Unraveling the Mechanism of Iridium Nanoparticle Exsolution through In Situ Scanning Transmission Electron Microscopy, Density Functional Theory and Machine-Learning Image-Analytics

Nanoparticle exsolution, <i>i.e.</i> metal atoms that, in a reducing environment, segregate to the surface from sites within a host oxide lattice to form anchored nanoparticles, is considered a remarkable alternative to traditional nanoparticle deposition techniques for catalytic and energy-related applications. Exsolved nanoparticles are found partially submerged (or ‘socketed’) into the surface of the host oxide, circumventing the major limitations of deposited nanoparticles, such as agglomeration or coking, which cause loss of catalytic activity. Several studies have suggested that the unique functional properties of exsolved particles are due to the structure and composition of this interface they form with the host oxide. However, many questions remain regarding the nucleation and growth mechanism of exsolved nanoparticles and the rearrangement of the host crystal structure during the exsolution process. This work presents a fundamental study where we monitor the early stages of nucleation of iridium nanoparticles during exsolution from a stoichiometric Ir-doped SrTiO₃ as a model structure. We apply in situ high-resolution scanning transmission electron microscopy (HR-STEM) to monitor in real-time exsolution in ultra-high vacuum while heating the sample from room temperature to 1100 °C, which allows us to observe atomic diffusion, preferential nucleation sites, the evolution of the host crystal structure, and the role of evolving host defects in the early stages of nucleation.<br/>We then use Density Functional Theory (DFT) and Molecular Dynamics (MD) simulations to investigate the Ir atomic interaction at the surface during the initial stages of exsolution. The simulations supported our in situ observation of nucleation via Ir atom clustering, furthermore indicating the influence of surface defects in trapping Ir atoms to initiate nanoparticle nucleation. In situ EELS studies allowed us to investigate in real-time variations in the composition, chemistry, and structure of the nanoparticles and the host during exsolution. To investigate the experimental in situ observation of a superstructure evolution in the host crystal structure, we then employed a machine-learning image-analytics approach, which revealed that, before nucleation, a superstructure forms at the surface of the monitored region, allowing the identification of this surface restructuring as the first step of the exsolution process. Finally, we investigated the socket formation, to which the high stability of exsolved nanoparticles is attributed. Although nanoparticle socketing into the host is undoubtedly one of the unique features of exsolution, the mechanistic reason behind its formation is, to date, not well documented and unclear. Through in situ STEM we monitored step-by-step the growth of the sockets around the exsolved particles, which allowed us to elucidate that the socketing only occurs once the particles are fully formed at the surface and at relatively high temperatures for our system; through in situ EELS we achieved high-resolution chemical characterization of the material’s composition at this anchoring interface. We propose that this unique socketing is generated via a similar mechanism to the Vapour-Liquid-Solid (VLS) one, where the exsolving metal nanoparticles catalyze the growth of SrTiO₃pedestals at the surface to lock in the exsolved nanoparticles. The mechanistic insights described in this ground-breaking study correlate the defect formation in the host oxide to the nanoparticle formation at the surface during exsolution, for the first time looking at the process holistically: considering both atoms-to-nanoparticles nucleation and the perovskite structural evolution at the same scale. This work elucidates new opportunities to effectively exploit structural and defect features to obtain exsolved particles, driving the progress in the tailored design of highly advanced systems for catalysis and energy-related applications more broadly.