Breakdown and Salvation of Statistical Rate Theory in Biomolecular Photoinduced Electron Transfer

Quantifying rates of photoinduced electron transfer (PET) is crucial to understand the remarkable efficiency of biological photosystems involved in photosynthesis and DNA repair. The ability to distinguish between specific PET channels and compare their timescales with competing photophysical events is also essential for the design synthetic light harvesting systems. However, commonly used statistical rate theories break down owing to strong electronic coupling and the nonequilibrium nature of PET. The assignment of specific time-resolved spectroscopic signatures to particular photophysical processes also remains a notorious challenge. Here, we employed nonadiabatic excited-state dynamics simulations to characterize ultrafast PET in a model system of two stacked adenine nucleobases. We next performed rate calculations for two of the identified PET channels with three different statistical rate theories. In particular, the widely used classical Marcus Theory and Fermi's Golden Rule cannot describe ultrafast PET and offer even qualitative rate trends. We demonstrate that these challenges can be overcome with microcanonical Rice-Rampsperger-Kassel-Marcus (RRKM) theory, which yields PET timescales that are in excellent agreement with excited-state dynamics simulations. We conclude that while (non-)adiabatic RRKM theory cannot grasp the dynamical nature of PET, it offers strong predictive capacity for the design of efficient photoredox-active chromophores.