IHSS&IWA26 / BRNO / CZECHIA / 23–28 August 2026 BOOK OF ABSTRACTS 46 Analysis and Characterization Monday, 24 August 2026 / Hall B+C SL17 Dissecting the Optical Properties of Humic Substances: from Fundamental Photophysics to the AI-driven Insights Evgeny Shirshin1, Boris Yakimov1, Anna Rubekina1, Ioanna Gorbunova1, Irina Perminova2 1 M.V. Lomonosov Moscow State University, Department of Physics, 119991, Moscow, Russia, eshirshin@gmail.com 2 M.V. Lomonosov Moscow State University, Department of Chemistry, 119991, Moscow, Russia Despite optical spectroscopy being a classical tool for the assessment of HS and DOM compositional variations and transfer, the origin of their optical properties remains debatable. The question of whether the optical spectra of HS are the result of superposition of spectra from individual chromophores or some universal physical mechanisms, which lead to striking similarity of spectral band shapes, exist, is the key to HS optics. Since the 2004 paper of Del Vecchio and Blough, several groups aimed at elucidating the role of charge transfer interactions in the formation of HS optical properties and identifying candidate molecules for HS fluorophores. Our group, with its background in fundamental photophysics and laser spectroscopy, also joined this exciting research area, which was supported by the Young Investigators Research Grant from IHSS. In scope of this work, two main conclusions were made: (1) ultrafast (~1 ps) decay component is present within all HS and DOM samples, being a universal signature of photophysical processes and excitation energy redistribution within a mixture of fluorophores, and (2) oxidation of simple aromatic molecules yield heterogenous systems of fluorophores with optical properties similar to that of HS, including ultrafast decay and spectral diffusion. During further research, we observed that HS-like processes are observed within a broad class of living systems, cells and tissues, are responsible for their NIR autofluorescence and can be successfully used for biomedical diagnostics. However, structural identification of the corresponding fluorophores was almost impossible because of their heterogeneity and low concentration, requiring completely novel approaches for shifting the understanding of optical properties to the next level. For this, we developed and verified the AI-based model, which predicts optical properties of molecules based on its structure, based on the open access datasets of >104 fluorophores. Next, using the generative AI models we simulated reactions in model systems in
RkJQdWJsaXNoZXIy NDA4Mjc=