Solving Ultra High Vacuum Leaks Has An Elementary Solution

When we think of a vacuum leak we generally think of a car that just doesn’t want to run quite right. Most normally aspirated internal combustion engines rely on the vacuum created by the pistons to draw in the air fuel mixture that’s produced by the carburetor or fuel injection system. Identifying the leak usually involves spraying something combustible around common trouble areas while the engine is running. Changes to the engine speed indicate when the combustible gas enters the intake manifold and the leak can be found.

What if your vacuum leak is in a highly specialized piece of scientific equipment where the pressures are about 12 times orders of magnitude lower than atmospheric pressure, and the leak is so small it’s only letting a few atoms into the vacuum chamber at a time? [AlphaPhoenix] takes dives deep into this very subject in his video “Air-tight vs. Vacuum-tight.” which you can watch below the break.

Not only does [AlphaPhoenix] discuss how a perfect pressure vessel is sealed, he also explains the specialized troubleshooting methods used which turn out not to be all that different from troubleshooting an automotive vacuum leak- only in this case, several magnitudes more complex and elemental in nature.

We also enjoyed the comments section, where [AlphaPhoenix] addresses some of the most common questions surrounding the video: Torque patterns, the scarcity of the gasses used, and leaving well enough alone.

Does talking about vacuums get you pumped? Perhaps you’d enjoy such vacuum hacks as putting the toothpaste back in the tube in your homemade vacuum chamber.

Thank you [Morgan] for sending this one in. Be sure to send in your own hacks, projects, and fantastic finds through the Tip Line!

34 thoughts on “Solving Ultra High Vacuum Leaks Has An Elementary Solution

        1. What is the lowness of atmospheric pressure? We can agree that X times lower = divide by X. Or we can agree it it is a syntax error and treat it like dividing by zero with a “don’t do that” warning.

          (Chasing vacuum leaks has been the primary occupation of many a physics grad student.)

  1. Very interesting video.

    Two things did surprise me as the repair was being completed:

    1.) You weren’t using a torque wrench to get all of the bolts torqued down to the same tightness.

    2.) You just went around the outside of the flange in order instead of tightening in a pattern that tightens bolts in a more diagonal pattern. Something like tighten the first bolt, move to the bolt 180 degrees away then move to a bold 90 degrees from that and then 180 degrees. This results in more even pressure on the flange and gasket surfaces.

    1. Yeah, i don’t know, star pattern is the first thing I learned when working on automotive stuff, as well as when a torque wrench is appropriate. In this case he should be using both, for the most reliable fitment. But what do I know, I’m a rando on the internet.

    2. Note that in the body of the article, it says:

      ” where [AlphaPhoenix] addresses some of the most common questions surrounding the video: Torque patterns,”

      Granted, going through YouTube comments is like searching a trash heap, so I’ll help you out, where he replies:

      “I think I’ve replied in more detail somewhere but as this is a top comment I’ll post here too. I was originally taught that you want to “cut in” the knife edge on these copper gaskets continuously, by tightening slowly in a circle, so that’s how I’ve always done it. That said, I know a lot of people that cross-tighten the whole way and also normally don’t get leaks – a great deal of vacuum science seems to be black magic and superstition. If a procedure works well enough, you stick with it, which unfortunately ends in a lot of competing-but-effective methods.”

    3. Also, the comment re: torque wrench was:

      “Regarding the torque wrench, in many many locations on this system, it’s almost impossible to fit a regular wrench around these flanges, let alone a necessarily bulkier torque wrench. This flange probably could have been handled by an open-end torque wrench, which I guess I assume exists but have never seen, but in general to work on these academic systems you need to develop a feel for it. Notice how I’m barely moving the wrench with each tighten.”

      As someone also in academia, I can *totally* understand the “just get a feel for it” problem due to one-off builds with too-tight clearances.

    4. So doing the thing where they are not tightening can often come back to bite you in the rear when working with conflats. Spec is you tighten them till the flanges meet. Torque does not matter and you generally just work your way around the flange, there is not torque pattern like a wheel hub or engine head. The knife edges dig into and deform the copper and the the groove is designed so that when they are flat against each other they copper deforms and take up the space. Generally if it leaks after you tighten them up either you got come crud where the knife edge is or someone dinged the knife edge which is generally unrepairable. If you have access to a machine shop you can recut the flange if you can mount it in a lathe.

  2. I’ve done some of this.

    The interesting bit about vacuum is “mean free path”, which is the distance something will travel in your chamber before hitting a molecule.

    So if you want to make an electron microscope, if the mean free path is less than your beam length, chances are that a lot of the electrons will collide before they reach the target (or collide after reflecting off the target).

    Medium vacuum, something you can get with a single vacuum pump, about 1.0 to 10^(-3) torr, is in the range of a couple of inches. That’s good enough for sputtering, but not enough distance for the electron microscope, and not enough for many applications.

    The Wikipedia article has a nice table in the middle that shows the mean free path by type of vacuum.

    https://en.wikipedia.org/wiki/Mean_free_path

    Once you are below medium vacuum, the molecules are so rare that they (statistically) stop acting like a gas. You have to start using kinetic means to remove individual molecules.

    1. I mean with sputtering you have to have gas in there for it to work! On the machine I built we pumped down as low as we could go with a turbo and then backfilled with argon.

      One interesting thing about electron microscopes is that during operation there is usually only one electron traveling down the column at any time. Kind of like a drip coffee maker.

      1. I recall that metals in a vacuum system, especially copper, have to be “electroless”, which I think means it has to come from a melt and not from electo-depositied billets. If it isn’t electroless it will outgas too much and too long to use. But searching shows that now “electroless” means chemical plating instead of electrolytic – maybe. This thesis was done at CERN https://cds.cern.ch/record/2653446/files/CERN-THESIS-2017-443.pdf

  3. This is interesting since I regularly work with vacuums down to low 10^-7 torr and our problems while similar are no where near as difficult, I.E. viton O-rings vs. knife edge seals. This makes my job seem less miserable when hunting leaks!

    1. Yeah, with viton seals 10^-7 is about as low as you can get in my experience. Trying to get past that is much harder because you’re limited by the gas permeability of the seal material itself.

  4. General Electric’s Glyptal enamel was the leak sealer of choice for our big radar tubes. Then an ion pump brings it down to where the tube will work again.

    Interestingly, those tubes sent up to space still needed envelopes and couldn’t use all that free vacuum, because it was poorer than what we achieve on earth. Satellites have lhin little clouds of gas that float along with them.

    1. @Col. Panek said: “General Electric’s Glyptal…”

      AFAIK the term “Glyptal” does not belong to General Electric in a strict sense. It has long been associated with Glycerine Phthalate. That does not mean General Electric can’t patent or trademark the common term Glyptal. The modern U.S. Patent and Trademark Office is one of the most inept and corrupt government entities anywhere on the planet.

      https://en.wikipedia.org/wiki/Glycerine_phthalate

      Excerpting from above the key point is: “This polyester (Glycerine Phthalate) does not form linear chains, but is built as a three-dimensional structure.” Glycerine Phthalate not having linear chains makes it a very good high voltage insulator, which was/is the primary purpose of Glyptal as a product. You can easily buy what many manufacturers call Glyptal today. But is what they are selling similar to Glycerine Phthalate without linear polyester chains? Who knows…

      https://duckduckgo.com/?t=ffab&q=Glyptal&ia=web

    2. Now stuff like TorrSeal or Hysol 1C is mostly used. I use Hysol 1C for everything since it does not outgas and flows very nice, I have used it to mount optics, glue the magnets back into the starter of my motorcycle, and even fixing some ceramic pieces at work.

  5. Ya, I’ve worked with high vacuum in sputterers, plasma ashers and electron gun milling where the vacuum is measured in terms of torrs and not in terms of inches of mercury column, and the pumps were either cryo or lobed rotary. No copper sealing was used, just large o rings that, when cleaned with alcohol, needed to be thoroughly dried out before reuse or the alcohol would outgas and indicate a leak when there wasn’t one (could be eventually pumped down but with a lot of extra time involved). Torquing down involved first simple hand tightening initially then gradually in round sequence rather than automotive star pattern. The big trick was taking it apart where if you loosened one side too much you couldn’t get the opposite bolts loose at all. I had to rescue a greenhorn maintenance tech who did that when he complained that he needed to drill a bolt out of the hole because it absolutely would not come loose. I re-torqued all the bolts, undid each one by the same method they were tightened, and all the bolts came out easily.

  6. Lots of great comments on here!

    One additional thing I’d add is about swagelok fittings. These are very commonly used on gas input lines for UHV systems, but they’re also widely abused. Often I’ll find a poor fitting that was the victim of someone doing their impression of a 300 lb gorilla. If a fitting is over tightened then the cone of the fitting will flare out, ruining the fitting. Over tightening also compresses the ID of the tubing thereby constricting flow. The single easiest thing to do is just buy a gauge tool from Swagelok, then you _know_ it’s tight but not too tight.

    Swageloks are also designed to be re-connectable. But contrary to popular belief, if they were properly tightened the front and back ferrules won’t come off the tubing when the nut is removed. To reconnect just reassemble and give it an 1/8-1/4 turn more. This works well for UHV and doesn’t ruin expensive fittings or make you cut and bend new tubes.

    1. Swagelok also also has a separate tool for setting the ferrules instead of putting them on using the fittings. And I think 1/4″ only needs 1.25 turns of the wrench for full seal on first setting and after that is just like 1/4 turn past snug. And like you say, they are good for UHV and you dont have to deal with the gaskets of VCR fittings.

  7. Not only is “twelve times lower/less-than” bad writing, 1/12 of atmospheric pressure is about 1.25 PSI, which is hardly “ultra high vacuum”.

    If “1/12 of” wasn’t what was meant by “twelve times lower”, that simply supports my point that “twelve times lower” is bad writing–objectively bad because it fails to communicate the writer’s ideas.

    I enjoy Hackaday quite a bit and read the site daily, but I’d like to see a little more attention to *correctness* in such matters.

  8. Hi, I’m guy who wrote the “12 times lower than” line. I clearly got that wrong and have updated the article to say “12 orders of magnitude lower than” instead. I appreciate the constructive feedback given!

  9. Today, I needed to use captions to watch this. To quote the video (as I experienced it): “There is uncertain out meaning but with the systematic to we and our commitment ở đường c1 siêu on it like that.”

  10. To make an efficient FINAL vacuum pump, use an electrostatic treadmill and scraper. With a fan-type vacuum pump eventually the atoms will rarely touch the fan blades. A piston pump means your odds of catching then squeezing the last atoms out is astronomically slow probability wise. The electrostatic treadmill acts like flypaper for sparse atoms and moves them to a catcher where they can be mechanically scraped off quickly, plus the larger surface area means a higher probability of catching those last sparse atoms.

    Also LAYERED MULTIWALL VACUUM VESSEL. Think of a pressure cooker, the pressure differential between walls determines stress. Imagine a pot within pots each has different pressure gradients. Outermost pot has 25% differential between inner pots and standard air pressure. Next sealed pot later has 25% pressure differential and so on. 4 layers inwards and the pressure is 0% in the core pot. Because the gradient is 25% you can actually use thinner cheaper materials and scale the whole thing up with Pressure Gradient Bricks for whole room inexpensive stable vacuum chambers. Ideally in the multilayer onion of pressure chamber pots, you’d use polymer beads to pad out the gaps between pots.

  11. We have a similar MBE system. What was the initial pressure level that made you to think there is a leak in the system? How is the base pressure after fixing the leak? Thank you.

Leave a Reply to maconaCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.