Publication date: 11th March 2026
Fluorescence resonance energy transfer (FRET) is a fundamental mechanism for non-radiative electronic energy transfer (EET) widely observed in materials science [1] and biophysics [2]. However, given the challenge of balancing accuracy and computational cost in quantum calculations of large systems, it is typically modeled within the dipole-dipole approximation, using frequently unsuitable statistical methods to estimate the orientational factor (κ). To overcome the so-called Kappaphobia [3], i.e., the reluctance to determine κ adequately, we combined Molecular Dynamics (MD) simulations with Density Functional Theory (DFT) and Time-Dependent Density Functional Theory (TD-DFT) calculations to properly assess the orientation factor on representative nonfullerene electron acceptors (NFAs). An MD-based solvent evaporation protocol was performed to model experimental spin-coating techniques, and the κ values were determined for 9000 pairs of molecules at thin-film conditions. As a result, the κ2 parameter showed broad dispersion, with significantly higher average values of 0.949 and 0.765 for ITIC-4F and Y6 films, respectively, compared to the standard 0.476 and 2/3 statistical values. Moreover, by considering the system-specific κ2 values, FRET rates became consistent with the experimental trend on exciton diffusion lengths. Finally, we assessed the limitations of the dipole-dipole approximation by employing the transition charges from electrostatic potentials (TrESP) method [4], and identified a simplified molecular descriptor that can be used as a cheap tool to extract initial insights into the orientation factor between NFA-like molecules.
This work is dedicated to the memory of Graziani Candiotto, who passed away during the development of this work. His passion for research and valuable contributions were essential to this study. Grazi was always very supportive and dependable, and his early departure represents a great loss for his friends and the scientific community. L.B. (grant E-26/202.091/2022 process 277806), G.C. (grant E-26/200.627/2022 and E-26/210.391/2022 process 271814), and M.G.M. (grants E-26/210.527/2024 and E-26/204.498/2024) are grateful for financial support from FAPERJ. R.B.R. and M.T.N.V. acknowledge support from FAPESP (grant 2022/04379-3). R.B.R. also thanks Bruna Bicudo for artwork support. M.G.M. also acknowledges the financial support from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and INCT-Materials Informatics. M. K. thanks INCT-Nanovida and the CNPq for financial support (301813/2025-6). The authors also acknowledge the computational support of Núcleo Avançado de Computação de Alto Desempenho (NACAD/COPPE/UFRJ), Laboratório Nacional de Computação Científica (LNCC), Sistema Nacional de Processamento de Alto Desempenho (SINAPAD), and Centro Nacional de Processamento de Alto Desempenho em São Paulo (CENAPAD-SP).
