I am interested in various coarse-grained modeling methods for polymers to investigate multiscale structural and dynamic properties. My research interests also include their applications in copolymer interactions, membrane surface analysis (such as membrane distillation), and environmentally sustainable technologies.
Overview of Current Research and Future Work:
Anisotropic coarse-grained (CG) modeling offers a promising approach to overcome limitations of isotropic models, such as poor directional packing and unphysical chain crossing. Polytetrafluoroethylene (PTFE) is a semi-rigid polymer, with rigidity arising from strong and highly electronegative C–F bonds. Anisotropic effects increase by over 200% as the coarse-graining level increases from 4 to 12 CF₂ units per bead.
In my PhD research, I develop ellipsoidal coarse-grained chain models for PTFE melt, aiming to accurately capture both local and global structural features [1]. By selecting appropriate bead sizes (representing 6 or 8 CF₂ units), the model maintains accurate geometric representation across different chain lengths and temperatures. The incorporation of ghost particles at the ends of each ellipsoid’s longest principal axis effectively prevents chain crossing and corrects torque applied to the bead’s center of mass. By using the analytical RE-squared potential, both NVT and NPT simulations can be reliably performed, supported by physically informed non-dimensional analysis. Moreover, angle-dependent radial distribution functions (RDFs) demonstrate strong performance in capturing anisotropic packing behavior.
While the current model shows promising results, several limitations remain. Its applicability to entangled PTFE chains and performance across phase transitions to the solid state have yet to be evaluated. Additionally, the iterative optimization process for structural matching is computationally expensive. Therefore, I am currently assessing the adaptability of the existing model. In parallel, I am also exploring the development of a machine learning-based coarse-grained potential for linear polymers using a graph neural network-based force-matching framework [2]. By automatically learning interparticle interactions, this approach has the potential to enhance the transferability of the model across different polymer systems and conditions.
[1] Hsu YT, Yu HY. Macromolecules, 2025. DOI: 10.1021/acs.macromol.4c02413
[2] Doerr S, Majewski M, et al. J. Chem. Theory Comput., 2021. DOI: 10.1021/acs.jctc.0c01343