Scalable QM-in-QM Embedding Framework in FHI-aims

Multiscale embedding techniques provide an optimal balance of high accuracy and computational efficiency for modeling extended materials. These methods focus on highly precise calculations for a small cluster subsystem, such as a catalytically active surface site, extracted from a larger system like a metal surface, while utilizing more efficient, lower-level methods for the surrounding periodic environment. The aim of this project is to develop a novel implementation of quantum mechanics (QM) embedding to describe cluster models embedded in extended periodic environments. The envisioned QM/QM approach takes an embedding potential, which is generated in a periodic Kohn-Sham Density Functional Theory calculation, and applies this as an embedding environment in an aperiodic higher-level calculation of the system. The approach enables the application of highly accurate theory in a localised manner, which will reduce computational cost and improve scalability on ARCHER2. Implementing this approach in FHI-aims will leverage its accurate atom-centered all-electron representation, improving simulations and scaling, particularly important for metallic systems studied by ARCHER2 users. Integration with the Atomic Simulation Interface (ASI) will enhance interoperability, streamline workflows, and extend benefits beyond the current FHI-aims community. This development facilitates highly accurate predictions for surface adsorption, bond dissociation, optical excitations, and charge transfer processes, significantly advancing the modeling of surface catalysis and spectroscopy.