Journal of Catalysis 357 (2018) 223–237
Contents lists available at ScienceDirect
Journal of Catalysis journal homepage: www.elsevier.com/locate/jcat
Hydrogenation of diesters on copper catalyst anchored on ordered hierarchical porous silica: Pore size effect Yujun Zhao a,c, Ziyuan Guo a, Haojie Zhang b, Bo Peng c, Yuxi Xu a, Yue Wang a, Jian Zhang a, Yan Xu a, Shengping Wang a, Xinbin Ma a,⇑ a Key Laboratory for Green Chemical Technology of Ministry of Education, Collaborative Innovation Center of Chemical Science and Engineering, School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China b School of Science, Tianjin University, Tianjin 300072, China c Institute for Integrated Catalysis, Pacific Northwest National Laboratory, Richland, WA 99352, USA
a r t i c l e
i n f o
Article history: Received 6 July 2017 Revised 2 November 2017 Accepted 5 November 2017
Keywords: Ester Hydrogenation Hierarchical Copper Diffusion
a b s t r a c t An ordered hierarchical porous silica (HPS) presenting both mesopores and micropores was fabricated and used for supporting copper catalysts (Cu/HPS). In the prepared Cu/HPS catalysts, high dispersion of both Cu0 and Cu+ species together with ordered porous structure were achieved. The anchoring effect of micropores hindered the agglomeration of copper species, while the formation of Cu-O-Si species, derived from the strong interaction between surface silica and copper precursor, was prompted by the ammonia evaporation approach. The Cu/HPS catalysts were tested in the hydrogenation of diesters to diols. It shows excellent activity on dimethyl adipate hydrogenation with a 1,6-hexanediol space-time yield of 0.72 g/(g∙h) under WHSV = 1.2 h1. It was further investigated that the activity in dimethyl adipate hydrogenation, in contrast to that of dimethyl oxalate hydrogenation, would be readily limited by pore diffusion, due to the porous structure of HPS. Ó 2017 Elsevier Inc. All rights reserved.
1. Introduction Diols, such as ethylene glycol (EG), 1,6-hexanediol (HDO), are value-added intermediates in the chemical industry. They have been applied in the synthesis of several polymers and fine chemicals, such as elastomers of polyurethane, adhesives, polyesters, plasticizers and pharmaceuticals [1–5]. Currently, HDO is mainly produced by hydrogenating carboxylic acids or their esters on industrial scale. In order to avoid the use of carboxylic acids which causes severe corrosion of the reactor and consequent catalyst deactivation, hydrogenation of dimethyl adipate (DMA) towards HDO attracted increasing interests from both fundamental research and industrial application. In a typical scaled-up operation, DMA hydrogenation takes place in the liquid phase, catalyzed by noble metals (Ru, Pd, Pt and Rh), facilitating the activation of C@O bond [1–4]. In addition to noble-metal catalysts, it has been further demonstrated that Co and Sn promote the reduction of ruthenium oxide to Ru(0) as the active species. The as-prepared Ru-Sn-Co/AlO(OH) catalyst showed excellent performance in the
⇑ Corresponding author. E-mail address:
[email protected] (X. Ma). https://doi.org/10.1016/j.jcat.2017.11.006 0021-9517/Ó 2017 Elsevier Inc. All rights reserved.
hydrogenation of DMA with a conversion of 98% and the selectivity of HDO as high as 95% at 493 K under a hydrogen pressure of 5 MPa [6]. However, the high cost of noble metal catalysts, as well as the technical difficulties involved in liquid phase operation, limited the further development of above-mentioned process. Therefore, considerable efforts aiming for the substitution of noble metals with copper-based catalyst on catalytic long-chain alcohol formation have also been reported [7]. The two major bottlenecks are, (1) the severe conditions (T = 523–623 K and P (H2) = 10–20 MPa) and the low yield of alcohols; (2) the use of toxic Cr as a promoter in catalysis. Therefore, the development of highly active Cr-free catalysts in gas-phase DMA hydrogenation is of great significance. Yuan et al. [8] reported a Cu-Zn-Al catalyst for the hydrogenation of DMA and a 1,6-hexanediol yield higher than 95% was achieved at WHSV of 0.5 h1. They also proposed that Cu/Zn ratio played an important role in maintaining high activity. However, the structure effect of the catalyst on the hydrogenation of DMA was not yet revealed and the catalytic activity deserved a further improvement from an industrial point of view. Examining other reaction systems involving C@O activation, the synthesis of EG via dimethyl oxalate (DMO) hydrogenation by using copper-based catalysts has attracted much attentions for the achievement of commercial scale [9–14]. As depicted in
224
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
Notation SBET Vmeso Vmicro D DCu DispCu SCu(0) XCu(I) SCu(I) CDMA SHDO CDMO SEG rsite Nw-p Cags as kg,i Sc DB,i Re JD
BET surface area (m2/g) mesopore volume (cm3/g) micopore volume (cm3/g) average pore diameter (nm) copper particle diameter (nm) dispersion of metallic copper in the catalyst (%) specific surface area of Cu0 (m2/g) the ratio of Cu+/(Cu+ + Cu0) (%) specific surface area of Cu+ (m2/g) DMA conversion (%) HDO selectivity (%) DMO conversion (%) EG selectivity (%) the conversion rate of the diesters normalized by reaction time and number of active sites (h1) Weisz-Prater criterion Carberry number the external surface area of the catalysts (m2 L1) the gas-solid diffusion coefficient of the catalysts 1 (m3g m2 ) S s Schmidt number the bulk diffusivity of specie i (m2 s1) Reynolds numbers Colburn J factor
Scheme 1, the hydrogenation of both DMO and DMA involves similar reaction sequences. The hydrogenation of DMO first generates methyl glycolate (MG), which can be subsequently hydrogenated to EG and ethanol via two-step hydrogenation. Similarly, the hydrogenation of DMA forms methyl 6-hydroxyhexanoate (MHH), HDO and 1-hexanol as the main products. Normally, it is proposed that there are two different active sites responsible for DMO hydrogenation, namely Cu0 and Cu+. The former mainly contributes to H2 activation, while the latter acts as Lewis acid site to activate C@O bond in carbonyl group [15]. In our previous work on Cu-MCM-41 catalyzed DMO hydrogenation, CuAOASi unit, generated from interaction between Cu and the support MCM-41, is demonstrated for its crucial roles on the formation and dispersion of Cu species, resulting in its high catalytic activity [16]. Copper phyllosilicate generated during the preparation of Cu/SiO2 was also reported to contribute to the formation of Cu+ species and improve the copper species dispersion [9,17].
u robs Deff,i R C⁄i T Vi Dp,i Dk.i dp Di,j Mi yi P I h
l qg e s
the linear velocity of the gas in catalyst bed (m s1) apparent reaction rate (mol L1 s1) effective diffusivity in the pores of the catalyst (m2 s1) the radius of the catalyst pellet (m) concentration of specie i at the particle surface (mol m3) temperature of catalyst bed (K) the atomic diffusion volume of specie i (cm3/mol) the diffusion coefficient in the pores (m2 s1) the Knudsen diffusion coefficient (m2 s1) pore diameter of the catalyst (m) binary diffusion coefficient (m2 s1) the molecular weight of the specie i (kg mol1) mole percent content of the diffusing specie i (dimensionless) reaction pressure (Pa) species (i = H2, DMA, HDO, Ethanol) porosity of catalyst bed molecular viscosity (kg m1 s1) density of gas (kg m3) porosity of porous medium (dimensionless) tortuosity factor (dimensionless)
For the rational catalyst design for DMA hydrogenation, in order to simultaneously activate the carbonyl functional group and hydrogen, one should be interested in a delicate CuAOASi unit or copper phyllosilicate to enhance the formation of Cu+ and high dispersion of Cu0 species. However, compared to DMO, the significant larger molecular size of DMA might lead to lower rate of reagents’ diffusion. Moreover, the preparation of copper catalyst supported on mesoporous silica by conventional impregnation approach resulted in rare interaction between SiO2 support and Cu species, leading to lack of accessible active sites. Herein, series of well-ordered hierarchical porous silica (HPS) with tunable pore size, featured by the co-presence of both mesopores and micropores, were prepared. Further incorporation of Cu species by ammonia evaporation allows the formation of CuAOASi unit, while the agglomeration of copper species is hindered by the micropores in HPS. Meanwhile, the mesoporous structure of HPS was also successfully maintained after loading the copper species.
Scheme 1. Main reaction pathway for the hydrogenation of (A) DMO and (B) DMA.
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
It was for the first time found that the activity in dimethyl adipate hydrogenation would be readily limited by pore diffusion while the dimethyl oxalate hydrogenation was controlled by the surface reaction. Therefore, the copper catalysts on hierarchical porous silica with larger mesopores presented excellent performance for the hydrogenation of diesters with relatively long carbon chain.
225
The sample after calcination was denoted as ‘‘calcined catalyst”. After calcination, all ‘‘calcined catalysts” were reduced in 100 mL min1 H2 under ambient pressure at 623 K, denoted as ‘‘reduced catalysts”. Prior to characterization, the ‘‘reduced catalysts” were sealed in Ar atmosphere in order to prevent further oxidation. 2.3. Catalyst characterization
2. Experimental 2.1. Chemicals F127 (template agent, tri-block copolymers EO106PO70EO106, EO: poly ethylene oxide, PO: poly propylene oxide, molecular weight 12000, MEILUN biology), HCl (37 wt%, Kermel), 1,3,5trimethylbenzene (TMB) (>99 wt%, Titan), tetraethyl orthosilicate (TEOS) (98 wt%, Kermel), KCl (99.5 wt%, Kermel), Cu(NO3)23H2O (99.0–102.0 wt%, GuangFu), aqueous ammonia solution (25 wt% NH3 basis, Kermel), colloidal silica (30 wt%, Qingdao Grand Chemical), dimethyl adipate (DMA) (99 wt%, J & K), dimethyl oxalate (DMO) (99.5 wt%, TIANXI), ethanol (>99.7 wt% anhydrous, YUANLI), methanol (99.9 wt% anhydrous, Kermel), HF (40 wt%, YINGDAXIGUI), KBr (spectroscopically pure, GUOYAO), H2 (99.999%, ZHENHAO), H3BO3 (99.9 wt%, Kermel) were used as received without any further purification. 2.2. Catalyst preparation HPS. 1 g F127 (template agent, tri-block copolymers EO106PO70EO106) was dissolved in 100 mL 1 M HCl (aqueous solution) at 288 K under stirring, then 2.5 g KCl, 1.2 g 1,3,5-trimethylbenzene (TMB) were charged into the F127/HCl solution under stirring for 1 h at 288 K. Afterwards 4.16 g tetraethyl orthosilicate (TEOS) were added under stirring for another 24 h. The resulted mixture was transferred and sealed into two autoclaves (100 mL for each, polytetrafluoroethylene lined) and hydrothermally treated for 24 h at different temperature (373, 403 and 443 K). The as-made products were filtered and dried in air at room temperature for 8 h. Finally, the organic template in the dried samples was removed by thermal treatment in air at 773 K for 4 h. The obtained samples (without organic template) were denoted as HPS-y, y represents the size of mesopores, which are directly dependent on the hydrothermal temperature. Cu/HPS. Cu/HPS catalysts were prepared by the ammonia evaporation (AE) method and the process was described as follows: certain amount of Cu(NO3)23H2O was dissolved in 10 mL deionized water followed by mixing with aqueous ammonia solution (25 wt%), resulting in an aqueous solution with a pH value of 11. Subsequently, 0.9 g HPS powder was introduced into the solution and the pH of the suspension was adjusted to ca. 11 by adding additional aqueous ammonia solution. After that, the suspension was aged at room temperature for 4 h under stirring (aging stage), followed by ammonia evaporation at 353 K in air under ambient pressure, in order to remove the ammonia and allow the deposition of copper species on HPS (AE stage). The precipitates were filtered and washed with deionized water, and the resulted solid was dried at 373 K for 8 h prior to the calcination at 673 K in air for 4 h. The obtained catalysts were named as xCu/HPS-y (x refers to the wt.% loading of Cu derived from the amount of Cu(NO3)23H2O fed at the beginning). 20Cu/SiO2. 20Cu/SiO2 was also prepared by the AE method, as previous reported.14 Different with the preparation method of Cu/HPS, 20Cu/SiO2 catalyst was prepared by using colloidal silica (30 wt%, Qingdao Grand Chemical Co.) instead of HPS as the silica source, and an excess aqueous ammonia solution was adopted to adjust the pH as 13.
N2 physisorption analysis was conducted at 77 K using a Micromeritics Tristar II 3000 Analyzer instrument. The surface area of different catalysts was calculated from the isotherms through the method of Brunauer-Emmet-Teller (BET), the size distribution of mesopores was calculated by the Barrett-Joyner-Halenda (BJH) method from the adsorption isotherms. The micropore information was obtained from Micromeritics ASAP-2020. The micropore volume was calculated by Micropore Analysis (MP) method18 and the size distribution of micropores was calculated by HorvathKawazoe (HK) method [18]. The content of copper (WCu) in the catalysts was analyzed through inductively coupled plasma (ICP-OES) (Varian, VistaMPX). The examination was conducted at a high frequency emission power of 1.5 kW in a plasma air flow (15.0 L min1). Before measurements, the solid sample was dissolved in the mixture of HNO3 and HF and neutralized by HBO3. X-ray powder diffraction (XRD) results were obtained on a Rigaku C/max-2500 diffractometer employing Cu Ka radiation (k = 1.5406 Å) at room temperature. The samples were scanned from 10° to 90° (2h) with a speed of 8° min1. Scherrer equation was used to calculate the average size of crystalline copper particle. Before the characterization, the catalyst was reduced with pure H2 of ambient pressure at 623 K for 2 h. H2-TPR profiles were collected by using a Micromeritics Autochem 2910 setup. 0.05 g catalyst was packed in a quartz tube and pretreated in Ar at 473 K for 2 h. Then the samples were reduced in 5% H2/Ar (total flow rate 30 mL min1) from room temperature to 1073 K with a temperature ramp of 10 K min1. A thermal conductivity detector (TCD) was used to record the hydrogen consumption during the reduction. Transmission electron microscopy (TEM) images of the catalysts were obtained on a Philips TECNAI G2 F20 system electron microscope at 100 kV. The reduced catalyst powder was ultrasonically dispersed in ethanol and located onto a copper grid by dipping method, which was protected in N2 to prohibit the oxidation before imaging. Fourier transform infrared (FT-IR) spectroscopy experiments were carried out on a Nicolet 6700 spectrometer (Nicolet) with a valid region from 400 to 4000 cm1. The standard KBr disk method was used to press sample wafers. Characterizations of X-ray photoelectron spectra (XPS) and Auger electron spectroscopy (XAES) were carried out on a PHI 1600 ESCA instrument (PE Company) equipped with an Al Ka Xray radiation source (hm = 1486.6 eV). It was operated with pass energy of 187.85 eV and the binding energies were calibrated by the binding energy of C 1s. Surface areas of metallic copper were measured by N2O titration method using a Micromeritics Autochem II 2920 instrument [19,20]. Prior to measurements, the catalysts were pretreated in He at 473 K for 1 h, followed by reduction in 5% H2/He at 623 K for 2 h. The consumed H2 was calculated and denoted as A1. After cooling to 363 K in He atmosphere, pure N2O was passed through the catalysts at a rate of 30 mL min1 for 1 h, ensuring the surface metallic copper was oxidized to Cu2O. Then hydrogen pulse reduction of exposed Cu2O to copper was carried out at 623 K until steady state was reached, namely, the pulse area was no longer changed. The consumed H2 in this step was denoted as A2. The quantity of hydrogen consumption was determined by TCD. The metallic copper surface area (SCu) was calculated by Eq. (1) based
226
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
on an atomic copper surface density of 1.46 1019 Cu atoms/m2 [21].
SCu ¼
2A2 NA W Cu 1:46 1019 A1 M Cu
ðm2 =gÞ
ð1Þ
The dispersion (DispCu) of metallic copper can be expressed as Eq. (2):
DispCu ¼
2A2 100% A1
ð2Þ
where NA: Avogadro’s constant, MCu: relative atomic weight, WCu: the content of copper (wt%). 2.4. Catalytic reactions Hydrogenation of DMA to HDO was performed in a continuous flow unit equipped with a stainless steel fixed-bed tubular reactor. About 0.4 g of calcined catalyst particle (40–60 meshes) was placed in the isothermal zone of the stainless steel reactor (inner diameter 8 mm). The catalyst was online reduced in pure H2 at 623 K for 4 h. After the catalysts bed was cooled down to 498 K, the substrate (20 wt% DMA in methanol) was injected by using an HPLC pump (Series II, Lab Alliance) from top of the reactor. The reaction was carried out at a system pressure of 3.0 MPa and a H2/DMA molar ratio of 175. The weight liquid hourly space velocity (WHSV) of DMA was varied from 0.3 to 3.3 h1. The by-products of the reaction included 1-hexanol, oxepane, cyclohexanol, 2methylcyclopentanol, 2-methylcyclopentanone and methyl 6hydroxy-hexanoate (MHH). The liquid products in the cold trap were collected and analyzed at the same conditions by GC (BEIFEN 3420A gas chromatograph) with an HP-INNOWAX capillary column (30 m 0.25 mm 0.25 lm) and an FID detector. Furthermore, the hydrogenation of DMO to EG was conducted in the same reactor to reveal the structure effect of the catalysts. The catalyst was packed in the reactor and reduced under the same condition as those for the hydrogenation of DMA. The substrate (20 wt% DMO in methanol) was also injected into the reactor by using HPLC pump. The reaction conditions were 473 K, 2.5 MPa and H2/DMO molar ratio of 80, the same as previously reported [14]. The WHSV of DMO was varied from 0.5 to 5.0 h1. The sampling and quantitative analysis methods were the same as those for DMA hydrogenation mentioned above. The conversion and selectivity for the reaction were calculated based on the following equations. The experimental error of the results for conversion (CDMA or CDMO) and selectivity (Si) is below 0.5%.
CDMA ðor CDMO Þ ¼ 100 ð%Þ
Amount of DMA ðor DMOÞafter reaction ðmolÞ Total amount of DMA ðor DMOÞ ðmolÞ
100ð%Þ ð3Þ Si ¼
Amount of DMA ðor DMOÞconsumed by the formation of product i ðmolÞ Total amount of DMA ðor DMOÞconverted ðmolÞ 100ð%Þ
ð4Þ 3. Results 3.1. Textural properties of HPS-y and 20Cu/HPS-y Textural properties of the mesoporous HPS-y, 20Cu/HPS-y catalysts and 20Cu/SiO2 are characterized and the results are listed in
Table 1 and Figs. 1 and 2. As depicted in Fig. 1A, for each catalyst sample, only one sharp peak is observed in the BJH mesopore size distribution curve. As shown in Fig. 1B, the HPS-y materials exhibit isotherms with characteristics of type IV patterns, indicating the presence of mesopores. The shape of the HPS-y hysteresis loop is of type H1 [22], as a result of the presence of column-shaped pores. These results confirm that ordered mesoporous silica with various mesopore sizes was successfully synthesized. Furthermore, besides the mesopores, micropores with diameter of 1 nm are also found in all the HPS materials (inset in Fig. 1A, this pore distribution was provided by Horvath-Kawazoe method). The formation of the large amount of micropores in HPS might take place in the calcination step, during which the organic template was removed by the oxidation of F127 and TMB at high temperature. The poly ethylene oxide blocks of F127 molecule presented in the skeleton of parent HPS might also contribute to the formation of micropores [23]. The coexistence of mesopores and micropores demonstrates the hierarchical structure of as-synthesized HPS materials. Moreover, by elevating the hydrothermal temperature from 373 to 443 K, the mesopore volume raises from 0.69 to 0.84 cm3 g1, while the average mesopore size increases from 12.4 to 22.5 nm. However, the micropore volume decreases from 0.88 to 0.55 cm3 g1 (Table 1), which might be the reason for the decline in specific surface area of HPS with the increasing pore size. The isotherms and pore size distribution curves of the reduced 20Cu/HPS-y samples are shown in Fig. 2. Apparently, although higher content of copper component was loaded, the mesoporous structure of HPS was retained in all the 20Cu/HPS-y catalysts (Fig. 2A). As shown in Fig. 2B, 20Cu/HPS-y catalysts present the hysteresis loop of type H1 [22], keeping unchanged compared to their corresponding HPS supports (Fig. 1B). It implies that the mesoporous structure of HPS was not damaged during the copper loading process. However, the amount of micropores in the HPS support shows a dramatic decrease after Cu loading (Fig. 3). The depression of micropore volume in HPS-12, HPS-17 and HPS-22 are 65%, 87% and 93%, respectively (Table 1). Most of the micropores is blocked by copper nanoparticles or destroyed in the alkaline solution during the preparation process. Moreover, the BET specific surface areas and pore volumes of these catalysts obviously decrease, in comparison with their original HPS supports. However, the average pore diameter for each 20Cu/HPS-y catalysts is nearly the same as their supports. This is the evidence of that most of the mesopores were reserved during Cu loading. The hysteresis loop shape of 20Cu/SiO2, compared to those of 20Cu/HPS-y catalyst, however, is totally different. The H4 type isotherm of 20Cu/ SiO2 reflects the slot hole, which might be resulted from the plate particle accumulation. The adsorption and desorption isotherms have an increasing step at relative pressure of 0.4–1.0 because of the capillary condensation of N2 inside the mesopores. Moreover, 20Cu/SiO2 exhibits much smaller average pore size with wider pore distribution as well as higher specific surface area than those of 20Cu/HPS-y samples. As listed in Table 1, the copper loading examined by ICP is close to the designed value which is derived from the amount of fed Cu (NO3)2 precursor for all the catalysts. This demonstrates that the Cu species is efficiently loaded on HPS via AE method without Cu loss. The copper particle size is examined by and XRD methods. It can be concluded that the copper species are well dispersed in all the catalysts except for 20Cu/HPS-22. As the copper loading in the catalyst is high (20 wt%), the pore volume and specific surface area of the HPS support are very important for the copper dispersion of the resulted catalyst. As revealed by our previous work [24], the copper precursor ([Cu(NH3)4]2+) can be pre-distributed in the ordered mesoporous silica during the aging stage and reacted with the silica during the evaporation stage, leading to the formation of high dispersion of Cu+ and Cu0 species in the final
227
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237 Table 1 Physical properties of synthesized HPS, 20Cu/HPS-y and 20Cu/SiO2 catalysts. Catalyst
Cu loadinga (%)
SBETb (m2/g)
Vmesoc (cm3/g)
Vmicrod (cm3/g)
Dc (nm)
DCue (nm)
HPS-12 HPS-17 HPS-22 20Cu/HPS-12 20Cu/HPS-17 20Cu/HPS-22 20Cu/SiO2
– – – 20.33 20.60 20.78 19.30
750 613 210 218 200 129 483
0.69 0.75 0.84 0.48 0.45 0.61 1.18
0.88 0.66 0.55 0.31 0.16 0.04 –
12.4 18.2 22.5 12.3 18.3 22.3 8.1
– – – ND ND 10 5
ND: Not detectable. a Determined by ICP-OES analysis. b Total specific surface area, measured by BET method. c Mesopore properties determined by BJH method (between 1.7 nm and 300 nm diameter). d Micropore properties determined by MP method (less than 2 nm) e Calculated from the XRD data based on the Scherrer equation.
Fig. 1. (A) BJH pore distribution of the HPS samples, (B) N2 adsorption-desorption isotherms of the HPS samples.
Fig. 2. (A) BJH pore distribution based on the adsorption isotherm of the reduced catalysts, (B) N2 adsorption-desorption isotherms of the reduced catalysts.
reduced catalyst. If the pore volume and specific surface area of the HPS are not large enough, the severe agglomeration of copper species will be inevitable due to the worse distribution of the copper precursor ([Cu(NH3)4]2+) in the support. In the present work, the anchoring effect of the hierarchical structure prohibits the severe aggregation of copper species during the AE stage and the follow-
ing calcination and reduction stages, resulting in the high dispersion of copper species. Noticeably, the average copper particle size in 20Cu/HPS-22 is larger than 10 nm, differing from other HPS supported Cu catalysts. Considering the variation of HPS structure before and after the loading of copper species, it was suggested that the amount of micropores in 20Cu/HPS-22 might be
228
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
not large enough for anchoring copper species, leading to the agglomeration of copper nanoparticles. All the other catalysts do not show aggregation of Cu species. Above investigation was further strengthened by the following TEM images. 3.2. TEM results The morphology and structure of the calcined 20Cu/SiO2 catalyst and HPS with different pore size are further characterized by TEM. As shown in Fig. 4, the 20Cu/SiO2 catalyst presents curly tube structure of the copper phyllosilicate [17,25], while all the HPS materials show highly ordered mesopores structure with various pore sizes, in accordance with their BJH pore size distributions. In this work, the 1D long-straight mesopores are easily observed in the TEM images (Fig. 4). The electron diffraction results of HPS (the upper inset images in Fig. 4b–d) do not exhibit the observable ordered lattice array, normally considered as a firm evidence for the presence of cubic mesostructure in some mesoporous silica [26]. It indicates that the as-synthesized HPS materials do not present cubic mesoporous construction units. TEM images of the reduced catalysts are presented in Fig. 5. It was also found that the morphology of 20Cu/HPS-y catalysts is different from that of 20Cu/SiO2 catalyst. Although the copper nanoparticles on 20Cu/SiO2 are as small as those on Cu/HPS catalysts, the supports are lack of ordered porous structure. It is also worth mentioning that the hierarchical porous structure of the supports is retained in the 20Cu/HPS-12 and 20Cu/HPS-17 catalysts. As presented in Fig. 5d, agglomeration of the copper species is observed on 20Cu/HPS-22. It is further confirmed that the micropores in HPS-22 are not sufficient to anchor the copper precursors during the Cu loading, leading to the aggregation of copper species in the calcined and reduction stages. Additionally, some ordered mesoporous structure of 20Cu/HPS-22 was destroyed, due to the thinner wall compared with those of other Cu/HPS catalysts. 3.3. XRD profiles X-ray powder diffraction patterns of calcined and reduced 20Cu/HPS-y are shown in Fig. 6. As shown in Fig. 6A, the broad signal in the range of 2h = 10–28° is attributed to amorphous silica for all tested catalysts. For the calcined 20Cu/HPS-12 and 20Cu/HPS17 catalysts, no characteristic peak of CuO can be found (35.4° and 38.6°, JCPDS05-0661) in their XRD patterns, while 20Cu/HPS22 presents characteristic diffraction peaks for CuO. It indicates that Cu species on the calcined 20Cu/HPS-12 and 20Cu/HPS-17 catalysts was highly dispersed. Furthermore, characteristic peaks of copper phyllosilicate (30.8°, 35.0°, 57.5° and 62.4°) were observed in the patterns of 20Cu/SiO2. However, similar features were not observed in any of the 20Cu/HPS-y catalysts. XRD patterns of the reduced copper catalysts are shown in Fig. 6B. The patterns for 20Cu/HPS-12 and 20Cu/HPS-17 give no obvious peak for any copper species (Cu or Cu2O), similarly due to the high dispersion as for the corresponding calcined samples (before reduction). In contrast, 20Cu/SiO2 and 20Cu/HPS-22 catalysts present distinct diffraction peaks of Cu2O and Cu (located at 33.6° and 43.2°). Although the Cu loading for each 20Cu/HPS-y catalyst is the same, the reason for higher crystallinity and larger particle size of Cu species observed in 20Cu/HPS-22 (10 nm based on Scherrer equation) can be attributed to the depressed micropore volume, compared to 20Cu/HPS-12 and 20Cu/HPS-17 catalysts. For 20Cu/HPS-12 and 20Cu/HPS-17, it was suggested that the Cu species was mainly located on the micropore in the HPS supports, hindering the agglomeration of Cu particles, in accordance with the anchoring effect proposed from BET results.
3.4. FT-IR results In order to clarify the interaction between Cu and HPS support as well as surface species, FT-IR technology was applied in addition to XRD. The IR spectra of calcined 20Cu/HPS-y and 20Cu/SiO2 samples were shown in Fig. 7. Evidently, the representative bands of the dOH at 938 and 694 cm1 are not found in all the samples, evidencing that Cu(OH)2 is not formed on all catalysts. The 20Cu/SiO2 catalyst shows characteristic bands for copper phyllosilicate [17] at 1047 and 670 cm1, which are not observed in the calcined 20Cu/ HPS-y samples. The shoulder band at 543 cm1 beyond the intensive IR band at 473 cm1 is attributed to CuO species, due to signal overlapping of CuO and SiAOASi unit [27,28]. The shoulder band at 970 cm1 (beyond the intensive band at 1080 cm1), assigned to the stretching vibrations of surface CuAOASi bond [16], evidences the strong interaction between Cu species and HPS supports. According to these results, it can be deduced that there is no copper hydroxide and copper phyllosilicate in the calcined 20Cu/HPSy samples. Therefore, the copper species present in 20Cu/HPS-y catalysts are mainly well dispersed CuO and CuAOASi species. In contrast, the formation of copper phyllosilicate during the preparation of Cu/SiO2 catalyst was evidenced, in accordance of previous work [9].
3.5. TPR results The H2-TPR profiles of calcined samples are shown in Fig. 8. For each sample, the reduction peak around 468 K belongs to the reduction of Cu2+ to Cu0 and Cu+ species [29,30]. Clearly, copper species with larger nanoparticle size are more difficult to be reduced [31,32]. Furthermore, there is a broad signal between 810 and 840 K (inset figure) owing to the deep reduction of CuAOASi species (formed via strong interaction between Cu and HPS, as discussed above) or copper phyllosilicate (Fig. 8) to Cu0 [29]. Apparently, the reduction temperatures attributed to CuAOSi for both 20Cu/HPS-12 and 20Cu/HPS-17 (800 and 810 K, respectively) are lower than that attributed to copper phyllosilicate for 20Cu/SiO2 (840 K). Combining with the FTIR results, it could be deduced that the deep reduction of copper phyllosilicate to Cu0 is much more difficult than that of CuAOASi structure present in Cu/HPS catalysts. Notably, small shoulder peaks at 527 K and 511 K were found for 20Cu/SiO2 and 20Cu/HPS-22 catalysts,
Fig. 3. Micropore distribution profiles for the HPS supports and Cu/HPS catalysts.
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
229
Fig. 4. TEM images of calcined 20Cu/SiO2 catalyst and HPS with different pore size. (a) 20Cu/SiO2, (b) HPS-12, (c) HPS-17, (d) HPS-22. Upper right insets: electron diffraction patterns. Lower right insets: amplified pore structure of HPS.
respectively. It could be ascribed to the consumption of H2 for large copper particle reduction. Moreover, the shift of the main reduction peak (467–480 K) to the high temperature is referred to the lower copper dispersion. 3.6. XPS spectra XPS measurements were carried out to evaluate surface composition and valence state of copper [33]. As illustrated in Fig. 9A, two intensive photoelectron peaks emerging at 932.9 eV and 952.7 eV are ascribed to Cu 2p3/2 and Cu 2p1/2. Generally, the binding energy of the Cu 2p peak at about 934.2 eV and the characteristic shakeup feature at 944 eV was used as fingerprint signal to confirm the presence of Cu2+ species [34]. Thus, the absence of the shakeup peak at 942–944 eV in the XPS spectra of Cu/HPS catalysts strongly demonstrates that all Cu2+ species have been reduced to low valence state of +1 or 0 [35]. The ratio of band intensity in LMM Auger electron spectra could be representative for the copper concentration on the catalyst surface and, therefore, was applied to distinguish the chemical states by their different kinetic energies. In general, the peaks at 913.1 and 916.7 eV are ascribed to Cu+ and Cu0, respectively [9,36,37]. The Cu LMM curves were deconvoluted and the results are shown in Fig. 9B. As listed in Table 2, the ratio of Cu+/(Cu+ + Cu0), denoted as XCu(I), was calculated from the integral area of deconvoluted curves. Based on the metallic Cu surface area determined by N2O titration, the surface area of Cu+ can be estimated with the help
of XCu(I). These results justify that the improved AE method ensures not only a reasonable copper dispersion, but also two sorts of copper species with almost identical XCu(I) (among 45.3–48.9%). The copper catalysts supported on HPS show lower surface area of Cu+ and Cu0 than 20Cu/SiO2, and the 20Cu/HPS-y catalyst with larger pore diameter give the lower surface area of copper species. Thus, the trend for the copper surface area of the various catalysts is: 20Cu/SiO2 > 20Cu/HPS-12 > 20Cu/HPS-17 > 20Cu/HPS-22. As depicted in TEM images and FTIR spectrum, the existence of phyllosilicate in 20Cu/SiO2 samples can be confirmed. This structure favors the formation of Cu+ during catalysts reduction due to the strong interaction between copper and silica [33]. In contrast, in the calcined Cu/HPS catalysts, only CuAO-Si species was found. Meanwhile, compared to the formation of phyllosilicate on Cu/ SiO2, the formation of CuAO Si species on HPS should be much more difficult due to the less surface silica provided during the preparation process. So the Cu+ surface areas in Cu/HPS catalysts are much less than 20Cu/SiO2 (Table 2). The anchoring function of micropore leads to the higher dispersion of copper species [38]. The presence of more micropores in 20Cu/HPS-12 leads to the enrichment of both Cu+ and Cu0, compared to 20Cu/HPS-17 and 20Cu/HPS-22. Surprisingly, the copper particle size in Cu/HPS is smaller than 20Cu/SiO2, but the surface area of Cu0 species is much lower than 20Cu/SiO2. Moreover, the Cu nanoparticles (ca. 3.0 nm, estimated by the N2O titration) are larger than the micropore diameter (about 1.2 nm) of HPS. Therefore, it is rational that some copper nanoparticles might be located
230
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
Fig. 5. TEM images of the reduced 20Cu/SiO2 and 20Cu/HPS-y catalysts. (a) 20Cu/SiO2, (b) 20Cu/HPS-12, (c) 20Cu/HPS-17, (d) 20Cu/HPS-22. Insets: amplified pore structure of Cu/HPS.
Fig. 6. XRD patterns of the (A) calcined catalysts, (B) reduced catalysts. (a) 20Cu/HPS-12, (b) 20Cu/HPS-17, (c) 20Cu/HPS-22, (d) 20Cu/SiO2.
at the micropore opening of the HPS supports, so that large fraction of copper surface inside the micropore cannot be observed by N2O titration method due to the diffusion effect, resulting in an overall lower Cu0 surface area in Cu/HPS than 20Cu/SiO2. Nevertheless, the presence of micropores in HPS favors a higher copper dispersion in Cu/HPS, due to its anchoring effect.
3.7. Diester hydrogenation on 20Cu/HPS-y and 20Cu/SiO2 Gas-phase hydrogenation of DMA over 20Cu/HPS-y and 20Cu/ SiO2 catalysts was performed. As shown in Fig. 10, the three catalysts present the same trend that with increasing WHSV, the DMA conversion decreases, while the HDO selectivity first increases then
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
Fig. 7. FTIR spectra of different samples (a) 20Cu/HPS-12, (b) 20Cu/HPS-17, (c) 20Cu/HPS-22, (d) 20Cu/SiO2.
231
inhibited. The HDO selectivity on 20Cu/SiO2 is much lower than on 20Cu/HPS-y under the same conditions. Meanwhile, even though 20Cu/SiO2 has the highest copper dispersion and surface area of copper species, its DMA conversion is much lower than that of 20Cu/HPS-22. A maximum conversion of DMA on 20Cu/HPS-22 was up to 99% and the selectivity of HDO was over 89% under the WHSV of 1.2 h1. Notably, only trace amounts of byproducts are detected over 20Cu/HPS-22 (Table S1). The products are mainly MHH and HDO whose total selectivity is more than 97.5% at the WHSV of 0.9 h1. While the total selectivity of MHH and HDO for 20Cu/SiO2 catalyst is only 87.1% under the same reaction conditions (Table S2), which is much lower than that of 20Cu/HPS-22. At the condition of WHSV = 1.2 h1, the 20Cu/HPS-22 catalyst shows excellent catalytic performances with a STYHDO (spacetime yield of HDO) of 0.72 g/(g∙h), which is more than double that (STYHDO = 0.33 g/(g∙h)) of the ever reported Cu-Zn-Al catalyst for the DMA hydrogenation [8]. Therefore, the design strategy of 20Cu/HPS-y catalysts have profound potential in preparing the industrial catalyst for DMA hydrogenation. To gain further insight into the unique nature of the 20Cu/HPS-y catalysts in the hydrogenation of esters, 20Cu/HPS-y catalysts had also been tested in the hydrogenation of DMO to EG and compared with 20Cu/SiO2 catalyst [39]. The results of catalytic DMO hydrogenation are shown in Fig. 11. When the WHSV is changed from 0.5 to 5, the DMO conversion decreases, while the EG selectivity first increases then drops down. The MG selectivity increases with increasing WHSV, while the ethanol selectivity decreases when the WHSV rises up to 5. The short residence time of the reactants doesn’t allow the total conversion of DMO and the further hydrogenation of EG to ethanol. When the WHSV is below 2.0 h1, competitive apparent activity is achieved on the tested catalysts. But the overall activity and EG selectivity on 20Cu/HPS-17 are much lower than those on 20Cu/HPS-12 and 20Cu/SiO2 with WHSV higher than 3 h1. Obviously, the 20Cu/SiO2 catalyst has the highest activity and EG selectivity among these catalysts in DMO hydrogenation. The activity sequence for DMO hydrogenation is different from that for DMA hydrogenation.
4. Discussion Fig. 8. H2-TPR profiles of the calcined samples.
drops down slightly. This phenomenon is rationalized by that at high WHSV, relatively low concentration of accessible active sites cannot meet the demand by the total conversion of DMA, and at the same time, further hydrogenation of HDO to hexyl alcohol is
4.1. Structure effect of HPS on the formation of catalyst On basis of the characterization results, it is demonstrated that a novel hierarchical porous silica material having both micropores (1 nm) and mesopores ( 12 to 22 nm) was successfully pre-
Fig. 9. (A) Cu 2p photoelectron spectra of the reduced samples, (B) Cu LMM photoelectron spectra of the reduced samples.
232
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
pared. The mesopore size of HPS can be easily tuned by varying the hydrothermal treatment temperature. It is noteworthy that the structure of the as-synthesized HPS is different from the previous reported FDU material by Fan et al. [26,40–42] FDU is known for its 3D ordered porous structure possessing both ordered mesopores and larger cages. However, no large cage was found in the HPS materials, but 1D straight channel instead. During the loading of copper catalyst by ammonia evaporation method, the ordered mesoporous structure of HPS was well maintained in the final Cu/HPS catalysts with a reasonable dispersion of copper species. Yin et al. [36] has reported a Cu/HMS catalyst synthesized by a one-pot method, but severe deterioration of the mesoporous structure of HMS was observed. The similar issue was also found in the Cu/MCM-41 catalyst reported in our previous work, the structure of MCM-41 was thoroughly destroyed during the ammonia evaporation process [16]. It can be observed in the TEM images that the wall of the mesopores is 4–5 nm thick for HPS-12 and HPS-17, and ca. 3 nm thick for 20Cu/HPS-22. These walls are much thicker than that of MCM-41 (less than 2 nm). This feature makes the ordered mesoporous structure of HPS stable during the AE stage, especially under the conditions of lower copper loading. Baldizzone et al. [43] found that mesoporous structure of graphitic hollow spheres could physically confine the metal particles and prevent their extensive sintering. Ding et al. [38] reported that the spatial restriction created by the small pores of the support would increase the proportion of the filled Cu particles ( 20Cu/HPS-12 > 20Cu/ HPS-17 > 20Cu/HPS-22. However, the activity of these catalysts in DMA hydrogenation is just contrary to this sequence. The Cu/ HPS-22 catalyst showed the highest activity regardless of its lowest surface areas of both Cu0 and Cu+. We suggested that the accessible Cu+ and Cu0 on 20Cu/HPS-y should be relatively sufficient enough for the DMA hydrogenation under the given reaction conditions. The reason beyond is that one should be concerned for diffusion effect when relatively large molecules of reactant or intermediate are involved in catalytic reaction occurred in the porous catalyst. According to the texture properties of these catalysts, the ordered mesoporous structure might acts as the determination factor in the DMA hydrogenation [44]. However, compared to that in DMA hydrogenation, these catalysts present a different activity sequence in the hydrogenation of DMO. Their catalytic activities are consistent well with their surface areas of copper species (Table 2). In other words, in the catalytic system of DMO hydrogenation, the number of accessible Cu sites under reaction condition might be very different from those for DMA hydrogenation. It can be concluded that the catalytic activities of 20Cu/HPS-y and 20Cu/SiO2 in DMO hydrogenation are mainly determined by the copper dispersion. The 20Cu/ SiO2 exhibits the highest activity due to its highest surface area of both Cu0 and Cu+, which are considered as the active sites for the DMO hydrogenation. Considering the results for the hydrogenation of DMA and DMO, it is also observed that the activity in the hydrogenation of DMO highly depends on the surface copper species, while the hydrogenation of DMA was controlled by pore size of the catalyst. According to the DFT calculations (SI), the molecular kinetic diameter of DMA is ca. 1.6 nm, which is much larger than DMO (ca. 0.6 nm). Taken their contrary activity sequence in consideration, the internal mass transfer should be a key factor for the hydrogenation of DMA. Therefore, the catalytic performance in the hydrogenation of DMA can be effectively improved by tuning the pore size of the support, in order to make more Cu sites accessible under reaction conditions.
4.3. Influence of mass transfer on the hydrogenation of diester To accurately evaluate the effect of the mass transfer in the hydrogenation of DMA, the mesoporous structure of HPS should
Table 2 Copper dispersion of different catalysts.
a
Cat.
DispCu (%)a
SCu(0) (m2/g)a
XCu(I) (%)b
SCu(I) (m2/g)c
10Cu/HPS-12 10Cu/HPS-17 10Cu/HPS-22 20Cu/HPS-12 20Cu/HPS-17 20Cu/HPS-22 20Cu/SiO2
35.05 31.25 26.48 31.87 26.53 11.39 25.8
14.41 11.32 10.26 20.89 17.62 9.25 27.3
53.20 53.12 53.88 48.86 45.29 47.06 46.62
16.38 12.83 11.99 19.96 14.58 8.23 23.84
Metallic Cu surface area determined by N2O titration. Cu+/(Cu+ + Cu0) calculated from Cu LMM XAES spectra. Cu+ surface area estimated on the basis of SCu(0) and XCu(I) under the assumption that the Cu+ ions and Cu0 atoms occupy identical areas and have identical atomic sensitivity factors. b
c
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
233
Fig. 10. Results for the hydrogenation of DMA on different catalysts. Reaction conditions: T = 498 K, p = 3 MPa, H2/DMA = 175.
Fig. 11. Results for the hydrogenation of DMO on various catalysts. Reaction conditions: T = 473 K, p = 2.5 MPa, H2/DMO = 80.
be maintained and the rational copper dispersion should be achieved for the catalysts. Therefore, series of 10Cu/HPS-y catalysts with 10% copper loading were prepared. The N2 adsorption (Fig. S1 and Table S3), XRD (Fig. S2), H2-TPR (Fig S3), TEM (Fig. S4), FTIR (Fig. S5), N2O titration (Table S3), XPS (Fig. S6) and in situ FT-IR of CO adsorption (Fig. S7) of 10Cu/HPS-y are carried out and the results are shown in Supporting Information. The ordered mesoporous structures of HPS are all retained in the reduced 10Cu/HPS-y catalysts and the copper species are
well dispersed on all the HPS with various mesopore diameter. Based on the characterization results of N2O titration and Cu LMM XAES, the surface areas of Cu+ and Cu0 were achieved and listed in Table 2. The rsite values (the conversion rate of the diesters normalized by reaction time and number of active sites (Cu+ or Cu0), h1) for these catalysts in the diester hydrogenation were also calculated and listed in Table 3 and 4. In order to evaluate the effect of mass transfer in the hydrogenation of DMA and DMO, Carberry criterion and Weisz-Prater criterion were applied to calculate
234
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
the external and internal mass transfer limitation in Cu/HPS catalysts. Carberry number (Ca), a ratio between the observed reaction rate and the maximum external mass transfer rate, is often used to estimate the external diffusion influence. Generally, the external diffusion may be neglected if the Carberry number is less than 0.05 [45]. The formula is showed in Eq. (5).
Cags ¼
r obs kg;i as ðC i C 0i Þ
ð5Þ
where Cags is the Carberry number, robs is the apparent reaction rate (mol L1 s1), C0i and C⁄i (mol m3) is the concentration of species i before and after the reaction at the bulk of catalysts bed, as is the external surface area of the catalysts and kg,i is the gas-solid diffusion coefficient of species i [46]. kg,i can be obtained by Eq. (6) and the relationship between JD and the Reynolds numbers (Re ¼ qg ud=l) is given in Eq. (7) [47]. Sc number (Sc ¼ l=qg DB;i ) reflects the effect of mass transfer on the physical properties of fluids. The notations l and qg represent the viscosity and density of the bulk gas phase, respectively. DB,i is the diffusion coefficient of species i in the bulk gas phase.
DB;i 1=3 Sc ReJD R 1 0:765 0:365 þ JD ¼ h Re0:82 Re0:386
kg;i ¼
ð6Þ ð7Þ
The catalytic performances for DMA hydrogenation over 10Cu/ HPS-y at various WHSV are showed in Fig. S8. The Carberry number for these catalysts were calculated and presented in Tables S4, S5, S6 and Fig. S9. When the WHSV is higher than 2.1 h1, the Carberry numbers for all the three catalysts are lower than 0.05, indicating the external diffusion can be neglected [48]. Weisz-Prater criterion is often applied to assess the internal diffusion effect in heterogeneous catalytic reactions. The physical meaning of Weisz-Prater number (NW-P) is the ratio between the observed reaction rate and the maximum internal mass diffusion rate. When NW-P is less than 0.3, the internal diffusion may be neglected and smaller NW-P implies less diffusion influence [49– 52]. Weisz-Prater number is given in Eq. (8).
NWP ¼
r obs R2 Deff ;i C i
ð8Þ
where robs is the apparent reaction rate (mol L1 s1). R is the radius of the catalyst pellet, which is 3.75 104 m in this study, C⁄i is the concentration of species i at the external surface of catalysts, which is generally assumed to be equal the bulk concentration. Deff,i is the effective diffusion coefficient in the pores of the catalysts and it can be calculated by Eqs. (9)–(13) [53]. The Deff,i is determined by diffusion coefficient in the pores (Dp,i), porosity (e) and tortuosity factor (s) of the catalyst sample. Furthermore, as presented in Eq. (10), the Dp,i can be calculated by considering both the Knudsen diffusion coefficient (Dk,i) and the bulk diffusivity of specie i (DB,i). The Dij represents the binary diffusion coefficient which means the diffusion coefficient of species i in the fluid of species j (Eq. (13)).
e s
ð9Þ
1 1 1 ¼ þ Dp;i Dk;i DB;i
ð10Þ
Deff ;i ¼ Dp;i
Dk;i
1 ¼ dp 3
sffiffiffiffiffiffiffiffiffi 8RT pMi
ð11Þ
n 1 1 X yi ¼ DB;i 1 yi j¼2 Dij
ð12Þ
1=2 0:001T 3=2 1 1 Dij ¼ h þ i 2 P P Mi Mj P ð VÞ1=3 þ ð VÞj1=3 i
ð13Þ
The internal mass-transfer calculation results for the hydrogenation of DMA and DMO are listed in Tables 3 and 4, respectively. It can be observed that the internal diffusion has significant influence on the DMA hydrogenation since the W-P numbers are much higher than 0.3 when the pore size is 12 nm. And the decrease of the W-P numbers with the increase of pore size implies the enhanced internal diffusion in the catalysts with larger pore diameter. When the pore size is 17 nm, the internal diffusion of HDO and DMA can be nearly neglected. If the pore size is further increased to 22 nm, the internal diffusion effect can be ignored in the reaction since the W-P numbers are lower than 0.3. The much lower rsite of 10Cu/HPS-12 may result from the pore diffusion effect in the reaction. In contrast, in the case of the DMO hydrogenation, all the NW-P for various catalysts are much lower than 0.3, (Table 4) indicating the absence of the internal mass transfer influence. Similar rsite for various catalysts in DMO hydrogenation can also confirm the absence of mass transfer effect. So the as-prepared catalysts with larger ordered mesopores will not benefit the DMO hydrogenation, because the surface reaction is the controlling step. Based on the results of catalyst characterization, the mass transfer calculations and catalytic performances in esters hydrogenation, we can conclude that the hydrogenation of DMA over the as-prepared copper-based catalyst is readily determined by the internal diffusion of dimethyl adipate when the pore diameter was less than 17 nm, while the hydrogenation of DMO lies on the surface reaction. Thus, larger pores in the catalysts can prompt the activity by enhance the mass transfer in DMA hydrogenation reaction system. This effect was depicted as a sketch diagram shown in Fig. 13. It is worth noting that only if the amount of active sites is sufficient for the reaction, the structure of the catalyst will play the key role in the reaction. Therefore, it is reasonable that higher copper loading catalysts (such as 20Cu/HPS-y) shows more severe internal diffusion influence, since more active sites are available for the reaction. Moreover, the ammonia evaporation method was proved to be an effective method to immobilize the active spe-
Fig. 12. Correlation between surface areas of copper species and HPS’s micropore volume.
235
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237 Table 3 Weisz–Prater number of different catalysts in DMA hydrogenationa. Cat. 10Cu/HPS-12 10Cu/HPS-17 10Cu/HPS-22 a b c
CDMA (%)
SHDO (%)
Nw-p
Nw-p
DMA
HDO
rsite-Cu(0) (h1)
rsite-Cu(I) (h1)
76.4 89.77 92.19
72.32 79.68 80.28
0.32 0.29 0.24
0.4 0.31 0.25
12(±0.2)b 21(±0.3)c 23(±0.1)c
10(±0.3)b 18(±0.3)c 20(±0.4)c
Reaction conditions: T = 498 K, p = 3 MPa, H2/DMA = 175, WHSV = 2.4 h1. The rsite results were calculated under WHSV = 2.4 h1. The rsite results were calculated under WHSV = 3.0 h1 (As listed in Tables S5, S6 and Fig. S8).
Table 4 Weisz–Prater number of different catalysts in DMO hydrogenationa. Cat. 10Cu/HPS-12 10Cu/HPS-17 10Cu/HPS-22 a
CDMO (%)
SEG (%)
Nw-p
Nw-p
DMO
EG
rsite-Cu(0) (h1)
rsite-Cu(I) (h1)
84.71 56.21 54.54
61.07 41.20 43.11
0.0110 0.0054 0.0039
0.0160 0.0110 0.0085
17 (±0.1) 14 (±0.3) 15 (±0.3)
15 (±0.2) 13 (±0.3) 13 (±0.4)
Reaction conditions: T = 473 K, P = 2.5 MPa, H2/DMA = 80, WHSV = 2.0 h1.
Fig. 13. Sketch diagram for the esters hydrogenation mechanism on Cu/HPS catalyst. Fig. 14. Catalytic stability of 20Cu/HPS-17 and 20Cu/SiO2 for DMA hydrogenation. Reaction conditions: T = 498 K, p = 3 MPa, H2/DMA = 175 and WHSV = 1.0 h1.
cies on the mesoporous material without changing its ordered mesoporous structure. The ordered hierarchical porous material with both mesopores and micropores will also benefit the catalyst dispersion due to the anchoring effect of its abundant micropores [38]. These investigations will be meaningful for the rational design and fabrication of the copper-based catalysts for ester hydrogenation. 4.4. Catalyst stability The stability of 20Cu/HPS-17 catalyst was evaluated in the hydrogenation of DMA. As depicted in Fig. 14, compared with 20Cu/SiO2, the 20Cu/HPS-17 exhibits a better stability during the 120-h hydrogenation with totally DMA conversion and high HDO yield of 89%. However, the conversion of DMA on 20Cu/SiO2 declines quickly. According to the FTIR results, the formation of CuAOASi species, owing to the strong interaction between copper species and the surface silica of HPS, is conducive to achieve a balance of Cu0 and Cu+ distribution [13]. Meanwhile, the XRD patterns (Fig. S10) of the used catalysts showed that Cu0 nanoparticle sizes were 5.8 and 8.9 nm for 20Cu/HPS-17 and 20Cu/SiO2, respectively. Obviously, the aggregation of the copper species happened on both 20Cu/HPS-17 and 20Cu/SiO2, whose initial metallic copper nanoparticle sizes were 3 and 5 nm, respectively. In the case of
Cu/SiO2, considering the much smaller average pore diameter (8.1 nm) than that of 20Cu/HPS-17 (18.3 nm), the copper nanoparticle size (5 nm) closer to pore diameter should definitely lead to serious diffusion resistance of the reactants in the pores of 20Cu/ SiO2 catalyst. In contrast, for Cu/HPS-17, above mentioned diffusion resistance become less crucial due to its larger mesopore size. Thus, 20Cu/SiO2 presented a worse stability than 20Cu/HPS-17 catalyst. The anchoring effect of the micropores might retard the severe aggregation of copper nanoparticles on HPS during the reaction [38]. Of course, some improvements were still necessary to further enhance the stability of Cu/HPS-17 catalyst by tuning the surface and interface structure of the catalyst and the mesopore size of HPS. 5. Conclusion An ordered hierarchical porous copper-based catalyst was prepared by ammonia evaporation technique for the gas-phase hydrogenation of diesters. The Cu/HPS catalysts with hierarchical porous structure show high copper dispersion and surface areas of copper species. These characters can be ascribed to the anchoring effect of the micropores, which can well disperse the copper species during
236
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
the ammonia evaporation process. Because of these outstanding properties, a remarkable HDO yield of about 89% was achieved in DMA hydrogenation at a nearly total conversion of DMA under the WHSV of 1.2 h1. This excellent catalytic performance of Cu/ HPS could result from the high surface areas of copper species, appropriate Cu0/Cu+ ratio and high sintering resistance. Additionally, the pore size of the catalysts plays a determining role in the hydrogenation of DMA. By tuning the pore size of the ordered mesoporous HPS, the catalytic activity and selectivity can be significantly improved. Furthermore, it is also revealed that the control step of diester hydrogenation is different: DMO hydrogenation is controlled by the surface reaction, while DMA hydrogenation will be easily influenced by the pore diffusion. These investigations give an important instruction to rationally design copper-based catalysts for the hydrogenation of various esters. Acknowledgement The authors are grateful to the financial support from the National Nature Science Foundation of China (21276186, 21325626, 91434127, U1510203), the Tianjin Nature Science Foundation (13JCZDJC33000) and the Chinese Scholarship Council. This work by Y. Zhao was also supported by the US Department of Energy, Office of Science and Office of Basic Energy Sciences. Pacific Northwest National Laboratory (PNNL) is operated by Battelle for the Department of Energy under Contract DE-AC05-76RL01830. We would like to express our special thanks to Donald M Camaioni and Kenneth G Rappé (PNNL) for their constructive suggestions to this study. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.jcat.2017.11.006. These data include MOL files and InChiKeys of the most important compounds described in this article. References [1] S.M. dos Santos, A.M. Silva, E. Jordão, M.A. Fraga, Performance of RuSn catalysts supported on different oxides in the selective hydrogenation of dimethyl adipate, Catal. Today 107–108 (2005) 250–257. [2] A.M. Silva, O.A.A. Santos, M.A. Morales, E.M. Baggio-Saitovitch, E. Jordão, M.A. Fraga, Role of catalyst preparation on determining selective sites for hydrogenation of dimethyl adipate over RuSn/Al2O3, J. Mol. Catal. A-Chem. 253 (2006) 62–69. [3] F.C.A. Figueiredo, E. Jordão, W.A. Carvalho, Adipic ester hydrogenation catalyzed by platinum supported in alumina, titania and pillared clays, Appl. Catal. A – Gen. 351 (2008) 259–266. [4] J. Fontana, C. Vignado, E. Jordão, W.A. Carvalho, Support effect over bimetallic ruthenium-promoter catalysts in hydrogenation reactions, Chem. Eng. J. 165 (2010) 336–346. [5] H.R. Yue, Y.J. Zhao, X.B. Ma, J.L. Gong, Ethylene glycol: properties, synthesis, and applications, Chem. Soc. Rev.. 41 (2012) 4218–4244. [6] H.B. Jiang, H.J. Jiang, K. Su, D.M. Zhu, X.L. Zheng, H.Y. Fu, H. Chen, R.X. Li, A RuSn-Co/AlO(OH) as a highly efficient catalyst for hydrogenation of dimethyl adipate to 1,6-hexanodiol in aqueous phase, Appl. Catal. A – Gen. 447–448 (2012) 164–170. [7] T. Turek, D.L. Trimm, N.W. Cant, The catalytic hydrogenolysis of esters to alcohols, Catal. Rev. 36 (1994) 645–683. [8] P. Yuan, Z. Liu, T. Hu, H. Sun, S. Liu, Highly efficient Cu-Zn-Al catalyst for the hydrogenation of dimethyl adipate to 1,6-hexanediol: influence of calcination temperature, Reac Kinet Mech Catal. 100 (2010) 427–439. [9] L.F. Chen, P.J. Guo, M.H. Qiao, S.R. Yan, H.X. Li, W. Shen, H.L. Xu, K.N. Fan, Cu/ SiO2 catalysts prepared by the ammonia-evaporation method: Texture, structure, and catalytic performance in hydrogenation of dimethyl oxalate to ethylene glycol, J. Catal. 257 (2008) 172–180. [10] Z. He, H. Lin, P. He, Y. Yuan, Effect of boric oxide doping on the stability and activity of a Cu-SiO2 catalyst for vapor-phase hydrogenation of dimethyl oxalate to ethylene glycol, J. Catal. 277 (2011) 54–63. [11] B. Zhang, S. Hui, S. Zhang, Y. Ji, W. Li, D. Fang, Effect of copper loading on texture, structure and catalytic performance of Cu/SiO2 catalyst for hydrogenation of dimethyl oxalate to ethylene glycol, J. Nat. Gas Chem. 21 (2012) 563–570.
[12] S.R. Wang, Q.Q. Yin, X.B. Li, Catalytic Performance and texture of TEOS based Cu/SiO2 catalysts for hydrogenation of dimethyl oxalate to ethylene glycol, Chem. Res. Chin. Univ. 28 (2012) 119–123. [13] S. Zhao, H. Yue, Y. Zhao, B. Wang, Y. Geng, J. Lv, S. Wang, J. Gong, X. Ma, Chemoselective synthesis of ethanol via hydrogenation of dimethyl oxalate on Cu/SiO2: Enhanced stability with boron dopant, J. Catal. 297 (2013) 142– 150. [14] S. Li, Y. Wang, J. Zhang, S. Wang, Y. Xu, Y. Zhao, X. Ma, Kinetics study of hydrogenation of dimethyl oxalate over Cu/SiO2 catalyst, Ind. Eng. Chem. Res. 54 (2015) 1243–1250. [15] E.K. Poels, D.S. Brands, Modification of Cu/ZnO/SiO2 catalysts by high temperature reduction, Appl. Catal. A – Gen. 191 (2000) 83–96. [16] X. Ma, H. Chi, H. Yue, Y. Zhao, Y. Xu, J. Lv, S. Wang, J. Gong, Hydrogenation of dimethyl oxalate to ethylene glycol over mesoporous Cu-MCM-41 catalysts, AIChE J. 59 (2013) 2530–2539. [17] T. Toupance, M. Kermarec, J.-F. Lambert, C. Louis, Conditions of formation of copper phyllosilicates in silica-supported copper catalysts prepared by selective adsorption, J. Phys. Chem. B 106 (2002) 2277–2286. [18] C. Zhang, Y. Wang, X. Yan, Liquid-phase adsorption: Characterization and use of activated carbon prepared from diosgenin production residue, Colloid Surf. A – Physicochem. Eng. Asp. 280 (2006) 9–16. [19] Y.J. Zhao, S.M. Li, Y. Wang, B. Shan, J. Zhang, S.P. Wang, X.B. Ma, Efficient tuning of surface copper species of Cu/SiO2 catalyst for hydrogenation of dimethyl oxalate to ethylene glycol, Chem. Eng. J. 313 (2017) 759–768. [20] C. Van Der Grift, A. Wielers, B. Jogh, J. Van Beumun, M. De Boer, M. VersluijsHelder, J. Geus, Effect of the reduction treatment on the structure and reactivity of silica-supported copper particles, J. Catal. 131 (1991) 178–189. [21] G. Chinchen, C. Hay, H. Vandervell, K. Waugh, The measurement of copper surface areas by reactive frontal chromatography, J. Catal. 103 (1987) 79–86. [22] E.W.H. Ralf Schmidt, Michael StiScker, Duncan Akporiaye, O.H. Ellestad, Pore size determination of MCM-41 mesoporous materials by means of 1H NMR spectroscopy, N2 adsorption, and HREM. a preliminary study, J. Am. Chem. Soc. 117 (1995) 4049–4056. [23] R. Ryoo, C.H. Ko, M. Kruk, V. Antochshuk, M. Jaroniec, Block-copolymertemplated ordered mesoporous silica: array of uniform mesopores or mesopore-micropore network?, J Phys. Chem. B 104 (2000) 11465–11471. [24] Y. Zhao, Y. Zhang, Y. Wang, J. Zhang, Y. Xu, S. Wang, X. Ma, Structure evolution of mesoporous silica supported copper catalyst for dimethyl oxalate hydrogenation, Appl. Catal. A – Gen. 539 (2017) 59–69. [25] P. Burattin, M. Che, C. Louis, Molecular approach to the mechanism of deposition-precipitation of the Ni(II) phase on silica, J. Phys. Chem. B 102 (1998) 2722–2732. [26] J. Fan, C. Yu, F. Gao, J. Lei, B. Tian, L. Wang, Q. Luo, B. Tu, W. Zhou, D. Zhao, Cubic mesoporous silica with large controllable entrance sizes and advanced adsorption properties, Angew. Chem. Int. Ed. Engl. 42 (2003) 3146–3150. [27] Y. Shi, S.P. Wang, X.B. Ma, Microwave preparation of Ti-containing mesoporous materials. Application as catalysts for transesterification, Chem. Eng. J. 166 (2011) 744–750. [28] G. Diaz, R. Perez-Hernandez, A. Gomez-Cortes, M. Benaissa, R. Mariscal, J.L.G. Fierro, CuO-SiO2 sol-gel catalysts: Characterization and catalytic properties for NO reduction, J. Catal. 187 (1999) 1–14. [29] A.J. Marchi, J.L.G. Fierro, J. Santamaría, A. Monzón, Dehydrogenation of isopropylic alcohol on a Cu/SiO2 catalyst: a study of the activity evolution and reactivation of the catalyst, Appl. Catal. A – Gen. 142 (1996) 375–386. [30] C.J.G. Van Der Grift, P.A. Elberse, A. Mulder, J.W. Geus, Preparation of silicasupported copper catalysts by means of deposition-precipitation, Appl. Catal. 59 (1990) 275–289. [31] A. Saadi, Z. Rassoul, M.M. Bettahar, Gas phase hydrogenation of benzaldehyde over supported copper catalysts, J. Mol. Catal. A – Chem. 164 (2000) 205–216. [32] K. Lanasri, A. Saadi, K. Bachari, D. Halliche, O. Cherifi, Gas phase hydrogenation of benzaldehyde over supported copper catalysts. Effect of copper loading, in: A. Gedeon, P. Massiani, F. Babonneau, (Eds.), Zeolites and Related Materials: Trends, Targets and Challenges, Proceedings of the 4th International Feza Conference, Elsevier Science Bv, Amsterdam, 2008, pp. 1279–1282. [33] J.L. Gong, H.R. Yue, Y.J. Zhao, S. Zhao, L. Zhao, J. Lv, S.P. Wang, X.B. Ma, Synthesis of ethanol via syngas on Cu/SiO2 catalysts with balanced Cu0-Cu+ sites, J. Am. Chem. Soc. 134 (2012) 13922–13925. [34] K.V.R. Chary, G.V. Sagar, C.S. Srikanth, V.V. Rao, Characterization and catalytic functionalities of copper oxide catalysts supported on zirconia, J. Phys. Chem. B 111 (2007) 543–550. [35] Y. Wang, Y. Shen, Y. Zhao, J. Lv, S. Wang, X. Ma, Insight into the balancing effect of active Cu species for hydrogenation of carbon-oxygen bonds, ACS Catal. 5 (2015) 6200–6208. [36] A. Yin, X. Guo, W.L. Dai, K. Fan, The nature of active copper species in Cu-HMS catalyst for hydrogenation of dimethyl oxalate to ethylene glycol: New insights on the synergetic effect between Cu0 and Cu+, J. Phys. Chem. C 113 (2009) 11003–11013. [37] X. Zheng, H. Lin, J. Zheng, X. Duan, Y. Yuan, Lanthanum oxide-modified Cu/SiO2 as a high-performance catalyst for chemoselective hydrogenation of dimethyl oxalate to ethylene glycol, ACS Catal. 3 (2013) 2738–2749. [38] J. Ding, J. Zhang, C. Zhang, K. Liu, H. Xiao, F. Kong, J. Chen, Hydrogenation of diethyl oxalate over Cu/SiO2 catalyst with enhanced activity and stability: Contribution of the spatial restriction by varied pores of support, Appl. Catal. A – Gen. 508 (2015) 68–79. [39] H. Yue, Y. Zhao, L. Zhao, J. Lv, S. Wang, J. Gong, X. Ma, Hydrogenation of dimethyl oxalate to ethylene glycol on a Cu/SiO2/cordierite monolithic
Y. Zhao et al. / Journal of Catalysis 357 (2018) 223–237
[40]
[41]
[42]
[43]
[44]
[45]
catalyst: Enhanced internal mass transfer and stability, AIChE J. 58 (2012) 2798–2809. D. Zhao, Q. Huo, J. Feng, B.F. Chmelka, G.D. Stucky, Nonionic triblock and star diblock copolymer and oligomeric surfactant syntheses of highly ordered, hydrothermally stable, mesoporous silica structures, J. Am. Chem. Soc. 120 (1998) 6024–6036. J. Fan, C.Z. Yu, J. Lei, Q. Zhang, T.C. Li, B. Tu, W.Z. Zhou, D.Y. Zhao, Lowtemperature strategy to synthesize highly ordered mesoporous silicas with very large pores, J. Am. Chem. Soc. 127 (2005) 10794–10795. G.C. Ma, X.Q. Yan, Y.L. Li, L.P. Xiao, Z.J. Huang, Y.P. Lu, J. Fan, Ordered nanoporous silica with periodic 30–60 nm pores as an effective support for gold nanoparticle catalysts with enhanced lifetime, J. Am. Chem. Soc. 132 (2010) 9596–9597. C. Baldizzone, S. Mezzavilla, H.W. Carvalho, J.C. Meier, A.K. Schuppert, M. Heggen, C. Galeano, J.D. Grunwaldt, F. Schuth, K.J. Mayrhofer, Confined-space alloying of nanoparticles for the synthesis of efficient PtNi fuel-cell catalysts, Angew. Chem. Int. Ed. Engl. 53 (2014) 14250–14254. M.N. Pahalagedara, L.R. Pahalagedara, C.H. Kuo, S. Dharmarathna, S.L. Suib, Ordered mesoporous mixed metal oxides: remarkable effect of pore size on catalytic activity, Langmuir 30 (2014) 8228–8237. Y. Zhao, J. Zhou, J. Zhang, D. Li, S. Wang, Selective hydrogenation of benzene to cyclohexene on a Ru/Al2O3/cordierite monolithic catalyst: Effect of mass
[46] [47] [48]
[49] [50]
[51]
[52]
[53]
237
transfer on the catalytic performance, Ind. Eng. Chem. Res. 47 (2008) 4641– 4647. H.S. Fogler, Essentials of Chemical Reaction Engineering, Prentice-Hall International, New Jersey, 2009. R. Venkatesan, H.S. Fogler, Comments on analogies for correlated heat and mass transfer in turbulent flow, AIChE J. 50 (2004) 1623–1626. L. Ronchin, L. Toniolo, Selective hydrogenation of benzene to cyclohexene catalyzed by Ru supported catalysts: influence of the alkali promoters on kinetics, selectivity and yield, Catal. Today 66 (2001) 363–369. M.A. Vannice, Kinetics of Catalytic Reactions, Springer Science+Business Media, New York, 2005. J.A. Moulijn, M. Makkee, R.J. Berger, Catalyst testing in multiphase micropacked-bed reactors; criterion for radial mass transport, Catal. Today 259 (2016) 354–359. M.J. Wulfers, R.F. Lobo, Assessment of mass transfer limitations in oligomerization of butene at high pressure on H-beta, Appl. Catal. A – Gen. 505 (2015) 394–401. N. Sudheesh, S.K. Sharma, M.D. Khokhar, R.S. Shukla, Kinetic investigations on the modified chitosan catalyzed solvent-free synthesis of jasminaldehyde, J. Mol. Catal. A – Chem. 339 (2011) 86–91. B.E. Poling, J.M. Prausnitz, J.P. O’Connell, Properties of Gases and Liquids, McGraw-Hill, New York, 2001.