Grain growth of polycrystalline ice doped with soluble impurities

The grain size of polycrystalline ice affects key parameters related to planetary evolution such as the rheological and dielectric properties of Earth's glaciers and ice sheets as well as the ice shells of ice satellites. Although past experiments have studied the grain growth of pure water ice as well as polycrystalline ice doped with air bubbles and insoluble particles, the effect of soluble ions, such as Cl^- and SO₄^2-, which are commonly found in glaciers, on the growth of polycrystalline ice is not clear. To investigate the effect of soluble impurities on the grain growth kinetics of polycrystalline ice, we conducted annealing experiments on polycrystalline ice samples doped with different concentrations of KCl (10^-2, 10^-3, 10^-4, 10^-5 mol/L) and MgSO₄ (10^-2, 10^-5 mol/L), respectively. Ice powders were obtained by spraying a fine mist of solution (ultra-pure water + KCl or MgSO₄) into liquid nitrogen. The powders were then dried and uniaxially pressed into a cylinder at 30 MPa and -30°C and then hydrostatically pressed into a cylindrical ice sample at 100 MPa and -30°C for 15 min. The samples were annealed for a maximum of 320 h at a hydrostatic pressure of 20 MPa (corresponding to about 2 km glacier depth) and different constant temperatures (-5, -10, -15, -20, -25°C). After each experiment, we took microscopic images of the polished sample surface using an optical microscope equipped with a cold stage. Machine learning methods combined with human quality check were utilized on the images to distinguish the grain boundaries. Then a grain size was measured as the average equivalent grain diameter with a geometry factor applied. For KCl, at -5°C (the eutectic point of KCl solution is -10.7°C), lowest-concentration-doped ice (10^-5 mol/L) grew slightly faster than higher-concentration-doped ice (10^-2, 10^-3, 10^-4 mol/L), and both are faster than pure ice; at -10°C and -15°C, there was no significant difference between the growth rates of both doped ice with different concentrations and pure ice; while at -20°C and -25°C (well below the eutectic point of ice and KCl), the growth rates of higher-concentration-doped ice (10^-2, 10^-3 mol/L) were slower than those of pure ice and lower-concentration-doped ice (10^-4, 10^-5 mol/L), which are similar to each other. For MgSO₄, at -5°C (the eutectic point of MgSO₄ solution is -3.6°C), the growth rates of doped ice with different concentrations are not significantly different from that of pure ice; while at -10°C and lower temperatures, the growth rates of doped ice are slower than that of pure ice. We propose that at temperatures well above the eutectic point, grain growth may be largely influenced by partial melts and temperatures well below the eutectic point, soluble impurities impede grain growth. These results help quantifying the contribution of different creep mechanisms, grain-size sensitive and insensitive, under natural conditions and will contribute to future estimation of the rheological strength of glaciers and ice shells.