Spectral background removal of MEMS FTIR spectrometer-based gas analyzer

The wide spectral range, compactness, low cost and high measurement speed of the MEMS FTIR spectrometers enable their use in real-time in-line gas analysis. They have the potential of identifying and quantifying several gases simultaneously compared to other infrared technologies such as the NDIR. However, the challenge of real-time spectral background removal from the measured spectrum has to be tackled first. In fact background removal is a common problem to spectroscopic applications relaying on the spectral shape and strength of the absorption lines. In this work, two of the most recent background correction algorithms, namely iterative averaging and morphological weighted penalized least squares, are adapted and applied on the MEMS FTIR spectrometer for gas mixture analysis. These algorithms don't require prior knowledge about the background or the peaks position and don't involve any manual selection of a suitable local minimum value. A 10-cm gas cell that contains known concentrations of SO₂, C₂H₄, N₂O and N₂ was measured using the MEMS FTIR spectrometer. The presence of several spectral absorption lines of these gases in the mid-infrared (MIR) region is considered a challenging case for automatic background correction algorithms. The spectra are measured in the MIR range of 1.6 μm - 4.9 μm with a resolution down to 33 cm^-1. The corrected spectra are compared with spectra measured with a standard bench-top spectrometer and the RMS error and Pearson's correlation coefficient are calculated and good values of 0.8 % and 98 %, respectively, are obtained. Overcoming the spectral background removal paves the way for the use of MEMS FTIR spectrometer in real-time monitoring of multiple gases simultaneously.