By hydrogenating carbon dioxide to value-added products such as methanol, heterogeneous catalysts can lower greenhouse gas emissions and generate alternative liquid fuels. The most common commercial catalyst for the reduction of CO2 to methanol is Cu/ZnO/Al2O3, where ZnO improves conversion and selectivity toward methanol. The structure of this catalyst is thought to be Zn oxy(hydroxyl) overlayers on the nanometer scale on Cu. In the presence of CO2 and H2 under reaction conditions, the Cu substrate itself can be restructured and/or partially oxidized at its interface with ZnO, or the Zn might be reduced, possibly completely to a CuZn alloy, making the exact structure and stoichiometry of the active site a topic of active debate. In this study, we examine Zn3 clusters on Cu(100) and Cu(111), as a subnano model of the catalyst. We use a grand canonical genetic algorithm to sample the system structure and stoichiometry under catalytic conditions: T of 550 K, initial partial pressures of H2 of 4.5 atm and CO2 of 0.5 atm, and 1% conversion. We uncover a strong dependence of the catalyst stoichiometry on the surface coverage. At the optimal 0.2 ML surface coverage, chains of Zn(OH) form on both Cu surfaces. On Cu(100), the catalyst has many thermally accessible metastable minima, whereas on Cu(111), it does not. No oxidation or reconstruction of the Cu is found. However, at a lower coverage of Zn, Zn3 clusters take on a metallic form on Cu(100), and slightly oxidized Zn3O on Cu(111), while the surface uptakes H to form a variety of low hydrides of Cu. We thus hypothesize that the 0.2 ML Zn coverage is optimal, as found experimentally, because of the stronger yet incomplete oxidation afforded by Zn at this coverage.
Read full abstract