First-principles and reduced-Hamiltonian modeling of charge-transfer dynamics in non-fullerene organic photovoltaics
Abstract
Organic photovoltaics (OPVs) based on non-fullerene acceptors (NFAs) have surpassed 21% power conversion efficiency, yet a predictive mechanistic understanding of charge-transfer (CT) processes at the donor:acceptor heterojunction remains incomplete. In particular, two distinct factors govern CT — energetic offsets (Marcus driving force) and electron-vibrational (vibronic) coupling — but their relative weights cannot be cleanly separated from experimental data alone, hampering the rational design of next-generation NFAs.
This project develops a quantitative, first-principles + reduced-Hamiltonian framework to disentangle these contributions for A-D-A-D-A type bipolar NFAs. The bipolar molecule A6 (a Y18-end-group-modified small-molecule that, uniquely, serves both as donor (paired with PCBM, PCE 7.2%) and as acceptor (paired with P3HT, PCE 5.4%)) provides a single-molecule internal control: the same chromophore exhibits a 1000-fold difference in LE→CT rate between the two pairings. Preliminary Marcus analysis shows this rate ratio cannot be explained by energetic offsets alone, implicating vibronically-gated electronic coupling.
The computational workflow involves:
(i) ground-state and excited-state DFT/TDDFT calculations on A6 and benchmark NFAs (Y6, ITIC) at the ωB97X-D/def2-TZVP level using ORCA 6, with truncation of solubilizing alkyl chains to keep the active π-system tractable (~120 heavy atoms);
(ii) Franck-Condon-weighted Marcus-Jortner rate analysis from DFT-derived reorganization energies (4-point method) and normal-mode projections;
(iii) electronic coupling V between A6 and partner molecules (PCBM, P3HT oligomer) for ~100 dimer geometries using fragment-orbital DFTB (FODFTB) screening, calibrated against 8-10 ωB97X-D + Fragment Excitation Difference / Charge Difference (FED-FCD) single points;
(iv) construction of a 5-site (A-D-A-D-A) Holstein-Peierls Hamiltonian, with intramolecular site energies and hopping integrals parameterized from (i); and
(v) open-quantum-system dynamics (HEOM via QuTiP-BoFiN, with Redfield as fallback) to simulate the LE→CT branching observed experimentally.
Outputs include two peer-reviewed publications — a methods paper presenting the open-source framework (target J. Chem. Phys.) and an A6-mechanism paper (target JPCL/JACS) — and a public Python package (MIT licensed) for reproducible vibronic CT modeling in donor-acceptor systems. The framework complements existing high-throughput AI/ML screening pipelines for OPV materials by providing the mechanistic backbone they currently lack, and is directly transferable to other Y-family NFA systems.
