Fabrication of NbC/GaN Nanofilm Sensor via Photolithography and its Investigation as a Sensor for Trimethylamine Mixed Gas Detection Using Dual-Feature Extraction and Deep Learning

In this study, we successfully synthesized NbC nanofilms on the GaN surface, and a more uniform size and thinner thickness of NbC were optimized by further fabricating circular hole arrays on GaN epitaxial wafers using photolithography and etching techniques. This sensor exhibits an ultralow detection limit of 200 ppb for TMA gas at room temperature, a high response value (84.14%) to 200 ppm of TMA, and low resistance fluctuation for uniform NbC nanofilms. The excellent performance after combination of the two can be attributed to the synergistic effects of p-n heterojunctions and Schottky barriers. Furthermore, the algorithm innovatively adopts dual-feature extraction via KPCA combined with polynomial feature engineering to systematically investigate the relationship within sensor array data. By integrating machine learning algorithms with the sensor array, the system achieves the precise identification of target components in gas mixtures, reaching 98% accuracy. Ultimately, this study demonstrates the significant application potential of gas sensors in the next generation robotic electronic nose.