At the time when Earth was hot liquid why the Si-Al layer covered only partially the Si-Mg layer?

At the time when Earth was hot liquid, why did the Si-Mg layer covered the entire surface, while a less denser Si-Al layer covered only partially the Si-Mg layer? I am very curious as to how a macroscopic liquid layer can have intrinsic cohesion between its particles, such that Si-Al was forced to spread only partially along the Earth’s surface. I thought this question would be more geological than physical, but given that it involves the liquid state and cohesion forces, I decided to post it here.Probably related to yolk over the egg white or water drop on the plate…