| Description |
Gianluca Levi University of Trieste and University of Iceland
Abstract:
Charge transfer and Rydberg excited states pose a significant challenge for electronic structure calculations, including practical implementations of the linear-response time-dependent density functional theory. I will illustrate how excited electronic states can be calculated by variationally optimizing the orbitals in state-specific, time-independent density functional calculations, which significantly improves the description of charge transfer and Rydberg excitations. The approach is based on finding saddle points on the electronic energy surface corresponding to excited states via direct optimization of the orbitals with quasi-Newton and mode following methods [1-2]. It has been implemented in the ORCA program as well as the GPAW software [3], where a plane wave or real space grid basis set can be used to represent the highly diffuse Rydberg orbitals. For charge transfer excitations, the error on the calculated excitation energy is found to be largely independent of the charge transfer distance, and the excited state energy exhibits the correct 1/R variation as a function of the donor-acceptor distance, even when local and semilocal functionals are used [1, 4]. Calculations on a charge transfer excited state of a molecular photoswitch [5] show that the shape of the energy surface as a function of the nuclear coordinates closely agrees with higher-level wave function calculations, and the results are largely independent of the functional. For Rydberg states of molecules, the excitation energy and dissociation energy curves obtained in orbital-optimized calculations with generalized gradient approximation functionals are in remarkably good agreement with experimental estimates and higher-level calculations, and are typically improved when exact exchange or self-interaction correction is included [6]. I will also illustrate how the approach can be combined with molecular dynamics simulations including explicit solvent effects. Such simulations have been used to model the ultrafast reorganization of the solvation shell of a polar solvent accompanying intramolecular charge transfer [5] and to identify pathways of flow of excess energy from a photoexcited metal complex to the solvent [7], thereby assisting the interpretation of time-resolved experiments in solution. Finally, I will discuss the conceptual and practical challenges of computing transition properties, such as the transition dipole moment, using orbital-optimized excited states. I will illustrate our efforts to address them and show preliminary results of calculations of transition dipole moments of molecules, a step toward making the orbital-optimized excited state framework a practical tool for computing optical spectra. [1] Y. L. A. Schmerwitz, E. Selenius, G. Levi, arXiv:2501.18568 (2025). [2] Y. L. A. Schmerwitz, G. Levi and H. Jónsson, J. Chem. Theory Comput., 19, 3634 (2023). [3] J. J. Mortensen et al., J. Chem. Phys. 2024, 160, 092503. [4] E. Selenius, A. E. Sigurdarson, Y. L. A. Schmerwitz and G. Levi, J. Chem. Theory Comput., 20, 3809-3822 (2024). [5] K. Mitterer et al., Research Square https://doi.org/10.21203/rs.3.rs-7775086/v1. [6] A. E. Sigurdarson, Y. L. A. Schmerwitz, D. K. V. Tveiten, G. Levi and H. Jónsson, J. Chem. Phys., 159, 214109 (2023). [7] B. O. Birgisson, A. O. Dohn, H. Jónsson, G. Levi, J. Chem. Phys. 162, 044306 (2025). |
CMSP Seminar (Atomistic Simulation Seminar Series): Orbital-Optimized Density Functional Calculations of Charge Transfer and Rydberg Excited States
Go to day