Ok, now we get to something real interesting that could go either here, or under metrology, but since you need a vacuum to do mass spectrometry, well, it had to go someplace and since this is vacuum metrology, why not here?
I am lucky (or well, I spent a ton of money, so maybe the luck was having the dough) to have a mass spectrometer here, on the large system. It is a fairly new Pfeiffer QMG-220, with a mass range to 200 amu and sub-integer resolution for mass, and a noise floor roughly e-14 millibar for partials when scanning at a decent speed -- it will go about one more order lower if you scan slower and it can average things out of the noise better. Mine does not have a channeltron in it -- I went for "rugged for a beginner-idiot" on the spec -- and this turns out to have not been a dumb idea.
After all, this pressure level is about 6 orders magnitude less than my process level for fusors....and I can still see contaminants fine.
Someone on another forum once told me I'd never learn to interpret gamma ray spectrometer results. Hah! Those are utter child's play compared to a mass spectrometer, believe me.
The reason is, at the pressures they will run at (a few times e-6 mbar at most for decent rez) the mean free path is long enough for some fairly odd things to happen. Let's say you have a perfect vacuum, with just a trace of water in it (close to the case here). You'd maybe expect to see line 18, or two ones (H's), and a 16 (O), right? Nope, you get a whole crowd of lines for that. You see O and O2, H and H2, OH, H30+ etc etc. This is because the ion source in one of these will bust up molecules, and charged particles in a vacuum will kind of cling to the nearest thing they hit -- which might have been a neutral, so you see compounds that never occur in chemistry, where they would fall apart as soon as they touched anything.
So the first thing to learn is that -- you see a bunch of lines, you should think that it might actually be simpler in there than it looks, and try to get a clue what would generate that "family" of lines. Unless you're putting exotic chemicals in there, this might not be too hard, actually. But you have to be looking at the evidence in the right way or it's going to totally fool you. So if you've got hydrogen in the tank -- you are going to see everything and everything+1 for example, normally. Some software attempts to sort this out, but none really gets it right compared to a smart human, or none I've seen. It is useful to keep a copy of the expected weights of what really is likely to be there handy (water, air, including Ar, COx, alcohol, acetone, whatever) and how it breaks up, that will help you a lot, not to mention knowing what you've put into the tank recently.
Ok, let's have some fun and look at the pretty eye candy a little bit. As always, click the pic to make it bigger.
As I said, this can be a bit daunting, but it's also clear why "water is the enemy" is the word in vacuum systems. The system was at about 2e-8 mbar at this point, and this truncated to mass 50 scan was done fast so there are some noise artifacts in it. You can pretty much count on those skinny lines, which are narrower than the normal resolution are in the noise and mean nothing. You will also notice that the masses are a bit off from the periodic table -- I've not yet recalibrated that since I got this upgraded lately, but hey, you can read through that, right? Well, we can make the little stuff go away by having it bin things into unit masses, but due to that error, we now have another one:
As you can see, all the "1" became "2" due to the shift off calibration. But we can also see less noise and see farther down as a result. Obviously I need to do that calibration though!
Now, this system had been under pumping for maybe a week or 10 days since the last tim a fusor run occured. It's been a long time since it was up to STP, maybe a month. What we are seeing, mainly, is what can get through the viton O rings that seal the door and a couple of feedthroughs - the rest of the tank is clean, and CF flanged. You'd think that a week or so at about e-8 mbar would have outgassed the tank, right? Nope. Let's turn on the internal quartz heaters for about 30 seconds and take another scan.
Whoa! Look at how much more stuff there is now (see scale factors). That water line set blasted up huge! Even at that low base pressure, water permeating into the tank through the viton tends to collect on things, doggone sticky polar molecule, but the light and heat blast it back off...from 3e-12 to 4e-11 -- more than a factor ten higher pressure (the pump was still running)! I didn't happen to note the ion gage reading this time, but it normally tracks this -- you see a real quick rise when you turn on the heaters, so mostly likely it jumped to 3-4 e-7 mbar or so.
Note that though this makes the tank pressure rise, this is helping get to a lower pressure -- the higher we can raise it this way, the quicker the pump can take it out -- look how effective a few seconds of heat and UV was compared to just waiting a week or more! When you turn the heaters off, it will go to a lower pressure than it was before turning them on, almost always (unless there is a leak, but I just did a He leak test that says, not really -- He stayed in the noise while hosing the tank down with it, pretty much, looked like a little might have gotten through the viton on the door at one point, but not much even then. When the worst I can get is pushing from e-15 to e-14, maybe, I'm a happy guy.
Ok, I then turned off the heat. Nothing really got hot much in there other than the heaters themselves, what we saw was mostly the UV from them blasting water off the walls and taking it apart. Note the water lines are lower than at the start, from just about one minute of that, and very little time to recover and cool....An hour bake and an hour cooling does a heck of a lot more good.
As I'm running out of attachment limits on this post, I'll reply to myself later with more mass spec goodness and explanations of why we see what we do. For one thing I've not yet shown you a full range scan up to mass 200 and there are "interesting" things there too, just not nearly as much as there is water, and residual air down here below 50.
For fun, here's a screen shot of what I see in the lab, split over two screens there. Note that I am running Linux (ubuntu 10.4) and running this windows XP .net DCOM app in virtual box, which did take a little trickery, but that's where all that nasty, crashy, slow, insecure stuff belongs -- in a sandbox on a real operating system. This has to be the worst, yet still kind of working software I've ever seen (and the next few on the list are all .NET). I'm not sure it's the fault of the tools, except maybe they make it too easy for people to think they are programmers, who should really stay out of the kitchen if they don't know better than to try and do realtime in things like this with polling loops, and highly wrapped comm protocols they don't understand the issues with.