On the Predictability and Fundamental Insights of Adsorption Models: Current Status and the Path Ahead

The electronic structure of materials contains the key features describing the energetic interaction of adsorbed species on their surfaces. The dimensionality of electronic feature space, however, is incredibly high. Primary features of the electronic structure must be identified to define models that are otherwise nonapplicable due to their complexity. In this paper, we discuss exemplary efforts as they pertain to surface science and heterogeneous catalysis describing electronic structure effects on adsorption. We discuss these examples under three categories: first principles-based physically informed models, correlations found from first principles data, and fully data-driven approaches. While there exists plenty of overlap between these categories, the primary philosophies of the three methodologies are compared and contrasted. Based on this discussion, a perspective on the path forward to continue advancing atomic understanding of chemisorption is presented. Specifically, the unique benefits and drawbacks of each philosophy are discussed, and the union of a collective effort from all fronts is proposed. While there are advantages and disadvantages of approaching a problem from either a first principles or data driven approach, the insights gained from all efforts have proven to be useful to the catalysis community in illuminating the unknown regarding atomic-scale catalytic reaction phenomena.