low temperature related electronics effects/devices?

[coppercone2]

So there is cryo stuff (superconductor), say <= liquid nitrogen.

But are there any effects or devices that rely on a pretty low temperature (say between -30 and -125C) ?

The only one I am familiar with is thermal imagers, this is well known.

But are there any other devices that can be made to do special things at the 'moderately cold' temperature range? Any other kinds of detectors, sensors, transducers  emitters, etc?

Like a circuit that gets 'possible' at that temperature region, not just stuff that has extremely marginal benefits from running colder (maybe noise stuff) that is basically not worth the effort ever.

Can you do something with this temperature range other then just characterizing something to see if its good for the north pole?

[RoGeorge]

AFAIK the Josephson junction (voltage reference dependent on physics universal constants only) can not work at room temperature.
https://en.wikipedia.org/wiki/Josephson_effect
https://en.wikipedia.org/wiki/Josephson_voltage_standard

[CaptDon]

M.R.I. machines rely on super cold temperatures to reduce the resistance and I2R losses in the field coils. Usually the cooling circuit is always on and if I remember the ones we had they were liquid helium cooled and ran around the 3 Tesla magnetic gauss. The newer ones may have been 5 Tesla.

[Terry Bites]

Icy Qbits- they things they use in fantasy "computers" that haven't solved any problems yet- will they ever, probably not.

[Kleinstein]

The very lowest noise amplifiers use a reduces temperature to reduce the noise. It is only a somewhat moderate effect with normal amplifiers, but still worth it when at the cutting edge (e.g. radio astronomy, radar).
Cooling can also be worth it for less noise from optical camera sensors, e.g. for the Astro photography.
For optical sensors cooling can well be worth the effort also in the visible range - it is not just a small effect, when looking at weak signals.

X-ray / gamma sensors made from pure Germanium for high end energy resolution need cooling (often liquid N2), not just for the sensor to operate but even just to keep the sensor alive for a reasonable time. So they need to be kept cold (e.g. < -20 C) essentially 24/7.

Chances are a few odd wavelength laser diodes may need a low temperature:  going to lower temperature can make some diodes reach the laser threshold at a lower current and this would allow to extend the usable range where one can still make a laser.
For some cases the temperature is also used to tune the wavelength - lower temperature extends the tuning range.

[mawyatt]

Some ultra low-loss microwave filters operated at LN2 temp (~77K).

Surprisingly SiGe bipolar transistors get better at low temps, regular Si bipolar transistors don't!

Best,

[SeanB]

LED's exhibit a very marked colour change when cooled down to liquid nitrogen temperatures, and this shows up wit even dry ice temperatures.

[ejeffrey]

Not just thermal imagers, lots of photo receivers are cooled to ~freezing or below to reduce dark noise.  This includes things like avalanche photodiodes, CCDs, and so on.

[MarkT]

Well tin undergoes a slow transformation at low temperature to a different allotrope, causing disintegration to dust, a phenomenon known as tin pest, so you have to be careful with choice of materials at very low temperatures.

[WatchfulEye]

M.R.I. machines rely on super cold temperatures to reduce the resistance and I2R losses in the field coils. Usually the cooling circuit is always on and if I remember the ones we had they were liquid helium cooled and ran around the 3 Tesla magnetic gauss. The newer ones may have been 5 Tesla.

Most MRI machines are superconducting i.e. they have zero resistance and zero I2R losses for the primary magnet; This is generally by the use of a titanium-niobium alloy cooled to 4 K. The magnets are also operated in persistent mode - i.e. once target current is acheived, the magnet terminals can be shorted with a superconducting shunt, and the power supply removed. The current will then freewheel indefinitely.

Most medical systems are 1.5 T, with high end systems being 3 T. A few odd-ball ones such as 4.0 or 4.7 T were developed, but those field strength have largely been abandoned. 7 T is where a lot of research is being focused at present. Cutting edge research is looking increasingly at 11.7 T, with a few research systems operating or in procurement.  A Dutch university have recently signed contracts for supply of a 14 T system; remarkably, this is not planned to use the conventional superconducting technology, but instead use Bi-2223 high temperature superconductor operating at 77 K (liquid nitrogen).

[coppercone2]

Ok. I know the super conducting stuff is the most interesting but that is what I am trying to filter out. I want to know what you can do with something hotter then -125C.

Because you can get that with refrigeration compressors and cooling systems that don't involve crazy.

Could well cooled CCD (without liquid nitrogen) maybe work as a particle detector for something in a enclosure? Like Muon or whatever?

The problem with the really cold stuff is obviously LN... I wanted to know if there is anything you can get with like closed loop refrigeration and electronics. I can call that open loop for the LN2 or He.... those are like gasoline appliances compared to electronic devices. I know you can buy a dewar etc... its just a different world when you need god damn liquified gas in thermos. I don't like how it evaporates it really bugs me that you can buy it and then it slowly decays its like paying a tax. Epoxy storage is bad enough. And food management.

[ejeffrey]

Could well cooled CCD (without liquid nitrogen) maybe work as a particle detector for something in a enclosure? Like Muon or whatever?

Yes a cooled CCD can be used for low signal detection.  This is pretty common, for instance a lot of astronomical cameras are cooled.  Also scientific imaging for detecting weak flourescence and other low signal applications.  Usually they are just cooled with peltiers, but even that can get to -20 C fairly easily.  I don't know specificially if they are used in particle detectors but it wouldn't surprise me.

The problem with the really cold stuff is obviously LN... I wanted to know if there is anything you can get with like closed loop refrigeration and electronics.

There are closed loop refrigerators about as cold as you want, but pretty common down to -40 or -80 C.  There are InGaAs telecom band (1550 mm) avalanche photodiodes that are cooled to this range to reduce dark counts.  In some cases if you didn't cool them the dark counts would completely saturate the detector.

[wraper]

M.R.I. machines rely on super cold temperatures to reduce the resistance and I2R losses in the field coils. Usually the cooling circuit is always on and if I remember the ones we had they were liquid helium cooled and ran around the 3 Tesla magnetic gauss. The newer ones may have been 5 Tesla.

Not just reduce resistance. They use superconductive magnets at ~4 Kelvin (-269^oC, -452°F). The only way to turn them off is to release over $10k worth of liquid helium.

[coppercone2]

how about visible light detectors? IR is not very exciting

[Nominal Animal]

The problem with the really cold stuff is obviously LN... I wanted to know if there is anything you can get with like closed loop refrigeration and electronics. I can call that open loop for the LN2 or He.... those are like gasoline appliances compared to electronic devices. I know you can buy a dewar etc... its just a different world when you need god damn liquified gas in thermos. I don't like how it evaporates it really bugs me that you can buy it and then it slowly decays its like paying a tax.

Liquid nitrogen isn't even expensive; we're talking about euro or two per kilogram of nitrogen, or about the price of milk or fruit juice, when you have a proper contract with a supplier.  (As a liquid, its about 80% as dense as liquid water.)  The flasks are designed to leak, because that makes them safe –– no pressure buildup! ––, and three quarters of air is already nitrogen so leakage is not an issue.  They have a simple very loose stopper, and that is enough for a double-walled dewar flask.  Just ensure good ventilation, so you keep acceptable oxygen ratio in the ambient air.

You really need to think of it as liquefied air with the other gas components removed.  That's how it's produced, too.
To cool something with liquid nitrogen, you can simply continuously boil the setup in nitrogen.

With liquid nitrogen, you can seriously overclock processors.  It matters for single-threaded performance for non-parallelizable non-distributable tasks (including cryptographic hashes).

Germanium, silicon, and gallium arsenide band gap energy all increase by 0.3eV - 0.8eV when temperature drops from 300 K (room temperature) to 77 K (what you get when you let nitrogen boil at one atmosphere pressure continuously).  Thermal (Johnson–Nyquist) noise in e.g. resistors is halved, from \$0.129 \sqrt{R} \text{nV} / \sqrt{\text{Hz}}\$ at 300 K to \$0.065 \sqrt{R} \text{nV} / \sqrt{\text{Hz}}\$ at 77 K.

Alloys benefiting from annealing can become even better (stronger and harder, with fewer lattice defects) at such temperatures; see cryogenic treatment.  Most flexible rubbers and plastics lose their elasticity, though.  (It is one way to machine otherwise too elastic materials.)

Both yttrium barium copper oxide (YBCO) and bismuth strontium calcium copper oxide (BSCCO) practical superconductors have critical temperatures above liquid nitrogen boiling point, and are quite well researched, so if one is sufficiently interested in superconductors these and liquid nitrogen is a valid starting point, with first- and second-generation conductors, 1G and 2G wires, commercially available.

These are the use cases I know of; I'm sure there are others.  I think N-MOSFET drain-source resistance at full conductance decreases, but the gate-source threshold voltage increases as temperature decreases, with total gate charge remaining approximately constant, for example.

[WatchfulEye]

how about visible light detectors? IR is not very exciting

Yes. Cooling reduces noise, in even fairly standard image sensors. This is widely used in "prosumer" astrophotography cameras, which take a commodity image sensor (e.g. Sony IMX533) and operate it at a temperature of -20 C.

[T3sl4co1l]

There's not a huge amount of stuff in "warm" temperatures, as the physics doesn't change much by then.  Weirdo crazy quantum stuff happens at cryo temps, or even LN2 with HTS, but mostly things just change by increment in the 200-300K range: junction leakage decreases significantly, thermal noise drops marginally, resistance decreases some, shot noise remains shot noise, etc.  Junction leakage is the most notable one, I suppose explaining cameras, even when they aren't narrow-bandgap (mid/far IR photodiode/CCD/CMOS) detectors.

For example, the super low noise figure amps (GaAsFETs etc. say used in radio astronomy receivers) are cryo to get the thermal noise down, all that matters is absolute temperature.  There's... something about doping also, I'm not sure how different they need to be to do that (build a functional amplifier at low temp), but maybe that's why FETs.  (Polarization still moves charges even if there aren't free charges from the dopant; note, instead of Vgs being limited to a diode drop, well I mean it still is, by definition, but at this temperature, Vgs could be, you know, 5V or more, at little current flow because the dopant charges are frozen out, versus the ~1.2V it would be at RT, and it can act like an enhancement mode JFET, as it were. But they may also be specific types customized for low temp operation. This should be easy enough to figure out from looking up a couple academic papers.)

Tim

[coppice]

Its interesting how the operating temperature range of some devices has expanded over time. Most MCUs now work down to fairly low temperatures, but in the 1970s, when they said 0-70C for MOS devices they really meant it. Using freezer spray to diagnose bad joints used to keep generating false positives, by just cooling a device to a few degrees below zero.

[ejeffrey]

For example, the super low noise figure amps (GaAsFETs etc. say used in radio astronomy receivers) are cryo to get the thermal noise down, all that matters is absolute temperature.  There's... something about doping also, I'm not sure how different they need to be to do that (build a functional amplifier at low temp), but maybe that's why FETs. 

The issue is that scattering still limits gain and noise whether it is from thermal phonons or crystal defects.  Dopants are "defects" so if you have to dope the semiconductor to get free carriers then you loose a lot of advantage of cooling.  In addition at very low temperature, the donor / acceptor states from dopants aren't thermally excited.

For this reason cryogenic amplifiers typically use other forms of band engineering such as HEMTs and HBTs.  "Normal" enhancement mode Si FETs can also work well since they rely on gate bias to make the channel conductive rather than  (only) dopants but from what I have seen they take 10x the power to get acceptable noise vs a HEMT or HBT.

[T3sl4co1l]

Makes sense, dopants would leave RRR much as they do in metals.  Just, in the opposite sort of direction, heh.  And as most doping differences disappear (frozen carriers), it's more like you're working with intrinsic semiconductor; for which enhancement still works.  Or photocurrent if you're into that, but not really for amplifiers, heh.  But more to the point, HEMTs and stuff, indeed can have a degenerate conductive state at or near 0K which makes them very good in general.

Tim

[mawyatt]

For example, the super low noise figure amps (GaAsFETs etc. say used in radio astronomy receivers) are cryo to get the thermal noise down, all that matters is absolute temperature.  There's... something about doping also, I'm not sure how different they need to be to do that (build a functional amplifier at low temp), but maybe that's why FETs. 

The issue is that scattering still limits gain and noise whether it is from thermal phonons or crystal defects.  Dopants are "defects" so if you have to dope the semiconductor to get free carriers then you loose a lot of advantage of cooling.  In addition at very low temperature, the donor / acceptor states from dopants aren't thermally excited.

For this reason cryogenic amplifiers typically use other forms of band engineering such as HEMTs and HBTs.  "Normal" enhancement mode Si FETs can also work well since they rely on gate bias to make the channel conductive rather than  (only) dopants but from what I have seen they take 10x the power to get acceptable noise vs a HEMT or HBT.

Well over 2 decades ago we coordinated the supply of SiGe HBT devices (~IBM 8HP) for use in an experimental LN2 cooled LNA for Radio Telescope use by Dr Sandy Weinreb at JPL. Recall these proved good enough to displace the presently used devices and went on to use in all the Radio Telescope LNAs at that time in the specific frequency range (which we can't recall, memory  ).

The IBM produced SiGe HBT devices in 7HP, 8HP, 8XP, 9HP and 9XP we utilized were quite good for all sorts of uses, including Cryogenic LNAs!!

Best,