Tunable diode laser absorption spectroscopy for stable isotope studies of ecosystem–atmosphere CO2 exchange

The stable isotope content of atmospheric CO2 provides information about ecosystem carbon–water relations and biosphere–atmosphere carbon exchange. Virtually every isotope study within these fields has required air sample collection at remote locations followed by isotope analysis at a laboratory. This requirement severely limits sampling frequency and experiment duration. In this paper, we evaluate a tunable diode laser absorption spectrometer (TDL) for measuring the carbon isotope content of CO2 at atmospheric mole fractions (350–700 μmol mol−1) and isotopic abundance (View the MathML source of −6 to −16‰). Using infrared absorption, the TDL system determines the mole fractions of View the MathML source and View the MathML source independently, rather than their ratio as in mass spectrometry (MS). The ability of the instrument to measure isotope ratios (View the MathML source) was tested outdoors in a grassland and compared to standard laboratory-based MS measurements made on field-collected flask samples. The TDL was operated at a sampling flow rate of 230 ml min−1 and a sampling interval of 2 min for two intake heights. There was a consistent offset for View the MathML source of 1.77‰ between the TDL and MS measurements, and the standard deviation of the error (MS−TDL) was 0.35‰ (n=82). Removal of two outliers improved this standard deviation to 0.25‰ (n=80). After removing the offset, 62 out of 82 samples had absolute differences less than 0.3‰. Subsequent laboratory experiments indicated that the TDL/MS offset was caused by pressure broadening, and can be avoided in the future by calibrating the TDL with CO2 mixed with air rather than nitrogen. Based on these results we estimate the precision for View the MathML source to be 0.25‰ for our sampling scheme. A similar comparison with flask-based measurements of CO2 mole fraction (View the MathML source) made with a calibrated infrared gas analyzer indicated a TDL precision of 0.4% (1.6 μmol mol−1 at 400 μmol mol−1). The TDL was used to investigate the vertical and temporal variation in the carbon isotope content of respired CO2 (View the MathML source) from the grassland. Measurements of View the MathML source of CO2 in air were made during four separate nights at 1 and 60 cm height above ground. View the MathML source did not vary with height, but it did vary from one night to the next. Hourly measurements of View the MathML source showed it changed as much as 6.4‰ (−29.1±0.4 to −22.7±0.8‰) in a single night. Temporal changes in View the MathML source during the night have not been reported in prior studies. Such observations could provide a new way to investigate temporal dynamics of the carbon substrates utilized for ecosystem respiration.