A g-C3N4-stabilized Au nanozyme-based sensor with AI-powered “recognition–extraction–interpretation–report” for bias-minimized in-field monitoring of mercury ions

Abstract

Mercury ions (Hg²⁺) pose serious threats to human health owing to their toxicity and bioaccumulative properties, requiring timely and accurate in-field monitoring. Although laboratory-based analytical techniques can offer excellent detection performance, they need expensive instrumentation and professional operators, making them unsuitable for on-site self-detection; existing portable optical approaches still rely on manual operation, inevitably introducing underestimated biases arising from operators’ subjectivity and hardly meeting the demand for reliable monitoring. To fill this gap, here we introduced g-C₃N₄ as a favorable support to stabilize agglomeration-prone gold nanoparticles (AuNPs) and constructed an AI-powered “recognition–extraction–interpretation–report” automatic signal-on optical sensor for Hg²⁺ quantification with minimized biases based on target-triggered enhancement of enzyme-like activity. The g-C₃N₄/AuNP nanozyme exhibited synergistic catalytic performance compared with individual g-C₃N₄ or AuNPs, providing a stable signal generation route for Hg²⁺ sensing. Importantly, the nanozyme response was integrated into a smartphone-based portable platform to achieve point-of-need monitoring at environmental sites. More importantly, by incorporating a standardized attachment and a deep-learning-assisted workflow, the developed sensor effectively eliminated subjective and objective biases associated with imaging conditions, manual region selection, operator dependency, and inconsistent signal interpretation. Our sensor was totally free from the need for spectrophotometers and human intervention, enabling user-independent Hg²⁺ in-field quantitative analysis. Our contribution offers a route to smart sensing by fusing advanced nanozyme materials and portable devices with AI algorithms, which can inspire the future development of intelligent sensors for broader analytical applications.