In recent years, quite a few production safety accidents caused by hazardous gas leakage have occurred in the petrochemical industry, causing great potential safety hazards and huge economic losses. Therefore, it is necessary to develop gas sensors with high sensitivity and accurate identification. Semiconductor gas sensor, which has the advantages of high sensitivity, fast response and high integration, is one of the most popular types in the sensing field. However, the semiconductor gas sensor has low specific recognition to reducing gases (such as H₂S, CO, H₂, etc.), and it is difficult to accurately achieve mixed-gas identification with a single sensor. With the development of micro-electromechanical systems (MEMS), the size of semiconductor sensor can be reduced to millimeters with high integration. In order to solve the cross-sensitivity problem, the concept of sensor array has been proposed and widely studied. Through the principal component analysis, the data having the most characteristic information can be selected from among the acquired data while preserving the original data information as much as possible, and they are projected onto the new orthogonal vector by linear transformation. This method can maximize data dispersion and minimize information loss after dimensionality reduction. Therefore, it is an effective way to identify the gas species by combining sensor array.

In this work, we synthesize four types of tungsten trioxide sensing materials with different morphologies or compositions by the hydrothermal method. The sensor array is fabricated by MEMS-based nano sensors. The gas sensitivities to the four single gases (H₂S, CO, H₂, NH₃) and their mixed gas are measured by sensor array, which can acquire four groups of data at the same time. Compared with single sensor, the sensor array has different responses to pure gas and mixed gas, which is the basis for gas identification. Furthermore, we use principal component analysis method to process the response of sensor array. The results show that different gases will occupy different areas in the diagram for pure gas, and show certain directionality according to different concentration distributions. By determining the position of the detected gas, the composition and concentration of the measured gas can be inferred. For mixed gas, the distributions of single gases show the same tendency. And the points of mixed gas always occupy the area between the fans formed by the two gas components, and each region keeps independent. Therefore, this method can also identify the compositions and the concentrations of gas species contained in mixed gas. These results prove that nano sensor array can provide direction and guidance for semiconductor sensor to identify the gas species and concentration.