Review of « Development of the deformational regimes and microstructures in the deep sections and overall layers structures of the Dome Fuji ice core, Antarctica »

by Tomotaka Saruya et al.

02/13/2024

This paper provides analyses of physical properties (grain size and crystallographic orientations) measured along the deepest part of the Dome Fuji ice core, between 2400 and 3035 m depth. Measurements were done using the dielectric tensor method (DTM) on thick sections of ice, the Laue X-ray diffraction method and the automatic ice fabric analyser on thin sections.

Based on these measurements, the authors provide an analysis of the deformation and recrystallization processes likely to take place in the deepest part of the ice core. They also observe and somehow quantify the impact of the shear component on the flow of ice below Dome Fuji.

Global comments :

This paper presents some valuable high resolution data of crystal orientation fabric (COF) in the deep part of the core where deformation heterogeneities are supposed to occur that could disrupt the dating signal and the radar echo sounding in the area.

They also provide some comparisons between three different means of measuring the COF that can be of interest. Nevertheless, I did not clearly understand the necessity to have the three different types of measurement, especially for Laue X-ray diffraction and fabric analyser that are strongly redundant (except that X-ray diffraction provide the a-axes orientations that are known to be isotropic, so yes, it is interesting to double check but it is not used in the analyses). The DTM method provides a higher resolution than classical diffraction or optical measurements nevertheless the method still requires to make sections (thick ones), and, in order to be statistically representative, it must integrate a minimum number of grains, therefore it may be viewed as a moving average of the COF obtained from the other means from thin section?

Overall, the part that analyses and discusses the results should be re-organised and strengthen by some more precise references to previous works. This part appears as a “brainstorming”, with many results and ideas mentioned, but they need to be converged into a “story” that shows what comes out of the results, what it brings to the existing story about ice core analyses (with more precise comparisons with existing work), and/or what new story it tells (although I didn’t see so many new results or observations in this work).

In particular, dynamic recrystallization along ice cores has been studied for a long time, both in terms of basis mechanisms that come into play and in terms of impact on the COF evolution. I think that COF evolution along Dome F ice core should be analysed regarding this existing frame. I provide some references in the detailed comments below but many other exist that the authors can refer to (since most of the references I provide are from my team… as it is more straightforward for me).

Some interpretations are not rigorous enough and should be strengthen by some physical concepts or adapted references. This is particularly the case when are mentioned grain boundary sliding and “microshear”. Grain boundary sliding is a specific mechanism that occurs under specific circumstances that are clearly not encountered here (small grains, high level of strain, accommodating mechanisms at grain boundary, etc.). The only observation of a few grain boundaries with specific shapes is not enough to support grain boundary sliding (and in 2D!). See for instance the works done in the metallurgy community (e.g. Doquet et al. 2016, Mechanics of Material, Linne et al. 2020, Int. J. Plasticity). I think this is not positive for the ice community to keep on mentioning GBS as an important mechanism although there exist so few evidence of it along ice cores. Especially since natural ice is characterized by very large grains, with very efficient accommodation processes that are grain boundary migration and dynamic recrystallization, there is no requirement of GBS to explain ice deformation along ice cores (even if it might take place in the specific conditions of the Goldsby et al. 1997 experiments on very fine-grained ice).

Many repetitions exist all along section 5. Maybe the authors should be more focused on the main message they want to provide, in order to avoid dispersion in the interpretations and some lack of rigor as just mentioned, but also to enable more focused and deeper comparisons with existing works.

In general, to make reading easier, please avoid acronyms in section titles, abstract and conclusion.

Specific comments along the text:

l. 24-25 (abstract): dynamic recrystallization is characterized by nucleation of new grains and grain boundary migration. The main basis processes have been described for a long time (see e.g. Poirier 1985, Humphreys and Haterly 2004) and please refer to the existing frame. Therefore this sentence is not clear “dynamic recrystallization related to this emergence”.

l. 43: what is meant by “dynamical analysis”?

l. 79: “remains” to be done?

Part 3.1:

What about measurement of fabrics that depart from cluster-shape with the DTM technique? We have observed some, for instance along the NEEM ice core (Montagnat et al. 2014, The Cryosphere), likely resulting from a tension component of the stress field.

Part 3.2 and 3.3:

Why is it necessary to perform all these different measurements? How complementary are they? Do you have enough grains in the X-ray thin sections?

Why don’t you measure the grain size from the automatic fabric analyser?

Part 4:

Figure 4 is of very low quality. Please improve it in the final version.

Content of appendix C is very often referred to and is highly necessary to understand the data. Maybe it should be put back in the main text.

Part 4.4: the quality of the writing should be improved in this part.

What does “an axis orthogonal to the shear plane and the IACC will deviate” means?

Part 4.5: to my point of view, the microstructure features analyses are based on too few grains, and on 2D observations of 3D mechanisms, and must therefore be taken with care unless they are statistically significant (for instance if the shape of grains have been characterized over a large enough number of grains and sections, etc.).

Moreover, extraction of the ice core leads to stress relaxations, especially around bubbles or clathrates in the deeper part of the core, that induce a lot of dislocations substructures (like subgrains). Care must be therefore taken when analyzing subgrain-scale features from these samples.

When recrystallization is dominated by grain boundary migration (such as in the bottom of the GRIP core, de La Chapelle et al. 1998 (J. Geophys. Res) or along the NEEM core, Montagnat et al. 2014, for instance), the grain boundaries are highly serrated, and it can be observed on a large number of grains, therefore with a statistical value that overcome the 2D sectioning effect, or the impact of stress relaxation.

Part 5:

As mentioned in the global statements, the discussion would need being re-organized, with a clearer focus on the selected results and observations that bring something new to the “story”.

Overall, the authors should bring on numerous existing observations and analyses, especially for dynamic recrystallization and its impact on COF and microstructures, and its interactions with impurity contents. Most analyses and interpretations remain vague and not precise enough, some are not far from being wrong. See below for details.

- Part 5.1.1: a highly concentrated clustered COF is related either to deformation by compression, with no dynamic recrystallization (since DRX opens the texture close to 45° from the compression direction, see e.g. work of Jacka’s team in the 80’s and 90’s or more recently Montagnat et al. 2009 PICR2, Montagnat et al. 2015, Frontiers in Earth Sciences), or to simple shear. In Montagnat et al. 2012 (Talos Dome ice core, EPSL), we have shown that compression alone could not explain a strong COF of the type the authors are measuring here, but that some simple shear is necessary. It seems to be the case here too.

Simple shear could be assisted by dynamic recrystallization (DRX) but it is less straightforward to distinguish since DRX strengthens simple shear textures (see Bouchez and Duval 1982, in Textures and Microstructures, or Journaux et al. 2019, The Cryosphere).

L. 287: dynamic recrystallization is not a deformation process! This is an accommodation process that reduces the local strain and stress heterogeneities and facilitated further deformation by dislocation creep.

Diffusional creep is very unlikely to occur along deep ice cores, first because the required densities of vacancies or interstitials are not met and also because it would not lead to the observed textures and DRX mechanisms.

L. 290-294: the relationship between climatic period (and impurity content) and microstructure has been studied a long time ago, and should be referred to here. See for instance Duval and Lorius 1980 (EPSL), Weiss et al. 2002 (Ann. Glaciol, at Dome Concordia), Durand et al. 2006 (JGR). The role of GB pinning on GB migration and therefore on the DRX mechanisms has been widely documented and even modeled in some of these articles.

The sentence “We interpret the wide scatter as indicative of grain nucleation and subsequent growth” is very vague. How is this interpretation related to already known mechanisms of DRX?

About the scatter of the permittivity signal, I could also interpret it as being related to the small measurement step regarding the size of the grains, that is the characteristic scale of the COF pattern. I would expect a measuring step lower that this grain size to create this variability as each measurement only account for a low number of grains. Then, the effect of the variability is reduced by the moving averaging.

- Part 5.1.2:

l. 315-316. It is very well know that dislocation creep and DRX mechanisms are temperature-controled mechanisms! No need to “speculate” it. Indeed, both dislocation velocity and grain boundary migration are temperature-dependent through Arhenius type of law…

I don’t see why molecular diffusion is mentioned here?

l. 321-322. What is meant by “more effective as a condition to trigger active emergence of recrystallization”? A strong clustered texture will be hard to compress, and therefore DRX is necessary as an accommodation process (or fracturing at higher strain rates) to enable further deformation, but such a clustered texture is very easy to shear (when the cluster orientation is normal to the shear plane, which is the case here), and therefore deformation can occur without or with a limited amount of DRX. This sentence is too vague, and it strengthens the confusion mentioned before about the role of simple shear in the origin of the highly clustered texture.

The texture is also likely more clustered in the glacial ice because DRX is expected to be less active due to GB pinning.

l. 330-333: there is nothing new in this statement… please refer to existing work about dynamic recrystallization along ice core that I have mentioned before.

- Part 5.2.1:

L. 350-351: “The recrystallization processes require thermally activated molecular diffusion”… This sentence is very vague and likely incorrect. Recrystallization occurs mostly by dislocation mobility, either within the crystals (formation of subgrains, dislocation pile-ups, etc.) or at GB by enabling grain boundary migration (under the driving force of dislocation-based strain energy). I don’t see why molecular diffusion is mentioned here.

L. 356: The shear strain component is only mentioned here. It should be evoked much before, see previous comments about part 5.1.1.

- Part 5.2.2:

l. 369-370: how is the mechanism mentioned here possible? The different layers are linked together and cannot rotate or shear independently? Dislocation creep accommodated by DRX tends to align the c-axes with the vertical to the shear plane (see references mentioned before), and the observed texture is, to my point of view, related to the fact that the shear plane is not horizontal?

Fig.10: for me, fig 10d represents bending and not rigid rotation? Can a layer experiment rigid rotation independently of the mechanical constraints that surround it? (“system’s rigid-body rotation”, l. 372).

l. 374-375: “Thus, the inconsistency of the angles in the circumstantial evidence for dominance of the simple shear…”. This sentence appears useless. Why are the angles “inconsistent”? Do we need that for inferring simple shear at these depths?

Fig 10a,b,c: the sketches are not precise enough since it depends on the dominant stress, if it is compression or shear. In particular, 10c holds for compression but not for shear. If shear is parallel to the layers, c-axes will be perpendicular to them (with or without DRX).

- Part 5.2.3:

There are many repetitions from previous parts, analyses/interpretations that are mixed with discussion… It goes in favor of the necessity to re-organise the discussion with a clearer story to tell, especially focusing on either the confirmation of existing works (that must be referred to) and/or on new interpretations (although they appear to be few in this work).

l. 435: about the impact of dust, work by Weiss et al., Durand et al., mentioned before suggest another explanation with dust impeding grain boundary motion during recrystallization, and therefore enhancing the impact of deformation versus recrystallization on the final COF. The final COF will be either less favorable for compression creep or more favorable for horizontal shear… Therefore not so straight forward!

l. 441-445: please refer to existing work! Dust particles (and/or insoluble particles?) are mainly located at GBs. They may enhance dislocation production (it has not been clearly proved) BUT they pin GBs. In the works mentioned before (Weiss et al. 2002 for instance), it is shown that the pinning force lead to a critical grain size very close to the one measured in glacial ice.

Durand, Gillet-Chaulet et al. 2007 (Climate of the Past) have modeled the impact of changes in ice viscosity with climatic transitions on ice flow. This work could be mentioned.

I think diffusion creep should not be mentioned here, as there are too few evidence of its likeliness to occur… Unless the authors can provide some NEW statements about it.

L. 481: please refer to Weiss et al. 2002 and Durand et al. 2006 regarding Dome C.

l. 481-482: “we can hypothesize that the physical phenomena…” this sentence is too vague. Please be more precise. Which mechanism does what, and what does your observation bring “to the story” already written by previous works?

- Part 5.2.4: again, what is the effect of the grain size, and therefore the number of grains in the measured area on the standard deviation?

Part 5.3:

- Part 5.3.1:

To my point of view, this part reveals a misunderstanding of the deformation mechanisms of ice (maybe this is due to the way it is written?). As mentioned in my general statement, GBS should not be evoked here with so few proof of its existence. This is a mechanism that is very unlikely to occur along ice cores (large grains, low strain rate, low stresses, efficient accommodation mechanisms…) and that can not be proven by a few 2D microstructure observations.

Microshear is mentioned, but I personally don’t know what it means! Any dislocation glide along a basal plane created a microshear…

How can GBS contribute to grain size reduction? How can it create new boundaries?? Please refer to the work of Ashby 1973 (Acta Metallurgica) for instance to better understand GBS. GBS is not a recrystallization mechanism! This is a deformation mechanism that is likely to take place in very fine-grained materials leading to superplasticity, in very specific conditions.

To correctly interpret fig. 12, one would need the measurement of misorientations in order to be able to provide a clear distinction between subgrains, grains, and interpret them in terms of mechanisms at play. For instance, figure 12a could also present an example of some GB pinning by a subgrain as observed in Montagnat et al. 2015. In figures 12f and g, it is not straightforward to interpret the observations as grain boundary migration. The shape of the GB is not sufficient to my point of view. In figure 12i I do not see quadruple junctions but the effect of a 2D sectioning of a larger grain. Just to say that these figures can easily be “over interpreted” in line with our scientific prejudice.

- Part 5.3.2:

l.570: the statement about the orientation of nucleus can not be proven with the presented results. Observations, with direct measurements, of nucleus orientations show that they are mainly oriented close to their parent grains (see e.g. Chauve et al. 2017, Phil Trans Roy Soc A).

l. 584: Figure 12 can not be used to prove the existence of GBM, please be more cautious in your statement. Some more proof is required.

Similarly, the only observations in figure 12 c1 and d1 of square-looking grains are not enough to demonstrate GBS since, first, they could also be a 2D sectioning effect, and second, more statistics is required to demonstrate GBS. Please be more cautious or remove.

“The orientations of the c-axis in nucleated grains…” you do not have any direct observation of nucleus orientations, and some existing observations (e.g. Chauve et al. 2017, but maybe not only) are not in favor of such a statement. Please be more cautious or remove.

l. 585: “The presence of numerous sGBs implies high stored strain energy”. This statement is wrong. Stored energy is related to the total dislocation density, during deformation, while sGBs only represent the population of geometrically necessary dislocations that remains after unloading! There exist not link between the density of geometrically necessary dislocations and the total dislocation density, and Lopez-Sanchez et al. 2023 (EPSL) have even shown that you can have a lot of strain in some given grains and NO sGBs remaining after unloading. Please remove.

For the remaining discussion related to nucleation and dynamic recrystallization, please refer and mention existing work in order not to “propose” explanations that have already been given, but rather to “confirm” that the processes you observe are in phase with previous observations and analyses of DRX along ice cores.

l. 596: please site Weiss et al. 2002 and/or Durand et al. 2006 closely related to what you observed instead of Cuffey and Patterson that is a general book.

l. 603: Please remove the following statement “Given the high concentration of dust particles, it is likely that the area experienced GBS via microshear” since, first, you have not enough proof for such an assertion, secondly, how can dust particle concentration be linked to GBS, where does that come from, and finally because the meaning of “GBS via microshear” is very uncertain!

Part 5.4:

The zoning must be put in relation with what has already been observed and analyzed in other ice cores. For instance, Dome C, Talos Dome, NEEM ice cores. Along Talos Dome we have shown that some dynamic recrystallization was necessary to reproduce the observed COF in complement to dislocation creep (this latter would lead to too strong COF) (Montagnat et al. 2012, EPSL). Along the NEEM ice core we have observed the impact of shear on the COF in exact connection with the change in the climatic origin of the ice, for instance.

Part 5.5:

Information about COF and the resulting layering of ice viscosity along ice cores can not be straightforward related to the dating of ice core or what is called “radioglaciology” (if I understood well?). See e.g. Durand et al. 2007 (Climate of the Past) or Buiron et al. 2011 about Talos Dome age-scale (Climate of the Past). Maybe some recent radar measurements studies could be mentioned here too.

Conclusion:

l. 689: I think it is not correct to mention here the “elongated shape” of crystals that has not been statistically characterized, and for which you can not quantify the relative occurrence. You could mention it in the text as an example of some local observations but not in the conclusion since it will give too much weight to this weak observation.

l. 690: same comment about this mention of GBS and microshear that has not been shown to occur significantly in your observation, with strong enough statistics and scientific evidence, to worth being mentioned as an important mechanism unless with the aim to orient the reader interpretation.

l. 705: temperature also play a role in the observed mechanisms, not only strain. For instance, along NEEM we observed dynamic recrystallization high along the core that was not observed for the same amount of strain along Dome C or Talos Dome.

Conclusive statement:

I recommend that major revisions are made prior to publication of this article. In particular, attention must be paid to providing sufficiently well-founded scientific evidence before concluding on the importance of a mechanism.

There is also a lack of reference to the work that have already been done for many years about ice core texture and microstructure measurements and their interpretation. This should be corrected.

At the end, the novelty of the study stands only on the fact that these deep core measurements have never been published. Care should be taken in the discussion in order to focus on what is potentially new or relevant for the community.

See attached pdf file