Zuojun Wei,Xinmiao Zhu,Xiaoshuang Liu,Haiqin Xu,Xinghua Li,Yaxin Hou,Yingxin Liu,*

1Key Laboratory of Biomass Chemical Engineering of the Ministry of Education,College of Chemical and Biological Engineering,Zhejiang University,Hangzhou 310027,China

2College of Pharmaceutical Science,Zhejiang University of Technology,Hangzhou 310014,China

Keywords:Catalyst Hydrogenation Selectivity Cinnamaldehyde Bimetal Reduced Graphene Oxide

A B S T R A C T In the present work,a series of Pt-based catalysts,alloyed with a second metal,i.e.,Re,Sn,Er,La,and Y,and supported on activated carbon,ordered mesoporous carbon,N-doped mesoporous carbon or reduced graphene oxide(rGO),have been developed for selective hydrogenation of cinnamaldehyde to cinnamylalcohol.Re and rGO were proved tobethemost favorable metaldopant and catalyst support,respectively.Pt50Re50/rGOshowed thehighestcinnamylalcoholselectivity of89%with94%conversionofcinnamaldehydeatthereactionconditions of 120°C,2.0 MPa H2and 4 h.

1.Introduction

Theselectivehydrogenation ofcinnamaldehyde(CAL)to cinnamylalcohol(COL)has been widely investigated because COL is an important intermediate in the manufacture of flavorings,perfumes and pharmaceuticals[1,2].However,it is quite difficult to obtain COL with high selectivity since the C＝C bond in CAL is more active than C＝O to hydrogenate,which leads to an unwanted byproduct of hydrocinnamaldehyde(HCAL)(Fig.1).A number of transition metals supported on various supports as catalysts have been developed for CAL selective hydrogenation to COL,although the main factors affecting the activity and selectivity remain unclear[3-6].

One way to increase the selectivity in CAL hydrogenation is to select a suitable metal with controlled size and shape of the nanoparticles.Gallezotetal.[7,8]testedaseriesofGroupVIIImetalcatalystsinCALhydrogenation,where Pd and Rh generally displayed high activity but ratherpoor selectivity toward COL,while Ir,Pt and Ru exhibited moderate selectivity.To further improve the selectivity of COL,alloying with a secondmetalisaneffectivestrategy,asitcanchangeboththeelectronic and geometric properties of the active metal[9].Zhao et al.[10]obtained 80%selectivity of COL at 96%conversion of CAL by a Sn-Pd/AC bimetallic catalyst,which is much greater than the 12.8%selectivity by a commercial Pd/AC.Vu et al.[11]compared the catalytic performance of different carbon nanotube-based catalysts(Pt,Ru and Pt-Ru)for the selective hydrogenation of COL,and found that the Pt-Ru system achieved higher selectivity(93%)and conversion(80%)after the catalyst was activated by high-temperature treatment.They deduced that the promoted selectivity was ascribed to the activation process which increased the electrical conductivity of carbon nanotube and thus probably improved the electron transfer from the support to the metal.

Catalyst supports also play an important role in the selectivity of COL,as the activity or selectivity may vary with the nature of the interactionsamongsubstrate,metalnanoparticlesandthesupport[10].Alot of materials such as SiO2,Al2O3,TiO2,zeolites and metal organic frameworks have been frequently reported in the hydrogenation of CAL[4,12-15].In our previous work[16-18],Au nanoparticles,Au/SiO2and Au/TiO2prepared by colloid deposition method,Au/TiO2prepared by deposition precipitation method and Ni/SiO2catalyst were used for the liquid-phase selective hydrogenation of CAL and other α,β-unsaturated aldehyde.Results showed that the selectivity to COL is strongly affected by the nature of supports,in which only Au/TiO2catalysts exhibited satisfactory selectivity.Durndell et al.[2]also found that the support's polarity influenced product selectivity,where the more polar SBA-15 used as support for Pt enhanced selectivity to α-methyl-COL from α-methyl-CAL by ca.30%than the less polar fumed silica.Carbon materials,such as activated carbon(AC)[10,19],carbon nanotube[20,21],carbon nanofiber[22],ordered mesoporous carbon(OMC)[23],andgraphene[3,10,15,24],whichhavehighspecificsurface area,high chemical and thermal stability,are all regarded as ideal supports for the dispersion of metal nanoparticles.Compared with the microporosity of AC,other carbon materials are mesoporous and are generally expected to prevent significant diffusion limitations.Moreover,it was reported that the special electric conductivity and the π-π interactions between CAL and the aromatic structure of graphite units in carbon nanotube,nanofiber and graphene also greatly improved the selectivity to COL[3,10,11,20,24,25].

Fig.1.Reaction pathways of hydrogenation of cinnamaldehyde.

Many studies have focused on supported Pt catalysts which possessed excellent activity and moderate selectivity to COL[4,10,15,26,27].In order to explore a Pt-based catalyst with enhanced selectivity to COL,in the present work,rGO was used as the support,and several metals(Re,Sn,Er,La,and Y)which are regarded as effective modifiers for the metal catalysts used in the reduction of C＝O group were introduced to form Pt-M bimetallic catalysts.Activated carbon,OMC and NOMCsupportedPt-basedcatalystswerealsopreparedforcomparison.Meanwhile,the structure-activity relationship of the catalysts was systematically investigated.

2.Experimental

2.1.Materials

H2PtCl6·6H2O and NH4ReO4were purchased from the Shanghai Jiuling Chemical Co.Ltd.Other chemicals were bought from the Sinopharm chemical reagent Co.Ltd.All the chemicals are analytical pure and were used without further purification.

2.2.Catalyst preparation

2.2.1.Preparation of Pt/rGO and Pt-M/rGO catalysts

Freeze-dried graphene oxide(GO)powder was first prepared through a modified Hummers'method as mentioned in our previous publications[28,29].Then the Re/rGO,Pt/rGO and Pt-M/rGO catalysts were obtained via a co-reduction method based on our previous work[28].In a typical procedure,0.1 g of GO powder was dispersed in 100 ml of double-distilled water under sonication.The corresponding metalsaltswithcalculatedamountswereadded,followedbyavigorous stirring for 1 h.Then 100 ml ofNaBH4aqueous solution(7.5 g·L?1)was dropwise added,and the mixture was incubated at 80°C for 4 h.The resulting Pt-M/rGO catalyst was obtained after being filtrated and vacuum dried.

2.2.2.Preparation of Pt nanocatalysts supported on other supports

For comparison,Pt nanocatalysts supported on AC,OMC and NOMC were prepared through an incipient wetness impregnation method according to our previous work[30-32].In a typical procedure,0.5 g of the supports was mixed with 0.04 mmol of H2PtCl6(corresponding to 3 wt%of Pt loading)in ca.1.0 ml of double-distilled water.The mixture was carefully mixed in an ultrasonic bath and kept overnight to form the catalyst precursor.The precursor was then dried in a vacuum oven at 110 °C for 8 h,calcined at 500 °C in nitrogen atmosphere,and reduced at 350°C in hydrogen for 4 h to obtain the final catalyst.

The OMC and NOMC materials were prepared in our laboratory followed by our previous work[33].The prepared OMC and NOMC have BET specific surface areas of 1473 m2·g?1and 666 m2·g?1,and averaged mesoporous pore sizes of 4.7 nm and 4.5 nm,respectively.

2.3.Catalyst Characterization

X-ray powder diffraction(XRD)patterns were recorded with an XRD-6000 diffractometer(Shimadzu Co.,Japan)usinga Cu Kαradiation(λ=0.15406 nm)in a Bragg-Brentano parafocusing optics configuration(40 kV,40 mA).Samples were scanned from 5°to 80°with a scanning rate of 4(°)·min?1and a step size of 0.02°.The crystalline phases were identified by reference to the JCPDS database.

High resolution transmission electron microscopy(HRTEM)images were obtained using a Tecnai G2 F30 S-Twin instrument(FEI Co.,USA)operated at an accelerating voltage of 300 kV.Samples were prepared by dispersing the catalyst powder in ethanol under ultrasound for 15-20 min and then dropping the suspension onto a copper grid coated with a carbon film.Particle size distribution of metal nanoparticles in each sample was determined from the corresponding TEM images by measuring the sizes of more than 200 particles.

Brunner-Emmet-Teller(BET)specific surface area and pore structures were measured by pulsed nitrogen adsorption-desorption method at?196 °C using an ASAP 2010 instrument(Micromeritics Instrument Co.).Prior to N2physisorption,the samples were degassed under vacuum at 250°C overnight.

X-ray photoelectron spectroscopy(XPS)spectrum was obtained using an Escalab Mark II X-ray spectrometer(VG Co.,United Kingdom)equipped with a magnesium anode(Mg Kα=1253.6 eV),50 eV pass energy,a 0.2 eV kinetic energy step,and 0.1 s dwelling time.Energy corrections were performed using a 1 s peak of the pollutant carbon at 284.6 eV.The sample was prepared by pressing the catalyst powder onto the surface with silver sol gel.

2.4.Catalyst test

TheliquidselectivehydrogenationofCALtoCOLwascarriedoutina 35 ml stainless steel autoclave equipped with a magnetic stirrer.In a typical procedure:10 mg of the Pt catalyst,1.5 mmol of CAL and 15 ml of isopropanol were introduced into the reactor.The reactor was then sealed,purged with N2for three times,and heated to desired temperature at a magnetic stirring speed of 1000 r·min?1.As is known to all,the external diffusion resistance should be greatly decreased at such a rotating speed[34].The reaction was then initiated by introducing H2at a designated pressure.After the reaction was completed,the reactor was cooled down to room temperature and the reaction mixture was centrifuged.The solid catalyst was washed with isopropanol for the next cycle;the supernatant was sampled and analyzed with gas chromatography.

Considering that the particle sizes of the supports used are less than 75 μm(the thickness of rGO is less than 20 layers,corresponding to only 10 nm),the internal diffusion can generally be neglected[35,36].Furthermore,Weisz-Prater criterion was employed to verify the negligence of internal diffusion in the reaction(See supplementary information).

2.5.Product analysis

The products were analyzed with anAgilent7890GC equippedwith an HP-5 capillary column(30.0 m × 0.32 mm × 0.25 μm)and a flame ionization detector.The injector temperature was set at 250°C and the sampling volume was 0.4 μl.The detector temperature was 260 °C.The split ratio was 1:10.The column temperature was raised from 100 °C to 260 °C at a heating rate of 5 °C·min?1and then maintained at 260°C for 8 min.CAL and COL were quantified by using n-dodecane asan internal standard.GC-MS analysis wasperformed withan Agilent 6890 GC system coupled to a mass spectrometer equipped with an Agilent 5973 quadrupole mass analyzer.Chromatographic analysis wasconductedwithaninjectortemperatureof280°CandanHP-5capillary column(30.0 m ×0.25 mm × 0.32 μm)with a helium(99.999%)fl ow rate of 2.0 ml·min?1and a 1:10 split ratio.The oven was heated using the following temperature program:the initial temperature of 100 °C increased to 250 °C at a heating rate of 5 °C·min?1and maintained for 10 min.The mass spectrometer was operated in an electron ionization mode at an energy of 70 eV.

3.Results and Discussion

3.1.Catalytic tests

The catalytic performance of different Pt-based catalysts for the selective hydrogenation of CAL is summarized in Table 1.Entries 2-5 in Table 1 reveal the effect of the supports on the performance of Pt catalysts,i.e.,AC,OMC,NOMC and rGO.It can be seen that at the same reaction conditions,Pt/rGO exhibited the highest selectivity of 46.8%to COL at a 23.3%conversion of CAL(Entry 5).To exclude the impact of support in catalytic reaction,a Pt-free rGO sample was used to catalyze this reaction and no obvious product was obtained(Entry 1).The excellent performance of Pt/rGO could be attributed to(1)the special electrical conductivity of the graphitic structure that may promote the electron transfer from support to metal active sites,which favors the back-bonding interactions with π*C＝Oto a large extent more than π*C＝C,and finally increases the selectivity to COL[37,38],and(2)the π-π interactions between CAL and the graphitic plane that would greatly improve the selectivity to COL[3].Therefore,rGO was chosen as the support of Pt catalysts for further investigation below.The OMC and AC supported Pt catalysts exhibited much higher activity.Nevertheless,the corresponding selectivities to COL were less than 10%(Entries 2 and 3).NOMC normally showed better catalytic performance than the inherent OMC in most hydrogenation reactions[33,39].However,in the hydrogenation of CAL,the nitrogen doping greatly inhibited the activity of Pt as the conversion of CAL decreased to only 14%(Entry 4).It should be noted that the conversions of CAL over all the tested catalysts were very low.This is because we focused on the selectivity of COL and conducted the reactions at a relatively low temperature of 80°C.It is known that higher temperature will lead to the deep hydrogenation of COL,giving rise to the downstream products and the decrease in the selectivity to COL.

Table 1 Selective hydrogenation of cinnamaldehyde over different Pt-based catalysts

Entries 6-10 in Table 1 demonstrate the performance of various metal modified Pt/rGO catalyst with the Pt/M molar ratio of 1:1(M=Re,Sn,and rare earth metals Er,La and Y).Rare earth metals are regarded as effective modifiers for heterogeneous metal catalysts to improve catalytic activity and stability[40-42].In general,the addition of rare earths could contribute to higher specific surface areas of oxide supports,and better thermal stability and poison-resistibility of catalysts.Unfortunately,itseemsthatnoneofthemcouldobtainsatisfactory results in our case.In the presence of La and Y,the activity of the Pt catalysts was fully inhibited with the CAL conversions of only 2.7%and 1.6%(Entries 9 and 10),respectively,which were much lower than that catalyzed by the single-metal catalyst Pt/rGO(Entry 5).The addition of Sn and Er achieved similar CAL conversions of around 20%(Entries 7 and 8),which were comparable to that catalyzed by Pt/rGO.However,their selectivities to COL were obviously lower than that of Pt/rGO.The Pt-Sn/SiO2catalyst has ever been reported for the same hydrogenation reaction,where greater enhancement of selectivity to COL from 6%to 80%was achieved with the addition of Sn[43],but the advantages of Sn did not show up in our catalyst,probably due to the variation of the supports and catalyst preparation methods.

Among all the tested metal modified Pt/rGO catalysts,Pt50Re50/rGO exhibited the highest COL selectivity of 83.3%(Entry 6).Changing the Pt:Re molar ratio to 75:25 resulted in an approximately 4%increase of CAL conversion from 20.3%to 24.4%,while the selectivity to COL decreases from 83.3%to 66.8%(Entry 11).The catalyst with the Pt:Re molar ratio of 25:75 presented a decrease in both the conversion of CAL and the selectivity to COL compared to Pt50Re50/rGO(Entry 12).Moreover,since single Re supported on rGO showed very low catalytic activity(Entry 13),we deduce that the synergistic effect of the bimetallic component plays an important role during the selective hydrogenation process of CAL.

Fig.2.Effect of(a)temperature,(b)pressure and(c)time on the selective hydrogenation of cinnamaldehyde over Pt50Re50/rGO catalyst.(Reaction conditions:1.5 mmol of CAL;5 ml of isopropanol;10 mg of catalyst.All the catalysts contain 3 wt%of total metal.)

The hydrogenation reaction of CAL over Pt50Re50/rGO was further optimized with reaction temperature,H2pressure and reaction time.As shown in Fig.2(a),the conversion of CAL is greatly improved with thereaction temperature risingfrom 60 °C to 140 °Cdue to theincrease in the molecule collision frequency.The selectivity to COL initially increased with the temperature up to 120°C and then began to decrease at 140 °C.This is because that at lower temperatures,C＝O bonds are primarily activated,which is beneficial to achieve a higher selectivity to COL;while a higher temperature with a higher energy input tends to activate more C＝C bonds,leading to the deep hydrogenation of COL to hydrocinnamylalcohol(HCOL).As shown in Fig.2(b),the conversion of CAL and the selectivity to COL increased by approximately 20%and 10%,respectively,with the increase in the hydrogen pressure from1MPato2MPa,respectively.ThereasonisthathigherH2pressure provides a larger density of adsorbed H atoms on the metal surface,which will not only improve the conversion of CAL,but also favor the deep hydrogenation of C＝C.As a result,the selectivity to COL decreases from 88.7%to 66.4%while H2pressure rises from 2 MPa to 3 MPa[Fig.2(b)].For reaction time[Fig.2(c)],after a 4-hour reaction,the conversion of CAL reached 94.1%and then changed little until the reaction was extended to 12 h;accordingly,the selectivity of COL peaked at 4 h and decreased with the reaction time extending due to the deep hydrogenation.

Within the experimental range,the highest selectivity of 88.7%to COL with 94.1%conversion of CAL was achieved at 120°C,2.0 MPa H2and4h.ThecorrespondingTOFcalculatedonthebasisoftheconversion of CAL at 2 h andthe molaramount of Pt in Pt50Re50/rGO wasestimated to be 653.5 h?1.Table 2 compares the state-of-the-art catalysts whichhave been employed specifically for the selective hydrogenation of CAL to COL so far.It shows that our catalyst(the last row in Table 2)exhibited obviously higher catalytic activity than most catalysts except for Pt-Co0.2/rGO in Ref.[24]with unusually high catalytic activity(TOF=15480 h?1)and Pt/rGO in Ref.[15]with a higher TOF of 1602 h?1but a much lower selectivity to COL of 73.7%.

Table 2 Comparison of the selective hydrogenation of CAL to COL over different catalysts

The recyclability of Pt50Re50/rGO was determined at the optimal reaction conditions.After the reaction,the catalyst was recovered by centrifugation and washed with isopropanol,and then returned to the reactor for reuse.As shown in Fig.3,the catalyst can be reused at least three times with less than 10%loss of CAL conversion with the selectivity remaining uniform.The activity decrease could be mainly ascribed to the mass loss of the catalyst during the recovery process.This is supported by the further experiment where 20%extra fresh catalyst was added into the reactor in the fourth run,and a similar CAL conversion and COL selectivity were achieved again.

Fig.3.Recyclability of 3 wt%Pt50Re50/rGO catalyst for selective hydrogenation of cinnamaldehyde.(Reaction conditions:1.5 mmol of CAL;5 ml of isopropanol;10 mg of catalyst;reaction temperature of 120°C,pressure of 2.0 MPa,and time of 4.0 h.In the fourth run,2 mg more catalyst was added).

3.2.Catalyst characterization

Fig.4.XRD patterns of Pt/rGO,Pt75Re25/rGO,Pt50Re50/rGO,Pt25Re75/rGO and Re/rGO.

The physico-chemical properties and morphology of Pt-Re/rGO bimetallic catalysts were characterized in order to gain a better understanding of the catalytic mechanism.The XRD patterns of Pt-Re/rGO with different Pt/Re molar ratios are shown in Fig.4.The diffraction peaks at around 2θ=25°(002)correspond to the parallel graphene layers[28,29],and other two peaks at 43°(100)and 77.5°(110)are characteristic of the 2D in-plane symmetry along the graphene sheets,respectively[14].The diffraction peaks at around 2θ=39.7°,46.4°,and 67.6°are assigned to the(111),(200),and(220)crystalline planes of the face-centered cubic Pt,respectively.There is no peak of oxidized Pt,indicating that the Pt(+4)was fully reduced to metallic Pt by NaBH4duringthepreparationofcatalyst,orPt(+4)wasnotdetectable by the XRD instrument used.Re was reported to have various valences of+2,+4,+6 and+7.In our Re/rGO sample,only the Re species of ReO2(020)at 31.7°,ReO3(220)at 34.2°,and a large amount of NH4ReO4at 17.2°are observed,which illustrates that Re(+7)in NH4ReO4could only be partly reduced to Re(+4)and Re(+6)species in the presence of NaBH4.However,after Re was doped into Pt/rGO catalysts,the diffraction peaks of these oxidized Re species were obviously weakened,suggestingsmaller particle sizes andhigherdispersion of oxidized Re species were obtained.Additionally,it is expected that more NH4ReO4would be reduced to metallic Re with the assistance of Pt,as evidenced by the diffraction peaks of Re(002)at 40.4°(JCPDS No.87-0715)[55]in the lower Pt/Re ratio samples(Pt25Re75/rGO and Pt50Re50/rGO),buttheRe(002)andRe(101)peaksaregreatlyobscured by the Pt(111)and C(100)diffraction peaks and cannot be discriminated in the Pt75Re25/rGO sample.Similar phenomena were also observed in the Au-Pd bimetallic catalyst in which Au promoted the reduction of Pd[56].No obvious Pt-Re alloy peak was observed in the Pt-Re/rGO catalysts,perhaps becauseofthehighdispersionof the particles.Nevertheless,it has been reported that highly dispersed Re species can alloy with Pt in all ranges of chemical composition due to their similar crystal structure,atomic radii and electronegativity[57].Therefore,it can be inferred that the interaction between platinum and rhenium species affected the development of metal particles during the reduction process,which might induce the decrement of metal particle size of Pt-Re bimetallic catalysts compared with Pt or Re monometallic catalyst,and improve the catalytic performance in the end.

The formation of Pt-Re alloy was further confirmed by HRTEM analysis.Fig.5 shows the scanning transmission electron microscopy(STEM)and energy-dispersive X-ray spectrum(EDS)mapping images for several metal particles of the Pt50Re50/rGO catalyst.As seen from the images,Pt and Re species are homogeneously dispersed within the metal particle domains(white dots),indicatingthat Pt-Re alloy formed.Fig.6 displays the typical TEM images together with the metal particle size distribution histograms of the as-prepared Pt-Re/rGO catalysts(3 wt%)with different Pt/Re molar ratios of 100:0,75:25,50:50,25:75 and 0:100.The dark particles in Fig.6(a)-(d)could be assigned to Pt and Re species according to EDS analysis.The fringe spacing of Pt50Re50/rGO(0.205 nm)is between the lattice plane spacing of Re(101)(0.211 nm)and Pt(200)(0.195 nm)[24,58],which also indicates the formation of Pt-Re alloy[59].

The Pt-Re alloy particles in Pt50Re50/rGO show the smallest average size and the narrowest distribution of(2.83±1.0)nm.Increasing the content of either Pt or Re in the catalysts would result in bigger particle sizesofthe alloy.Wetherefore deducetheformationof Pt-Realloywith smallerparticlesizesandbetterinteractionbetweenPtandRemayplay animportant role inprovidingexcellentcatalytic performance,whichis supported by the hydrogenation reactions that Pt50Re50/rGO achieved the best conversion of CAL and the highest yield of COL.For the Re/rGO catalyst,no metallic Re but only ReOxsingle crystals with the plane spacing of 0.286 nm are observed[Fig.6(e)],which is in accordance with the aforementioned XRD results and inference that it is too difficult to reduce NH4ReO4to metallic Re by NaBH4.

XPS spectra were used to investigate the surface composition of Pt-Re/rGO catalysts.The Pt 4f spectra of Pt/rGO are shown in Fig.7(a),with the expected doublets for Pt 4f7/2and Pt 4f/5/2.The intensest peak pair at 71.3 and 74.5 eV is assigned to metallic Pt,occupying 59%of the total Pt element(Table 3).The second intensest peak pair at 72.4 and 75.9 eV can be identified as Pt(+2),and the third intensest peak pair at 74.9 and 78 eV is indexed to Pt(+4)[14].While in Pt50Re50/rGO,the metallic Pt content increases to 66%,indicating that Pt-Re alloy with a composition of 50:50 promotes the reduction of Pt oxides into the metallic state.As a result,more hydrogenation active sites of metallic Pt exist in the Pt50Re50/rGO catalyst to facilitate better catalytic performance.

Fig.5.STEM and EDS elemental mapping images of 3 wt%Pt50Re50/rGO catalysts.(a)STEM with elemental mapping for Pt(green)and Re(red);(b)STEM-EDS line scanning;(c):STEMEDS spectra for Pt(green)and Re(red).

The chemical valances and composition of Re species serve as key factorsinmodifyingcatalyticactivity[60,61].Accordingly,theoxidation states of Re in the as-prepared Pt-Re/rGO catalysts were also examined by XPS analysis,and the curves were deconvoluted to quantitatively determine every single component.As shown in Fig.7(b),the Re 4f7/2peaks of Pt-Re/rGO catalysts vary greatly with the Pt:Re molar ratios.In the monometallic Re/rGO catalyst,the most common species are Re(+6)at 45.2 eV and Re(+4)at 42.2 eV,occupying ca.35%and 32%of the total Re element,respectively.The next one is Re(+7)at 46.2 eV,accounting for 22%of the total Re element,while the content of metallic Re only represents 10%(Table 3).This is further verified with theconclusionfromXRDand TEM analysis that NH4ReO4ismostly reducedtoRe(+6)andRe(+4)(whicharenonmetallic Respecies)by NaBH4.However,when Re is alloyed with Pt at different Pt/Re ratios from 25:75 to 75:25,there is an increase of the metallic Re content with a peak of 78%at 50:50(Table 3).In the meantime,the peak area of Re(+7)decreases remarkably with the increase of Pt,indicating that the existence of Pt is beneficial to the reduction of Re.

3.3.Discussions

The selective hydrogenation of unsaturated CAL has been vigorously studied.Themainthrustofthisresearchistoidentifytherequiredcharacteristics of a catalyst with high activity and selectivity to unsaturated alcohol.Generally,the reactive metal,support and metal modifier of a catalyst all feature in achieving a high selectivity to COL,which can adsorbC＝OattheactivesitesratherthanC＝C[6].Inthisstudy,aseries of Pt-based catalysts were designed with focus on the nature of the support and metal modifier.Results showed Pt alloyed with Re at 50:50 molar ratio supported on rGO was the most favorable catalyst for the selective hydrogenation of CAL to COL among the catalysts used in this work.The effects of Re can be divided into two parts based on the catalyst characterization results.Firstly,a part of Re(+7)was reduced to metallic Re to be alloyed with Pt(0).The formation of this alloy structure is generally recognized to generate a combination of electronic and geometric effects,such as higher particle size distribution and dispersion of Pt,which is beneficial to obtaining high activity or selectivity[55,62,63].Secondly,more than a half of Re(+7)was reduced to hydrophilic ReOx(mainly Re(+4)and Re(+6)).So that C＝O is more likely to be adsorbed than C＝C in CAL on ReOx,i.e.,the vicinity of active center Pt in the well-mixed metal particles.As a result,C＝O is easier to be hydrogenated than C＝C in CAL,and a high selectivity to COL is achieved.

It has been frequently reported that the COL selectivity exhibited an inverse proportionality to the size of the metal nanoparticles[2,5,64].The reasons were explained as follows:A smaller metal particle would favor the adsorption of C＝C over C＝O due to the abundance of lowcoordinated surface sites[64].While for a larger one,the density of Pt(111)facets relative to lower coordination sites over larger nanoparticles would increase,which hampers the close approach of C＝C and hence favor the C＝O hydrogenation[2].Moreover,a larger metalparticle would repel the benzene ringof CAL[65].However,in our work,the Pt50Re50/rGO catalyst with the finest average particle size of 2.8 nm(Fig.6)exhibited the highest selectivity to COL.This inconsistence is perhaps because the size differences among these Pt-Re/rGO catalysts were less than 0.7 nm(Fig.6),such minor change may only bring slight effects on the COL selectivity,which are much less than that from the stronger interactions between ReOxin the Pt-Re particles with C＝O in CAL.

Fig.6.HRTEM of Pt-Re/rGO catalysts.(a)Pt/rGO;(b)Pt75Re25/RGO;(c)Pt50Re50/RGO;(d)Pt25Re75/rGO;and(e)crystal of unreduced NH4ReO4.

Reduced graphene oxide is demonstrated as the most favorable support for Pt-Re catalysts among the tested materials.Graphene has a theoretical specific surface area ashigh as 2630 m2·g?1[66].Actually,due to the re-folding of graphene layers during the reduction of graphene oxide produced by Hummers'method,the specific surface area of rGO is generally less than 200 m2·g?1,corresponding to 10-15 folded layers[67,68].Thespecific surface area of Pt50Re50/rGO is around 62 m2·g?1,with almost 100%of mesoporous and macroporous structures(Fig.8),and the interlayer spacing of the folded graphene sheets is too small to detect by nitrogen adsorption analysis.We therefore deduce that Pt-Re nanoparticles are mainly supported on the outer surface of rGO,which is much more accessible to both hydrogen and CAL with diffusion resistance in micropores eliminated.Moreover,the giant π bonds of graphene surface prefer to orient parallelly to the plane of the benzene ring in CAL through π-π interactions[3,28],which also promotes the adsorption of CAL onto the surface of rGO and in the end accelerates the hydrogenation rate.This might be the reason why other mesoporous carbon materials like OMC and NOMC have larger specific surface areas but lower catalytic activity than rGO.

4.Conclusions

Fig.7.XPS spectra for Pt 4f(a)and Re 4f(b)in Pt-Re/rGO catalysts.

Pt-Re/rGOcatalystswithdifferentmetalratioswerepreparedforthe selectivehydrogenationofCALtoCOL.Pt50Re50/rGOshowedthehighest COLselectivityof89%and94%conversionofCALattheoptimalreaction conditions of 120°C,2.0 MPa H2and 4 h.The excellent catalytic performance could be ascribed to:(1)the successful formation of Pt-Re alloy,whichcanimprovethemetalparticlesizedistributionanddispersionof Pt,(2)theReOxthatcanselectivelyadsorbC＝Otothevicinityofmetallic active sites,which is beneficial to the selective hydrogenation;and(3)the mesoporous,macroporous,and aromatic graphite structure of rGO that is conducive to the adsorption of CAL onto the support,which will accelerate the hydrogenation process of CAL.

Supplementary Materials

Supplementary data tothis article canbefoundonlineathttps://doi.org/10.1016/j.cjche.2018.04.022.

Table 3 Surface element scanning of Pt-Re/rGO catalysts by XPS

Fig.8.(a)Nitrogen adsorption-desorption isotherms and(b)the BJH pore size distributions of Pt50Re50/rGO catalyst.