Key points are not available for this paper at this time.
Catalytic processes are rather complex in which a large number of reaction pathways are possible, and identifying the preferred reaction pathway either from experiments or theoretical modeling is rather challenging. The approach combining density functional theory (DFT) calculations and microkinetic modeling is receiving increasing attention recently to gain more insights into surface catalytic reactions. However, the error propagation from DFT to those properties determined remains largely overlooked in the literature. In order to assess the uncertainty of the DFT-determined mechanisms and kinetics of the dry reforming of methane (DRM) reaction over Ni(111) and Pt(111), we employed the well-trained Bayesian error estimation functional with van der Waals correlation (BEEF-vdW) tailored for describing surface chemical properties in the DFT calculations. With a large ensemble of 2000 exchange–correlation functionals generated around the optimal BEEF-vdW functional, an ensemble of DFT energetics can be calculated, resulting in an ensemble of complex reaction networks that were generated from given elementary steps. Further complexity reduction of the reaction networks offers an estimate of the uncertainty of the reaction mechanism and kinetics. We found that our approach can not only reproduce the detailed reaction pathways as reported before for the DRM reaction but also give the confidence of these pathways being dominant, which will be very helpful in providing more insights into the reaction system. More importantly, our method can find some overlooked rate-determining states in previous studies. For example, two rate-determining steps observed experimentally but overlooked computationally, that is, CH* + O* and CO2 dissociation, were identified using our method. The approach developed here can also be readily adapted for studying more complex catalytic reactions and accelerating future catalyst design.
Wang et al. (Wed,) studied this question.