The standard model of computational catalysis is based on assumptions of active sites or active surfaces, DFT calculations for energy barriers and reaction energies, the harmonic approximation for pre-exponentials, and micro-kinetic mean-field models. Quantum chemistry can remove the ambiguity connected with parameter fitting in microkinetic models, but only when rate and equilibrium constants for crucial steps can be predicted with chemical accuracy (4 kJ/mol for energies and one order of magnitude for rate constants). This has recently been achieved for the methylation of ethene, propene and butene in H-ZSM-5. Following a divide-and-conquer strategy, the common energy calculations on periodic models based on density functional theory are augmented with the more accurate wave-function type calculations for cluster models of the reaction site (Møller-Plesset perturbation theory (MP2) and Coupled Cluster theory (CCSD(T)).1 The availability of rigorous methods made it possible (i) to revise the well-established “Lunsford” mechanism for the oxidative coupling of methane on Li-doped MgO and to predict an alternative one that stimulated new experiments. (ii) to identify cases for zeolites (proton exchange barriers for alkanes on Brønsted sites) and MOFs (CH4/CO2 separation) where missing agreement between predicted reactivity and adsorption data, respectively, points to imperfections of the samples. Our method brings theory to a new level: from descriptive to predictive, from qualitative to quantitative, and from ideal, perfect structures to real defective structures.

Ab initio predictions with chemical accuracy for heterogeneous catalysis and adsorption

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.