An integrated quantum-classical protocol for the realistic description of solvated multinuclear mixed-valence transition metal complexes and their solvatochromic properties

Linear cyanide-bridged polymetallic complexes, which undergo photoinduced metal-to-metal charge transfer,

represent prototypical systems for studying long-range electron-transfer reactions and understanding the role played by specific

solute−solvent interactions in modulating the excited-state dynamics. To tackle this problem, while achieving a statistically

meaningful description of the solvent and of its relaxation, one needs a computational approach capable of handling large

polynuclear transition-metal complexes, both in their ground and excited states, as well as the ability to follow their dynamics in

several environments up to nanosecond time scales. Here, we present a mixed quantum classical approach, which combines largescale

molecular dynamics (MD) simulations based on an accurate quantum mechanically derived force field (QMD-FF) and selfconsistent

QMD polarized point charges, with IR and UV−vis spectral calculations to model the solvation dynamics and optical

properties of a cyano-bridged trinuclear mixed-valence compound (trans-[(NC)5FeIII(μ-CN)RuII(pyridine)4(μ-NC)FeIII(CN)5]4−).

We demonstrate the reliability of the QMD-FF/MD approach in sampling the solute conformational space and capturing the local

solute−solvent interactions by comparing the results with higher-level quantum mechanics/molecular mechanics (QM/MM) MD

reference data. The IR spectra calculated along the classical MD trajectories in different solvents correctly predict the red shift of the

CN stretching band in the aprotic medium (acetonitrile) and the subtle differences measured in water and methanol, respectively. By

explicitly including the solvent molecules around the cyanide ligands and calculating the thermal averaged absorption spectra using

time-dependent density functional theory calculations within the Tamm−Dancoff approximation, the experimental solvatochromic

shift is quantitatively reproduced going from water to methanol, while it is overestimated for acetonitrile. This discrepancy can likely

be traced back to the lack of important dispersion interactions between the solvent cyano groups and the pyridine substituents in our

micro solvation model. The proposed protocol is applied to the ground state in water, methanol, and acetonitrile and can be flexibly

generalized to study excited-state nonequilibrium solvation dynamics.