Exploring the dynamics of inward core solidification using analogue tank experiments

Given their small sizes and low central pressures, the cores of most asteroids are expected to have started crystallising at the core mantle boundary (CMB) instead of at their centre, as is the case for the Earth. This so-called top-down crystallisation is thermally unstable but compositionally stable, making the conditions for dynamo generation more difficult to achieve. Nevertheless, modern observations of Ganymede show an active magnetic field, where it has been suggested that solidification occurs away from the CMB as an iron snow. This model proposes that iron crystals grow in a snow zone and subsequently sink into the interior and melt, releasing dense fluid that drives convection and a magnetic field. However, whether this process could have occurred in asteroid cores is uncertain due to the significantly smaller adiabatic temperature difference between the CMB and the centre of their cores. This weak temperature gradient may also prevent crystallisation away from the CMB. Therefore, the power for a compositional dynamo may result from an increase in convective velocities caused by the formation of dense crystals at the CMB or turbulence caused by the settling of the crystals themselves.

To investigate these possibilities, we employ analogue tank experiments to explore the possible mechanisms driving convection during inward asteroid core crystallisation. An ammonium chloride solution is cooled from above with a layer of buoyant propanol separating the solution from the cold plate to prevent the growth of crystals on this boundary. Instead, the crystals form below the buoyant layer in a ‘snow zone’. We vary the temperature difference across this buoyant layer to investigate the different regimes that may exist. At each driving temperature difference, we measure the velocity fields of any fluid flow within the ammonium chloride solution using particle imaging velocimetry. This enables us to compare the convective velocities with and without crystallisation as well as develop scaling laws to apply the results of these experiments to models of core thermal evolution.

We find that the mean convective speeds increase by over an order of magnitude when the fluid is crystallising. This increase in speed is driven by an increase in the bulk density of the fluid in the snow zone due to the presence of a small crystal fraction. While the motion of crystals themselves do not induce any turbulence in the fluid due to their small size, they act to locally increase the density of the fluid, causing dense, crystal-rich plumes to emanate from the snow zone, which drive faster convective speeds throughout the fluid. This result provides a new mechanism for dynamo generation in inwardly crystallising cores, especially if remelting of falling iron crystals is delayed until deep within the core’s interior, as has recently been proposed for Mars, or if there is a nucleation barrier that causes significant undercooling before the onset of crystallisation. We also measure the temperature and composition as a function of depth within the tank, from which we may assess whether thermal equilibrium can be assumed when modelling snow zones in cores.