Machine learning-assisted chromium speciation using a single-well ratiometric fluorescent nanoprobe

Chromium is widely recognized as a significant pollutant discharged into the environment by various industrial activities. The toxicity of this element is dependent on its oxidation state, making speciation analysis crucial for monitoring the quality of environmental water and assessing the potential risks associated with industrial waste. This study introduces a single-well fluorometric sensor that utilizes orange emissive thioglycolic acid stabilized CdTe quantum dots (TGA-QDs) and blue emissive carbon dots (CDs) to detect and differentiate between various chromium species, such as Cr (III) and Cr (VI) (i.e., CrO₄^2- and Cr₂O₇^2-). The variations of fluorescence spectra of the proposed probe upon chromium species addition were analyzed using machine learning techniques such as linear discriminant analysis and partial least squares regression as a classification and multivariate calibration technique, respectively. Linear discriminant analysis (LDA) demonstrated exceptional accuracy in differentiating single-component and bicomponent samples. Additionally, the findings from the partial least squares regression (PLSR) showed that the sensor created has strong linearity within the 1.0-100.0, 1.0-100.0, and 0.1-15 μM range for Cr₂O₇^2-, CrO₄^2-, and Cr³⁺, respectively. Furthermore, appropriate detection limits were successfully achieved, which were 2.6, 2.9, and 0.7 μM for Cr₂O₇^2-, CrO₄^2-, and Cr³⁺, respectively. Ultimately, the successful capability of the sensing platform in the identification and quantification of chromium species in environmental water samples provides innovative insights into general speciation analytics.