H₂TiO₃ ion-sieve modified pencil graphite electrode integrated with artificial neural network calibration for...

The widespread use of lithium (Li) has resulted in lithium-ion (Li⁺) contamination across a wide range of concentrations in the environment, posing potential risks to environmental safety and ecological health. However, the accurate and reliable quantification of Li⁺ over such a broad concentration range remains challenging. Herein, a lithium ion sieve (H₂TiO₃)-engineered electrochemical sensing platform was developed by integrating selective ion recognition with artificial neural network (ANN)-assisted signal modeling. Lithium titanate (Li₂TiO₃) was prepared via a sol-gel route and subsequently converted to H₂TiO₃ through proton exchange in an acidic medium. The resulting H₂TiO₃ was then immobilized onto a pencil graphite electrode to obtain a working electrode for Li⁺ detection. Electrochemical characterization revealed that the working electrode markedly enhanced the redox peak currents of potassium ferrocyanide and lowered the charge-transfer resistance after incubation in a lithium-ion solution. To overcome the difficulty of pinpointing signals when the entire cyclic voltammetry curve varies rather than individual peak intensities, an ANN module was applied to correlate the cyclic voltammetry response with the Li⁺ concentration. The electrode exhibited a concentration-dependent response toward Li⁺ in an ultra-wide dynamic range of 1.0 × 10^-11 to 1.0 × 10^-4 g L^-1. The sensor exhibited excellent stability and reproducibility, with relative standard deviations remaining below 8%, and delivered recoveries between 97.9% and 103.9% when applied to water samples. The integration of ion-sieve recognition with machine learning modeling enabled Li⁺ quantification across an unprecedentedly wide concentration window.