Mitigating the Poisoning Effect of Formate during CO2 Hydrogenation to Methanol over Co-Containing Dual-Atom Oxide Catalysts

During the hydrogenation of CO2 to methanol over mixed-oxide catalysts, the strong adsorption of CO2 and formate poses a barrier for H2 dissociation, limiting methanol selectivity and productivity. Here we show that by using Co-containing dual-atom oxide catalysts, the poisoning effect can be countered by separating the site for H2 dissociation and the adsorption of intermediates. We synthesized a Co- and In-doped ZrO2 catalyst (Co–In–ZrO2) containing atomically dispersed Co and In species. Catalyst characterization showed that Co and In atoms were atomically dispersed and were in proximity to each other owing to a random distribution. During the CO2 hydrogenation reaction, the Co atom was responsible for the adsorption of CO2 and formate species, while the nearby In atoms promoted the hydrogenation of adsorbed intermediates. The cooperative effect increased the methanol selectivity to 86% over the dual-atom catalyst, and methanol productivity increased 2-fold in comparison to single-atom catalysts. This cooperative effect was extended to Co–Zn and Co–Ga doped ZrO2 catalysts. This work presents a different approach to designing mixed-oxide catalysts for CO2 hydrogenation based on the preferential adsorption of substrates and intermediates instead of promoting H2 dissociation to mitigate the poisonous effects of substrates and intermediates.


■ INTRODUCTION
Hydrogenation of CO 2 to methanol is an effective way to recycle CO 2 , 1−4 because methanol can be transported in liquid phase and converted to chemicals and fuels. 5Achieving high methanol selectivity during CO 2 hydrogenation is a challenge because of competitive CO formation via the reverse water− gas shift (RWGS) reaction.In the first step of CO 2 hydrogenation, adsorbed formate species are formed, which are an intermediate for both methanol and CO.While methanol is obtained by hydrogenation of formate, its decomposition produces CO. 6−9 Therefore, to increase methanol selectivity, it is important to suppress CO formation by promoting the hydrogenation of formate.
Recently, oxide catalysts like In 2 O 3 and mixed oxides of ZrO 2 10−16 have emerged as catalysts for CO 2 hydrogenation with good methanol selectivity.These catalysts exhibit oxygen vacancies adjacent to a metal atom with hydrogen dissociation ability. 17−26 For example, catalysts containing atomically dispersed In and Zn on ZrO 2 cannot dissociate H 2 efficiently under a CO 2 environment. 27,28−39,41,42,46−48 Adding metal promoters for hydrogen dissociation creates independent active sites that catalyze the RWGS reaction and total reduction of CO 2 to methane. 49,50ere we introduce an alternative approach for mitigating the poisonous effect by creating a site for the adsorption of CO 2 and stabilization of formate by introducing a Co single atom into the ZrO 2 structure.This preferential adsorption of CO 2 at a site different from the metal center for H 2 dissociation reduces the poisoning effect and improves both methanol

Catalyst Preparation
Doped ZrO 2 catalysts were prepared by using the coprecipitation method.In a typical method to synthesize the Co−In−ZrO 2 catalyst, Co(NO 3 ) 2 •6H 2 O (1 mmol), In(NO 3 ) 3 •3H 2 O (0.5 mmol), and ZrO(NO 3 ) 2 •2H 2 O (8.5 mmol) were dissolved in a round-bottomed flask containing 150 mL of deionized water.Diluted aqueous NH 4 OH (9 wt % NH 3 ) solution was added dropwise to the metal solution under vigorous stirring until the pH became 9.The formed precipitate was aged for 1 h under stirring at room temperature.The precipitate was recovered by centrifugation and washed with deionized water until the pH of the supernatant became neutral.The solid was dried at 130 °C for 12 h.The final catalyst was obtained after calcination in a muffle furnace at 500 °C for 3 h (ramp rate of 2 °C min −1 ) under static air.Dual-atom oxide catalysts of Zn and Ga were prepared using the same method, and the catalysts were named Co-M-ZrO 2 (M = In/ Zn/Ga).Single-atom oxides of Co-doped ZrO 2 (Co−ZrO 2 ), Mdoped ZrO 2 (M-ZrO 2 ), and undoped ZrO 2 were prepared using the same procedure.Metal loadings were calculated in atom % of the respective metal in relation to the total metal content.The loadings of Co and M (In, Zn, Ga) were kept constant at 10 and 5 atom %, respectively.
For example, the loading of Co (L Co ) in atom % was calculated as i k j j j j j y where n Co , n M , and n Zr are the mol of Co, M, and Zr, respectively.

Catalytic Testing Procedure
Catalytic activity for CO 2 hydrogenation was evaluated in a custombuilt stainless-steel fixed-bed flow reactor system (Figure S9).Typically, 200 mg of catalyst was loaded into the reactor and held in place with quartz wool.A thermocouple was inserted into the reactor to measure the catalyst bed temperature.H 2 and CO 2 were supplied by two mass-flow controllers (Bronkhorst Japan K.K.) and mixed before the inlet of the reactor.The reactor was pressurized using a mixture of H 2 and CO 2 having the ratio H 2 /CO 2 = 4:1 via a backpressure regulator (Swagelok Japan FST Co., Ltd.).After the system pressure was stable, the reactor temperature was increased to the desired value.The total flow rate was 100 mL min −1 to maintain the gas-hourly space velocity at 30,000 mL h −1 g cat −1 . Reaction products and unreacted CO 2 were quantified online by a Shimadzu GC 8A gas chromatograph (Shimdazu Corp.) equipped with two channels and a thermal conductivity detector.Porapak Q and Molsieve 5 Å packed columns were used for the separation of CO 2 , MeOH, and CO, respectively.The gas line from the outlet of the reactor to the inlet of the GC was heated to 170 °C to prevent condensation of the liquid products.See Supporting Information for the mathematical equations for calculating CO 2 conversion, selectivity of products, and space-time yield of CO and MeOH.

Preparation and Structural Characterization of Oxide Catalysts
All of the doped oxides were prepared by coprecipitation of metal nitrates with ammonia solution followed by washing, drying, and calcination at 500 °C under air.The loading of Co and In in doped catalysts was 10 and 5 atom %, respectively.The surface concentration of Co for Co−In−ZrO 2 and Co− ZrO 2 was similar to the theoretical value at 11 atom % as measured by X-ray photoelectron spectroscopy (XPS) (Table S1).The surface concentration of In in Co−In−ZrO 2 was 11 atom %, roughly two times the theoretical loading.The surface concentration of In in In−ZrO 2 was 12 atom %, suggesting that In prefers to migrate toward the surface when doped in ZrO 2 .Consequently, the ratio of Co and In on the surface of Co−In−ZrO 2 was approximately 1:1.
Doping did not introduce any morphological changes and catalysts showed similar N 2 adsorption isotherms and surface areas in the range of 65−73 m 2 g −1 (Figure S1 and Table S1).In the X-ray diffraction (XRD) analysis (Figure 1a), diffraction peaks attributed to t-ZrO 2 were observed and peaks for oxides of dopants such as CoO, Co 3 O 4 , and In 2 O 3 were not apparent.The 2θ value corresponding to the (101) plane of t-ZrO 2 shifted to a higher value (Figure 1b) because of a decrease in .Arrhenius plot for the calculation of the apparent activation energy of (e) methanol formation and (f) CO formation over doped oxide catalysts.Determination of the order of methanol formation with respect to (g) H 2 and (h) CO 2 .The rate of products was taken in mol g cat −1 h −1 units for calculations.
interplanar distance in t-ZrO 2 owing to the smaller ionic radii of the dopants (r Co 2+ = 0.74 Å, r In 3+ = 0.8 Å) as compared to Zr 4+ (r Zr 4+ = 0.84 Å). 53 The lattice parameter of the t-ZrO 2 was reduced with respect to the ionic radii and concentration of the dopant (Figure S2).These results indicate homogeneous doping of metals in the ZrO 2 matrix without the formation of oxides of In and Co.
In the high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of Co−In−ZrO 2 only the t-ZrO 2 phase was visible (Figure 1c).Elemental mapping using energy dispersive X-ray (EDX) analysis (Figure 1d

Catalytic Performance of Oxide Catalysts
The performance of all catalysts for catalytic CO 2 hydrogenation was evaluated in a stainless-steel fixed-bed flow reactor (Figure S4).Undoped ZrO 2 showed negligible CO 2 conversion (Table S2).While the CO 2 conversion over the Co−In−ZrO 2 catalyst was slightly higher than that over binary oxide catalysts, the methanol selectivity was very different (Figure 2a and Table S2).At 270 °C the methanol selectivity over Co−In−ZrO 2 was 86%, whereas the methanol selectivity over In−ZrO 2 and Co−ZrO 2 was 58 and 39%, respectively.High methanol selectivity was achieved at the same CO 2 conversion level, further confirming the promotional effect of Co−In−ZrO 2 (Figure 2b).The difference in selectivity between dual-and single-atom-doped catalysts was maintained at 300 °C (Figure 2b and Table S2).The space-time yield (STY) of methanol over Co−In−ZrO 2 at 300 °C was 1.3 μmol g −1 s −1 , which was higher than the combined methanol STY of the binary oxides (Table S2).The selectivity and STY of CO were lower over Co−In−ZrO 2 in comparison to those over binary oxides (Table S2 and Figure S5).These results suggest that the methanol formation was favored over the Co−In− ZrO 2 catalyst, and the RWGS reaction was suppressed.Figure 2c shows the change in methanol selectivity at the same conversion level with respect to change in composition of Co and In, while keeping the total dopant loading constant at 15 atom %.Methanol selectivity was independent of total cation loading, and the highest methanol selectivity was obtained at an optimum loading of Co 10 atom % and In 5 atom %.It is evident that the presence of Co and In atoms in the vicinity of the ZrO 2 surface promoted methanol formation.The spent Co−In−ZrO 2 catalyst was analyzed using XRD and EDX analysis and no change in the catalyst structure was observed (Figure S6).For comparison, we also prepared a Co/In−ZrO 2 catalyst having metallic Co nanoparticles, which showed a high selectivity for methane during CO 2 hydrogenation (Figure S7), emphasizing the role of Co single atoms in avoiding the formation of side products.
Methanol formation in CO 2 hydrogenation is associated with the hydrogen utilization ability of the catalyst. 54Changing the H 2 /CO 2 ratio from 4:1 to 1:1 over the Co−In−ZrO 2 catalyst, the methanol selectivity decreased by only 7% of the original value (Figure 2d).In comparison, over Co−ZrO 2 and In−ZrO 2 , the methanol selectivity decreased by 38 and 30%, respectively.Therefore, the dual-atom oxide catalyst was more effective in terms of utilization of hydrogen for methanol production under hydrogen-lean conditions.To better understand the ability of Co−In−ZrO 2 to produce better methanol under low H 2 partial pressure, we calculated the order of the methanol formation reaction with respect to the reactants and the apparent activation energy (E a ) for the formation of methanol and CO.The E a for methanol formation (Figure 2e) was the highest for Co−ZrO 2 (139 kJ mol −1 ) followed by In−ZrO 2 (116 kJ mol −1 ) and the lowest for the Co−In−ZrO 2 oxide (80 kJ mol −1 ).The E a for the formation of CO followed the reverse trend (Figure 2f).For Co−ZrO 2 , the order of methanol formation with respect to H 2 (n Hd 2 ) was 1.0 and that with respect to CO 2 (n COd 2 ) was −0.5 (Figure 2g,h).The values of n Hd 2 and n COd 2 for In−ZrO 2 were similar at 0.9 and −0.4,respectively.Therefore, over singleatom doped oxide catalysts the activation of H 2 was the limiting process and a higher partial pressure of H 2 was needed for methanol formation.The n COd 2 values of −0.4 and −0.5 for single-atom doped oxides indicate that CO 2 adsorption over the surface inhibited methanol formation.These results are in line with the reported literature 55 and it has been shown that methanol selectivity reduces over oxide catalysts because the adsorption of intermediates and H 2 dissociation occur on the same site. 27,28In comparison, over the Co−In−ZrO 2 catalyst the n Hd 2 and n COd 2 were 0.3 and −0.2, respectively.The higher n COd 2 was due to the mitigation of the adverse effect of poisoning due to the adsorbed intermediates.Also, the much lower value of n Hd 2 indicates that H 2 activation over the dualatom oxide catalyst was more facile than that over single-atom catalysts.We propose that over the Co−In−ZrO 2 catalyst, the H 2 activation and CO 2 adsorption occur on different yet proximal sites, and methanol formation is promoted because of the cooperative effect between Co and In.

Site for CO 2 Adsorption in the Dual-Atom Oxide Catalyst
For all catalysts, in the H 2 temperature-programmed reduction (H 2 -TPR), no characteristic reduction peaks were observed until 500 °C, indicating that the dopants did not reduce under the reaction condition (Figure 3a).The In and Co dopants maintained their ionic states (In 3+ and Co 2+ , respectively) as observed in the XPS analysis of the catalysts after the reaction (Figures S8 and S9).We have previously shown that inherent oxygen vacancies are formed in Co−ZrO 2 due to the charge imbalance and difference in coordination environments between Co 2+ and Zr 4+ . 56The density of defective oxygen species, measured by O 1s XPS of the fresh catalyst (the peak around 531.6 eV), is indicative of the abundance of inherent oxygen vacancy.The relative abundance of defective oxygen species in Co−In−ZrO 2 (16%) was higher than that in In− ZrO 2 (12%) and ZrO 2 (11%), but the same as that in Co− ZrO 2 (16%) (Figure 3b).However, determination of defective oxygen species by XPS is influenced by the presence of surface −OH species.Because of this reason, estimation of change in the abundance of oxygen vacancy on the surface solely based on O 1s XPS can be ambiguous.To confirm the change in abundance of −OH and oxygen vacancy after Co and In doping on the surface, we checked the relative intensities of the bicarbonate species (HCO 3 *) and carbonate (CO 3 *) species in IR analysis after adsorbing CO 2 (Figure S10).The formation of HCO 3 * is indicative of the presence of −OH species and increase in CO 3 * species is indicative of the increase in oxygen vacancy on the surface (see Figure S10 and the following discussion).For doped oxides, carbonate formation increased compared to that of pure ZrO 2 (Figure 10a).For Co−In−ZrO 2 and Co-ZrO 2 , the carbonate peak intensities were similar and the highest among all of the catalysts.This is in line with the oxygen vacancy abundance indicated by the O 1s XPS analysis.As compared to pure ZrO 2 , addition of dopant decreased the relative abundance of surface −OH species and it decreased the most when Co was present (Figure S10b).Therefore, it can be concluded that doping of Co and In increased the oxygen vacancy on the surface rather than increasing the number of −OH groups.
Oxygen-vacant sites are oxophilic in nature and enhance the adsorption of strongly bonded CO 2 . 17In the CO 2 temperature-programmed desorption (CO 2 -TPD) analysis, two desorption features were observed for all of the catalysts (Figure 3c).The low-temperature feature (100−300 °C) was due to the desorption of physisorbed and weakly adsorbed CO 2 , while the feature around 320 °C was assigned to the desorption of chemisorbed CO 2 .For Co−In−ZrO 2 both physisorbed and chemisorbed CO 2 were present in higher amounts.Moreover, the TPD profile of Co−In−ZrO 2 was similar to that of Co-ZrO 2 , especially in the chemisorbed CO 2 region (in line with the similar CO 3 * peak intensity in the IR analysis of adsorbed CO 2 shown in Figure S10).Comparing the amounts of chemisorbed CO 2 for all catalysts, it was evident that CO 2 chemisorption in Co−In−ZrO 2 was related to the doping of Co (Figure 3d).Therefore, Co sites are responsible for the CO 2 chemisorption in Co−In−ZrO 2 .

Stabilization of Formate on the Dual-Atom Oxide Catalyst
In CO 2 hydrogenation, formate is formed by hydrogenation of adsorbed CO 2 and it is the key intermediate for methanol formation over oxide catalysts. 11,57In order to observe the nature of formate species over different catalysts, in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) (Figure S11) analysis was performed at 300 °C, 0.1 MPa, and H 2 /CO 2 = 4:1.Figure 4a shows the comparison of formate peak positions for all catalysts after 100 min of reaction.In comparison to undoped ZrO 2 , the positions of the formate peaks were shifted in the presence of doped oxides.Bidentate formate bonded with two Zr atoms is the most stable formate configuration in undoped ZrO 2 . 14,58,59In the doped oxides, stabilization of formate at the interfacial M-Zr (M = Co or In) site causes a shift in peak position. 28,56,60It is evident that the nature of formate species over Co−In−ZrO 2 and Co−ZrO 2 was the same, owing to the similar peak position in the DRIFTS analysis.Therefore, formate species in the dual-atom catalyst were present on the Co−Zr interfacial site.
Further confirmation of the tendency of formate species to preferentially adsorb on Co sites was obtained by temperatureprogrammed decomposition of adsorbed formic acid over the catalyst surface (Figure 4b).The decomposition of adsorbed formic acid over Co−In−ZrO 2 resulted in CO formation with a peak top at 340 °C.The decomposition profile was similar to Co−ZrO 2 but different from In−ZrO 2 , which exhibited a broad peak centered at 390 °C.These results confirm that the oxygen-vacant site created by doping of Co in Co−In−ZrO 2 was responsible for CO 2 adsorption and stabilization of formate after the hydrogenation of CO 2 .

Cooperation between Co and In Atoms for Methanol Formation
For methanol formation, the reaction pathway follows successive hydrogenation of formate to methoxy species adsorbed on the surface, followed by desorption of methoxy as methanol.Time-resolved evolution of adsorbed species was measured by in situ DRIFTS under CO 2 hydrogenation condition (Figure 5a,b).The complete DRIFTS spectra for all catalysts and detailed peak assignment are shown in Figures S12−S14 and Table S3.To check the possibility of methanol formation via CO hydrogenation, we performed in situ DRIFTS analysis under a CO + H 2 environment (Figure S15), which resulted in no methoxy formation and confirmed the formate hydrogenation pathway for methanol formation.Because methoxy species were not observed over the Co− ZrO 2 catalyst, it is evident that hydrogen dissociation over In atoms promotes the hydrogenation of formate species adsorbed on Co−In−ZrO 2 .This cooperative effect was further confirmed by the temperature-dependent in situ DRIFTS study.Formate species appeared over the Co−In−ZrO 2 catalyst at a temperature as low as 100 °C, followed by Co− ZrO 2 and In−ZrO 2 (Figure 5c).This result shows that In atoms promote the hydrogenation of CO 2 adsorbed over Co− Zr sites to formate.The peaks for the adsorbed methoxy species also appeared at a lower temperature over Co−In− ZrO 2 than over In−ZrO 2 (Figure 5d).Therefore, the separation of sites for CO 2 adsorption and H 2 dissociation over Co−In−ZrO 2 promotes hydrogenation of adsorbed species to yield methanol.

Theoretical Confirmation of the Cooperative Effect
Density functional theory (DFT) calculation was used to investigate the impact of dual-atom (In and Co) doping on the  Using the most stable structure shown in Figure 6a, oxygen vacancy formation on the surface of Co−In−ZrO 2 was investigated.Seven possible sites near the doped Co and In atoms were considered (Figure S17).Oxygen vacancies adjacent to Co atoms were comparatively easier to form because of the difference in the coordination between Co and Zr atoms.The most stable structure was with oxygen vacancy adjacent to the Co atom, as depicted in Figure 6c.A cavity was generated near the Co atom, which could serve as an active site for the CO 2 and H 2 activation.
Next, we considered the dissociation of H 2 over the In and Co sites of Co−In−ZrO 2 in the absence and presence of adsorbed formate species at the Co−Zr interface as shown in Figure 6d.On a clean surface, H 2 dissociation was exothermic over both Co and In sites (Figure 6e).Therefore, the first H 2 dissociation, responsible for the formation of formate species, was possible over both Co and In sites, with a barrier of 0.22 and 0.42 eV, respectively.H 2 dissociation over Co was more exothermic with a lower activation energy likely due to the presence of undercoordinated O atoms nearby Co that would accept H + during the heterolytic dissociation of H 2 . 61This is in accordance with the experimental results because formate easily forms over the surface of single-atom Co−ZrO 2 catalysts.Next, we investigated the influence of adsorbed formate species on the Co site.The geometry of formate-adsorbed Co−In− ZrO 2 was based on experimental results having the formate species coordinated with Co and Zr atoms (Figure 6d).Using this structure, H 2 dissociation was not possible over the Co site with the adsorbed formate species.H 2 dissociation over the In site was favorable with a barrier of 0.38 eV (Figure 6f).H 2 dissociation over Zr atoms was endothermic, as expected.The proximity of In to the adsorbed formate species would facilitate transfer of hydrogen dissociated over In atoms to formate species, resulting in further hydrogenation of formate to methanol.

Mechanism
Based on the above results, we propose the following mechanism for CO 2 hydrogenation to methanol over the Co−In−ZrO 2 catalyst through the cooperative effect of Co and In (Figure 7).Doping of Co 2+ creates an oxygen-vacant site for the adsorption of CO 2 .First, H 2 dissociation might happen on either Co or In atoms for hydrogenation of adsorbed CO 2 , although the presence of In promotes hydrogenation of adsorbed CO 2 to formate.Following formate formation, the ability of Co to dissociate H 2 and hydrogenate formate is hindered.Instead, H 2 dissociated over adjacent In atoms facilitates the hydrogenation of formate to methoxy species and promotes the desorption of methoxy species as methanol.Over the dual-atom catalyst, because formate gets selectively hydrogenated to methoxy by the cooperation of Co and In atoms, both the selectivity and productivity of methanol increase.Thus, the presence of a Co single atom reduces catalyst poisoning by strong adsorption of CO 2 and formate.In atoms assisted in H 2 dissociation and hydrogenation of adsorbed intermediates.Consequently, a higher methanol selectivity can be obtained even at a low H 2 partial pressure.

Generalizing Dual-Atom Co-Containing Catalysts
To evaluate the generality of the idea of mitigating the poisonous effect by introducing a Co single atom, we prepared Co−Zn−ZrO 2 and Co−Ga−ZrO 2 dual-atom oxides for CO 2 hydrogenation to methanol.In both cases, the methanol selectivity and methanol STY increased compared to the single-atom doped oxides (Figure 8).Characterization of the catalysts and investigation of reaction mechanism for Co−Zn− ZrO 2 and Co−Ga−ZrO 2 revealed a similar mode of action for increased methanol formation by separation of sites for CO 2 adsorption and H 2 dissociation (see Supporting Information Sections 3 and 4).The above results establish that this strategy can be effectively used to enhance the methanol selectivity and yield using Co single atoms to create an active site for the adsorption of CO 2 and stabilization of intermediates.

■ CONCLUSIONS
During the hydrogenation of CO 2 to methanol, it is essential to maintain high selectivity while increasing the methanol yield.Improving the reactivity of the mixed-oxide catalyst requires countering the poisoning effect of the strongly adsorbed CO 2 and formate species, which reduces the H 2 dissociation ability.Counter to the strategy reported so far of introducing metal nanoparticles to dissociate H 2 , we demonstrate that the separation of the adsorption sites for CO 2 and H 2 over a dual-atom mixed-oxide catalyst having a Co single atom is an effective way to promote both the selectivity and productivity of methanol by mitigating the poisonous effect.We prepared a Co-and In-doped ZrO 2 (Co−In−ZrO 2 ) catalyst where Co and In atoms were atomically dispersed.Structural characterization indicated the atomic distribution of Co and In atoms.Co−In−ZrO 2 showed better methanol selectivity and productivity than the corresponding single-atom-doped oxides.Kinetic analysis showed that Co−In−ZrO 2 had the lowest apparent activation energy for methanol formation, with a concomitant highest apparent activation energy for CO formation.It was found that Co−In−ZrO 2 was less dependent on the H 2 and CO 2 partial pressure than were binary oxides during methanol formation.Co sites were responsible for the CO 2 adsorption and formate formation.In sites were responsible for H 2 dissociation and promoted formate hydrogenation to methoxy and methoxy desorption as methanol selectively.This cooperative effect was not limited to the Co−In system and was also observed in Co−Zn and Co−Ga systems.This study shows a different approach to selective catalyst design based on the preferential adsorption of substrates and intermediates.We believe that this strategy will help in the design of selective catalysts for complicated reactions involving multiple substrates and intermediates.

Figure 1 .
Figure 1.(a) XRD patterns of doped oxides along with (b) the shift of the (101) reflection of t-ZrO 2 .(c) HAADF-STEM analysis of Co−In−ZrO 2 showing the (101) plane of t-ZrO 2 and the corresponding d spacing.(d) HAADF-STEM image of the Co−In−ZrO 2 catalyst with the corresponding elemental mapping of (e) Co (red), (f) In (green), and (g) Zr (blue).(h) Double-aberration-corrected HAADF-STEM image of Co−In−ZrO 2 showing Co (red circle) and In (green circle) single atoms.(j, n) Line intensity profiles of the selected lines shown in (i).

Figure 2 . 1 , 1 , H 2 /
Figure 2. (a) Comparison of the catalytic activities of Co−ZrO 2 , In−ZrO 2 , and Co−In−ZrO 2 .Reaction conditions: 270 °C, 3 MPa, 30,000 mL h −1 g cat −1 , H 2 /CO 2 = 4.(b) Comparison of methanol selectivity at 300 and 270 °C over different catalysts at the same CO 2 conversion.CO 2 conversion was maintained at 1 and 2.7% for all catalysts at 270 and 300 °C, respectively.The same CO 2 conversion was achieved by changing the space velocity.(c) Selectivity of methanol over doped ZrO 2 catalysts with a total dopant loading of 15% at the same CO 2 conversion level of 2.7%.Reaction conditions: 300 °C, 3 MPa, SV = 30,000−45,000 mL h −1 g cat −1 , H 2 /CO 2 = 4.(d) Methanol selectivity under varying H 2 /CO 2 ratio for Co−ZrO 2 , In−ZrO 2 , and Co−In−ZrO 2 .The percentages denote a relative decrease in methanol selectivity.Reaction conditions: 300 °C, 3 MPa, 30,000 mL h −1 g cat −1 −g) revealed a homogeneous distribution of Co and In atoms.HAADF-STEM image and elemental mapping of Co− ZrO 2 (Figure S3a−c) and In-ZrO 2 (Figure S3d−f) also showed a homogeneous distribution of Co and In atoms, respectively.The individual In and Co atoms in Co−In−ZrO 2 were identified by double-aberration-corrected HAADF-STEM analysis.The line intensity profile showed areas of higher and lower intensity for the presence of In and Co atoms (Figure 1h−n), respectively, according to the Z-contrast, relative to their atomic number.These results show that In and Co atoms replaced Zr atoms in the crystal structure and dopant atoms were observed in proximity of each other by way of random distribution within the ZrO 2 matrix.

Figure 4 .
Figure 4. (a) Comparison of formate peak positions during the in situ DRIFTS experiment.Reaction conditions for DRIFTS experiment: 300 °C, 0.1 MPa, and H 2 /CO 2 = 4:1.For undoped ZrO 2 , the reaction temperature was 340 °C.(b) Mass profile of CO during formic acid decomposition to CO over doped oxide catalysts.

Formate
species appeared rapidly at the start of the reaction along with carbonate and bicarbonate over all doped catalysts.Methoxy species appeared with a delay of about 10 min over Co−In−ZrO 2 and In−ZrO 2 .Methoxy formation over Co−In− ZrO 2 was faster in comparison to In−ZrO 2 .Methoxy species were not detected over Co−ZrO 2 .After 100 min, the flow of CO 2 was stopped, and formate species reduced rapidly over Co−In−ZrO 2 and Co−ZrO 2 .The rate of the disappearance of formate over In-ZrO 2 was slower.Simultaneously, the abundance of methoxy increased marginally, as more sites for adsorption were available after stopping CO 2 .The rate of disappearance of methoxy over Co−In−ZrO 2 was comparatively faster than that over In−ZrO 2 .These results indicate that formate adsorbed over the Co−Zr interface was more reactive, either to form methoxy or decompose to CO.In addition, the desorption of methoxy species adsorbed over the Co−In− ZrO 2 surface was also facile.

Figure 5 .
Figure 5. (a, b) Normalized intensity of formate and methoxy species during in situ DRIFTS experiment over Co−ZrO 2 , In−ZrO 2 , and Co−In− ZrO 2 , based on the DRIFTS spectra shown in Figures S10−S12.For each catalyst, the peak intensity was normalized using the peak intensity of the respective species at 100 min during the reaction.Reaction condition: 300 °C, 0.1 MPa, H 2 /CO 2 = 4:1.(c, d) Temperature-dependent evolution of formate and methanol over different catalysts during in situ DRIFTS analysis.
Figure S16 shows the optimized geometry for different positions of In and Co atoms in the Co−In-doped ZrO 2 (101) surface.Most stable structures were obtained when both In and Co atoms were on the first layer (Figure 6a,b).The energy difference between the structure with In and Co atoms adjacent to each other (Figure 6b) and with In and Co separated from each other (Figure 6a) was only 0.07 eV.This corroborates the experimental observation of atomic dispersion of Co and In atoms in the ZrO 2 matrix leading to a random arrangement of Co and In atoms in proximity.The surface distortion of ZrO 2 with doped In was smaller than the distortion with the doping of Co. Co atoms were coordinated with 5 oxygen atoms, resulting in an imbalance in coordination number between cations.

Figure 6 .
Figure 6.Optimized geometries of Co-and In-doped ZrO 2 (blue, cyan, and brown correspond to Co, Zr, and In, respectively.)with (a) Co and In atoms far apart and (b) adjacent to each other.(c) Optimized geometry of Co−In−ZrO 2 having oxygen vacancy near the Co atom (the dotted circle represents the original position of removed oxygen).(d) Geometry of the catalyst with formate stabilized at the Co−Zr interface, used to calculate H 2 dissociation energy.(e) H 2 dissociation over Co and In atoms over the surface of Co-and In-doped ZrO 2 having oxygen vacancy shown in (c).(f) H 2 dissociation over the In atom and two Zr atoms over the surface of Co-and In-doped ZrO 2 having formate species stabilized at the Co−Zr interface shown in Figure 6d.The detailed structures for H 2 dissociation steps including the transition state are shown in the Supporting Information (Figures S18−S19).

Figure 7 .
Figure 7. Schematic representation of the proposed mechanism for CO 2 hydrogenation to methanol over the ternary Co−In−ZrO 2 showing the site separation of CO 2 adsorption and H 2 dissociation over Co and In sites, respectively.