Excited-state nonadiabatic dynamics in explicit solvent using machine learned interatomic potentials
This work addresses a computational bottleneck for researchers studying photoinduced processes in chemistry, offering a more efficient method for excited-state dynamics simulations, though it is incremental as it builds on existing ML potentials and QM/MM frameworks.
The paper tackled the high computational cost of excited-state nonadiabatic simulations in explicit solvent by replacing traditional QM/MM with a machine-learned interatomic potential (FieldSchNet), demonstrating that the ML/MM model reproduces electronic kinetics and structural rearrangements from QM/MM reference simulations for furan in water with five coupled singlet states.
Excited-state nonadiabatic simulations with quantum mechanics/molecular mechanics (QM/MM) are essential to understand photoinduced processes in explicit environments. However, the high computational cost of the underlying quantum chemical calculations limits its application in combination with trajectory surface hopping methods. Here, we use FieldSchNet, a machine-learned interatomic potential capable of incorporating electric field effects into the electronic states, to replace traditional QM/MM electrostatic embedding with its ML/MM counterpart for nonadiabatic excited state trajectories. The developed method is applied to furan in water, including five coupled singlet states. Our results demonstrate that with sufficiently curated training data, the ML/MM model reproduces the electronic kinetics and structural rearrangements of QM/MM surface hopping reference simulations. Furthermore, we identify performance metrics that provide robust and interpretable validation of model accuracy.