questions about bandgap
the mos parameters are not well controlled, so the variation in your reference will be large and it will need a lot of trim to get good precision.
for a ieee paper, it's fun to see how many different references you can make to get your name published. when you're shipping 1,000,000 units/month you are not interested in fun, only yield.
as far as the tempco of the resistors goes - haha you'll have to write the loop equations to figure it out!
no just kidding, if you use all the same resistors the % change is proportional and it washes out for the most part - perhaps moving your inflection point a little bit higher or lower, but that's no big deal. you build one, find the "magic" voltage of that cell and tell your test engineer to trim them all to Vmagic.