Efficient hydrogenation of dimethyl oxalate to ethylene glycol via nickel stabilized copper catalysts

Abstract

CuNi/SiO₂ nanocatalysts with Ni-stabilized Cu nanoparticles of around 10 nm were synthesized. After H₂ reduction, the catalysts with grain size of around 25 nm showed very high performance in the catalytic hydrogenation of dimethyl oxalate to ethylene glycol under mild reaction conditions. The composition and structure of these nanocatalysts were characterized. This study showed that Ni played a key role in stabilizing Cu against deactivation. To meet the requirements of industrial application, the optimal CuNi/SiO₂ nanocatalyst was tested under continuous reaction for over 2000 hours. The conversion and product selectivity were maintained at 99% and above 95%, respectively.

---

Introduction

With the depletion of crude oil resources and increasing environmental concerns, development of alternative processes that are both economical and energy-efficient is crucial.¹ Producing chemicals from syngas is one of the most promising routes. Ethylene glycol (EG),² as one of the most widely produced and consumed chemicals,³ has been intensely studied in recent years as a product of syngas hydrogenation.^4–8 Industrial processing of syngas to EG (STE) mainly involves three steps: (1) obtaining high-purity carbon monoxide (CO), (2) coupling of CO to dimethyl oxalates (DMO)⁹ using nitrite esters and (3) selective hydrogenation of DMO to EG.^10,11 Selective hydrogenation of DMO to EG is the key reaction in the STE route, and it has drawn much attention from the industry and academia. As shown in Scheme 1, DMO with three kinds of active groups (C=O, C–O and CH₃–O) could lead to some different reaction routes. Methyl glycolate (C₃H₆O₃) is the preliminary intermediate, which could lead to an equilibrium reaction with the side-product b² and then to the desired product, EG. Over hydrogenation of EG could lead to EtOH (b¹) easily, whereas acidic/basic catalysis could lead to an unexpected polymerization, such as b³ or b⁴ etc.¹² To meet industrial demands, the catalysts for hydrogenation of DMO to EG must possess some or all of the following properties: (1) high conversion rate and high selectivity for the formation of EG, (2) high stability for service life, and (3) running at no more than 270 °C to avoid an unexpected over-polymerization.¹³

Scheme 1 The proposed reaction pathways for DMO hydrogenation.

Cu-Based catalysts have been regarded as good candidates and are well investigated; they can trigger the reaction under mild conditions with high conversion, and produce less coke. Some of the previously reported works were performed with supported catalysts,^14–18 some used assisted agents^19–23 and others relied on the acid–base properties of the catalysts.^12,24 However, as Cu-based catalysts are very sensitive to reaction temperature and impurities in the raw material, their selectivity and stability are always a key problem for industrial application. Generally, the activity of Cu-based catalysts is extremely low below 200 °C and they are easily deactivated at temperatures over 230 °C due to the deposition of coke and aggregation Cu nanoparticles.²⁵ To meet industrial requirements, it is important to improve the activity by stabilizing the Cu catalyst or by decreasing the reaction temperature, as is done in a Ni-based bimetallic system.^26–28 However, efficient Cu-based catalysts are still hard to develop. In our previous work, we reported a facile method for the fabrication of copper–nickel surface sites on SiO₂ with high dispersity with grain size of around 10 nm in diameter. The Cu⁺–Ni^δO_x surface sites greatly improved the catalytic performance in the hydrogenation of LA or ethyl levulinate (EL) to γ-valerolactone under mild conditions.²⁹ Herein, we report the development of an STE system based on such a Cu–Ni–Si catalyst. This type of catalyst could maintain high activity (over 99% conversion and 95% selectivity) for over 2000 hours without decay, which conforms well with the requirements of industry.

Results and discussion

Fig. 1 show the H₂-TPR profiles of a series of CuNi/SiO₂ samples. For the CuNi(0)/SiO₂ sample, the temperature of the main reduction peak was 219.1 °C. With the introduction of Ni, the main reduction peak gradually shifted to lower temperature; the lowest temperature was 211.5 °C, recorded when the Ni amount was 8%. This phenomenon illustrated that there existed some kind of interaction between Cu–Ni–Si, which had also been observed by Zhao et al.³⁰ This interaction made the Ni-containing catalysts more prone to get reduced than CuNi(0)/SiO₂, which did not contain Ni.

Fig. 1 H₂-TPR profiles of the as-prepared catalysts.

The pore size distribution curves (shown in Fig. 2) exhibited wide bimodal distributions around 15.0 nm and 4.6 nm when Ni was introduced. It can be proposed that the large volume around 15.0 nm could solve the problem of coking well. Interestingly, the nickel-modified Cu/SiO₂ had a new pore diameter of 4.6 nm, indicating that the textural structure of the catalyst was determined not only by the copper species but also by the nickel species. Moreover, the diameter of the wide peak distribution decreased from 16.2 nm to 13.7 nm, followed by an increase to 18.7 nm, which implied that there was a threshold for the amount of Ni introduced.

Fig. 2 Pore size distribution of the as-prepared catalysts.

Fig. 3a and b show a comparison of the X-ray diffraction (XRD) patterns of the as-prepared and reduced samples of CuNi(0–4)/SiO₂. The XRD patterns of the as-prepared sample without Ni showed characteristic peaks of CuO and chrysocolla Cu_2−xSi₂O₅(OH)_x structures. Cu and Cu₂O structures were obtained after reduction by H₂. With the introduction of Ni, the chrysocolla structure collapsed step by step, whereas all CuO structures remained when the amount of Ni was over 4%. Thus, the introduction of Ni clearly restricted the crystallization of chrysocolla. Since the peaks at around 37° were affiliated to both Cu₂O and NiO, and the peaks at around 43° were affiliated to Cu or NiO, we had to identify the structures of the Ni and Cu species based on the peaks at 50.4° and 62.8° that appeared after H₂ reductions. From the XRD results of the reduced samples, the peaks affiliated to metallic Cu were very obvious, but the peaks corresponding to copper oxides and nickel oxides were not so clear. Calculation by the Scherrer formula gave the particle sizes as 12.7, 16.9, 25.3, 25.7 and 25.9 nm for the samples in series from CuNi(0)/SiO₂ to CuNi(4)/SiO₂. One important observation was that the Cu particle size increased with an increase in the amount of Ni.

Fig. 3 XRD curves of (a) as-prepared and (b) H₂ reduced catalysts.

Fig. 4 displays the H₂ consumption by pulse method. By comparing the series of CuNi/SiO₂ catalysts, the consumption number for CuNi(0)/SiO₂, CuNi(1)/SiO₂, CuNi(2)/SiO₂ and CuNi(4)/SiO₂ were 7.7, 5.5, 4.5 and 3.4, respectively. The corresponding Cu dispersions of 7.5%, 5.6%, 4.7%, and 3.4%, respectively, could be calculated. The particle sizes were calculated as 14.3 nm, 17.8 nm, 23.0 nm and 31.0 nm, which were in well agreement with the XRD results.

Fig. 4 Pulse consumption of H₂ for the series of CuNi/SiO₂ catalysts.

Fig. 5 compares the valence state of nickel over the series of catalysts before and after reduction. As can be seen from Fig. 5a, virtually all nickel species in the four catalysts existed as Ni²⁺. After reduction by H₂, most of Ni remained in the 2+ oxidation state (Fig. 5b); however, one new small peak occurred at about 853 eV, implying that a few Ni²⁺ species had been partly reduced to Ni⁺. These Ni⁺ ions may have been able to act as stabilizers to the adjacent Cu species. Moreover, no peak was found at 852.3 eV in the XPS, which meant that no Ni⁰ existed in the reduced CuNi/SiO₂ catalysts. Fig. 6 shows the catalytic performance of the CuNi/SiO₂ catalysts with different ratios of Cu to Ni in gas-phase hydrogenation of DMO. Steady-state product compositions were obtained after 120 h on stream. It is known that the hydrogenation of DMO proceeds via methyl glycolate (MG) to EG, whereas EG can be dehydrated further to ethanol (Scheme 1). The reaction between EG and ethanol on the basic sites yields 1,2-BDO.³¹ As shown in Table 1, the conversion of DMO increased with an increase in the Ni content, from 97.33% on the CuNi(0)/SiO₂ to 99.99% on CuNi(2)/SiO₂ catalysts, followed by a slight decrease to 95.63% on the CuNi(4)/SiO₂ catalyst. Almost all the DMO were converted to MG and EG. The degree of over-hydrogenation to ethanol remained <1% and no 1,2-BDO was detected. Fig. 6 compares the activities versus the different particle sizes. Traditionally, smaller Cu nanoparticles result in a better catalytic performance. However, our results showed that the larger Cu particles of about 25 nm had much better catalytic performance than that of those with 14 nm size. In 2007, an innovative discovery was made regarding shape-dependent catalysis, which was considered as the new face of catalysis.³² Research showed that “different atom arrays naturally display crystallographically distinct facets, whose surface atoms have distinct catalytic activities”.³³ Our results implied that the introduction of Ni into Cu-based catalysts made the Cu surfaces more specific for the conversion of DMO to final products.

Fig. 5 Ni 2p_3/2 XPS diagram of the samples (a) before and (b) after reduction. Series of CuNi/SiO₂ catalysts with different Ni content are shown as different colors: black line, Ni(0); red line, Ni(1); blue line, Ni(2); green line, Ni(3); pink line, Ni(4).

Fig. 6 Effect of Ni content and Cu particles size on the catalytic performance of the CuNi/SiO₂ catalysts. Reaction conditions: 3.0 MPa, 215 °C, and H₂/DMO ratio 100 (mol mol⁻¹), LHSV_DMO = 0.55 h⁻¹.

Table 1 The DMO hydrogenation performance of catalysts with different Ni content^a

Catalyst	XDMO, %	SEG, %	SMG, %	SEtOH, %
a Note: X – conversion, S – selectivity. Reaction conditions: 215 °C, 3.0 MPa, H2/DMO = 100 (mol mol^−1), LHSVDMO = 0.55 h^−1. The data were obtained after reaction for 120 hours.
Cu–Ni(0)/SiO2	97.33	87.47	9.24	0.45
Cu–Ni(1)/SiO2	98.90	91.16	6.24	0.84
Cu–Ni(2)/SiO2	99.99	97.22	1.45	0.75
Cu–Ni(3)/SiO2	97.54	90.01	3.79	0.63
Cu–Ni(4)/SiO2	95.63	85.88	6.71	0.38

---

On the basis of XRD, XPS, chemisorption and reaction investigations, it could be speculated (as shown in Scheme 2) that without the introduction of nickel, Cu species are unstable during the reaction and easily aggregate to form larger grains due to the weak interaction between copper and silica. When small amounts of nickel were introduced, the Ni–O species could insert between Cu and Si, resulting in the stabilization of Cu species. With further increase in the nickel content, an optimized dispersion of Cu could be obtained based on the threshold of dispersing Ni–O species on silica. However, an excess nickel led to an aggregation of nickel oxides, on which the tiny copper sites had the chance of over growth.

Scheme 2 Schematic for the variation of Cu species with an increase in the nickel content.

Since the CuNi(2)/SiO₂ catalyst exhibited the highest hydrogenation efficiency, it was chosen for further observation and practical testing of the catalytic hydrogenation of DMO to EG. High resolution aberration-corrected scanning transmission electron microscopy (HRSTEM) images (Fig. 7a and b) revealed the active sites on the substrates were very fine nanoparticles with about 10 nm diameter. Fig. 7c–e show the EDX mapping analysis on 50 nm domain, which revealed that the copper and nickel were well mixed on the SiO₂ surface and the appearance of Cu was always coupled with the presence of Ni. The HRSTEM observation, therefore, was in agreement with our speculation in Scheme 2. Fig. 8 displays the detailed catalytic results using the optimized CuNi(2)/SiO₂ catalyst. During a 2000 hours test, we investigated conversion, temperature-dependent selectivity, stability and distribution of byproducts. The CuNi(2)/SiO₂ catalyst maintained over 99% conversion and 95% selectivity at a temperature range of 215 °C to 230 °C without any decay.

Fig. 7 (a) and (b) HRSTEM images and (c)–(e) STEM-EDS elemental mapping on a 50 nm domain of CuNi(2)/SiO₂ catalyst.

Fig. 8 Catalytic performance and stability test of optimized CuNi(2)/SiO₂ catalyst at from 215 °C to 230 °C, 3.0 MPa, H₂/DMO ratio 100 (mol mol⁻¹), LHSV_DMO = 0.55 h⁻¹. Conversion, selectivity and temperature are shown by black red and blue lines, respectively.

The nature of the catalyst after the 2000 hours test was then obtained by XRD and STEM. Fig. 9a compared the XRD curves of freshly H₂-reduced and reacted CuNi(2)/SiO₂ catalyst. The two traces had the same featured peaks, meaning that the reacted one kept its original structure unchanged. Shaper peaks implied that the particle size of the reacted sample was a little larger than that of the freshly reduced one. With the Scherrer formula, the particle size of the reacted sample was calculated to be 26.5 nm. Fig. 9b shows the STEM images. The average copper particle size was around 25 nm and separated obviously. Both results revealed that the catalyst retained its particle size and dispersibility, which corresponded to its super stability.

Fig. 9 (a) XRD curve and (b) STEM image of the reacted CuNi(2)/SiO₂ catalyst.

Conclusions

In conclusion, Cu-based model catalysts were explored by adding nickel to improve the performance for the reaction of DMO to EG. The appropriate amount of nickel was vital for the optimized catalysis. The Ni–O species effectively stabilized Cu and was ideal for use in industry, as shown by its over 99% conversion, 95% selectivity, and over 2000 hours' stability. This strategy for the synthesis of the stable mixed oxides provided a simple catalyst preparation and offered selectivity control. Results from the current study demonstrated the high potential of mixed metal oxide catalysts for practical application.

Experimental

The Cu-based catalysts were prepared by a co-precipitation method, in which 0.20 M aqueous solution of Cu(NO₃)₂·3H₂O, Ni(NO₃)₂·3H₂O and the proper amount of silica sol were taken and precipitated using 0.2 M aqueous sodium carbonate at an ambient temperature. The precipitate was aged further for 8 h at 60 °C. Then, the mixture was separated by filtration and washed with deionized water to remove traces of sodium. The solid, thus obtained, was dried in a static oven at 110 °C for 24 h and calcined at 400 °C for 4 h. The weight percentage of copper in the Cu/SiO₂ catalyst was held at 35%. The samples were named as CuNi(0)/SiO₂ without introducing Ni and CuNi(1–4)/SiO₂ with 2%, 4%, 6% and 8% Ni, respectively.

X-ray powder diffraction (XRD) patterns were obtained in the range of 5–80° by a Bruker D8 diffractometer using Cu Kα radiation with a scanning step 0.002°, voltage 40 kV, and current 100 mA. High resolution Transmission Electron Microscopy (TEM) images were obtained on a FEI TECNAI-20 instrument with an accelerating voltage of 200 kV. TEM specimens were prepared by dispersing the powder in alcohol by ultrasonic treatment and dropping onto a holey carbon film supported on a copper grid, and then dried in air. Scanning transmission electron microscopy (STEM) was performed on a double-corrected Titan Cubed 60-300 and a cold field emission gun was operated at 200 kV. STEM images were obtained using a high-angle annular dark-field (HAADF) detector. Temperature Programmed Reduction (TPR) was performed using a Micromeritics AutoChem 2950. N₂ adsorption (BET and pore size analysis) was performed using a Micromeritics ASAP2000M. The dispersion and metallic copper surface areas of the catalysts were determined by N₂O chemisorption at 333 K, according to the reports.³⁴ 100 mg of Cu-based catalysts was calcined at 623 K and reduced by 5% H₂-95% Ar (25 mL min⁻¹) at 623 K for 2 h and cooled to 333 K. Then, pure N₂O was introduced to oxidize Cu atoms completely on the surface. 2Cu(s) + N₂O → Cu₂O(s) + N₂. The quantity of irreversibly chemisorbed O₂ was measured by a hydrogen pulse chromatographic technique on a Micromeritics Autochem II 2920 equipped with a TCD. Hydrogen pulse-dosing was repeated until there was no change. The consumed amount of hydrogen was the value obtained by subtracting the small area of the first few pulses from the area of the other pulses. Copper loading of all the reduced catalysts was analyzed by X-ray fluorescence (XRF) on a Bruker S4 Pioneer. Copper dispersion was calculated by dividing the amount of chemisorption sites into the total supported copper atoms. X-ray photoelectron spectroscopy (XPS) studies were performed using a Kratos AXIS Ultra DLD spectrometer equipped with a high temperature gas reaction cell. All the spectra were obtained with monochromatic Al K_α. The C 1s peak at 284.8 eV was set as the reference for the binding energy calibration. All the spectrum processing and peak fittings were performed with CasaXPS.

All the reactions were carried out in a continuous flow, fixed-bed reactor and 5 g catalyst was packed in the reactor. Prior to the reaction, the catalyst was reduced at 260 °C in H₂ flow. DMO and H₂ were fed into the reactor using an injection pump, and the LHSV flow rate of DMO was 0.55 h⁻¹. The reaction temperature was adjusted between 210 °C and 230 °C. The reaction pressure was maintained at 3.0 MPa. The molar ratio of H₂ to DMO was set as 100 (mol mol⁻¹). The hydrogenation products were analyzed using gas chromatography (Agilent 7890) equipped with a flame ionization detector and a capillary column (DB-200). The main products and by-products were identified by GC-MS on Agilent 5975C inert XL EI/CI MSD.

Acknowledgements

The authors acknowledge the partially financial support from the SINOPEC, and the National Natural Science Foundation of China (NSFC 21373272 and 2013CB93410).

References

1. G. W. Huber, S. Iborra and A. Corma, Chem. Rev., 2006, 106, 4044–4098.
2. M. W. Forkner, J. H. Robson, W. M. Snelling, A. E. Martin, F. H. Murphy and T. E. Parsons, Glycols–Ethylene Glycols, 2004.
3. World Petrochemical (WP) report on Ethylene Glycol http://chemical.ihs.com/WP/Public/Reports/eg/.
4. A. Y. Yin, X. Y. Guo, W. L. Dai and K. N. Fan, Chem. Commun., 2010, 46, 4348–4350.
5. Z. He, H. Q. Lin, P. He and Y. Z. Yuan, J. Catal., 2011, 277, 54–63.
6. A. Yin, C. Wen, X. Guo, W. L. Dai and K. Fan, J. Catal., 2011, 280, 77–88.
7. A. Yin, X. Guo, W. Dai and K. Fan, J. Phys. Chem. C, 2009, 113, 11003–11013.
8. G. Xu, Y. Li, Z. Li and H. Wang, Ind. Eng. Chem. Res., 1995, 34, 2371–2378.
9. Z. Xu, J. Sun, C. Lin, X. Jiang, Q. Chen, S. Peng, M. Wang and G. Guo, ACS Catal., 2013, 3, 118–123.
10. L. Chen, P. Guo, M. Qiao, S. Yan, H. Li, W. Shen, H. Xu and K. N. Fan, J. Catal., 2008, 257, 172–180.
11. Y. Huang, H. Arig, X. Zheng, X. Duan, S. Takakusagi, K. Asakur and Y. Yuan, J. Catal., 2013, 307, 74–83.
12. Y. Song, J. Zhang, J. Lv, Y. Zhao and X. Ma, Ind. Eng. Chem. Res., 2015, 54, 9699–9707.
13. S. G. Hovenkamp and J. P. Munting, J. Polym. Sci., Part A: Gen. Pap., 1970, 8, 679–682.
14. J. Ding, J. Zhang, C. Zhang, K. Liu, H. Xiao, F. Kong and J. Chen, Appl. Catal., A, 2015, 508, 68–79.
15. C. Wen, A. Yin, Y. Cui, X. Yang, W. Dai and K. Fan, Appl. Catal., A, 2013, 458, 82–89.
16. A. Yin, X. Guo, W. Dai, H. Li and K. Fan, Appl. Catal., A, 2008, 349, 91–99.
17. Y. Zhu, Y. Zhu, G. Ding, S. Zhu, H. Zheng and Y. Li, Appl. Catal., A, 2013, 468, 296–304.
18. H. Lin, X. Duan, J. Zheng, X. Zheng, P. He, Y. Yuan and Y. Yang, RSC Adv., 2013, 3, 11782–11789.
19. S. Zhao, H. Yue, Y. Zhao, B. Wang, Y. Geng, J. Lv, S. Wang, J. Gong and X. Ma, J. Catal., 2013, 297, 142–150.
20. Z. He, H. Lin, P. He and Y. Yuan, J. Catal., 2011, 277, 54–63.
21. X. Zheng, H. Lin, J. Zheng, X. Duan and Y. Yuan, ACS Catal., 2013, 3, 2738–2749.
22. H. Miyazaki, T. Uda, K. Hirai, Y. Nakamura, H. Ikezawa and T. Tsuchie, US4585890, 1986.
23. A. Yin, J. Qu, X. Guo, W. Dai and K. Fan, Appl. Catal., A, 2011, 400, 39–47.
24. Y. Zhu, X. Kong, X. Li, G. Ding, Y. Zhu and Y. Li, ACS Catal., 2014, 4, 3612–3620.
25. B. Putrakumar, N. Nagaraju, V. P. Kumar and K. V. R. Chary, Catal. Today, 2015, 250, 209–217.
26. J. Zhang, M. Ibrahim, V. Collière, H. Asakura, T. Tanaka, K. Teramura, K. Philippot and N. Yan, J. Mol. Catal. A: Chem., 2016, 422, 188–197.
27. J. Zhang, J. Teo, X. Chen, H. Asakura, T. Tanaka, K. Teramura and N. Yan, ACS Catal., 2014, 4, 1574–1583.
28. S. De, J. Zhang, R. Luque and N. Yan, Energy Environ. Sci., 2016, 9, 3314–3347.
29. J. Zhu, Y. Tang and K. Tang, RSC Adv., 2016, 6, 87294–87298.
30. Y. Zhao, S. Zhao, Y. Geng, Y. Shen, H. Yue, J. Lv, S. Wang and X. Ma, Catal. Today, 2016, 276, 28–35.
31. C. Carlini, M. D. Girolamo and A. Macinai, J. Mol. Catal. A: Chem., 2003, 200, 137–145.
32. D. L. Feldheim, Science, 2007, 316, 699.
33. P. D. Yang, Nature, 2012, 41, 482–483.
34. C. J. G. Van Der Grift, A. F. H. Wielers, B. P. J. Jogh, J. Van Beunum, M. De Boer, M. Versluijs-Helder and J. W. Geus, J. Catal., 1991, 131, 178–188.

---