Local decomposition of hybridization functions: Chemical insight into correlated molecular adsorbates.

Hybridization functions are an established tool for investigating the coupling between a correlated subsystem (often a single transition metal atom) and its uncorrelated environment (the substrate and any ligands present). The hybridization function can provide valuable insight into why and how strong correlation features such as the Kondo effect can be chemically controlled in certain molecular adsorbates. To deepen this insight, we introduce a local decomposition of the hybridization function, based on a truncated cluster approach, enabling us to study individual effects on this function coming from specific parts of the systems (e.g., the surface, ligands, or parts of larger ligands). It is shown that a truncated-cluster approach can reproduce the Co 3d and Mn 3d hybridization functions from periodic boundary conditions in Co(CO)4/Cu(001) and MnPc/Ag(001) qualitatively well. By locally decomposing the hybridization functions, it is demonstrated at which energies the transition metal atoms are mainly hybridized with the substrate or with the ligand. For the Kondo-active 3dx2-y2 orbital in Co(CO)4/Cu(001), the hybridization function at the Fermi energy is substrate-dominated, so we can assign its enhancement compared with ligand-free Co to an indirect effect of ligand-substrate interactions. In MnPc/Ag(001), the same is true for the Kondo-active orbital, but for two other orbitals, there are both direct and indirect effects of the ligand, together resulting in such strong screening that their potential Kondo activity is suppressed. A local decomposition of hybridization functions could also be useful in other areas, such as analyzing the electrode self-energies in molecular junctions.

Exchange Spin Coupling from Gaussian Process Regression.

Heisenberg exchange spin coupling between metal centers is essential for describing and understanding the electronic structure of many molecular catalysts, metalloenzymes, and molecular magnets for potential application in information technology. We explore the machine-learnability of exchange spin coupling beyond linear regression, which has not been studied yet. We employ Gaussian process regression, since it can potentially deal with small training sets (as likely associated with the rather complex molecular structures required for exploring spin coupling) and since it provides uncertainty estimates ("error bars") along with predicted values. We compare a range of descriptors and kernels for 257 small dicopper complexes and find that a simple descriptor based on chemical intuition, consisting only of copper-bridge angles and copper-copper distances, clearly outperforms several more sophisticated descriptors when it comes to extrapolating toward larger experimentally relevant complexes. Exchange spin coupling is similarly easy to learn as the polarizability, while learning dipole moments is much harder. The strength of the sophisticated descriptors lies in their ability to linearize structure-property relationships, to the point that a simple linear ridge regression performs just as well as the kernel-based machine-learning model for our small dicopper data set. The superior extrapolation performance of the simple descriptor is unique to exchange spin coupling, reinforcing the crucial role of choosing a suitable descriptor and highlighting the interesting question of the role of chemical intuition vs systematic or automated selection of features for machine learning in chemistry and material science.

Modeling adsorbate-induced property changes of carbon nanotubes.

Because of their potential for chemical functionalization, carbon nanotubes (CNTs) are promising candidates for the development of devices such as nanoscale sensors or transistors with novel gating mechanisms. However, the mechanisms underlying the property changes due to functionalization of CNTs still remain subject to debate. Our goal is to reliably model one possible mechanism for such chemical gating: adsorption directly on the nanotubes. Within a Kohn-Sham density functional theory framework, such systems would ideally be described using periodic boundary conditions. Truncating the tube and saturating the edges in practice often offers a broader selection of approximate exchange-correlation functionals and analysis methods. By comparing the two approaches systematically for NH3 and NO2 adsorbates on semiconducting and metallic CNTs, we find that while structural properties are less sensitive to the details of the model, local properties of the adsorbate may be as sensitive to truncation as they are to the choice of exchange-correlation functional, and are similarly challenging to compute as adsorption energies. This suggests that these adsorbate effects are nonlocal. © 2017 Wiley Periodicals, Inc.