Synergistic interaction of metal-acid sites for phenol hydrodeoxygenation over bifunctional Ag/TiO2nanocatalyst

Andrew Ng Kay Lup,Faisal Abnisa,*,Wan Mohd Ashri Wan Daud,*,Mohamed Kheireddine Aroua

1Department of Chemical Engineering,Faculty of Engineering,University of Malaya,50603 Kuala Lumpur,Malaysia

2Centre for Carbon Dioxide Capture and Utilization,School of Science and Technology,Sunway University,Bandar Sunway,47500 Petaling Jaya,Malaysia

3Department of Engineering,Lancaster University,Lancaster LA1 4YW,UK

Keywords:Silver based catalyst Physicochemical properties Hydrogen spillover Metal-acid sites Hydrodeoxygenation

A B S T R A C T The use of silver metal for hydrodeoxygenation(HDO)applications is scarce and different studies have indicated of its varying HDO activity.Several computational studies have reported of silver having almost zero turnover frequency for HDO owing to its high C--O bond breaking energy barrier and low carbon and oxygen binding energies.Herein this work,titaniasupportedsilver catalystsweresynthesized andfirstly usedtoexamineits phenolHDO activity via experimental reaction runs.BET,XRD,FESEM,TEM,EDX,ICP-OES,Pyridine-FTIR,NH3-TPD and H2-TPD analyses were done to investigate its physicochemical properties.Phenomena of hydrogen spillover and metalacid site synergy were examined in this study.With the aid of TiO2reducible support,hydrogen spillover and metal-acid site interactions were observed to a certain extent but were not as superior as other Pt,Pd,Ni-based catalysts used in other HDO studies.The experimental findings showed that Ag/TiO2catalyst has mediocre phenol conversion but high benzene selectivity which confirms the explanation from other computational studies.

1.Introduction

Theuse of bifunctional metal-acid catalyst has been known to be effectiveforhydrodeoxygenation(HDO)ofphenoliccompoundsinbio-oil[1,2].These catalysts contain metal and acid sites which are instrumental in catalyzing several reactions occurring in hydrodeoxygenation.For instance,metal sites catalyze hydrogenation and hydrogenolysis reactions while acid sites catalyze dehydration,transalkylation,isomerization,alkylation and condensation reactions[3,4].Likewise,metal sites promote hydrogen adsorption and activation whereas acid sites promote adsorption and activation of O-containing model compounds throughformationofoxygenvacancies[5,6].ForphenolicHDO,removal ofhydroxylgroup(--OH)fromphenolicscouldoccurthrough:(i)direct hydrogenolysis of Caromatic--OH bond to form deoxygenated aromatics;(ii)phenylringhydrogenationtoformcycloalcoholintermediatewhich dehydrates and hydrogenates to form cycloalkane;(iii)phenolic tautomerization to form keto intermediates that undergoes-C＝O hydrogenation and dehydration to form deoxygenated aromatics[1].

The catalytic roles of acid and metal sites of HDO catalyst were well established in past studies.For HDO catalyst to exhibit excellent HDO activity,validation of the existence and nature of acid and metal sites in catalyst is not suffice.In fact,the synergistic interactions of both sites had to be analyzed in order to further understand the HDO mechanism occurring at both sites within the catalyst.Synergy of acid and metal sites is also important for the surface migration of activated hydrogen species from metal site to acid site which is known as hydrogen spillover[7].Hydrogen spillover is commonly made more possible by having both sides existing in close proximity.Nevertheless,physicochemical properties of the selected metal and support are also important in influencing chemisorption and activation of phenolics in the presence of hydrogen.Therefore,proper designing of a bifunctional catalyst model for HDO is necessary to ensure efficient hydrogen spillover and optimal surface acidity for excellent hydrogenolysis of C--OH bond in phenolics.

Metal-promoted zeolite is one of the bifunctional metal-acid catalysts reported to have good HDO activity owing to the presence of metal sites and high surface acidity[8].However,high surface acidity has also been known to favor secondary condensation reactions which would lead to severe catalytic coking[9].Thus,mild acid supports such as Al2O3,TiO2,ZrO2and CeO2may offer better coking resistance while having optimal acid sites for hydrogenolysis in HDO.Among these,TiO2has been increasingly studied on due to its support reducibility,metal-support interaction and higher oxophilicity for C--O bondscission[10-13].BesidesusingacidicmetaloxidesasHDOcatalyst supports,they have also been used as promoters[14]and ligands[15]which were respectively reported to improve hydrogenolysis activity and to tune the Br?nsted acid strength of catalyst.Likewise,many transitionmetalswereusedinpastHDOstudiessuchasCo,Cu,Fe,Ga,Ir,Mo,Ni,Pd,Pt,Re,Rh,Ru,Sn,W,Zn and Zr[13].The use of transition metals in HDO was also favored as they contained metal sites which effectively promote hydrogen adsorption and activation.However,one of their limitations is their high material cost which would be a major challenge to their applications at commercial scale.Silver,on the other hand,iscategorizedwithCu,Fe,Ni,TiandZnaslowcosttransitionmetals[16,17].

In several studies,silver-based catalysts were reported to have excellent activity and selectivity for chemoselective reduction of styrene oxides,stilbene oxide,carbonyls,alcohols,nitro-aromatics and epoxides[18-21].Mitsudome et al.have also further reported that heterolytic H2cleavage by silver and basic oxide interface favored chemoselective reduction of nitro compounds containing=C＝C[22].However,there were also several HDO studies that reported of the low HDO activity of Ag.For instance,Jalid et al.[23]reported of the transition metal activity for ethanol conversion to ethane with the order:Co＞Ru＞Ir＞Rh＞Ni＞Fe＞Pt＞Pd＞Cu＞Re＞Ag＞Au.Likewise,Lausche et al.[24]also reported of high C--O bond breaking energy over Ag(211)surface.The stark differences within these Ag-catalyzed HDO may very likely be due to the use of different model compounds.Thus,the physicochemical properties of Ag/TiO2have to be prior investigated without the influence of model compound to determine whether it exhibits suitable physicochemical properties for HDO process.In this work,synergistic interaction of metal-acid sites and catalytic performance of Ag/TiO2catalyst for phenol HDO were experimentally assessed.

2.Experimental

2.1.Materials

Silver nitrate(ACS reagent)and titanium(IV)dioxide(technical,≥97%)were purchased from Sigma-Aldrich.All chemicals were used as received without any purification.Purified argon,helium and gas mixtures of 10%NH3in He and 5%H2in N2were purchased from Linde Malaysia Sdn.Bhd.

2.2.Synthesis of Ag/TiO2catalysts

Titaniasupportedsilvermetalofvarioussilverloadings(1wt%,3wt%,5wt%and10wt%)wereusedascatalysts.Theyweresynthesizedviaincipient wetness impregnation method[13,25-27].In each production batch,100 g of pure anatase TiO2was impregnated with 10 ml silver nitrate aqueous solution of different concentrations to produce catalyst precursor of varying silver loadings.Impregnation suspension was stirred at room temperature for 1 h to achieve complete homogenization.Samples were then oven dried at 120°C for 24 h and calcined in air at 400°C for 24 h in a furnace.The catalyst samples were denoted as Ag/TiO2-1,Ag/TiO2-3,Ag/TiO2-5 and Ag/TiO2-10 respectively.A sample of titania support was also calcined in air at 400°C for 24 h and characterized to investigate the effect of support on catalyst.

2.3.Characterization of Ag/TiO2catalysts

Textural properties of catalysts and bare TiO2support were determined by N2isothermal adsorption-desorption using a Micromeritics ASAP 2020 surface area and porosity analyzer.Samples were outgassed under vacuum at 90 °C for 1 h and subsequently at 250 °C for 2 h before textural analysis.X-ray diffraction(XRD)patterns of catalysts were obtained using a PANalytical X'Pert diffractometer with Cu Kα(λ=0.1540598 nm)radiation.The working voltage and current of X-ray tube were 40 kV and 40 mA respectively.Scanning range was from 10°to 90°with a step size of 0.026°and scan rate of 3(°)·min?1.JCPDS card number 04-0783,42-0874 and 21-1272 for Ag,Ag2O and TiO2were used.

Surface morphology and topography of catalysts were examined using a field emission scanning electron microscope(FESEM),ZEISS GeminiSEM 300 equipment.Elemental composition and mapping of catalyst surfaces were done using energy dispersive X-ray spectroscopy(EDX)and elemental mapping features of ZEISS GeminiSEM 300.Morphology and size of catalyst particles were analyzed using a transmission electron microscope(TEM),TEM LEO 912 Omega equipment.Inductively coupled plasma optical emission spectroscopy(ICP-OES)wasdonetodeterminetheactualsilverloadingofcatalystsamples[28].

Ammonia temperature programmed desorption(NH3-TPD)was doneto study catalystacidity usinga MicromeriticsChemisorb 2720 instrumentina10vol%NH3/Hegasflow.Samples(50mg)wereloadedin a quartz U-tube reactor and heated from 298 K to 573 K at 20 K·min?1andheldat573Kfor30minunder20ml·min?1heliumflowforsample outgassing.Outgassed samples were subsequently cooled down to 298 K and used for ammonia chemisorption by changing gas flow from pure He to 10 vol%NH3/He for 30 min.Samples were then purged with He for 30 min at 373 K for physisorbed molecule removal.Next,samples were heated from 373K to1173 K at 40K·min?1usingHecarrier gas(20 ml·min?1)for ammonia desorption measurement.For the Ag/TiO2-10 sample,NH3-TPD was conducted using different heating rates(30,40,50,60 K·min?1)for determination of NH3desorption activation energy.Procedures for baseline correction and stabilization of thermal conductivity detector(TCD)signal were also done to enhance signal-to-noise ratio of NH3-TPD spectra.

The nature of surface Br?nsted and Lewis acid sites was determined by FTIR analyses of pyridine adsorbed on samples(Py-FTIR)using a Bruker TENSOR 27 FTIR spectrometer.Similar Py-FTIR procedures were employed with reference to other works[29,30].IR spectra were collected after pyridinedesorption atdifferentoutgassingtemperatures(770,860,950 K).

H2-TPD analysis was done to study desorption properties of adsorbedhydrogenspeciesonAg/TiO2catalystsusingtheMicromeritics Chemisorb 2720 instrument.Samples were reduced under 5 vol%H2/N2flow(20ml·min?1)at573Kfor0.5handwereallowedtocooldownto 298 K for hydrogen adsorption.After hydrogen adsorption for 0.5 h,gasflow was changed to 20 ml·min?1helium flow for physisorbed molecule removal.When TCD signal was stabilized,hydrogen desorption tests were then carried out by heating samples from 298 K to 1000 K at40K·min?1under20ml·min?1heliumflow.ForAg/TiO2-10sample,H2-TPD was done at several heating rates(30,40,50,60 K·min?1)for determination of H2desorption activation energy.

2.4.Catalytic reactivity test

HDO activity of Ag/TiO2catalyst was examined by conducting gasphase hydrodeoxygenation of phenol over Ag/TiO2catalysts in a continuous-flow microreactor.Before reaction,catalyst(230 mg)was pretreated with hydrogen at 400 °C under a 20 ml·min?1of 5%H2/N2flow for 2 h in the microreactor.Reactions were then performed at 460,480,500 and 520 K with similar catalyst mass and hydrogenflow rate.All experiments were done at atmospheric pressure and with 3.5 μmol·s?1average phenol flow rate.Reaction product mixtures were collected and analyzed via GC-MS.Phenol conversion was calculated as percentage ratio of the number of moles of reacted phenol to thenumberofmolesofinitialphenol.Productselectivitywascalculated as percentage ratio of the number of moles of a certain product to the number of moles of total products[31].

3.Results and Discussion

3.1.Physical properties of Ag/TiO2catalysts

3.1.1.Textural properties

Fig.1.A)N2adsorption-desorption isotherms and B)respective pore size distributions of samples with different Ag loadings:(a)TiO2(calcined at 400°C),(b)Ag/TiO2-1,(c)Ag/TiO2-3,(d)Ag/TiO2-5 and(e)Ag/TiO2-10.

Based on N2adsorption-desorption isotherms(Fig.1A),type IV isotherms with H1 hysteresis loop type were obtained which indicate mesoporous character of Ag/TiO2samples[32-34].At P/Po＜0.04,the initial sharp increase in N2adsorbed quantity was attributed to monolayer N2adsorption on sample.For relative pressure of 0.04 to 0.8,the steady increase in N2adsorbed quantity was attributed to multilayer N2adsorption on sample.At P/Po＞0.8,the sharp adsorption volume increase was due to Kelvin type condensation which occurs in mesoporous pore structures.The H1 hysteresis loops as exhibited by the adsorption-desorption isotherms of TiO2and Ag/TiO2samples also indicate the characteristics of mesoporous cylindrical channels with uniform size and shape[35].Fig.1B shows the BJH pore size distributions of the samples.All samples exhibited two distinct mesopore size ranges:2-4 nm and 4-30 nm with respective modes of 3 nm and 10 nm.With the increase in Ag loading,average pore size,pore volume and specific surface area of sample were noted to decrease which are due to Ag deposition within the pores(Table 1).Specific surface areas of stock and calcined TiO2(400°C)samples were 10.531 and 10.242 m2·g?1respectively which indicated negligible sintering effect by calcination.

3.1.2.XRD patterns

XRD diffractograms(Fig.2)confirmed the presence of anatase TiO2and metallic silver in samples(b)-(e).Diffraction peaks of anatase at 2θ =25.46°,37.11°,37.96°,38.71°,48.20°,54.04°,55.21°,62.25°,62.82°,68.89°,70.42°,75.17°,76.15°,82.79°and 83.15°with their corresponding crystallographic planes(101),(103),(004),(112),(200),(105),(211),(213),(204),(116),(220),(215),(301),(224)and(312)were observed in all of the titania-supported samples[36].Ag2O small crystallites were also well dispersed as the diffraction peaks of Ag2O for(111),(200),(220)and(311)planes were not significantly detected by XRD.

Diffraction peaks at 2θ =38.11°,44.79°,65.01°and 78.17°were attributed to metallic silver with the corresponding crystallographic planes of(111),(200),(220)and(311)[37].For Ag(200)crystallites,their average crystallite sizes decreased with the increase in Ag loading(Table 1).The Ag diffraction peak at 44.79°for Ag(200)crystal plane was noted to experience a slight peak shifting effect since the actual diffraction peak for pure Ag was noted to be 44.295°instead[37].Based on Bragg's law,the XRD peak shifting of Ag(200)peak to a higher 2θ from 44.295°to 44.79°would cause the lattice spacing of Ag(200),d200to decrease from 0.20433 nm to 0.20218 nm.This may indicate the presence of compressive stress on Ag crystal in this particular crystallographic direction during its addition onto TiO2support.The compressive stress was also observed to be anisotropic as the peak shifting effect was not observed in other diffraction peaks of silver crystal planes.The absence of silver nitrate peaks in XRD patterns would indicate that all of the silver nitrate precursors in TiO2support were thermally decomposed into Ag2O or metallic Ag during calcination at 400°C.

Table 1 Physicochemical properties of TiO2and Ag/TiO2samples

Fig.2.XRD patterns of(a)TiO2(calcined at 400 °C),(b)Ag/TiO2-1,(c)Ag/TiO2-3,(d)Ag/TiO2-5 and(e)Ag/TiO2-10 where:(●)TiO2;(◇)metallic Ag.

All XRD patterns of samples exhibited similar degree of crystallinity and minimal peak broadening effect which indicated that the addition of silver from 1 wt%to 10 wt%marginally affected the crystallinity of sample.XRD peaks of silver crystallites(Fig.2)were not as prominent as compared with TiO2XRD peaks.This would be mainly attributed to the low Ag proportion and also the good dispersion of Ag nanoparticles inTiO2support.However,severaltinyAgcrystalliteswerealsodetected by XRD to give the characteristic Ag diffraction peaks at higher Ag loadings(Fig.3).The absence of 81.53°diffraction peak which corresponds to(311)crystal plane of Ag in Fig.3 may thus be due to the well dispersion of Ag nanoparticles in TiO2support and its nature of being a minor characteristic peak of Ag metal.The existence of Ag in TiO2support was also confirmed by ICP-OES and EDX analyses(Section 3.2.1).

3.1.3.Morphology and topography

Based on the FESEM micrographs(Fig.4),all Ag/TiO2samples were observed to have circularslab-like particles.The particles were arranged in numerous stacking layers in various orientations to form aggregates of Ag/TiO2.The addition of silver had no effects on the particle shape of catalyst since similar particle shape was also seen in the calcined TiO2support.The FESEM micrographs of Ag/TiO2-1 and Ag/TiO2-3 showed no visible silver deposition on catalyst surface.However,Ag/TiO2-5 micrograph(Fig.4D)showed little silver nanoparticles deposited on the surface while this phenomenon was more obvious in Ag/TiO2-10 sample(Fig.4E).These were due to the dominant pore filling effect by Ag at low loadings and the onset of Ag deposition on catalyst surface at higher loadings which were in corroboration with BET results.

Fig.3.Close-up version of XRD patterns(60°-85°)of(a)TiO2(calcined at 400 °C),(b)Ag/TiO2-1,(c)Ag/TiO2-3,(d)Ag/TiO2-5 and(e)Ag/TiO2-10 where(●)TiO2;(◇)metallic Ag.

Fig.4.FESEM micrographs of calcined TiO2,Ag/TiO2-1,Ag/TiO2-3,Ag/TiO2-5 and Ag/TiO2-10 at 50 k× (a,b,c,d,e)and 100 k× (A,B,C,D,E)magnifications.

Fig.5.TEM micrographs of(A)calcined TiO2,(B)Ag/TiO2-1,(C)Ag/TiO2-3,(D)Ag/TiO2-5 and(E)Ag/TiO2-10 at 200 k×magnification.

Identification of these fine particles as silver nanoparticles on catalyst surface was also further analyzed using TEM.TEM micrographs of samples(Fig.5)also showed similar circular slab-like particles which were attributed to titania.Extremely fine particles surrounding and within TiO2particles were also observed in TEM micrographs upon theadditionofsilver.TheseAgnanoparticleswereobservedtodecrease in particle size with the increase of metal loading(Table 1).Their particle sizes were also noted to be slightly smaller than their crystallite sizes,indicating Ag nanoparticles are anisotropic and not entirely spherical.Particle sizes of TiO2support were also analyzed from themicrographs based on more than 100 sampled particles for each sample.Average particle size of sample was observed to increase with the increase in Ag loading(Table 1).The increase in particle size with Ag loading could be due to the Ag deposition on the catalyst particles which corroborates with the decreasing trend of specific surface area of samples in textural analysis.

3.2.Chemical properties of Ag/TiO2catalysts

3.2.1.Elemental analysis and mapping

Elemental analyses via ICP-OES and EDX showed that all of the Ag/TiO2catalysts consisted only of Ag,Ti and O elements with increasing Ag loadings(Table 1).Based on the elemental mapping analysis(Fig.6),all Ag/TiO2catalysts were reported to have similar and uniform spatial dispersion of Ag and Ti on catalyst surfaces and pores.The uniform distribution of Ag and Ti would also indicate that good dispersion of metal and acid sites within catalyst was achieved.Thus,this would significantly enhance the accessibility of catalytic sites by phenolics for hydrodeoxygenation.

Fig.6.Elemental mapping of Ti(green)and Ag(red)for Ag/TiO2-10 catalyst with FESEM micrographs of 10 k×and 40 k×magnifications.

The slight discrepancies between the stoichiometric silver loadings and the actual silver loadings reported using ICP-OES were attributable to the slight sample loss during catalyst synthesis or calcination of catalysts with thermally decomposable compounds such as nitrates[38].The silver compositions of catalysts reported by EDX analysis were slightly lower than that of ICP-OES analysis.The detection of characteristicX-raypeaksduringEDXanalysiswouldrequiresufficientexcitation of elements within samples by primary electrons and sufficient generation of characteristic X-ray by excited elements which are primarily affected by the depth of the samples.The generation of characteristic X-ray by excited elements is isotropic in nature and may be reabsorbed by other atoms in samples before reaching to the surface for detection[39].Thus,the detection of silver deposited in pores by EDX may not be as effective as ICP analysis.However,it should also be noted that silver composition values reported by EDX were determined based on a particular region of catalysts,making them to be localized in nature.

3.2.2.NH3-TPD and Py-IR analyses

Fig.7 and Table 2 show the strength distribution of acid sites of bare TiO2support and Ag/TiO2catalysts as determined by NH3-TPD analysis.Based on NH3-TPD spectra,the area under the curve represents the amount of NH3desorbed from catalysts which correlates to the acid site density of catalysts.For calcined TiO2support,mild surface acidity was observed with the existence of a broad desorption peak ranging from 670 K to 800 K.The mild surface acidity of TiO2was attributed mainlytoTi4+ionsastheLewisacidsitesandmarginallytothebridging hydroxyl groups within TiO2as Br?nsted acid sites[40-42].

The addition of Ag has resulted in the increase of acid site density which is attributed to the increase in surface hydroxyl formation during Ag deposition on TiO2[43-45]and its strong proton and hydride ion donor properties[46].Acid site density of TiO2support was signi ficantly lower than that of Ag/TiO2catalysts in which addition of silver wasobservedtoproducesynergisticeffectsontheacidsitesofcatalysts.The first desorption peak which had almost similar desorption temperature range asTiO2sample may involve the alteration of acid sites upon silver addition.Thealterationof natureand abundanceof acid sites may be due to the metal-support interaction between Ag and TiO2in which silvermetalwasinvolvedintheadditionalbridginghydroxylgroupsbetweentheTiandAgmetals.Forinstance,intheIRstudiesof2,6-lutidine adsorption on titania supported Nb and W based catalysts,Onfroy et al.[47]showed the progressive increase of Br?nsted acidity and decrease of TiO2Lewis acidityin thecatalysts withtheincreaseof metalloadings.Similar observations were also reported in other metal combinations such as Mo-Ti[48],polytungstate[49],Pd-Fe[50],Ru-Ti[51],Zr-Zr[52],Pd-Al[53],Al-Si[40]and Re-Al[54].

Fig.7.NH3-TPD spectra of catalyst samples at 40 K·min?1(left)and Ag/TiO2-10 at different heating rates(right).

Table 2 NH3-TPD analysis of Ag/TiO2catalysts

The nature of surface Lewis and Br?nsted acid sites in Ag/TiO2was furthervalidated viaPy-FTIR analysis.Based on Fig.8,adsorbed pyridine on Ag/TiO2resulted in observed IR bands at 1641 and 1540 cm?1that are characteristic of pyridinium ions at surface Br?nsted acid sites(PyH+)while at 1619 and 1445 cm?1that are characteristic of pyridinium ions at surface Lewis acid sites(PyL)[29,30,55].The 1490 cm?1band is attributed to adsorbed pyridine on both surface Br?nsted and Lewis acid sites.Fig.8(A)shows that the increase in Ag loading resulted in higher amount of surface Br?nsted and Lewis acid sites as observed from the increase in PyH+and PyL band intensities.Based on Fig.8(B),significant portion of PyH+bands were detected at 770 K while PyL bands were notably detected at 950 K.This observation isconsistentwithNH3-TPDresultonthehigherNH3desorptiontemperatureatLewisacid site,indicatingthegreateracidstrengthof Lewisacid site as compared with Br?nsted acid sites in Ag/TiO2catalyst.

Desorption kinetics at Ag/TiO2acid sites were also investigated via NH3-TPD spectra of Ag/TiO2-10 catalyst at various heating rates(Fig.7).Desorption rate equation was assumed to be of first order Polanyi-Wigner desorption model in this case:

where rdis desorption rate,β is heating rate,θ is coverage,A is preexponential factor,m is kinetic order of desorption(m=1),Edis desorptionactivationenergy,Risidealgasconstant,andTisabsolutetemperature.By correlating peak temperature(Tp)and surface coverage at peak temperature(θp)with heating rate(Fig.9),desorption activation energies and pre-exponential factors were determined via Redhead equations(Eqs.(2),(3))[56,57].

Fig.8.Pyridine-FTIR spectra after desorption for:(A)Ag/TiO2with different Ag loadings at 770 K,(B)Ag/TiO2-10 at different outgassing temperatures.

Fig.9.Linear plot of ln(β/Tp2)versus 1/Tpfor A)peak 1 and B)peak 2 of Ag/TiO2-10 catalyst.

The activation energies of NH3desorption from Ag/TiO2-10 were reported to be 77.84 kJ·mol?1for Br?nsted acid site and 96.24 kJ·mol?1for Lewis acid site.Both desorption phenomena showed similar range of pre-exponential factors which are 1.75× 103s?1and 1.94×103s?1respectively.The presence of Br?nsted and Lewis acid sites in Ag/TiO2catalysts is instrumental in catalyzing hydrodeoxygenation.In fact,hydrogenolysis of phenolic compounds during HDO is made possible through the respective roles of Lewis acid site as adsorption sites of phenolic compounds and Br?nsted acid site as proton donor to adsorbed phenolic intermediates[5].The significant desorption activation energies at both acid sites of Ag/TiO2would indicate its excellent adsorption capability to adsorb and activate phenolic compounds for subsequent C--OH bond scission via hydrogenolysis.In previous studies,bifunctional metal-acid catalysts with high surface acidity were noted to achieve high hydrogenolysis activity.Nevertheless,high surface acidity has also been reported to cause secondary condensation reactions to form excess phenolic pool which is a coking precursor[9].Thus,metal oxide supports may serve as alternatives due to their better coking resistances and optimal acidities for HDO.The acidities of metal oxide supported HDO catalysts which achieved more than 70%HDO conversion and selectivity were compared in Table 3.The similar range of acidities of Ag/TiO2catalyst as compared with these excellent HDO catalysts would also further support the capability of Ag/TiO2in catalyzing hydrogenolysis via its synergistic Br?nsted and Lewis acid sites.

Table 3 Comparison of various HDO catalysts in terms of acid and metal site densities

Fig.10.H2-TPD spectra of catalyst samples at 40 K·min?1(left)and Ag/TiO2-10 at different heating rates(right).

3.2.3.H2-TPD analysis

H2-TPD analysis of Ag/TiO2catalysts was done to investigate their relative catalytic activities.Based on Fig.10,the area under the curve represents the amount of hydrogen desorbed from catalyst.A catalyst with higher amount of hydrogen desorbed would be correlated with its higher amount of metal active site which is an important parameter for its catalytic activity[6,63].With the increase in Ag loading,total H2desorption amount was noted to increase which indicated the role ofsilverinfunctioningasthemainmetalactivesites.SmallerH2desorption peak was also observed in TiO2support which indicated the ability of TiO2as a reducible support for H2desorption and activation[64].Investigation of the nature and the quantity of desorbed hydrogen species via H2-TPD analysis was done to assess the propensity of hydrogen dissociation over Ag metal site and hydrogen spillover from metal to support sites which are the crucial elements for the catalysis of HDO by Ag/TiO2[7,51,65].Likewise,these would also draw further insightsonthenatureoftheadsorptionandactivationmechanismsofhydrogen over Ag/TiO2catalyst during HDO process.Peak deconvolution was done to resolve overlapping peaks.It was found that the first broad peak consisted of two overlapping desorption peaks and the second peak as the only peak by itself(Fig.11).For Ag/TiO2-1 catalyst(Fig.11b),peak deconvolution was done only on the first broad peak while the second peak was analyzed separately as a single desorption peak.The first two deconvoluted peaks which were below 673 K would be attributed to H2desorption from metal sites whereas the third deconvoluted peak which was above 673 K would be due to hydrogen spillover species[25,66,67].

Existence of two distinct deconvoluted peaks below 673 K would indicate the occurrence of H2chemisorption on metal active sites in two dissimilarmanners:molecularanddissociativechemisorption.Molecular H2chemisorption generally involves lesser energy than dissociative H2chemisorption since the latter requires additional dissociation energy to dissociate H2molecule into atomic hydrogen prior to its adsorption.Thus,from H2-TPD perspective,the first deconvoluted peak accounts for the molecular H2desorption from metal sites whereas the second deconvoluted peak accounts for the atomic H desorption from metal sites[68].Based on Table 4,it can be noted that the ratio of molecular H2desorption to atomic H desorption decreased significantly during initialaddition of Ag but subsequentlyincreased whenmore Agwas added.Adsorption of atomic hydrogen on metal active sites may be preferred as they are more readily accessible for reactions,causing the aforementionedparameterasoneofthefactorsinconsideringitsoptimalcatalytic activity.Nevertheless,it should also be realized that the increase of Ag loading may have greater compensating effects for catalytic activity due to the higher abundance of catalytic sites for reaction.

Hydrogen spillover which involves surface migration of activated H atoms from metal sites to catalyst support was also observed in Ag/TiO2catalysts.Ag functioned as the metal active sites which readily adsorb and activate hydrogen.The activated hydrogen subsequently diffused to the support surface which is thermodynamically more favored as compared with the direct generation of adsorbed H atoms on support surface.This phenomenon is instrumental for the facile activation and reaction of model compounds by hydrogen[1,59,69,70].The hydrogen spillover ratio was noted to increase from 0.018 to 0.144 during the initial increase of Ag loading from 1 wt%to 3 wt%.At higher loadings,hydrogen spillover ratio remained almost similar.Kinetic parameters of H2desorption from Ag/TiO2-10 catalyst were also determined using Redhead analysis at different heating rates.With the increase in heating rate,similar desorption modes were observed while desorption peak temperatures were shown to increase(Fig.10).By using Eq.(2),Eqs.(3)and(4)can be used to correlate desorption peak temperature with heating rate for first(m=1)and second(m=2)order H2desorption respectively(Fig.12).For instance,molecular H2desorption has first order of desorption while atomic H and hydrogen spillover species desorption have second order of desorption to account for their recombination of two hydrogen atoms into molecular hydrogen.

FormolecularH2desorptionmodefromAg/TiO2-10catalyst,desorption activation energy and pre-exponential factor were respectively determined to be 15.09 kJ·mol?1and 0.17 s?1.For atomic hydrogen desorption mode from Ag/TiO2-10 catalyst,desorption activation energy and pre-exponential factor were respectively determined to be 47.66 kJ·mol?1and 40.83 s?1.Likewise,desorption activation energy and pre-exponential factor for hydrogen spillover species desorption were 49.91 kJ·mol?1and 6.28 s?1respectively.The higher desorption activation energies of the atomic hydrogen and hydrogen spillover species desorption would indicate the stronger interaction of such hydrogen species with the metal active sites and support surface respectively.The presence of metal sites in Ag/TiO2is necessary to create a synergy with the acid sites for C--OH bond hydrogenolysis of phenolics because activated hydrogen species is an important precursor for such bond scission.Likewise,formation of activated hydrogen species on metal sites is highly dependent on hydrogen sticking probability and electronic interaction of hydrogen with the metal species.Silver which is a transition metal was also compared with other transition metals such as Ni,Pd,Pt and Ru which have high sticking probabilities(Table 3).Silver was noted to have similar range of hydrogen sorption and activation which would infer its role in producing sufficient activated hydrogen species for the hydrogenolysis of adsorbed phenolic intermediates at acid sites.The acid site with activated hydrogen ratio of Ag/TiO2also indicated the abundance of activated hydrogen for hydrogen spillover to TiO2support.Elemental mapping has shown the uniform spatial distribution of metal and acid sites with both of them existing in close proximity.Thus,with these advantages,about 12.8 mol%to 14.4 mol%of adsorbed hydrogen was observed to have undergone surface migration to TiO2support.By observing the synergistic role of acid and metal sites and the efficient hydrogen spillover over TiO2support,Ag/TiO2catalyst could be a potential HDO catalyst for phenolics in whichrigorous kinetic studies have to be donein future studies for the elucidation of its catalytic interaction.

Fig.11.Deconvoluted H2-TPD spectra of(a)TiO2,(b)Ag/TiO2-1,(c)Ag/TiO2-3,(d)Ag/TiO2-5 and(e)Ag/TiO2-10.

Table 4 H2-TPD analysis of TiO2support and Ag/TiO2catalysts

Fig.12.Linear plot of ln()versus 1/Tpfor A)peak 1,B)peak 2 and C)peak 3 of Ag/TiO2-10 catalyst.

3.3.HDO activity of Ag/TiO2catalysts

Benzene yield and phenol conversion over Ag/TiO2-10 catalyst in phenol HDO at different reaction temperatures are shown in Table 5.A further temperature increase to 550 K had a lesser increase in phenol conversion due to the counter effect of the thermodynamic limitation of H2adsorption at higher temperature[1,72].Ag/TiO2catalyst was notedtohavemediocrephenolconversionbuthighbenzeneselectivity.Other minor HDO products formed were cyclohexanol,cyclohexene,cyclohexane,cyclohexanone and benzenediols which indicated side reactions of phenyl ring hydrogenation and trans-hydroxylation overAg/TiO2.With the increase in Ag loading,phenol conversion increased proportionally while benzene selectivity remained relatively constant.Higher phenol conversion at higher Ag loading was due to the increase in available Ag metal sites for HDO process[2,13].Likewise,the smaller Ag particle size at higher loading has also resulted in better Ag dispersion which in turn improved phenol conversion.

Table 5 Catalytic performance of Ag/TiO2-10 catalyst for phenol HDO.Reaction conditions:Pressure=101325 Pa,Wcat=230 mg,100 ml·min?15%H2/N2flow,WHSV=5.09 h?1

Benzene turnover frequency(TOFbenzene)was calculated in order to quantify the specific deoxygenation rates of the catalyst.Based on Tables 5 and 6,TOFs of Ag/TiO2catalyst increased with the increase in reaction temperature and metal loading which suggests that higher number of Ag sites enhances phenol deoxygenation activity.The deoxygenation TOF of Ag/TiO2(0.78 wt%Ag,TOF500K=0.0055 s?1)was reported to be lower than TOFs of Pd and Pt-based catalysts such as Pd/ZrO2(0.89 wt%Pd,TOF573K=0.09 s?1),Pd/SiO2(1 wt%Pd,TOF523K=0.14 s?1)and Pt/SiO2(1 wt%Pd,TOF523K=0.17 s?1)[12,73]which indicates Ag has lower phenol deoxygenation activity than Pd and Pt-based HDO catalysts.Likewise,this is also in corroboration with computational findings that associated the low HDO activity of Ag to its high C--O bond breaking energy barrier[23,24].Decent HDO activity by Ag was able to be achieved in spite of its high C--O bond breaking energy barrier which is attributed to the good dispersion of Ag nanoparticles,proper distribution of metal-acid sites for efficient diffusion of adsorbed intermediates across metal-support interfaces.Among other supported metal catalysts such as Pd,Pt,Ni and Mobased catalysts,the phenol conversion of Ag/TiO2catalyst is certainlyinferior to theirs which could reach up to about 90%conversion[13].Oneoftheconsiderationsis thatthis particularwork wasdoneatatmospheric pressurewhilemanystudiesinvolved HDO processes withhigh H2pressure.

Table 6 Effect of Ag loading on catalyst performance.Reaction conditions:Pressure=101325 Pa,T=500 K,Wcat=230 mg,100 ml·min?15%H2/N2flow,WHSV=5.09 h?1

4.Conclusions

Effect of silver loading on titania supported silver catalyst and its physicochemical properties were examined in detail in this study.The incorporation of silver as metal active site onto titania support mainly involved thefillingof pores with silver at low silver loadings and subsequently the deposition of silver aggregates on catalyst surface at high silver loadings.Addition of silver onto catalyst has also resulted in the increase of acid site density with silver metal sites as Lewis acid sites andthebridginghydroxylgroupsbetweenTiandAgmetalsasBr?nsted acid sites.The Br?nsted and Lewis acid sites were noted to have NH3desorption activation energies of 77.84 kJ·mol?1and 96.24 kJ·mol?1respectively.Addition of silver onto catalyst has also resulted in the increase of H2uptake due to the increase in metal active sites as shown in H2-TPD analysis.Molecular and dissociative H2chemisorption and hydrogen spillover species were noted to be the main chemisorption modes of hydrogen on Ag metal sites.Hydrogen spillover was observed due to the close proximity of acid and metal sites as indicated by elemental mapping.The synergy between acid and metal sites in Ag/TiO2catalyst would indicate its catalytic activity for hydrogenolysis of adsorbedphenolic intermediatesatacidsitebyactivatedhydrogenspeciesthatundergonesurfacemigrationfrommetalsites.Withthevalidation of HDO activity of Ag/TiO2catalysts despite its mediocrity,further optimization studies on catalyst and reaction conditions are necessary to improve its HDO activity.

Acknowledgments

TheauthorsthankGSP-MOHE,UniversityofMalayaforfullyfunding this study through the project number“MO008-2015”.The first author would also like to thank Ministry of Higher Education of Malaysia(MOHE)for MyBrain15(MyPhD)program and IPPP for project“PG081-2016A”.