Dear Polleke;
IMHO: The A-A, B-B, A-B, B-A model is wrong, because it is pure static, but we need a dynamic model to describe solubility and find then the major players of this “game”.
If we declare the crystal as A-A, and the Solvent as B-B:
In my model the driving force is only each particular concentration difference which attracts/forces the molecules in both directions: into the crystal, and into the solution/solvent!
The measure would be both Diffusion Coefficients: fdA in B–B(...), and fdB in A-A(...).
Therein fdx() depends very strong at the polarity difference of A and B, and maybe at some other substance specific properties.
As limiting parameter we must use also the saturated concentrations of each in each.
Maybe we need as a contrary force the gap between both saturated concentrations too.
I believe, that only if we use such a model as above we are able to describe, what will going on during the solution task, and what solubility “finally” means.
I have drawn some pictures, but I haven’t finished the whole model with all equations yet.
But that's maybe only my opinion!
Good Luck!
ARGOS++