A new method for the treatment of coupled electron and ion dynamics will be developed, using time-dependent density functional theory (TDDFT) for electrons and semiclassical dynamics for nuclei. The method incorporates branching, electronic relaxation and decoherence, as well as essential nuclear quantum effects in a new and more theoretically sound way than current methods do. More accurate non-empirical exchange-correlation functionals will be developed from first-principles for the simulation of real-time electron dynamics in non-equilibrium situations, correcting deficiencies of functionals used today. In this way, we will develop an accurate methodology to model dynamical processes in solar cell device operation, and eventually to be used in their design.
For modeling the electronic structure and dynamics of solar-cells, TDDFT is the favored quantum mechanical method, due to its system-size scaling. However, (i) the exchange-correlation functionals in use today lack known features of the exact functionals that are particularly important when the electronic system is excited out of its ground-state. This impacts their ability to accurately and reliably describe processes in solar energy conversion, especially charge-transfer dynamics and level alignment. Further, (ii) the usual treatments in coupling electrons to nuclei are either qualitatively inadequate in their description of electron-nuclear correlation (Ehrenfest), or suffer from over-coherence (surface-hopping) for which a consistent solution has remained elusive. An ab initio approach to the electron-ion coupling problem would lead to greater reliability. We will develop new methods, based on first-principles, that confront these problems. For (i), we focus on implementing an exact condition that we recently derived, that prevents spuriously detuned resonances. The spurious detuning plagues approximate functionals in use today when the electronic system has evolved far from its ground-state, and impacts level alignment after photoexcitation and the ensuing charge-transfer dynamics. We also will develop new approximations that either utilize initial-state dependence to minimize the impact of step structures that we recently found to be present in the exact functional that are missing in approximations, or that model important parts of these structures, and we will determine the system-size scaling of these non-adiabatic features. For (ii), we will develop a new coupled-dynamics method based on the recently introduced exact factorization approach. Here we have identi ed features of the time-dependent exact potential energy surfaces on which the nuclei evolve, that are responsible for wavepacket branching and decoherence, and, with this knowledge, we are ready to develop a practical approach from first principles. Our proposed approach treats the nuclei semiclassically, coupled to TDDFT electrons via simplifed electron-ion correlation terms arising from the exact factorization. Success of the project will yield accurate, practical, and predictive methods, sound in their basic theory, to model electron and nuclear dynamics in solar energy conversion. The ultimate goal is to use these tools for the computational design of new materials for solar cell devices of high efficiency.
