Fluorescent nanosensors for metal ions

Abstract

Precise detection of pH and metal ion concentrations is crucial across fields such as environmental monitoring and biomedical diagnostics, as both serve as vital indicators of chemical and biological processes. Carbon dots (CDs), known for their photostability, low toxicity, and tunable fluorescence, have emerged as promising fluorescent sensors. This thesis investigates the development of CD-based sensors for these applications, integrating machine learning to improve sensitivity, selectivity, and analytical robustness.

The research began with CDs synthesised from fructose and p-phenylenediamine for pH monitoring across a broad range (pH 3–10). Their fluorescence response, influenced by quantum confinement and surface states, was systematically evaluated. Gaussian process regression (GPR) enabled accurate pH prediction from spectral data, while linear discriminant analysis (LDA) facilitated precise classification. Together, these approaches significantly improved the performance of CD-based pH sensors.

The metal ion sensing capability of CDs was first explored using biomass-derived systems. Building on these findings, fructose-based CDs were engineered for improved selectivity towards Cu(II) and Fe(III). A four-CD array, developed under optimised hydrothermal conditions with varied N-doping levels, was tested using SVM and LDA, successfully distinguishing 14 metal ions, including Cu(II), Co(II), and Ni(II) in mixtures. Real sample testing in environmental water and human urine confirmed the array’s reliability and practicality.

Overall, this research demonstrates the significant potential of combining CDs with machine learning to develop robust, cost-effective, and highly selective sensors. The outcomes establish a foundation for CD-based systems in environmental, biomedical, and industrial applications requiring accurate chemical detection.