Selectivity Enhancement of ZnO Toward Reducing Gases Using Characteristic Response Features Based on Surface Kinetics below 250°C

Abstract

Biomarkers in human breath are proven to be associated with the metabolic and pathological conditions of personal health. For instance, acetone can be an indicator of diabetes and carbon monoxide is a biomarker of asthma. Therefore, monitoring these biomarkers in human breath can be used for disease diagnosis. Moreover, due to its non-invasive and convenient diagnosis attribute, breath analysis is gaining its popularity. Metal oxide semiconductor (MOS) gas sensors are extensively studied because of their rapid response and high sensitivity. In addition, MOS gas sensors may become a reliable and inexpensive tool for breath analysis because fabrication of MOS gas sensors is simple and fully compatible with existing IC fabrication processes. However, although MOS gas sensors exhibit a sensitive response toward various gases, conventional MOS gas sensors usually show poor selectivity toward target gases. To enhance the selectivity of MOS gas sensors toward target gases, sensor arrays are generally used to monitor input gases under a steady sensing condition and the electrical response of the sensor signals is used as the input variable for gas classification algorithms, such as linear discriminant analysis (LDA) and principle component analysis (PCA). In this work, we studied the sensing response property of ZnO toward various reducing gases, and characteristic response features were examined for selectivity improvement. The ZnO sensor was prepared by sputter-depositing a 20 nm-thick ZnO thin film on the Al electrodes, which were thermal evaporated on a thermal oxide substrate and patterned by a stainless steel (SS) hard mask. During the gas sensing test, the ZnO sensor was exposed to a dry air gas mixture of acetone, ethanol, carbon monoxide, ammonia and hydrogen of various concentrations at 250 oC. The ZnO gas sensor demonstrates a characteristic electrical response toward reducing gases. The sensor shows a prompt electrical conductance rise upon the target gases exposure followed by a gradual drop; the conductance will then slowly level off. The dynamic characteristic sensing response is ascribed to be associated with various surface processes taking place on the sensor during gas sensing. According to our previous works, PdO demonstrates a similar sensing response characteristic but in an opposite manner because it is a p-type semiconductor while ZnO is an n-type one. The PdO sensor exposed to a dry air gas mixture of VOCs at 250oC shows a valley shaped response feature. Upon the gas exposure, the PdO sensor shows promptly drop in conductance, followed by a gradual rise and then by a conductance saturation. The first prompt conductance drop can be ascribed to adsorbed and lattice oxygen removal, releasing negative charge to PdO. The latter conductance rises is due to negative charge transfer to adspecies from Pd nanoclusters formed as a result of PdO reduction. The nanoclusters can be finally reoxidized leading to the conductance saturation. We proposed herein similar surface processes on the ZnO sensor for the characteristic response features observed in this work. The characteristic gas sensing features can be utilized as chemical fingerprints for gas classification. By using the slope of the conductance rise, which suggests the effect of oxygen removal and the slope of the conductance drop, which suggests the effect of the Zn nanocluster formation and charge donation to adspecies from Zn nanocluster as variables for principal component analysis, different gases can be satisfactorily classified into separated clusters. In summary, characteristic electrical response features are observed for the ZnO sensor depending on target gases. The response features are a result of surface processes simultaneously occurring on the ZnO sensor as a result of the exposure to reducing gases. The distinct response feature can be utilized for characteristic feature extraction in pattern recognition algorithms to improve the selectivity of the sensor.