Reac Kinet Mech Cat DOI 10.1007/s11144-017-1267-y

Role of initial water content in glycerol hydrogenolysis to 1,2-propanediol over Cu–ZnO catalyst Miaomiao Hou1 • Hong Jiang1 • Yefei Liu1 • Rizhi Chen1

Received: 15 June 2017 / Accepted: 1 September 2017 Ó Akade´miai Kiado´, Budapest, Hungary 2017

Abstract The work aims to investigate the role of initial water content in the glycerol hydrogenolysis to 1,2-propanediol over Cu–ZnO in depth by characterizing the fresh and spent catalysts in detail with ICP, XRD, XPS, EDS, HRTEM, N2 sorption, N2O chemisorption, H2-TPD and NH3-TPD. The glycerol conversion and 1,2-propanediol selectivity gradually reduce simultaneously with increasing initial water content. At higher initial water content, more CuO are formed, bringing about a decrease in the 1,2-propanediol selectivity. The sizes of Cu and ZnO crystallites increase tremendously with increasing initial water content, lowering the active surface area and acidity. Furthermore, the surface Cu content reduces along with the increase of surface Zn content at higher initial water content. The two aspects make a decrease in the glycerol conversion. This work would aid the glycerol hydrogenolysis to 1,2-propanediol over Cu–ZnO with high catalytic efficiency. Keywords Glycerol  Hydrogenolysis  1,2-propanediol  Cu–ZnO  Initial water content

Electronic supplementary material The online version of this article (doi:10.1007/s11144-017-1267-y) contains supplementary material, which is available to authorized users. & Rizhi Chen [email protected] 1

State Key Laboratory of Materials-Oriented Chemical Engineering, Nanjing Tech University, Nanjing 210009, Jiangsu, China

123

Reac Kinet Mech Cat

Introduction Biodiesel is a new type of clean and environmentally friendly fuel, transformed from triglycerides in vegetable oils and animal fats by transesterification with methanol, making it increasingly significant due to the worldwide concerns over dwindling fossil fuel reserves and the impact of global warming on climate change [1–3]. During the production of biodiesel, a large quantity of glycerol as the low value-added by-product is available [4], lowering the total economical efficiency. Thus, the effective utilization of this crude glycerol by converting into high valueadded chemicals has become highly desirable. To date, many processes have been developed to transform glycerol into important chemicals such as 1,2-propanediol, 1,3-propanediol, acrolein, acetol, alcohol, epichlorohydrin, biopropanols and lactic acid [4, 5]. Among the various possible chemicals from glycerol, 1,2-propanediol (1,2-PDO) is an important product and widely used in cosmetics, pharmaceuticals, polyester resins, de-icing, etc. [3, 6], which can be produced by the selective hydrogenolysis of glycerol. In order to achieve high performance of glycerol hydrogenolysis to 1,2-PDO, various catalysts have been extensively studied, including the noble metals [7–11] and non-noble metal [11–18] catalysts. The noble metal catalysts have high cost and often exhibit lower selectivity to 1,2-PDO because their activity is stronger towards C–C bond cleavage than C–O bond [10, 18]. In contrast, the non-noble metal catalysts like Cu-based catalysts are more competitive because of their low price and high selectivity to 1,2-PDO, attracting considerable interest [15–17]. For instance, Wang et al. [16] investigated the glycerol hydrogenolysis over Cu–ZnO catalysts, and achieved a 1,2-PDO selectivity of 86.3% at a glycerol conversion of 22.5% at 473 K. Meanwhile, a two-step glycerol hydrogenolysis reaction pathway including glycerol dehydration to acetol and glycidol intermediates on acidic ZnO surfaces followed by their hydrogenolysis on Cu surfaces was proposed. Up to date, there have been numerous reports about the hydrogenolysis of glycerol to 1,2-PDO over Cu based catalysts [3, 11, 14–20]. During these investigations, the adopted glycerol concentration was mainly in the range of 10–80 wt% [16–25]. Regrettably, it is still not elaborated how to select the glycerol concentration, namely the initial water content. Dasari et al. [19] discussed the effect of initial water content on the hydrogenolysis of glycerol from the reaction equilibrium. Water is generated in this reaction and it is always preferable to eliminate the water from the initial reaction mixture to drive the equilibrium in the forward direction. Montassier et al. [20] proved that the size of copper particles in a Cu/C catalyst increased if the catalyst was stirred in water at elevated temperatures. Bienholz et al. [14] also found that the presence or formation of water could cause the increase in Cu particle size and led to the deactivation of CuO/ZnO catalyst. These works confirmed that water played an important role in the hydrogenolysis of glycerol over Cu based catalysts. Unfortunately, the relationships between the initial water content and the microstructures of Cu based catalysts and the corresponding performance of glycerol hydrogenolysis are not well established.

123

Reac Kinet Mech Cat

The aim of the work is to explore the role of initial water content on the glycerol hydrogenolysis to 1,2-PDO over Cu–ZnO catalyst prepared in depth by a coprecipitation method. To reach the goal, the fresh (CuO–ZnO), reduced (Cu–ZnO) and spent Cu–ZnO catalysts after reaction at different initial water contents (60 and 90 wt%) were characterized in detail by ICP, XRD, XPS, EDS, HRTEM, N2 adsorption–desorption, N2O chemisorption, H2-TPD and NH3-TPD.

Experimental Chemicals Copper nitrate trihydrate (Cu(NO3)23H2O, AR) was purchased from Shanghai Xinbao Fine Chemical Factory, China. Zinc nitrate hexahydrate (Zn(NO3)26H2O, AR), sodium carbonate (Na2CO3, AR) and glycerol (AR, C 99%) were purchased from XiLong Science Co., Ltd., China. 1,4-butanediol (AR, 98%) was provided by Aladdin Chemical Reagent Co., Ltd., China. Hydrogen (99.99%) was purchased from Jiangsu Tianhong Chemical Co., Ltd., China. Deionized water with an electrical conductivity below 12 lS cm-1 was homemade. Preparation and characterization of Cu–ZnO catalyst The CuO–ZnO catalyst was prepared by a co-precipitation method [26]. Typically, a mixed aqueous solution of Cu(NO3)23H2O and Zn(NO3)26H2O with a fixed molar ratio of Cu2?/Zn2? (1:1.4) was applied as the precursor solution. The solution was heated to 70 °C and maintained for 1 h, where the pH was kept at 8.0 by adding 2 mol/L Na2CO3 with a peristaltic pump. Then, the precipitate was filtrated, washed with deionized water, heated at 110 °C overnight, and calcined at 450 °C for 3 h to yield the CuO–ZnO catalyst. After that, the as-prepared CuO–ZnO catalyst was pulverized into particles with 10–14 mesh to obtain sufficient mechanical strength and size suitable for use in a fixed bed reactor. To measure the total metal elements composition of the catalysts, inductively coupled plasma optical emission spectrometry (ICP-AES, Optima 2000 DV) was performed. The samples were digested using concentrated HNO3 acid prior to ICPAES analysis. Each sample was measured for at least three times to acquire the average value. X-ray diffraction (XRD) patterns were obtained on a Rigaku MiniFlex600 diffractometer (Cu Ka radiation, 40 kV, 15 mA, scanning 2h range: 20°–80°, scanning rate: 20/min). To characterize the surface species on the catalysts, the X-ray photoelectron spectroscopy (XPS) measurements were performed on a Thermo ESCALAB 250 spectrophotometer with a monochromatic Al Ka (ht = 1486.6 eV) radiation source at 15 kV. The C1 s peak (284.8 eV) was used for the calibration of binding energy. The surface metal elements distribution was obtained by energy dispersive X-ray spectrum (EDS, HORIBA). High resolution transmission electron microscopy (HRTEM, JEM 2100) images were adopted to verify the distribution and morphology of CuO, Cu and ZnO particles. The specific surface area and pore volume were obtained by N2 physisorption and

123

Reac Kinet Mech Cat

desorption isotherms at its normal boiling point using a Micromeritics ASAP 2020 analyzer after samples were evacuated (\2.66 Pa) at 393 K for 4 h. Thermogravimetric analysis (TGA) was performed on a NETZSCH STA449F3. In each test, the sample was heated in a N2 flow (30 mL/min) from 30 to 600 °C at 10 °C/min. The total Cu specific surface area, the dispersion of Cu species and the Cu crystallite size were determined by dissociative N2O chemisorption and hydrogen temperatureprogrammed reduction (H2-TPR) methods using a Micromeritics AutoChem 2920 equipment. Typically, the pre-reduction of CuO in the catalysts (0.1 g) to Cu0 was carried out with 5% H2/Ar (20 mL/min) at 300 °C (10 °C/min) for 2 h, then the sample was cooled to 80 °C in the Ar atmosphere. After that, the surface Cu atoms were oxidized to Cu2O with 5% N2O/He at 80 °C for 30 min, and the sample was cooled to room temperature by Ar purge. Finally, the H2-TPR was applied again for reducing Cu2O to Cu0. Catalyst capacity to activate hydrogen molecular was evaluated by temperature-programmed desorption of hydrogen (H2-TPD) using a Micromeritics AutoChem 2920 equipment as follows. 0.1 g of the sample was added into a U-shape quartz reactor, degasified in the Ar atmosphere at 350 °C for 20 min, then the gas was switched to 10% H2/Ar (20 mL/min) for 2 h. After that, the sample was cooled to 50 °C by Ar purge, and adsorbed in the 10% H2/Ar atmosphere for 1 h. Followed by switching the gas to Ar for 2 h, the physically adsorbed H2 was removed. Finally, the temperature was increased to 600 °C (10 °C/ min) for desorption. Temperature-programmed desorption of ammonia (NH3-TPD) was carried out with the same procedure as H2-TPD to determine the acidities of the catalysts. Measurement of dissolved oxygen A galvanic DO sensor (PCD650, Thermo Fisher Scientific Inc., USA) was adopted to measure the concentration of dissolved oxygen in pure water and aqueous glycerol solutions with different initial water contents (60–90 wt%). Glycerol hydrogenolysis The hydrogenolysis of glycerol over Cu–ZnO (35 mL, 10–14 mesh) was carried out in a fixed-bed stainless steel reactor consisting of a stainless steel tube with an internal diameter of 20 mm and a coaxially centered thermocouple with its tip located in the middle of the bed using three-stage temperature control. Prior to the hydrogenolysis experiment, the CuO–ZnO catalyst was reduced in a H2 flow (300 mL/min) at 250 °C for 3 h under atmospheric pressure, and then the reactor was cooled to room temperature in the N2 atmosphere; named as Cu–ZnO. After that, the aqueous glycerol solution with a certain initial water content (35 mL/h) was continuously pumped into the bottom side of the fixed bed reactor by a doubleplunger micropump (Szweico 2ZB-2L20A, Beijing Spacecrafts, China) and the hydrogen (233 mL/min, 4 MPa) was delivered into reactor by a mass flow controller (D08-1F, Beijing Sevenstar Huachuang Electronics Co., Ltd., China), while the reactor was heated to 200 °C with a heating rate of 2 °C/min. As the temperature reached the target value, the reaction system was considered to be

123

Reac Kinet Mech Cat

steady-going, and the hydrogenolysis was performed for 10 h. In each test, the initial water content of the feedstock was constant during the whole hydrogenolysis process. For exploring the role of initial water content in detail, the glycerol hydrogenolysis at each fixed initial water content was replicated for three times, and the average glycerol conversion and 1,2-PDO selectivity were obtained. The liquid products were collected every 2 h, and analyzed with 1,4-butanediol as an internal standard using a gas chromatograph (Shimadzu 2014) equipped with a flame ionization detector (FID) by separating them on Inno wax capillary column (EC-1, 30 m 9 0.53 mm 9 1.2 lm). 1,2-PDO was the target product of glycerol hydrogenolysis, and acetol, n-propanol, ethylene glycol and methanol were the by-products. The spent Cu–ZnO catalyst is marked as Cu–ZnO–S–x, where x represents the initial water content. The glycerol conversion and product selectivity were calculated on the basis of the following equations [27]. Conversion ð%Þ ¼ Selectivity ð%Þ ¼

Moles of glycerol converted  100 Moles of glycerol initially ch arg ed

Moles of carbon in specific product  100 Total moles of carbon precent in all products

Results and discussion Effect of initial water content on glycerol hydrogenolysis

Fig. 1 Effect of initial water content on the glycerol conversion and 1,2-propanediol/ ethylene glycol selectivity in the glycerol hydrogenolysis over Cu–ZnO catalyst. Reaction conditions: reaction temperature 200 °C, hydrogen pressure 4.0 MPa, H2/glycerol solution 400 (volume ratio), LHSV 1.0 h-1, catalyst loading volume 35 mL

Conversion or Selectivity (%)

The glycerol hydrogenolysis over Cu–ZnO catalyst was carried out at four different initial water contents to investigate the role of initial water content. Fig. 1 shows the changes of the glycerol conversion and 1,2-propanediol selectivity with the initial water content, and the ethylene glycol selectivity is given for comparison. Obviously, with increasing initial water content from 60 to 90%, the glycerol conversion and selectivity of 1,2-PDO decrease from 80.3 to 74.8 and 93.7 to 100

80

Glycerol conversion 1,2-Propanediol selectivity Ethylene glycol selectivity

12 8 4 0 60

65

70

75

80

85

90

Initial water content (%)

123

Reac Kinet Mech Cat

87.9%, respectively. However, the selectivity of ethylene glycol increases from 5 to 10.3%. These results highlight that the initial water content significantly affects the glycerol hydrogenolysis over Cu–ZnO catalyst: the higher the initial water content the worse the hydrogenolysis performance. Similar results were also reported. Wolosiak-Hnat et al. [28] studied the effect of glycerol concentration on the hydrogenolysis of glycerol over a Cu/Al2O3 catalyst, and found that higher initial water content was not in favor of the glycerol hydrogenolysis. But they did not investigate the reasons in depth, especially for the decrease in 1,2-PDO selectivity. Study on the effect of initial water content by catalyst characterization To evaluate the role of initial water content in the glycerol hydrogenolysis over Cu– ZnO, the fresh and spent catalysts were characterized in detail. As shown in Fig. 1, as the initial water content increases from 60 to 90%, the glycerol conversion and 1,2-PDO selectivity almost linearly decrease. Thus, the evolution of catalyst microstructure with the increase in initial water content may have the similar trend. Compared to the initial water contents of 70 and 80%, as the initial water contents are 60 and 90%, the glycerol conversion and 1,2-PDO selectivity change more obviously, and then the evolution of catalyst microstructure can be more distinct. Hence, in the present work, only the two typical spent catalysts (Cu–ZnO–S–60 and Cu–ZnO–S–90) were selected for investigation. First of all, ICP-AES was carried out to evaluate the leaching of Cu and Zn. Table 1 gives the contents of Cu and Zn metal elements in different catalysts. As compared to the fresh catalyst, the total composition of Cu–ZnO catalyst does not change significantly after the reaction, irrespective of the initial water content. These results indicate that the decrease in hydrogenolysis performance with increasing initial water content is not due to the leaching of Cu and Zn. XRD measurements were performed to assess the influence of initial water content on the crystalline structure of Cu–ZnO catalyst. Fig. 2 shows the XRD patterns of CuO–ZnO, Cu–ZnO, Cu–ZnO–S–60 and Cu–ZnO–S–90 catalysts. The diffraction peaks ascribed to CuO and ZnO are observed in the fresh CuO–ZnO catalyst, and the peak at 2h = 26.7° may be assigned to the additive for the catalyst molding [29]. The CuO crystallites diffraction peak disappears after H2 reduction, while the characteristic peaks of Cu crystallites can be clearly observed (Fig. 2b), which indicates that CuO has been completely reduced to metallic Cu through H2 reduction, providing active centers for hydrogenolysis of glycerol. However, the Table 1 Element analysis of the fresh and spent catalysts after the glycerol hydrogenolysis at different initial water contents calculated by ICP Catalysts

Cu (wt%)

Zn (wt%)

CuO–ZnO

23.39 ± 1.21

25.29 ± 1.13

Cu–ZnO–S–60

23.71 ± 0.98

24.69 ± 1.05

Cu–ZnO–S–90

23.87 ± 1.15

24.16 ± 1.09

123

Reac Kinet Mech Cat

SiO2 CuO Cu ZnO

Intensity (a.u.)

Fig. 2 XRD patterns of the fresh, reduced and spent catalysts after the glycerol hydrogenolysis at different initial water contents: a CuO– ZnO, b Cu–ZnO, c Cu–ZnO–S– 60 and d Cu–ZnO–S–90

(d) (c) (b) (a) 20

30

40

50

60

70

80

2 Theta (degree)

CuO crystallites diffraction peak [30] appears again in the spent catalysts (Figs. 2c and 2d). Moreover, the intensity of CuO characteristic peak increases with increasing initial content water. A possible explanation is as follows. The active center Cu0 may react with the oxygen in water to form Cu2?, and at higher initial water content, the CuO is more easily formed due to more oxygen. To confirm this, the amounts of dissolved oxygen in pure water and aqueous glycerol solutions with different initial water contents, i.e., 90, 80, 80 and 60 wt%, were measured by a galvanic DO sensor. The corresponding dissolved oxygen concentrations are 6.85, 5.88, 5.78, 5.70 and 5.61 mg/L, indicating that the oxygen content in the aqueous glycerol solution increases with increasing initial water content, leading to the formation of more CuO. Furthermore, Cu0 is favorable for hydrogenation reaction, while Cu2? is favorable for retro-aldol reaction [20]. Ethylene glycol is produced easily by the retro-aldol reaction [31]. Therefore, the amount of produced ethylene glycol increases with increasing initial water content, leading to a decrease in the 1,2-PDO selectivity and an increase in the ethylene glycol selectivity (Fig. 1). It is also seen from Fig. 2 that the intensities of Cu and ZnO crystallites diffraction peaks increase after the reaction, suggesting the increase in the sizes of Cu and ZnO during the glycerol hydrogenolysis. The sizes of CuO, Cu and ZnO grains in the catalysts were calculated by the Scherrer formula, as shown in Table 2. It is evident that the size of Cu or ZnO increases and the trend is more obvious at higher initial water Table 2 Crystallite size of the fresh, reduced and spent catalysts after the glycerol hydrogenolysis at different initial water contents calculated from XRD diffractions Catalysts CuO–ZnO

CuO/Cu (nm)

ZnO (nm)

6.1

6.5

Cu–ZnO

10.3

14.8

Cu–ZnO–S–60

15.7

16.4

Cu–ZnO–S–90

17.5

17.6

123

Reac Kinet Mech Cat

(c)

Zn 2p Zn 2p

Cu 2p

Zn LMM1 O 1s Cu LM2

C 1s

Zn 3p Zn 3s

Zn 3d

Fig. 3 XPS survey spectra of the fresh and spent catalysts after the glycerol hydrogenolysis at different initial water contents: a CuO–ZnO, b Cu– ZnO–S–60 and c Cu–ZnO–S–90

Intensity (a.u.)

content. For example, the size of Cu increases from 10.3 to 15.7 nm and then to 17.5 nm as the initial water content changes from 60 to 90 wt%. These results indicate the poor dispersion and accumulation of Cu and ZnO crystallites [32]. According to the reaction mechanism [16], the dispersion of Cu and ZnO would have a crucial effect on the glycerol conversion, and better dispersion is beneficial for the reaction. Thus, the glycerol conversion will reduce with increasing initial water content due to the increase in the sizes of Cu and ZnO. In order to further verify the presence of Cu2? in the spent Cu–ZnO catalysts, the XPS was performed to characterize the surface species on the CuO–ZnO, Cu–ZnO– S–60 and Cu–ZnO–S–90 catalysts. Fig. 3 shows the XPS total energy spectra of the three catalysts. All the samples mainly contain Cu, Zn and O elements. As shown in Fig. 4, except for the Cu peaks (932.4 eV), clear CuO peaks (934.0 eV) can be observed in the two spent Cu–ZnO catalysts [33], confirming the existence of Cu2?. As compared to Cu–ZnO–S–60, the peak intensity of Cu2p for Cu–ZnO–S–90 is significantly lower, indicating that the Cu content on the catalyst surface reduces with increasing initial water content (Table 3). The Cu2p XPS peaks of Cu–ZnO–S– 60 and Cu–ZnO–S–90 can be well fitted to Cu and CuO peaks (Fig. 5). Based on the peak area, the calculated CuO percentages are about 29.9 and 49.7% for Cu–ZnO– S–60 and Cu–ZnO–S–90, respectively. Therefore, with the increase of initial water content, the Cu2? content also increases in consistent with the XRD analysis (Fig. 2), which would be responsible for the decrease in the 1,2-PDO selectivity. Fig. 6 presents the Zn2p XPS energy spectra of the three catalysts. In comparison with the fresh catalyst, the peak intensity of the spent catalyst increases, and the trend is more obvious at higher initial water content, indicating that the content of Zn on the catalyst surface rises with increasing initial water content (Table 3). It is possible that the microstructure of the Cu–ZnO catalyst can evolve during the glycerol hydrogenolysis, and parts of Cu are covered by the ZnO crystals, and the change is more obvious as the initial water content increases, resulting in a decrease in the glycerol conversion.

(b)

(a) 0

200

400

600

Binding energy (eV)

123

800

1000

Reac Kinet Mech Cat Fig. 4 Cu2p XPS spectra of the fresh and spent catalysts after the glycerol hydrogenolysis at different initial water contents: a CuO–ZnO, b Cu–ZnO–S–60 and c Cu–ZnO–S–90

Intensity (a.u.)

(c)

(b)

(a)

930

940

950

960

970

Binding energy (eV)

Table 3 Atomic concentration of different elements on the surface of the fresh and spent catalysts after the glycerol hydrogenolysis at different initial water contents obtained by XPS Catalysts

Atomic concentration (%) Cu2p

CuO–ZnO

Zn2p

O1s

C1s

10.74

12.56

47.52

29.18

Cu–ZnO–S–60

8.75

18.52

41.59

31.14

Cu–ZnO–S–90

6.28

54.47

24.51

14.74

Fig. 5 Cu2p XPS peaks split spectra of the spent catalysts after the glycerol hydrogenolysis at different initial water contents: a Cu–ZnO–S–60 and b Cu–ZnO–S–90

The EDS mapping images were obtained by FESEM in combination with EDS to further investigate the distribution of elements on the catalyst surface, as presented in Fig. S1. The distribution of Cu and Zn elements on the surfaces of the fresh and reduced catalysts are more homogeneous. In contrast, the distribution of Cu and Zn elements in the spent catalysts become worse, and some Cu are covered by Zn as

123

Reac Kinet Mech Cat

Intensity (a.u.)

Fig. 6 Zn2p XPS spectra of the fresh and spent catalysts after the glycerol hydrogenolysis at different initial water contents: a CuO–ZnO, b Cu–ZnO–S–60 and c Cu–ZnO–S–90

(c)

(b) (a) 1020

1030

1040

1050

Binding energy (eV)

indicated by the panes in Figs. S1h–S1l. Table 4 lists the molar percentages of Cu and Zn elements on the four catalyst surfaces from the EDS mapping analyses. From the data we can see that the surface composition almost keeps stable during the H2 reduction. However, after the reaction, the surface Cu content reduces along with the increase of surface Zn content, and the change becomes more obvious with the increase of initial water content. These results are consistent with the XPS analyses (Figs. 4 and 6; Table 3). The structure and morphology of CuO–ZnO, Cu–ZnO, Cu–ZnO–S–60 and Cu– ZnO–S–90 catalysts have been characterized in more detail by HRTEM, as shown in Fig. S2. The fresh and reduced catalysts are composed of nearly uniform spherical grains (Figs. S2a and S2d). We performed selected area electron diffraction (SAED) studies (Figs. S2c and S2f), the regular diffraction rings are obtained especially for the CuO–ZnO, further proving the spherical particle structure for the two samples [34]. Lamellar, rod-like structure and spherical particles are presented in the Cu–ZnO–S–60 catalyst (Fig. S2g). Cu nanoparticles with an average particle size about 9 nm are dispersed on the ZnO with lamellar and rod-like structures (Fig. S2h). It can be seen from the SAED (Fig. S2i), in addition to the nanoparticles expression by the regular annular structure, there are also scattered points diffracted by the lamellar and rod-like structures [35]. More rodlike structures with larger size are presented in the Cu–ZnO–S–90 catalyst, where Table 4 Cu and Zn elements molar percentage of the fresh, reduced and spent catalysts after the glycerol hydrogenolysis at different initial water contents obtained by EDS mapping analysis Catalysts

Cu

Zn

Cu/Zn

CuO–ZnO

47.75

52.25

91.39

Cu–ZnO

47.95

52.08

92.07

Cu–ZnO–S–60

46.06

53.94

85.38

Cu–ZnO–S–90

44.38

55.62

79.79

123

Reac Kinet Mech Cat

the Cu nanoparticles with an average diameter of about 20 nm are dispersed. The results indicate that the Cu nanoparticles easily aggregate and the ZnO tends to form rods with larger size with increasing initial water content, in accordance with the XRD results in Fig. 2. These changes should take into account the loss of catalytic activity and glycerol conversion. To assess the influence of initial water content on the specific surface area and pore volume of Cu–ZnO catalyst, nitrogen sorption measurements were carried out. Fig. 7 gives the nitrogen adsorption–desorption isotherms of CuO–ZnO, Cu–ZnO, Cu–ZnO–S–60 and Cu–ZnO–S–90 catalysts. All the samples exhibit Langmuir type IV isotherms. At low pressure, the overlapped adsorption–desorption curves are observed. Simultaneously, at high pressure, obvious hysteresis loops appear especially for the fresh Cu–ZnO catalyst, which is the characteristic of mesoporous and macroporous materials [36, 37]. However, compared to the fresh catalyst, the hysteresis loops for the spent catalysts become narrower, indicating the smaller pore sizes [38]. Table 5 lists the specific surface areas and pore volumes of all the catalysts. It can be seen that the specific surface area and pore volume of the catalyst increase by H2 reduction, indicating that the hydrogen reduction process may make the catalyst dispersion better. However, after the glycerol hydrogenolysis, the specific surface area and pore volume significantly reduce, especially for the hydrogenolysis at higher initial water content. These results suggest that the Cu– ZnO crystals tend to aggregate with increasing initial water content in agreement with XRD and HRTEM characterizations (Figs. 2 and S2), leading to a reduction in specific surface area and active sites, then to a loss of catalytic activity and glycerol conversion. In addition, in the present work, after the glycerol hydrogenolysis, the spent Cu–ZnO catalyst was directly used for the N2 adsorption–desorption analysis without especial treatment. Thus, the adsorption of organics on the Cu–ZnO catalyst during the glycerol hydrogenolysis (Fig. S3) may also be a reason for the reduced specific surface area and pore volume. The dispersion, specific surface area and crystallite size of Cu were further determined by means of N2O chemisorption [11], as shown in Table 6. This is an important parameter for the evaluation of Cu-based catalysts [18]. Compared to the Cu–ZnO–S–60 catalyst, for the Cu–ZnO–S–90 catalyst, both the dispersion and 120

Quantity sorbed (cm3/g STP)

Fig. 7 Adsorption-desorption isotherms of the fresh, reduced and spent catalysts after the glycerol hydrogenolysis at different initial water contents: a CuO–ZnO, b Cu–ZnO, c Cu– ZnO–S–60 and d Cu–ZnO–S–90

100 80

(b) 60 40 20

(a) (c)

0

(d) 0.0

0.2

0.4

0.6

0.8

1.0

P/P0

123

Reac Kinet Mech Cat Table 5 Specific surface area and pore volume of the fresh, reduced and spent catalysts after the glycerol hydrogenolysis at different initial water contents BET surface area (m2/g)

Catalysts

Pore volume (cm3/g)

CuO–ZnO

54.78

0.151

Cu–ZnO

64.92

0.174

Cu–ZnO–S–60

46.41

0.132

Cu–ZnO–S–90

33.77

0.102

Table 6 Cu dispersion, specific surface area and crystallite size of the spent catalysts after the glycerol hydrogenolysis at different initial water contents measured by N2O chemisorption Catalysts

Dispersion (%)

Specific surface area (m2/g)

Cu crystallite size (nm)

Cu–ZnO–S–60

2.2

5.08

15.9

Cu–ZnO–S–90

1.7

3.95

20.4

specific surface area of Cu reduce, and the average particle size increases from 15.9 to 20.4 nm. It is further certified that the size of Cu nanoparticles will increase at higher initial water content, leading to the decrease in the active specific surface area. It is expected as a result of the XRD and HRTEM analyses (Figs. 2 and S2), and further confirms the causes for the reduction of glycerol conversion. The hydrogen activation and spillover capacity can be represented by the adsorption amount of hydrogen, which can reflect the catalytic activity of metal catalysts [39]. Hence, H2-TPD measurements were performed to evaluate the hydrogen adsorption capacity of Cu–ZnO catalyst. Fig. 8 shows the H2-TPD curves of the Cu–ZnO–S–60 and Cu–ZnO–S–90 catalysts. For each sample, two H2 desorption peaks occur between 250 to 350 °C and 350 to 550 °C, which indicates that there are two different active sites for hydrogen adsorption on the Cu–ZnO Fig. 8 H2-TPD diagrams of the spent catalysts after the glycerol hydrogenolysis at different initial water contents: a Cu– ZnO–S–60 and b Cu–ZnO–S–90

Intensity (a.u.)

(b)

(a)

100

200

300

400

Temperature (°C)

123

500

600

Fig. 9 NH3-TPD diagrams of the spent catalysts after the glycerol hydrogenolysis at different initial water contents: a Cu–ZnO–S–60 and b Cu– ZnO–S–90

Intensity (a.u.)

Reac Kinet Mech Cat

(b)

(a)

200

400

600

800

Temperature (°C)

surface [40]. It can be seen from the peak temperature, both active sites strongly adsorb hydrogen [41]. Clearly, the amount of hydrogen desorption for both sample is significantly different, and the Cu–ZnO–S–90 catalyst has lower hydrogen desorption as compared to Cu–ZnO–S–60. The results indicate that the hydrogen adsorption capacity of Cu–ZnO catalyst decreases with increasing initial water content. As presented in Fig. 2 and Table 2, the Cu size rises with increasing initial water content, resulting in few specific surface areas for the adsorption of hydrogen (Table 6), thereby lower hydrogen adsorption capacity [42]. It has been reported that the amount of adsorbed hydrogen can reflect the number of active centers [41]. Thus, at higher initial water content, the number of active centers reduces, thereby the reduction in glycerol hydrogenation ability. Acidic catalysts have an excellent effect for hydrogenolysis of glycerol [43, 44]. Thus the acidities of Cu–ZnO catalysts were detected by NH3-TPD, as shown in Fig. 9. Two NH3 desorption peaks are observed in the range of 350–550 and 550 to 700 °C, indicating that both catalysts have a moderate strength and a strong acid sites [44, 45]. By comparing the peak area, we can conclude that the acid content of Cu–ZnO–S–90 catalyst is fewer than Cu–ZnO–S–60 because of the increase in the ZnO size (Figs. 2 and S2). According to the reaction mechanism of glycerol hydrogenolysis [16], glycerol is dehydrated to acetol on ZnO acidic sites followed by the hydrogenation on Cu active sites. Therefore, at higher initial water content, the acid content of ZnO will reduce, leading to the decreased glycerol conversion.

Conclusions In this study, the effect mechanism of initial water content in the hydrogenolysis of glycerol to 1,2-propanediol over a Cu–ZnO catalyst was explored through the elaborative characterization of the fresh and spent catalysts. The initial water content has a significant influence on the glycerol hydrogenolysis. The active center Cu0 reacts with water to form Cu2?, so as the water content increases, the Cu0 on the

123

Reac Kinet Mech Cat

catalyst surface is more easily oxidized. The presence of Cu2? makes it easier to produce ethylene glycol, resulting in a decrease in the selectivity of 1,2-propanediol. At higher initial water content, the Cu and ZnO crystals tend to aggregate, and parts of Cu are covered by ZnO crystals, leading to a reduction in active sites and acidity, then to a loss of catalytic activity and glycerol conversion. In summary, lower initial water content, i.e. higher glycerol concentration, is beneficial for the glycerol hydrogenolysis over Cu–ZnO catalyst with higher reaction performance. This work would provide some references for the selection of reaction conditions of glycerol hydrogenolysis, the preparation of high-performance catalysts and better understanding the hydrogenolysis process. Acknowledgements The financial supports from the National Natural Science Foundation (91534110, 21606124), the National Key R&D Program (2016YFB0301503), the Jiangsu Natural Science Foundation for Distinguished Young Scholars (BK20150044), the Natural Science Foundation of Jiangsu Province (BK20160978) and the Foundation from State Key Laboratory of Materials-Oriented Chemical Engineering (ZK201402, ZK201407) of China are gratefully acknowledged.

References 1. Pagliaro M, Ciriminna R, Kimura H, Rossi M, Della Pina C (2007) Angew Chem Int Edit 46:4434–4440 2. Foo GS, Wei D, Sholl DS, Sievers C (2014) ACS Catal 4:3180–3192 3. Rajkhowa T, Marin GB, Thybaut JW (2017) Appl Catal B-Environ 205:469–480 4. Peng GM, Wang XC, Chen XF, Jiang YJ, Mu XD (2014) J Ind Eng Chem 20:2641–2645 5. Zhou CHC, Beltramini JN, Fan YX, Lu GQM (2008) Chem Soc Rev 37:527–549 6. Long W, Hao F, Xiong W, Liu PL, Luo HA (2017) Modified sepiolite supported nickel and tungsten oxide catalysts for glycerol hydrogenolysis to 1,2-propanediol with high selectivity under mild conditions. Reac Kinet Mech Cat. doi:10.1007/s11144-017-1194-y 7. Li YM, Liu HM, Ma L, He DH (2014) RSC Adv 4:5503–5512 8. Jin SH, Xiao ZH, Li C, Williams CT, Liang CH (2014) J Energy Chem 23:185–192 9. D’Hondt E, de Vyver SV, Sels BF, Jacobs PA (2008) Chem Commun 45:6011–6012 10. Gallegos-Suarez E, Guerrero-Ruiz A, Rodriguez-Ramos I, Arcoya A (2015) Chem Eng J 262:326–333 11. Xia SX, Yuan ZL, Wang LN, Chen P, Hou ZY (2011) Appl Catal A-Gen 403:173–182 12. Florez-Rodriguez PP, Pamphile-Adria´n AJ, Passos FB (2014) Catal Today 237:38–46 13. Perosa A, Tundo P (2005) Ind Eng Chem Res 44:8535–8537 14. Bienholz A, Schwab F, Claus P (2010) Green Chem 12:290–295 15. Niu L, Wei RP, Li C, Gao LJ, Zhou MH, Jiang F, Xiao GM (2015) Reac Kinet Mech Cat 115:377–388 16. Wang S, Liu HC (2007) Catal Lett 117:62–67 17. Geng GL, Wei RP, Liang T, Zhou MH, Xiao GM (2016) Reac Kinet Mech Cat 117:239–251 18. Du Y, Wang CC, Jiang H, Chen CL, Chen RZ (2016) J Ind Eng Chem 35:262–267 19. Dasari MA, Kiatsimkul PP, Sutterlin WR, Suppes GJ (2005) Appl Catal A-Gen 281:225–231 20. Montassier C, Dumas JM, Granger P, Barbier J (1995) Appl Catal A-Gen 121:231–244 21. Pudi SM, Biswas P, Kumar S (2016) J Chem Technol Biot 91:2063–2075 22. Gao Q, Xu BL, Tong Q, Fan YN (2016) Biosci Biotech Bioch 80:215–220 23. Montassier C, Me´ne´zo JC, Hoang LC, Renaud C, Barbier J (1991) J Mol Catal 70:99–110 24. Schmidt SR, Tanielyan SK, Marin N, Alvez G, Augustine RL (2010) Top Catal 53:1214–1216 25. Lee M, Hwang YK, Chang JS, Chae HJ, Hwang DW (2016) Catal Commun 84:5–10 26. Wang CC, Jiang H, Chen CL, Chen RZ, Xing WH (2015) J Membr Sci 489:135–143 27. Akiyama M, Sato S, Takahashi R, Inui K, Yokota M (2009) Appl Catal A-Gen 371:60–66 28. Wolosiak-Hnat A, Milchert E, Grzmil B (2013) Chem Eng Technol 36:411–418 ¨ vec¸og˘lu ML (2017) Int J Refract Meth 65:19–24 29. Ovalı D, Ag˘aog˘ulları D, O

123

Reac Kinet Mech Cat 30. Zelazny A, Samson K, Grabowski R, Sliwa M, Ruggiero-Mikolajczyk M, Kornas A (2017) Reac Kinet Mech Cat 121:329–343 31. Maris EP, Davis RJ (2007) J Catal 249:328–337 32. Rekha V, Raju N, Sumana C, Douglas SP, Lingaiah N (2016) Catal Lett 146:1487–1496 33. Kim ND, Oh S, Joo JB, Jung KS, Yi J (2010) Korean J Chem Eng 27:431–434 34. Rogal Ł, Korpala G, Dutkiewicz J (2015) Mat Sci Eng A-Struc 624:291–299 35. de la Parra-Arciniega SM, Garcia-Gomez NA, Garza-Tovar LL, Garcı´a-Gutie´rrez DI, Sa´nchez EM (2017) Ultrason Sonochem 36:95–100 36. Ye XY, Xiao XQ, Zheng C, Hua NB, Huang YY (2015) Mater Lett 152:237–239 37. Chen MZ, Liang C, Zhang F, Li HX (2014) ACS Sustain Chem Eng 2:486–492 38. Yang HW, Tang DL, Lu XN, Yuan YZ (2009) J Phys Chem C 113:8186–8193 39. Xia SX, Nie RF, Lu XY, Wang LN, Chen P, Hou ZY (2012) J Catal 296:1–11 40. Fan YJ, Wu SF (2016) J CO2 Util 16:150–156 41. Li L, Mao DS, Yu J, Guo XM (2015) J Power Sources 279:394–404 42. Iriowen E, Orefuwa S, Goudy A (2015) J Alloy Compd 645:S242–S246 43. Liao XY, Li KW, Xiang XM, Wang SG, She XC, Zhu YL, Li YW (2012) J Ind Eng Chem 18:818–821 44. Nie RF, Lei H, Pan SY, Wang LN, Fei JH, Hou ZY (2012) Fuel 96:419–425 45. Li TZ, Fu C, Qi JS, Pan J, Chen SH, Lin JJ (2013) Reac Kinet Mech Cat 109:117–131

123