Charge-Transfer-State Descriptors from Donor–Acceptor Interfaces Enable Improved Prediction of Indoor Organic Photovoltaic Efficiency
- 1Department of Chemistry, McGill University, 801 Sherbrooke St. W, Montreal, Quebec, Canada, H3A 0B8
Abstract
Organic photovoltaic (OPV) devices have attracted significant attention as lightweight, flexible, and solution-processable alternatives to conventional solar cells. In this work, a novel computational workflow was developed to generate donor–acceptor interfacial adducts for non-fullerene acceptor (NFA)-based indoor OPV systems and to investigate their charge-transfer (CT) characteristics using density functional theory (DFT) and time-dependent DFT calculations. Molecular docking followed by excited-state analysis enabled identification of important CT descriptors, including excitation energies, electron–hole separation, participation ratios, and charge-transfer characteristics. A dataset containing 30 donor–acceptor pairs were subsequently constructed by combining these CT descriptors with molecular fingerprints and frontier molecular orbital (FMO) parameters. Machine learning (ML) models based on ElasticNet and XGBoost algorithms were then trained to predict OPV power conversion efficiency (PCE). Our results demonstrate that incorporating physically meaningful CT descriptors derived from donor–acceptor interfaces significantly improve the ML-based prediction of OPV efficiencies from a small dataset.
Related
- Carbon Nanocones Geometry-Dependent Descriptors Evidence Levels and Polymer Interfaces
- A Reductionist Molecular Dynamics Framework for Identifying Molecular Descriptors of Hydration at Solid–Liquid Biomaterial Interfaces
- Universality of Lithium Metal and Solid Electrolyte Interphase Descriptors of Coulombic Efficiency Governed by Lithium Connectivity
- Accessible-State Descriptors Transfer Across Families but Fail at Larger Scales
- Efficiency and Accessibility Are Not Enough: Chemical-Validity Filtering in High-Throughput Screening of Organic Photovoltaic Materials
BibTeX
@misc{mukherjee2026chargetransferstate,
title = {{Charge-Transfer-State Descriptors from Donor–Acceptor Interfaces Enable Improved Prediction of Indoor Organic Photovoltaic Efficiency}},
author = {Anagh Mukherjee},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010876/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010876/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Charge-Transfer-State Descriptors from Donor–Acceptor Interfaces Enable Improved Prediction of Indoor Organic Photovoltaic Efficiency
AU - Anagh Mukherjee
PY - 2026
DA - 2026/10/22
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010876/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010876/v1
AB - Organic photovoltaic (OPV) devices have attracted significant attention as lightweight, flexible, and solution-processable alternatives to conventional solar cells. In this work, a novel computational workflow was developed to generate donor–acceptor interfacial adducts for non-fullerene acceptor (NFA)-based indoor OPV systems and to investigate their charge-transfer (CT) characteristics using density functional theory (DFT) and time-dependent DFT calculations. Molecular docking followed by excited-state analysis enabled identification of important CT descriptors, including excitation energies, electron–hole separation, participation ratios, and charge-transfer characteristics. A dataset containing 30 donor–acceptor pairs were subsequently constructed by combining these CT descriptors with molecular fingerprints and frontier molecular orbital (FMO) parameters. Machine learning (ML) models based on ElasticNet and XGBoost algorithms were then trained to predict OPV power conversion efficiency (PCE). Our results demonstrate that incorporating physically meaningful CT descriptors derived from donor–acceptor interfaces significantly improve the ML-based prediction of OPV efficiencies from a small dataset.
ER -