Abstract
Rationale: Since their introduction more than a decade ago, isotope ratio infrared spectroscopy (IRIS) systems have rapidly become the standard for oxygen (δ18O) and hydrogen (δ2H) isotope analysis of water samples. An important disadvantage of IRIS systems is the well-documented sample-to-sample memory effect, which requires each sample to be analyzed multiple times before the desired accuracy is reached, lengthening analysis times and driving up the costs of analyses. Methods: We present an adapted set-up and calculation protocol for fully automated analysis of water samples using a Picarro L2140-i cavity ring-down spectroscopy instrument. The adaptation removes memory effects by use of a continuously moisturized nitrogen carrier gas. Water samples of 0.5 μL are measured on top of the water vapor background, after which isotope ratios are calculated by subtraction of the background from the sample peaks. Results: With this new technique, single injections of water samples have internal precisions (1σ) below 0.05‰ for δ18O values and 0.1‰ for δ2H values, regardless of the isotope ratio of the previous sample. Precision is worse, however, when the isotope difference between the sample and background water is too large (i.e., exceeding approximately 9‰ for δ18O values and 70‰ for δ2H values). Isotope ratios show negligible drift across the four weeks within which the experiments were performed. The single-injection 1σ precision for 17O excess (Δ′17O) determined with this method is 60 per meg. Conclusions: Our experiments demonstrate that by removing sample-to-sample memory effects with a moisturized carrier gas, the time for measurement of δ18O and δ2H values using an IRIS system can be reduced markedly without compromising the analytical precision and accuracy. Thorough replication is needed to achieve sufficiently low uncertainties for Δ′17O.
Original language | English |
---|---|
Article number | e9055 |
Number of pages | 12 |
Journal | Rapid Communications in Mass Spectrometry |
Volume | 35 |
Issue number | 8 |
Early online date | 17 Feb 2021 |
DOIs | |
Publication status | Published - 30 Apr 2021 |
Externally published | Yes |