Influence of Hubbard U Parameter in Simulating Adsorption and Reactivity on CuO: Combined Theoretical and Experimental Study

Abstract

Transition metal oxides are an important class of catalytic materials widely used in the chemical manufacturing and processing industry, owing to their low cost, high surface area, low toxicity, and easily tunable surface and structural properties. For these strongly correlated transition metal oxides, standard approximations in the density functional theory (DFT) exchange-correlation functional fail to describe the electron localization accurately due to the intrinsic errors arising from electron self-interactions. DFT+U method is a widely used extension of DFT, where the Hubbard U term is an onsite potential which puts a penalty on electron delocalization, successfully describing such systems at only slightly higher computational cost than standard DFT methods. The U-value is usually chosen based on its accuracy in reproducing bulk properties like lattice parameters and band structure. However, chemical reactions on transition metal oxide surfaces involve complex surface−adsorbate interactions, and using the bulk properties based U-values in a locally changing surface environment may not describe reaction energetics correctly. Hence, in the current DFT+U benchmarking work, using CuO as a model transition metal oxide, we perform DFT+U calculations to investigate the dissociative chemisorption of H 2 on it. It is observed that the U-value impacts computed adsorption enthalpies by over 100 kJ mol −1 . The DFT+U calculated adsorption enthalpy is compared with the experimental adsorption enthalpy, and equilibrium adsorption configurations are confirmed using infrared analysis. We reveal that the commonly used U-value of 7 eV (fitted against CuO bulk properties) overestimates the adsorption enthalpy by 20−40 kJ mol −1 . The U-value between 4.5 and 5.5 eV correctly predicts the adsorption of H 2 on CuO. The DFT+U benchmarking procedure elucidated in this article, encapsulates surface−adsorbate interactions, surface reactivity, and the dynamic surface reaction environment and, thus, provides an appropriate U-value to be used to model reactions on metal oxide surfaces.

A comparative test of different density functionals for calculations of NH3-SCR over Cu-Chabazite

“…This comparison is in agreement with previous reports 59 and indicates that a Hubbard-U term is needed to describe the electronic structure and bonding in copper oxides. 45,61,62 The conclusions regarding the effect of including a Hubbard-U term for oxidization are similar for the side-on (C4) and bis (C5) configurations, see Fig. 3c and d.…”

High-Throughput Screening of Synergistic Transition Metal Dual-Atom Catalysts for Efficient Nitrogen Fixation

“…In addition, note that the theoretical limiting potential as well as the potential-liming step based on the Hubbard U approach are in good agreement with the PBE results (Figure S4). Since the NRR involves complex surface–adsorbate interactions, using the bulk properties-based U values may not describe reaction energetics correctly. , In this regard, we forego the use of U and solvation correction throughout the computations since the ignorable energy change was witnessed. In comparison to the Fe/ g -CN SAC counterpart, where the hydrogenation of *N 2 is the PDS with a much higher limiting potential of 1.02 V, the synergetic effect of dimer Fe-sites assuredly promotes the N 2 activation and therefore lowers the limiting potential.…”

Scalable approach to high coverages on oxides via iterative training of a machine‐learning algorithm

“…The DOS will depend on computational parameters, and in particular is known to be strongly affected by the addition of a Hubbard U value . Given the lack of consensus around an appropriate U value for TiO 2 , we have elected to omit it here . However, since the DOS is post‐processed using PCA which emphasizes differences between the DOS of different systems, we hypothesize that these differences will remain relatively systematic regardless of the computational approximations used, and that the main conclusions will remain valid as long as a consistent computational method is used for all DFT calculations.…”