On the importance of initial conditions in nonadiabatic molecular dynamics

A significant number of computational strategies have been developed over the past decades to simulate the excited-state dynamics of molecules beyond the Born-Oppenheimer approximation. When combined with advanced electronic-structure methods, these techniques – often based on coupled or uncoupled trajectories – have the potential to unravel the mechanistic details of photochemical reactions of direct interest to physical, atmospheric, and material chemists.

Despite all these extensive developments, the very first step of any photochemical reaction – photoexcitation – is dramatically approximated in most applications of nonadiabatic dynamics.[1] In an ideal world, the initial ground-state nuclear wavefunction, representing the molecule of interest, should be coupled, for example, to the time-dependent electromagnetic field of a laser pulse, leading to a transfer of nuclear amplitude to an excited electronic state based on the precise characteristics of the field. In practice, two strong approximations are made to simplify the photoexcitation process.[2] (1) A harmonic Wigner distribution is often used to represent the ground-state distribution of the molecule, from which initial nuclear positions and momenta can be sampled. (2) These pairs of nuclear positions-momenta are then promoted to a given excited electronic state and used to initiate the nonadiabatic molecular dynamics – a so-called sudden excitation approximation.

In this talk, I will discuss the limitations of these two approximations, supported by numerical simulations of realistic molecular systems. I will then present more rigorous strategies to describe the photoexcitation process: ab initio molecular dynamics with a quantum thermostat to sample pairs of nuclear positions-momenta in the ground electronic state[3] and the promoted nuclear density approach to incorporate the effect of a laser pulse at the level of the initial conditions at no cost.[4]