|In this revision of “Stratospheric carbon isotope fractionation and tropospheric histories of CFC-11, CFC-12 and CFC-113 isotopologues” by Thomas et al. the authors have clarified many of my previous questions. However, my most serious concerns about the quality of the data did not lead to any improvements. Those concerns addressed the following issues:|
1. The authors used a method for isotope measurement which lacked ground-truthing. Very little information about the performance of this method was provided (linearity test, analytical precision for measurement of samples). A method development showing parameters such as reproducibility, accuracy/ trueness etc has not been provided. In addition, there are no similar methods for 13C/12C ratio analysis by GCMS in the literature even showing that this is feasible.
The authors mentioned in their rebuttal some studies to prove, that 13C/12C ratio analysis by GCMS with a single detector is an established method. However, none of these examples provides evidence for the functionality of the method used for the current study because of two major differences. None of these SIMS and Fourier-transform MS (Orbitrap) used in those papers was connected to a GC. Therefore those methods measured isotope ratios on a constant signal and did not derive them from transient signals (peaks from the GC) which is a fundamental difference and an additional cause of uncertainty for GC-based systems. For instance GC-IRMS (transient signal)is by about an order of magnitude less precise than Dual Inlet IRMS (constant signal). I also would like to add that most of these cited papers are method papers where the performance of the method, other than the current one, was evaluated by using appropriate standard materials (e.g. the cited study by Neubauer et al 2020). This leads to my second major criticism of this study:
2. Basic principles of stable isotope analysis are disregarded. In particular, the authors did not use any appropriate reference material which are directly linked to the isotope-delta zero-point. In addition, realization of a two-point normalization was not carried out in order to correct for any scale contraction effects.
The authors argue that this scale compression/expansion (usually smaller than 10%) is due to cross-contamination between samples, isotope exchange, or blank effects. I would disagree with this theory because scale compression also occurs for pure substances without exchangeable ions and for relatively large signal sizes. It is commonly known that this effect occurs in the mass spec due to slight variations in ionization efficiency, calculation of the H3-factor (for hydrogen) etc. If cross contamination and exchange effects influence the results then the whole method is actually quite useless. It is true that scale compression effects are usually smaller than 10% for IRMS instruments but, as pointed out in my previous review, they are larger for regular GCMS-systems. I mentioned a study by Bernstein et al (2011) who showed that for chlorine isotopes the measured scales of GCMS instruments varied by up to 30 % for chlorine. Given the lower abundance of 13C and the very small amounts of carbon in these samples (lower pmol range) this variation may be much larger. The only “reference material” used in this study is AAL-071170, a gas mixture with a similar composition as air. This mixture is certainly an appropriate standard for concentration measurements of trace compounds in air but it does not fulfill the requirements of an isotopic reference material. Reference materials are, above all, no mixtures of compounds but usually pure substances. In this way they can be cross-checked with bulk and offline methods.
3. The authors use the Rayleigh model to describe degradation in the stratosphere. The Rayleigh model is, however, only appropriate to describe first-order or pseudo-first-order reactions (Mariotti 1981 Doi 10.1007/Bf02374138, or any textbook). The authors argue that “The Rayleigh model is independent of the reaction order. It applies for any process, for which the relationship dc(13C)/ dc(12C) / [c(13C) / c(12C)] = 1 + ε = const. holds”. The problem here is, however, that we deal with several processes and not just one: Degradation by Photolysis and O1D and in addition transport and mixing. This means that the relationship above may be changed by any of these four processes. Hence the reported epsilons only characterize the current data but a different data set may yield completely different epsilons only if mixing patterns change.
Overall, I have to state again that this study still does not implement important principles of (natural abundance) stable isotope analysis. I understand the challenge of obtaining stable isotope ratios from precious samples of limited volume. The performance of a method should, however, be properly evaluated and documented. Without using appropriate reference materials and without defining the scale of the mass spec the already quite large analytical uncertainty may become substantially and unpredictably larger and the data basically useless.