First-principles calculations of the structural and electronic properties of cobaltites for neuromorphic applications*

Transition metal oxides (TMOs), particularly the cobaltites La_1-xSr_xCoO_3-d (LSCO), are promising materials for neuromorphic computing. They display a metal-to-insulator transition (MIT) under the action of physical stimuli, which can be exploited to realize low power resistive switching devices. Here, using first principles calculations we investigate how to control the oxygen vacancy concentration in cobaltite to trigger a MIT. Experiments¹ revealed a series of topotactic transitions in LSCO from perovskite to brownmillerite and to Ruddlesden-Popper phases, which displayed various magnetic states and a MIT. Our calculations² based on DFT+U unraveled the complex interplay between crystal structures, magnetic and electronic properties that leads to the MIT. We found that cooperative structural distortions and concurrent magnetic state transitions during the topotactic transition are ultimately responsible for driving the MIT. To guide the design of resistive switching devices, we developed a first-principle model³ to predict the electrical bias needed to trigger the MIT, and provided strategies to minimize the threshold voltage. Further, to measure the oxygen vacancies concentration in thin film cobaltites, we combined experiments and theory to identify fingerprints of oxygen vacancies in the X-ray absorption spectra of LSCO and provided a robust protocol to determine oxygen stoichiometry⁴. Finally, we found that a metallic interface may arise in a heterostructure formed by two insulating phases in cobaltites⁵, and we identified the mechanism leading to the formation of such an interface. Our results point at the possibility of realizing energy-efficient resistive switching processes at a two-dimensional interface within a single material.