Molecular photochemistry at the nanoscale and developments towards modeling nonadiabatic dynamics on many electronic states

Nonadiabatic molecular dynamics (NAMD) simulation methods are useful for gauging the presence of and characterizing pathways for efficient nonradiative recombination (NRR), which can greatly reduce the efficacy of optoelectronic materials. This utility is herein applied to lead-halide perovskites, which have received great attention for use in solar energy conversion, since they readily absorb visible light and have a low propensity for NRR, to further the understanding of why these materials do not readily undergo NRR. More specifically, we investigate the energetic accessibility of conical intersections (CIs) in an archetypal lead-halide perovskite, CsPbBr3, through simulation of the dynamics of an excited molecule sized cluster model (Cs4PbBr6) and time-independent optimizations of a nanoparticle model (Cs12Pb4Br20) of the CsPbBr3 surface. Using knowledge of the perovskite potential energy surfaces (PESs) gained from NAMD simulations of the electronically excited cluster model, geometries of both cluster and nanoparticle model minimal energy CIs (MECIs) were optimized. Energies of the observed MECIs were corrected for both dynamic electron correlation and spin-orbit coupling and were observed to be greater than the optical band gap of bulk CsPbBr3. This suggests that this particular set of intersections is energetically inaccessible after optical excitation and does not facilitate efficient NRR in this material. Characterization of the PESs surrounding the MECIs suggests that the ionic nature of the bonds in CsPbBr3 could be a contributing factor to the high energy of the observed CIs. Thus, the investigation of other, ionic materials for optoelectronic applications would be a promising direction for future research.Though NAMD methods have been successfully applied to study photophysical processes in a range of fields, such as molecular photochemistry, spectroscopy and materials science, the application of these methods has been limited to processes in which a small number of electronic states are populated when they use properties of the PESs to refine trajectory dynamics. This limitation arises from the immense computational cost of computing a large number of PESs at each simulation time step. Here, we introduce the collapse to a block (TAB) correction to Ehrenfest dynamics and its adaptation for dense manifolds of states (TAB-DMS), which model electronic coherence loss in a state-pairwise fashion. The employed state-pairwise formulation is demonstrated to be necessary for maintaining the most physical description of coherence loss when a model exists in a coherent superposition of three or more electronic states. Additionally, the TAB-DMS adaptation employs the history of the time-dependent Ehrenfest wavefunction to efficiently generate a small number of approximate electronic states, eliminating the need to compute a large number of electronic PESs. We demonstrate through the computation of TAB and TAB-DMS ensemble probabilities of transmission for a series of one-dimensional model problems, that TAB-DMS retains the accuracy of TAB simulations when only a modest number of approximate eigenstates are computed. The accuracy of the TAB-DMS methodology is also demonstrated to be systematically improvable through the computation of additional approximate electronic states. These results indicate pairing TAB-DMS with ab-initio electronic structure theory and efficient GPU propagation schemes would be fruitful next steps towards modeling real photophysical processes occurring over many electronic states.