Almost every nontrivial chip has a temperature sensor inside (used for temperature compensation of factory calibrated stuff). I expect RFID chips are no different.
Can't you just get an off the shelf NFC memory chip or something, then figure out what combination of register reads are necessary to read it's internal temperature sensor?
NFC gear is low bandwidth and low power. They don’t need to monitor temperature like you would in a high performance CPU or memory chip.
Rough estimate based on experience +/- 2 C at best.
I'm not a chemist or microbiologist but I imagine they'd need better than that?
Having said that, single point calibration is cheap to do (just calibrate at room temp) and pretty effective.
That level of accuracy would also be useful in candy making, a secret hobby of mine. The cheap thermometers tend to have that margin of error and are pretty terrible at showing just how uneven the heat in a pan can be but I am not sure ic that has more to do with the sensors sensitivity or the refreshing on the cheap LCD display they tend to put on them.
No, 2 degrees C is about the right accuracy you need for a "proper" cup of tea or coffee. You'll lose that much heat in the transfer from kettle to cup. Inaccuracy of 5 degrees C in brewing will be indistinguishable for most people.
Precision also has a placebo effect. Coffee or tea will taste better to you if you put a lot of work into it.
This is very true but, tangentially, this is the opposite when you make drink/food for others. I found that you get the best effect when you spend a lot of effort to cook and perfect the dish but when asked, you have to pretend you basically winged it and it wasn't a hassle
This is commonly known as the IKEA effect.
For example, https://ams.com/wireless-sensor-tags-interfaces
My related experience was positioning a temperature sensor above a Raspberry Pi Zero and the couple watts produced by the Pi was enough to affect the measurement.