Experimental Physics for Morons, #47

I spent the bulk of yesterday afternoon doing vacuum system work, specifically working on the system to feed gas into the atomic beam source. My feelings about this can be inferred from the Facebook status message I set at the time: “Chad Orzel abhors a vacuum.”

The apparatus I’m building uses laser cooling to decelerate an atomic beam of krypton atoms in a particular metastable state. This works brilliantly to slow metastable krypton atoms down, but the only atoms affected by the laser are krypton atoms– everything else continues along unimpeded. As a result, the entire experiment needs to be carried out in an ultra-high vacuum chamber, so I have a large and shiny apparatus with multiple vacuum pumps to ensure that all the air is removed from the path of the krypton atoms:

i-2d974e663fb5b59a0a6159bda96390e7-WMD.jpg

Of course, in order to work with metastable krypton, I need to get krypton into the system and excited to the proper state. And that’s where the “moron” part comes in.

The process of getting krypton into the chamber is pretty involved. The krypton gas starts in a lecture bottle where is is stored at a pressure of several atmospheres. There’s a valve on the top of the bottle, then a gas regulator that limits the pressure at which the gas exits into the rest of the system. Then there are two more valves, with about four inches of copper tubing between them, then about five feet of copper tubing, before a needle valve with a micrometer readout so the opening can be precisely controlled. After that, there’s a gauge, then a length of glass tubing in which we excite a plasma discharge to prepare a small fraction of the atoms into the proper state for laser cooling.

The big series of valves is because the gas in the bottle is at a pressure of several atmospheres, while the pressure inside the vacuum chamber is about one billionth of an atmosphere (10-7 – 10-6 torr). We need to be able to carefully control the pressure of the gas in the discharge region, and cut it down to a level that doesn’t overwhelm the pumps.

Of course, we also need to be able to pump all the air out of all that tubing. The glass tubing is a half-inch in (outer) diameter, and the copper tubing is a quarter-inch OD. There’s about ten feet of tubing in total, and getting the air out of it is a little like sucking water through a ten-foot drinking straw– it takes forever, and the only thing you can do to speed it up is to open all those valves as far as possible.

Of course, it’s important that you close all the valves back down after pumping the air out. It’s also critically important to close the regulator down, as well, before you open the valve on the gas bottle. Otherwise, you wind up filling six feet of narrow tubing with krypton gas at about three atmospheres, and you get to spend an entire afternoon pumping it back out again.

Which is why I abhor a vacuum…

32 thoughts on “Experimental Physics for Morons, #47

  1. Oh, I’ve had to deal with pump oil everywhere, too. That’s not usually my fault, though– it happens because the power goes on and off without warning, and they never tell me when it does, so I can’t shut off the mechanical pumps to prevent oil backstreaming.

  2. If they told you when it was going to go off and on, it wouldn’t be “without warning” now would it?

  3. True.
    The real problem is that they don’t tell me after it’s gone off, either, despite multiple requests to notify me when the power goes out.

    If I get timely notification, I can avoid having a mess made of the apparatus. But I need to know that an outage has happened.

  4. Thank you for making me properly grateful that we’re oil-free (pumping on Cs/Rb means ion pumps for the normal operation) and hooked into a solid UPS backbone. I sometimes take that for granted.

  5. 1) Distaff techie in a lab bikini with wildly teased hair, heavy punk eye makeup, torn black fishnet stockings, bitch boots, a prominent TiN ion-plated nose ring, and a whip. Request NIH grant funding.

    2) Differentially evacuate with a trapped pump on the high pressure side so you don’t suck an aircraft hanger through the hard vacuum end. Gas conductance is forth power of tubing inner radius. Big hit for every bend and constriction.

    3) Lower left corner – duct tape!

    4) A quality model of Robbi the Robot adds scale. “The big machine, 8000 cubic miles of klystron relays, enough power for a whole population of creative geniuses – operated by remote control! Operated by the electromagnetic impulses of individual Krell brains. In return, that machine would instantaneously project solid matter to any point on the planet. In any shape or color they might imagine. For any purpose, Creation by pure thought! (Forbidden Planet) That would be DARPA, not NIH.

    “The secret devil of every soul on the planet, all set free at once to loot and maim! And take revenge, Morbius, and kill!” Definitely DARPA, with SBIR polish. The CIA would FUBAR, DoD REMFs would hit Pacoima.

  6. I’m still convinced that the aparatus is actually a giant death laser. Anyone else notice what he named the picture?

  7. It’s consoling to hear that They turn off and on the power at developed countries as well…

    (o from Jerusalem, Israel)

  8. We bought some high pumping speed M12C Maxima direct-drive two stage oil pumps made by Welsh for Fisher and they are rugged – we run continuously and they are pretty cold and quiet. They have a built-in automatic back-flow shutoff valve so when the power goes off we get no oil sucked in the vacuum manifold lines and cold traps. They cost us about $2500 apiece and they have about 200L/min air displacement speed and nominal vacuum below 0.5mTorr. We have about ten of them, for about 2 years now, and so far so good. Only the delivery times were sometimes annoyingly long.
    (But please don’t buy their smaller M8C Maxima – those are overdesigned piece of clunk and tend to leak or break down after few months)

    Otherwise, your experimental rig would make Holywood people happy. No way for a chemist to match it 🙂

  9. Apropos past blogs about the overselling of giant accelerators, notice the reaction of the readers who have never seen standard AMO experimental equipment!

    You need to pay a bit more attention to this kind of p.r., starting with this wonderful picture. Always look at the background, not the foreground, when shooting for public consumption. A couple of black sheets in the background, then some colorful side fill lighting, and you can out shine the biggest detector at CERN. After all, you can’t really tell how big that thing is in a picture. Put a toy sci-fi character in the shot and it will look gargantuan.

    Must be awesome with the laser on.

    I think National Geographic needs to visit your lab to even things up after their recent LHC story.

    PS – Can’t you suggest an undergrad design project to make a device that would send you an e-mail or text your phone when the power goes off in the lab?

  10. For the finding-out-if-the-power-has-gone-out bit, have you considered getting a $10 electric alarm clock?

  11. Of course, power going out on Saturday and you finding out on Monday probably doesn’t count as “timely” so, um, never mind. You probably already know just as well using more expensive equipment instead of a $10 alarm clock.

    I now have thoughts of some Rube-Goldberg-esque arrangement where something determines that the power has gone out and sends you a text message. Actually, that ought to be easy; have a computer whose startup script does that. Probably a bit more than $10 though.

  12. Another bad vacuum experience:

    Phase one: Pump it down – pressure won’t quite get low enough, and slowly creeps back up. Pump it down – pressure won’t quite get low enough and slowly creeps back up. Repeat ad suicidum.

    Phase two: Determine the maximum number of times that a student with severe obsessive-compulsive disorder can check, re-check, and re-re-check every seal, port, and pump.

    Phase three: Disassemble everything. Discover the deceased mouse.

    Those little guys really can outgas. All vacuum systems should come with a factory-supplied hungry cat. Seriously, don’t leave any opening uncapped if you’re away on semester break. Compassionate faculty should veto any senior-project proposal involving any vacuum < 0.5 atmospheres.

  13. Actually, that ought to be easy; have a computer whose startup script does that.

    Apart from the minor insignificant detail that the message will not be sent until power is restored. Which, for anything more than a minor glitch that wouldn’t be a problem anyway, is too late.

    Which reminds me of the storm last April which left my boss’s house with neither power nor phone service. So he couldn’t turn on a radio or a computer, or call the storm information line, to find out that the university was closed for the day; he actually had to come in to find this out. I had the opposite problem that morning: since I use a battery-powered alarm clock in my bedroom, I had no way of knowing that the overnight power glitch at my house (I did notice that the clock on my microwave had reset) had been more than a glitch at the university (which is why they were closed for the day: at the time they had to make the decision they didn’t know when the power would be back on), and I didn’t think to call because the university doesn’t normally close for the sort of minor snowfall we’d had overnight.

  14. So what’s the advantage of conflat seals rather than KF stuff for these pressures? We regularly reach 10^-7 mBar with KF.

  15. “Apart from the minor insignificant detail that the message will not be sent until power is restored. ”

    Hook the computer up to a UPS with a serial or USB interface. Then the computer would find out the power is out and would be able to immediately send out a message before shutting down.

  16. So how much does the big bulbous silver metal part cost? Is that something you order or something you had fabricated in-house?

  17. If the power outages last only for a few moments, it would make more sense to hook the pumps to the UPS.

  18. We’ve got the luxury of pneumatic valves on one of our vacuum systems; they’re open when powered, and closed when the power gets cut (or when any critical failure takes place, really). It saved us when the seals on an internal cooling block failed, and water started filling the entire vacuum chamber.

  19. So what’s the advantage of conflat seals rather than KF stuff for these pressures? We regularly reach 10^-7 mBar with KF.

    The pressure in the source region is relatively high, particularly when the gas is on. that’s when it’s at 10^-6, and why there are two big pumps at that end (one of which I’m going to have to repair this weekend, annoyingly). There is at least one KF connection in there, and the glass tube enters through a compression gasket fitting, because as you say, it doesn’t really matter

    The pressure down at the WMD chamber end is more like 10^-9 torr, probably 10^-8 with the valves open and the gas on. That needs ConFlat.

    So how much does the big bulbous silver metal part cost? Is that something you order or something you had fabricated in-house?

    It was a custom job by Kurt J. Lesker company. I forget exactly how much it cost, but it was something like three grand.

    If the power outages last only for a few moments, it would make more sense to hook the pumps to the UPS.

    Unfortunately, the big turbopump is really finnicky, and cuts out even for the short spike you get when the UPS comes on. Ditto the lasers– they shut off for spikes that don’t even trip the turbo.

    We’ve got the luxury of pneumatic valves on one of our vacuum systems; they’re open when powered, and closed when the power gets cut (or when any critical failure takes place, really). It saved us when the seals on an internal cooling block failed, and water started filling the entire vacuum chamber.

    Yeah, I should really get a pneumatic valve for the backing line. I have one pneumatic gate valve between the source region and the experimental chamber, to keep that clean (the turbo there turns itself back on when the power comes back, so backstreaming isn’t an issue). Annoyingly, we don’t have compressed air service in the building, so I had to buy a compressor at Lowe’s to run it. It sounds like the end of the world when it kicks on, a couple of times a day.

    They’re expensive, though, and we had a turbo die in grad school when the pneumatic valve failed (snapped closed, and the pump cooked itself when the backing pressure got too high), so I haven’t gotten around to it.

    My bigger worry with the power outages is the effect on the pump lifetime. I’ve lost a few lasers to power spikes, too.

  20. Sounds like you need a separate power filter in front of the UPS, if not a better power block in the first place. That is, a high-end power supply/UPS should not be glitching at switchover! And yeah, a throwaway computer would be fine for notifying you and/or shutting down pumps as needed.

  21. The thing about the notification is, I shouldn’t need to spend my limited funding buying and installing gadgets to automatically notify me in the event of a power outage. We have a Facilities office and a Campus Safety office, both of which are staffed 24/7/265, and have my cell phone number.

    They’re supposed to call me (and several other people) when the power goes out on campus, and when they remember to do the things they’re supposed to, everything works out fine. If they can’t do that, they should pay for the automatic system, not me.

  22. both of which are staffed 24/7/265

    I think I may have identified the problem here 😛

  23. Phase three: Disassemble everything. Discover the deceased mouse. …

    ROTFLBOTWSTN (rolling on the floor laughing, bouncing off the walls, scaring the neighbours).

  24. “They’re expensive, though, and we had a turbo die in grad school when the pneumatic valve failed (snapped closed, and the pump cooked itself when the backing pressure got too high), so I haven’t gotten around to it.”

    Modern turbo controllers should shut down if the backing pressure gets too high.

    Also, does a heater vent or a hair dryer on the long tubes help pump them down faster?

  25. Also, does a heater vent or a hair dryer on the long tubes help pump them down faster?

    I think that would only help significantly if the pump-out rate is limited by outgassing from the walls of the tube. My guess, though, is that it’s just a very low-conductance path, and as such, it takes gas forever to diffuse out to the pump.

    Also — oil-free scroll pumps FTW. They don’t go very low, but they go low enough for backing a turbo, which is all you care about anyway.

    A friend of mine in grad school had his turbopump impeller disintegrate inside its housing. I’m not sure why it failed, but the carnage inside was quite impressive regardless. Luckily, the pump housing didn’t have any problem containing the bits of shrapnel that were formerly rotating at 90,000 RPM (it was a small turbo pump).

  26. Modern turbo controllers should shut down if the backing pressure gets too high.

    Sadly, one of my turbopump controllers is pre-modern. It was a donation from my old group at NIST. I have two newer ones that actually start back up after a power outage (or can be configured to do so), and they’re great.

    Also, does a heater vent or a hair dryer on the long tubes help pump them down faster?

    That would help with the initial pumpdown, but the current problem, as Asad notes, is just a matter of conductance. It’s not outgassing that’s the problem, it’s a large amount of gas at the far end of a really thin tube.

    Also — oil-free scroll pumps FTW. They don’t go very low, but they go low enough for backing a turbo, which is all you care about anyway.

    Yeah, I’ve got one backing a turbo on a test source built for a student project. It’s a great device, and nice and quiet, too. At some point, I’ll move it over to the main system, and life will be good, but I still have some stuff to try over there.

    And now, I need to stop procrastinating, and tear apart an oil diffusion pump. Whee!

Comments are closed.