Hydrogen production from steam reforming ofmethanolover CuO/ZnO/Al2O3 catalysts:Catalytic performance and kinetic modeling☆

Yu Wan,Zhiming Zhou*,Zhenmin Cheng

State Key Laboratory ofChemicalEngineering,East China University of Science and Technology,Shanghai200237,China

1.Introduction

Methanolas an on-board hydrogen source has received much attention in recent years on polymer electrolyte membrane fuel cells(PEMFC)in mobile applications,owing to its convenience in low cost,safe handling and easy synthesis from various renewable sources[1–3].Methanolis liquid at ambient conditions with a high H:C ratio of4:1 and no C–C bond thathas to be broken,lowering the risk of the high temperature needed to generate hydrogen and coke formation during the reaction[4,5].

It is well known that hydrogen can be obtained from methanol through three different ways:methanoldecomposition(Eq.(1))[6,7],partial oxidation of methanol(Eq.(2))[8,9]and steam reforming of methanol(Eq.(3))[10,11].

Among these approaches,steam reforming ofmethanol(SRM)is the most attractive to generate H2for PEMFC because it can produce high purity of hydrogen under a relatively low reaction temperature of 473–573 K[10–12].Furthermore,CO,which is toxic to PEMFC,is produced by the reverse water-gas shift reaction(r-WGS)(Eq.(4))rather than the direct SRM process[13,14].

The most widely used catalysts in the SRM process are Cu-based catalysts,especially Cu/ZnOand Cu/ZnO/Al2O3due to their high activity and selectivity[15–17].In situ X-ray diffraction and absorption spectroscopy characterization of CuO/ZnO carried out by Günter et al.[15]revealed that the bulk structure and the active surface were modi fied by the intimate Cu–ZnOinterface,and the interaction between Cu and ZnOhad a pronounced in fluence on the activity ofcatalyst.With the aid of the diffuse re flection Fourier transform infrared spectroscopy,Chang etal.[16]elucidate that Cu played an importantrole in adsorbing methanol and converting it into methoxy,while ZnO promoted dehydrogenation ofmethoxy to formate,which was in turn converted into carbonate and finally decomposed to H2and CO2.Thus,the coexistence of Cu and ZnO facilitated the SRMreaction.Tong et al.[17]proposed that the reorganization ofatoms occurred during the reaction at the interface between Cu and the oxide support,and the Cu–ZnO interfaces seemed to be the active site.In addition,in order to keep the chemical and thermal stability of Cu/ZnO catalysts in industrial application,Al2O3is often added to inhibit the thermalsintering[14,18,19].However,the amount of Al2O3should be appropriate due to the negative effect of Al2O3on the catalyst activity[16,20].

Apart from CuO/ZnO and CuO/ZnO/Al2O3catalysts,other dopants such as ZrO2[16,20,21]and CeO2[22–26]are introduced into Cubased catalysts to improve their activity performance.Agrell et al.[21]reported that binary CuO/ZnO had poor stability in temperatureprogrammed reduction(TPR)and oxidation(TPO)redox cycles,but the catalysts containing ZrO2were resistantto redox cycles and showed high stability.Huang etal.[20]found thatintroduction ofZrO2improved the reducibility and stability of CuO/ZnO/Al2O3.CeO2has attracted much attention for its high oxygen storage capacity that is believed to improve the thermalstability of the parent oxides[22].Zhang et al.[23]claimed that addition of CeO2increased oxygen storage capacity of Cu-based catalysts and consequently decreased the CO content in the productgas through the r-WGSreaction.Pateland Pant[24]showed that an optimum of 10 wt%of doping of cerium into CuO/ZnO/Al2O3greatly improved the catalyst performance in terms of conversion of CH3OH,selectivity to CO and production rate of H2,although the weakening effectof CeO2on the SRMwas reported by other researches[20,25].

Despite a lot ofwork devoted to the Cu-based catalysts in the SRM,the relationship between the speci fic surface area of copper and the activity of catalyst remains the subject of debate.Some investigators[23,27,28]found a good correlation between the catalytic activity and the Cu surface area,i.e.,the larger the Cu surface area,the higher the catalytic activity,while others[25,29]did not observe such a correlation.In addition,in many studies the amount of CuO[14,25,30–32]or Al2O3[25,31]is out of the practicalrange(CuO:30–60 wt%,Al2O3:10–20 wt%)[20,23,33,34]for industrialapplication,which may obscure some important information concerning the structureperformance relationship of the Cu-based catalysts in the SRM.

In this context,a series of CuO/ZnO/Al2O3catalysts are firstprepared by coprecipitation,with a CuO content of 30–60 wt%and a Al2O3content of 10 wt%.The structuralproperties and catalytic performance of these catalysts are evaluated and correlated.Next,a certain amount of ZrO2or CeO2is introduced into a screened CuO/ZnO/Al2O3and the effects of ZrO2and CeO2dopants on the structure and performance of catalyst are explored.Finally,the intrinsic kinetics of SRM is studied over the best catalyst developed.The results obtained in this work are helpful for the development of high-performance Cu-based catalysts for the SRMprocess.

2.Experimental

2.1.Materials

Cu(NO3)2·3H2O(≥99%),Zn(NO3)2·6H2O(≥99%),Al(NO3)3·9H2O(≥99%),Zr(NO3)4·5H2O(≥99%)and Ce(NO3)3·6H2O(≥99%)were commercially available from Sinopharm Chemical Reagent and employed as Cu,Zn,Al,Zr and Ce sources,respectively.Na2CO3(≥99.8%)were purchased from Shanghai Lingfeng Chemical Reagent CO.and used as the precipitant.

2.2.Catalyst preparation

The Cu-based catalysts were prepared by coprecipitation method in an aqueous solution.In a typicalsynthesis,a 0.5 mol·L-1solution of a mixture ofmetalnitrate Cu(NO3)2,Zn(NO3)2and Al(NO3)3(Ce(NO3)3or Zr(NO3)4when needed)and a 0.5 mol·L-1solution of Na2CO3were first prepared,respectively.Next,the two solutions were added slowly and simultaneously into 50 ml of deionized water(kept at 343 K in water bath)with vigorous stirring,and the pH of the solution was maintained at 7.0–7.5 by carefully controlling the addition speeds of the above two solutions.Upon completion of the addition,the solution was aged at343 K for 2 h with stirring,after which the precipitate obtained was filtered and washed with deionized water severaltimes to remove the residual Na+.Then,the precipitate was dried at 383 K for 12 h,followed by calcination at623 Kfor 4 h under air flow.The catalyst powder obtained was finally pelletized,crushed and sieved to particles with average size dp=0.34 mm for test.The metal composition of the catalyst was adjusted by varying the initialnitrate concentration.For convenience in the following discussion,the as-prepared catalysts were denoted as CuaZnbAlc,CuaZnbZrcAldand CuaZnbCecAld,where a,b,c and d are the mass percentages of metal oxides.For example,Cu30Zn60Al10represents the catalyst with 30 wt%CuO,60 wt%ZnO and 10 wt%Al2O3.A commercial CuO/ZnO/Al2O3catalyst(Stock no 45468,Alfa Aesar)with about 50 wt%CuO,30 wt%ZnO and 20 wt%Al2O3was used as a reference.

2.3.Catalyst characterization

2.3.1.N2adsorption-desorption

The Brunauer–Emmet–Teller(BET)surface area,pore volume and pore diameter of the catalyst sample were acquired from N2physisorption data that were collected at 77 K on the Micromeritics ASAP 2020.Before the measurements the sample was degassed at 423 K and 133 Pa for 6 h.

2.3.2.Temperature-programmed reduction(TPR)

TPR experiments were performed on the Micromeritics AutoChem 2920.The sample(about 50 mg)was first purged with Ar at 423 K in a flow rate of 30 ml·min-1for 2 h to remove physically adsorbed water,and then the temperature ramped from room temperature to 773 K with a heating rate of 5 K·min-1in a 10%H2/Ar flow(30 ml·min-1).The consumption of H2was monitored by TCD.

2.3.3.X-ray diffraction(XRD)

XRD patterns were obtained on a Rigaku diffractometer(D/MAX 2550 VB/PC)operating at 40 kV and 100 mA with CuKαradiation(λ=0.154056 nm).Measurements were performed by step scanning 2θfrom 10°to 80°with a step size of0.02°.

2.3.4.N2O titration

The speci fic surface area and dispersion of Cu of the catalyst were determined by N2O titration using the Micromeritics AutoChem 2920.First,the copper species was reduced into Cu0by the same procedure as that applied for the TPR analysis,after which the sample was cooled to room temperature in Ar.Next,the Ar flow was replaced by a mixture of10%N2O/Ar at323 K for 1 h,during which Cu0on the catalyst surface was transformed into Cu2O(Eq.(5))[35,36].Finally,the reduction step was repeated to reduce Cu2O into Cu0.From the amount ofconsumed H2monitored by TCD in the two reduction steps,the Cu surface area was estimated by assuming the surface density of 1.46×1019copper atoms per square meter,and the Cu dispersion was calculated as the ratio of Cu atoms on the catalyst surface to the total number of Cu atoms in the catalyst.

2.3.5.High-resolution transmission electron microscopy(HRTEM)

The micro-morphology of the catalysts was analyzed using a JEOL JEM-2100 electron microscope operating at200 kV.The ground sample wasultrasonically dispersed in ethanolfor15 min.Adrop of the suspension was then transferred onto a carbon-coated coppergrid,followed by drying at room temperature.

2.4.Catalyst test

Fig.1.Schematic diagram of the experimentalsetup.1—hydrogen,2—argon,3—pressure regulator,4—ball valve,5—mass flow controller,6—check valve,7—liquid feedstock,8—HPLC pump,9—evaporator,10—heater and thermal insulator,11—reactor,12—condenser,13—low temperature circulating bath,14—gas–liquid separator,15—gas chromatograph,16—soap bubble flowmeter.

The SRMreaction was conducted in a fixed-bed quartz tube reactor(8 mm i.d.)that was placed in a three-section furnace with separate temperature controlof each zone,as shown in Fig.1.0.2 g of catalyst particles(dp=0.34 mm)was diluted with inert quartz sand(dp=0.34 mm)with a mass ratio of 1:5,and the mixture was loaded on quartz woolpositioned atthe center of the reactor.Prior to the reaction test,the catalystwas reduced in a stream of10%H2/Ar with a total flow rate of50 ml·min-1according to a temperature program:from room temperature to 573 K at 2 K·min-1and kept at 573 K for 2 h.

After reduction the reactor was cooled to the desired temperature in pure Ar(50 ml·min-1).When the temperature was stabilized,Ar was switched offand the liquid reactant,a mixture ofwater and methanol(H2O/CH3OH molar ratio of1.2),was first injected into the evaporator using a HPLC pump and then to the reactor for the SRMreaction.The reaction product was cooled in an ice bath to condense the unreacted steamand methanol.The dry gas was analyzed by a gas chromatograph(GC,HP 6890)equipped with a thermalconductivity detector(TCD)and a TDX packed column(2 m)was used for separation.The carrier gas was Ar and the temperatures at the injector,the oven and the detector were 423,393 and 453 K,respectively.The condensate was analyzed by another GC(Agilent 7890 A)equipped with a TCD and a CP-Wax column(50 m×0.53 mm).The carrier gas was H2and the temperatures at the injector,the oven and the detector were 393,363 and 393 K,respectively.

The conversion of methanoland the selectivity to CO were de fined as follows:

where Fi,inand Fi,outrepresentthe inlet and outlet molar flow rates of component i,respectively.

2.5.Kinetic study

A screened catalyst with good activity,selectivity and stability in the SRMprocess was employed for the kinetic study,and the experiments were performed using the same setup as in Fig.1.Preliminary tests were carried outto eliminate the externaland internaldiffusion effects,the former being achieved by increasing the space velocity(or decreasing the space time)of reactants and the later being obtained by reducing the catalyst particle size[37,38].Before collection of the kinetic data the catalyst had been on stream over 50 h in order to eliminate the in fluence of catalyst deactivation.Moreover,the system was stabilized for 2 h under each target reaction condition before collection ofsamples for analysis.The kinetic experiments were conducted by varying the flow rate of reactants and the H2O/CH3OH molar ratio at 0.1 MPa over a temperature range of 503–543 K.

3.Kinetic Model

The reaction schemes for SRMconsistofthree reversible overallreactions:(1)methanol–steam reforming(CH3OH+H2O ? CO2+3H2),(2)water–gas shift reaction(CO+H2O?CO2+H2)and(3)methanol decomposition(CH3OH?CO+2H2).A classic and comprehensive kinetic modeldeveloped by Peppley et al.[39]is used to describe the SRM over the Cu-based catalyst prepared in this work.This modelis based on several important assumptions:(a)hydrogen and the oxygen-containing species adsorb on different active sites;(b)the active sites for methanol decomposition are different from those for methanol-steam reforming and the WGS reaction;and(c)for both methanol-steam reforming and methanol decomposition the ratedetermining step(RDS)is the dehydrogenation of the adsorbed methoxy,while for the WGS reaction the formation ofan intermediate formate species is the RDS.

Corresponding to the three reversible reactions,the Langmuir–Hinshelwood rate expressions are derived as follows[39]:

Methanol–steam reforming:

Water–gas shiftreaction:

Methanoldecomposition:

where kR(KR),kW(KW)and kD(KD)are the rate constants(equilibrium constants)of methanol–steam reforming,WGS reaction and methanol decomposition,respectively;biis the adsorption constant of species i;and piis the partial pressure of component i.The temperature dependence of each constant can be expressed as the Arrhenius equation

where,,andare the pre-exponentialfactors;ER,EWand EDare the activation energies of different types of reactions;andΔHiis the adsorption heat of species i.These parameters are unknown and can be estimated from the kinetic data.The equilibriumconstantofeach reaction is calculated by[40]

For a fixed-bed reactor,the continuity equations for CO2and CO are given by

where FCO2and FCOare the molar flow rates of CO2and CO,respectively,and W is the variable of the catalyst weight.The initialconditions for Eqs.(18)and(19)are FCO2=0 and FCO=0 for W=0.According to the stoichiometry of the three reactions,the partialpressure ofdifferent component can be expressed as the function of FCO2and FCOas follows

where N represents the H2O/CH3OH molar ratio of the liquid feedstock and pt.is the total pressure,which is maintained at 0.1 MPa in this study.The ordinary Eqs.(18)and(19)are solved using the fourthorder Runge–Kutta method,and the kinetic and adsorption parameters involved in Eqs.(11)–(14)are estimated with the kinetic data by the Levenberg–Marquardt algorithm,which minimizes the sum ofsquares of relative residuals between the experimentally measured FCO2,outand FCO,outand the calculated counterparts

where M is the number of experimental runs used for parameter estimation,M=45.There are twenty kinetic and adsorption parameters to be estimated,but seven of them(adsorption heats of seven surface species)can be adopted from those suggested by Peppley et al.[39],since these values have been proven reasonable over Cu/ZnO/Al2O3 catalysts[39,41].They areΔH CH3O(1),ΔH HCOO(1),ΔH OH(1),ΔH H(1a),ΔHCH3O(2),ΔHOH(2),and ΔHH(2a),which are equal to-20,100,-20,-50,-20,-20 and-50 kJ·mol-1,respectively.

4.Results and Discussion

4.1.Cu/ZnO/Al2O3 catalysts

Table 1 lists the texturalproperty,the Cu dispersion and the Cu surface area of different Cu-based catalysts.Except for Cu60Zn30Al10,the BET surface area and pore volume of CuO/ZnO/Al2O3catalysts decrease with increasing the CuO/ZnO mass ratio,probably because the porous structure of the support is occupied by metal precursors[31].This trend is in accordance with that reported by Shishido et al.[27,42]who found that the BET surface area of CuO/ZnO/Al2O3decreased,in general,as the CuO content increased from about 10.2 wt%to 82.8 wt%(the Al2O3content was fixed at 6.6 wt%),but with the exception of Cu61.7Zn31.6Al6.6(or Cu:Zn:Al(molar ratio)=60:30:10)that had a higher surface area compared to Cu30.6Zn62.8Al6.6,Cu41.0Zn52.4Al6.6and Cu51.3Zn42.1Al6.6.On the contrary,Chang et al.[16]presented that the BET surface area of CuO/ZnO/Al2O3increased with increasing the CuO content from 10 wt%to 40 wt%.The difference probably results from differentmethods for preparing catalysts and varying preparation conditions.The Cu dispersion ofcatalyst varies in inverse proportion to the CuOcontentor CuO/ZnOratio,butthe Cu surface area increases generally with the CuO/ZnO ratio,which is understandable since the contact interface between CuO and ZnO increases with the CuO/ZnO ratio.However,too much CuO may lead to agglomeration of CuO particles and in turn decrease the active Cu surface area.This explains why Cu50Zn40Al10possesses the highest Cu surface area of 89.9 m2·g-1among all CuO/ZnO/Al2O3catalysts.Similar results on the variation of the dispersion and surface area of Cu with the amountofCuO in the catalyst were also observed by Huang et al.[20].In addition,both dispersion and speci fic surface area of Cu of the CuO/ZnO/Al2O3catalysts prepared in this work are larger than those of the commercialcatalyst.

Fig.2 shows the XRD patterns of different catalysts that were calcined at 623 K in air but without reduction.The diffraction peaks assigned to CuO(JCPDS 45-0937)and ZnO(JCPDS 36-1451)are clearly observed,butno peaks belonging to Al2O3are detected,likely due to the high dispersion or the amorphous phase of Al2O3[43,44].With an increase in the CuO content,the diffraction peak strength of CuO(e.g.,2θ=38.8°)increases accordingly.

Table 1 Physicochemicalproperties ofdifferent catalysts

The H2-TPR experiments are applied to analyze the reducibility of the catalyst.As shown in Fig.3,there are two reduction peaks in the H2-TPR pro files ofCuO/ZnO/Al2O3catalysts.The firstpeak located ata lowertemperature can be ascribed to the reduction ofhighly dispersed and smaller CuO crystallites,while the second peak ata higher temperature indicates larger CuOand/or the presence ofstrong copper-supportinteractions[14,21].According to the temperature of the main reduction peak(the first peak),the reducibility of CuO/ZnO/Al2O3samples follows the order Cu30Zn60Al10＜Cu40Zn50Al10＜Cu50Zn40Al10≈Cu60Zn30Al10,which mirrors the trend in the Cu surface area.The catalystwith higherreducibility or with lower reduction temperature normally displays high activity in the SRM[14,33],and therefore Cu50Zn40Al10and Cu60Zn30Al10are expected to be superior to Cu30Zn60Al10and Cu40Zn50Al10in terms of the catalytic activity.

Fig.2.XRD patterns ofdifferent catalysts(without reduction).

Fig.3.H2-TPR pro files ofdifferent catalysts.

Fig.4 shows conversion ofCH3OHand selectivity to CO of CuO/ZnO/Al2O3atdifferentreaction temperatures.As a consequence of the endothermic character of the SRM process,the conversion of CH3OH increases with temperature.The selectivity to CO also increases with temperature,indicating that CO is formed through the endothermic r-WGS reaction.Among these catalysts Cu50Zn40Al10and Cu60Zn30Al10are really more active than Cu30Zn60Al10and Cu40Zn50Al10,in accord with the expectation from TPR analysis,but Cu50Zn40Al10with the largestsurface area of Cu exhibits the highestactivity in the SRM,implying that the activity depends not only on the reducibility,but more importantly also on the Cu surface area.Some researchers have reported that the catalyst with higher Cu surface area has better activity in the SRM[23,27,28].The catalyst activity appears to have no correlation with its surface area since,although Cu50Zn40Al10has the lowestsurface area,its activity is the highest.In addition,Cu30Zn60Al10and Cu60Zn30Al10have very similarsurface areas,buttheir activities are different.In the following section,the in fluence of ZrO2and CeO2dopants on the SRMwillbe analyzed by comparison with Cu50Zn40Al10.

Fig.4.Comparison of CuO/ZnO/Al2O3 catalysts with various CuO/ZnO mass ratios in the SRM:(a)Conversion of methanoland(b)Selectivity to CO.P t=0.1 MPa,H2O/CH3OH(molar ratio)=1.2,W cat=0.2 g,d P=0.34 mm.

4.2.Effectof ZrO2 and CeO2 dopants

Cu50Zn30Zr10Al10and Cu50Zn30Ce10Al10are prepared and employed to study the effectofZrO2and CeO2dopants on the SRM.First,the structure of CuO/ZnO/Al2O3changes after doping with ZrO2and CeO2.Compared to Cu50Zn40Al10both BET surface area and Cu dispersion of Cu50Zn30Zr10Al10and Cu50Zn30Ce10Al10increase(Table 1),indicating that ZrO2and CeO2dopants can improve the catalyst dispersion[16,20,45].In addition,the intensity of the diffraction peaks of CuO and ZnOfor Cu50Zn30Zr10Al10and Cu50Zn30Ce10Al10is weakened(Fig.2),especially for Cu50Zn30Ce10Al10,and the particle size of CuO(black dots in the HRTEMimages shown in Fig.5)after addition of ZrO2(Fig.5b)and CeO2(Fig.5c)becomes smaller,further demonstrating the improved dispersion ofZrO2-and CeO2-doped catalysts.However,both Cu surface area(Table 1)and reducibility(Fig.3)ofcatalyst are ranked as follows:Cu50Zn30Zr10Al10＞Cu50Zn40Al10＞Cu50Zn30Ce10Al10,implying that Cu50Zn30Zr10Al10has the highestactivity.

Fig.5.HRTEMimages of(a)Cu50Zn40Al10,(b)Cu50Zn30Zr10Al10 and(c)Cu50Zn30Ce10Al10 catalysts after calcination treatment but without reduction.

Second,the activity and selectivity of CuO/ZnO/Al2O3catalysts change after doping with ZrO2and CeO2.As presented in Fig.6,at various reaction temperature the catalytic activity decreases in the order of Cu50Zn30Zr10Al10＞Cu50Zn40Al10＞Cu50Zn30Ce10Al10.This is also the order of the Cu surface area ofcatalyst.Indeed,an approximate linear relationship exists between the conversion of CH3OH and the Cu surface area for allcatalysts used here,including the commercialone(Fig.7),indicating that the catalytic activity of the Cu-based catalyst is close associated with its Cu surface area.A strange phenomenon is that although Cu50Zn30Ce10Al10possesses a higher Cu dispersion than that of Cu50Zn40Al10,as con firmed by XRD and HRTEManalysis,the Cu surface area of the former is lower than thatof the latter.This is probably because addition of CeO2increases the Al2O3contenton the catalyst surface[20],which in turn covers a certain amount of CuO and thus lowers the Cu surface area.This also makes the reduction of CuO of Cu50Zn30Ce10Al10more dif ficult than that of Cu50Zn40Al10(Fig.3).As far as the selectivity to CO is concerned,Cu50Zn30Ce10Al10displays a lower selectivity compared to Cu50Zn30Zr10Al10and Cu50Zn40Al10,which can be attributed to the oxygen storage capacity of CeO2that can promote the WGS reaction and suppress the formation of CO[23,46,47].

Fig.6.Comparison of Cu50Zn40Al10,Cu50Zn30Zr10Al10 and Cu50Zn30Ce10Al10 in the SRM.P t=0.1 MPa,H2O/CH3OH(molar ratio)=1.2,W cat=0.2 g,d P=0.34 mm.

Fig.7.Variation ofconversion of CH3OH with Cu surface area of catalyst.P t=0.1 MPa,H2O/CH3OH(molar ratio)=1.2,W cat=0.2 g,d P=0.34 mm.

Fig.8 shows the conversion ofCH3OHand selectivity to COwith time on stream at543 K(the highestreaction temperature used in this study)for Cu50Zn40Al10,Cu50Zn30Zr10Al10,Cu50Zn30Ce10Al10and the commercial catalyst.During the initial 10 h of operation the catalyst activity declines quickly,which is mostly due to thermal sintering[48],and after about 20 h allcatalysts exhibit stable activity.Among allthe four catalysts the commercialone shows the lowest SRMactivity,which is probably caused by its higher Al2O3content(20 wt%)because an increased amount of Al2O3was found to decrease the activity of CuO/ZnO/Al2O3catalysts[20].Buton the other hand,the preparation method may also contribute to the difference.In a preliminary experiment(not shown here),a Cu50Zn30Al20catalyst with the same composition as the commercialone is prepared by the coprecipitation method as described above,and itshows a higher SRMactivity than the commercialcatalyst.In addition,the catalystactivity at473–543 K is found to decrease in the order of Cu50Zn40Al10＞Cu50Zn30Al20＞Cu50Zn30Ce20(commercial).Unfortunately,the preparation method for the commercialcatalyst is unclear.Although Cu50Zn30Zr10Al10has the highestactivity,its selectivity to COis also the highest.On the contrary,Cu50Zn30Ce10Al10shows a moderate activity butits selectivity to COis the lowest.Considering that CO acts as a poison for PEMFC anodes,Cu50Zn30Ce10Al10is superior to Cu50Zn30Zr10Al10as a SRMcatalyst,and hence Cu50Zn30Ce10Al10is chosen for the kinetic study.

Fig.8.Variation of conversion of CH3OH and selectivity to CO with time on stream for different catalysts.T=543 K,P t=0.1 MPa,H2O/CH3OH(molar ratio)=1.2,W cat=0.2 g,d P=0.34 mm.

4.3.Reaction kinetics

Five sizes ofcatalystpowder,i.e.,dP=0.64 mm,dP=0.34 mm,dP=0.22 mm,dP=0.17 mm and dP=0.13 mm respectively,are used to evaluate the effect of internal diffusion.As presented in Fig.9(a),when dP≤0.22 mm the conversion of CH3OH remains almost unchanged,indicating that the internaldiffusion effect is insigni ficant.Fig.9(b)shows that when Wcat/Fin≤ 3.5 kg·s·mol-1no difference in the conversion ofCH3OH is observed under two differentcatalystloadings of 0.1 g and 0.2 g,implying that the external mass transfer resistance can be ignored.Based on these results,45 experimental runs are conducted to collect data from which kinetic and adsorption parameters can be estimated.The experimentalconditions are as follows:T=503–543 K,Pt=0.1 MPa,Wcat/Fin=1.3–3.5 kg·s·mol-1,H2O/CH3OH(molar ratio)=1.0–1.8,Wcat=0.2 g,dP=0.17 mm,under which the internal and external mass transfer effects are negligible.

The effectofspace time on ef fluentconcentrations ofCOand CO2ata fixed H2O/CH3OH molar ratio is shown in Fig.10.An increase in the space time undoubtedly increases the ef fluent CO2concentration as a result of increased conversion of CH3OH,which in turn promotes the formation of CO by the r-WGS.Fig.11 displays the in fluence of H2O/CH3OHmolar ratio ata fixed feed flow rate.The ef fluent CO2concentration decreases with increasing the H2O/CH3OH ratio,which can be ascribed to the decreased CH3OH concentration in the feed when the feed flow rate is keptconstant.Accordingly,COformation is suppressed via the r-WGS and its ef fluent concentration decreases.

Fig.10.Effectofspace time on ef fluentconcentrations of(a)CO2 and(b)CO.P t=0.1 MPa,H2O/CH3OH(molar ratio)=1.2,W cat=0.1 g,d P=0.17 mm.

Fig.9.Effects of(a)particle size and(b)space time on the conversion ofCH3OH.(a)T=543 K,P t=0.1 MPa,F in=0.05 ml·min-1,H2O/CH3OH(molar ratio)=1.2,W cat=0.1 g;(b)T=543 K,P t=0.1 MPa,H2O/CH3OH(molar ratio)=1.2,d P=0.17 mm.

Fig.11.EffectofH2O/CH3OHmolar ratio on ef fluentconcentrations of(a)CO2 and(b)CO.P t=0.1 MPa,F in=1 mg·s-1,W cat=0.1 g,d P=0.17 mm.

Table 2 Estimated parameters for the kinetic modelof SRM

Fig.12.Comparison of model-calculated and measured flow rates of(a)CO and(b)CO2 at the outlet of the reactor.

Table 2 lists the estimated parameters involved in the kinetic model and the corresponding statisticalanalysis.The activation energies for the reactions ofmethanol–steam reforming,water–gas shift and methanoldecomposition are estimated to be 93.1,85.1 and 116.5 kJ·mol-1,respectively.The estimated activation energy for methanol-steam reforming falls wellwithin the range of74–122 kJ·mol-1over the Cubased catalysts such as Cu/ZnO,CuO/ZnO/Al2O3,CuO/MnO/Al2O3and CuO/ZnO/Cr2O3/Al2O3,as summarized by Lee et al.[41],and also approaches 92 kJ·mol-1reported by Agarwal et al.[31],87.7 kJ·mol-1by Pateland Pant[49],106.7 kJ·mol-1by Sá etal.[32]and 86.9 kJ·mol-1by Silva et al.[50].As for WGS and methanol decomposition involved in the SRM,to the best of our knowledge,very few literature have studied the kinetics of the two reactions[31,39].Peppley et al.[39]showed that EWand EDover the Cu40Zn40Al20catalyst were 87.6 and 170 kJ·mol-1,respectively,while Agarwal etal.[31]reported 79.7 kJ·mol-1of EWand 82.5 kJ·mol-1of EDover Cu10Zn5Al85.It is apparent that our estimate of EWis close to those reported in the literature,but the values of EDfor different catalysts vary over a wide range.From the above comparison it appears that the activation energies for methanol-steam reforming and WGS are independentof the catalystused,butthe activation energy for methanol decomposition relies on the type and composition of catalyst.The reason for this phenomenon is unclear at present and more work is needed in this area.

The statisticalsigni ficance ofa parameter estimate is measured by the t-value and the 95%con fidence interval.Ifa parameter has a very small tvalue or its con fidence intervalincludes zero,the parameteris statistically insigni ficant.As shown in Table 2,allparameter estimates are signi ficant atthe 95%con fidence level.Fig.12 compares the model-calculated molar flowrates ofCOand CO2atthe outletof the reactor with the experimental counterparts.Itcan be seen that the calculated values closely match the experimentaldata,and the average relative errors between them,de fined asare calculated to be 2.3%and 1.7%for CO and CO2,respectively.Therefore,the kinetic modeland the estimated parameter values are reliable,and can accurately describe the steam reforming of methanolover the Cu50Zn30Ce10Al10catalyst.In addition,as shown in Figs.10 and 11,the calculated ef fluent concentrations of COand CO2are in excellentagreementwith the experimentaldata.

5.Conclusions

Hydrogen production by steam reforming of methanol(SRM)was carried out over Cu-based catalysts(CuO/ZnO/Al2O3,CuO/ZnO/ZrO2/Al2O3and CuO/ZnO/CeO2/Al2O3),which were synthesized by coprecipitation method and had compositions in the range ofpractical interest.These catalysts were characterized in detailby severaltechniques and their catalytic performances in the SRMwere tested in a fixed-bed reactor.It was found that the catalytic activity of Cu-based catalysts was not related to the surface area and Cu dispersion ofcatalyst,but depended strongly on the catalyst reducibility and the speci fic surface area of Cu.In particular,at different reaction temperatures an approximate linear relationship was found between the catalytic activity and the Cu surface area for all catalysts investigated(including a commercialcatalyst).The effect of ZrO2dopant on the activity and selectivity ofCuO/ZnO/Al2O3was differentfromthatofCeO2dopant.Doping with ZrO2(CuO/ZnO/ZrO2/Al2O3)increased the catalyst activity but simultaneously produced more CO.On the contrary,the CeO2-doped catalyst(CuO/ZnO/CeO2/Al2O3)exhibited a slightly lower activity,but the selectivity to COwas decreased.The intrinsic kinetic study conducted on CuO/ZnO/CeO2/Al2O3showed thatincreasing the space velocity of reactants and the H2O/CH3OHmolar ratio helped to reduce the ef fluent concentration of CO.The kinetic data were welldepicted by the classic Langmuir–Hinshelwood rate expressions developed by Peppley et al.[39],and the average relative errors between the model-derived molar flow rates of CO and CO2in the ef fluent and the experimental counterparts were 2.3%and 1.7%,respectively.In addition,allparameter estimates involved in the rate expressions were statistically signi ficant at the 95%con fidence level.

[1]P.Kruger,Electric power requirement for large-scale production of hydrogen fuel for the world vehicle fleet,Int.J.Hydrog.Energy 26(2001)1137–1147.

[2]R.Schl?gl,The role of chemistry in the energy challenge,ChemSusChem 3(2010)209–222.

[3]J.D.Holladay,J.Hu,D.L.King,Y.Wang,An overview of hydrogen production technologies,Catal.Today 139(2009)244–260.

[4]S.Onsuratoom,T.Puangpetch,T.Chavadej,Comparative investigation of hydrogen production over Ag-,Ni-,and Cu-loaded mesoporous-assembled TiO2–ZrO2mixed oxide nanocrystal photocatalysts,Chem.Eng.J.173(2011)667–675.

[5]G.A.Olah,Towards oil independence through renewable methanol chemistry,Angew.Chem.Int.Ed.52(2013)104–107.

[6]A.W.Jasper,S.J.Klippenstein,L.B.Harding,B.Ruscic,Kinetics of the reaction of methyl radicalwith hydroxylradicaland methanol decomposition,J.Phys.Chem.A 111(2007)3932–3950.

[7]I.V.Yudanov,A.V.Matveev,K.M.Neyman,N.R?sch,How the C–O bond breaks during methanol decomposition on nanocrystallites of palladium catalysts,J.Am.Chem.Soc.130(2008)9342–9352.

[8]J.R.Lattner,M.P.Harold,Autothermal reforming of methanol:experiments and modeling,Catal.Today 120(2007)78–89.

[9]I.Eswaramoorthi,V.Sundaramurthy,A.K.Dalai,Partial oxidation of methanol for hydrogen production over carbon nanotubes supported Cu–Zn catalysts,Catal.Today 313(2006)22–34.

[10]S.Sá,H.Silva,L.Brand?o,J.M.Sousaa,A.Mendes,Catalysts for methanol steam reforming—A review,Appl.Catal.B 99(2010)43–57.

[11]B.Frank,F.C.Jentoft,H.Soerijanto,J.Kr?hnert,R.Schl?gl,R.Schom?cker,Steam reforming ofmethanolover copper-containing catalysts:In fluence ofsupportmaterialon microkinetics,J.Catal.246(2007)177–192.

[12]D.R.Palo,R.A.Dagle,J.D.Holladay,Methanolsteam reforming for hydrogen production,Chem.Rev.107(2007)3992–4021.

[13]K.Takehira,T.Shishido,Preparation of supported metal catalysts starting from hydrotalcites as the precursors and their improvements by adopting“memory effect”,Catal.Surv.Jpn.11(2007)1–30.

[14]J.P.Breen,J.R.H.Ross,Methanolreforming for fuel-cellapplications:Development of zirconia-containing Cu–Zn–Alcatalysts,Catal.Today 51(1999)521–533.

[15]M.M.Günter,T.Ressler,R.E.Jentoft,B.Bems,Redox behavior of copper oxide/zinc oxide catalysts in the steam reforming of methanol studied by in situ X-ray diffraction and absorption spectroscopy,J.Catal.203(2001)133–149.

[16]C.C.Chang,J.W.Wang,C.T.Chang,B.J.Liaw,Y.Z.Chen,EffectofZrO2on steam reforming ofmethanolover CuO/ZnO/ZrO2/Al2O3catalysts,Chem.Eng.J.192(2012)350–356.

[17]W.Y.Tong,K.Cheung,A.West,K.M.Yu,S.C.E.Tsang,Direct methanol steam reforming to hydrogen over CuZnGaOxcatalysts without CO post-treatment:mechanistic considerations,Phys.Chem.Chem.Phys.15(2013)7240–7248.

[18]L.Alejo,R.Lago,M.A.Pe?a,J.L.G.Fierro,Partial oxidation of methanol to produce hydrogen over Cu–Zn-based catalysts,Appl.Catal.A 162(1997)281–297.

[19]M.V.Twigg,M.S.Spencer,Deactivation of copper metal catalysts for methanol decomposition,methanolsteam reforming and methanol synthesis,Top.Catal.22(2003)191–203.

[20]G.Huang,B.J.Liaw,C.J.Jhang,Y.Z.Chen,Steam reforming of methanol over CuO/ZnO/CeO2/ZrO2/Al2O3catalysts,Appl.Catal.A 358(2009)7–12.

[21]J.Agrell,H.Birgersson,M.Boutonnet,I.M.Cabrera,R.M.Navarro,J.L.G.Fierro,Production of hydrogen from methanolover Cu/ZnO catalysts promoted by ZrO2and Al2O3,J.Catal.219(2003)389–403.

[22]G.Nahar,V.Dupont,Hydrogen production from simple alkanes and oxygenated hydrocarbons over ceria–zirconia supported catalysts:Review,Renew.Sust.Energ.Rev.32(2014)777–796.

[23]L.Zhang,L.W.Pan,C.J.Ni,T.J.Sun,S.S.Zhao,S.D.Wang,A.J.Wang,Y.K.Hu,CeO2–ZrO2-promoted CuO/ZnO catalyst for methanolsteam reforming,Int.J.Hydrog.Energy 38(2013)4397–4406.

[24]S.Patel,K.K.Pant,Selective production ofhydrogen via oxidative steam reforming of methanolusing Cu–Zn–Ce–Aloxide catalysts,Chem.Eng.Sci.62(2007)5436–5443.

[25]S.D.Jones,H.E.Hagelin-Weaver,Steam reforming ofmethanolover CeO2-and ZrO2-promoted Cu-ZnO catalysts supported on nanoparticle Al2O3,Appl.Catal.B 90(2009)195–204.

[26]J.Papavasiliou,G.Avgouropoulos,T.Ioannides,Effectofdopants on the performance of CuO–CeO2catalysts in methanolsteam reforming,Appl.Catal.B(2007)226–234.

[27]T.Shishido,Y.Yamamoto,H.Morioka,K.Takaki,K.Takehira,Active Cu/ZnO and Cu/ZnO/Al2O3catalysts prepared by homogeneous precipitation method in steam reforming of methanol,Appl.Catal.A 263(2004)249–253.

[28]S.G.Sanches,J.H.Flores,R.R.de Avillez,M.I.P.da Silva,In fluence of preparation methods and Zr and Y promoters on Cu/ZnO catalysts used for methanol steam reforming,Int.J.Hydrog.Energy 37(2012)6572–6579.

[29]J.E.Park,S.D.Yim,C.S.Kim,E.D.Park,Steam reforming of methanol over Cu/ZnO/ZrO2/Al2O3catalyst,Int.J.Hydrog.Energy 39(2014)11517–11527.

[30]P.Kurr,I.Kasatkin,F.Girgsdies,A.Trunschke,R.Schl?gl,T.Ressler,Microstructural characterization of Cu/ZnO/Al2O3catalysts for methanolsteam reforming—A comparative study,Appl.Catal.A 348(2008)153–164.

[31]V.Agarwal,S.Patel,K.K.Pant,H2production by steam reforming of methanol over Cu/ZnO/Al2O3catalysts:Transient deactivation kinetics modeling,Appl.Catal.A 279(2005)155–164.

[32]S.Sá,J.M.Sousa,A.Mendes,Steam reforming of methanolover a CuO/ZnO/Al2O3catalyst,Chem.Eng.Sci.66(2011)4913–4921.

[33]J.P.Shen,C.S.Song,In fluence ofpreparation method on performance ofCu/Zn-based catalysts for low-temperature steam reforming and oxidative steam reforming of methanol for H2production for fuelcells,Catal.Today 77(2002)89–98.

[34]L.Guczi,A.Erd?helyi,Catalysis for alternative energy generation,Springer Science&Business Media,2012.

[35]J.W.Evans,M.S.Wainwright,A.J.Bridgewater,D.J.Young,On the determination of copper surface area by reaction with nitrous oxide,Appl.Catal.7(1983)75–83.

[36]G.C.Chinchen,C.M.Hay,H.D.Vandervell,K.C.Waugh,The measurement of copper surface areas by reactive frontalchromatography,J.Catal.103(1987)79–86.

[37]Z.M.Zhou,X.Li,T.Y.Zeng,W.B.Hong,Z.M.Cheng,W.K.Yuan,Kinetics of hydrogenolysis of glycerolto propylene glycol over Cu–ZnO–Al2O3catalysts,Chin.J.Chem.Eng.18(2010)384–390.

[38]Y.Qi,Z.M.Cheng,Z.M.Zhou,Steam reforming ofmethane over Nicatalysts prepared from hydrotalcite-type precursors:Catalytic activity and reaction kinetics,Chin.J.Chem.Eng.23(2015)76–85.

[39]B.A.Peppley,J.C.Amphlett,L.M.Kearns,R.F.Mann,Methanolsteam reforming on Cu/ZnO/Al2O3catalysts.Part 2.A comprehensive kinetic model,Appl.Catal.,A 179(1999)31–49.

[40]C.L.Yaws,Chemicalproperties handbook,McGraw-Hill,New York,1999.

[41]J.K.Lee,J.B.Ko,D.H.Kim,Methanolsteam reforming over Cu/ZnO/Al2O3catalyst:Kinetics and effectiveness factor,Appl.Catal.A 278(2004)25–35.

[42]T.Shishido,M.Yamamoto,D.Li,Y.Tian,H.Morioka,M.Honda,T.Sano,K.Takehira,Water–gas shift reaction over Cu/ZnO and Cu/ZnO/Al2O3catalysts prepared by homogeneous precipitation,Appl.Catal.A 303(2006)62–71.

[43]J.Baneshi,M.Haghighi,N.Jodeiri,M.Abdollahifar,H.Ajamein,Urea-nitrate combustion synthesis of ZrO2and CeO2doped CuO/Al2O3nanocatalyst used in steam reforming of biomethanolfor hydrogen production,Ceram.Int.40(2014)14177–14184.

[44]R.Shokrani,M.Haghighi,N.Jodeiri,H.Ajamein,M.Abdollahifar,Fuelcellgrade hydrogen production via methanolsteam reforming over CuO/ZnO/Al2O3nanocatalyst with various oxide ratios synthesized via urea nitrates combustion method,Int.J.Hydrog.Energy 39(2014)13141–13155.

[45]S.D.Jones,L.M.Neal,M.L.Everett,G.B.Ho flund,H.E.Hagelin-Weaver,Characterization of ZrO2-promoted Cu/ZnO/nano-Al2O3methanol steam reforming catalysts,Appl.Surf.Sci.256(2010)7345–7353.

[46]L.Cao,C.J.Ni,Z.S.Yuan,S.D.Wang,Correlation between catalytic selectivity and oxygen storage capacity in autothermalreforming ofmethane over Rh/Ce0.45Zr0.45RE0.1catalysts(RE=La,Pr,Nd,Sm,Eu,Gd,Tb),Chem.Commun.10(2009)1192–1195.

[47]W.J.Shan,Z.C.Feng,Z.L.Li,J.Zhang,W.J.Shen,C.Li,Oxidative steam reforming ofmethanolon Ce0.9Cu0.1Oycatalysts prepared by deposition–precipitation,coprecipitation,and complexation–combustion methods,J.Catal.228(2004)206–217.

[48]S.T.Yong,C.W.Ooi,S.P.Chai,X.S.Wu,Review ofmethanolreforming Cu based catalysts,surface reaction mechanisms,and reaction schemes,Int.J.Hydrog.Energy 38(2013)9541–9552.

[49]S.Patel,K.K.Pant,Experimental study and mechanistic kinetic modeling for selective production of hydrogen via catalytic steam reforming of methanol,Chem.Eng.Sci.62(2007)5425–5435.

[50]H.Silva,C.Mateos-Pedrero,P.Ribeirinha,M.Boaventura,A.Mendes,Lowtemperature methanol steam reforming kinetics over a novel CuZrDyAl catalyst,React.Kinet.Mech.Catal.115(2015)321–339.