the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Dec 2025
Saltwater exposure accelerates ice grain growth and may increase fracture vulnerability
Abstract. Natural ices often fail at stresses much lower than those measured in laboratory settings, complicating our understanding of glacial failure and icy moon crustal fracture. This may be because strength models, which depend on the size of individual ice grains, do not account for the saltwater commonly found in terrestrial and planetary ices. We conducted grain growth experiments, finding that saltwater always modifies grain growth compared to pure ice, and that increasing volume of saltwater introduces a pinning effect limiting this growth. Ice grain size therefore depends directly on liquid fraction, controlled by salinity and temperature. Modeled effects of grain growth on tensile strength following saltwater infiltration find that low-salinity water can reduce ice strength by up to 46 % within 24 hours, narrowing gaps between observations and experiments.
- Preprint
(2078 KB) - Metadata XML
-
Supplement
(1900 KB) - BibTeX
- EndNote
Status: open (until 26 Jan 2026)
- RC1: 'Comment on egusphere-2025-6040', Anonymous Referee #1, 28 Dec 2025 reply
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 103 | 41 | 9 | 153 | 19 | 6 | 5 |
- HTML: 103
- PDF: 41
- XML: 9
- Total: 153
- Supplement: 19
- BibTeX: 6
- EndNote: 5
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
This paper is a welcome attempt to examine the effect of a liquid phase on grain growth in ice, using a carefully designed set of rigorous experiments in the H2O-NaCl system. While the assessment of grain size seems to be well done, the quantification and critical analysis of the grain-scale brine distribution is weak. The most obvious weakness is that the authors do not appear to be aware of the abundant literature on quantification of dihedral angles in natural systems: as McCarthy et al. (2019) point out, there’s been an enormous amount of work done on the controls of interfacial energies on melt interconnectivity in the olivine-basalt system, but none of this is mentioned, despite the fundamental understandings it has resulted in, applicable to all texturally-equilibrated liquid-bearing systems. Perhaps most significantly, a large body of work has been done on melt-bearing natural systems undergoing grain growth, in which it can be shown that the grain-scale distribution of the melt phase departs from that expected for an ideal isotropic system. Several studies have found that the melt distribution changes as the grain size increases, attributed to the non-negligible anisotropy of grain boundary and interfacial energies: critically, the proportion of grain boundaries fully wetted by melt (i.e. with a dihedral angle of zero) increases as rocks become more coarse-grained, with significant implications for rheology (Mu & Faul, 2016). Walte et al. (2003) find similar disequilibrium and transient melt films in growing aggregates of norcamphor, attributable to disappearance of small grains as the average grain size increases.
Any treatment using grain growth models in a liquid-bearing system needs to take into account the variety of pore geometries and locations, instead of assuming, as the authors appear to do, the brine distribution is purely one of interconnected channels on three-grain junctions (ie the ideal distribution for an isotropic system not undergoing any grain growth). While this study is a welcome foray into this complex problem in the context of ice rheology, a more detailed treatment is required before the results of any calculations are realistic.
Specific comments
Line 68 – do you mean the concentration of salt in the liquid, or the volumetric proportion of the system that is liquid?
Line 108 – what is the mechanism by which the sample was flooded with brine? Was the ice entirely sintered together to form a 100% solid block, or was it formed of a porous framework with air in the pores? The reason this is important is that spontaneous infiltration of a previously entirely solid non-porous polycrystal will occur if the solid-solid-liquid dihedral angle is <60˚, driven by the reduction in interfacial energy (e.g. Stevenson, 1986)
Line 147 – the cited reference goes into a lot of detail, which is entirely missing here. How many individual measurements were made for each composition (there is no information on this, even in Figure 4, despite the importance for working out the uncertainties on the true value of the dihedral angle)? Which of the two methods of McCarthy et al. (2019) did you use? What is the error on individual measurements? Did you assess the extent to which the true 3D dihedral angle might vary due to anisotropy of the interfacial and grain boundary energies based on the spread of angles you measured in the 2D sections (see Jurewicz & Jurewicz (1986))? It is clear from figure S3 that some planar ice-fluid interfaces are present in the pores at three-grain junctions, denoting anisotropy – this should be discussed. It should also be pointed out that the connectivity of the melt phase is critically dependent on the extent of anisotropy of interfacial and grain boundary energies, which leads to a range of true 3D dihedral angles and the stabilization of some grain boundary melt/liquid films (I can see something that looks like a film in Figure S3c, for example, suggestive that for those pairs of grains their relative crystallographic orientation results in a grain boundary energy that stabilizes a dihedral angle of zero). In pretty much all systems, there is some anisotropy so there will be a range of porosities over which connectivity is established as the median dihedral angle is reduced below 60˚. This is covered by Didier Laporte in various publications, and should be addressed here as the presence of stable melt films will have significant effects on ice strength (which you discuss later in the manuscript). It should also be mentioned that connectivity departs from the ideal model even in fully isotropic systems if there is grain growth, due to the transient formation of melt films as small grains get consumed by the surrounding growing grains (Walte et al., 2003).
Line 205 – this section on dihedral angles needs quite a bit of work. Firstly, you need to state that you are presenting the populations of observed angles in 2D sections, not true 3D angles. Then you need to provide the median (not the mean!) value of this population of 2D angles, which gives you the value of the true 3D angle (Riegger & Van Vlack, 1960), using Stickels & Hucke (1964) to determine the uncertainties on that value. Then you need to assess how close your populations actually are to those expected for random cuts through a sample with a single value of true 3D angle in order to assess how much anisotropy there is in the system (using Jurewizc & Jurewizc, 1986). What you refer to as melt veins along grain boundaries (line 208) are either sections cut through channels along three-grain junctions or sections cut through melt films on two-grain junctions. If they are indeed melt films on two-grain junctions, then the dihedral angle is zero (no angles of zero are shown in Figures 4c – 4f)
Line 215 – No, this is exactly the opposite. The Xu et al. (2023) study involves an entirely solid system in which, yes, grain growth does lead to an increase in the proportion of low energy grain boundaries, as they move at a slower rate than high energy boundaries. BUT the melt-solid-solid dihedral angle increases as the energy of the grain boundary decreases (see your equation 3). For the melt-solid-solid dihedral angle to decrease, you need the average energy of the grain boundary to increase.
Here, a relevant study is that of Mu & Faul (2016), who looked at the olivine-basalt system and found an increase in melt films (i.e. dihedral angles of zero) with increasing grain growth (see also the earlier work they refer to in their admirably comprehensive introductory and background sections) – they put this down to an increase of the proportion of high-angle boundaries. A major result of their study is that as the grain size increases, a greater proportion of the melt will occupy films on two-grain boundaries instead of forming channels on three-grain boundaries, leading to significant changes in the rheology of the system. It seems also to be the case in your system, in that the mean of the population of dihedral angles observed in 2D sections decreases with time as the grain size increases (shown in Figure 4). It doesn’t appear that you took into account a change in the proportion of films to channels when calculating your pinning pressure, but assumed an ideal grain-scale melt distribution (though I could find no details of how you determined the (single valued) radius of the melt pocket for your equation 4. Neither could I find a definition of the term f in equation 4).
Figure 4 – the usual way of reporting dihedral angles from a 2D section is to use the median, not the mean. This is closest to the value of a single true 3D angle for a system in which the angles are being measured in a randomly oriented 2D section (Harker & Parker, 1945; Riegger & Van Vlack, 1960). You also need to explain how you derived the uncertainties on the median values. The usual way of doing it is the method of Stickels & Hucke (1964). Given the observations of the changing melt geometry in systems undergoing grain growth (Mu & Faul, 2016), it would also be good to see some sort of analysis of the grain-scale distribution of the brine.
Figure 5a – there are well-established patterns of dihedral angle variation with liquid composition in the metallurgical literature, such as Ikeuye & Smith (1949), Passerone et al. (1977) and Shimizu & Takei (2005), and these should be referenced here to support your extrapolated curve. Replace the term “all grain boundaries are fully lubricated” with something like “melt films are present on all grain boundaries”.
Line 271 – this is only a valid calculation if you use the true 3D value of the dihedral angle, which is closely approximated by the median of the population of apparent angles in a randomly oriented 2D section, not the mean! And of course, you should also make some estimate of the range of interfacial energies, given an analysis of the extent to which the true 3D dihedral angle varies as a consequence of anisotropy.
Line 331 – “water infiltrates through grain boundaries” needs rectifying. Are you talking about the infiltration driven by the reduction of interfacial energy in a system with low dihedral angle (Stevenson, 1986)? In which case, it should be “water infiltrates along three-grain junctions in the ice, driven by the reduction of interfacial energy as the equilibrium ice-brine dihedral angle is < 60˚”. You should also address the possibility of sufficient anisotropy to stabilize melt films on two-grain boundaries (ie the dihedral angle is at, or very close to, zero), which will have a major effect on rheology – the extent of anisotropy can be constrained from your 2D data sets, and also from the observations of melt films in your samples.
Technical corrections
Firstly, you use the words brine and saltwater interchangeably in the manuscript. I am not familiar with this literature so don’t know what the accepted usage is, but would expect brine to be the correct scientific term.
Line 72 – Cooper & Kohlstedt is not in the reference list.
Line 99 – here, and in many other places, the punctuation of references is incorrect. This should read Cole (1979), and not (Cole, 1979).
Line 150 - The reference for the onset of permeability at 60˚ is Smith (1948), not Bulau et al. (1979).
Line 180 – a quick plateau? Do you mean a plateau reached after only a short time period?
Line 182 – once grain size begins to plateau
Line 192 – delete the first use of the word “than” to make this sentence make sense.
Line 210 – you need to make it clear that you are reporting values of apparent 2D angles, not the true 3D value of the angle.
Line 235 – do you mean the interfacial energy?
Line 252 – what is the relevance of this sentence? Are you saying that the control of dihedral angle on melt interconnectivity means that there are variations of physical properties with dihedral angle? I would delete the reference to crystallization history as the dihedral angle does not control melt interconnectivity in virtually all solidifying magmatic systems other than perhaps in the deep mantle.
Line 274 – per unit micron? Do you mean per unit area?
Line 373 – you should reference the abundant literature on the relationships between liquid composition (and temperature) and dihedral angle already known for simple eutectic systems in the metallurgy literature (Ikeuye & Smith, 1949; Passerone et al., 1977; Shimizu & Takei, 2005 among many others), to set your result in context.
References
Harker, D. & Parker, E. R. (1945). Grain shape and grain growth. Transactions of the American Society of Metals 34, 156–195.
Ikeuye, K. K. & Smith, C. S. (1949). Some studies of interface energies in some aluminium and copper alloys. Metals Transactions 185, 762–768.
Jurewicz, S. R., & Jurewicz, A. J. (1986). Distribution of apparent angles on random sections with emphasis on dihedral angle measurements. Journal of Geophysical Research: Solid Earth, 91(B9), 9277-9282.
Laporte, D., & Provost, A. (2000). Equilibrium geometry of a fluid phase in a polycrystalline aggregate with anisotropic surface energies: dry grain boundaries. Journal of Geophysical Research: Solid Earth, 105(B11), 25937-25953.
Laporte, D., & Watson, E. B. (1995). Experimental and theoretical constraints on melt distribution in crustal sources: the effect of crystalline anisotropy on melt interconnectivity. Chemical Geology, 124(3-4), 161-184.
Mu, S. & Faul, U.H. (2016) Grain boundary wetness of partially motlen dunite. Contributions to Mineralogy and Petrology, 171:40.
Passerone, A., Eustathopoulos, N. & Desre, P. (1977). Interfacial ten- sions in Zn, Zn-Sn and Zn-Sn-Pb systems. Journal of the Less- Common Metals 52, 37–49. https://doi.org/10.1016/0022- 5088(77 )90233- 8.
Riegger, O. K. & Van Vlack, L. H. W. (1960). Dihedral angle measurement. Transactions of the Metallurgical Society of the AIME 218, 933–935.
Shimizu, I. & Takei, Y. (2005). Thermodynamics of interfacial energy in binary metallic systems: influence of adsorption on dihe- dral angles. Acta Materialia 53, 811–821. https://doi.org/10.1016/ j.actamat.2004.10.033.
Smith, C. S. (1948). Grains, phases and interfaces: an interpretation of microstructure. Transactions of the Metallurgical Society of the AIME 175, 15–51.
Stevenson DJ (1986) on the role of surface tension in the migration of melts and fluids. Geophys Res Lett 13:1149–1152
Stickels, C.A.; Hucke, E.E. Measurement of Dihedral Angles. Trans. Metall. Soc. AIME 1964, 230, 795.
Walte ND, Bons PD, Pashier CW, Koehn D (2003) Disequilibrium melt distribution during static recrystallization. Geology 31:1009–1012