Towards a consensus on the pressure and composition dependence of sound velocity in the liquid Fe–S system

Recent advances in techniques for high-pressure and high-temperature experiments enable us to measure the velocity of sound in liquid Fe alloys. However, reported velocities in liquid Fe–S differ among research groups (e.g., by >10% at 5 GPa), even when similar methods are used (i.e., the ultrasonic pulse–echo overlap method combined with a large volume press). To identify the causes of the discrepancies, we reanalyzed previous data and conducted additional sound velocity measurements for liquid Fe–S at 2–7 GPa, and evaluated the potential error sources. We found that the discrepancy cannot be explained by errors in the sound velocity measurements themselves, but by inaccuracies in determining the temperature, pressure, and chemical composition in each experiment. Of particular note are the significant errors introduced when determining pressures from the unit-cell volume of MgO, which is a temperature-sensitive pressure standard, using inaccurate temperatures. To solve the problem, we additionally used h-BN as a pressure standard, which is less sensitive to temperature. The pressure dependence of the sound velocity became smaller than that of the original data because of the revised pressure values. Our best estimate for the seismic velocity of the Moon's liquid outer core is 4.0 ± 0.1 km/s, given a chemical composition Fe₈₃S₁₇.