I did this analysis several years ago, but was hoping someone else would reply because I can't remember the result precisely.

I believe the noise at the output is the thermal noise of the opamp modulated by a +/-1 square wave with frequency Fchop. This will have "carriers" at 3*Fchop, 5*Fchop, etc, so you have aliasing. However, when you add up and integrate all the noise that is aliased back it turns out to be the same as if you did nothing. At least that is how I remember it. The Enz/Temes paper has an analysis, but I found it easier to do the math myself.

rg