In analysing the interactions of adsorbates with extended surfaces, theory, and to a large degree also experiments, tend to rely on adsorption models such as the famous Newns-Anderson model. There, the interaction of adsorbate and substrate is represented by the interaction of the adsorbate's one-electron frontier orbitals with the delocalised Bloch states of the substrate. Together with its simplified cousin, the d-band model, such adsorption models yielded significant insight into the chemistry of extended surfaces.The Newns-Anderson Model gives rise to a so-called chemisorption function, a density of states weighted by the electronic coupling between adsorbate and substrate states, measuring the line broadening of an adsorbate frontier state upon adsorption. In my contribution, I will describe our efforts to compute the chemisorption function from first principles, employing a diabatization scheme to estimate electronic couplings between states. Recognizing the periodic nature of the substrate and the finite extent of computational supercells, we extend the original chemisorption function by a Brillouin zone integration to account for the (generally localized) adsorbate states' coupling to different substrate states at different points in reciprocal space. Including this integration, the chemisorption function not only converges well with both lateral simulation cell size and slab depth, but also compares well with experiments. We test our approach against highly accurate experimental measurements of ultrafast charge transfer of a probe Ar atom on various metal surfaces and find that the chemisorption function reproduces both absolute transfer rates and rate asymmetries due to spin on magnetic surfaces. Next to its predictiveness, though, the Newns Anderson chemisorption function allows us to investigate the charge transfer mechanism in terms of e.g. contributions of states with different angular momentum character, showing the importance of phase cancellation effects in interpreting charge transfer at surfaces.
Read full abstract