Deep learning-assisted smartphone-based ratio fluorescence for “on–off-on” sensing of Hg2+ and thiram

Abstract

Rapid, accurate, and low-cost detection of heavy metals and pesticide residues are crucial to environmental pollution control, but it is still a challenge. In this work, the fluorescence sensing platform system of the iron-based metal–organic framework (Fe-MIL-88NH2) and gold nanoclusters (Au NCs) based on a smartphone portable device coupled with a deep learning-driven applet of WeChat to measure Hg2+ and thiram were proposed. Meanwhile, Hg2+ quenched the fluorescence mechanism of Au NCs was explored by density functional theory (DFT). Interestingly, thiram can restore the fluorescence intensity of Au NCs. As a result, the ratiometric fluorescence sensor can accurately detect Hg2+ in the “on–off” model and detect thiram in the “off–on” model, which possessed high sensitivity and low detection limits are 7 nM and 0.083 μM, respectively. Meanwhile, the visual changes of fluorescence color from red to purple and blue for determination of Hg2+ and the color returned to red when detecting thiram. Based on the RGB or HSV values reflected in the images, the linear range individually quantified of Hg2+ and thiram in the broad linear range of 0.002–30 μM and 0.083–49.910 μM, respectively, which are equivalent to or better than that attained from fluorescence spectrometer. In addition, the developed sensor in combination with deep learning can accurately predict Hg2+ and thiram concentration levels in actual samples. Besides, our strategy provides a powerful sensing platform in the analysis of water samples and crops and suggests great application potential in environmental monitoring.