摘要

Metal clusters have broad applicability in catalysis due to their unique reactivity and chemical selectivity, and density functional theory has become an important method for understanding catalysis and attempting to design better catalysts. In the present paper, a main focus is on the correlation part of the exchange-correlation functional, and we tested the reliability of the Kohn-Sham density functional theory with local correlation functionals and with the nonlocal random phase approximation (RPA) correlation functional for the water splitting reaction on monatomic Fe(0) and, by implication, for transition-metal-catalyzed reactions more generally. We computed four barrier heights and six energies of reaction in the catalytic mechanism. If the results are judged by deviation from CCSD(T) calculations, it is found that many modern exchange-correlation (xc) functionals (about half of the functionals tested) with local correlation perform better than those using RPA nonlocal correlation; for example, the PWB6K, B97-3, omega B97X-D, MPW1K, M06-2X, and M05-2X hybrid xc functionals with local correlation have overall mean unsigned deviations of 1.9 kcal/mol or less from the CCSD(T) results, in comparison to a mean unsigned deviation of 3.5 kcal/mol for EXX-RPA@PBE. We also find significant differences between the predictions for catalysis at the Fe(100) surface. This work provides guidance and challenges for future theoretical investigations of transition-metal catalysis.

  • 出版日期2015-4