Graphene Quantum Dots with Improved Fluorescence Activity via Machine Learning: Implications for Fluorescence Monitoring

Graphene quantum dots (GQDs) have shown great potential in physical–chemical-biological applications, especially for fluorescence monitoring. However, the low fluorescence activity, safety issues, and unclear synthesis mechanism restrict their application. Here, we investigate the synthesis process of B,N-GQDs by oxidizing 3-aminophenylboronic acid monohydrate and study their core synthesis process parameters (synthesis temperature, H2O2 additional volume, and synthesis time) and corresponding synergic/antagonistic effects in a multidimensional and wide-ranging region. By collecting the optical properties of B,N-GQDs in varied synthesis conditions and utilizing different machine learning models to fit the data, we select the polynomial regression 7 model and the 675/500 peak intensity ratio to evaluate the best synthesis process parameters. Furthermore, through the weight analysis method, we demonstrate that the weight of H2O2 additional volume (0.0260) is obviously higher than those of synthesis temperature (−0.0058) and synthesis time (0.0172), exhibiting that H2O2 additional volume dominates in the synthesis process of B,N-GQDs. Meanwhile, by the greedy random walk method, we could confirm that B,N-GQDs synthesized in the condition of “184-10-2.23” proved to be the best synthesis condition among the different conditions tested in this work. The sample shows a high 675/500 peak intensity ratio (0.285) and photoluminescence quantum yield (PLQY) (0.74%). More importantly, the sample in the laboratory rat reveals a bright fluorescence, indicating that optimized B,N-GQDs are suitable for fluorescence monitoring.