Machine-learning-assisted Cr2O3 nanoparticle gas sensor for methamphetamine screening in complex backgrounds

Methamphetamine (MET) is highly addictive and poses serious social hazards, with its abuse presenting severe challenges to global health and safety. This study employed a simple solution combustion method to synthesize high specific surface area (SSA) Cr 2 O 3 nanoparticles, enabling highly selective detection of trace MET vapors in complex alcohol backgrounds. Structural analysis confirms the surface is enriched with Cr 3 + Lewis acid sites and surface-adsorbed oxygen species. This Cr 2 O 3 sensor exhibits optimal sensing performance to MET at 200 °C, demonstrating fast response/recovery, excellent repeatability, moisture resistance, and long-term stability. Pyridine-infrared spectroscopy and electron paramagnetic resonance (EPR) spectroscopy confirm the presence and Lewis acidity of Cr 3+ centers. Density functional theory (DFT) calculations further show that MET preferentially coordinates with surface Cr 3+ sites through its secondary amine group. By integrating principal component analysis (PCA) with a one-dimensional residual convolutional neural network (1D-ResNet), the Cr 2 O 3 sensor exhibited good MET discrimination capability against multiple drugs, volatile organic compounds (VOCs), and environmental interfering gases (such as NO 2 , H 2 , and CO). This study demonstrates the selective detection and identification of MET in complex backgrounds using a Cr 2 O 3 sensor, providing a novel strategy for developing portable smart drug screening sensors suitable for complex field environments.