Accuracy of Density Functional Theory for Predicting Kinetics of Methanol Synthesis from CO and CO₂ Hydrogenation on Copper

Abstract

Density functional theory (DFT) is widely used for investigating heterogeneous catalysis; however, the predictive power of DFT is determined by the approximation used in the exchange−correlation (XC) functionals. In this work, we systematically investigate how the kinetics of methanol synthesis predicted by DFT depends on the choice of XC functionals. Microkinetic modeling is performed based on the Gibbs energies calculated with XC functionals that represent three levels of accuracy: Perdew−Burke−Ernzerhof (PBE) functional, Heyd−Scuseria−Ernzerhof (HSE) hybrid functional, and the random phase approximation (RPA) functional. We show that the predicted kinetics strongly depends on the choice of XC functionals. Methanol's turnover frequencies predicted by PBE and HSE are about 30 times faster than the predictions from RPA. PBE predicts that the overall barrier of CO hydrogenation is 0.56 eV lower than that of CO 2 hydrogenation, therefore suggesting CO as the carbon source for methanol synthesis on copper. This contradicts previous isotope-labeling experiments that supported CO 2 as the carbon source in industrial methanol synthesis; therefore, PBE suggests that metallic copper cannot be the active site for CO 2 hydrogenation. On the other hand, the overall barrier of CO hydrogenation, predicted by HSE and RPA, is lower than the overall barrier of CO 2 hydrogenation by 0.22 and 0.14 eV, respectively. This suggests that CO 2 hydrogenation is also competitive for methanol production, and we cannot completely rule out the possibility that metallic copper is the active site for catalyzing CO 2 hydrogenation. In addition, the prediction of the dominating adsorbates also strongly depends on the choice of XC functionals. Our results show that different XC functionals can predict different kinetics for methanol synthesis, which calls attention to the accuracy of DFT for modeling methanol synthesis.

Recent Developments in the Modelling of Heterogeneous Catalysts for CO2 Conversion to Chemicals

“…Tameh et al . examined the accuracy of DFT for prediction of kinetics in CO 2 hydrogenation to methanol using microkinetic modelling.…”

“…One have suggested that metallic copper is the active site and addition of oxides would sustain the large copper surface and reduce CuO to metallic copper. [36,143] Other suggested the effect of "CuÀ Zn" synergy for being essential for the active site [144,145] where methanol is synthesized over Cu + at the Cu/ZnO interface, or over Cu cations that dissolve in the ZnO matrix and metallic copper only promotes the dissociation of H 2 . According to the Behrens' et al [146] results where they combined experimentally obtained data with DFT calculations for studying the active sites of methanol synthesis over Cu/ZnO/Al 2 O 3 industrial catalyst, the most active surface was found to be a Cu step with Zn alloyed into it.…”

“…Tameh et al [143] examined the accuracy of DFT for prediction of kinetics in CO 2 hydrogenation to methanol using microkinetic modelling. They compared RPA method to different functionals including RPBE and BEEF-vdW functionals.…”

Unraveling the Role of the Rh–ZrO2 Interface in the Water–Gas-Shift Reaction via a First-Principles Microkinetic Study

“…However, there is no a priori guarantee that hybrid functionals or the inclusion of +U give better results than GGAs, meta-GGAs, or vdW corrected functionals. [84][85][86] The reason is that the quality of the underlying GGA, amount of exact-exchange, and the value of the +U parameter all impact reaction and activation energies. 87 In the case of a hybrid functional, the amount of the exact exchange to be included depends on the screening properties of the materials, 88,89 and it is unclear whether universally parametrized (screened)…”

Advances and challenges in modeling solvated reaction mechanisms for renewable fuels and chemicals

“…This adaptive QM/MM scheme is very useful, but it could still benefit from a reduction of spatial artifacts that affect multiscale modeling . We expect to see substantially more applications of these methods in the coming years as they can allow higher levels of QM theory for improved insights into catalytic reactions …”