Quantum Dot-Based Ratiometric Optical Nanosensor for Machine Learning-Assisted Visual and Spectral Speciation of Arsenite and Arsenate

Arsenic contamination in drinking water remains a major global public health concern. Because arsenic toxicity depends strongly on its chemical form, speciation analysis is essential for understanding its environmental behavior, toxicity, and bioavailability. Although laboratory-based techniques provide accurate measurements, there is an increasing demand for portable and cost-effective platforms capable of on-site detection and discrimination of inorganic arsenic species. In this study, a ratiometric fluorescent platform was developed for the spectral and visual speciation of arsenite (AsO 3 3– ) and arsenate (AsO 4 3– ). The sensing strategy relies on fluorescence quenching of a mixture of green-emissive thioglycolic acid-capped CdTe quantum dots (GQDs) and red-emissive mercaptopropionic acid-capped CdTe quantum dots (RQDs) by glutathione (GSH). Upon exposure to AsO 3 3– or AsO 4 3–, fluorescence was restored to different extents, generating distinct response patterns that enabled arsenic speciation within 25 min. Under optimized conditions, the proposed “on–off–on” nanosensor enabled determination of AsO 3 3– and AsO 4 3– over concentration ranges of 0.70–50 and 0.15–20.00 μmol L –1, respectively, with detection limits of 0.21 μmol L –1 for AsO 3 3– and 0.06 μmol L –1 for AsO 4 3– . Linear discriminant analysis (LDA) was identified as the optimal machine-learning model, achieving classification accuracies above 93.0%. Application to spiked groundwater and wastewater samples yielded recoveries of 90.0–110.0%, RSDs of 0.3–8.9%, and a classification accuracy of 97.2%. RGB data extracted from fluorescence color patterns using a smartphone-based imaging cabinet were also successfully applied to arsenic discrimination and quantification. By integrating ratiometric fluorescence sensing, smartphone imaging, and machine-learning-assisted classification, the proposed platform provides a portable, cost-effective, and user-friendly strategy for arsenic speciation with potential for environmental monitoring and point-of-use water quality assessment.