Combining Atomic Layer Deposition with Surface Organometallic Chemistry to Enhance Atomic-Scale Interactions and Improve the Activity and Selectivity of Cu–Zn/SiO2 Catalysts for the Hydrogenation of CO2 to Methanol

The direct synthesis of methanol via the hydrogenation of CO2, if performed efficiently and selectively, is potentially a powerful technology for CO2 mitigation. Here, we develop an active and selective Cu–Zn/SiO2 catalyst for the hydrogenation of CO2 by introducing copper and zinc onto dehydroxylated silica via surface organometallic chemistry and atomic layer deposition, respectively. At 230 °C and 25 bar, the optimized catalyst shows an intrinsic methanol formation rate of 4.3 g h–1 gCu–1 and selectivity to methanol of 83%, with a space-time yield of 0.073 g h–1 gcat–1 at a contact time of 0.06 s g mL–1. X-ray absorption spectroscopy at the Cu and Zn K-edges and X-ray photoelectron spectroscopy studies reveal that the CuZn alloy displays reactive metal support interactions; that is, it is stable under H2 atmosphere and unstable under conditions of CO2 hydrogenation, indicating that the dealloyed structure contains the sites promoting methanol synthesis. While solid-state nuclear magnetic resonance studies identify methoxy species as the main stable surface adsorbate, transient operando diffuse reflectance infrared Fourier transform spectroscopy indicates that μ-HCOO*(ZnOx) species that form on the Cu–Zn/SiO2 catalyst are hydrogenated to methanol faster than the μ-HCOO*(Cu) species that are found in the Zn-free Cu/SiO2 catalyst, supporting the role of Zn in providing a higher activity in the Cu–Zn system.


Experimental part
Catalyst preparation. SiO2 (Aerosil 300) was compacted, sieved (180−300 m), calcined in static air (500 C, 12 h, 1 C min −1 ), evacuated to ca. 10 −5 mbar (500 C, 20 h) and then transferred, while hot, to a nitrogen-filled glovebox. This support is referred to as SiO2−500. Silica-grafted copper mesityl, CuMes/SiO2−500, was prepared via a surface organometallic chemistry approach, 1 using Cux(Mesityl)x, where x = 2, 4, 5, denoted CuMes (Strem Chemicals), as described previously. 2,3 Briefly, in a glovebox, SiO2−500 (1 g, 0.59 mmol of SiOH) was dispersed in 5 mL of dry toluene (dried by molecular sieve packed column, water content < 5 ppm) and contacted with a solution of CuMes (72 mg, 0.39 mmol of Cu) in 5 mL of toluene. The reaction mixture was stirred at 100 r.p.m for 3 h at room temperature. After this time, the solid was washed with toluene (3 × 5 mL) and dried (ca. 10 −5 mbar) at room temperature for 3 h to give the material denoted CuMes/SiO2−500. This material (300 mg) was then exposed to pulses of diethylzinc (Pegasus Chemicals, pulse duration was 0.1 s; 2, 5, 10 or 20 pulses were used) at 150 C in the ALD chamber (Picosun R-200). Materials obtained are referred to as CuMes-Et 2 Zn(n)/SiO 2 , where n is the number of pulses used. Before the catalytic tests, the materials were loaded, inside a glovebox, to a plug flow reactor and pretreated under undiluted H2 (50 mL min −1 ) at 500 C for 2 h. The reference methanol synthesis catalyst (Alfa Aesar) contained 63.5 wt% CuO, 25 wt% ZnO, 10 wt% Al2O3, and 1.5 wt% MgO. This material was reduced under H2 at 250 C for 3.5 h, i.e., pretreated according to published protocols. 4,5 Catalyst characterization. The surface area and pore volume of the materials were determined by N2 physisorption (Quantachrome NOVA 4000e) with the Brunauer-Emmet-Teller (BET) model 6 (using the adsorption data) and Barrett-Joyner-Halenda (BJH) model 7 (using the desorption data), respectively. Before the measurements, the samples were outgassed at 250 °C for 2.5 h. The Cu and Zn loadings of the catalysts were determined, after digestion in aqua regia, by inductively coupled plasma optical emission spectroscopy (ICP-OES) using an Agilent 5100 VDV instrument. Fourier-Transform Infrared (FTIR) spectroscopy experiments were performed on self-supporting wafers using a Bruker Alpha spectrometer in transmission mode (24 scans, 4 cm −1 resolution) under a N2 atmosphere. Intensities were normalized to the Si−O−Si overtones of the silica support. Powder X-ray diffraction (XRD) data were collected on a PANalytical Empyrean X-ray diffractometer equipped with a Bragg-Brentano HD mirror operated at 45 kV and 40 mA using Cu Kα radiation (λ = 1.5418 Å). The materials were examined within the 2θ range of 5−90° using a step size of 0.0167°. The scan time per step was 3 s. Transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM) with a high-angle annular dark-field (HAADF) detection, and energy-dispersive X-ray (EDX) spectroscopy were carried out on an FEI Talos F200X transmission electron microscope. Gold grids were used. If not noted otherwise, all samples were passivated under 1% O2/N2 at room temperature for 2 h before the transfer to the TEM instrument.
X-ray photoelectron spectra (XPS) were recorded on a SPECS (Germany) photoelectron spectrometer using a hemispherical PHOIBOS-150-MCD-9 analyzer (Mg K radiation, h = 1253.6 eV, 150 W). The binding energy (BE) scale was pre-calibrated using the positions of the peaks of Au 4f7/2 (BE = 84.0 eV) and Cu 2p3/2 (BE = 932.67 eV) core levels. The Si 2p peak at 103.5 eV of the SiO2 support was used as an internal standard. The survey and narrow spectra were recorded at a pass energy of the analyzer of 50 and 20 eV, respectively. To determine the chemical (charge) state of elements on the surface of the samples, the regions Si 2p, C 1s, Cu LMM, O 1s, Cu 2p, and Zn 2p 3/2 were measured. The atomic ratios of the elements on the catalyst surface were calculated from the integral photoelectron peak intensities which were corrected using theoretical sensitivity factors based on Scofield's photoionization cross sections. 8 The residual gas pressure S8 during the measurements did not exceed 8 × 10 −9 mbar. To carry out experiments with the high-pressure cell of the SPECS photoelectron spectrometer, all samples were rubbed into a stainless steel mesh spot welded onto a standard holder.
Ex situ X-ray absorption spectra (XAS) at the Cu and Zn K-edges were measured at the SuperXAS beamline (X10DA) at the Swiss Light Source (SLS, PSI, Villigen, Switzerland), operating in top-up mode at a 2.4-GeV electron energy and a current of 400 mA. Cu and Zn XAS spectra were collected at the K-edge using a Si (111) monochromator in transmission mode with continuous scanning between 8760 and 10792 eV with a step size of 0.1 eV. 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 eV or 9659 eV for Cu or Zn K-edges, respectively. The ex situ samples were sealed in a capillary in the glovebox and analyzed without exposure to air.
In situ Cu and Zn XAS spectra of Cu-Zn(5)/SiO2 were acquired simultaneously at the SuperXAS beamline at SLS. The incident photon beam was selected by a liquid nitrogen cooled Si (111) quick-EXAFS monochromator and the rejection of higher harmonics and focusing were achieved by a rhodium-coated collimating mirror. The beam size on the sample was approximately 2000  500 μm. During measurement, the quick XAS monochromator was rotating with a frequency of 1 Hz in a 3° angular range and X-ray absorption spectra were collected in transmission mode using ionization chambers specifically developed for rapid data collection with a frequency of 1 MHz. The in situ EXAFS spectra used for fitting were collected after cooling (<50 C), for EXAFS 600 scans were acquired (10 minutes acquisition) and averaged. For Cu and Zn, a Zn reference foil was used for energy calibration (9659.0 eV). The edge energy is set at the maximum of the first derivative of the normalized XANES. For the in situ experiment, approximately 20 mg of the passivated powder sample (180−300 μm) was packed into a 3 mm quartz capillary (i.d. 2.8 mm, bed length ca. 1 cm) which was integrated into a pressurizable gas flow system consisting of 2 parallel arrays, each consisting of 3 mass flow controllers (Bronkhorst), while the total pressure was maintained by a back-pressure regulator (Bronkhorst EL-Press). Switching between the two systems (i.e. switching the MFC array that was feeding the capillary) was performed using a remotely controlled 6-port 2-way switching valve (VICI, Valco) that could be operated from outside the experimental hutch. While one gas mixture was flowing to the cell, the other was directed via a bypass to the exhaust. Samples were heated using a custom-built infrared heater (Elstein-Werk M. Steinmetz GmbH & Co. KG (Germany), 30 mm length, with two heating elementsone above and one below the sample capillary), and the temperature was controlled using a 0.3-mm K-type thermocouple placed in direct contact with the catalyst bed. Ar and H2 were purified by passing through a trap containing molecular sieves and Q5 catalyst prior to introduction to the XAS quartz cell. CO2 was purified by passing through a trap containing activated molecular sieves.
In the in situ experiment, Ar (10 sccm, 1 bar) was flowed over the passivated catalyst for 10 minutes while the spectra were recorded for 10 minutes. The gas flow was then changed to H2 (10 sccm, 1 bar), and hydrogen treatment was performed under the same flow using a temperature ramp of 5 C min −1 , reaching a final temperature of 300 C. Spectra were recorded continuously during the hydrogen treatment. The sample was then cooled, (<50 C) under a flow of H2 to enable collection of data to be used for EXAFS analysis of the as-reduced material (600 scans, 10 minutes), before being heated to 230 C under H2. The gas composition was again changed to H2/Ar (3:2, 10 sccm), and subsequently pressurized to 11 bar. Once the pressure had stabilized, data acquisition was started, and the sample was measured continuously for 30 minutes. After 10 minutes of acquisition, the gas composition was switched to H2/Ar/CO2 (3:1:1, 10 sccm) S9 using the 6-port valve, to capture any changes upon the introduction of CO2 to the reaction gas. The delay between switching of the gas compositions and CO2 reaching the catalyst bed was estimated to be approximately 40 seconds, based on the reduced absorption of the beam (drop in baseline of spectrum prior to normalization) upon replacing a fraction of the Ar in the feed with less-absorbing CO2, using a strategy similar to a previously reported protocol. 9 After reaction, the sample was depressurized and cooled under a flow of Ar (10 sccm) and data was collected for EXAFS fitting (600 scans, 10 minutes). A schematic representation of the experimental protocol is given in Figure S51.
The processing of the XAS data was performed with the ProQEXAFS and the Athena software, 10,11 and EXAFS fittings were conducted with the Artemis software. 11 The S0 2 value (0.876) for the Cu K edge was obtained by fitting a Cu foil (See Figure S54). Coordination numbers were fixed for this fit. MCR analysis of the XANES data was performed using the in-built feature of the ProQEXAFS software. For this purpose, an energy range of 8900−9100 eV was used for Cu K-edge data, and 9600−9800 eV was used for Zn Kedge data.
H2 TPD and TPR experiments were performed using an AutoChem system (Micromeritics) with a thermal conductivity detector (TCD). In a typical H2 TPD experiment, ca. 50 mg of a passivated material was loaded in air and reduced in situ at 500 °C under a H2 flow (50 mL min −1 ) for 2 h. Subsequently, the specimen was purged with Ar for 30 min (50 mL min −1 ) at 500 C and cooled down to −50 C under Ar. The sample was then saturated in 5% H2/Ar flow (50 mL min −1 ) for 30 min and purged with Ar for another 30 min. Finally, the sample was heated to 500 C at 10 C min −1 under an Ar flow and the desorbed H2 was monitored with the TCD detector. The curves of H2 TPD were fitted using the Origin Software and Gaussian peak shapes. For the N2O titration measurements, the passivated materials were first pretreated under H2 at 500 C, followed by purging and cooling down to 50 C under Ar. The gas was then switched to 5% N2O/Ar for 60 min to oxidize the surface Cu 0 sites to Cu +1 and then reduced with 5% H2/Ar from room temperature to 500 C (10 C min −1 ). Similarly, the TPR experiments were carried out under 5% H2/Ar from room temperature to 500 C after the sample was oxidized by 5% O2/He at room temperature.
Solid-state NMR spectra were recorded on a Bruker 400 MHz spectrometer using a double resonance 4 mm CP-MAS probe. Samples were packed in 4 mm zirconia rotors inside an Ar-filled glovebox, and recorded at 298 K. In all cases, the downfield 13 C resonance of adamantane (38.4 ppm) was used as an external secondary reference to calibrate the chemical shifts. The MAS frequency was set to 10 kHz. For the 1 H-13 C HETCOR experiment, DUMBO homonuclear ( 1 H-1 H) decoupling was used during t1. To prepare 13 Clabelled surface intermediates, the activated Cu-Zn(5)/SiO2 catalyst was loaded into a thick-walled glass reactor inside an argon-filled glovebox. The reactor was evacuated, and 13 CO2 was introduced (1 bar), and then condensed under liquid nitrogen cooling (−196 C). Then, 1 bar of H2 was introduced while still maintaining cooling with liquid nitrogen at −196 °C. The reactor was then heated to 230 °C (10 C min −1 ) and kept for 12 h. After 12 h, the reaction vessel was cooled to −196 o C, evacuated at 10 −5 mbar, and allowed to return to room temperature over the course of 20 minutes whilst dynamic vacuum was maintained. The resulting solid was transferred to an argon-filled glovebox and packed to 4 mm zirconia rotors for solid-state NMR analysis. To prepare 13 C-labelled surface intermediates on Cu/SiO2, it was treated an identical way as Cu-Zn(5)/SiO2, as described above.
CO2 hydrogenation tests. CO2 hydrogenation was performed in a tubular fixed-bed reactor (304.8 mm total length, 9.1 mm internal diameter, Hastelloy X, Microactivity Effi, PID Eng&Tech). 12 In a typical experiment, the catalyst (100 mg) was first pretreated under H2 at 500 C for 2 h. The catalytic test was S10 performed at 230 C under 25 bar. The gas flow of H2/CO2/N2 (3/1/1), N2 as internal standard, was passed through the catalyst bed and the products were analyzed online by a GC (PerkinElmer Clarus 580) equipped with thermal conductivity and flame ionization detectors, and a methanizer. Different contact times (space velocities) were probed by changing the gas flow rate from 100 to 15 NmL min −1 . The formation rate, CO2 conversion, and methanol selectivity were calculated using the following equations: where Fx,out is the outlet flow rate of methanol or CO [mol h −1 ]; Cx,out is the outlet gas fraction of species x; Fx,in is the inlet flow rate of species x [mol h −1 ]; rx is the formation rate of methanol or CO [g h −1 gCu −1 ]; mCu is the mass of Cu used in the reaction [g]; MWx is the molecular weight of methanol or CO [g mol −1 ]; XCO2 is the conversion of CO2; SMeOH is the selectivity to methanol. Intrinsic formation rates and selectivities were extrapolated using a second-order polynomial fit of the experimental data. At least three experimental data points were averaged for each presented data point.
Operando DRIFTS. The catalyst powder (50−100 mg) was placed in a cylindrical cavity of a Harrick Praying Mantis High Temperature Reaction Chamber and the cell was mounted in a diffuse reflection (DRIFTS) accessory. The spectra were collected using a Thermo Scientific Nicolet 6700 FT-IR spectrometer equipped with a liquid-nitrogen-cooled MCT detector at a resolution of 4 cm −1 . The flow of gases was set using mass flow controllers (Bronkhorst). The switching between two reactant gas streams was performed using a 4-way valve. The pressure of the two gas streams (to the cell and to the vent) was controlled by backpressure regulators (Bronkhorst). The outlet gas stream was analyzed by a Pfeiffer OmniStar GSD 300C mass spectrometer. Prior to the measurements, the specimen was reduced in situ at 500 °C in a H2 stream (20 Nml min −1 ) for 2 h and subsequently cooled in H2 to the reaction temperature (230 °C). The cell was pressurized to 20 bar and immediately exposed to the reactant mixture (H2/CO2 = 3/1, total flow 20 Nml min −1 ) at 20 bar via a switching valve. The spectra were acquired continuously every 20 seconds (50 scans) in a time-resolved manner, to monitor the appearance and the evolution of surface species. Transient DRIFTS utilizes a periodic perturbation of a system by external parameters (stimulation) to influence the concentration of the active species. 13 The transient DRIFTS experiment was performed in the above-mentioned setup by using a switching valve to change the stream of the reactant gases allowing for a periodic perturbation of the gas concentration, as shown in Figure S65. No baseline correction was applied to the time-resolved spectra. Multivariate spectral analysis was performed using the multivariate curve resolution-alternating least squares (MCR-ALS) algorithm, as described elsewhere. 14 MCR is a chemometric method used for data processing and deconvolution of complex spectra down to individual components based on kinetic resolution. MCR can provide the response profiles (e.g. spectra, time profiles, etc.) of the individual chemical species of an unresolved mixture when no previous information is available about the nature and composition of these mixtures. S11

Supplementary Figures
Optimization of the catalyst preparation method and H2 pretreatment temperature. This section provides results and a discussion of additional catalysts to those presented in the main text. These additional catalysts differ in their preparation method. In particular, if a H2 pretreatment temperature is lowered from 500 C to 300 C, it gives a Cu-Zn(5)/SiO2−300−H2 catalyst with inferior methanol formation rate and methanol selectivity relative to Cu-Zn(5)/SiO2−500−H2 presented in the main text (ca. 2.1 g h −1 gCu −1 and 63%, respectively, Figure S1). Considering the similar Cu particle size (1.8  0.4 nm, Figure S2) in these two materials and an even slightly higher amount of surface Cu 0 sites in Cu-Zn(5)/SiO2−300−H2 (169 mol gcat −1 , Table S4), this result indicates that the formation of the more active structure of Cu and Zn requires high pre-treatment temperatures (ca. 500 C) that exceed the reduction temperatures determined in the TPR experiments ( Figure S3).

Order of Cu and Zn introduction.
We have also synthesized a catalyst by reversing the order of introducing Zn and Cu, that is, first pulses of Et2Zn were performed onto SiO2−500, followed by the grafting of Cu via SOMC. This material is denoted as Zn(5)-Cu/SiO2 and after H2 treatment (500 C, 2 h) its methanol formation rate is 2.3 g h −1 gCu −1 , which is notably lower than that of Cu-Zn(5)/SiO2 ( Figures S4  and S5). A similar Zn enrichment around the perimeter of Cu NPs, as in Cu (5)-Zn/SiO2, is observed in Zn(5)-Cu/SiO2 from the TEM-EDX maps, however, the formed Cu NPs have a size of 3.4  1.0 nm ( Figures  S6 and S7), which is larger than those in Cu-Zn(5)/SiO2.  Note that the formation rates of CH3OH and CO for Cu/SiO2 with respect to contact time have been reported. 3 S14 Figure S6. TEM-EDX of Zn(5)-Cu/SiO2. Scale bar: 6 nm.

Reduction of CuMes before Zn ALD.
To investigate if the interaction between Cu and Zn can be engineered from a preformed Cu/SiO2, we reduced CuMes/SiO2 under H2 at 500 C and subsequently deposition Zn using 5 pulses of Et2Zn. After a H2 treatment (500 C, 2 h), the material denoted Cured-Zn(5)/SiO2 was obtained (see Figures S8, S9, and S10 for characterization). Interestingly, Cured-Zn(5)/SiO2 shows a low intrinsic methanol formation rate of 0.9 g h −1 gCu −1 (Figures S11 and S12). TEM reveals that the Cu particle size in Cured-Zn(5)/SiO2 is 3.1  0.6 nm and the mappings show that Zn is not enriched around the Cu NPs. These results indicate that the enrichment of Zn around Cu NPs only occurred when molecular silica-grafted CuMes sites interact with Et2Zn species.  Figure S11. Formation rates of CH3OH and CO for Cured-Zn(5)/SiO2 with respect to the contact time; extrapolated to zero conversion (zero contact time) using second-order polynomial fits.
Air exposure. Exposure of CuMes-Et2Zn(5)/SiO2 to air leads to a rapid change in its color from dark brown to green, indicating the formation of CuO in the air-exposed material ( Figure S12). Interestingly, Cu-Zn(5)/SiO2−air shows, after the typical H2 pretreatment at 500 C, a low activity in CO2 hydrogenation to methanol (0.6 g h −1 gCu −1 methanol formation rate, Figure S13). TEM-EDX maps show that the Cu particle size in Cu-Zn(5)/SiO2−air−H2 grew to 2.7  0.4 nm ( Figure S14), and Zn is selectively enriched around the Cu NPs ( Figure S15). The hydrogen desorption peak, after saturation of Cu-Zn(5)/SiO2−air−H2 in 5% H2/Ar, shifts to a lower temperature of 46 C, i.e., by ca. 12 C as compared to Cu-Zn(5)/SiO2 ( Figure S3). This data indicates that the active and selective Cu-Zn sites characteristic for Cu-Zn(5)/SiO2 do not form after the exposure of CuMes-Et2Zn(5)/SiO2 to air. However, if CuMes-Et2Zn(5)/SiO2 is first converted to Cu-Zn(5)/SiO2 under H2 (500 C, 2 h), and then exposed to air at room temperature, the activity of the catalyst (pretreated in H2 before the test) is only slightly lower than that of the catalyst that has not been exposed to air ( Figure S13). Lastly, we passivated Cu-Zn(5)/SiO2 under 1% O2/N2 for 2 h and then exposed the material to air. In contrast to the exposure to air of CuMes-Et2Zn(5)/SiO2, the color of the passivated Cu-Zn(5)/SiO2 did not change to green and remained black ( Figure S12). The activity and selectivity of the passivated catalyst in CO2 hydrogenation can be fully recovered after H2 pretreatment (500 C, 2 h, Figures S13 and S16). To conclude, passivation is an effective measure to handle Cu-Zn(5)/SiO2 under air without compromising its catalytic performance. Figure S12. Images of Cu-Zn(5)/SiO2 catalysts after different pretreatments: (a) CuMes-Et2Zn(5)/SiO2 without air exposure. (b) CuMes-Et2Zn(5)/SiO2 after air exposure. (c) CuMes-Et2Zn(5)/SiO2 after reduction (H2, 500 C, 2 h), passivation (1% O2/N2, room temperature, 2 h), and air exposure. S18 Figure S13. Formation rates normalized per mass of Cu of Cu-Zn(5)/SiO2 after different pretreatment procedures together with the selectivities for CH3OH, specified above the respective bars (230 °C, 25 bar, H2/CO2/N2 = 3:1:1, contact time 0.06 s g mL −1 ). Air-free: the catalyst was loaded inside a glovebox with exposure to air. Exposed to air: the catalyst was loaded in air. Reduced before exposure to air: the catalyst was reduced (H2, 500 C, 2 h) and then loaded in air. Passivated: the catalyst was reduced (H2, 500 C, 2 h), passivated (1% O2/N2, room temperature, 2 h), and loaded in air. All the catalysts were in-situ reduced under H2 (500 C, 2 h) before CO2 hydrogenation.                         Fitting results are summarized in Table S7.  Table S7. We noted in the main text that the presence of a minor amount of Cu 0 in the passivated sample prior to H2 treatment cannot be excluded. 16,17 Indeed, the inclusion of a Cu-Cu path in the EXAFS fitting of the passivated sample results in a Cu-M path with a low degeneracy (RCu-  Differences between before and after the gas switch were not resolvable. Fitting results are summarized in Table S7.   Note that a related Cu/SiO2 material reported earlier did provide a formate peak at 168 ppm in the 13 C CP-MAS spectrum and no methoxy peak. 15 The difference to the result of the experiment in this work likely stems from the different experimental conditions to prepare the labelled surface intermediates, i.e., the reaction vessel was cooled to −196 o C and then evacuated at 10 −5 mbar in this work (see experimental section for details).      Tables   Table S1. Cu and Zn loading of Cu-Zn/SiO2 materials.

Material
BET surface area (m 2 g −1 ) Pore volume (cm 3 g −1 ) Average pore size (nm)  For quantitative EXAFS analysis, least square fittings implemented in the ARTEMIS software were applied. 11 Theoretical phases and amplitudes were obtained through self-consistent ab initio calculations with the FEFF6 code using the Cu f.c.c structure. The fitted variables were: coordination number (CN), S52 interatomic distance R, bond length disorder factors (Debye Waller factors, DW), and energy shift for the first Cu−Cu coordination shell. The photoelectron reference energy ∆E0 was fitted for each spectrum independently. The amplitude reduction factor S0 2 = 0.77 was obtained from fitting of the corresponding Cu foil, by fixing the CN to 12. Data fitting was carried out in the range of 1.7−2.8 Å and a window dr of 0.5, and the Fourier transform was carried out for k = 3.0−11.3 Å −1 .