Multivariate Analysis of Photoacoustic Spectra for the Detection of Short-Chained Hydrocarbon Isotopologues

Abstract

We report, to our knowledge, the first optical detection scheme for short-chained hydrocarbon isotopologues. The sensor system is based on photoacoustic spectroscopy (PAS). Two continuous wave, thermoelectrically cooled, distributed feedback interband cascade lasers (DFB-ICLs) with emission wavelengths around 3.33 and 3.38 μm, respectively, served as light sources. The investigations comprised the main stable carbon isotopologues of methane (12CH4, 13CH4), ethane (12CH3-12CH3, 13CH3-12CH3, 13CH3-13CH3), and propane (12CH3-12CH2-12CH3, 13CH3-12CH2-12CH3). They were selected because of their importance for numerous applications from climate and planetary research to natural gas exploration. Multiple measurements of single components in nitrogen and synthetic mixtures were conducted at room temperature and atmospheric pressure. Depending on the investigated hydrocarbon isotopologue, detection limits ranging from 0.043 ppmv to 3.4 ppmv were achieved. For a selective concentration determination, multivariate analysis (MVA) was applied. Partial least-squares regression (PLSR) was used to calculate concentrations from the PA spectra. The implementation of MVA has shown that the PA setup in principle works reliably and that the selective concentration determination of short-chained hydrocarbon isotopologues is possible.

PAS-based isotopologic analysis of highly concentrated methane

“…Considering the complexity of the collected spectra a sensor calibration in the form of a quantitative multivariate approach is implemented to validate the methane isotopologue concentrations and to test the suitability of the method. In particular, Partial Least Squares Regression (PLSR) proved to be applicable for determining the concentrations of mixture components from photoacoustic spectra (Loh and Wolff, 2020;Saalberg and Wolff, 2018). Figure 9 shows the true and the predicted methane isotopologue concentrations, after a leave-one-out crossvalidation based on PLSR was applied to all nine investigated mixtures.…”

Assessment of vibrational-translational relaxation dynamics of methane isotopologues in a wet-nitrogen matrix through QEPAS

“…QEPAS and PAS were demonstrated as leading techniques for CH 4 detection in wet and dry nitrogen-based matrixes [27] , [28] , [29] , [30] , in samples containing both nitrogen and oxygen [31] or in air [32] , [33] . The possibility of performing multi-gas detection employing a single sensor was also exploited: methane was detected together with several infrared absorbers such as ammonia, oxides of nitrogen, water vapor, hydrogen sulfide for several applications like breath analysis, environmental monitoring, gas leaks detection in natural gas plants [34] , [35] , [36] , [37] , production monitoring during petroleum exploration, both at trace level and high concentration [38] , [39] , [40] , [41] , [42] , [43] , [44] .…”

Multivariate spectral analysis in quartz-enhanced photoacoustic spectroscopy

“…In this work, a review of different approaches to QEPAS sensing of complex gas mixtures employing PLSR as multivariate tool is reported. This regression method was selected based on its proved capability to deal with spectroscopic data [37][38][39] . The main sources of interference are investigated, and three different case studies are presented.…”