Geometry Optimization in Polarizable QM/MM Models: The Induced Dipole Formulation

We present the mathematical derivation and the computational implementation of the analytical geometry derivatives for a polarizable QM/MM model (QM/MMPol). In the adopted QM/MMPol model, the focused part is treated at QM level of theory, while the remaining part (the environment) is described classically as a set of fixed charges and induced dipoles. The implementation is performed within the ONIOM procedure, resulting in a polarizable embedding scheme, which can be applied to solvated. and embedded systems and combined with different polarizable force fields available in the literature. Two test cases characterized by strong hydrogen-bond and dipole-dipole interactions, respectively, are used to validate the method with respect to the nonpolarizable one. Finally, an application to geometry optimization of the chromophore of Rhodopsin is presented to investigate the impact of including mutual polarization between the QM and the classical parts in conjugated systems.