So some folks were pestering me to repost some info I had sent out via email about vacuum systems. Here y'all go. This is just a high level overview based off of vacuum system work I did back in grad school. It's been several years so some details might be a bit fuzzy in my head. Don't build anything based off this and then come complain to me if it sets your cat on fire.
Vacuum stuff can be kind of confusing so kids, gather around the Hi-fi and let unka Dan explain how it works...
To break it down, most vacuum work can be broken down into 4 different regimes:
- Weak vacuum - what most really cheap pumps can achieve. This is where the vacuum is measured in mm mercury. (standard atmospheric pressure is about 760 mm mercury.) It's good enough to degas stuff but that's about it. A $20 pump can do most of what you need here. Mostly you end up paying for pumping rate in these pumps.
- Low vacuum or roughing vacuum - vacuum down to about 10^-4 torr. A torr is a mm of mercury pressure or 1/760th of an atmosphere. You can do some basic plasma and sputtering at the low end of this range. A decent mechanical pump can get you part of the way into this regime but to get to the sub-millitorr range, you really have to throw down some serious money on a deluxe mechanical pump. You can do some plasma stuff at this range, sputtering and some other things but you run into a problem. It is impossible for mechanical pumps to practically go below 10^-4 torr.
The issue is that once you get below 10^-4 torr, you start getting ballistic flow. What ballistic flow means is that the air is so thin that the air molecules no longer hit each other very much so that 'pressure' doesn't really work the way you expect anymore. With regular air, when the pump pulls out some of the remaining air molecules push on each other so a fresh batch get shoved into the pump maw on the next cycle. When you get to the ballistic range, that pushing effect doesn't happen. Basically, it's like a game of hungry hungry hippos. In normal conditions, picture the hippo arena full of sand. As the hippo takes bites, the sand flows to fill the bites the hippo takes out. In ballistic flow, it's like all you have left is a thin, patchy layer of sand left and so it doesn't flow into the hole left by the hippo bites anymore.
In reality, the molecules are still bouncing around and some will get into the pump mouth on subsequent pump cycles but the overall pump efficiency trends to zero. In the ballistic flow regime, the pump effectiveness is determined by two things, the size of the pump mouth and the rate of backflow. If you look at high vacuum designs, the pipes are always as huge as possible so that the remaining molecules are more likely to blunder into the pump. Also, you have to worry about various molecules evaporating up into the chamber that you wouldn't normally think of as being a problem. For example, normal oil has an evaporation rate high enough that it will be evaporating into the chamber as fast as any practical mechanical pump (that needs the oil to not melt) can pull it out.
To get to the lower pressures you need for other things, you need radically different pump technologies and chamber designs.
I tell you all of this to bring across an important point: Spending a lot on the mechanical pump is largely a waste of money. If you have a specific purpose that requires right about 10^-4 torr or fast pumping speeds, spending $3000 on a mechanical pump is worth your time. However, if you can tolerate waiting a bit for pump-down, you can get by with a $100 pump and some cleverness. Now, back to the pumping regimes...
- High vacuum going up to 10^-7 torr. This is the realm that is realistic to achieve via DIY. This is what you need to have if you're doing experiments with any sort of particle beam. E.g. generating x-rays, making an electron microscope, making vacuum tubes or CRTs, etc. You have to be exceedingly careful about the chamber materials. No rubber or normal plastic can be in the chamber - they act as sponges for gas. A single rubber gasket is not only like a sieve at these pressures, a little chunk of it in the chamber can contain enough gas to counteract the pumps for days. Even a single fingerprint in the chamber can render a high vacuum system useless. These chambers are made entirely of metal and some teflon. They are carefully cleaned with solvents and you have to be extremely careful about the samples that go in the chamber. Some high vacuum chambers (and pretty much all ultra high vacuum ones) go through a 'bake-out' process when they are built and are heated up to hundreds of degrees while being pumped down to burn impurities off the interior and pull them out.
This is also where you start seeing really exotic pumps.
The first pump I'll describe is the diffusion pump. http://en.wikipedia.org/wiki/Diffusion_pump The basic principle is this. You take a very high molecular weight oil that has very low evaporation. You put it in a tube and heat it up. The oil evaporates up a chimney and is released through a series of downward pointing slots. (look at the wikipedia article for a diagram.) The oil molecules are sent down and they catch stray air molecules and then the whole lot splat against the cooled sides, condense and fall into the bottom. When the oil is reheated in the base of the chimney, the air is released, in a concentrated area where the pressure is high enough to grab it with the mechanical pump.
Diffusion pumps are fairly cheap (generally less than $1000 for a small one) and have no moving parts other than the oil. The downsides are that they are dirty. Inevitably, some of the oil escapes and coats everything in the chamber. This leads to all sorts of issues as you might imagine. Also, if there is a leak and air gets into the chamber, all that superheated oil catches on fire. You can get lots of cheap junked diffusion pumps on Ebay where this has happened and they're garbage. Don't waste your money unless you feel like trying to chisel burnt oil residue for a few weeks.
This has led to the invention of turbomolecular pumps: http://en.wikipedia.org/wiki/Turbomolecular_pump These look like a standard turbine but aren't. For one, the blades are flat, not curved like a standard turbine. That's because the blades aren't generating lift, they're basically just angled hunks of metal that whack stray air molecules downward. They spin at ungodly speeds and either use special ceramic or magnetic bearings. They are extremely clean and work great. They are also stupid expensive - $5000 and up. Also, if you get an air leak into the chamber, the shock of full pressure air hitting the front blades turns your expensive pump into metal dust.
So, how, you ask can you get to high vacuum on a budget? With a sorption pump. Basically if you have a very high surface area material, it can molecularly 'grab' a monolayer of gas on its surface. This has to do with surface energy of dangling bonds and other esoteric stuff. Just think of solid surfaces as being like velcro for gas molecules. If you have, say activated carbon power and heat it up, that layer of gasses is boiled off. When you cool it back down, any gas molecules that hit it stick and the pressure drops. A well designed sorption pump can get down to 10^-7 torr. (very good diffusion pumps can usually get to about 10^-8 unaided and turbo pumps can get to 10^-10 on a good day) So why do people use expensive pumps when a few dollars worth of carbon powder does the trick? Convenience. Sorption pumps have limited capacity and fill up. when they do, the pump has to be reheated, etc. This is a huge disadvantage if you are doing a lot of experiments. Therefore you use the more expensive pumps. As amateurs, we can afford to be be profligate with our time to save money.
So, if we were to make our own high vacuum chamber? The best way to do it would to acquire a large glass bell jar. We then make a very well polished base plate for the jar and get some vacuum grease such as Apeizon. That stuff has a boil-off rate of something like 10^-8 torr so it won't give us too much grief. The base plate will have a tray has heaters in it and is exposed on the bottom so we can cool it down with liquid nitrogen. We put activated carbon into it to act as the sorption pump. So, to pump the chamber down, we carefully clean the interior with pure alchohol to remove any grease or other contaminants. We use a decent mechanical pump to get down to the 10^-3 torr range and then turn the heaters on. The activated carbon heats up to several hundred degrees and the adsorbed gasses boil off and are pulled out by the mechanical pump. A good idea at this point is to purge the system with pure nitrogen to help flush out the gasses in the chamber, specifically hydrogen, helium and neon which are poorly picked up by the sorption bed. We then close the pump valve and turn the heaters off. As the activated carbon cools, it grabs most of the remaining air molecules out of the system. We then further cool the pumps with LN2 to help pull gasses out of the chamber.The ideal design will be a large carbon tray with separate segments we can use sequentially.
We're basically guaranteed 10^-6 torr with such a system, possibly even better. Which takes us to...
- Ultra high vacuum. This regime is what you need to do things like XPS and other surface science analytical techniques. It's a bitch. Basically, you use either an extremely good diffusion pump, a high end turbo pump or one of the above with an ion getter pump or titanium sublimation pump. http://en.wikipedia.org/wiki/Ion_pump_%28physics%29 http://en.wikipedia.org/wiki/Titanium_sublimation_pump These odd things can get you down to about 10^-12 torr or so but do so very slowly.
The reason we probably can't realistically get UHV is that the chamber design gets insane. Basically you are looking at large stainless steel vessels with half inch thick walls. That's because metal is slightly porous and you have to take into account the rate of air simply oozing through the chamber walls. Vacuum grease is no longer acceptable at these pressures and all the joints are sealed with crushable metal gaskets made of very pure copper or preferably gold. Things like simply running a wire into the chamber become major engineering challenges. It is possible to have mechanical linkages going into a UHV chamber but you can expect to pay kilodollars per linkage.
Yes, we could make a sorption pump with an ion getter that could theoretically get to 10^-10 torr but we'd be very hard pressed to make a chamber that could support that.
the good news is that if we are clever, we can probably make an entire vacuum system that can reliably get to 10^-6 torr for under $400-500 depending on how good we are at scoring deals. The biggest variable will probably be the glass jar. Those can get pretty spendy. We could make a metal chamber but then we lose the ability to see whats going on inside.