Neural-Network-Assisted Impedance Analysis for Humidity and Ammonia Detection Using MXene and PtSnO ₂ Sensors with Cross-Interference Suppression

Nonlinearity, baseline drift, humidity interference, and selectivity are among the primary challenges associated with direct current measurements in semiconducting gas sensors today. This study demonstrates that integrating PtSnO₂ and MXene-based sensors enables high-performance gas sensing by effectively mitigating humidity-induced interference. Impedance measurements across multiple frequencies were employed, yielding sensors with tunable gas responses, low noise, extended dynamic range, enhanced baseline stability, and minimal humidity cross-sensitivity. PtSnO₂ was optimized for simultaneous NH₃ and relative humidity (RH) detection, while MXene served as a dedicated RH sensor. A multilayer perceptron was trained on the impedance dataset to deconvolute and accurately predict RH and NH₃ concentrations. The proposed sensor system and analytical framework were benchmarked against commercial DHT22 (humidity) and DFrobot (NH₃) sensors, demonstrating superior performance and sensitivity. This methodology is extendable to other material systems, including metal oxides and transition metal dichalcogenides, for advanced gas-sensing applications. These findings advance gas-sensing technology by improving detection accuracy and robustness in industrial safety, environmental monitoring, and biomedical diagnostics, including early disease detection.