why making a surface "super" smooth increases the coefficient of friction? I read that:
If you take a rough surface and make it smooth, the coefficient of friction decreases. But if you make it super smooth, then the coefficient of friction increases. How come?
 A: If I think about friction geometrically, i.e. ignore for now chemical interactions, (although they are very important) you get an argument about the surface area in contact. If you take two flat smooth surfaces, they will contact along a plane. If you roughen up, one or more of the surfaces, contact will only take place at high spots, i.e. there will be a lot of gaps where there is no contact, so the area in contact is much lower. Conversely the forces/stresses on the much smaller area that is in contact will be higher.
A: What do you mean by super smooth?
I remember a note (again) by Feynmann that said if you made your test surface so clean that there is absolutely no dirt or impurities on it, then the super clean surface would actually attach to anything sliding on it, making the apparent friction coefficient higher. 
Is this what you mean by super smooth? 
In that case, take a copper plate. It is ideally made of copper atoms. Its surface will be dirty, filled with other molecules. 
If you now imagine that you have the tool to clean it so well that just the copper atoms are on the surface and nothing else, you will actually get a very reactive - in the chemical sense - surface. "Naked" atoms will bind to anything that passes by, and if you try to make something slide over it, they will make bounds and stick very well. 
A: There is chemical reason behind it. Interlocking is the reason for the friction between rough surfaces. In case of super smooth surfaces, interlocking can't take place.
Chemically, when two super smooth surfaces comes in contact with each other, their chemical structures disturbs.
Then their outer surface shells get merged into one another so they oppose relative motion, hence friction increases.
