A User's Guide to Vacuum Pumps Part 1: Noisy Pumps

A great many physics experiments need to be conducted at low pressures, in order to avoid sample contamination, thermal effects, or dissipative forces produced by interaction with air. Some experiments don't require all that much in terms of vacuum, while others require pressures so low that they're limited by the diffusion of gasses through stainless steel.

To cover the wide range of pressures needed for different experiments, there are lots of different types of vacuum pumps, and there are nearly as many schemes for classifying them. As an experimental physicist and blogger, though, I've come to the conclusion that vacuum pumps are rather like the political views of people on the Internet: not only do they all suck, they can best be described by considering their position on two orthogonal axes. Internet political quizzes use plot some social measure versus some economic measure, and inevitably end up arguing for the superiority of libertarianism. The axes of interest for vacuum pumps are quiet/noisy and clean/dirty.

I had toyed with the idea of splitting this into four separate posts for the four quadrants of the unified pump classification plot, and scheduling those for last week when we were out of town. I didn't get around to that, though, and a couple of the quadrants would be kind of thin, so we'll condense it into two. Today, we'll tackle the class of Noisy Vacuum Pumps, the left-hand half of the graph above:

Noisy and Dirty: the canonical noisy and dirty pump is a Rotary Vane Pump, consisting of a rotating cylinder with sliding vanes offset from the center of a cylindrical chamber. As the vanes pass by the inlet, gas rushes in to fill an expanding volume, as the rotation continues, the inlet is closed off, and then the gas experiences a decreasing volume and increases in pressure until the outlet opens, allowing gas to escape.

These pumps tend to be noisy, as they require a motor to drive the revolution of the rotor, and they have a lot of moving parts. When they reach their minimum pressure, they have a tendency to rattle quite a bit. They also tend to be dirty, as the air-tight seal between the vanes and the walls of the pump chamber is maintained by a layer of oil. Unless precautions are taken to prevent it, the pump oil has a tendency to creep out of the pump and into the sample chamber. As they age, these pumps also inevitably develop leaks, and they can also blow a seal, in which case they tend to spill a liter or two of oil all over the floor. Vacuum pump oil is very viscous, and is a bitch to clean up.

These pumps are good for reaching a pressure of a few millitorr or so (atmospheric pressure is about 780 torr), which isn't terribly impressive, but is good enough for many simple experiments. They do tend to be fairly cheap, though, as every physics department in the world has acquired dozens over the years, and they're fairly robust. Anybody with the basic skills needed to be an experimental physicist can do basic maintenance on them, and there are few repairs that can't be done on-site by a competent technician.

Also in the "Noisy and Dirty" category is the oiled bearing turbomolecular pump. A turbomolecular pump is basically like a small jet engine-- an array of rotating fans that spin very rapidly, forcing gas to move from the inlet to the outlet. The small turbopump in my lab spins at 1,500 rpm, while the larger pumps rotate at 27,000 rpm.

These are noisy pumps because that rapid rotation vibrates the pump housing a bit, producing a constant high-pitched whine while the pump is in operation. The bearings in a typical turbopump are astoundingly good, so they're much quieter than any rotary vane pump, but the noise can get a little annoying.

These are classed as "dirty" because, as with anything else that vibrates at that speed, they require some protection against friction, in the form of oil or grease applied to bearings supporting the shaft. Unlike a rotary vane pump, the oil is not directly in contact with the vacuum side, though a tiny bit will sometimes creep in. The dirtiness is really only apparent when you need to change the oil or grease, which must be done periodically. Inevitably, the oil gets all over everything in that process.

Turbmolecular pumps are high-capacity pumps, able to reach extremely low vacuum pressures. They require a "backing" pump to take away the exhaust gases, typically a mechanical pump like a rotary vane pump of a scroll pump, and the ultimate vacuum pressure reached with a turbo pump will depend on what sort of backing pump you use, and how clean it is.

Noisy and Clean: The best example of a noisy and clean pump is a magnetic-bearing turbomolecular pump. These are identical to the turbo pumps described above, except instead of bearings lubricated with grease or oil, they have magnetically levitated bearings, that (in theory at least) don't come into direct contact with anything during normal operation.

A cool thing about magnetically levitated turbo pumps: if you cut the power when they're spinning, the rotating magnetic bearings will act like a generator, and can be used to supply power to the pump controller for a short time. The controller for the magnetic-bearing pump I used at NIST would stay lit up for a good ten or fifteen minutes, even if you physically unplugged it from the wall outlet.

Turbopumps, of either variety, are frighteningly expensive. The smaller one in my lab cost about $6,000, while the larger system was about $13,000. You use these only if you have a really large gas load to deal with, and you need the system to be clean.

Another clean and noisy pump is a scroll pump. These work in more or less the same way that a rotary vane pump does, but instead of a single rotating shaft with sliding vanes, they use an interlocking pair of spirals. The spiral design gives a much higher pumping efficiency, which means that these can operate without requiring oil to make a seal, and thus, they can be very clean. They still rattle, but they're also quieter than rotary vane pumps. They're considerably more expensive than rotary pumps, while still only reaching millitorr pressures, so they're used only where you really need to avoid oil contamination.

For high-vacuum applications, a type of pummp that's cleaner than a turbo pump, but still in the noisy category is a Cryopump. A cryopump, as the name suggests, uses cold to pump gas. A circulating gas is liquified and used to cool an assembly that sticks into the vacuum chamber (which is sometimes coated with charcoal or some other material to increase the surface area). Gases whose boiling point is higher than the temperature of the cooling liquid will tend to freeze to the surface, removing them from the vapor inside the chamber and thus lowering the pressure.

Cryopumps are noisy because the best ones use a built-in compressor to liquify the gas on the spot. The kind I used in grad school used ultra-pure helium gas to cool the internal assembly down to about 8K, and the compressor made a constant chunka-chunka noise that you could hear out in the hall. On a couple of occasions, I was able to diagnose a problem in the lab from my office, just because I couldn't hear that noise.

They're extremely clean, though, because there's nothing at all dirty to get into the vacuum system. They're also frighteningly expensive, not only for the pump and compressor, but because they require ultra-pure helium gas, which needs to be replenished from time to time.

And those are the Noisy vacuum pumps. Next time: Quiet Pumps