CO 2 hydrogenation on Cu-catalysts generated from Zn II single-sites: Enhanced CH 3 OH selectivity compared to Cu/ZnO/Al 2 O 3 Journal of Catalysis

The hydrogenation of CO 2 to CH 3 OH is mostly performed by a catalyst consisting mainly of copper and zinc (Cu/ZnO/Al 2 O 3 ). Here, Cu-Zn based catalysts are generated using surface organometallic chemistry (SOMC) starting from a material consisting of isolated Zn II surface sites dispersed on SiO 2 – Zn II @SiO 2 Grafting of [Cu(OtBu)] 4 on the surface silanols available on Zn II @SiO 2 followed by reduction at 500 (cid:1) C under H 2 generates CuZn x alloy nanoparticles with remaining Zn II sites according to X-ray absorption spectroscopy (XAS). This Cu-Zn/SiO 2 material displays high catalytic activity and methanol selectivity, in particular at higher conversion compared to benchmark Cu/ZnO/Al 2 O 3 and most other catalysts. In situ XAS shows that CuZn x alloy is partially converted into Cu(0) and Zn(II) under reaction conditions, while ex situ solid state nuclear magnetic resonance and infrared spectroscopic studies only indicate the pres- ence of methoxy species and no formate intermediates are detected, in contrast to most Cu-based catalysts. The absence of formate species is consistent with the higher methanol selectivity as recently found for the related Cu-Ga/SiO 2 . (cid:3) 2020 The Authors. Published by Elsevier Inc. ThisisanopenaccessarticleundertheCCBYlicense(http://


Introduction
The conversion of carbon dioxide (CO 2 ) to value added products would allow the mitigation of CO 2 emissions that are recognized as a major contributor to climate change [1,2]. One strategy to mitigate the deleterious effect of CO 2 emissions would be to convert it by hydrogenation to methanol (CH 3 OH), an important bulk chemical that can also be used for the generation of energy, thereby forming a closed carbon-fuel-cycle, referred to as the ''methanol economy" [3][4][5][6]. This entails the sustainable production of H 2 , the efficient capture of CO 2 as well as the use of highly active and selective hydrogenation catalysts. Currently, copper-based catalysts are among the most common hydrogenation catalysts used for the production of CH 3 OH from CO, a mixture of CO/CO 2 as well as CO 2 [7]. The most studied and industrially used catalyst is Cu/ ZnO/Al 2 O 3 , where the role of the different components is still under debate. The role of ZnO has been particularly discussed and has mainly assigned to the formation of highly active Cu-ZnO interfacial sites or a CuZn surface alloy [8][9][10][11][12][13][14][15]. However, these catalysts still suffer from low activity, selectivity and stability in CO 2 rich streams [16]. Alternatively, Cu/ZrO 2 has also been reported to be an efficient catalyst for the formation of CH 3 OH [17][18][19][20]. We recently showed that the role of ZrO 2 is to act as a Lewis acidic surface site at the periphery of Cu nanoparticles to stabilize reaction intermediates (formate and methoxy) [21]. Furthermore, by using a surface organometallic chemistry approach, we could show that Cu nanoparticles supported on silica decorated with isolated Zr IV surface sites (Cu-Zr/SiO 2 ) [22] display the same performance as Cu/ZrO 2 by also providing Lewis acidic Zr IV sites at the interface with Cu thereby increasing CH 3 OH selectivity. The same effect is observed with Lewis acidic isolated Ti IV surface sites on SiO 2 as a support [23,24]. However, all these catalysts suffer from fast erosion of selectivity with increasing conversion due to competitive adsorption of methanol/water on the Lewis acid sites needed for CO 2 activation and conversion to methanol (competitive adsorption).
More recently, we have shown that this approach could be used to improve the CH 3 OH selectivity when starting from a silica support consisting of well-defined Ga III sites. In this case, grafting of the Cu precursor followed by a hydrogen treatment yields CuGa x alloy nanoparticles (Cu-Ga/SiO 2 ) along with remaining isolated Ga III sites [25]. This catalyst shows high activity and excellent CH 3 -OH selectivity, especially at higher conversion, in sharp contrast to Cu-M/SiO 2 with M = Ti or Zr. In situ X-ray absorption spectroscopy (XAS) showed that, under reaction conditions, such catalysts evolve to generates Cu 0 and fully oxidized gallium sites. Compared to other catalysts prepared by surface organometallic chemistry (SOMC), no formate but only methoxy surface species are observed in the case of Cu-Ga/SiO 2 , which correlates with and can explain the increase in selectivity at higher conversion [26].
We thus decided to investigate the formation of the corresponding Cu/Zn systems starting from the silica-supported isolated Zn II surface sites [27] using an SOMC approach to explore its catalytic performance and to compare it with Cu/ZnO/Al 2 O 3 and other SOMC CO 2 hydrogenation catalysts.

Synthesis of Cu-Zn/SiO 2
A solution of [Cu(OtBu)] 4 (110 mg, 0.20 mmol) in 20 mL of toluene was added to 1 g of Zn II @SiO 2 wetted with toluene. The suspension was stirred for 4 h, washed three times with toluene (5 mL) and dried at 10 À5 mbar for 1 h. The solid was then reduced under H 2 at 500°C for 5 h (100°C h À1 ) cooled down to room temperature under H 2 , evacuated under high vacuum (10 À5 mbar) and stored in an argon filled glovebox.

Material characterization
Elemental analyses of all materials were performed by the Mikroanalytisches Labor Pascher, Remagen, Germany. Powder Xray diffraction (pXRD) patterns were recorded on a PANalytical X'Pert PRO-MPD diffractometer at a voltage of 40 kV and a current of 40 mA by applying Cu-Ka radiation (c = 1.54060 Å). Catalyst morphology was obtained by transmission electron microscopy (TEM) on a Hitachi HT7700 microscope within the facilities of Sco-peM at ETH Zurich. For the determination of the particle size distribution, >100 individual particles were considered, and the mean particle size and standard deviation are given according to a lognormal distribution function. Fourier-Transform Infrared (FTIR) spectroscopy experiments were performed on self-supporting wafers using a Bruker Alpha FT-IR spectrometer in transmission mode (24 scans, 4 cm À1 resolution) under exclusion of air. The specific surface area of the catalysts was measured from a N 2 physisorption isotherm recorded at 77 K on a BEL JAPAN BELSORP-mini II apparatus. The samples were degassed at 300°C under vacuum (10 À3 mbar) for 3 h prior to measurement. The data was analyzed by the BET method with a p/p 0 range between 0.1 and 0.3. H 2 chemisorption isotherms were obtained using a BELSORP-max apparatus on the reduced samples at 40°C and fitted according to a Langmuir isotherm (Eq. (1)), where P H 2 ;eq is the equilibrium hydrogen pressure, Q H 2 the hydrogen uptake (lmol g cat À1 ), Q H 2 ;max the saturation uptake of H 2 and K H 2 the thermodynamic constant for the dissociative hydrogen chemisorption.
Metal surface area was determined by N 2 O titration. In a typical experiment 30-50 mg of sample were weight into a U-shape quartz tube and connected to the instrument (BEL Japan, INC., BELCAT-B). Prior to analysis, the samples were pretreated under a flow of 50% H 2 /He at 300°C for 2 h, after which 25-30 successive pulses of the titration gas mixture (1% N 2 O in He) were introduced by a calibrated injection valve (2.77 mL N2O (STP) per pulse). The amount of N 2 O consumed was determined by monitoring the amounts of N 2 O and N 2 in the exhaust with a thermal conductivity detector. The quantity of surface metal sites are then determined considering the titration equation Pyridine adsorption experiments were performed on a selfsupporting pellet of the Cu-based catalysts and monitored by infrared spectroscopy (Nicolet NEXUS 6700) in transmission mode with a 4 cm À1 spectral resolution. After exposure of pyridine in the gas phase, the pellet was subsequently placed under high vacuum (10 À5 mbar) at room temperature (rt), 100°C, 200°C, 300°C, 400°C and 500°C (300°C/min) for 15 min prior to measurement of the IR spectrum. Similar to pyridine adsorption, CO adsorption was performed on a self-supporting pellet of the Cu-based catalyst by exposure of ca. 90 mbar of CO at room temperature followed by recording the infrared spectrum in transmission mode.

X-ray absorption spectroscopy (XAS)
X-ray absorption spectra at the Cu and Zn K-edge were measured at the SuperXAS beamline at the Swiss Light Source (SLS). The SLS was operating in top-up mode at a 2.4 GeV electron energy and a current of 400 mA. The incident photon beam provided by a 2.9 T super bend magnet source was selected by a Si(1 1 1) quick-EXAFS monochromator [32] and the rejection of higher harmonics and focusing were achieved by a silicon collimating mirror at 2.5 mrad. During the measurements the monochromator was rotating with 10 Hz frequency and X-ray absorption spectra were collected in transmission mode using ionization chambers specially developed for quick data collection with 1 MHz frequency [32]. The resulting spectra were averaged over 5 min. Calibration of the monochromator energy position was performed by setting the inflection point of a Cu or Zn foil spectrum recorded simultaneously with the sample to 8979 or 9662 eV for Cu or Zn K-edges, respectively.
In a typical in situ experiment, about 10-20 mg of the powder sample was packed into a 3 mm thick quartz capillary (0.1 mm wall thickness), which was connected with a pressurizable gas flow system. The catalysts were reduced under a H 2 /N 2 mixture (15%, 1 bar) at 300°C for 60 min, and then cooled down to reaction temperature (230°C). The reduction gas was flushed with N 2 for 15 min and then changed to the reaction gas mixture (CO 2 :H 2 : N 2 = 1:3:1, 5 mL min À1 ). Under reaction gas, the set-up was pressurized to 5 bar using a back-pressure regulator and the spectra was recorded every 15 min for one hour or until no changes in the spectra occurred. The spectra were background-corrected and normalized using the Demeter software package. Ex situ samples were pressed in pellets with optimized thickness for transmission detection and placed in aluminized plastic bags (Polyaniline (14 lm), polyethylene (15 lm), Al (12 lm), polyethylene (75 lm)) from Gruber-Folien GmbH & Co. KG using an impulse sealer inside an argon filled glovebox to avoid air contamination. References (ZnO, Cu, Zn and a-brass) were mixed with cellulose (in case of CuZn and ZnO), pressed into wafers and sealed in Kapton tape.

Solid state nuclear magnetic resonance spectroscopy
Solid-state NMR experiments on 1 H and 13 C were recorded on a Bruker 400 MHz AVANCE III HD spectrometer with a 4 mm MAS triple resonance probe operating in double resonance mode with a magic angle spinning frequency of 10 kHz. The chemical shift scale was calibrated using adamantane as an external secondary reference. Ramped cross polarization ( 1 H-13 C) was used for experiments with 1 H excitation frequency at 100 kHz. The contact time was 2 ms for 1D experiments and for 1 H-13 C HETCOR experiments. Additionally, for 1 H-13 C HETCOR experiment, DUMBO homonuclear ( 1 H-1 H) decoupling was used during t 1 . The static magnetic field was externally referenced by setting the 13 C higher frequency peak of adamantane to 38.4 ppm. The 1 H excitation and decoupling radiofrequency (rf) fields were set as 100 kHz. CP conditions were optimized to fulfill the Hartmann-Hahn condition under magicangle spinning with minor adjustments to reach optimal experimental CP efficiency. All samples were packed in an argon filled glovebox. For preparing the ex situ sample, to 100 mg of Cu-Zn/ SiO 2 (reduced at 300°C under H 2 after exposure to air) in a thick-walled glass reactor was introduced 1 bar of 13 CO 2 , which was then condensed under liquid nitrogen cooling. Then 1 bar of H 2 was introduced while still maintaining cooling of liquid nitrogen at À196°C. The reactor was then heated up to 230°C which leads to a pressure increase to 5 bar. After 12 h, the reaction vessel was cooled down to room temperature and evacuated under high vacuum (10 À5 mbar) and the resulting solid was stored in an argon filled glovebox.
2.6. Catalytic testing in CO 2 hydrogenation CO 2 hydrogenation reactions were conducted in a fixed-bed tubular reactor (9.1 mm ID) in down-flow configuration (PID Eng&Tech). In a typical experiment 250 mg of catalyst powder oxidized in air was mixed with 5.0 g of SiC and loaded in the reactor under ambient conditions (20 and 30 mg was used for Cu/ZnO/ Al 2 O 3 and Cu/ZnO/Al 2 O 3 Katalco, respectively). First, the catalyst was reduced for 1 h under a flow of 15% H 2 /N 2 (50 mL min À1 ) at 300°C and atmospheric pressure. After cooling down to 230°C, the reactor was pressurized to 25 bar with a flow of CO 2 :H 2 :N 2 (1:3:1, 50 mL min) for 30 min. The reactor was then set to measurement conditions (230°C, 25 bar) and the gas phase was analyzed via online gas chromatography (Agilent 7890B) equipped with an FID for CH 3 OH and TCD for N 2 , CO 2 , CO and CH 4 . Different contact times were probed by changing the gas flow rate from 100 mL (STP) min À1 to as low as 6 mL (STP) min À1 . Finally, activity data was collected at the initial flow rate of 100 mL min À1 to check for potential deactivation of the catalyst. The reaction rates, conversions and selectivities were calculated using the following set of equations (Eqs. (2)-(5)): where F in is the total gas inlet flowrate [mol h À1 ], F out is the total gas outlet flow rate [mol h À1 ], c x;in and c x,out are the inlet and outlet gas fraction of species x, r Cu;x is the formation rates of species x per gram copper g x h À1 g À1 Cu h i ; m cat is the mass of catalyst in the reactor g cat ½ ; w Cu the weight loading of copper wt Cu % ½ ; S x the product based selectivity of product x, F i;out the flowrates of the products, and X CO2 the conversion of CO 2 . Intrinsic formation rates (with respect to the contact time) are obtained by using a second/third order polynomial fit on the experimental data at conversions below 7%.
The specific surface area of this material is 187 m 2 g À1 as determined by N 2 physisorption isotherms and Brunauer-Emmett-Teller [36] (BET) analysis (Table S1), which is close to the initial SiO 2 material (ca. 200 m 2 g À1 ). Transmission electron microscopy (TEM) studies show the presence of small and narrowly distributed CuZn x alloy (vide infra) nanoparticles of Cu-Zn/SiO 2 (3.9 ± 1.0 nm) (Fig. 1b). The particle size is slightly larger than for the corresponding Cu/SiO 2 (2.9 ± 1.3 nm) prepared via a similar approach and more similar to corresponding gallium based materials (Cu-Ga/ SiO 2 ) (4.6 ± 1.4 nm) [22]. A surface metal nanoparticle concentration of 52 lmol g cat À1 was determined by N 2 O titration for Cu-Zn/ SiO 2 (assuming a 1:2 stoichiometry between N 2 O and the surface sites) (Table S1), which is similar to what is obtained for Cu/SiO 2 , [22] considering the larger particle sizes for Cu-Zn/SiO 2 . This increased N 2 O consumption for Cu-Zn/SiO 2 is likely due to the reaction of N 2 O with reduced zinc sites arising from CuZn x (surface)alloy (vide infra) [37]. Chemisorption experiments using H 2 at 40°C was performed since it was shown to be a reliable method to obtain metal dispersions for Cu-based systems with similar physicochemical properties [31]. A metal surface site concentration of 64 lmol g cat À1 for Cu-Zn/SiO 2 (assuming a 1:2 stoichiometry between H 2 and the metal surface site) was obtained, consistent with the number obtained from N 2 O titration (Table S1 and Fig. S2). Powder X-ray diffraction show no crystalline phases, due to the amorphous nature of the SiO 2 support and the presence of small metal nanoparticles (Fig. S3). IR spectroscopy of Cu-Zn/SiO 2 in the presence of 90 mbar CO at room temperature shows stretching bands at 2092 cm À1 which is red-shifted with respect to what is observed for pure Cu/SiO 2 at 2106 cm À1 evidencing a different copper morphology/structure (Fig. S4). Furthermore, the presence of Lewis acidic zinc sites is shown by pyridine adsorption and IR spectroscopy [38], where the ring vibrational band of pyridine at 1611 cm À1 for Cu-Zn/SiO 2 is observed, likely associated with its adsorption on Zn II sites (Fig. S5). Pyridine on Cu-Zn/SiO 2 is fully desorbed at 500°C under high vacuum (10 À5 mbar) (Fig. S5).
In order to obtain further information regarding the oxidation states and structural environments of zinc and copper in Cu-Zn/ SiO 2 , the XAS spectra at the zinc and copper K-edges are recorded for the as-prepared material ex situ under inert conditions (Fig. 1c). The zinc K-edge for Cu-Zn/SiO 2 , shows an edge energy at 9658 eV while for Zn II @SiO 2 and ZnO the edge energy is higher at 9662 eV (Fig. S6). This decrease in edge energy for Cu-Zn/SiO 2 corresponds to reduced Zn species [39]. A feature at 9662 eV shown by the first derivative of the Zn K-edge XANES spectrum of Cu-Zn/SiO 2 also evidences that a fraction of the zinc sites remain as Zn II (Fig. 1c). The XANES spectrum of the Cu K-edge has an edge energy at 8979 eV indicative of reduced copper (Fig. S7). Linear combination fits using Zn II @SiO 2 and a-brass shows that 51% of the sites can be fitted as Zn II and 49% as a-brass (Fig. S8). Overall, the XAS spectra show that reduction of the samples after Cu grafting (500°C under H 2 ) leads to a partial reduction of Zn II with the formation of CuZn x alloys along with remaining Zn II sites. These findings are similar to what was found for corresponding Cu-Ga/SiO 2 system where the formation of a CuGa x alloy is also observed [25], but contrasts with what was observed for Cu-Ti/SiO 2 [23] and Cu-Zr/SiO 2 [22] systems that remained as isolated Ti IV and Zr IV sites upon Cu nanoparticle formation.

Catalytic performance in CO 2 hydrogenation
Cu-Zn/SiO 2 was tested in CO 2 hydrogenation at 230°C and 25 bar (Fig. S9). Following exposure to air, the material was first reduced at 300°C under H 2 . The catalyst was then tested by varying the gas flow rate to examine the effect of contact time (Fig. S10) on the catalytic activity/selectivity at conversions below 10% (ca. 15% conversion for thermodynamic equilibrium).
The intrinsic formation rates obtained from extrapolating to zero contact time are evaluated and compared with Cu/SiO 2 , Cu-Zr/SiO 2 and Cu-Ga/SiO 2 . These materials have similar Cu loadings, particle size distribution as well as metal (M) site densities in Cu-M/SiO 2 (M = Ga, Zn and Zr), albeit slightly lower for Zr (Table S2), thus allowing a direct comparison between these catalysts and Cu-Zn/SiO 2 . Two Cu/ZnO/Al 2 O 3 catalysts, one commercially available and one prepared from a malachite precursor were used as benchmark materials [30]. The intrinsic CH 3 OH formation rate is 1.6 g h À1 g Cu À1 for Cu-Zn/SiO 2 which is 5 times higher than Cu/ SiO 2 and also slightly higher than Cu-Zr/SiO 2 or Cu-Ga/SiO 2 (Fig. 2a). Note that the catalytic activity of the support by itself (Zn II @SiO 2 ) is below detection limits. The intrinsic CO formation rates for Cu/SiO 2 and Cu-Zr/SiO 2 (0.3 g h À1 g Cu

À1
) are similar to Cu-Zn/SiO 2 . This leads overall to a CH 3 OH selectivity of 86% for Cu-Zn/SiO 2 , with CO being the only byproduct. Thus, Cu-Zn/SiO 2 has a higher CH 3 OH selectivity than unpromoted Cu/SiO 2 (48%) or even Cu-Zr/SiO 2 (77%), and is similar to, albeit slightly lower than Cu-Ga/SiO 2 (90%) [22]. Both the CH 3 OH and CO formation rates decrease at longer contact times (Fig. S11), indicating product inhibition involved in both pathways for Cu-Zn/SiO 2 , similar to Cu-Ga/SiO 2 [25]. It is particularly noteworthy that for the corresponding Cu-Ti/SiO 2 and Cu-Zr/SiO 2 catalysts, only CH 3 OH formation rates decrease with longer contact times [22,23]. Since both CH 3 OH and CO formation rates decrease with increasing contact time, a high selectivity toward CH 3 OH is maintained for Cu-Zn/SiO 2 (>70%) at conversions up to 5% (Figs. 2b and S12). The high CH 3 OH selectivity at higher conversions is not observed for similar type of catalysts that are more affected by conversion: Cu/SiO 2 and Cu-Zr/ SiO 2 only reaches 30% and ca. 40% CH 3 OH selectivity at a conversion of 5% (Fig. 3b).
In comparison to Cu-Zn/SiO 2 , Cu/ZnO/Al 2 O 3 [29] (Figs. S13-S18) shows higher formation rates for CH 3 OH (3.9 g h À1 g Cu À1 ) under the same reaction conditions, but it also favors the formation of CO (0.9 g h À1 g Cu

À1
), hence the overall lower intrinsic CH 3 OH selectivity (79% vs. 86%) (Fig. 2b). The CH 3 OH selectivity of Cu/ZnO/Al 2 O 3 drops more drastically with conversion (50% CH 3 OH selectivity at around 5% conversion) compared to Cu-Zn/SiO 2 . It shows that the Cu-Zn based catalyst generated via SOMC can maintain a higher CH 3 OH selectivity at higher conversion (Fig. 2b). The main difference is that the CO formation rate is less affected by contact time for Cu/ZnO/Al 2 O 3 compared to Cu-Zn/SiO 2 suggesting different reaction mechanisms for CO formation between the two materials.
A similar particle size distribution by TEM is obtained for the spent Cu-Zn/SiO 2 catalyst (4.2 ± 1.3 nm (Fig. S19)) compared to the fresh catalyst (3.9 ± 1.0 nm). This is further confirmed by absence of any crystalline phases by powder X-ray diffraction (Fig. S3).

In situ X-ray absorption spectroscopy
In order to obtain further insights of the structure of zinc and copper in Cu-Zn/SiO 2 , the material was further investigated by in situ XAS at the Zn and Cu K-edges (Fig. 3). The X-ray absorption spectra of Cu-Zn/SiO 2 were measured after oxidation of the catalyst in air, followed by reduction at 300°C under H 2 . The temperature was then decreased to 230°C and the reaction gas mixture consisting of CO 2 :H 2 :N 2 (1:3:1) was introduced and the pressure was increased to 5 bar (see materials and methods section).
The zinc K-edge after exposure to air shows an edge energy at 9662 eV consistent with the complete oxidation of zinc. Upon reduction of the oxidized catalyst at 300°C under H 2 , the white line intensity decreases and a feature towards lower energy (9658 eV) appears, which is indicative of reduced zinc sites (Fig. 3). A similar feature is observed for the as-prepared catalyst but with a higher intensity of the signal at lower energy (9658 eV), indicating that the reduction of the catalyst exposed to air results in a lower fraction of reduced zinc than in the as-prepared catalyst. Under reaction conditions -at 230°C and at 5 bar under a mixture of CO 2 : H 2 :N 2 (1:3:1) -the intensity of the white line is intermediate between these of the material after exposure to air vs. after reduction. The feature at lower energy (9658 eV) persists, indicating the presence of remaining reduced zinc sites for the Cu-Zn system. In comparison, the XAS spectrum of Zn II @SiO 2 has a higher white line intensity and lacks the feature at 9658 eV characteristic of reduced zinc sites under H 2 at 300°C (Fig. S20). This shows that in order to have reduced zinc sites, the presence of copper is necessary. In order to obtain a more quantitative ratio of the reduced zinc sites, linear combination fits using isolated Zn II surface sites (Zn II @SiO 2 ) and a-brass are performed; this analysis shows that after H 2 treatment at 300°C, 71% of the zinc sites are fitted as Zn II and the remaining sites can be fitted as a-brass (Fig. S21). Linear combination fits of the XANES spectrum under the reaction gas mixture show that 84% of zinc are present as Zn II sites, with the remaining sites being fitted as a-brass (Fig. S22). This contrasts with what was found for Cu-Ga/SiO 2 where all the gallium sites are oxidized under the same reaction conditions [25]. Since the in situ XAS measurements were only carried out at 5 bar in contrast to 25 bar for the catalytic test due to instrumental limitation, one may expect that increasing the total pressure (e.g. to 25 bar) could further favor the oxide form of zinc and Cu 0 due to a higher CO 2 conversion and thus a higher partial pressure of H 2 O. However, this experiment at low pressure already indicates the subtle differences between Cu-Zn/SiO 2 and Cu-Ga/SiO 2 . At the Cu K-edge, copper in Cu-Zn/SiO 2 is fully oxidized under air and fully reduced under H 2 and remains so under reaction conditions (Fig. S23). In summary, in situ XAS shows that the oxidation state of zinc in Cu-Zn/SiO 2 is highly dependent on the reaction conditions, where Zn II and Zn 0 sites are coexisting under reaction conditions.
The common features between all the Cu-based CO 2 hydrogenation catalysts prepared via SOMC is that they contain well-defined Lewis acidic surface metal sites on SiO 2 (Zr IV , Ti IV , Ga III and Zn II ). They likely play an important role in driving CH 3 (Fig. S25) also shows the absence of formate and the presence of only methoxy species by the 13 C-H stretches at around 2955 and 2855 cm À1 . It is noteworthy that Cu-Zn/SiO 2 and Cu-Ga/SiO 2 [25], which only show the presence of methoxy surface species, are also both highly selective for CH 3 OH at higher conversions. This likely indicates that the absence of stable formate surface species could play a major role in improving the CH 3 OH selectivity over CO. In fact, we have recently shown that highly stabilized formate species as in the case of Cu/Al 2 O 3 leads to the preferential formation of CO likely via formation of methyl formate at higher conversion that can readily decompose in CO and methanol [26].

Conclusion
A Cu-Zn based catalyst was generated by surface organometallic chemistry forming CuZn x alloy nanoparticles along with residual Zn II sites on SiO 2 . This material contrasts with the previously prepared Cu-Zr or Cu-Ti based systems were no reduction of Zr IV and Ti IV occurred but display similar feature to reported Cu-Ga based systems, which also consist of CuGa x alloy in the assynthetized material. The Cu-Zn based catalyst also shows high activity in CO 2 hydrogenation, mainly forming CH 3 OH as the main product, even at relatively high conversion. Comparing to Cu/ZnO/ Al 2 O 3 , the Cu-Zn based catalyst generated via SOMC shows higher CH 3 OH selectivity especially at higher conversions. Under reaction conditions, zinc is present in both its reduced state and as Zn II sites according to in situ XAS, which contrast with what is observed with CuGa x , where only Ga III sites are present. Noteworthy, no formate species are intercepted and only methoxy surface species are observed according to ex situ solid state NMR and IR spectroscopy; this observation is consistent with the higher CH 3 -OH selectivity at higher conversion. The Cu-Zn based catalyst shows structural and catalytic similarities to the previously reported Cu-Ga based system, and thus indicates important features required for highly active and selective catalysts for the hydrogenation of CO 2 to CH 3 OH. We are currently working on transposing these findings to develop improved industrial catalysts.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.