Machine-learning-assisted n-GaN-Au/PANI gas sensor array for intelligent and ultra-accurate ammonia recognition

Gallium nitride (GaN) based gas sensors have attracted considerable attention owing to their excellent physical and chemical properties, and especially mature and controllable metal–organic chemical vapor deposition (MOCVD) synthesis process. However, the practical applications of GaN-based gas sensors are constrained by the high detection limit and poor selectivity in the case of multiple types of gases. Herein, a novel sandwich film structure of n-GaN-Au/PANI (PANI: polyaniline) is fabricated based on controllable magnetron sputtering technique and in-situ oxidative polymerization. The n-GaN-Au/PANI sensor with etching time of 50 min (n-GaN-Au/PANI-ET50) demonstrates ultra-low detection limit of 10 ppb, good moisture resistance, high response value for 200 ppm NH3 (195 %), and fast response/recovery (28 s/49 s) at room temperature. This can be attributed to the synergistic enhancement by the ternary materials as well as the large specific surface area of PANI caused by the incorporation of Au nanoparticles. In addition, machine learning is used in combination with a sensor array for the high-precision identification of mixed gas components, achieving an accuracy of 94 %. Overall, this study demonstrates the immense potential of n-GaN-Au/PANI sensors in the field of NH3 detection.