Multicolored CQDs Synthesis via Solvent Effects: Machine Learning to Establish a Prediction Model and Portable Detection of NO2−

Abstract

In this study, carbon quantum dots (CQDs) were synthesized via microwave pyrolysis using o-phenylenediamine (o-PDA) and 2,2-azobisisobutyronitrile (AIBN) as precursors. Based on the solvent effect, multicolored CQDs were prepared. In order to quantitatively analyze the solvent synergistic effect and establish a prediction model for the emission wavelength of CQDs, we established four regression prediction models. The results showed that the Ensemble model was able to associate solvent parameters and predict emission wavelengths most effectively (R² = 0.9007). DIW-CQDs were prepared taking deionized water (DIW) as the dispersion solvent. DIW-CQDs enable sensitive NO₂⁻ detection. The detection limits were 0.38 and 0.03 μM in the ranges of 2.5–25 and 25.5–50 μM, respectively. Furthermore, a rapid detection swab of DIW-CQDs-Fe³⁺@swab was constructed and applied to the analysis of actual food samples. Taking preserved mustard as an example, the NO₂⁻ content was measured to 18.84 mg/kg, which is close to the limit value of 20 mg/kg stipulated by Chinese national food safety standard (GB 2762-2025).