Determining charge transport regimes in organic molecular crystals: a machine learning framework

Charge transport performance in organic molecular crystals (OMCs) is crucial for advancements in nanotechnology. Experiments have shown metallic-like and semiconducting charge transport regimes in OMCs, mediated by free electrons with Bloch-like oscillations (BOs) and polaronic states. In metallic-like regimes, the charge propagates as a wave, while in semiconducting regimes, it travels as a quasi-particle coupling charge with a cloud of lattice phonons. While the conditions for polaronic states in OMCs are well-established, those enabling BOs still need to be understood. In this study, we identify the electronic and structural properties of OMCs that favor the formation of polarons or BOs by analyzing their linear and wave transport properties. We employ semiempirical non-adiabatic dynamical simulations at the picosecond scale and machine learning methods to map the parameter spaces where the charge transport occurs via polarons or BOs. The dynamical simulations are based on a general model Hamiltonian developed to address OMCs. Our results reveal that increasing the electronic transfer rate between molecules, the crystal's speed of sound, and its metallicity favors the formation of BOs. BOs in OMCs can exhibit frequencies around 2 THz and current amplitudes up to 3000 |e| ps^-1, opening up possibilities for high-frequency applications. Conversely, large polarons are predominantly formed based on the interplay between intra- and intermolecular electron-lattice interactions.