This preprint manuscript presents new findings on the spatial distribution of sulfur (S), chlorine (Cl), and sodium (Na) in polar ice using high-resolution laser ablation inductively coupled plasma mass spectrometry (LA-ICP-TOFMS). The study focuses on samples from the EPICA Dome C (EDC) core in Antarctica and the EGRIP core in Greenland, aiming to understand impurity localization and its impact on post-depositional diffusion. The results show that S, Cl, and Na are predominantly localized along grain boundaries, with limited evidence for accumulation at triple junctions or within grain interiors, except in dust-rich ice. These findings suggest that diffusion may occur primarily along grain boundaries rather than through interconnected veins, challenging assumptions that post-depositional mobility varies significantly with climatic periods. The study highlights the need to integrate impurity mapping with modeling efforts to better constrain diffusion

processes that affect paleoclimate records preserved in deep ice cores.

While the fundamental science discussed in the preprint is a significant contribution to the field and is clearly within the scope of The Cryosphere, the manuscript text requires substantial reworking before it is suitable for publication. In particular, the framework established in the introduction and methods sections lacks clarity and cohesion, making the central aims and motivations of the study difficult to follow. This undermines the accessibility of the results and their implications, even for readers familiar with ice core and/or LA-ICP-MS science. Once the manuscript has undergone major revision, it will be a lovely addition to The Cryosphere.

Please see the attached document for full comments and corrections.

The paper by Bohleber et al. presents the first attempt at two-dimensional mapping of S and Cl in EDC and EGRIP deep ice cores using LA-ICP-MS. Although both elements are relatively abundant in these cores, due to analytical challenges with ICP-MS, they have not been previously targeted by LA-ICP-MS studies. This paper shows that signals of S and Cl can be detected even in samples that do not contain elevated concentrations of volcanic S and Cl. Since both S and Cl are important for understanding mechanisms and extents of impurity diffusion in ice—crucial for interpreting the chemical signatures of “Oldest Ice”—the new methods developed in this study have great potential for future ice core research. The paper also demonstrates that not only the most abundant isotope of S, but also less abundant isotopes can be detected. This would contribute to investigating sources of S. Therefore, I strongly support the publication of the fascinating new results presented in this paper.

However, I have a question or concern regarding the conclusion about the localization of S. The authors find a high level of localization of S at grain boundaries, with only minor occurrences within grain interiors even in glacial ice. Previous studies using Raman spectroscopy (Ohno et al., 2006; Sakurai et al., 2011; Stoll et al., 2021) reported numerous S-containing salts or minerals such as Na₂SO₄ and CaSO₄. I wonder why the samples used in this study did not contain many of these S-containing particles in grain interiors. Although there may be differences between Dome Fuji (Ohno et al., 2006; Sakurai et al., 2011) and EDC samples (this study), Stoll et al. (2021) used EGRIP samples, as in this study. I wonder if this discrepancy is due to a sample-to-sample (or layer-to-layer) difference, or if it might be an artifact of the analytical methods used here. I would like to see some explanations about the difference between the findings of this study and those of previous studies.

The paper states that the new impurity maps support a view of diffusive transport not only through ice veins but also along grain boundaries, yet do not show clear differences between samples from the Holocene and the last glacial period in the EDC core. Based on these findings, the authors seem to reject the hypothesis by Rhode et al. (2024) that the diffusion mechanism changed between the Holocene and the last glacial period due to changes in localization. However, if the amount of S in grain interiors changed between the Holocene and the last glacial period but was not detected by the methods used in this study or not observed in the specific samples analyzed by chance, the authors cannot completely deny the hypothesis. Therefore, I would like the authors to confirm that the very low amount of S detected in grain interiors in glacial EDC ice is robust and not an artifact caused by analytical issues.

Additionally, I have some minor editorial comments, which are listed in the “Detailed comments” below.

Detailed comments:

1. Section 2.1 – This section needs more details on how this study succeeded in detecting S and Cl using LA-ICP-TOFMS. I recommend moving a major part of Lines 223–251 to Section 2.1, as this part is more appropriate in the Methods section rather than the Discussion.
2. Line 145, pre-ablation – Please explain more about pre-ablation. If this involves sublimation of the sample surface, I have a concern that grain boundaries might be preferentially sublimated compared to grain interiors, which could lead to concentrated impurities at grain boundaries. Please confirm that this is not the case.
3. Table 2 – Please add concentrations of S (or SO₄²⁻ would be fine), Cl, and Na for the samples used in this study. If the data from exactly the same depths are unavailable, concentrations from similar depths would be helpful to better understand the differences among samples.
4. Figures 1, 2, 4, and S2 – It is difficult to see grain boundaries and air bubbles in the optical mosaic images. Please add optical mosaic images with marked grain boundaries and air bubbles for each figure. Figure S1 also needs an optical mosaic image to show grain boundaries.
5. Lines 166–167, “In EGRIP 2286…… (Fig. 2)” – It is hard to identify the isolated pixels within the grain interior showing high S values. Please mark these pixels clearly.
6. Lines 167–169, “The EGRIP cloudy……… (40 µm EDC)” – It is hard to see the intensity difference between EGRIP and EDC. To confirm this, the authors need to show the intensity data.
7. Lines 169–170, “Both maps show………… a few mm” – Please show the intensity data for the grain boundaries.
8. Lines 173–174, “At 40 µm and ……… at triple junctions” – I’m not convinced by this sentence. Some triple junctions appear to show strong intensities. The authors need to provide more quantitative discussion here, as I wrote in comments 6 and 7. It is indeed hard to assess intensity variability just from visual images.
9. Figures 3 and 4 – Please add the spot size in the figure captions. Although it is mentioned in the text, including it in the captions would make it easier to follow.
10. Figures 1, 3, 4, 5, 6, S1, S2, S3 – Please use larger fonts in the diagrams.
11. Line 226, “Fig. 1–7” – This should be “Fig. 1–6.” There is no Fig. 7.
12. Line 250, “It is likely that also ³⁷Cl is mass-shifted and detected at ³⁹K instead.” – If this is the case, please explain why the authors can still distinguish between K and Cl signals. Why can we be sure that the signals in the maps represent Cl?
13. Lines 286–288 – Why do the intensities at grain boundaries depend on scan direction? Why does this dependence only appear for the 1 µm spot size? Unless a clear explanation is given, I’m concerned about the authors’ discussion on differences in signal intensities.
14. Lines 313–322 – Considering previous studies, I would expect higher concentrations of Na in grain interiors. However, this is hard to see in Figures 2, 3, or 4. Please mark Na in grain interiors more clearly.
15. Lines 332–334, “At least for sections… bulk S concentration.” – I agree that it is difficult to see systematic differences between the data presented here. However, if the ratio of S at grain boundaries to grain interiors changes (but not detected by the methods used in this study), it could affect the apparent diffusivity. To draw a conclusion, I think more quantitative analysis is needed.
16. Lines 346–349 – From Figures 3 and 4, it is difficult to see partially interrupted impurity populations at grain boundaries and air bubbles in glacial maps. Please show these features more clearly.
17. Lines 357–365 – If S/Na and Cl/Na at grain boundaries are similar for Holocene and glacial samples, I don’t think this necessarily argues against a relative difference. Changes in the amounts of these elements in grain interiors—possibly not observed by the methods used—could result in apparently different diffusivity.
18. Lines 384–385, “but do not show any …. EDC ice core.” – If impurity localization at grain boundaries and veins shows no clear differences, differences in grain size would change the ratio of grain boundaries and veins in a unit volume, potentially affecting diffusivity. However, smaller grain sizes in glacial ice would give larger ratios of grain boundaries and veins, leading to faster diffusion, correct? I think grain size data are also important for considering diffusion mechanisms.