Abstract:

Thiolated gold nanoclusters have attracted intense research attention due to their
applications in catalysis. Calculations of reactivity descriptors are important to
understand catalytic behavior, identify and predict active sites, and ultimately, design
new nanocatalysts. Recent advancements in synthesis techniques paved the way to the
production of bimetallic counterparts of these systems and the availability of a large
number of possible isomers makes studies on catalytic properties very challenging. I will
discuss an approach based on density functional theory calculations to explain the
catalytic activity of poly-dispersed Ag-alloyed Au25 nanoclusters for CO oxidation
reaction. Results of the condensed-to-atom Fukui functions indicate that the reduction in
the catalytic activity of Au25-xAgx(SR)18 is likely related to the reduction in the electron
donating capability of the outer shell sites. I will also present a machine learning model
designed to predict adsorption energies of thiolated nanoclusters. Machine learning
methods provide a faster route to materials property prediction. The features of the
model are based only on geometric properties of adsorbate-free, non-relaxed isomers.
The predictive capability of the model was tested using three different Ag doped
clusters: Au25, Au36 and Au133. Features based on the distribution of Ag atoms relative to
the CO adsorption site are the most important in predicting adsorption energies. The
orientation of the ligands have a significant effect on predicting the adsorption energies
of Au36, which is less spherical than both Au25 and Au133.