ak-mueller

We present a multistate molecular mechanics model that captures both the ground (S0) and triplet excited (T1) states of hemithioindigo-based photoswitches — enabling nanosecond-scale molecular dynamics with near quantum mechanical accuracy.

Curious about how machine learning is transforming excited-state molecular dynamics? Our latest overview dives into best practices for applying machine learning in non-adiabatic dynamics simulations, from data pre-processing and surface fitting to post-simulation analysis. Learn how machine learning approaches can help overcome computational bottlenecks and uncover patterns in complex photochemical systems — paving the way for data-driven design of photochemistry and -physics.

SHNITSEL-data (Surface Hopping Nested Instances Training Set for Excited-state Learning) is an open-access dataset containing 418,870 high-accuracy ab initio data points for nine organic molecules. It includes quantum chemical properties in ground and electronically excited singlet and triplet states, such as energies, forces, dipole moments, nonadiabatic couplings, transition dipoles, and spin-orbit couplings. Generated with state-of-the-art methods, SHNITSEL-data supports the development of machine learning models for excited-state processes in photochemistry and photophysics.

How does Near-Infrared-Photoswitching work? Our newest research delivers an ultrafast molecular movie of the all-red-light photoswitch peri-Anthracenethioindigo (PAT) in action! Guided by excited-state simulations, its mechanism of motion is fully revealed. We show that PAT double bond rotation occurs exclusively from the triplet state – but it is stable in air due to very favorable energy levels.