Abstract
The variability among prior data for FLiBe is 11% for the liquid density and 61% for the thermal expansivity. New liquid density and thermal expansivity data are collected, with particular attention to uncertainty quantification. We discuss and quantify bounds for possible sources of variability in the measurements of liquid density: salt composition (<0.6% per 1 mol % BeF2), salt contaminants at 100 s ppm to <1 mol% (2%), Li isotopic composition (2%), sample isothermal conditions (0.2%), dissolved gases (<0.3%), and evolution of bubbles with temperature transients - depending on Ar or He cover gas (0.1 or 0.6% for dilatometry, 1 or 5% for hydrostatic measurements). To aid in quantifying thermal expansivity sensitivity to composition, we review and generalize the ideal molar volume prediction for FLiBe; to improve this model, measurements are needed for the thermal expansivity of BeF2. We collect new data on the density of liquid FLiBe using the hydrostatic method and 170 g of hydrofluorinated FLiBe with less than 0.13 mol % impurities (dominantly Al, K, Na, Mg, Ca), as determined by ICP-MS. We obtain the following: The dominant sources of uncertainty are the bobber volume uncertainty (0.15%), the mass measurement uncertainty (0.2%), and possibly the wetting angle of the salt on the wire (<0.3%). Occasional noise and <0.2% deviation from linearity may be due to the formation of gas bubbles on the bobber surface from the temperature-dependence of gas solubility; repeatable results for heating and cooling runs provide confidence that bubble effects are well managed in this experimental setup. These are the first measurements of the liquid density of FLiBe that report error analysis and that measure the liquid composition before and after density measurements.