Understanding the Trend in Core–Shell Preferences for Bimetallic Nanoclusters: A Machine Learning Approach

Abstract

Finding out the driving factors in core–shell preference of nanoscale binary metal alloys is important due to their ubiquitous presence in applications ranging from catalysis to biomedical. We consider binary-alloyed metallic nanoparticles encompassing a vast range of alkali, alkaline, basic, 3d, 4d, and 5d transition metals, and p-block metals and determine the core–shell preference by calculating the segregation energies of single-atom alloy clusters by density functional theory. Application of machine learning to this large database, built on features characterizing the constituents, leads to the identification of four key factors: (i) cohesive energy difference, (ii) atomic radius difference, (iii) coordination number difference, and (iv) magnetism, providing the core-to-shell preference of a given constituent. Interestingly, the relative importance of one key feature over another is found to be decided by the metal type. Our analysis also predicts that, for very small and very large differences of cohesive energy of the constituents, instead of core–shell structure, mixed and Janus structures are stabilized, respectively. Our exhaustive study will be useful in designing bimetallic nanoalloys with specific chemical ordering of the constituent species.