Comparative performance analysis of unmixed and mixed metal oxide sensors for dual-sensing leveraging machine learning.

Abstract

This paper presents the synthesis of mixed metal oxide (BaTiO3: ZnO) (B: Z) sensors with various molar ratios using a low- temperature hydrothermal method for dual sensing applications (gas and acceleration). The sensor developed with an equal molar ratio of 1B:1Z, showcases superior performance compared to unmixed and alternative mixed metal oxide sensors. This equilibrium in ratios optimally enhances synergistic effects between elements B and Z, resulting in improved sensing properties. Furthermore, it contributes to structural stability, enhancing performance in gas and acceleration sensing. A decreased band gap of 2.82eV and a rapid turn-on voltage of 0.18V were achieved. The acceleration performance of 1B:1Z sensor exhibits a maximum voltage of 2.62 V at a 10 Hz resonant frequency and an output voltage of 2.52 V at 1 g acceleration, achieving an improved sensitivity of 3.889 V/g. In addition, the proposed gas shows a notable sensor response of ~63.45% (CO) and 58.29% (CH4) at 10 ppm with a quick response time of 1.19s (CO) and 8.69s (CH4) and recovery time of 2.09s (CO) and 8.69s (CH4). Challenges in selectivity are addressed using machine learning, employing various classification algorithms. Linear Discriminant Analysis (LDA) achieves superior accuracy in differentiating between CO and CH4, reaching 96.6 % for CO and 74.6 % for CH4 at 10 ppm. Understanding these concentration-dependent trends can guide the optimal use of the sensors in different current applications. .