Atomic-Dispersed and Iron-Doped Carbon Dots by Efficient One-Pot Synthesis for High Peroxidase-like Catalytic Activity

Carbon-based nanozymes are expected to reach the catalytic performance of natural enzymes, but their multiparameter synthesis resists rational design. We combined a deep neural network trained on 59 formulations with a differential evolution search algorithm to predict a synthetic route outside the training space, and the measured mass-normalized efficiency matched the prediction. The optimized VS-1 showed a particle size of 2.2 nm, with Fe atoms dispersed and coordinated to approximately 7–8 light atoms (N/O). The optimized VS-1 exhibits a Michaelis constant (Km) of 0.149 mM and a maximum reaction velocity of 2.67 × 10–6 M s–1, demonstrating a substrate affinity approximately three times higher than that of natural horseradish peroxidase. Furthermore, mechanistic investigations based primarily on electron paramagnetic resonance spectroscopy, supported by radical-scavenger trends, identify the superoxide anion radical as the dominant reactive species for substrate oxidation. This work provides a highly active carbon-based catalytic carbon dot for sensing.