Vacuum doesn't suck, Part 2.

At atmospheric pressure (like you're in now!), any old fan can pump air out of a chamber. All the air molecules are pressing against each other and trying to expand. The fan blade hits them and pushes them away. But after a while there isn't as much air and the fan struggles. Air is pushing back against the blade, and a lot of those molecules get through.

When air molecules are bumping against each other and pressing in all directions, the pumped stream is called viscous flow. When the air molecules are hitting the walls of the chamber more than they're hitting each other, it's called molecular flow, and a simple fan won't do. What will pump the remaining molecules? Why, yes! Boiling oil!

There are many kinds of vacuum pumps, but one of the most common is called a diffusion pump. They are cylinders with silicone-based oil inside that is boiled by a heating element. They look like this:

The rising oil vapor goes into a device that directs the flow of the vapor back downward.

How does it work? Well, the remaining air molecules in the chamber are zooming around like crazed banshees on speed. Some of them go downward by chance, some hit a wall or another molecule and lose energy, and all of them have to deal with gravity. If they get low enough, the downward-flowing oil vapor pushes them down further. The more-concentrated air at the bottom of the pump is then removed by a simpler pump.

Here is the basic procedure which is used on a lot of vacuum equipment. There is a pump outside the production area (for cleanliness) that is similar to a fan, but the blades spin through oil to reduce leakage. Most of the time that pump, called a mechanical pump, is used to remove whatever molecules are loose in the diffusion pump. When it's time to pump down a chamber, a valve redirects the mechanical pump's action from the diffusion pump to the chamber - if the diffusion pump were exposed to atmospheric pressure, it would be overwhelmed. When the air pressure in the chamber reaches approximately 6% of atmospheric pressure, the mechanical pump switches back to pumping on the diffusion pump, and a large valve opens that exposes the diffusion pump to the chamber, and pumping the molecular flow of air begins.

Just a brief word about measuring vacuum. Atmospheric pressure is about 14.7 pounds per square inch, which is simply the weight of the air on that square inch. On half of a square inch, the weight would be about 7.4 pounds. That's not really handy, since it's area-dependent. So systems are used that measure the weight of the gases independent of area. Imagine that you have a U-shaped tube that has all the air removed and is partially filled with a liquid - mercury, for example. The height of the mercury in each half of the tube would be the same. But if you then opened the top of only one of the tubes, air would press down on the mercury in that side and the other side would rise. The amount that side rises represents the pressure of the air on the other side. It doesn't matter how big the two sides are, as long as they're equal.

Using that method of measurement, standard atmospheric pressure is about 29.92 inches of mercury. Using our U-tube, the airless side of the tube would see the mercury rise that high. That's equivalent to 760 millimeters of mercury (why is this country too dumb to switch to the metric system?). When measuring vacuum, 1 millimeter of mercury is called 1 torr, because "torr" is a lot easier to say than "millimeter of mercury".

If you're still reading, here is the last thing. When we started pumping our chamber, the pressure inside was 760 torr. To reach high vacuum, the pressure has to be reduced to 1/1,000,000 of a torr, and that's kind of low high vacuum. A billionth of a torr is common pressure for many processes.

Okay, only one more of these to go, since time is running out. The last part will be about finding leaks in vacuum equipment, which is probably the most fun part. It was for me, for sure.