Viscosity of Liquid Ferric Sulfate Solutions and Application to the Formation of Gullies on Mars

Introduction: Almost all flow models for martian gullies to date have assumed liquids that exhibit viscosities on the order of 1 cP (10 Pa.s) such as pure water [1,2] and liquid carbon dioxide [3]. Alternatively, salt solutions have been suggested based on thermodynamic and kinetic arguments as a way to stabilize liquid water in present day martian conditions [4,5]. We have recently proposed that concentrated aqueous solutions of ferric sulfate Fe2(SO4)3 might be a liquid under martian conditions because it has amongst the lowest eutectic temperature (205 K) of common ionic materials [6]. This paper reports on experimental measurements of the visocity of ferric sulfate solutions as a function of temperature and concentration. We then use this information in a numerical flow model of a standard gully as a bridge between the composition and temperature of the brine and the fluid dynamics, to identify geomorphological characteristics that might be observed on Mars [7,8]. Specifically, in this study we estimate the flow velocity and the drag forces on boulders within the gullies for these highly viscous fluids. Methods: Viscosity of various concentrations of Fe2(SO4)3 brines was measured using EZ/Zahn (ASTM) Dip Viscosity Cups (#1 – 5). These cups were filled with ferric sulfate brines and the time for the solution to drain through a calibrated hole in the bottom was measured. Empirical equations produced by the manufacturer were used to determine the viscosity from the drain time. Results: The viscosity of ionic solutions increases with higher concentrations and lower temperatures (Fig. 1). Values range from 7.0 10 Pa.s for 38.8 wt% at 285 K to 4.6 Pa.s for 58.2 wt% at 260 K. The total error on the final dynamic viscosity was estimated to be approximately 10%. To determine the dependency of viscosity on concentration and temperature, we used an extended version of the usual Arrhenius-type law. Fitting the dependency of our data, we obtain the following semiempirical equation: