Sound Velocity and Compressibility of Liquids at High Pressures from in-situ Ultrasonic Measurements

Liquids with a variety of compositions have been proposed to be present in the Earth's deep mantle. Examples include dehydration induced silicate melts on the top of the mantle transition zone (e.g., Revenaugh and Sipkin, 1994) or lower mantle (e.g., Schmandt et al., 2014), carbonatite melts in the asthenosphere (e.g., Gaillard et al., 2008) or the mantle transition zone (e.g., Rohrbach and Schmidt, 2011), primordial silicate melts (e.g., Labrosse et al., 2007) or metallic melts (Zhang et al., 2016) in the deep mantle, and slab-derived metallic liquids in the lowermost mantle (Liu et al., 2016). The presence of liquids would dramatically modify the seismic wave velocities and elastic properties of the mantle, and thus is one of the possible sources of mantle heterogeneities observed as low seismic velocity anomalies. However, the sound velocity and elastic property of liquids have not been well constrained especially for silicate and carbonatite liquids due to technical challenges in high pressure experiments. In the past a few years, we have made a series of technical developments which enable us to apply high-pressure ultrasonic sound velocity measurements to a wide range of liquid compositions. Combined with synchrotron X-ray imaging and diffraction techniques, we have successfully determined the sound velocities of several silicate, carbonate, and Fe-alloying liquids up to 7 GPa and 2373 K. The velocity data were combined with density data to place tight constraints on the equations of state and compressibilities of these liquids. With these data, we evaluated how various liquids, if present, would affect the seismic velocity and elastic properties of the mantle. These results would help to better interpret the low seismic anomalies in the deep mantle. For silicates, our results show distinct velocity-density systematics between liquids, solids, and glasses, reflecting the differences in compression mechanisms. This means glasses cannot be used as good analogs for liquids when studying compressional properties. The use of glass velocities would result in significant underestimation of melt fractions in the mantle.