Metal-organic framework-derived porous cerium-doped tricobalt tetraoxide dodecahedrons for efficient detection of...

Highly selective detection of toxic hydrogen sulfide (H₂S) is crucial, as it is a by-product of many industrial processes. Metal oxide semiconductors (MOSs)-based chemiresistive gas sensors have been extensively utilized in gas detection; nevertheless, developing MOSs with innovative nanostructures is essential to boost their gas-sensing capabilities. Herein, we prepared porous dodecahedral cobalt tetraoxide (Co₃O₄) with varying cerium (Ce) doping concentrations using metal-organic framework (MOF) templates for H₂S sensing. The sensor based on 3 at.% Ce-doped Co₃O₄ demonstrated optimum H₂S-sensing performance, characterized by high sensitivity (response = |R_g- R_a| / R_a × 100 % = 151.6 %), a short response time (41 s), and excellent selectivity toward 100 ppm H₂S at 220 °C. These enhancements are attributed to a large specific surface area, high Co³⁺/Co²⁺ ratio, rich surface adsorbed oxygen, and strong H₂S adsorption. Density functional theory (DFT) calculations confirmed that Ce-doped Co₃O₄ exhibited a higher H₂S adsorption energy and a greater charge transfer capacity compared to pristine Co₃O₄. By combining this optimal sensor with the bald eagle search (BES)-optimized random forest (RF) (BES-RF) algorithm, a high prediction accuracy for H₂S concentration was achieved. This study provides a new strategy to improve the practicality of MOSs-based gas sensors for indoor toxic gas detection through machine learning assistance.