We demonstrate how a full QM/MM derivatization of the recently developed GOCAT model can be utilized in the global optimization of molecular embeddings. To this end, we provide two distinct examples: An $$\text {S}_\text {N}2$$ S N 2 reaction, and one enzymatic example of recent interest, the ketosteroid isomerase. These serve us to highlight the advantages of such an approach and sketch the roadmap for further improvements.

A promising application for design and deployment of molecular machines is nanoscale transport, driven by artificial cilia. In this contribution, we present several further steps toward this goal, beyond our first-generation artificial cilium (Raeker et al., J. Phys. Chem. A 2012, 116, 11241). Promising new azobenzene-derivatives were tested for use as cilium motors. Using a QM/MM partitioning in on-the-fly photodynamics, excited-state surface-hopping trajectories were calculated for each isomerization direction and each motor version. The methods used were reparametrized semiempirical quantum chemistry together with floating-occupation configuration interaction as the QM part and the OPLSAA-L forcefield as MM part. In addition, we simulated actual particle transport by a single cilium attached to a model surface, with varying attachment strengths and modes, and with transport targets ranging from single atoms to multi-molecule arrangements. Our results provide valuable design guidelines for cilia-driven nanoscale transport and emphasize the need to carefully select the whole setup (not just the cilium itself, but also its surface attachment and the dynamic cilium-target interaction) to achieve true transport. © 2018 Wiley Periodicals, Inc.