What sort of heat loads/thermal mass are you looking at, and what sort of cycle and transition times? The feasibility of a TEC here depends a lot on those.
It's certainly possible to use a TEC for heating as well as cooling, it's just not often done because a heater is simpler. In your case, a TEC may be a good choice if you need to be able to go from 90 to 60C relatively quickly and the sample is relatively small. However, if you can afford a slower cooling cycle, it's going to be much simpler to use a heating element and a cooling fan. If you need fast cycle times and the sample is big...maybe water cooling, or even a conventional refrigerant-based heat pump?
If you still want to try using a TEC, the design case for heating with a TEC is a bit easier because the power lost in the TEC itself contributes to productive heating rather than being a counter term. So essentially you add the power dissipated in the TEC to your heat flow rather than subtracting it, otherwise the design procedure is basically the same as designing for cooling but with some of the signs flipped.
However, there are a few significant challenges based on your temperature ranges:
- 90C puts you into "high temperature" TEC range. 80C is a typical max operating temp.
- If your sample is swinging between 60 and 90C, your heatsink temp may swing even farther. This means lots of thermal stress, and your mechanical design must account for it
- Because of the above, you probably want to look at something like Laird's
PC series, which is designed for thermal cycling and has a max operating temperature of 120C.
- 90C at 25C ambient means your delta T is at least 65C, which means coefficient of performance is going to be poor. This is not such a problem for heating (when Qc drosp to zero, the TEC just becomes a fragile and expensive heater, but still a heater!), but is a big problem for cooling. Even the Laird PC TECs are only spec'ed for dTmax of 67C. You'll need a two-stage system of some sort. Possible two TECs or a TEC plus something else. In your case, since even your minimum temp is above any reasonable ambient, you might be able to use some sort of big thermal mass with a heater attached instead of a typical heatsink. By holding the mass at something like, I dunno, 50C you reduce your TEC's delta T and if you give it enough thermal mass the TEC will be able to dump all of the cooling-cycle heat into it without getting into trouble.
You should start by figuring out the thermal capacity of your sample and test chamber. This, combined with combined with your required transition time will allow you to determine the rate of heat flow you need to apply during transitions. Figure out the rate of heat loss from your test chamber, and the amount of heat (if any) generated by whatever's in your test chamber. Add all of these up, and you now have the total heat flow (Qc) your TEC needs to provide for the heating and cooling transitions. Post those numbers here and you may get more useful advice on the feasibility of your project.