Abstract
Pollution by heavy metals represents a serious threat to both the environment and human health. A facile multi-emission fluorescence sensor array based on carbon dots (QR-CDs) and a novel lanthanide complex (EDTA-Tb3+) was constructed, and is capable of obtaining simultaneous, multidimensional data, which can improve the detection efficiency and accuracy when it comes to multiple heavy metal ions. To meet the challenges of establishing a unified model, we built an innovative unified model (SX-model) by the “stepwise prediction” strategy combined with machine learning methods to obtain optimal screening methods. This model integrates classification and concentration models under the framework of the tree-based pipeline optimization technique (TPOT). Then, the extreme random forest (ERF) was selected as the classification model method with the highest accuracy among various methods through TPOT. This sensor array demonstrated sensitive detection of seven heavy metal ions in the range of 0.05–50 μM with an accuracy of 95.6%. The ability to identify binary mixed samples simultaneously and effectively was greatly enhanced. Furthermore, the metal ions in 288 real samples (obtained from lake water and soil samples) were effectively identified with 93.3% and 100% accuracy, respectively. The proposed original SX-model-assisted multi-emission sensor not only overcomes issues regarding low sensibility but also breaks the bottleneck of analysis methods, showing great application potential in the array detection field.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have