Deep learning-assisted self-cleaning cellulose colorimetric sensor array for monitoring black tea withering dynamics.
Journal:
Food chemistry
Published Date:
Sep 30, 2025
Abstract
The withering process is a critical stage in developing the aroma profile of black tea. In this study, we presented an eco-friendly cellulose film-based colorimetric sensor array (CSA) for detecting volatile organic compounds (VOCs) and assessing withering stages using deep learning. TiO was attached to the cellulose film surface, resulting in a self-cleaning TiO-cellulose film. Functionalized cellulose film featuring hydrophobic non-sensing areas were fabricated via site-specific deposition of octadecyltrichlorosilane (OTS). The OTS/TiO-CSA was prepared by drop-coating multiple dyes onto the hydrophilic sensing area of the functionalized cellulose film, exhibiting improved humidity resistance. By assisting with a deep learning model (Long Short-Term Memory), the OTS/TiO-CSA achieved 90 % accuracy in identifying withering stages. Notably, dyes on the OTS/TiO-CSA surface degraded under limited UV exposure, most exceeding 70 % degradation. This study introduces a fabrication strategy for a smart, eco-friendly OTS/TiO-CSA, while demonstrating its potential as a sustainable tool for monitoring tea withering stages.
