A data-driven approach for the guided regulation of exposed facets in nanoparticles

Nanomaterials with high-index facets have desirable properties but are often challenging to synthesize. One way to realize such structures is by incorporating guest metal or metalloid atoms that can stabilize high-index facets by influencing surface energies. However, the effect of different guest atoms can vary substantially, and the vast parameter set (possible combinations of host nanoparticles and guest species) makes a trial-and-error experimental approach to explore every combination impractical. Here we report a data-driven approach incorporating high-throughput density functional theory calculations to assess surface energies of low- and high-index facets of nanoparticles (9 transition metals) with surfaces modified by 13 guest atoms. Machine-learning techniques are then used to understand the critical features leading to energetically favoured high-index facet formation in the context of tetrahexahedron. The predictions are validated by chemical synthesis, demonstrating the efficacy of this approach in accelerating the synthesis of tetrahexahedron materials with exposed {210} facets.