There's this "atomic" explanation of the freezing-point phenomena on Wikipedia that leaves me really intrigued.
Consider the problem in which the solvent freezes to a very nearly pure crystal, regardless of the presence of the solute. This typically occurs simply because the solute molecules do not fit well in the crystal, i.e. substituting a solute for a solvent molecule in the crystal has high enthalpy. In this case, for low solute concentrations, the freezing point depression depends solely on the concentration of solute particles, not on their individual properties. The freezing point depression thus is called a colligative property.
The explanation for the freezing point depression is then simply that as solvent molecules leave the liquid and join the solid, they leave behind a smaller volume of liquid in which the solute particles can roam. The resulting reduced entropy of the solute particles thus is independent of their properties. This approximation ceases to hold when the concentration becomes large enough for solute-solute interactions to become important. In that regime, the freezing point depression depends on particular properties of the solute other than its concentration. (source: http://en.wikipedia.org/wiki/Freezing_point_depression)
I'd like to understand better how the enthalpy and entropy "behave" on this process and how they explain it.