Hachgenei N., Vaury V., Nord G., Spadini L., Duwig Céline. (2022). Faster and more precise isotopic water analysis of discrete samples by predicting the repetitions' asymptote instead of averaging last values. MethodsX, 9, 101656 [11 p.].
Titre du document
Faster and more precise isotopic water analysis of discrete samples by predicting the repetitions' asymptote instead of averaging last values
Hachgenei N., Vaury V., Nord G., Spadini L., Duwig Céline
Source
MethodsX, 2022,
9, 101656 [11 p.]
Water stable isotope analysis using Cavity Ring-Down Spectroscopy (CRDS) has a strong between-sample memory effect. The classic approach to correct this memory effect is to inject the sample at least 6 times and ignore the first two to three injections. The average of the remaining injections is then used as measured value. This is in many cases insufficient to completely compensate the memory effect. We propose a simple approach to correct this memory effect by predicting the asymptote of consecutive repeated injections instead of averaging over them. The asymptote is predicted by fitting a y = a/x + b relation to the sample repetitions and keeping b as measured value. This allows to save analysis time by doing less injections while gaining precision. We provide a Python program applying this method and describe the steps necessary to implement this method in any other programming language. We also show validation data comparing this method to the classical method of averaging over the last couple of injections. The validation suggests a gain in time of a factor two while gaining in precision at the same time. The method does not have any specific requirements for the order of analysis and can therefore also be applied to an existing set of analyzes in retrospect. We fit a simple y = a/x + b relation to the sample repetitions of Picarro L2130-i isotopic water analyzer, in order to keep the asymptote (b) as measured value instead of using the average over the last couple of measurements. This allows a higher precision in the measured value with less repetitions of the injection saving precious time during analysis. We provide a sample code using Python, but generally this method is easy to implement in any automated data treatment protocol.
Plan de classement
Sciences fondamentales / Techniques d'analyse et de recherche [020]
;
Hydrologie [062]
