High-precision oxygen isotope ( <i>δ</i> ¹⁸ O) measurements of atmospheric dioxygen using optical-feedback cavity-enhanced absorption spectroscopy (OF-CEAS)

Abstract. Atmospheric dioxygen (O2) concentration and isotopic composition are closely linked to the carbon cycle through anthropic carbon dioxide (CO2) emissions and biological processes such as photosynthesis and respiration. The measurement of the isotopic ratio of O2, trapped in ice core bubbles, brings information about past variation in the hydrological cycle at low latitudes, as well as past productivity. Currently, the interpretation of those variations could be drastically improved with a better (i.e., quantitative) knowledge of the oxygen isotopic fractionation that occurs during photosynthesis and respiration processes. This could be achieved, for example, during experiments using closed biological chambers. In order to estimate the isotopic fractionation coefficient with good precision, one of the principal limitations is the need for high-frequency online measurements of isotopic composition of O2, expressed as δ18O of O2 (δ18O(O2)) and O2 concentration. To address this issue, we developed a new instrument, based on the optical-feedback cavity-enhanced absorption spectroscopy (OF-CEAS) technique, enabling high-temporal-resolution and continuous measurements of O2 concentration as well as δ18O(O2), both simultaneously. The minimum Allan deviation occurred between 10 and 20 min, while precision reached 0.002 % for the O2 concentration and 0.06 ‰ for δ18O(O2), which correspond to the optimal integration time and analytical precision before instrumental drift started degrading the measurements. Instrument accuracy was in good agreement with dual-inlet isotope ratio mass spectrometry (IRMS). Measured values were slightly affected by humidity, and we decided to measure δ18O(O2) and O2 concentration after drying the gas. On the other hand, a 1 % increase in O2 concentration increased the δ18O(O2) by 0.53 ‰. To ensure the good quality of O2 concentration and δ18O(O2) measurements we eventually proposed to measure the calibration standard every 20 min.