Research

The methods that capture the right physics often cannot be applied to realistically sized systems. Embedding approaches partition a system into an active region treated at a high level and an environment treated at a lower level. The challenge is doing this self-consistently for periodic systems with long-range electrostatics and polarization response.

With Prof. Sierka’s group I work on density-functional embedding methods inside TURBOMOLE, aiming for embedding that stays dependable across heterogeneous environments, interfaces, and condensed-phase systems. I am equally interested in excited-state and time-dependent properties (e.g. the response of materials under laser fields), connecting naturally to the CRC/SFB 1375 NOA at Jena.

Temporary anion states, or negative ion resonances, sit just above the electron-detachment threshold and play a role in electron-driven chemistry and charge transfer in extended systems. Standard DFT struggles with them because the underlying states are not square-integrable and conventional local exchange–correlation potentials fail at the right asymptotic behaviour.

In a JPCL cover article (Ghosal, Joshi, Voora) we showed that combining exact exchange with the RPA correlation potential, probed by a complex absorbing potential, gives positions and widths far more reliably than purely local treatments. I am extending these nonlocal frameworks to dynamical long-range polarization and to settings where resonances couple to external fields, surfaces, or condensed-phase environments.

Noncovalent interactions (hydrogen bonds, metal–solvent coordination, dispersion forces, carbene reactivity) shape condensed-phase chemistry and have long been a test for DFT, especially when fragments are open-shell. Errors in the underlying density propagate into interaction energies, so even reasonable functionals are unreliable in these regimes.

In a JCP paper (Joshi, Voora) we introduced perturbative singles corrections (RPAS) to the random-phase approximation. RPAS improves interaction energies across hydrogen-bonding, metal–solvent, carbene–solvent, and dispersion regimes, with the biggest gains in the open-shell cases that have been hardest for post-Kohn–Sham methods. The broader interest is in many-body methods that stay dependable for realistic, multi-reference, and weakly bound systems.