Oscillometric conductivity apparatus I once used had a temperature coefficient of 12KHz per degree C. Using a timer/counter it was therefore possible to easily resolve temperature changes in the region of 1 part in 10exp3. Buildling an electrolyte based Oscillometer is fiddly but once operational it can become a very sensitive thermometer. See:- www.sciencedirect.com/science/article/pii/0956566390800123
1. Is there a diffrential temperature measurements need; 2. is there a dynamic temperature measurement need (temperature derivative) or 3. is it a cryogenic temperature measurements need (i.e.e absolute temperature below 1 mK)?
Case 1, a thermopile sensor could be a good approach having the capability to resolve tens of nanokelvins. Case 2. look at the nano-calorimetric sensors used in fast DSC such as those in www.xensor.nl with scanning rate u to 100000 K/s; Case 3, cryogenic rodium-iron thermometer are standard sensors in the temperature region.
Barrie, The problem (as it seems for me) is the size of the setup size. it is too large for the sample I am using (3x18x.1 mm)
Kebbati, I will take a look (thanks)
Viro, our sample will be at room temperature using a feedback system. We plan to detect the temperature changes via the error signal on the feedback. The changes are on the Micro Kelvin range. The will check the sensors you suggested but I really skeptical about the thrmopile!
Try a micro PTC thermistor in an oscillator circuit. Positive feedback through the thermistor is likely to have enormous thermal sensitivity. Any overall non-linearity is not likely to bother you if you are looking for very small changes over a very small range.