Helium leaks are a nightmare. During my PhD I worked with a self-built dilution cryostat that would often have leaks in the custom-built Indium seals. To find the leaks you'd have to pump the cryostat to high vacuum, hook up a portable mass spectrometer tuned to Helium to the pump circuit and then spray different parts of the cryostat with Helium from a regular gas canister. The He would then get sucked in through the leak and show up on the mass spectrometer, which was coupled to a loudspeaker so it would cause a sound whose pitch increased with the measured He density. Once you found the leak you had to vent the entire system, remove the faulty seal and replace it with a new one. All seals were handmade, i.e. you took a small filament of Indium, placed it between the flange and the housing (after carefully cleaning everything with Acetone) and carefully screwed it shut, turning each screw only a tiny bit at each turn and going around all the screws until you could see the Indium squeeze out of the edges.
Even worse, some leaks would only show up when the system got cooled down to liquid Helium temperature. When that happened you were out of luck as you can't cool the system down to 4K before spraying it with Helium, so you had to just guess where the leak might be and replace all seals in that area until you found the right one. Going even deeper in temperature would eventually turn the He4 suprafluid, which means that it loses all internal friction. In that state it would squeeze through even the tiniest molecular cracks, so again if that happened you just had to redo all the seals and hope they would hold.
AlphaPheonix on YouTube has an excellent and informative video of a very similar process of testing seals, though the vacuum is pulled for a different reason: https://www.youtube.com/watch?v=VD69crOFx10
One would imagine there must be some kind of paint you can paint onto the outside of a vacuum flask, and have it be sucked into any gaps and solidify there.
It seems they already make a product for that - vacuum grease. It can get sucked into even a fairly large crack, and plug it up, as long as the depth of the crack is much longer than the width, then atmospheric pressure won't provide enough force to send it deeper into the crack.
And the grease itself is designed not to evaporate into the vacuum.
> paint onto the outside of a vacuum flask, and have it be sucked into any gaps and solidify there.
Stuff like paint will tend to be very permeable- if it relies on a solvent drying out, then obviously its permeable enough to let the solvent through. Even if it does dry to a very solid state, there will be a very very long tail of evaporation of volatiles inside the substance.
> It seems they already make a product for that - vacuum grease.
Vacuum grease isn't perfect- it's not recommended for the high end of high vacuum or for UHV. Or for single digits of kelvin.
Another big factor is that if you're doing science, you probably want to go out of your way to eliminate contamination like that. Guessing about the possible chemical interactions of a secret proprietary substance under extremely low temperature and pressure is not a fun way to get your doctorate.
Yes, a huge factor is the outgassing of materials. While just solidifying and a long time after that, it gases off solvents and other small particles that you constantly have to pump away and that raise your pressure noticeably.
In cryogenic environments this is less of an issue because things are more sticky, but in a room temperature apparatus targeting ultra high vacuum this is troublesome or impossible.
For a schlenk line in chemistry where you need 0.1 - 1 mbar, sure! But this is insane physics territory... Like other people said, grease isn't behaving like you think it would at millikelvin temperatures, or at pressures lower than those in the interstellar medium.
If you have a big aluminum chamber under ultra low pressure, guess what you're concentrating in there? Hydrogen and helium from the atmosphere that's seeping through your metal. Crazy stuff that's (sadly) not fixed by slapping PTFE paste or vacuum grease onto it.
Not sure if there's vacuum grease that can work at very low temperatures, the upper stage of the cryostat would go to 300 mK and the lower stage to 20 mK, at that temperature non-metallic compounds tend to get porous. There are materials that can work in there but to my knowledge nothing that was an easy solution, at least at the time.
If you're dealing with single molecule cracks and extreme vacuum, even vacuum grease is going to offgas and contaminate things.
Hydrogen is EXTREMELY small, the temps the poster is talking about are very very low, and XHV/UHV is extremely low pressure vacuum.
Like 'single molecules we don't want rattling around is a problem', 'more vacuum than outer space vacuum', and 'oxygen long ago turned into a solid' temperatures.
Those materials exist (see Apiezon vacuum grease and Lesker vacuum sealant). In less-demanding or desperate applications, they can work.
But like similar last-ditch repairs (think automobile muffler tape or radiator sealant), they only occasionally last. Usually the best thing to do is to go actually fix the problem.
The above advice assumes you're trying to maintain high vacuum (10^-5 torr or lower pressure). At low vacuum (10^-3 torr and above), that trickery can work wonders.
Oh yes, way too familiar situation... dry cryostats were really a game changer. Apart of the incredible advantage of not having to refill them frequently and on Sunday, they are much more reliable and without any cold seal
For an experimental physicist you basically debug until you get the data for your paper or until you find your sample is bad. So you’re debugging 99% of the time.
In many fields it’s rare to build new instrumentation. You fab with well known techniques with one subtle modification, so it’s not even clear what part of your time isn’t spent debugging.
Basically 2-3 years of getting everything to work and testing each component, than 2-4 weeks of good measurements that comprise the main results of the PhD thesis.
Shame they don’t give you any extra certification or recognition for more than two years of learning about the care and maintenance of cryogenic high vacuum systems… that’s a whole persons job in some places!
I’m not looking forward to thermal vacuum testing for my robots… because I don’t think there’s a big enough chamber in Australia… option one is (assuming funding) building a thermal vacuum chamber that can get down to GEO level vacuum with all the fun a large volume vacuum system entails… option two is shipping the entire thing to a friendly country with larger chambers (ESA in EU or NASA in the USA) but then there’s the space technology export paperwork … and I don’t know which would be worse.
And that does not include the problems with electric wires. We had an unofficial rule: after half an hour debugging the electric wires just change all of them. I still have nightmares about BNC connectors.
To that rule, my team added another: when throwing away the cable, cut it in half. So nobody scrounges a cable out of the garbage can only to waste more time debugging a bad cable.
That’s a pretty terrible rule. Unless you’re rolling over them with chair wheels, a bad BNC cable should be easy to spot by visual inspection. In my 20 years of instrumentation work it’s rarely the cable.
It was a students' lab, where each device is connected and disconnected 2 or 3 times per day, 5 days per week. The wires get a lot of use and abuse.
Usually the problem is the tip that got slightly loose and the signal is now intermittent or null. IIRC, when we detected the wrong wire we had to put it in a box and the guy in charge of the lab equipment would fix it later.
If you’ve got an apparatus with 20 parts (a low estimate) and they’re all independently 95% reliable (a high estimate), it’s only gonna be in a usable state 36% of the time.
A good friend of mine made a leak checker from scrap parts while I was working at a university lab. It was just for a UHV system (no cryo). I wish I still worked with him.
Ha, I also remember using a helium canister to look for leaks in a vacuum system in the lab. At least ours wasn’t at cryo temperatures though. Sounds like a real pain.
Not my area, but I know He is very expensive (as is getting it cold enough to make He4) so using He4 as a leak detector is likely going to blow the entire experiment's budget.
Nuclear reactors offer all the fun of leak tracing and invisible cracks with the added bonus that the fluids involved are extremely radioactive and contaminate everything they touch. This is why all the "just mass produce small reactors" and "just try thorium / molten salt" stuff hasn't taken off, and may never: all the beautiful theory disintegrates into man-years of laborious leak testing.
LWR coolant is not that bad. They keep the water very clean with ion exchange resins because otherwise you can have problems with corrosion. Workers sometimes float in the water during refueling where they flood the area about the reactor and spent fuel pond and open the lid of the reactor vessel.
Leaks in liquid metal fast reactors are much more obnoxious but still manageable. Sodium catching on fire when it hits air frequently isn't as bad as it sounds (a "pool fire" isn't particularly hot or dangerous but a "spray fire" can be) but it is important to catch sodium leaks quickly without false alarms and many development projects had trouble with that.
I always enjoy the handwaving by nuclear enthusiasts. These are some of the most difficult engineering systems routinely built and operated by mankind, where a coolant loss accident doesn't just destroy the reactor but can turn the facility into a multi-billion dollar cleanup effort in minutes.
I would liken this to airplanes. They are similarly difficult to engineer and maintain, they can also go very very badly in mere minutes if operated incorrectly, they also have very costly recovery procedures if they crash somewhere populated.
For both systems there have been catastrophic accidents that were national tragedies, from which we learned a lot. "The rules of aviation are written in blood," and that is also true for reactors. Past failures like the one you linked don't mean that reactors are unsafe; if anything, because we had that failure to learn from, reactors are now much safer. If we had a volatile technology like nuclear reactors and none of them had ever had any accidents, then I would be very hesitant to have one in my home town since it could very well be the first to ever blow up.
I think the reason they have seen a lot of success where nuclear reactors haven't is simply up to the fact that they are both cheaper to build and that the public can very directly see where the benefit to them comes from, whereas nuclear power is abstract and indistinguishable from coal to the average person.
I agree with your analogy to an extent, but I think it's got a very significant flaw.
The consequences of aeroplane disaster are localised in space (to the immediate vicinity of the vehicle and whatever it crashes into) and time (once the crash has happened, it is more or less over).
Nuclear disasters can have global effects, and can last for decades.
I'm cautiously pro-nuclear in theory, but am awestruck by the incredible forces unleashed and the extreme impact they can have when mishandled. It's far more potent than anything merely mechanical.
Monju was a poorly designed machine which was particularly badly run. I recently found a book they wrote about it and when I looked at the plans I thought "I can't believe they built that in an earthquake zone". I don't know if anyone is ever going to build another loop-type LMFBR as many of the problems you can have with a loop-type can't happen in pool-type reactors.
What was most shocking about the fire at Monju wasn't that it happened but that they tried to cover it up. Neither that nor the incident where they dropped the refueling machine into the reactor vessel were dangerous to people off site but the latter sure convinced everyone they didn't know what they were doing.
Contrast that to Superphenix where the fuel transfer drum failed and they struggled with a steam turbine system that was procured under corrupt contracts but overall had a good operational record. Or the three pre-1970s cases where there was significant fuel damage in the US but they found that a core melt in a LFMBR isn't as bad as it sounds because the iodine (most dangerous radioactive element in the fission products) reacts with the sodium and the NaI dissolves in the sodium so it doesn't go anywhere. Or EBR-II and the FFTF which performed flawlessly, or the highly successful fast reactors in Russia.
"Construction started in 1986". Is this an example of an accident that would not occur today due to eg better technologies and practices, or an example of something inherent to nuclear technology that can never be made safe?
Bringing up unrelated highly controversial topics is not good. It tends to derail threads and accumulate emotionally charged, poorly informed comments.
But anyway, leaks of radioactive materials are inherently much easier to detect than other kinds of leaks because they are radioactive. Every major hospital in the developed world has a radiation safety department or equivalent performing regular leak and contamination testing in association with scintigraphy and brachytherapy, but this has not prevented the widespread use of radionuclides in medicine.
The notion of "contaminat[ing] everything they touch" is also a misconception that the radiation protection community has tried to combat for decades: radiation is only a meaningful hazard insofar as it reaches levels comparable to the natural background radiation produced by 40K, 14C, and other sources pervasive in the natural environment. There is therefore a level of dilution beyond which radioactive contamination ceases to be of meaningful concern, just as is true with all other toxic substances.
Source: as a medical physicist in training, I work with a radiation safety department at a major US hospital.
As someone who is currently in a game of cat and mouse with an intermittent drip from underneath the kitchen sink, this was exactly the first thing that came to my mind also!
I recently played this same cat and mouse game with no success, so I built a better mouse trap using a disposable aluminum foil cooking pan and a Moen water leak sensor. I put the drip pan on a steep angle and bent/reshaped it to collect water at a single point where I placed the sensor. I wrapped the sensor in a small piece of paper towel so that even a single drop of water would be absorbed by the paper towel and trigger the sensor. I tested it with a single drop of water from a syringe, replaced the paper towel to reset the trap, and eventually I got a notification in real time on my phone as it dripped. Over kill? Yes. But it lead to me finally finding the issue.
I had a bathroom sink where we put in a new faucet that had a pinhole leak in the tubing which you couldn't see when installed. Everything would look good, but the water would come out and drip down the drain pipe.
I feel like I would just replace all of it than play cat and mouse. Surely it would be cheaper in time, and low cost until you figure out you have to replace the whole sink.
Even a new vessel need new plumbing to connect it to the vacuum pump (or pumps, sometimes you need one to to go from ambient pressure to low pressure and another to go from low pressure to very low pressure) and the pressure measuring device(s) and also sealing the holes where the wires go (I guess they have some sensor inside the vessel connected to things that are outside). So a brand new vessel means restarting all the seals from zero.
I have replaced the p-trap under sink already, but your comment actually got me thinking maybe it's a leak around the caulk seal where the sink joins the counter or faucets!
My family (avid do-it-yourselfers, who renovated each house we lived in) used to score project size by trips to the hardware store, both proposed/planned/expected, and also actual. :P
Even worse, some leaks would only show up when the system got cooled down to liquid Helium temperature. When that happened you were out of luck as you can't cool the system down to 4K before spraying it with Helium, so you had to just guess where the leak might be and replace all seals in that area until you found the right one. Going even deeper in temperature would eventually turn the He4 suprafluid, which means that it loses all internal friction. In that state it would squeeze through even the tiniest molecular cracks, so again if that happened you just had to redo all the seals and hope they would hold.