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

High-index facet nanoparticles, such as tetrahexahedron (THH) and hexoctahedron (HOH), offer significant potential in catalysis due to their unique surface structures, yet their synthesis is challenging due to thermodynamic instability. We present a data-driven approach to address these challenges by incorporating high-throughput density functional theory (DFT) calculations and machine learning techniques. We explore the surface energies of low- and high-index facets for nine transition metals modified by 13 guest atoms. Machine learning focuses on understanding key factors stabilizing high-index facets, particularly {210} facets of THH, and extends material discovery from monometallic hosts to multimetallic structures. Our findings are validated through chemical synthesis, successfully producing THH nanoparticles with exposed {210} facets.

In parallel, we demonstrate a method to control the shape evolution of Cu nanoparticles between THH and HOH by tuning Te atom concentration. DFT calculations show that the surface density of Te atoms alters the stability of {210} and {421} facets. Through controlled annealing and dealloying of CuTe nanoparticles, we achieve selective synthesis of THH or HOH, driven by Te surface concentration.