Water coordinated on Cu(I)-based catalysts is the oxygen source in CO2 reduction to CO

Catalytic reduction of CO2 over Cu-based catalysts can produce various carbon-based products such as the critical intermediate CO, yet significant challenges remain in shedding light on the underlying mechanisms. Here, we develop a modified triple-stage quadrupole mass spectrometer to monitor the reduction of CO2 to CO in the gas phase online. Our experimental observations reveal that the coordinated H2O on Cu(I)-based catalysts promotes CO2 adsorption and reduction to CO, and the resulting efficiencies are two orders of magnitude higher than those without H2O. Isotope-labeling studies render compelling evidence that the O atom in produced CO originates from the coordinated H2O on catalysts, rather than CO2 itself. Combining experimental observations and computational calculations with density functional theory, we propose a detailed reaction mechanism of CO2 reduction to CO over Cu(I)-based catalysts with coordinated H2O. This study offers an effective method to reveal the vital roles of H2O in promoting metal catalysts to CO2 reduction.

: Total (a.u.) and relative (kcal/mol) energies of the reaction species at B2PLYP/cc-pVTZ/Aug-cc-pVTZ-PP levels with inclusion of B2PLYP/cc-pVTZ/Aug-cc-pVTZ-PP zero-point vibrational energies (ZPVE) Supplementary Table 3: Total (a.u.) and relative (kcal/mol) energies of the reaction species and transition states for CO2 reduction catalyzed by singlet Cu + at B2PLYP/cc-PVTZ/Aug-cc-pVTZ-PP levels with inclusion of B2PLYP/cc-PVTZ/Aug-cc-pVTZ-PP zero-point vibrational energies (ZPVE) Supplementary Table 4: Total (a.u.) and relative (kcal/mol) energies of the reaction species and transition states for CO2 reduction catalyzed by triplet Cu + at B2PLYP/cc-PVTZ/Aug-cc-pVTZ-PP levels with inclusion of B2PLYP/cc-PVTZ/Aug-cc-pVTZ-PP zero-point vibrational energies (ZPVE) To further confirm the current conclusion in actual heterogeneous catalysis, we employed a water-gas shift reaction (WGSR) apparatus equipped with gas chromatography (Supplementary Figure 6a) to online investigate the effect of H2O content on the reduction of CO2 to CO, and commercial available Cu/ZnO/Al2O3 particles were used as catalyst. To well control the H2O content, a certain amount of H2O was continuously introduced to the reaction system by adjusting the flow rate of syringe pump. As shown in Supplementary Figure

S11
of CO demonstrated a first increasing trend followed by a declining one, in which it gave the optimal performance when the H2O partial pressure in reaction system was 50 kPa. Such a pattern was in good agreement with the previous theoretical calculations by Sun et al. 1 Namely, H2O could kinetically accelerate the hydrogenation on CO2 to COOH, promoting the reverse WGSR to produce CO, whereas the too high initial partial pressure of H2O would thermodynamically inhibit the CO2 conversion. Although the improvement using the current reaction system was not so impressive, it indicated that the presence of H2O could indeed promote the conversion of CO2 to CO in heterogeneous catalysis. Supplementary Figure 6c compares CO2 conversion efficiency in the absence and presence of optimized H2O content (50 kPa). Although a pretty low conversion efficiency was observed, the results suggested that the presence of H2O favored CO2 conversion to CO. For the purpose of confirming the origin of O atom in resulting CO from CO2 reduction, we employed the above reverse WGSR to reduce CO2 in the presence of 50 kPa of H2 16 O and H2 18 O, respectively. Afterward, the reaction products were collected and measured off-line with an Agilent Technologies 7890B gas chromatograph equipped with a GS-CarbonPLOT capillary column and an Agilent Technologies 5977A mass spectrometer. Supplementary Figures 6d and 6e show the corresponding chromatograms. Due to the limited separation capacity of GS-CarbonPLOT capillary column in a manual injection mode, the peaks of air and CO could not be resolved well, but both could be observed clearly. Subsequently, we identified the components of CO from the systems of H2 16 O and H2 18 O using mass spectrometry. As demonstrated in Supplementary Figures 6f  and 6g, a more abundant peak of m/z 30 appeared for the product in the presence of H2 18 O than that in the presence of H2 16 O. According to the current reaction system, the peak of m/z 30 could be assigned to C 18 O from the interaction of CO2 and H2 18 O. More importantly, after comparing the peak intensity, it is apparent that an improvement of 46.9-fold in peak intensity of m/z 30 was observed for the system of H2 18 O than that in the presence of H2 16 O. These results not only suggest the peak m/z 30 was not from background, and also brought further solid evidence that the O atom in resulting CO originated the involved H2O.

S12
In-situ DRIFTS could give more convincing results to the current study. The experiment was carried out at 150 o C and 250 o C with Cu/γ-Al2O3 as catalyst, respectively, and the corresponding results are displayed in Supplementary Figure 7a and b. It is apparent that for both reaction temperatures, the in situ DRIFT spectra had an analogous pattern by adding H2O into the reaction system. Namely, after introducing CO2 into the system (without H2O), the absorption peak intensity of bicarbonate on γ-Al2O3 (1230 cm -1 , 1439 cm -1 , and 1670 cm -1 ) demonstrated a gradually Supplementary Figure S7 | In situ DRIFT spectra of the resulting products on Cu/γ-Al2O3 or Pt/γ-Al2O3 with addition of H2O into the reaction system of CO2 and H2 at different reaction temperatures: (a) 150 o C @Cu/γ-Al2O3, (b) 250 o C @Cu/γ-Al2O3, and (c) 50 o C @Pt/γ-Al2O3 (flow rate of 5%CO2/95%Ar: 4 mL min -1 ; flow rate of H2: 3 mL min -1 ).
increasing trend and reached a plateau. Simultaneously, the absorption peak of bidentate formate (1602 cm -1 ) on γ-Al2O3 steadily emerged, but this type of absorption peak on Cu (1649 cm -1 ) was not obvious. However, after H2O was introduced to the reaction system, on the one hand, the absorption peak of bidentate formate on γ-Al2O3 (1379 cm -1 , 1398 cm -1 , and 1602 cm -1 ) gradually increased, along with the gradual decay till disappearance of the absorption peak of bicarbonate occurred at 1230 cm -1 and 1439 cm -1 . More importantly, the absorption peak of bidentate formate on Cu (1649 cm -1 ) exhibited a gradually increasing trend. These results indicated that H2O played a crucial role in the generation of formate on both Cu and γ-Al2O3. According to our discussion on Fig. 5a, formate was a critical intermediate in the reduction of CO2 to CO, which confirmed the critical role of H2O in CO2RR. Despite this, no obvious absorption peaks of CO adsorbed on either γ-Al2O3 or Cu were observed, due to its fast desorption rate or weak adsorption on both supports under the current conditions. 2 To get more compelling results on the effect of H2O in CO2RR, Pt/γ-Al2O3 was employed as a catalyst to perform the experiments. As shown in Supplementary Figure 7c, after introducing H2O into the reaction system, the absorption peak of CO adsorbed on Pt (2000 cm -1 and 2061 cm -1 ) demonstrated a gradual increasing pattern with the extension of reaction period. Such a result indicated that the introduced H2O was indeed favorable to the generation of CO in CO2RR S14                          Supplementary Figure    As is well-known, the Q2 region of commercial TSQ mass spectrometer ( Fig. 1) is generally employed for collision-induced dissociation (CID) of gas phase ions, in which the selected ions are accelerated by applying an electrical potential (5 V of AC voltage in this work) to increase the ion kinetic energy and then allowed to collide with neutral molecules (e.g., argon in this work). In the collision, some of the kinetic energy is converted into internal energy which results in bond breakage and the fragmentation of the molecular ion into smaller fragments (https://en.wikipedia.org/wiki /collision-induced dissociation). In the current investigation, it involves the reactants of [Cu(H2O)] + , Cu + , H2O, and CO2 in the Q2 region during the reaction/collision between [Cu(H2O)] + (1) and (2), and the resulting H + ions would supply the necessary protons in Fig. 5a in the main text. On the other hand, many prior studies [3][4][5][6][7] have indicated that electrons could be dissociated from anions [e.g., PtBr6 2-, Pd(CN)4 2-, Ru 3 Co(CO) 13-, CO3 -, and CO2 -] in the CID of mass spectrometry. Based on the above fact, it is speculated that the required electrons in this study could be generated following Eqs. (3)- (5)

Supplementary
If the above assumption was correct, O2 should be produced. After examining the mass spectrum from the reaction of [Cu(H2O)] + and CO2, [Cu(O2)] + ions, namely m/z 95 for [ 63 Cu(O2)] + and 97 for [ 65 Cu(O2)] + , were indeed captured as shown in Supplementary Figure 28a and b. This fact suggested that in CO2RR, the dissociation of H2CO3 was likely to be one of the electron sources.
In addition, we also explored the possibility of generating necessary electrons from Cu-based species in the CID, namely Eq. (6) or (7). If this route was feasible, a higher collision energy in CID would favor a more amount of electrons, thereby leading to a decrease in the total amount of Cu(I)-based species (e.g., [Cu(H2O)] + To confirm the above possibility, in the reaction between [Cu(H2O)] + and CO2 we enhanced the collision energy ranging from 4 to 20 V by maintaining other parameters constant. With the increase in the collision energy, the dissociation possibility of Cu + to Cu 2+ or [Cu(H2O)] + to [Cu(H2O)] 2+ would increase. To evaluate the total amount of Cu-based species, we summed the peak intensity of [Cu(H2O)] + , Cu + , [Cu(CO)] + , and [Cu(CO2)] + collected from the corresponding mass spectra. As shown in Supplementary Figure 28c and d, the amount of Cu(I)-based species presented a deceasing trend with increasing the collision energy, indicating that in the reaction of [Cu(H2O)] + and CO2, Cu(I)-based species would become others such as Cu(II)-based species via Eqs. (6) and (7), metallic Cu by reacting with the generated electrons from H2CO3 or transferring their charges to Ar gas by generation of Ar + (m/z 40). However, no direct evidence was gained to confirm the generation of Cu(II)-based species and Ar + using the current technique. In our opinion, the reduction of Cu(I)-based species to metallic Cu was highly possible because the required electrons were available in the current system, as well as the documented references. 8,9 From the above discussion, it is apparent that there is at least more than one route to generate necessary electrons for supplying CO2RR in the current study. Although there was no H2 introduction or generation involved in the current work, the electrons offered opportunity to reduce CO2 to CO.