Computational investigation of curcumin–polymer interactions: a DFT–ML integrated framework for electronic structure and biodegradation analysis of PGA and PLA complexes

Abstract

Biodegradable polymers such as poly(glycolic acid) (PGA) and poly(lactic acid) (PLA) are widely used in biomedical applications due to their biocompatibility and tunable degradation behavior. Here, a combined Density Functional Theory (DFT) and Machine Learning (ML) framework is employed to investigate the structural and electronic properties of Curcumin–polymer complexes. Geometries were optimized at the B3LYP/6-311++G(d,p) level, and frontier-orbital descriptors were refined using ωB97X-D single-point calculations. ESP mapping shows that Curcumin–PGA exhibits a broader potential range than Curcumin–PLA, indicating stronger electrostatic polarization. FMO analysis reveals a pronounced contrast in electronic response: Curcumin–PGA displays a significantly reduced HOMO–LUMO gap (2.52 eV), whereas Curcumin–PLA retains a much larger gap (6.86 eV). AIM analysis confirms medium-strength hydrogen bonding. ML modeling, used qualitatively due to the small dataset, identifies ΔEgap and ESP range as dominant descriptors of degradation behavior. This integrated DFT–ML approach clarifies Curcumin–polymer interactions and supports rational design of biodegradable drug-delivery systems.