Catalytic conversion of carbon dioxide into dimethyl carbonate using reduced copper-cerium oxide catalysts as low as 353 K and 1.3 MPa and the reaction mechanism

Synthesis of dimethyl carbonate (DMC) from CO2 and methanol under milder reaction conditions was performed using reduced cerium oxide catalysts and reduced copper-promoted Ce oxide catalysts. Although the conversion of methanol was low (0.005–0.11%) for 2 h of reaction, DMC was synthesized as low as 353 K and at total pressure of as low as 1.3 MPa using reduced Cu–CeO2 catalyst (0.5 wt% of Cu). The apparent activation energy was 120 kJ mol−1 and the DMC synthesis rates were proportional to the partial pressure of CO2. An optimum amount of Cu addition to CeO2 was 0.1 wt% for DMC synthesis under the conditions at 393 K and total pressure of 1.3 MPa for 2 h (conversion of methanol: 0.15%) due to the compromise of two effects of Cu: the activation of H2 during reduction prior to the kinetic tests and the block (cover) of the surface active site. The reduction effects in H2 were monitored through the reduction of Ce4+ sites to Ce3+ based on the shoulder peak intensity at 5727 eV in the Ce L3-edge X-ray absorption near-edge structure (XANES). The Ce3+ content was 10% for reduced CeO2 catalyst whereas it increased to 15% for reduced Cu–CeO2 catalyst (0.5 wt% of Cu). Moreover, the content of reduced Ce3+ sites (10%) associated with the surface O vacancy (defect sites) decreased to 5% under CO2 at 290 K for reduced Cu–CeO2 catalyst (0.1 wt% of Cu). The adsorption step of CO2 on the defect sites might be the key step in DMC synthesis and thus the DMC synthesis rate dependence on the partial pressure of CO2 was proportional. Subsequent H atom subtraction steps from methanol at the neighboring surface Lewis base sites should combine two methoxy species to the adsorbed CO2 to form DMC, water, and restore the surface O vacancy.


INTRODUCTION
Carbon dioxide is one of major green house gases. The conversion of CO 2 has been widely investigated to reduce the atmospheric concentration of CO 2 (Izumi, 2013). In the viewpoint of global warming, fixation methods of CO 2 and/or converted compounds from CO 2 are also critical. Transferring captured CO 2 to the bottom of the sea in a supercritical state is partially in practical use, but it incurs huge investment costs. The conversion of CO 2 to dimethyl carbonate (DMC) is attractive because DMC can be used as an electrolytic solution of lithium ion battery, methylating reagent, and feedstock for engineering plastics (Ono, 1997).
In this context, the conversion of CO 2 and methanol to DMC under milder conditions was studied and the mechanism was investigated by X-ray absorption near-edge structure (XANES). Methanol could be synthesized photocatalytically from CO 2 ( (Ahmed et al., 2011(Ahmed et al., , 2012; Morikawa et al. under review). In future, the DMC synthesis reported in this work could be combined with photocatalysis to synthesize DMC from CO 2 as a single starting material.

PREPARATION OF CeO 2
Cerium oxide samples were prepared from cerium nitrate hexahydrate (Wako Pure Chemical, >98.0%). It was dissolved in deionized water (<0.06 μS cm −1 ) to make the concentration to 0.2 mol L −1 . A 5% ammonia aqueous solution (Wako Pure Chemical) was added to the solution to reach the pH 10. Obtained yellow precipitate was filtered using a polytetrafluoroethene-based membrane filter (Omnipore JGWP04700, Millipore) with a pore size of 0.2 μm and washed several times with deionized water. The obtained powder was calcined in air at 673 K for 4 h. Then, the powder was connected to a vacuum system using rotary and diffusion pumps (10 −6 Pa) and the temperature was elevated at a ramping rate of 5 K min −1 from 290 to 673 K and kept at 673 K for 1 h.
A part of freshly-prepared CeO 2 above was reduced under 25 kPa of hydrogen. The temperature was elevated from 290 to 673 K with the ramping rate of 10 K min −1 and maintained at 673 K for 1 h.

PREPARATION OF Cu-CeO 2
3.8-950 mg of copper nitrate trihydrate (Wako Pure Chemical, >99.9%) was dissolved in 10 mL of deionized water. The Cu 2+ solution was added to 1.0 g of CeO 2 powder prepared in section Preparation of CeO 2 . Then, 25% of ammonia aqueous solution was added to the suspension until the pH reached 9.5. The mixture was reacted at 290 K for 1 h with magnetically stirred at a rate of 300 rpm. The color of precipitate was yellow, yellow green, and dark brown when the Cu content was 0.1, 1, and 20 wt%, respectively. The precipitate was filtered using JGWP04700 membrane and washed by several times with deionized water. The obtained powder was dried at 353 K for 12 h and denoted Cu-CeO 2 . The Cu-CeO 2 samples were treated under H 2 (25 kPa). The temperature was elevated from 290 to 673 K with the ramping rate of 10 K min −1 , maintained at 673 K for 1 h, and evacuated (10 −6 Pa) at 673 K for 30 min.

CHARACTERIZATION
Nitrogen adsorption isotherm measurements were performed at 77 K within the pressure range 1.0-90 kPa in a vacuum system connected to diffusion and rotary pumps (10 −6 Pa) and equipped with a capacitance manometer (Models CCMT-1000A and GM-2001, ULVAC). The Brunauer-Emmett-Teller (BET) surface area (S BET ) was calculated on the basis of eight-point measurements between 10 and 46 kPa (P/P 0 = 0.10-0.45) on the adsorption isotherm. The sample was evacuated at 423 K for 90 min before the measurements.
The electronic state of cerium in catalysts was investigated by the synchrotron X-ray measurements. The catalyst powder samples were prepared in vacuum (10 −6 Pa) and transferred directly to a Pyrex glass cell equipped with 50 μm-thick Kapton (Dupont) windows on both sides. The samples in N 2 (60 kPa) or CO 2 gas (60 kPa) were sealed with fire and transported to beamline.
Ce L 3 -edge X-ray absorption fine structure (XAFS) spectra were measured at 290 K in a transmission mode in the Photon Factory at the High-Energy Accelerator Research Organization (Tsukuba, Japan) on beamline 9C and also in SPring-8 (Sayo, Japan) on beamline 01B1. The X-ray energy was calibrated at the first intense peak top energy (5731.1 eV) for the Ce L 3 -edge spectrum of CeO 2 . The XAFS data were analyzed with a software package XDAP version 2.2.7 (Vaarkamp et al., 2006).

DMC SYNTHESIS TESTS
An autoclave (Taiatsu Glass Kogyo, inner volume 120 mL; Model TVS-N2-120) was used for the DMC synthesis tests using CeO 2 and Cu-CeO 2 catalysts. The inner space of autoclave was purged with argon gas (>99.998%) at a rate of 300 mL min −1 . 10 mL of dehydrated methanol (Kanto Chemical, 99.8%) and 100 mg of catalyst were introduced in the autoclave with the flow of Ar, not in contact with air. Next, CO 2 was flowed at a rate of 300 mL min −1 for 15 min. The outlet valve of reactor was closed and the interior pressure was increased to 2.0, 1.0, 0.50, 0.10 MPa at 290 K. Then, the temperature of the reactor was elevated from 290 to 393, 373, or 353 K with the ramping rate of 4 K min −1 , and maintained at the destination temperature for 2-6 h. As a comparison, 10 mL of dehydrated methanol, 50 mg of catalyst, and 3.6 MPa of CO 2 were introduced in the autoclave in similar way at 290 K. Then, the temperature of reactor was elevated from 290 to 403 K with the ramping rate of 4 K min −1 , and maintained at 403 K for 2 h. The reaction suspension was filtered using a JGWP04700 membrane. The filtrate was analyzed by gas chromatograph equipped with frame ionization detector (Model GC-18A, Shimadzu) equipped with a capillary column Ultra ALLOY-5 (Frontier Laboratories; inner diameter 250 μm, length 30 m). The conversion (%) of methanol to DMC was defined as Conversion(%) = 2 × molar amount of DMC formed molar amount of methanol introduced × 100.

Pretreatment effects in H 2
In the test under the 2.8 MPa CO 2 (initial pressure) at 393 K for 6 h, DMC formation rate using incipient CeO 2 catalyst was 0.44 mmol h −1 g cat −1 and that for reduced CeO 2 was 0.70 mmol h −1 g cat −1 (Table 1A). Total initial pressure of CO 2 and methanol was 3.5 MPa at 393 K. By the pretreatment effect in H 2 at 673 K, the synthesis rate increased by a factor of 1.6 times. The conversion (%) of methanol to DMC was improved to 0.33% ( Table 1A).

Effects of the Cu addition
The DMC synthesis rate using reduced Cu-CeO 2 (0.5 wt% Cu) was compared to that using reduced CeO 2 at 393 K for 2 h. The rates were 1.8 and 1.5 mmol h −1 g cat −1 , respectively (Table 1B). By the inclusion of 0.5 wt% of Cu in the catalyst, the rate increased by a factor of 1.2 times. The conversion of methanol to DMC was improved to 0.32% (Table 1B).
Next, the effects of Cu addition were investigated by progressively changing the Cu content between 0 and 20 wt% under  (reduced CeO 2 ) to 0.95 mmol h −1 g cat −1 (Table 1C; Figure 1). However, further increase of Cu content between 0.3 and 1 wt% in Cu-CeO 2 catalysts was not effective compared to the test results for the Cu-CeO 2 catalyst, 0.1 wt% of Cu (Table 1C). When the Cu content was between 1 and 20 wt%, the synthesis rates gradually approached to constant, similar to the one for undoped CeO 2 (0.41 mmol h −1 g cat −1 ) (Table 1C; Figure 1).

Reaction pressure effects
In the reaction tests at 393 K using reduced Cu-CeO 2 catalyst (0.5 wt% Cu), partial pressure of CO 2 introduced at 290 K was varied between 2.0 and 0.10 MPa. The partial (initial) pressure of CO 2 increased to between 2.8 and 0.13 MPa at reaction temperature of 393 K ( Table 1D). The DMC synthesis rates were plotted as a function of initial pressure of CO 2 and initial total pressure of CO 2 + methanol at 393 K (Figure 2). The DMC synthesis was possible under the total pressure of 1.3 MPa, but the amount of produced DMC was below detection limit (3 μmol) under the total pressure of 0.77 MPa at 393 K ( Table 1D). The DMC synthesis rates were proportional to partial pressure of CO 2 (Figure 2).
The reaction pressure effects were also tested under severer reaction conditions: at 403 K and 6.6 MPa using Cu-CeO 2 catalysts (0.1 and 0.5 wt% Cu) (Table 1o,p). The synthesis rates (3.1-1.9 mmol h −1 g cat −1 ) were higher by a factor of 3.3-3.2 times compared to corresponding test results at 393 K and 1.3 MPa (Table 1C,D). Thus, major reason of relatively low DMC synthesis rates in this paper was mild reaction conditions at 393-353 K and 3.5-1.3 MPa.

Reaction temperature effects
Further, the reaction temperature was varied between 393 and 353 K using reduced Cu-CeO 2 catalyst (0.5 wt% Cu) under the CO 2 partial pressure of 2.0 MPa introduced at 290 K. The CO 2 pressure increased to between 2.8 and 2.5 MPa at reaction temperatures of 393-353 K ( Table 1E). Under the total pressure of 2.7 MPa, DMC synthesis was possible as low as 353 K: 0.031 mmol h −1 g cat −1 (Table 1E; Figure 3). The apparent activation energy was estimated to 120 kJ mol −1 based on the Arrhenius plot (Figure 3, inset).

BET SURFACE AREA AND Ce L 3 -EDGE XANES
The BET surface area was 78 and 94 m 2 g cat −1 for Cu-CeO 2 samples consisting of 0.1 and 0.5 wt% Cu, respectively ( Table 2).
Ce L 3 -edge XANES spectra taken for Ce-based catalysts and also standard Ce compounds were depicted in Figure 4. Twin peaks appeared at 5731 and 5738 eV in the XANES spectrum for as-synthesized CeO 2 (spectrum a), indicating that valence state of Ce 4+ was predominant (Zhang et al., 2004).
When the CeO 2 was reduced under hydrogen at 673 K (spectrum b), a shoulder peak gradually grew at 5727 eV, on the lower side of peak at 5731 eV. As an intense whiteline peak appeared at 5726.7 eV in the spectrum for Ce III (NO 2 ) 3 ·6H 2 O (Figure 4A-f), the shoulder peak for spectrum b suggested the partial reduction of initial Ce 4+ to Ce 3+ . The partial reduction of CeO 2 is traditionally known to promote electron-donating catalysis, e.g., ammonia synthesis (Izumi et al., 1996;Aika et al., 1997). The spectrum b was fitted with the spectrum a for fresh CeO 2 and spectrum f for Ce(NO 2 ) 3 ·6H 2 O by changing the mixing ratio of standard spectra (Izumi and Nagamori, 2000;   . Spectra a and f were used as models of Ce 4+ and Ce 3+ states, respectively. The goodness of fit was evaluated based on the residual-factor (R f ) The spectrum b was best fitted with the mixing ratio of Ce 4+ :Ce 3+ = 90 : 10 ( Figure 4B-b).
The shoulder peak at 5727 eV also appeared in the XANES spectrum for reduced Cu-CeO 2 (0.5 wt% Cu; spectrum c), and the intensity was greater than that in spectrum b for reduced CeO 2 . The spectrum c was also fitted with the spectrum a (Ce 4+ ) and spectrum f (Ce 3+ ) by changing the mixing ratio. The best fit was realized with the mixing ratio was Ce 4+ :Ce 3+ = 85:15 (Figure 4B-c).
The Ce L 3 -edge XANES spectrum for reduced Cu-CeO 2 catalyst (0.1 wt% Cu; spectrum d) was essentially identical with that for reduced CeO 2 catalyst (spectrum b). The best-fit ratio with the mixing standard spectrum component of Ce 4+ and Ce 3+ was 90 and 10%. However, the shoulder peak at 5727 eV in the spectrum significantly weakened when 60 kPa of CO 2 was introduced to the reduced Cu-CeO 2 (0.1 wt% Cu) at 290 K (spectrum e). The bestfit ratio with the mixing standard spectrum component of Ce 4+ and Ce 3+ was 95 and 5%. The decrease of shoulder peak intensity at 5727 eV suggested re-oxidation of Ce 3+ to Ce 4+ by the reaction with CO 2 .

DMC SYNTHESIS UNDER MILDER CONDITIONS
DMC synthesis from CO 2 and methanol was reported at a synthesis rate of 1.8-5.1 mmol h −1 g cat −1 using CeO 2 at 403 K and 8.7 MPa for 2-4 h (Yoshida et al., 2006). DMC synthesis rate from CO 2 and methanol in this work using reduced CeO 2 at 393 K and 3.5 MPa for 6 h was lower: 0.70 mmol h −1 g cat −1 (Table 1A) due to lower reaction temperature and lower pressure. Because the forward reaction reduces the molar amount of materials in system from three to two and is uphill reaction (Pacheco and Marshall, 1997) (Equation 1), reaction conditions of lower reaction temperature and lower pressure are disadvantageous for the DMC synthesis reaction. The synthesis rate was enhanced by a factor of 1.6 times by the pre-reduction in H 2 for CeO 2 ( Table 1A).
The disadvantage of moderate reaction conditions was compensated by the Cu addition to CeO 2 catalysts. At 393 K and 3.5 MPa for 2 h, the DMC synthesis rates increased to 1.8 mmol h −1 g cat −1 by the addition of 0.5 wt% of Cu (Table 1B).
The effects of Cu addition to the DMC synthesis rates were compared at even milder reaction conditions: at 393 K and 1.3 MPa for 2 h (Table 1C). Under the reaction conditions, the 0.1 wt% of Cu was most effective and it promoted the synthesis rate by a factor of 2.3 times (Figure 1). One of the plausible explanations is that the positive effects to induce the Ce 4+ site reduction to facilitate H 2 dissociation and spillover on the catalyst surface and negative effects to block (cover) the surface active sites for DMC synthesis, e.g., the adsorption/activation sites for methanol, compromised to make a synthesis rate maximum at the Cu amount of 0.1 wt%.
The maximal DMC synthesis rate using Cu-CeO 2 catalyst (0.1 wt% Cu) at 393 K and 1.3 MPa was 0.95 mmol h −1 g cat −1 (Table 1f), but the rate at 403 K and 6.6 MPa was quite higher (3.1 mmol h −1 g cat −1 ), nearly equivalent to those in literature using CeO 2 (1.8-5.1 mmol h −1 g cat −1 ) at even severe conditions (403 K and 8.7 MPa) (Yoshida et al., 2006), demonstrating the effects of pre-reduction and/or the Cu addition to CeO 2 found in this work. The S BET values (78-94 m 2 g cat −1 ; Table 2) for Cu-CeO 2 catalysts (0.1-0.5 wt% Cu) were also similar to those for CeO 2 catalyst reported (80 m 2 g cat −1 ) (Yoshida et al., 2006).
The effects of ZrO 2 mixed with CeO 2 (Tomishige et al., 2001;Zhang et al., 2011b) were also interpreted to enhance the redox chemistry between Ce 4+ and Ce 3+ . In this sense, the redox of Cu + and Cu 2+ may enhance the redox between Ce 4+ and Ce 3+ . We tested Co-CeO 2 catalyst under the reaction condition of Table 1d. The conversion of methanol to DMC was 0.23% (not listed), slightly inferior to Cu-CeO 2 catalyst. Furthermore, in our preliminary results, the conversions to DMC using Fe-CeO 2 and Ni-CeO 2 catalysts were nearly equivalent to that using Co-CeO 2 catalyst. Thus, the effects of hydrogen activation and/or redox of added metal (Fe, Co, Ni, or Cu) to CeO 2 may work in similar way: enhancing effects of mixed metal ions and/or activating effects of hydrogen during pretreatment.
The reactions at lower reaction temperatures were tested (Table 1E). At 2.7 MPa using Cu-CeO 2 catalyst (0.5 wt% of Cu), DMC was formed at as low as 353 K ( Table 1E). The temperature dependence of DMC synthesis rates nicely followed the Arrhenius equation to give the apparent activation energy: 120 kJ mol −1 (Figure 3, inset). Similar range of apparent activation energy (107 kJ mol −1 ) was obtained using homogeneous Sn catalysts in the temperature range of 357-403 K (Kalhor et al., 2011). The dependences of DMC synthesis rates on the CO 2 pressure and total reactant pressure were also investigated at 393 K using Cu-CeO 2 catalyst (0.5 wt% of Cu; Table 1D). DMC was synthesized as low as 1.3 MPa.
The progress of catalysts to synthesize DMC from CO 2 is quite fast, especially under relatively mild conditions (see the Introduction section). The Cu-CeO 2 catalysts in this study are one of the good catalysts to work at relatively mild conditions. The dependence of synthesis rates on pressure and temperature ( Table 1) was interpreted based on X-ray spectroscopy in next section by monitoring the oxygen defect sites and Ce 3+ .

ACTIVE SITES OF DMC SYNTHESIS
The dependence of DMC synthesis on the CO 2 pressure (previous section) was proportional (Figure 2). This fact suggested that the key reaction step of DMC synthesis depended linearly on the CO 2 concentration.
To provide the insight into the surface reaction mechanism, the electronic state and structure for Ce sites were investigated using Ce L 3 -edge XANES. 10-15% of the Ce 4+ sites of FIGURE 5 | Proposed reaction mechanism of CO 2 conversion to DMC using partially-reduced Ce oxide surface.
as-prepared CeO 2 or Cu-CeO 2 (0.5 wt% of Cu) were reduced to Ce 3+ based on the shoulder peak intensity at 5727 eV (Figure 4b,c). Similarly, 10% of the Ce 4+ sites of as-prepared Cu-CeO 2 (0.1 wt% of Cu) were reduced to Ce 3+ under H 2 at 673 K, but a half of the Ce 3+ sites were re-oxidized to Ce 4+ by the introduction of CO 2 at 290 K (Figure 4d,e). These changes in the XANES spectra suggested the adsorption of CO 2 at the surface defect sites over the catalyst and the associated, neighboring Ce 3+ sites to the defects were re-oxidized to Ce 4+ . This reduction and re-oxidation mechanism was already reported on CeO 2 layers grown over Cu(111) surface (Staudt et al., 2010).
The proposed reaction mechanism was shown in Figure 5. Based on the dependence of DMC synthesis rates on the CO 2 pressure and the change of a shoulder peak at 5727 eV in the Ce L 3 -edge XANES spectra, O vacancy was assumed as defect site and worked to adsorb CO 2 . The population of O vacancy should increase by the reduction in H 2 and/or by the presence of Cu sites in catalysts. In order to synthesize DMC, surface Lewis base sites are required to subtract H atom from methanol ( Figure 5). If each H atom was subtracted at the Lewis base site from two methanol molecules, DMC and water molecules are formed to restore an O vacancy site.

CONCLUSIONS
Reduction of CeO 2 and Cu-promoted CeO 2 catalysts in hydrogen at 673 K was effective to enhance the DMC synthesis from CO 2 and methanol by a factor of 1.6-1.2 times. Added Cu worked cooperatively with CeO 2 catalysts as it accelerated the partial reduction of Ce 4+ sites to Ce 3+ . At the same time, doped Cu sites may block surface active sites. As a compromise, the DMC synthesis rate was maximal: 0.95 mmol h −1 g cat −1 at 393 K and 1.3 MPa (total pressure) in 2 h when the Cu amount was 0.1 wt% for reduced Cu-CeO 2 catalyst. The DMC synthesis was possible at the reaction temperature as low as 353 K (2.7 MPa) using the reduced Cu-CeO 2 catalyst. The apparent activation energy was calculated to be 120 kJ mol −1 . Based on the Ce L 3 -edge XANES, 10% of Ce sites were reduced to Ce 3+ by the reduction in H 2 for Cu-CeO 2 (0.1 wt% of Cu) while half of them were re-oxidized to Ce 4+ by the introduction of CO 2 at 290 K. A linear rate dependence on CO 2 pressure and the re-oxidation in CO 2 suggest that the adsorption of CO 2 might be the key step in DMC synthesis. H subtraction from methanol needs to occur at the neighboring sites of adsorbed CO 2 . Two methoxy groups and adsorbed CO 2 combine then to form DMC and water and restores surface O vacancy (defect site).