IsoAdvector unphysical alpha separation at sharp, high curvature boundary layers

I am using this library on a case that involves argon bubbles rising in liquid gallium. I have an issue where the alpha field "leaks" near the edges of the bubbles, and subsequently forms smaller bubbles roughly the size of a mesh cell. This was way more intense with MULES, however using isoAdvector still doesn't solve it.

There are sharp gradients at the edges of the bubbles, however this separation shouldn't happen in my setup.

Here are some isoFaces that showcase this behaviour.

The isoFaces look weird to me. It looks like a second surface layer is forming at the "leakage" zone.

I have tweaked all of the main solver parameters for alpha and p_rgh fields, and it doesn't seem to help. Lowering the Courant number also doesn't help. My p_rghfinal is 1e-10 and soFaceTol is 1e-6 and surfCellTol is 1e-6. The artefact sizes seem to correlate with cell sizes. I have tested this in v6 and v2012.

Any help would be greatly appreciated.