Machine learning-guided analysis of room-temperature H₂ gas sensing using Au- and Pd-decorated ZnO sensors

Developing room-temperature (RT) gas sensors for selective detection of H 2 gas is crucial to minimize power consumption, eliminate the need for integration of a micro-heater, and longer service life. Utilizing UV light is an effective technique to reduce sensing temperature down to RT. Here, we decorated Au and Pd nanoparticles (NPs) on the surface of commercial ZnO NPs and investigated the effects of various UV light wavelength (254, 365, and mixed (254+365 nm)) on the H 2 gas sensing properties of fabricated gas sensors. All gas sensors revealed higher sensing response in the presence of UV light relative to dark condition. In particular, Pd-decorated gas sensor showed enhanced selectivity to H 2 gas in the presence of mixed UV illumination (254 + 365 nm) at RT. Principal component analysis (PCA) and convolutional neural networks (CNNs) were performed on different sensor types (ZnO, Au-decorated ZnO, and Pd-decorated ZnO) and the epoch accuracy for H 2 , CO, NH 3 , and C 7 H 8 gases were 89.58%, 92.92%, 95.00%, 95.83%, 97.92%, 97.71%, 97.08%, and 97.92%, respectively. Thus, CNNs was able to accurately distinguish between different gas species. We believe that present work can open new doors for further investigations related to machine learning analysis of UV light illuminated noble metal decorated gas sensor with possibility of working at RT. • Fabricated low-power gas sensors capable of operating at room temperature, using UV-activated ZnO nanomaterials decorated with Au and Pd. • Mixed UV illumination (254 + 365 nm) provided superior sensing performance compared to single-wavelength sources. • Noble metal nanoparticles improved selectivity through catalytic activity and electronic sensitization. • Principle component analysis enabled near-perfect identification of gas species with high selectivity.