Density Functional\nTheory and Machine Learning of\nTransition Metals in Mo<sub>2</sub>C for Gas Sensors

Gas accumulation is the primary cause\nof explosions in\nunderground\nmines, and preventing it requires effective gas detection. To address\nthis, we propose an approach combining machine learning (ML) and density\nfunctional theory (DFT) for designing nanoscale gas sensors. Our study\ndemonstrates that a back-propagation neural network (BPNN) model,\noptimized with suitable hyperparameters, achieves high accuracy with\nan R<sup>2</sup> (coefficient of determination) of 0.92 and a low\nRMSE (root-mean-square error) of 0.24 in predicting the substrate\nmaterial formed by transition metal (TM)-doped Mo<sub>2</sub>C and\nits interaction with key gas molecules (CO, H<sub>2</sub>S, CH<sub>4</sub>, and C<sub>2</sub>H<sub>6</sub>). Based on these interaction\nstrengths, we have analyzed the materials in more depth. Additionally,\nwe find that certain features directly affect the increase or decrease\nof interaction strengths within a specific range, providing insights\nthat contribute to the design of more efficient nanoscale sensors.