[EN] Oxide-supported copper nanoparticles exhibit promising properties as catalysts for the selective hydrogenation of CO2 to methanol. Both reaction rate and selectivity depend conspicuously on the nature of the oxide support/promoter at the metal periphery. However, a major challenge is the achievement of a quantitative description of such metal/oxide promotion effects, which is an essential step toward a rational catalyst design. We investigate structure-performance relationships with a series of model catalysts consisting of Cu nanoparticles dispersed on a mesoporous gamma-Al2O3 carrier overlaid with different transition metal oxides spanning a broad range of Lewis acidity (YOx, ScOx, ZrOx, TaOx). Remarkably, the apparent activation energy (E-a) for methanol formation is found to downscale linearly with the relative Lewis acidity of coordinatively unsaturated metal surface sites (cus) exposed on the oxide support, making this single physicochemical parameter a suitable reactivity descriptor in the whole study space. In correspondence with this performance trend, in situ Fourier transform infrared spectroscopy reveals that both the ionic character and the relative reactivity of bidentate formate species, developed on the catalyst surface under reaction conditions, vary systematically with the surface Lewis acidity of the oxide support. These findings support the involvement of oxide-adsorbed bidentate formate species as reaction intermediates and point to the relative electron-accepting character of the Lewis cus on the oxide surface as the factor determining the stability of these intermediates and the overall energy barrier for the reaction. Our results contribute a unifying and quantitative description for support effects in CO2 hydrogenation to methanol on oxide-supported copper nanoparticles and provide a blueprint for a predictive description of metal-oxide promotion effects, which are ubiquitous in heterogeneous catalysis.

The authors are grateful to P. Bussian and Sasol for providing the alumina precursor. S. Ruthe and K. Jeske (MPI-KOFO) are acknowledged for assistance with chromatographic product quantification. J. M. Salas (ITQ) and J. P. Holgado (ICMS-CSIC, Spain) are acknowledged for contributions to the in situ FTIR and XPS experiments, respectively. This research received funding from the Max Planck Society, the Bundesministerium ffir Bildung and Forschung (project 01DG17019), the Spanish Ministry of Science, Innovation and Universities (Severo Ochoa Excellence award SEV-20160683), and the Generalitat Valenciana (Scientific Excellence of Junior Researchers, SEJI2018/011). B.S. acknowledges the Humboldt foundation for a postdoctoral fellowship.