Welcome to the new version of CaltechAUTHORS. Login is currently restricted to library staff. If you notice any issues, please email coda@library.caltech.edu
Published March 2018 | public
Conference Paper

Embedded mean-field theory for high-efficiency electronic structure

Abstract

Quantum embedding has emerged as a powerful strategy for electronic structure calcns., in which the total system is divided into a small, chem. active subsystem described by a high level of theory, while the surrounding subsystem is treated with a lower level of theory. However, popular realizations of the approach, such as ONIOM, can exhibit drawbacks that include the need to specify fixed particle no. and spin state for each subsystem and uncontrolled errors assocd. with subsystem interactions. Here, we present embedded mean-field theory (EMFT), a quantum embedding approach that alleviates these restrictions by describing the subsystems at different levels of mean-field theories. In EMFT, subsystems are partitioned at the level of the at.-orbital (AO) basis (or the block-orthogonalized AO basis), which imposes a block-like structure on the one-particle reduced d. matrix. The EMFT ground state is obtained self-consistently via minimization of the energy functional of the d. matrix with different levels of treatment for each sub-block. Thus, the potential energy surfaces are guaranteed to be smooth functions of the nuclear coordinates, and response theory (including nuclear gradients) can be implemented more straightforwardly than other formulations. EMFT has been demonstrated to be accurate over a wide range of benchmark systems and chem. applications, including applications that involve subsystem partitioning across complicated bonding networks. Recently, EMFT has been successfully applied in large-scale ab initio mol. dynamics (MD) simulations of H-atom reactive scattering on graphene. We have also developed a linear-response formulation of time-dependent EMFT (TD-EMFT), which describes excited electronic states within the framework of EMFT. We have demonstrated the performance of TD-EMFT for different excitations in several org. mols., and applications to several realistic systems including a light-harvesting complex, the solvatochromic shift of a chromophore, and the near-edge X-ray absorption spectroscopy (XAS). TD-EMFT offers advantage over linear-response TDDFT in terms of the computational efficiency. Compared to other embedding approaches, TD-EMFT enables efficient calcn. of excited-state gradients and deriv. couplings, which is promising for ab initio simulations of condensed-phase processes involving non-adiabatic transitions.

Additional Information

© 2018 American Chemical Society.

Additional details

Created:
August 19, 2023
Modified:
October 18, 2023