Flexible Sensor Array Based on Polythiophene/Cu-MOF-Derived CuO and Polyaniline/MXene Composites for Room-Temperature Detection of Hydrogen Sulfide and Ammonia

Abstract

Real-time detection of hydrogen sulfide (H2S) and ammonia (NH3) is important for occupational safety in humid, mixed-gas environments, yet conventional chemiresistive sensors remain limited by cross-sensitivity, humidity dependence, and poor portability at room temperature. Here, we report a flexible dual-channel sensor array coupled with a gated recurrent unit-convolutional neural network (GRU-CNN). A polythiophene/Cu-MOF-derived CuO composite (PTh/CuO) targets H2S through sulfur-affinitive CuO surface sites, whereas a polyaniline/Ti3C2Tx MXene composite (PANI/MXene) targets NH3 through PANI deprotonation and interfacial charge modulation. At 25 ± 1 °C, the optimized PTh/CuO and PANI/MXene channels exhibited responses of 87.5% to 100 ppm H2S and 189% to 100 ppm NH3, respectively, together with linear responses over 1−10 ppm (R2 = 0.997 and 0.998) and theoretical limits of detection of 73.3 and 40.4 ppb. Distinct responses were retained at 300 ppb H2S and 200 ppb NH3, with response/recovery times of 15/22 and 25/43 s, respectively. A GRU-CNN model trained using dynamic four-channel signals supported gas classification and concentration estimation across 0−10 ppm and 30%−80% relative humidity within the tested dataset; the maximum concentration-prediction mean squared errors were 0.020 for H2S and 0.015 for NH3. The channels retained 98.4% and 98.1% of their initial responses under 30° bending. Integration with a wireless wristband enabled remote signal acquisition and a wearable monitoring demonstration. This work combines chemically differentiated sensing channels, humidity-inclusive data analysis, and flexible wireless readout within a single room-temperature platform.