocco. Wallcatalyzed watergas shift reaction in multitubular Pd and 80wt%Pd20wt%Cu membrane reactorsat1173K,J.Membr.Sci.298(2007)1423.

[10] G.Marbn, A.B.Fuertes. Highly active and selective CuOx/CeO2 catalyst prepared by a singlestep citrates method for preferential oxidation of carbon monoxide, Appl.Catal.B:Environ.57(2005)4353. [11] M.J.Kahlich, H.A.Gasteiger, R.J.Behm. Kinetics of the selective CO oxidation in H2richgasonPt/Al2O3,J.Catal.171(1997)93105. [12] F.Mario, C.Descorme, D.Duprez. Noble metal catalysts for the preferential oxidation of carbon monoxide in the presence of hydrogen (PROX), Appl. Catal. B:Environ.54(2004)5966.

188

Bibliografa

[13] M.Watanabe, H.Uchida, K.Ohkubo, H.Igarashi. Hydrogen purification for fuel cells: selective oxidation of carbon monoxide on PtFe/zeolite catalysts, Appl. Catal.B:Environ.46(2003)595600. [14] R.J.H.Grisel, C.J.Weststrate, A.Goossens, M.W.J.Craj, A.M.van der Kraan, B.E.Nieuwenhuys. Oxidation of CO over Au/MOx/Al2O3 multicomponent catalystsinahydrogenrichenvironment,Catal.Today72(2002)123132. [15] W.Liu, M.FlytzaniStephanopoulos. Total oxidation of carbon monoxide and methane over transition metalfluorite oxide composite catalysts II. Catalyst characterizationandreactionkinetics,J.Catal.153(1995)317332. [16] Z.Liu, R.Zhou, X.Zheng. Comparative study of different methods of preparing CuOCeO2 catalysts for preferential oxidation of CO in excess hydrogen, J. Mol. Catal.AChem.267(2006)137142. [17] D.H.Kim, J.A.Cha. A CuOCeO2 mixedoxide catalyst for CO cleanup by selective oxidationinhydrogenrichmixtures,Cat.Lett.86(2003)107112. [18] D.Gamarra, A.Horns, Zs.Koppany, Z.Schay, G.Munuera, J.Soria, A.Martnez Arias. Catalytic processes during preferential oxidation of CO in H2rich streams overcatalystsbasedoncopperceria,J.PowerSources169(2007)110116. [19] G.Avgouropoulos, T.Ioannides, C.Papadopoulou, J.Batista, S.Hocevar, H.K.Matralis. A comparative study of Pt/Al2O3, Au/Fe2O3 and CuOCeO2 catalysts for the selective oxidation of carbon monoxide in excess hydrogen, Catal.Today75(2002)157167. [20] M.Manzoli,R.DiMonte,F.Boccuzzi,S.Coluccia,J.Kaspar.COoxidationoverCuOx CeO2ZrO2 catalysts: Transient behaviour and role of copper clusters in contact withceria,Appl.Catal.B:Environ.61(2005)192205. [21] T.ValdsSols, G.Marbn, A.B.Fuertes. Nanosized catalysts for the production of hydrogenbymethanolsteamreforming,Catal.Today116(2006)354360. [22] M.Kramer, T.Schmidt, K.Stowe, W.F.Maier. Structural and catalytic aspects of solgel derived copper manganese oxides as lowtemperature CO oxidation catalyst,Appl.Catal.A:Gen.302(2006)257263. [23] S.A.Makhlouf. Magnetic properties of Co3O4 nanoparticles, J. Magn. Magn. Mater.246(2002)184190.

189 [24] H.Yamaura, K.Moriya, N.Miura,N.Yamazoe.Mechanismof sensitivitypromotion in CO sensor using indium oxide and cobalt oxide, Sensor Actuat. BChem. 65 (2000)3941. [25] J.Jansson, A.Palmqvist, E.Fridell, M.Skoglundh, L.sterlund, P.Thormhlen, V.Langer. On the catalytic activity of Co3O4 in Lowtemperature CO oxidation, J. Catal.211(2002)387397. [26] F.Grillo,M.M.Natile,A.Glisenti.Lowtemperatureoxidationofcarbonmonoxide: the influence of water and oxygen on the reactivity of a Co3O4 powder surface, Appl.Catal.B:Environ.48(2004)267274. [27] E.Y.Ko, E.D.Park, K.W.Seo, H.C.Lee, D.Lee, S.Kim. A comparative study of catalysts for the preferential CO oxidation in excess hydrogen, Catal. Today 116 (2006)377383. [28] K.Omata, Y.Kobayashi, M.Yamada. Artificial neural networkaided design of Co/SrCO3 catalyst for preferential oxidation of CO in excess hydrogen, Catal. Today117(2006)311315. [29] Z.Zhao, M.M.Yung, U.S.Ozkan. Effect of support on the preferential oxidation of COovercobaltcatalysts,Cat.Comm.9(2008)14651471. [30] R.J.Wu, J.G.Wu, T.K.Tsai, C.T.Yeh. Use of cobalt oxide CoOOH in a carbon monoxide sensor operating at low temperatures, Sensor Actuat. BChem. 120 (2006)104109. [31] X.Li,T.Xia,H.Dong,Y.Wei.StudyonthereductionbehaviorofCoOOHduringthe storage of nickel/metalhydride battery, Mater. Chem. Phys. 100 (2006) 486 489. [32] G.A.Somorjai, Y.G.Borodko. Research in nanosciences Great opportunity for catalysisscience,Catal.Lett76(2001)15. [33] J.Livage, M.Henry, C.Sanchez. SolGel Chemistry of TransitionMetal Oxides, Prog.SolidStateChem.18(1988)259341. [34] J.Agrell, G.Germani, S.G.Jaras, M.Boutonnet. Production of hydrogen by partial oxidation of methanol over ZnOsupported palladium catalysts prepared by microemulsiontechnique,Appl.Catal.A:Gen.242(2003)233245. [35] S.Zhu, H.Zhou, M.Hibino, I.Honma, M.Ichihara. Synthesis of MnO2 Nanoparticles Confined in Ordered Mesoporous Carbon Using a Sonochemical Method, Adv. Funct.Mater.15(2005)381386.

190

Bibliografa

[36] G.Marbn, A.B.Fuertes, T.ValdsSols. Templated synthesis of high surface area inorganic oxides by silica aquagelconfined coprecipitation, Microp. Mesop. Mat.112(2008)291298. [37] A.B.Fuertes. Lowcost synthetic route to mesoporous carbons with narrow pore size distributions and tunable porosity through silica xerogel templates, Chem. Mater.16(2004)449455. [38] M.Kurmoo, J.L.Rehspringer, A.Hutlova, C.D'Orlans, S.Vilminot, C.Estourns, D.Niznansky.FormationofnanoparticlesofFe2O3fromyttriumirongarnetina silicamatrix:Anunusuallyhardmagnetwithamorinliketransitionbelow150K, Chem.Mater.17(2005)11061114. [39] E.L.Pires, U.Arnold, U.Schuchardt. Amorphous silicates containing cerium: selective catalysts for the oxidation of cyclohexane, J. Mol. Catal. AChem. 169 (2001)157161. [40] G.Marbn, T.ValdsSols, A.B.Fuertes. High surface area CuMn2O4 prepared by silica aquagelconfined coprecipitation. Characterization and testing in steam reformingofmethanol.,Cat.Lett.118(2007)814. [41] C.Galletti, S.Specchia, G.Saracco, V.Specchia. Development of a lab scale catalytic metal platechannels reactor for CO preferential oxidation, Chem. Eng. J.154(2009)246250. [42] P.Chin, X.Sun, G.W.Roberts, J.J.Spivey. Preferential oxidation of carbon monoxide with ironpromoted platinum catalysts supported on metal foams, Appl.Catal.A:Gen.302(2006)2231. [43] C.Gu,S.Lu,J.Miao,Y.Liu,Y.Wang.MesomacroporousmonolithicCuOCeO2// Al2O3 catalysts for CO preferential oxidation in hydrogenrich gas: Effect of loadingmethods,Int.J.HydrogenEnergy35(2010)61136122. [44] Y.Li, B.Tan, Y.Wu. Freestanding Mesoporous QuasiSingleCrystalline Co3O4 NanowireArrays,J.Am.Chem.Soc.128(2006)1425814259. [45] Y.Li, B.Tan, Y.Wu. Mesoporous Co3O4 nanowire arrays for lithium ion batteries withhighcapacityandratecapability,NanoLett.8(2008)265270. [46] G.Marbn, A.Lpez, I.Lpez, T.ValdsSols. A highly active, selective and stable copper/cobaltstructurednanocatalystformethanoldecomposition,Appl.Catal. B:Environ.99(2010)257264.

191 [47] P.Scherrer.Bestimmungder Gre undderinneren Struktur von Kolloidteilchen mittelsRntgenstrahlen,GttingerNachrichten2(1918)98100. [48] M.Kruk, M.Jaroniec, A.Sayari. Application of large pore MCM41 molecular sieves to improve pore size analysis using nitrogen adsorption measurements, Langmuir13(1997)62676273. [49] G.Sedmak, S.Hocevar, J.Levec. Kinetics of selective CO oxidation in excess of H2 overthenanostructuredCu0.1Ce0.9O2ycatalyst,J.Catal.213(2003)135150.

InstitutoNacionaldel d Carbn 2010