Xin Ren, Li Leng, Yueqiang Cao,*, Jing Zhang, Xuezhi Duan, Xueqing Gong, Jinghong Zhou,*,Xinggui Zhou

1 State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai 200237, China

2 Key Laboratory for Advanced Materials,Centre for Computational Chemistry and Research Institute of Industrial Catalysis,East China University of Science and Technology,Shanghai 200237, China

Keywords:Glycerol hydrogenolysis Ir-Re/SiO2 catalyst Deactivation mechanism Recycling performance Amberlyst-15 promoter

ABSTRACT Recycling performance of heterogeneous catalysts is of crucial importance especially for a batch reaction system.In this work, we demonstrate a strategy for enhancing recycling performance of Ir-Re/SiO2 catalyst synergized with amberlyst-15 in glycerol hydrogenolysis to produce 1,3-propanediol.Comprehensive characterization results reveal that the Re sites in the Ir-Re/SiO2 catalyst undergo irreversible segregation and oxidation.These hinder the formation of active Re-OH species and thus contribute to a complete and irreversible deactivation.However, the introduction of amberlyst-15 into the reactant mixture can restrain the oxidation process of Re sites and favor the formation of Re-OH species,and thus significantly enhance the catalytic recycling performance.The results demonstrated here could guide the development of excellent bimetallic catalysts with the desirable recycling performances for the reaction.

1.Introduction

The conversion of biomass resources into valuable chemicals has been proposed as a promising strategy for addressing the environmental issues and the potential deplete of fossil resources [1-10].Glycerol as a by-product of biodiesel is widely recognized as a biomass platform molecule[11-14],which can be converted into high value-added chemicalsviaseveral strategies,such as selective oxidation [15-17], reforming [18-20]and hydrogenolysis[13,14,21-26].Hydrogenolysis of glycerol among these is known as the most promising process for producing diols,i.e., 1,2-propanediol (1,2-PD) and 1,3-propanediol (1,3-PD), which are significant building block chemicals for the production of useful polymers in our daily life[11,27].1,3-PD,with two terminal hydroxyls,is more valuable relative to 1,2-PD.However,the primary and secondary hydroxyls of glycerol show similar reactivity, meanwhile the secondary hydroxyl has more steric hindrance,resulting in that the selective hydrogenolysis of glycerol into 1,3-PD is yet challenging [21,27-31].Catalysts consisting of two functional active components, which are responsible for the activation of hydrogen and the second hydroxyl group of glycerol, have been indicated as promising catalysts for this reaction and thus drawn intensive interests [21-23,32-38].

Bimetallic catalysts in architectural configurations of core/shell[39-42],alloy/intermetallics[43-48]and Janus structures[49-51]are extensively studied in the many catalytic processes, including the hydrogenolysis of glycerol.Tomishige and co-workers reported several types of Ir-ReOxcatalysts, in which Ir sites are responsible for the activation of hydrogen while ReOxspecies for the activation of glycerol molecule, and these bifunctional catalysts showed promising catalytic performances for the hydrogenolysis of glycerol [21,52-54].Our group previously prepared Ir-Re alloy catalysts by a direct reduction method without a prior calcination process which leads to the formation of rhenium oxides difficult to be reduced [22].In comparison with the counterpart of Ir-ReOxstructures resultantviaemploying a calcination procedure before the reduction, the Ir-Re alloy catalyst exhibits remarkably enhanced 1,3-PD formation rate using water as a solvent [22].Although these bimetallic systems present superior activity and/or selectivity against monometallic counterpart, achieving excellent stabilities still remains challenging for bimetallic catalysts[55,56].Several factors could lead to the deactivation of bimetallic catalysts during catalytic processes, especially the reconstructions of bimetallic catalysts[57,58],which destroy the synergistic effects of two metal components and deteriorate the catalytic performances.Although Ir-Re based catalysts among the reported ones show the highest formation rate of 1,3-PD in the hydrogenolysis of glycerol using water as a solvent [21,22,32,37,59], the strong oxyphilicity of Re could drive Re atoms in the bimetallic catalyst system move outward and then degrade hydrogenolysis performances [60-63].Understandings regarding the stabilities of Ir-Re bimetallic catalysts and then enhancing the catalytic recycling performance are vital for their further potential applications.

Herein, we demonstrate a strategy for enhancing the catalytic recycling performance of Ir-Re/SiO2catalysts for the hydrogenolysis of glycerol to 1,3-PD by adding solid acid,i.e., amberlyst-15, in the reaction system.Based on systematic investigations for the structures of catalysts, we proposed two scenarios about the changes in the structure of the Ir-Re/SiO2catalyst during the sequential reaction processes with and without adding the solid acid for better understanding the enhanced recycling performance endowed by the solid acid.These mechanistic understandings revealed here could point to the direction for improving the stability/durability of bimetallic Ir-Re catalysts and/or other bimetallic catalysts for the hydrogenolysis of glycerol.

2.Experimental

2.1.Catalyst preparation

Based on our previous work [37], silica as the support benefits the formation of Ir-Re bimetallic catalysts.Thus, SiO2(G-6, ＜100 mesh) was purchased from Fuji Silysia as the inert support for the catalysts, which was calcined at 500 °C for 3 h before use.H2-IrCl6(Strem Chemical, 99.9%) and NH4ReO4(Strem Chemical,99.999%) were impregnated on a commercial SiO2by a sequential incipient wetness method to prepare the bimetallic Ir-Re/SiO2catalyst according to our previous work [22].Typically, a certain amount of H2IrCl6aqueous solution was first impregnated onto 2 g of SiO2to give a 4%(mass)of nominal loading and the impregnated sample was aged at room temperature for 12 h and dried at 120°C for 12 h.Afterwards,a certain amount of NH4ReO4aqueous solution based on 4% (mass) of nominal loading was impregnated onto the dried sample,which was subsequently aged at room temperature for 12 h and then dried at 120°C for 12 h.The as-obtained sample was reduced in a fixed-bed reactor at 500 °C by H2(70 ml·min-1) for 3 h and then passivated at 25 °C by O2/Ar (V/V= 1:99, 30 ml·min-1) for 20 min.The monometallic Ir/SiO2and Re/SiO2catalysts were also prepared by impregnating the same amount of H2IrCl6aqueous solution and NH4ReO4aqueous solution, respectively, onto 2 g of SiO2, which were also aged, dried and reducedviathe similar procedures.

2.2.Catalytic performance tests

An autoclave (Parr 4848, 100 ml) coupled with a temperature controller and magnetic stirrer was used for the catalyst performance tests.Typically, 20 g of glycerol aqueous solution (20%(mass))mixed with 150 mg of the catalyst sample freshly reduced by pure hydrogen at 500°C were sealed into the reactor.In the case of solid acid promoted catalysts, around 50 mg of amberlyst-15 was added to the vessel with the reaction mixture.The vessel was purged with 3 MPa of H2for three times to remove the air and then pressurized to 8 MPa.The reaction mixture was then heated to 120 °C, and the stirring was switched on with a rate maintained at 500 r·min-1.After reaction at 120 °C and 8 MPa for 12 h, the vessel was cooled down to the room temperature,and then the products were collected from the solid-liquid mixture by certification.The products were analyzed using UPLC(Waters 2414) equipped with a C18 AQ column (Shiseido) and a refractive index detector.In a typical reuse experiment, the used catalyst mixture recycled by centrifugation was quickly transferred to the vessel containing a fresh 20 g of glycerol aqueous solution.About 15 mg of fresh catalyst was further added to empirically compensate the catalyst loss during centrifugation and transfer processes.In the case of performance tests for the solid acid promoted ones,20 mg of solid acid was also added,based on the previous experience[63].In all the tests,the carbon balance was over 95%.The glycerol conversion and the products selectivity were calculated as follows:

2.3.Catalyst characterizations

HAADF-STEM characterization was performed using a Tecnai G2 F20 S-Twin (FEI) with an accelerating voltage of 200 kV and a point resolution of 0.23 nm.Images were captured in STEM mode with the annular darkfield detector.The samples were dispersed in ethanol by sonication and further dispersed on copper grids.The average particle sizes were yielded by statistics analysis for more than 200 metal particles randomly selected.TGA experiments were carried out on a thermogravimetric analyzer (PerkinElmer, Pyris 1 TGA)in an air flow.Approximately 10 mg of catalyst was loaded in the sample cell and then heated from 20 to 800°C at a ramp rate of 5 °C·min-1.The temperature was held at 100 °C for 1 h to remove the possibly bound water on the samples.The used catalyst before characterization was treated as following: the sample was separated by centrifugation, and then washed with deionized water.The resulting solids were dried in vacuum at 50 °C for 12 h.The metal loadings of the fresh and used catalysts were determined by inductively coupled plasma atomic emission spectroscopy(ICP-AES, Vanan 710ES).Before the tests, a certain amount of freshly reduced sample or the used catalysts after treatment was dissolved in hydrofluoric acid.

X-ray photoelectron spectroscopy (XPS) measurements were performed on a Multilab 2000 spectrometer(Thermo VH Scientific)using an Al Kα X-ray (1486.6 eV) in anex-situway.The operating voltage and electricity of the instrument were 15 kV and 15 mA,respectively.The samples were reduced at 500 °C by H2flow in a furnace and then passivated with 1% O2/Ar at ambient temperature.Then, the processed samples were directly introduced to the analysis chamber.Binding energies were corrected by referring the binding energy of the C 1s peak at 284.6 eV.The Fourier-transform infrared(FT-IR)spectra of the catalysts were obtained on a Nicolet 6700 infrared spectrometer in anex-situway.The blank and used samples were finely grounded and mixed with KBr with a mass ratio of 1/50 and pressed into a wafer.The FTIR spectra were collected at room temperature in the air environment with a spectral resolution of 4 cm-1and accumulation of 32 scans.Powder X-ray diffraction (XRD) was performed with a Rigaku D/Max 2550 VB/PC diffractometer using Cu Kα radiation.The 2-theta degree scans covered a range of 10°-80° with a step of 0.02°.

CO chemisorption were acquired using a Micromeritics Auto-Chem 2920 equipped with a thermal conductivity detector.About 80 mg of sample was added in a U-shaped quartz reactor.Prior to each measurement, the sample was reducedin situat 500 °C for 1 h under H2/Ar(V/V= 1:20, 40 ml·min-1), then purged with Ar at 50°C for 30 min(40 ml·min-1).After cooling to 45°C,pulses of 5%CO/He were introduced until the CO peak area was stable.The mole ratio of adsorbed CO/Ir was set as 1.0.NH3temperature programmed desorption (NH3-TPD) was carried out to analyze the acidity of the catalysts on a Micromeritics AutoChem Ⅱinstrument.About 100 mg of catalyst were loaded in a U-shaped quartz tube and pretreated in Ar flow(50 ml·min-1)at 500°C for 1 h and then cooled to 120°C.The sample was subsequently exposed to a mixed gas flow of 10% NH3and 90% He (30 ml·min-1) for 0.5 h and then switched to an Ar flow (50 ml·min-1) for 1 h to remove the physically adsorbed ammonia.Finally, the temperature was raised to 800°C at 10°C·min-1under an Ar flow(30 ml·min-1),and the desorption curve was recorded at the same time.

Diffuse reflection infrared Fourier transformed spectra of CO adsorption(CO-DRIFTS)were recorded on a Perkin Elmer Spectrum 100 spectrometer with a MCT detector, using anin-situcell equipped with a sample cup, heater and ZnSe windows (Harrick Praying Mantis).About 20 mg of catalyst was placed into the sample cup and reduced at 350°C by H2for 1 h.The sample was cooled to room temperature, and then 2% CO/Ar was introduced to the cell.After the adsorption of CO reaching equilibrium, an Ar flow(50 ml·min-1) was flushed into the cell for 0.5 h to remove CO,and then the IR spectra were recorded at a resolution of 4 cm-1and accumulation of 16 scans.The spectra of the reduced catalysts under Ar flow were also recorded as the background spectra.

3.Results and Discussion

Catalytic performance tests were performed for the Ir-Re/SiO2catalyst and amberlyst-15 promoted one for four runs.The conversion of glycerol and selectivities to products are compared for these two cases in Fig.1(a), where the conversion reaches 31% with 28.6%of 1,3-PD selectivity in the first run for the Ir-Re/SiO2catalyst without adding amberylst-15.The catalyst was then recycled and returned to the reactor with a new fresh reactant mixture for a second run without any conditioning in between.In the second run,the total glycerol conversion is nearly zero, and no products can be determined, indicating a complete deactivation of the catalyst.Furthermore, the third and fourth runs were carried out, and the results are similar to those seen with the second run.This deactivation phenomenon is not consistent with that of previously reported Ir-ReOx/SiO2catalyst which showed partially decreased hydrogenolysis activity in the second catalytic run [63].These could be a result of different bimetallic structures between Ir-ReOx/SiO2and Ir-Re alloys as mentioned in the Introduction.

In contrast, with adding ambeylst-15, which is inert for the hydrogenolysis reaction (Table S1), into the reaction mixture, the Ir-Re/SiO2catalyst shows 47.2% of glycerol conversion with 39.8%of 1,3-PD selectivity.These values are higher than those obtained without amberlyst-15.Moreover, the yield of 1,3-PD achieved on the catalyst with adding ambeylst-15 is 18.8%, much higher than that on the catalyst without adding ambeylst-15 (8.9%).In addition,the catalytic performance of the Ir-Re/SiO2catalyst with adding amberlyst-15 outperforms the majority of previously reported catalysts in the term of 1,3-PD formation rate, except the Ir-ReOx/Rutile catalyst reported by Tomishige’s group (Table S2).After the first run, the catalyst was recycled and returned into the reactor with a new reaction mixture for the second catalytic run, and 35.2%of glycerol conversion is achieved over the recycled catalyst.This is clearly different from the results obtained without adding amberlyst-15, in which no hydrogenolysis activity is seen in the second run.Moreover, in the third and fourth catalytic runs, the conversion of glycerol are similar to that in the second run.Besides,the products selectivities change slightly in these four catalytic runs except a little increased selectivity to 1,2-PD along with decreased selectivity to 1-PO, as shown in Fig.1(b), due to the slightly suppressed hydrogenolysis activity.The effects of acid additives other than amberlyst-15, including HZSM-5 and H2SO4,were also compared in Table S3.It can be clearly seen that HZSM-5 can also promote the reusability of Ir-Re/SiO2for glycerol hydrogenolysis, while the conversion of glycerol decreases significantly after each runs,which is inferior to the amberlyst-15.Unexpectedly,the introduction of H2SO4into the reaction system causes the deactivation of Ir-Re/SiO2catalyst for the reaction,which could be resulted from the poison of SO2-4on the catalyst according to the previous reports [64-66].These results clearly demonstrate the difference in the durability of the Ir-Re/SiO2catalyst with adding amberlyst-15 or not, that is, the introduction of amberlyst-15 as a promoter can significantly improve the durability of the Ir-Re/SiO2catalyst.

Three conditioning ways were then employed to test if the deactivation is reversable: reducing the used catalyst at (I)200 °C and (II) 500 °C as well as (III) calcinating the used catalyst at 500°C followed by a further reduction at 200°C.Such processed catalysts were mixed fresh reactant mixtures to test the catalytic performances.After reaction for 12 h, the total glycerol conversions are close to zero over all the processed catalysts(Fig.S1),suggesting that the deactivation of the Ir-Re/SiO2catalyst is almost irreversible.The catalyst structures should presumably encounter some irreversible changes during the reaction process and/or between the sequential runs.Along this line, the stability of the Ir-Re/SiO2catalyst during the reaction process was checked by performing a similar test with reducing the reaction time by half (i.e.,6 h).As shown in Table 1, the total glycerol conversion is 16.2%,close to the half of that obtained in the 12-h catalytic run (Fig.1(a)), which implies that the change in the structure of Ir-Re/SiO2is negligible during the reaction process.The increased selectivity to 1,3-PD together with the decreased selectivity to 1-PO and 2-PO is mainly a result of suppressed over-hydrogenolysis of 1,3-PD.Subsequently,the recycled catalyst was recycled and returned to the reactor with a new fresh reactant mixture for a second run,and hydrogenolysis activity is hardly seen with this catalyst(Table 1), similar to that observed with the Ir-Re/SiO2-12 h.These results indicate that the deactivation of the catalyst is probably not from the reaction process.

Table 1Performance of Ir-Re/SiO2 catalyst without adding amberlyst-15 at 120 °C and 8 MPa

The structures of the Ir-Re/SiO2catalyst used without adding amberlyst-15 and the reference fresh one were comparatively investigated by several characterization techniques.The possibility of active metals leaching was first checked by ICP-AES measurements.Fig.1(c)shows that the loadings of Ir and Re of the used catalyst are close to those of the fresh one, suggesting a negligible leaching of the active metals.Thus, the deactivation behavior is not mainly resulted from the leaching of active metals.The possibility of the organic compounds deposition on the catalyst was further explored by TGA analysis.To exclude the interference from a potential residue of glycerol on the catalyst, a fresh catalyst was mixed with the same aqueous glycerol solution for the same time as that for the catalytic reaction, and then filtered and dried to remove the solution.The finally obtained powder catalyst(denoted as blank catalyst) was used as a reference for the TGA tests.Fig.1(d) shows both TGA profiles for the used and blank catalysts, on which the slight mass loss before and after 100°C can be assigned to the volatile of water and decomposition of glycerol on the catalyst, respectively.The similar thermal evolution phenomena suggest that the deposition of organic components after reaction for 12 h is inconspicuous,and thus cannot account for the deactivation behavior of the used catalyst.

Fig.1. (a)Glycerol conversion and(b)products selectivity within 4 catalytic runs over the Ir-Re/SiO2 catalyst and the amberlyst-15 promoted one.1,3-PD,1,2-PD,1-PO and 2-PO represent the product of 1,3-propanediol,1,2-propanediol,1-propanol and 2-propanol,respectively.Reaction conditions for each run:T=120°C,P=8 MPa,time=12 h.(c)Comparison for Ir and Re loading of the fresh Ir-Re/SiO2 catalyst(Ir-Re/SiO2)and the Ir-Re/SiO2 catalyst used without adding amberlyst-15(Ir-Re/SiO2-used).(d)TGA profiles for Ir-Re/SiO2-used and the blank catalyst.HAADF-STEM images and corresponding histograms for particle size distributions of (e) Ir-Re/SiO2 and (f) Ir-Re/SiO2-used.

Considering that the size of active metal particles plays an important role in catalytic processes [36,67-69], the particle size of Ir-Re bimetallic particles was then probed by HAADF-STEM for the fresh Ir-Re/SiO2catalyst and the one used without adding amberlyst-15.As seen in Fig.1(e)and(f),the nanoparticles are uniformly dispersed on SiO2after the catalyst reaction for 12 h,similar to that observed for the fresh one.The histograms of the particle size distribution for the used Ir-Re/SiO2catalyst is similar to that of the fresh one, and the average diameter of bimetallic Ir-Re nanoparticles over the used catalyst is calculated to be 3.2 nm,which is slightly larger than that over the fresh one.This slight increase in the particle size might deteriorate its catalytic performance, but could not cause the complete deactivation in the second catalytic run.

Note that the composition and structures of surfaces of bimetallic catalysts is vital for the catalytic performances [43,44,70].The surface properties of the fresh Ir-Re/SiO2catalyst and the one used without adding amberlyst-15 were thus comparatively investigated by XPS characterization.It can be observed that Re XPS spectrum of Ir-Re/SiO2-used in Fig.2 is remarkably different from that of the fresh one,while Ir XPS spectra are similar for both catalysts.More specifically, the intensity of highly valent Re 4f peaks increase significantly for the used Ir-Re/SiO2, and the ones of metallic Re species attenuate by a large extent.These results clearly indicate a remarkable change in the chemical environment of Re species on the surfaces of bimetallic particles,i.e., the oxidation of Re species to higher valence state.Furthermore,the ratios of Ir/Re determined by XPS are 0.52 and 0.32 for Ir-Re/SiO2and Ir-Re/SiO2-used, respectively, which demonstrate the enrichment of Re species on the bimetallic surface of the used catalyst.

Fig.2. (a) Re and (b) Ir 4f XPS spectra of the fresh Ir-Re/SiO2 catalyst and the one used without adding amberlyst-15, which are denoted as Ir-Re/SiO2 and Ir-Re/SiO2-used,respectively.

DRIFTS spectra of CO adsorption on the fresh and used Ir-Re/SiO2catalysts were further carried out to investigate their surface properties.As shown in Fig.3(a),the CO-DRIFTS spectrum of the Ir/SiO2catalyst exhibits a sharp absorption band at 2067 cm-1and a broad band at range of 1650-1870 cm-1,which can be assigned to the vibration of linearly adsorbed CO and bridge-bonded one on the Ir sites, respectively [22,52].In contrast, the absorption band corresponding to the bridged CO is hardly observed for the fresh Ir-Re/SiO2catalyst,and the one attributed to the linearly adsorbed CO shifts to 2059 cm-1along with an obvious attenuation in the intensity.The redshift of linear adsorption band can be interpreted by that the presence of Re species endow an electronic interaction with Ir sites and thus affect the adsorption strength of CO onto these sites [71,72].Meanwhile, the attenuated band intensity could be a result of the diluted Ir sites by Re on the surface.

It is noteworthy that the absorption band seen with the fresh Ir-Re/SiO2catalyst is broader than that with Ir/SiO2catalyst as indicated by the higher FWHM of Ir-Re/SiO2(67 cm-1) than that of Ir/SiO2(45 cm-1).This could be a result of the additional adsorption of CO on Re sites, as confirmed by CO-DRIFTS spectrum for the monometallic Re/SiO2catalyst shown in the inset of Fig.3(a),which demonstrates a band at 2036 cm-1assigned to the linearly adsorbed CO on metallic Re sites according to previous studies[73-75].However, the CO-DRIFTS spectrum of the used Ir-Re/SiO2catalyst exhibits a sharp and symmetric absorption band at 2080 cm-1, a higher wavenumber than that seen with the fresh one, indicating the absence of the metallic Re sites on the surface.This is in consistent with the above XPS analysis for the used catalyst that metallic Re species was oxidized to ReOxin a higher valent state which cannot adsorb CO [52,76,77].Moreover, the intensity of this band is obviously weaker than that observed with the fresh catalyst, and the position concurrently blue-shifts.CO chemisorption were further carried out to explore the amount of CO adsorbed on the fresh and the used catalyst.As seen in Table S4,the amount of CO adsorbed on the fresh catalyst is determined to be 7680 mmol·g-1, while that on the used catalyst without amberstly-15 is 6720 mmol·g-1.The decreased amount of CO adsorption is in consistent with decreased intensity of the peaks seen with CO-DRIFTS spectra in Fig.3(a).implying the segregation of Re from the bulk to the surface, which is confirmed by the XRD pattern in Fig.S2.These results imply that the Ir sites are partially covered by the ReOxspecies over the used catalyst, which is also consistent with the decreased ratio of Ir/Re in the used catalyst determined by XPS.

Fig.3. (a) CO-DRIFTS spectra of the fresh and used Ir-Re/SiO2 bimetallic catalysts as well as the monometallic ones.The inset shows the enlarged spectra of the Re/SiO2 catalyst.(b) NH3-TPD profiles for the fresh and used Ir-Re/SiO2 bimetallic catalysts together with those for the referred monometallic ones and the support.

Previous studies have well shown that the higher valent Re species on the surface are unfavorable for the activation of glycerol during the catalytic process, probably due to the suppressed formation of Re-OH species which favors the activation of glycerol during the catalytic process [52,78].The strong bond of Re-O within Re-OH species leads to a weak bond of H-O and thus high electron affinity to conjugate basic species, based on which NH3-TPD measurements were then performed to explore the surface acidic properties of fresh and used Ir-Re/SiO2catalysts.As shown in Fig.3(b), the TPD profile for the fresh monometallic Ir/SiO2and Re/SiO2catalysts show no apparent desorption peaks.The profile for the fresh Ir-Re/SiO2catalyst exhibits a remarkable desorption peak at around 240 °C, which can be assigned to the acidity induced by Re-OH [22,62].In contrast, an attenuated desorption peak is seen with the profile for the used Ir-Re/SiO2catalyst.Specifically, the amount of NH3adsorbed on the catalysts were calculated according to the peak area and listed in Table S5.The total amount of NH3adsorbed on the fresh Ir-Re/SiO2catalyst is 52.84 μmol·g-1, which is significantly higher than that of the catalyst after the first run (i.e., 5.87 μmol·g-1).Based on the amount of NH3adsorbed on catalysts, the molar ratios of NH3/Re for the fresh Ir-Re/SiO2catalyst and the one after first run are calculated to be 2.01 and 0.22, respectively.These results clearly suggest the remarkably decreased acidity of the catalyst after the first run without adding amberlyst-15.Based on a combination with the analysis by XPS and CO-DRIFTS, such weakened acidity is deduced to be that the over-oxidation of Re sites to higher valence state suppresses the formation of Re-OH.It should be recalled that the used catalyst processed by reduction at high temperature cannot regain the hydrogenolysis activity of the used Ir-Re/SiO2catalyst,implying that the oxidized Re species are unable to be reduced to the initially low valence state, probably due to the strong interaction between highly valent Re species and the support surface[22,79-81].Thus, the remarkable change in the surface properties of the Ir-Re/SiO2catalyst could dominate the deactivation of the Ir-Re/SiO2catalyst.

Based on the discussion above, we propose that the Re species in the bulk of the Ir-Re/SiO2catalyst easily segregate onto the surface due to its strong affinity toward the oxygen-containing group,such as the hydroxyl from the glycerol and the oxygen dissolved in the glycerol aqueous solution during the reaction process(Scheme 1).Moreover,the exposure of the catalyst to the air during the sequential catalytic runs causes the oxidation of Re species,which gives rise to the ReOx-rich surface and thus suppresses the formation of Re-OH.On such reconstructed surface, the adsorption and activation of glycerol molecule are restrained mainly due to lack of Re-OH species,leading to the deactivation behavior.

Scheme 1. Schematic of the changes in the structure of the Ir-Re/SiO2 catalyst during the sequential reaction processes without adding amberlyst-15,where the blue,yellow and red balls represent Ir, Re and ReOx species, respectively.

Introduction of amberlyst-15 as a promoter can significantly improve the durability of the Ir-Re/SiO2catalyst as shown in Fig.1(a).To gain insights into the promotional role of the amberlyst-15, the structures of the fresh Ir-Re/SiO2catalyst and the one used with adding amberlyst-15 in the reaction were further comparatively studied by several characterization techniques.The loadings of Ir and Re after the four catalytic runs decrease by 0.14%and 0.38%(Table S6),respectively,in comparison with those of the fresh catalyst.Furthermore, the TGA measurements were carried out for the used catalyst to explore the possibility of organic compounds decomposition.Similar to the abovementioned TGA tests, the fresh catalyst with the same amount used for the catalytic run was mixed with amberlyst-15 and reaction mixture for the same time as that for the reaction, and then recycled and dried to be employed as a blank catalyst for the TGA tests.As shown in Fig.4(a),the TG-DTG profiles of the catalyst used for four catalytic runs are similar to those of the blank catalyst,and the corresponding weight losses at around of 300 °C can be assigned to the decomposition of possible residues of reactants, products and/or amberlyst-15.The determined weight loss of the used catalyst is slightly higher than that of the blank catalyst, indicating a small number of organic compounds accumulated on the catalyst during these sequential catalytic runs, which might cause the slightly decreased conversion of glycerol in the fourth catalytic run.Fig.4(b) further shows the FTIR spectra of the used catalyst and the blank catalyst, in which no remarkable differences can be observed for these two catalysts.The TGA and FTIR results indicate a negligible deposition of organic compounds on the Ir-Re/SiO2catalyst promoted by amberlyst-15, which agrees well with previous observation that Re-based catalysts show good anticoking ability [82].

Fig.4. (a) TG-DTG profiles and (b) FTIR spectra of the blank catalyst and the fouth used Ir-Re/SiO2 catalyst promoted by amberlyst-15.HAADF-STEM images and corresponding histograms for particle size distributions of Ir-Re/SiO2 catalysts after (c) the first and (d) the fourth runs promoted by amberlyst-15.

Fig.5. (a)CO-DRIFTS spectra of the fresh Ir-Re/SiO2 catalyst and Ir-Re/SiO2 catalysts used after the first and third catalytic runs with adding amberlyst-15 as the promoter.(b)NH3-TPD profiles for Ir-Re/SiO2 catalysts used for the first run with and without adding amberlyst-15 as the promoter.(c) Re and (d) Ir 4f XPS spectra of the Ir-Re/SiO2 catalysts used after the first run with adding amberlyst-15 as the promoter.

HAADF-STEM characterizations were then performed for the catalysts after the first run and the fourth catalytic runs.As seen in Fig.4(c)and(d),the size distributions of the bimetallic nanoparticles are similar on these two used Ir-Re/SiO2catalysts.The average particle sizes are 3.1 and 3.5 nm for the Ir-Re/SiO2catalyst used after the first run and the fourth run,respectively, which are close to that of the used catalyst without adding amberlyst-15 while somewhat larger than that of the fresh catalyst.Fig.5(a) exhibits CO-DRIFTS spectra of the fresh Ir-Re/SiO2catalyst as well as those of the used catalysts after different catalytic runs.The spectrum of the Ir-Re/SiO2catalyst after the first catalytic run with adding amberlyst-15 shows a tailing absorption band centered at 2059 cm-1, similar to that of the fresh catalyst, suggesting the CO adsorption on both Ir and Re sites.Table S4 also shows the amount of CO adsorbed on the used catalyst after the first catalytic run with adding amberlyst-15 as 7264 mmol·gIr-1, which is similar to that on the fresh catalyst.In addition, XPS measurements were performed for the catalyst after the first run with adding amberlyst-15.As shown in Fig.5(c) and (d), the Re 4f and Ir 4f XPS spectra are quite similar to those of the fresh catalyst,without observing distinct increase in the intensity of peaks assigned to the high-valence Re and Ir species.Further combining with the XRD pattern in Fig.S2, adding amberlyst-15 can suppress the segregation and oxidation of Re species of the Ir-Re/SiO2catalyst during the durability tests.

Differently,the spectrum of the catalyst after the third run exhibits two split absorption bands at 2079 and 2039 cm-1, which are assigned to the adsorption of CO on Ir and Re sites, respectively,and the intensity of later is much stronger.Considering that the adsorption of CO on the Ir sites is stronger than that on the Re sites[22,75,83], such obviously stronger intensity of the band associated with the CO adsorbed on Re sites indicates higher density of Re on the surface,i.e., the enrichment of Re sites on the surface.Notably, the absorption band seen with the CO-DRIFTS spectrum attenuates slightly after the first run, probably due to the deposition of organic compounds on the catalyst surface.These could be an explanation for the slightly decreased conversion of glycerol in the second catalytic run (Fig.1(a)).

Based on the above results,it is reasonably to conclude that the amberlyst-15 could retard the over oxidation process of Re species as schematically shown in Scheme 2,which alleviates the deactivation process during the sequential catalytic runs,though the segregation of Re sites onto the surface is inevitable after several runs.Such improved anti-oxidation ability of Re by amberlyst-15 would the favor the formation of Re-OH, and it would possibly generate more Re-OH species on the surface due to the increased Re sites by segregation.These are verified by the results of NH3-TPD tests in Fig.5(b), in which the profile for the used catalyst with adding amberlyst-15 exhibits more desorbed NH3species than the catalyst used without adding amberlyst-15.On the other hand,the segregation of Re is not serious enough to cover all the surface that seen with the catalyst used without adding amberlyst-15, and the amount of Ir sites on the surface is still enough to trigger this bi-functionally catalytic process [8,22,30,84].As a result, the hydrogenolysis activity of Ir-Re/SiO2catalyst is only slightly lower in the second and subsequently catalytic runs in comparison with that of the fresh catalyst (Fig.1(a)).

Scheme 2. Schematic of the changes in the structure of the Ir-Re/SiO2 catalyst during the sequential reaction processes with adding amberlyst-15,where the blue,yellow and red balls represent Ir, Re and ReOx species, respectively.

Moreover, previous studies have indicated that solid acid promoters can effectively protonate the surface to promote the formation of Re-OH species and thus favor the hydrogenolysis of glycerol[21,30,63].The experimental results in Table S7 show that the recycled catalyst after the first catalytic run without adding amberlyst-15 exhibits significantly enhanced hydrogenolysis activity in the second run when adding 50 mg of amberlyst-15 in between.These results not only further confirm that the deactivation of the used Ir-Re/SiO2catalyst is mainly caused by the decrease in amount of Re-OH species due to the oxidation of Re sites, but also suggest that this deactivation process can be compensated by introducing a solid acid to pronate the oxidized species to favor the formation of Re-OH.

4.General Discussion

Based on the discussion above,we propose two scenarios about the changes in the structure of the Ir-Re/SiO2catalyst during the sequential reaction processes with and without adding the solid acid.For the Ir-Re/SiO2catalyst without being promoted by amberlyst-15, the Re species in the bulk easily segregate onto the surface due to its strong affinity toward the oxygen-containing group, such as the hydroxyl from the glycerol and the oxygen dissolved in the glycerol aqueous solution during the reaction process.Moreover, the exposure of the catalyst to the air during the sequential catalytic runs causes the over oxidation of Re species,which gives rise to the ReOx-rich surface and thus suppresses the formation of Re-OH.On such reconstructed surface, the adsorption and activation of glycerol molecule are restrained mainly due to lack of Re-OH species,leading to the complete deactivation.For the amberlyst-15 promoted catalytic process, the oxidation of Re sites during the sequential catalytic runs can be effectively alleviated,although it can be still observed the segregation of Re onto the surface.Thus,the bimetallic surface of Ir-Re/SiO2catalyst after several catalytic runs transformed into Re-rich surface rather than the ReOx-rich one, which can still afford the Re-OH species and promote the hydrogenolysis of glycerol.However,due to the segregation of Re and thus somewhat loss of Ir sites on the surface, the hydrogenolysis activity will decrease to a lower level.

Segregation of one metal onto the surface in the bimetallic catalysts is a normal phenomenon,which could be dominated by two factors,i.e., the relevantly higher binding strength of some adsorbates on this metal against the other one and/or the significantly different surface energies between these two metals [85].For bimetallic Ir-Re catalyst,the surface energy of Ir is obviously lower than that of Re [86,87], and thus it would be thermodynamically favorable for Ir rather than Re segregate onto the surface in light of the surface energy dominated factor.Thus, the segregation of Re is probably resulted from its stronger adsorption strength toward some adsorbates against Ir.Given that Re has much higher affinity to oxygen than Ir [61], and the segregation of Re could be induced by oxygen-containing adsorbates on the catalyst surface.This is well verified by the XPS (Fig.2) and CO-DRIFTS (Fig.3(a))analysis, which show highly valent Re species on the surface of the used Ir-Re/SiO2catalyst.In fact,the catalytic performance tests for the Ir-Re/SiO2catalyst were carried out in oxygen-containing environment: the hydrogenolysis of glycerol was performed in an aqueous solution, in which a small amount of oxygen dissolved;the treatments for recycling the used catalyst between two catalytic runs were also performed under the environment containing high-concentration oxygen (i.e., air).The later one should be the dominant aspect leading to the segregation of Re and the further complete deactivation of Ir-Re/SiO2catalyst.However, these results shown in this work cannot point to the origin of improved anti-oxidation ability of Re by the solid acid during the segregation process, which is of significance but out of the scope focused by this work.The further research associated with this issue is undergoing and will be discussed in the future work.

5.Conclusions

In summary,we have demonstrated a remarkably distinct recycling performance of Ir-Re/SiO2catalyst and the amberlyst-15 promoted one during the sequential catalytic runs for the hydrogenolysis of glycerol, the former of which shows an irreversible and complete deactivation while the later one exhibits significantly improved recycling performance.Based on the comparative characterizations for the structures of these catalysts,the deactivation has been demonstrated to be mainly resulted from the segregation of Re sites onto the surface and the overoxidation of Re sites into highly valent state, which suppress the formation of Re-OH species.The oxidization and segregation processes for Re sites are irreversible,and the structure of Ir-Re/SiO2is difficult to be recovered into the initial states.Adding amberlyst-15 in the reactant mixture as a solid acid promoter can alleviate this deactivation of the Ir-Re/SiO2catalyst, and the catalyst exhibits enhanced durability for the hydrogenonlysis of glycerol.The introduction of amberlyst-15 restrains the over-oxidation process of Re sites, which can still afford the Re-OH species and promote the hydrogenolysis of glycerol, though cannot completely suppress the segregation of Re.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was supported by Ministry of Science and Technology of the People’s Republic of China, under the Research Fund for National Key R&D Program of China(2018YFB0604700), the Natural Science Foundation of China (22008067, 22008074, 22008072,21991103), the China Postdoctoral Science Foundation(2020M681202 and 2021T140204) and Natural Science Foundation of Shanghai (20ZR1415700).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2021.07.015.