Deep Learning-Assisted Room-Temperature Phosphorescence Sensor Array Based on Host–Guest Doping for Visual Discrimination of Triazole Fungicides

Detection and discrimination of structurally similar triazole fungicide (TF) subtypes remain highly desirable yet challenging. This work presents a straightforward room-temperature phosphorescence (RTP) sensor array for the visual discrimination of TFs, based on host-guest doping-induced RTP signal amplification. Five TF subtypes were doped into a boric acid (BA) matrix via thermal treatment, yielding intense, multicolored, and long-lived afterglow composites. The rigid BA matrix amplified the phosphorescence of guest molecules by reducing the singlet-triplet energy gap and suppressing nonradiative decay. The composites exhibited concentration-dependent RTP fingerprints in terms of emission color, intensity, and lifetime, enabling the discrimination of TFs, including binary and ternary mixtures, through linear discriminant analysis and hierarchical cluster analysis. Time-resolved RTP signal collection effectively eliminated background interference from autofluorescence and scattering, ensuring robust detection in real samples. Furthermore, an intelligent artificial vision platform utilizing the DenseNet algorithm achieved automated identification of TF types and concentrations directly from afterglow images with high accuracy (>91%) and speed (<1 s). This study offers a visual strategy for trace-level TF discrimination, demonstrating significant potential for on-site environmental and food safety monitoring.