Development and application of UV-visible and mid-IR differential absorption spectroscopy techniques for pollutant trace gas monitoring

Spatial representativeness is an important quality criterion in trace gas monitoring, especially if measurements are intended for regulatory and model validation purposes. Open-path absorption spectroscopy techniques meet the representativeness requirement by providing concentrations averaged over atmospheric paths ranging from some hundred meters to some kilometers. This research concerns the characterization and application of a UV-visible differential optical absorption spectroscopy (DOAS) system, and the development and demonstration of a trace gas detection technique based on tunable, mid-infrared (mid-IR) quantum-cascade lasers (QCL). The spectral accuracy, stability and resolution of the DOAS spectrometer, and the efficiency of its PMT1-based light detection system were characterized and tested in laboratory and field conditions. Additional laboratory experiments included exploratory test of a photodiode array (PDA) as multichannel analyzer, and spectral analysis of the arc of high-pressure Xe lamps. PDA measurements of NO2 in the 360-530 nm range show an effective lower detectable fractional absorbance of 8·10-4 (1-sigma) for a 3-s integration time, overpassing the detection capabilities of the scanner/PMT system. DOAS measurements of SO2, NO2, and O3 were extensively validated through laboratory calibration and intercalibration, and field intercomparison with conventional point monitors and EPA2-certified DOAS systems, showing good agreement among the different instruments. A nitrous acid (HONO) generator was developed and used for characterizing the DOAS detection capability. Laboratory measurements indicate a lower detection limit (LDL) of ~0.2 ppm·m HONO (2-sigma) for a 30-min integration time. Concerning formaldehyde (HCHO), a field intercomparison with HPLC3-analyzed DNPH4 samples shows a non-aleatory, high degree of correlation (r2 = 97%) between the two techniques, but unexplained, significant deviation from the 1:1 line (which is nevertheless within the 85% confidence interval of the correlation curve). A mathematical method was developed for the estimation of detection limits through the analysis of sequential lamp spectra. LDL estimations with this method are in very good agreement with operational detection limits. Other data processing and concentration retrieval algorithms were developed and used for analyzing laboratory and field measurements (12 campaigns). The analysis and interpretation of observations in two of these field studies are reported in detail. The first study concerns the dynamics and photochemical production regime of HCHO in the Grenoble region, France. DOAS measurements at a suburban location of Grenoble allowed estimating the sources of formaldehyde, and served at validating the results of a mesoscale photochemical grid model. Measurement-derived emission and photochemical production ratios are in good agreement with literature values, the emission inventory, and model calculations. A large fraction (>80%