Answers are grounded in this paper and cite their sources.
Heavy metal contamination of water resources, particularly by Cu²⁺, Cd²⁺, and Pb²⁺, remains a persistent environmental and public health concern, creating a continued demand for ion-selective sensors capable of detecting these metals reliably. While potentiometric membrane sensors are widely used for this purpose, the rational design of new ionophores has been hampered by reliance on QSPR models built from abstract, computationally derived descriptors that are difficult to relate to actual chemical structure. In this study, three interpretable MLR-based QSPR models were developed to predict the potentiometric sensitivity of ion-selective membranes toward Cu²⁺, Cd²⁺, and Pb²⁺, using datasets of 43–45 structurally diverse ionophores. Unlike earlier models built on topological or autocorrelation indices, the descriptors used here correspond to recognizable structural features, including atom counts, donor-group identities, and specific coordinating fragments such as amides, amines, diamides, polyethers, and phosphine oxides. The resulting models showed strong predictive performance, with R² values of 0.9064, 0.9373, and 0.9457 for Cu²⁺, Cd²⁺, and Pb²⁺, respectively, and corresponding external validation coefficients of 0.8558, 0.9184, and 0.9302. Williams-plot analysis confirmed that the models operate within a well-defined applicability domain; although one training compound in the Cd²⁺ model was identified as a high-leverage influential object, the models generally demonstrate high reliability and stability. Beyond their predictive accuracy, the models reveal clear structure–sensitivity relationships for each metal ion, offering practical guidance for selecting or modifying ligand structures to improve sensor performance. These findings provide a transparent and chemically grounded framework that can support the rational design of next-generation potentiometric sensors for environmental heavy-metal monitoring.
No takes yet. Share an insight, caveat, or question.
Keshavarz et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: