Reac Kinet Mech Cat DOI 10.1007/s11144-016-0975-z

The influence of impregnation sequence on glycerol hydrogenolysis over iridium-rhenium catalyst Wenting Luo1,3 • Yuan Lyu1 • Leifeng Gong1 Hong Du1,3 • Miao Jiang1,3 • Yunjie Ding1,2

•

Received: 9 October 2015 / Accepted: 11 January 2016 Ó Akade´miai Kiado´, Budapest, Hungary 2016

Abstract A series of iridium-rhenium catalysts with different impregnation sequences were prepared and the catalytic properties of these catalysts for glycerol hydrogenolysis were evaluated. The catalyst prepared by the impregnation of Ir prior to Re afforded the best selectivity to 1,3-propanediol of 49.2 % with 43.7 % glycerol conversion. These catalysts were also characterized by temperature programmed reduction (TPR), X-ray diffraction (XRD), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), NH3-temperature programmed desorption (NH3-TPD), Fourier transform infrared spectroscopy (FTIR) and H2-chemisorption. It was found that Ir0 species played a key role on the activities of glycerol conversion, and Re species might act as an efficient promoter to improve the dispersive state of Ir species as well as the reducibility. Kinetic studies have also been conducted and it was observed that the number of Ir0 may affect the extent of consecutive hydrogenolysis of 1,3-propanediol over catalysts, and an appropriate number of Ir0 was of importance to 1,3-propanediol selectivity. Keywords

Glycerol  1,3-propanediol  Hydrogenolysis  Iridium  Rhenium

Electronic supplementary material The online version of this article (doi:10.1007/s11144-016-0975-z) contains supplementary material, which is available to authorized users. & Yunjie Ding [email protected] 1

Dalian National Laboratory for Clean Energy, Chinese Academy of Sciences, 457 Zhongshan Road, Dalian, Liaoning 116023, China

2

State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, 457 Zhongshan Road, Dalian, Liaoning 116023, China

3

University of Chinese Academy of Sciences, No.19A Yuquan Road, Beijing 100049, China

123

Reac Kinet Mech Cat

Introduction Biomass is widely thought as a sustainable replacement of finite fossil energy in the future. As a result, plenty of glycerol will be mass produced as a by-product of the developing biodiesel industry [1]. But now glycerol is in limited demand as an industrial chemical. So it is necessary to find out an effective use of glycerol. Among many glycerol transformation processes, hydrogenolysis of glycerol is a very effective one of them. A lot of studies about glycerol hydrogenolysis to 1,2propanediol have been reported [2, 3]. But 1,3-propanediol as another product of glycerol hydrogenolysis has greater economic value because it is the monomer of polytrimethylene terephthalate (PTT). So hydrogenolysis of glycerol to 1,3propanediol is drawing more attention these years. Recently, many kinds of catalysts have been developed for glycerol hydrogenolysis to produce 1,3-propanediol [4–18]. He et al. [4] investigated the influence of Re oxide added into Pd catalyst and found that Re oxide could increase the conversion of glycerol and the selectivity to 1,3-propanediol. Kaneda et al. [6] synthesized boehmite-supported Pt nanoparticles/WOx material (Pt/ WOx/AlOOH) and found that this material is a highly efficient catalyst for glycerol hydrogenolysis reaction. They thought the good performance of Pt/WOx/ AlOOH is caused by the active Al–OH species on the surface of boehmite. Zhang et al. used mesoporous Ti–W oxides as support to develop a catalyst with a loading of 2 wt% Pt and this catalyst showed a high 1,3-propanediol selectivity(40.3 %) and glycerol conversion(24.2 %). Tomishige et al. [13–16] observed high selectivity to 1,3-propanediol of Ir-ReOx/SiO2 with an Ir loading of 4 wt% when reacted with the addition of a solid acid co-catalyst in a batch reactor and found that the reduced Re oxides interacted with the Ir were the active sites for glycerol hydrogenolysis. Zhou et al. prepared bimetallic Ir–Re catalysts by reducing them before calcination and demonstrated that the interaction of metallic Ir and Re was important for this reaction [17, 18]. According to the reports of Tomishige and Zhou, the catalysts were prepared by sequential impregnation of H2IrCl6 and NH4ReO4 aqueous solution over SiO2 [13– 18]. But the influence of different impregnation sequences on Ir–Re catalysts for the title reaction has never been studied in the literature, which might influence the interaction of the two metal components with support and the number of Ir0 on the surface. In this paper, we report our investigation on the influence of impregnation sequences of Ir and Re components on the performance of glycerol hydrogenolysis. The catalytic performance of SiO2 supporting Ir–Re catalyst was tested in a continuous flowing trickle bed reactor, which we thought to be more suitable than batch reactor to investigate the initial activity, induction period as well as the stability of the catalysts. It was found in this paper that the catalyst prepared by sequential impregnation of Re species followed by Ir on SiO2 give better glycerol conversion than the catalyst prepared by co-impregnation of Ir and Re components, while the later catalyst gives a higher selectivity to 1,3-propanediol. The trickle bed reactor was found effective for the title reaction even at an Ir loading as low as 2.0 wt% with the absence of acid co-catalyst.

123

Reac Kinet Mech Cat

Experimental section Catalysts preparation The Ir/Re/SiO2 catalyst was prepared by sequential incipient wetness impregnation. SiO2 (Qingdao Haiyang Chemical) was first impregnated with an aqueous solution of NH4ReO4, followed by drying at 393 K for 12 h and calcination at 773 K for 3 h to obtain Re/SiO2. Then Ir was deposited on the Re/SiO2 by the same procedure with an aqueous solution of H2IrCl6. The Ir/Re/SiO2 catalyst was obtained after drying at 393 K for 12 h and calcination at 773 K for 3 h. For comparison, the Re/ Ir/SiO2 catalyst was obtained through an inverse impregnation sequence, and Ir–Re/ SiO2 was prepared by co-impregnation of SiO2 with the two metal solutions. The drying and calcination conditions during the preparation of the Re/Ir/SiO2 and Ir– Re/SiO2 catalysts were the same as those of the Ir/Re/SiO2 catalyst. Both of the contents of iridium and rhenium are 2 wt%. Catalytic performance tests Glycerol hydrogenolysis reactions were performed in a trickle bed reactor (9 mm i.d.) using 1.5 g of catalyst. The hydrogen flow was 40 ml min-1, which corresponded to a space velocity of 1600 ml (h g)-1. The glycerol flow was 0.65 ml h-1, which corresponded to a space velocity of 0.44 ml (h g)-1. The reactions were conducted at 403 K and 8 MPa. In addition to the data for pre-reduced catalyst, the catalytic performance data for comparison were the analysis results for product streams collected from 48 to 60 h. The product was collected from 24 to 36 h when reacting over pre-reduced catalyst. Effluent gaseous products were analyzed on-line using an Agilent GC-7890. Liquid phase products were esterified by acetic anhydride catalyzed by pyridine and heated at 328 K for 1.5 h. Then the sample was quantified with Agilent GC-7890 fitted with a flame ionization detector and a HP-5 capillary column. The conversion of glycerol and the selectivity to products were calculated by the following equations: Conversion of glycerol ð%Þ moles of glycerol initially added  moles of glycerol that remained ¼  100 moles of glycerol initially added Selectivity ð%Þ ¼

C mole of specific product  100 sum of C mole of glycerol consumed

Characterization Temperature programmed reduction (TPR) analyses were performed on an AMI300 catalyst characterization system. Approximately 100 mg of each sample was placed in quartz reactors and reduced in a stream of 30 mL min-1 H2 (10 vol% in Ar) from 323 to 773 K at a heating rate of 10 K min-1.

123

Reac Kinet Mech Cat

The X-ray powder diffraction (XRD) data was collected using X’pert PRO/ PANalytical Diffractometer with Cu Ka radiation. The X-ray patterns were measured at the 2h range of 10–80° at 40 kV and 40 mA. The particle size of the samples was observed using TEM using a JEM-2100 microscope operated at an accelerating voltage of 200 kV. The catalysts through the processes of grinding were placed on a copper grid after being briefly dispersed in ethanol with ultrasonic processing. X-ray photoelectron spectra (XPS) were collected on a Thermo ESCALAB 250Xi equipped with a monochromatized Al Ka radiation (1486.6 eV, 15 kV, 10.8 mA). The pressure in the analysis chamber was around 7.1 9 10-5 Pa. Temperature programmed desorption of ammonia (NH3-TPD) was carried out on an AMI-300 catalyst characterization system equipped with a thermal conductivity detector (TCD). A 100 mg sample was pretreated in Ar flow at 873 K for 0.5 h. Then, after cooling to 373 K under continuous flow of argon, ammonia adsorption was performed by admitting a flow of 5 % NH3/Ar flow for 2 h using the static adsorption method. In order to remove all the physically adsorbed ammonia, the sample was purged in a helium flow of 30 ml min-1 at 373 K for 0.5 h. Then the NH3-TPD pattern was recorded by heating from 373 to 873 K with a ramp of 10 K/min in the flow of Ar. At the same time, the desorbed NH3 was monitored by online TCD analysis. Fourier transform infrared spectra of adsorbed pyridine were determined on a BRUKER EQUINOX 55 spectrometer equipped with a DTGS detector. The sample (20 mg) was prepared by pressing into a self-supported wafer. Then the sample was first outgassed at 673 K for 1 h in vacuum (10-2 Pa). When the sample was cooled down to room temperature, a background spectrum was recorded. After the adsorption of pyridine at room temperature, the sample was evacuated at 423 K for 0.5 h (10-2 Pa), and then spectrum was recorded at room temperature. H2-chemisorption was performed on an AMI-300 catalyst characterization system. About 100 mg of the catalyst samples were reduced in flowing 10 % H2/Ar at 463 K for 2 h, followed by purging with argon at 463 K for 0.5 h. Then the catalysts were cooled down to 313 K as the sample for pulse adsorption. The adsorbed gas was 10 % H2/Ar and the carrier gas was argon.

Results and discussion Catalysts characterization The TPR profiles of catalysts with different impregnation sequence and monocomponent catalysts are shown in Fig. 1. The TPR profile of Re/SiO2 showed a single reduction peak at 583 K. The H2-consumption peak which belonged to the reduction of iridium component was observed at 493 K in the TPR profile of Ir/SiO2 catalyst. Compared with the TPR profiles of Re/SiO2 and Ir/SiO2, obvious shifts of the reduction peak towards lower temperature were observed for all bi-component catalysts, especially Ir–Re/SiO2 and Ir/Re/SiO2 catalysts, which might be caused by smaller IrO2 particles. There was only one reduction peak at 482 K in TPR profile of

123

Reac Kinet Mech Cat

Fig. 1 Temperature programmed reduction (TPR) profiles of the catalyst with different impregnation sequence

Table 1 The results of TPR

Catalyst

H2 consumption (lmol/g)

Re valence

Ir/Re/SiO2

386.9

3.7

Re/Ir/SiO2

366.6

4.0

Re/SiO2

345.6

4.4

the Re/Ir/SiO2 catalyst, while two peaks at 402 and 435 K were observed in TPR profiles of the Ir–Re/SiO2 and Ir/Re/SiO2 catalysts. The appearance of two peaks could be due to different sizes of Ir–Re particles. Based on the above observations, it is obvious that the metallic Ir particles as the active sites for the title reaction are difficult to form in the case of Re/Ir/SiO2 catalyst at reaction temperature (403 K). The valence of Re, which was calculated by the amounts of H2 consumption, was around 4 (Table 1). So Re species on the reduced catalysts existed in the form of oxides, and we denoted them as ReOx. The XRD patterns of fresh catalysts as well as used ones are shown in Fig. 2. The diffraction peak at 2h = 21.4° is attributed to amorphous silica gel. The diffraction peaks of IrO2(110) at 2h = 26.7°, IrO2(101) at 2h = 34.2°, IrO2(211) at 2h = 53.1° are observed (PDF 01-088-0288) on all fresh catalysts. The peak due to ReOx species was not detected, suggesting that the ReOx species is highly dispersed. A diffusive diffraction peak of Ir metallic particle at 2h = 40.5° (PDF 01-0870715) was observed, accompanied with the disappearance of diffraction peaks of IrO2 on the spent Ir–Re/SiO2 and Ir/Re/SiO2 catalysts, while the XRD patterns of Re/Ir/SiO2 showed only the diffraction peaks of IrO2 without any diffraction peak of Ir metallic particle. It can be concluded that IrO2 was easily reduced to Ir metallic

123

Reac Kinet Mech Cat

Fig. 2 XRD profiles of the catalyst with different impregnation sequence before and after reaction

particles on Ir–Re/SiO2 and Ir/Re/SiO2 catalysts during the title reaction, but IrO2 was very difficult to reduce into Ir metallic particles in the case of Re/Ir/SiO2 catalyst. These results are consistent with TPR experiments. Fig. 3 shows the TEM images of the catalysts with different impregnation sequence before and after glycerol hydrogenolysis reaction. The average particle sizes of these catalysts are listed in Table 2. The average particle size of the Ir species over fresh Ir–Re/SiO2 was 2.87 nm, and it reduced to 2.33 nm during the title reaction. Similar Ir metal dispersion results were observed for Ir/Re/SiO2 catalyst. However, the average particle sizes of fresh and spent Re/Ir/SiO2 catalysts were 6.08 and 2.68 nm, respectively, which means a worse dispersive state. According to the TPR and XRD results in Figs. 1 and 2, the Re species were partially reduced under the reaction condition to highly dispersed ReOx particles, which were difficult to observe in XRD pattern in Fig. 2 [19]. So the particles existed in Fig. 3 were thought to be Ir particles instead of ReOx particles. HRTEM images (insets in Figs. 3a and 3b) show the lattice fingers of 0.23 nm, 0.22 and 0.32 nm on these fresh and spent catalysts, ascribed to IrO2(200), Ir(111) and IrO2(110). Fig. S1 shows the Ir 4f XPS spectra of Ir–Re/SiO2, Ir/Re/SiO2 and Re/Ir/SiO2 before and after glycerol hydrogenolysis, which is summarized in Table 3. The peak at 61.8 eV over Ir–Re/SiO2 and 61.7 eV over Ir/Re/SiO2 is assigned to the binding energy of Ir 4f7/2 for Ir4? species before reaction [20]. After reaction, the XPS spectra of Ir–Re/SiO2 and Ir/Re/SiO2 showed single peak at 60.7 and 60.4 eV, respectively, corresponding to the binding energy of Ir 4f7/2 for Ir0 species [21]. In the case of the fresh Re/Ir/SiO2 catalyst, the XPS spectra showed a peak at 61.4 eV which is assigned to the Ir4?, in the case of the Re/Ir/SiO2 catalyst before reaction, but two types of binding energies were found over spent sample, one of which

123

Reac Kinet Mech Cat

Fig. 3 TEM image of the catalysts with different impregnation sequence before and after reaction. a Ir– Re/SiO2 before reaction, b Ir–Re/SiO2 after reaction, c Ir/Re/SiO2 before reaction, d Ir/Re/SiO2 after reaction, e Re/Ir/SiO2 before reaction, f Re/Ir/SiO2 after reaction

Table 2 Particle size of the catalysts with different impregnation sequence before and after reaction Catalyst The Ir average particle size before reaction/nm Ir dispersion before reaction calculated by TEM (%) The average particle size after reaction/nm Ir dispersion after reaction calculated by TEM (%)

Ir–Re/SiO2 2.87 38.5 2.33 47.4

Ir/Re/SiO2 2.94 37.6 2.34 47.2

Re/Ir/SiO2 6.08 18.2 2.68 41.2

corresponding to Ir metallic particles at 60.6 eV and the other at 61.5 eV corresponding to IrO2 species. The XPS results demonstrate that the IrO2 particle on Ir/Re/SiO2 and Ir–Re/SiO2 catalysts could completely be reduced into Ir metallic particles during reaction at 403 K, while IrO2 on the Re/Ir/SiO2 sample was

123

Reac Kinet Mech Cat Table 3 The summarized information of XPS for catalysts prepared with different impregnation sequence Catalyst

Ir species

Re species

Bonding energy (Ir 4f7/2)/eV

Valence

Bonding energy (Re 4f7/2)/eV

Valence

Fresh

Spent

Fresh

Spent

Fresh

Spent

Fresh

Spent

Ir–Re/SiO2

61.8

60.7

Ir/Re/SiO2

61.7

60.4

?4

0

46.1

45.1

?6 to ?7

?6 to ?7

?4

0

45.9

45.2

?6 to ?7

Re/Ir/SiO2

61.7

61.5; 60.6

?6 to ?7

?4

?4; 0

45.7

45.1

?6 to ?7

?6 to ?7

partially reduced into Ir metallic particles. In other words, higher temperature is necessary to reduce these IrO2 species. It was obvious that most part of Ir species on Re/Ir/SiO2 remained oxidative state during reaction. Fig. S2 presents Re 4f XPS spectra for catalysts prepared by different impregnation sequence before and after reaction. The XPS spectra for these catalysts showed a peak at 45.7–46.1 eV. The binding energies of Re 4f7/2 for Re6? and Re7? are 45.0 and 46.9 eV, respectively [22, 23]. The binding energies of Re 4f7/2 showed in the Fig. S2b were among 45.0 and 46.9 eV. It is well known that NH4ReO4 can be pyrolyzed wholly into Re2O7 in air, so valence state of the Re species here should be ?7 [24]. The observed lower values of binding energy for Re 4f7/2 than that of the standard Re7? species may be caused by Re species interacting with support or IrO2 particles [22]. The binding energies of ca. 45.1 eV for Re 4f7/2 were obtained on these catalysts after reaction, just a little lower than fresh catalysts. This is due to the partial reduction of Re species during the reaction under the H2 atmosphere. H2 chemisorption was carried out in order to determine the number of metallic Ir active sites and the results are shown in Table 4. The highest H2 uptake of 0.928 lmol g-1 was observed on the Ir/Re/SiO2 catalyst. The adsorption amount of Ir–Re/SiO2 and Re/Ir/SiO2 were 0.373 and 0.101 lmol g-1, respectively. No obvious H2 uptake was detected on the Ir/SiO2 and Re/SiO2 catalysts reduced at the reaction temperature. This result suggested that the oxidized Ir and Re species did not adsorb H2 molecules. The metal dispersion of the catalysts calculated from H2 chemisorption is much smaller than that calculated from the particle size determined from TEM. So it is deduced that H2 adsorption is probably suppressed by the coverage of ReOx species on Ir metal surface, because the oxidized Re species could not adsorb H2 molecule. The FT-IR spectra of pyridine chemisorption of these catalysts after evacuation at 200 °C were detected and the results are shown in Fig. 4. The band at 1450 cm-1 corresponds to the coordinated pyridine adsorbed on Lewis acid sites, and the band at 1540 cm-1 is attributed to pyridine ions formed by interaction with Brønsted acid Table 4 Uptake of H2 of catalyst with different impregnation sequence Catalyst

Ir–Re/SiO2

Ir/Re/SiO2

Re/Ir/SiO2

Uptake of H2 (lmol g-1)

0.373

0.928

0.101

Dispersion calculated by H2 chemisorption (%)

0.7

1.8

0.2

123

Reac Kinet Mech Cat

Fig. 4 FT-IR spectra of adsorbed pyridine of the catalysts with different impregnation sequence

sites [25]. While the band at 1495 cm-1 is attributed to both Lewis and Brønsted types of acidic sites. As shown in Fig. 4, no IR absorption of Brønsted acid sites was observed on the spectra of SiO2 and Ir/SiO2, except the signal of pyridine physisorption on Si–OH of SiO2 at 1447 cm-1. There were IR absorption bands at 1450 cm-1, 1495 cm-1 and 1540 cm-1 on the Re/SiO2, Ir–Re/SiO2, Ir/Re/SiO2 and Re/Ir/SiO2 catalysts, implying that these acidic sites resulted from the ReOx doping. Fig. 5 shows the NH3-TPD profiles of the catalysts with different impregnation sequence and mono-component catalysts. A very weak NH3 desorption peak with a wide temperature range centered at 573 K was detected on the Ir/SiO2 catalyst, while there was an apparent NH3 desorption peak at temperature from 473 to 493 K on each catalyst containing ReOx. The NH3 uptake of the samples calculated by integral of the NH3 desorption is listed in Table 5. The NH3 desorption amount of the Ir–Re/SiO2 and Ir/Re/SiO2 catalysts were 233.2 lmol g-1 and 213.5 lmol g-1, respectively, which were obviously lower than that of the Re/Ir/SiO2 sample (312.1 lmol g-1). This result is consistent with the Py-IR measurements. On the basis of the above observations, it is concluded that Re species is inferred to be the main supplier for acidity of the catalysts. This conclusion is confirmed from results of He [4] who tested acid amounts of 5PdxRe/SBA-15(x = 1, 3, 5, 10). The addition of Re from 1 to 10 % would increase the acid amount of the catalysts from 108.2 to 210.6 lmol g-1. ReOx might also supply the hydroxyl group in the existence of water that could interact with glycerol to form 2,3-dihydroxypropoxide [19, 26]. Catalytic activity tests The catalytic performances of the glycerol hydrogenolysis over the SiO2-supported Ir–Re, Ir and Re catalysts are listed in Table 6. The products of glycerol hydrogenolysis are mainly 1,3-propanediol (1,3-PD), 1,2-propanediol (1,2-PD),

123

Reac Kinet Mech Cat

Fig. 5 NH3-TPD profiles of the catalysts with different impregnation sequence

Table 5 Uptake of NH3 of catalyst with different impregnation sequence

Catalyst

Uptake of NH3 (lmol g-1)

Ir–Re/SiO2

233.2

Ir/Re/SiO2

213.5

Re/Ir/SiO2

312.1

Ir/SiO2

27.4

Re/SiO2

239.6

1-propanol (1-PO), 2-propanol (2-PO), propane with trace amounts of ethanol and ethane. It is shown in Table 6 that glycerol conversion varied in the order: Ir/Re/ SiO2 (54.4 %) [ Ir–Re/SiO2 (43.7 %) [ Re/Ir/SiO2 (3.2 %), and the selectivity to 1,3-propanediol followed the order Re/Ir/SiO2 (64.4 %) [ Ir–Re/SiO2 (49.2 %) [ Ir/Re/SiO2 (39.9 %). The 1,3-propanediol yield of Ir/Re/SiO2(21.7 %) and Ir–Re/SiO2(21.5 %) were better than those of other Ir–Re system catalysts without the existence of acid co-catalysts even with an Ir loading (4.0 wt%) a facotr of two higher than ours [13, 17, 18]. The catalytic performance of Ir/Re/SiO2 and Ir– Re/SiO2 were obviously much higher than that of Re/Ir/SiO2. Ir/SiO2, Re/SiO2 and their mechanical mixture catalysts showed poor activity for glycerol conversion. The reaction path of glycerol hydrogenolysis is displayed in Scheme 1. First, glycerol reacts to form glycols (1,3-propanediol and 1,2-propanediol). Then, glycols continue reacting to form monohydric alcohols (1-propanol and 2-propanol). Finally, monohydric alcohols react to form alkanes. It was always found that the selectivity to 1-propanol and propane increased with the decrease of selectivity to 1,3-propanediol. This might be caused by further hydrogenolysis of 1,3-propanediol and 1,2-propanediol, which we will confirm below by kinetic studies. The plot of

123

Reac Kinet Mech Cat Table 6 Catalytic performance in the hydrogenolysis of glycero0 l over catalyst with different impregnation sequence Catalyst

Conv./ %

Selectivity/ % 1,3-PDb

1,2-PDc

1-POd

2-POe

C3H8

Otherse 0.1

Ir–Re/SiO2

43.7

49.2

7.0

34.5

5.5

3.8

Ir/Re/SiO2

54.4

39.9

4.3

44.4

5.5

5.8

0.1

Re/Ir/SiO2

3.2

64.4

13.1

15.0

5.3

2.2

0

0.6

65.3

18.7

12.1

3.8

0.0

0

–

–

–

–

–

–

0.8

14.1

0.0

69.0

16.9

0.0

0.0

54.5

33.5

9.5

42.2

11.0

3.2

0.6

Ir/SiO2 Re/SiO2

Trace

Ir/SiO2 ? Re/SiOg2 Re/Ir/SiOh2 a

-1

Reaction conditions: P = 8.0 MPa, T = 403 K, WHSV = 0.5 h , 1.5 g catalyst, 80 wt% glycerol aqueous solution,VH2 = 40 ml/min, the analysis product sample was the stream collected from 24 to 36 h except pre-reduced catalyst b

1,3-PD = 1,3-propandiol

c

1,2-PD = 1,2-propanediol

d

1-PO = 1-propanol

e

2-PO = 2-propanol

f

Others include ethanol, ethane

g

Mixture of 1.5 g Ir/SiO2 and 1.5 g Re/SiO2 which was reduced at 650 K for 2 h; the reaction condition was the same as above h The catalyst was reduced at 463 K for 2 h before reaction. Reaction condition: P = 8.0 MPa, T = 403 K, WHSV = 0.5 h-1, 1.5 g catalyst, 80 wt% glycerol aqueous solution,VH2 = 40 ml/min. The analysis product sample was the stream collected between 24 to 36 h

Scheme 1 Reaction path of glycerol hydrogenolysis

conversion of glycerol and selectivity of products versus time on stream was shown in Fig. S3. A distinct initial activity induction period (ca. 48 h) was observed when reaction proceeded, which might be related to the gradual formation of catalytic

123

Reac Kinet Mech Cat

active sites under the reaction conditions. The catalytic performance of catalysts varied little after induction period when reaction proceeding within 108 h. TPR results showed that IrO2 in the Ir–Re/SiO2 and Ir/Re/SiO2 catalysts are reducible at reaction temperature, but higher temperature is necessary to reduce IrO2 on the Re/Ir/SiO2 and Ir/SiO2 catalysts. The XPS results confirmed the TPR results. XRD patterns also showed that IrO2 was the solely observed species over used Ir/SiO2 and Re/Ir/SiO2 samples, but a diffusive diffraction peak of Ir metallic particles was observed on the spent Ir–Re/SiO2 and Ir/Re/SiO2 catalysts. Based on the characterization results, it can be concluded that lower active for glycerol hydrogenolysis is due to most part of Ir species on Re/Ir/SiO2 remaining oxidative state, and the impregnation sequence would influence the reduction of IrO2 on these catalysts during the title reaction. If we reduced Re/Ir/SiO2 at 463 K for 2 h before the reaction (Table 6), the glycerol conversion and selectivity of 1,3-propanediol over Re/Ir/SiO2 were 48.9 and 37.0 %. The induction period of pre-reduced Re/Ir/SiO2 catalyst was shortened to 24 h. This result proves that Ir0 for this reaction is really of great importance. Diffraction peaks of IrO2 for fresh Re/Ir/SiO2 were observed more distinct than those for Ir–Re/SiO2 and Ir/Re/SiO2. So the grain size of Ir on Ir–Re/SiO2 and Ir/Re/ SiO2 was smaller than that on Re/Ir/SiO2. Particle sizes of IrO2 showed by TEM images also proved that Ir on Ir–Re/SiO2 and Ir/Re/SiO2 were better dispersed than that on Re/Ir/SiO2. In the reduction process, IrO2 was first reduced and then the active hydrogen atom formed on Ir0 particles transferred into Re oxides to reduce the high value of Re oxides species. Smaller IrO2 particles are easy to reduce, so the reduction temperature of catalysts with smaller particles is lower, which is consistent with the results of TPR. So Re species on Ir–Re/SiO2 and Ir/Re/ SiO2 would act as an efficient promoter to improve the dispersive state of Ir species. For Re/Ir/SiO2, the IrO2 particle was first formed without existence of Re species. Accordingly, IrO2 on Re/Ir/SiO2 has a worse dispersion. H2 chemisorption data showed the order of H2 adsorption capacity is as follows: Ir/Re/SiO2 [ Ir–Re/ SiO2 [ Re/Ir/SiO2, which might be caused by the fact that the promotional effect of ReOx on the dispersion of IrO2 for Ir/Re/SiO2 and Ir–Re/SiO2 leads to more Ir0 sites formed during the initial activity induction period. TPR results also showed that Ir and Re species interacted with each other on bicomponent catalysts because the reduction peaks had shift to low temperature compared with mono-component catalysts. It was also found that the reduction peaks in high temperature region disappeared, which indicated that there was no isolated ReOx on the bi-component catalysts. In addition, there were almost no activities for glycerol hydrogenolysis over Re/SiO2, Ir/SiO2 and their mixture catalysts. So the active sites for this reaction must be the interface between Ir0 and Re species. What we concluded above was well consistent with what Tomishige et al. [16] observed by X-ray absorption fine structure method and the study of Zhou et al. [17, 18]. with bimetallic Ir–Re alloy. The reaction pathway of glycerol hydrogenolysis might be that glycerol form intermediates with the presence of acidic sites, followed by hydrogenation of the intermediates with the hydrogen atoms produced on the Ir metallic surface. The co-existence of Ir0 and ReOx is essential for this reaction. XRD patterns suggest that ReOx dispersed well on the Ir–Re catalysts. NH3-TPD data and FT-IR results showed that the acid amounts might be in proportion to the

123

Reac Kinet Mech Cat

amount of ReOx on the surface. ReOx acted as the active sites to adhere glycerol, should play an important role on glycerol conversion. More Re species should adsorb more substrates which might result in the increase of conversion. But the catalytic performance was not consistent with this inference. However, Ir–Re/SiO2 with a little larger acid amounts (233.2 lmol g-1) exhibited a lower conversion. But higher conversion was achieved on the catalyst with larger number of Ir0. This might be due to the fact that the step of the transfer of active hydrogen species formed on Ir0 to adsorbed glycerol is the rate determining step. It has also been mentioned in other literature that the active hydrogen species transferring is the ratedetermining step [13]. Kinetic studies The glycerol hydrogenolysis on solid catalyst is a typical example for gas–liquidsolid reaction system. For such systems, the models representing the reaction kinetics cannot be directly developed using experimental data, unless the absence of mass and heat transfer resistances is ensured. The reaction system ruled out the effect of internal diffusion and external diffusion by decreasing the size of particles and improving the space velocity, and temperature gradient has been ensured to be negligible across the catalyst bed. The rate expression characterizing the glycerol hydrogenolysis reaction can be given as follows: rgly ¼ 

dCgly n ¼ kCm gly pH2 dt

where k is the apparent reaction rate constant, m and n represent the reaction orders with respect to glycerol and hydrogen, respectively [27, 28]. If the concentration of glycerol remain unchanged, simplifying the above equation gives: rgly ¼ 

dCgly ¼ k1 pnH2 dt

Here k1 = kcm gly. Solving the above equation gives: C0 Xgly ¼ k1 pnH2 t Here Xgly is the conversion of glycerol on Ir–Re/SiO2 and C0 is the initial concentration of glycerol. The curve of Xgly vs. pH2 (Fig. 6) shows that Xgly was proportional to pH2 , so the reaction order with respect to hydrogen was 1. This result was consistent with Tomishige’s reports [29, 30]. If hydrogen pressure remains unchanged, simplifying the above equation gives: rgly ¼ 

dCgly ¼ k2 Cm gly dt

here k2 ¼ knpH : 2

Assuming m = 1, solving the above equation gives:

123

Reac Kinet Mech Cat

Fig. 6 The plot of Xgly versus pH2 and X1,3-PD versus pH2. Filled square catalyst: Ir–Re/SiO2, substrate: glycerol; filled circle catalyst: Ir/Re/SiO2, substrate: glycerol; filled triangle catalyst: Ir–Re/SiO2, substrate: 1,3-propanediol; filled inverted triangle catalyst: Ir/Re/SiO2, substrate: 1,3-propanediol. R Reaction condition: P = 8.0 MPa, T = 403 K, WHSV = 2.3 h-1, 0.5 g catalyst, 80 wt% substrate aqueous solution, VH2 = 50 ml/min


 ln 1  Xgly ¼ k2 t: The curve of –ln(1 - Xgly) – ln(1 - Xgly) vs t (Fig. 7) gives a straight line passing through the origin, hence confirms that the reaction is a pseudo-first-order reaction of glycerol, which was also investigated on other catalysts for glycerol hydrogenation [31]. So the rate of reaction could be given as: rgly ¼ 

dCgly ¼ kCpH2 : dt

The rate constant of glycerol hydrogenolysis on Ir–Re/SiO2 sample is 0.030 h-1, which is obtained from the slope of the curve in Fig. 7. The rate constant of glycerol hydrogenolysis on Ir/Re/SiO2 catalyst was found to be 0.039 h-1 by the same method used above. The rate constants of 1, 3-propanediol hydrogenolysis on Ir–Re/SiO2 and Ir/Re/ SiO2 catalysts were found to be 0.0040 h-1 and 0.0046 h-1, respectively. According to the above results, higher rate constant of glycerol hydrogenolysis accompanied with higher rate constant of 1,3-propanediol hydrogenolysis were obtained over Ir/Re/SiO2 catalyst. It means that the catalyst which is effective for glycerol hydrogenolysis to 1,3-propanediol would also favor the consecutive hydrogenolysis of 1,3-propanediol. From the results of NH3-TPD and Py-IR, it was found that the ReOx sites on the Ir–Re/SiO2 and Ir/Re/SiO2 catalysts were similar. But there was a very large difference in the number of Ir0 sites obtained from the

123

Reac Kinet Mech Cat

Fig. 7 The plot of –ln(1 - Xgly) - ln(1 - Xgly) versus t and – ln(1 – X1,3-PD) - ln(1 - X1,3-PD) versus t. Filled square catalyst: Ir–Re/SiO2, substrate: glycerol; Filled circle catalyst: Ir/Re/SiO2, substrate: glycerol; filled triangle catalyst: Ir–Re/SiO2, substrate: 1,3-propanediol; filled inverted triangle catalyst: Ir/Re/SiO2, substrate: 1,3-propanediol. Reaction condition: P = 8.0 MPa, T = 403 K, 0.5 g catalyst, 80 wt% substrate aqueous solution,VH2 = 50 ml/min

results of H2-chemisorption between the Ir–Re/SiO2 and Ir/Re/SiO2 catalysts. It suggests that the reaction rates of glycerol and 1,3-propanediol hydrogenolysis were mainly affected by the number of Ir0, and the reaction rates were higher when reacting over Ir/Re/SiO2 catalysts with a larger number of Ir0 sites. Therefore, we think that there was a more appropriate number of Ir0 sites over the Ir–Re/SiO2 catalyst than over the Ir/Re/SiO2 catalyst. An appropriate number of Ir0 supply an appropriate reaction rate of glycerol and 1,3-propanediol hydrogenolysis, and thus a higher selectivity to 1,3-propanediol was obtained.

Conclusions Different impregnation sequences were applied to prepare SiO2 supporting Ir-ReOx catalysts with an iridium loading of 2 wt%, and their catalytic performances were tested in glycerol selective hydrogenolysis to 1,3-propanediol. It was found that the catalyst prepared by co-impregnation of Re and Ir species on support obtained a glycerol conversion of 43.7 % and a 1,3-propanediol selectivity of 49.2 %, and those for catalyst prepared by impregnation of Re prior to Ir were 54.4 and 39.9 %. But the catalyst with the impregnation of Ir prior to Re on the SiO2 support got worse result for glycerol hydrogenolysis without reduction pretreatment, and its activity recovered when Ir species was reduced to metallic. Except for adsorbing glycerol, Re species might act as an efficient promoter to improve the dispersive state of Ir species, and enhanced the reducibility of Ir species. So for catalyst prepared by the impregnation of Ir prior to Re species on the SiO2 support, the particle size of Ir species that formed without the existence of Re species was larger,

123

Reac Kinet Mech Cat

resulting in higher reduction temperature and leading to a poor activity for glycerol conversion without reduction treatment before reaction. Combining with the results of kinetic studies, it was observed that the number of Ir0 sites would affected the hydrogenolysis reaction rates of glycerol and 1,3-propanediol, and an appropriate number of Ir0 sites were important to selectivity to 1,3-propanediol. Acknowledgments This work was financially supported by National the Natural Science Foundation of China (Contract No. 21403217).

References 1. Cardona C, Posada J, Montoya M (2007) Proc Eur Congr Chem Eng September:16–20 2. Zanin CICB, Jorda˜o E, Mandelli D, Figueiredo FCA, Carvalho WA, Oliveira EV (2015) React Kinet Mech Cat 115:293–310 3. Marinoiu A, Cobzaru C, Carcadea E, Capris C, Tanislav V, Raceanu M (2013) React Kinet Mech Cat 110:63–73 4. Li Y, Liu H, Ma L, He D (2014) RSC Adv 4:5503 5. Zhu S, Qiu Y, Zhu Y, Hao S, Zheng H, Li Y (2013) Catal Today 212:120–126 6. Arundhathi R, Mizugaki T, Mitsudome T, Jitsukawa K, Kaneda K (2013) Chem Sus Chem 6:1345–1347 7. Zhu S, Gao X, Zhu Y, Zhu Y, Xiang X, Hu C, Li Y (2013) Appl Catal B: Environ 140–141:60–67 8. ten Dam J, Djanashvili K, Kapteijn F, Hanefeld U (2013) Chem Cat Chem 5:497–505 9. Zhang Y, Zhao X-C, Wang Y, Zhou L, Zhang J, Wang J, Wang A, Zhang T (2013) J Mater Chem A 1:3724–3732 ¨ Y, Ding Y, Lin R, Li J, Dong W, Wang T, Chen W (2009) Chin J Catal 30:1189–1191 10. Gong L, LU 11. Mizugaki T, Yamakawa T, Arundhathi R, Mitsudome T, Jitsukawa K, Kaneda K (2012) Chem Lett 41:1720–1722 12. Liu L, Zhang Y, Wang A, Zhang T (2012) Chin J Catal 33:1257–1261 13. Nakagawa Y, Ning X, Amada Y, Tomishige K (2012) Appl Catal A: Gen 433–434:128–134 14. Nakagawa Y, Mori K, Chen K, Amada Y, Tamura M, Tomishige K (2013) Appl Catal A: Gen 468:418–425 15. Koso S, Watanabe H, Okumura K, Nakagawa Y, Tomishige K (2012) J Phys Chem C 116:3079–3090 16. Amada Y, Watanabe H, Tamura M, Nakagawa Y, Okumura K, Tomishige K (2012) J Phys Chem C 116:23503–23514 17. Deng C, Leng L, Duan X, Zhou J, Zhou X, Yuan W (2015) J Mol Catal A: Chem 410:81–88 18. Deng C, Duan X, Zhou J, Zhou X, Yuan W, Scott SL (2015) Catal Sci Technol 5:1540–1547 19. Nakagawa Y, Shinmi Y, Koso S, Tomishige K (2010) J Catal 272:191–194 20. Riga J, Tenretnoel C, Pireaux JJ, Caudano R, Verbist JJ, Gobillon Y (1977) Phys Scr 16:351–354 21. Campos C, Torres C, Oportus M, Pen˜a MA, Fierro JLG, Reyes P (2013) Catal Today 213:93–100 22. Tsoncheva T, Vankova S, Bozhkov O, Mehandjiev D (2007) Can J Chem 85:118–123 23. Youzhu Yuan Y, Iwasawa Y (2002) J Phys Chem B 106:4441–4449 24. Bai M, Liu Z-H, Zhou L-J, Liu Z-Y, Zhang C-F (2013) Trans Nonferrous Met Soc China 23:538–542 25. Parry EP (1963) J Catal 2:371–379 26. Ciftci A, Peng B, Jentys A, Lercher JA, Hensen EJM (2012) Appl Catal A: Gen 431–432:113–119 27. Sharma RV, Kumar P, Dalai AK (2014) Appl Catal A: Gen 477:147–156 28. Rekha V, Sumana C, Douglas SP, Lingaiah N (2015) Appl Catal A: Gen 491:155–162 29. Tomishige K, Nakagawa Y, Tamura M (2014) In: Nicholas KM (ed) Selective catalysis for renewable feedstocks and chemicals. Springer Int Publishing Ag, Cham, pp 127–162 30. Amada Y, Shinmi Y, Koso S, Kubota T, Nakagawa Y, Tomishige K (2011) Appl Catal B: Environ 105:117–127 31. Xia S, Nie R, Lu X, Wang L, Chen P, Hou Z (2012) J Catal 296:1–11

123