 Figure 1. Illustration of unfrozen briny water (blue) shown within a large pore-size ice (cyan)-regolith (grey) mixture.  Through a 3D pore space the flow of unfrozen water (permeability) or ions (electrical conductivity) is dominated via iceice liquid vein networks when above the eutectic temperature and by ice-grain liquid vein networks when below the eutectic temperature.  Introduction: In order to improve the accuracy of cryosuction modeling on Mars, we preformed laboratory experiments on ice-regolith mixtures to find the unfrozen water content, pore size, diffusion coefficient and tortuosity of ice-regolith mixtures. Excess ice (ice content that exceeds the natural regolith porosity) has been discovered at mid- and high-latitudes on Mars via shallow trenching by the Phoenix lander [1], recent impact cratering [2], neutron and gamma ray spectrometry [e.g., 3], and orbital radar [4-5]. The cause of this shallow ice formation remains uncertain. A likely mechanism is cryosuction or regelation, a process by which horizontally segregated ice forms via migration of unfrozen water and produces frost heave [6]. Brines are possible within the martian regolith as salts that can significantly depress the freezing temperature are ubiquitous. Quantitative cryosuction modeling is difficult because models estimating unfrozen water content and permeability of Mars-like brines within iceregolith mixtures as a function of temperature are in early stages of development and lack laboratory verification [6,7]. Thus, the purpose of our laboratory study is to measure the amount of unfrozen water and permeability as a function of temperature, salt type, salt concentration, and grain size.   Unfrozen water: The amount of unfrozen water in an ice-regolith mixture is a function of temperature, salt type, initial salt concentration, and grain-size distribution. Unfrozen water, e.g. thin films, adsorbed water, premelted water [8-9], occurs at a microscopic scale anytime water is frozen within a pore space (Fig. 1). As the water freezes it excludes the majority of the dissolved salts, thus further lower the freezing point of the remaining brine. This forms liquid vein networks (LVNs) at the boundaries between two and three ice grains and between ice and the regolith. The unfrozen water at the ice-ice boundaries freezes at the eutectic temperature. At the ice-regolith boundary, the unfrozen water can remain unfrozen below the eutectic temperature due to interfacial premelting and the GibbsThompson effect [e.g., 10].  Magnetic resonance: Magnetic Resonance (MR) was used to study model ice-regolith samples. Samples were constructed by densely packing monodispersed PMMA particles (0.4 µm or 102 µm) in 30 mM MgCl2 solution and then freezing. The MR signal allows determination of the unfrozen water content φ in the samples as a function of temperature from −20oC upwards. The preliminary data (Fig. 2) compares in order of magnitude with results from models and further data analysis refinement will provide better quantification, MR T2 relaxation distributions provide access to pore size changes. Preliminary results indicate that whilst the unfrozen water pore sizes evolve to larger sizes with increasing temperature in the 102 µm bead sample, they do not seem to change size as significantly in the 0.4 µm bead sample (Fig. 3). We also will present preliminary results from pulsed gradient spin echo (PGSE) MR measurements of the diffusion coefficient of the unfrozen liquid, the surface-area to volume ratio S/V and tortuosity α [11,12]. Permeability ( )2/ VSak αφ=  where the empirical constant a varies between 1.7 and 3 and can then be estimated from MR data.  Dielectric spectroscopy: The electrical properties of ice-regolith mixtures can also be investigated to jointly estimate the permeability. The direct-current (DC; zero-frequency) electrical conductivity is proportional to the tortuosity [13]: mixDC LVN DC σφσα = , where LVN DCσ and mix DCσ  are the conductivity of the LVNs and mixture. Our measurements show that LVNs through ice dominate the DC conductivity in large grain (sands) above the eutectic (Fig. 4), but that unfrozen water on mineral grains dominates at smaller grain sizes and below the eutectic. Furthermore, the tortuosity within smaller-grain soils is significantly higher than that in larger-grain soils (Fig. 4). As the permeability of the unfrozen water is controlled by the tortuMAGNETIC RESONANCE AND DIELECTRIC SPECTROSCOPY DERIVED VALUES OF UNFROZEN WATER CONTENT IN ICE-REGOLITH MIXTURES.  D. E. Stillman1, S. L. Codd2, J. D. Seymour2, P. Lei2, M. Young2, H. G. Sizemore3, R. E. Grimm1, and A. W. Rempel4, 1Dept. of Space Studies, Southwest Research Institute, Boulder, CO (dstillman@boulder.swri.edu), 2Magnetic Resonance Lab, Montana State University, Bozeman, MT, 3Planterary Science Institute, Tucson, AZ, 4Dept. of Earth Sci., University of Oregon, Eugene, OR. Fig. 4. The temperature dependence of the DC conductivity of ice-regolith mixtures as a function of temperature, regolith type, and amount of initial brine. All plots slowly increase with temperature as the amount of unfrozen water and brine conductivity increases. The plots also increase rapidly around the eutectic temperature. The pure ice and 110 µm sand sample increase the most near the eutectic as the low conductivity values below the eutectic come from the conductivity of the solidified hydrated MgCl2 network [14]. On the 1.4 µm regolith samples, the conductivity below eutectic temperature are nearly equivalent as there is a similar amount of unfrozen water on mineralogical grains. However, the jump at the eutectic informs us that the conductivity above the eutectic is dominated by LVN in the ice. At 0.25 µm there is a slope break in the conductivity curve, suggesting a change in conductivity between unfrozen water on mineralogical grain well below the eutectic to LVN through the ice at temperatures above the eutectic. Lastly, the eutectic temperature of −33.7°C appears too low for the sand and pure ice sample, but correct for the 1.4 µm and too high for the 0.25 µm sample.    Figure 2. Previously measured (gray) unfrozen water content vs. temperature for differing soil types from [15]. The color solid lines were used in [6] to calculate frost heave rates on Mars. MR data from our 0.4 (blue stars) and 102.2 (red starts) µm sample mixed with 30 mM of MgCl2.  Fig. 3. Qualitative plot of T2 distributions as a function of temperature for 30 mM of MgCl2. The 0.4 µm sample has a log-normal like distribution indicative of the domain size of unfrozen water on the ice-regolith interface. The 102 µm sample is trimodal possibly representing 2-grain and 3-grain (triple junction) ice-ice interfaces. osity of the LVNs, we may be able to estimate permeability with DS [e.g., 13]. Conclusions: Small grain-size particles retain a significant amount of unfrozen water at low temperature (0.4 µm sample in Fig. 2) due to the interfacial premelting and the Gibbs-Thompson effect. This water is locked in small pores (0.4 µm sample in Fig. 3) that have a large tortuosity (0.25 µm sample in Fig. 4) and hence low permeability. Large grain-size particles have the majority of their water at ice-ice interfaces and thus have a low amount of unfrozen water (102 µm sample in Fig. 2), but large pores (102 µm sample in Fig. 3) and a small tortuosity (110 µm sample in Fig. 4). This new data can be compared to theoretical predictions and previous extrapolation of empirical data to low temperature (Fig. 2). Additionally, it can be directly input into a model of martian frost heave to predict heave rates in regolith mixed with 30 mM MgCl2 at these grain sizes.  References: [1] Mellon et al. (2009) JGR, 114, E00E07. [2] Dundas et al. (2014) JGR, 119, 109-127. [3] Feldman et al. (2004) JGR, 109, E09006. [4] Bramson et al. (2015) GRL, 42, 2015GL064844. [5] Stuurman et al. (2016) GRL, 43, 9484-9491. [6] Sizemore et al. (2015) Icarus, 251, 191-210. [7] Rempel (2012) Vadoze Zone J., 11, vzj2012.0045. [8] Rempel et al. (2004) J. Fluid Mech. 498, 227-244. [9] Möhlmann (2008) Icarus 195, 131-139. [10] Wettlaufer & Worster (2006) Annu. Rev. Fluid Mech. 38, 427-452. [11] Brown et al. (2014) Biotechnology Reports, 3, 60-64.  [12] Brox et al. (2015) J. Glac., 61, 225. [13] Sen (2004) Concepts in Magn. Res., 23A, 1-21. [14] Stillman et al. (2010) J. Phys. Chem. B, 114, 6065-6073. [15] Andersland, O. B., and B. Ladanyi (2004), An Introduction to Frozen Ground Engineering, 363 pp., CRC Press, Boca Raton, FL. 
