Performance improvement of on-chip mid-infrared waveguide methane sensor using wavelet denoising and Savitzky-Golay filtering

Abstract

Due to mode interference and reflection, waveguide gas sensor is prone to generate noise, which affects gas detection performance. In order to improve the performance of waveguide methane (CH4) sensor, wavelet denoising (WD) and Savitzky-Golay (SG) filtering were combined to process the noisy signal. With the fabricated silicon-on-insulator (SOI) ridge waveguide, a waveguide CH4 sensing system was built, and the absorbance signal was processed by denoising schemes. Experimental results show that compared with the traditional WD scheme, the WD-SG scheme has obvious advantages in suppressing low-frequency random noise and interference noise. Using WD-SG scheme, the signal-to-noise ratio (SNR) of the absorbance signal is increased by 4.95 times. According to the Allan deviation analysis, the 1σ limit of detection (LoD) of the sensor is reduced from 2929 parts per million (ppm) to 397 ppm when the WD-SG scheme is adopted. The sensor shows a fast response time of less than 2.5 s. Our work shows that the WD-SG scheme can effectively improve the SNR and reduce the influence of noise on the waveguide sensor system on the basis of real-time monitoring.