3

I am building a thermometer for a cryostat which operates at low temperatures (until 4 kelvin).

I would like to know the temperature of some parts in my cryostat during the cooling (so from 300 K to 4 K). The idea is to send a small current (in order not to heat the cryostat too much) in a resistive material which would get a higher resistance for low temperatures, and from the resistance measurement I could get a value for the temperature.

For a maximal precision I would need a material for which the resistivity changes a lot in this temperature range. Which material and which type of resistor (or even diodes, transistors, capacitors, etc.) would you suggest for this application?

For now I tried a thick film Ruthenium oxide resistor because this material seems to be used in some cryogenic temperature sensors, but it only goes from 10 kΩ to 10.4 kΩ when changing its temperature from 300 K to 77 K (liquid nitrogen) which doesn't seem to be enough change in resistivity.

I would be thankful for any advice.

Nicolas Schmid
  • 133
  • 5
  • Although I never routinely went much below -200 C, I had no problem reaching that low with phosphor thermometry. Uses short pulses of light only as often as you need to make a measurement, so you can very much minimize any self-heating at the target site. And, technically, you can just paint the phosphor and use optics to focus down on it from a distance, as well, avoiding entirely the idea of a fiber optic cable. I have no data down towards 4 K, though. Just down to about 50 K, where the phosphors I used worked fine. Sensitivity is quite good -- moreso at low temps. – jonk May 04 '22 at 23:52
  • If you need to use fiber, I've used as small as 10 micron with success. Just FYI. (Though the researchers at the Brain Institute weren't as happy, as they would have preferred a 1 micron fiber since they were trying to heat and measure individual axons and 10 microns was pretty clubby.) – jonk May 04 '22 at 23:58

1 Answers1

9

I take it that you don't want to pay for silicon sensors such as the DT-670. Ge resistance sensors have great resolution at 4K but are not so great at higher temperatures (over a balmy 10K or so) and have impressive price tags.

Ordinary 1N4148 diodes should be okay at least down to 77K. You can use maybe 10uA excitation and measure the forward voltage. They may do weird things like turn into oscillators as you approach liquid He temperatures. Pt1000 sensors will work and will be more 'calibrated' at higher temperatures, but they go kind of flat below around 50K.

With all these sensors to minimize milliwatts or microwatts dumped into your cryostat you may be involving yourself in 'interesting' noise-limited electronics. As usual, specialized suppliers offer off-the-shelf solutions, at a price. If you're only interested in roughly tracking the cooling you may be able to get away with less. You may be able to avoid Kelvin connections, for example, at 10uA, even with typical fine low thermal conductivity conductors.

The disappearance of heat capacity as you approach 0K also has interesting effects.

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • 2
    Thank you! sorry for the delay between your response and the moment I accepted it. I wanted to test what you suggested first. The 1N4148 diode at 10uA works like a charm ! – Nicolas Schmid May 16 '22 at 18:30
  • 1
    Wow. I never would have thought the lowly 1N4148 could work here. What prior work did you with the diode that led to this knowledge? – Marla May 16 '22 at 19:00
  • @Marla I did quite a bit of work with superconducting geophysical instrumentation that runs at 4K, however cool down involves LN2 before burning up expensive liquid helium. To monitor and control the cryostat and cool-down there are quite a few sensors involved. MOSFETs work well at very low temperatures, bipolar parts less so. There are some useful papers out there where folks have tested batches of common parts to see what they actually do at cryogenic temperatures. – Spehro Pefhany May 16 '22 at 19:54