AI-Enhanced SERS with Probe Combinations for Concurrent Identification and Quantification of Coexisting Metal Ions in Water

Efficient and simultaneous detection of coexisting metal ions is essential for preserving water quality and mitigating ecological risks during environmental emergencies. However, traditional methods often fall short, either lacking the sensitivity required for trace-level detection or the ability to quantify multiple ions concurrently. Here, we propose an AI-enhanced surface-enhanced Raman spectroscopy (SERS) approach that combines advanced spectral analysis with tailored probe combinations, enabling precise identification and quantification of four coexisting metal ions, even in complex environmental water. By using chelating agents used in spectrophotometric assays as probes, we constructed a high-dimensional spectral array that generates comprehensive "super profiles" of 84 combinations of coexisting metal ions at various concentrations. A self-attention-based deep learning architecture was applied to leverage this extensive data set, addressing challenges associated with spectral overlap and nonlinear responses. Validation across various environmental samples, including river water, industrial effluents, and domestic sewage, demonstrated exceptional accuracy, with prediction errors for mixtures of four ions consistently below 10%. This work provides a scalable solution for the simultaneous detection of coexisting metal ions, enabling on-site, high-throughput applications in environmental monitoring and advancing water quality management strategies.