Machine learning assisted spectroscopic investigation of fluorescence quenching in Hypocrellin B with magnetic nanoparticles

Abstract

Abstract The present study investigates the interaction behavior of the photosensitizer Hypocrellin B (HB) with Fe 3 O 4 nanoparticles and Fe 3 O 4 /CdTe nanocomposites using a combined spectroscopic and machine learning approach. The structural characteristics of the synthesized nanoparticle were confirmed by UV-Visible and Fourier transform infrared (FTIR) spectroscopy. Steady-state and time-resolved fluorescence studies revealed the formation of ground-state complexes accompanied by static fluorescence quenching. Fluorescence titration experiments demonstrated that HB exhibits stronger binding affinity towards Fe 3 O 4 nanoparticles compared with Fe 3 O 4 /CdTe nanocomposites. The calculated bimolecular quenching constant (k q ) in the range of 5.4 × 10 13 to 9.4 × 10 13 M − 1 s − 1 further supported the static quenching mechanism. The thermodynamic feasibility of possible photoinduced electron- transfer interaction was also evaluated using the Rehm-Weller equation and the energy level analysis. In addition, principal component analysis (PCA) and Support vector regression (SVR) were employed to analyze spectral variation and predict interaction-related parameters. The PCA results provided insight into the spectral changes associated with HB-nanoparticle complex formation, while the SVR model demonstrated promising predictive capability with R 2 values above 0.98 and prediction errors below 5%. The combined experimental and multivariate analysis offers valuable insight into HB-nanoparticle interactions and highlights the potential of integrating spectroscopic techniques with data-driven modelling for fluorescence quenching studies relevant to photodynamic and biomedical applications.