Enhancement of NO2 sensing performance in Ni-doped ZnO gas sensors through machine learning approaches

This study presents the development of Ni-doped ZnO nanostructures integrated with machine learning algorithms for enhanced NO2 gas sensing. The materials were synthesized via a modified wet chemical approach, with varying Ni concentrations forming Zn1-xNixO. XRD analysis confirmed successful Ni incorporation with crystallite size reduction from 12.6 nm to 11.0 nm. The Zn0.90Ni0.10O composition demonstrated optimal sensing performance, achieving a sensitivity factor of 11.57 towards 10 ppm NO2 at 200°C, with response and recovery times of 166s and 59s respectively. Implementation of machine learning algorithms, particularly XGBoost regression, enabled precise gas concentration prediction (RMSE = 0.22 ppm) and reduced false positive rates by 87%. The ML-enhanced system achieved real-time monitoring capabilities with sub-100ms latency and maintained 92% of initial response after 100 measurement cycles. This integrated approach combining materials engineering with intelligent data processing demonstrates significant potential for practical environmental monitoring applications.