Basics 2:The Complete Dummy's Guide to Operation of a Typical Diffusion Pumped High Vacuum System |
This page will take some time to load, due to the included drawing. Text browser users can download draw1 for use on another viewer.
This section is just for all those people who have never had anything to do with a vacuum system of any kind but need to use one. Diffusion pumped systems are still quite common, and used for all sorts of applications e.g. Vacuum Evaporators for making metal coatings. The diagram below is of a small bell jar system that we use for a variety of things here in the lab.
Since the introduction of the SI units, the unit "torr" is no longer to be used. Preferred units are now to be Pascal, it's derivative millibarr, and bar. Here are the Conversion Tables for Pressure Units from Edwards High Vacuum's web page for those who, like us, are still using older equipment calibrated in torr.
I have taken tongue in cheek "artistic license" and changed a few things in the drawing. The real-life glass belljar has been replaced by a metal vacuum chamber in the picture and the ion gauge is shown in the recommended inlet-down position to prevent problems with detritus dropping into the gauge. The diagram below is pictorial, there are a standard set of schematic symbols for vacuum systems which will also be included here eventually.
![]() |
The system shown here is typically used for vacuum evaporation, etc. This guide assumes that everything is off initially and that the belljar is at atmospheric pressure to start with.
Be sure you have liquid Nitrogen available before starting this type of system, and that you have some familiarity with proper LN2 handling techniques. You should wear eye protection, a long-sleeved lab coat, and protective gloves. Frostbite is painful!
Switch power on to the gauge controllers but don't try starting the ion gauge yet.
Turn on the water cooling lines to the diffusion pump and the water cooled baffle. Verify water flow before proceeding further.
Before going too much further, I would like to add a little sermon on water cooling lines.
There have been several floods in this area due to plastic water lines coming off of equipment left running overnight. Soaking someone's precious collection of books or having water get into electrical conduit tends to create a lot more waves than the relatively small expense of taking some precautions.
A water flow switch controlling a water inlet solenoid will prevent problems and insure your getting invited to the next bush party or whatever. All of our lab systems are plumbed in with soft copper tubing, have flow interlocks, and haven't had flooding problems in years.
All of the vacuum valves in the drawing should be closed to start with. The pump vent valve may be as shown, or in some cases it is part of the rotary vacuum pump and closes automatically when the pump is running.
Switch the power on to the rotary pump. The pump will be quite noisy and spew oil vapour out it's exhaust when starting from atmospheric pressure. Don't be alarmed, that's normal operation and the noise and vapour should subside as the vacuum improves.
If the racket and oil vapour spewing persist more than a few minutes (Or longer for larger systems.) it's quite likely that you have a gross leak or major contamination and you will have to shut the pump down and investigate.
Open the foreline to the diffusion pump and wait until the pressure on thermocouple gauge 1 is below .15 mbars.(200 millitorr.) Switch the diffusion pump heater on and wait for a minimum 1/2 hour. Larger pumps often take longer, in fact just about all diffusion pumps take about an hour to really get going but may be usable in less time with less performance.
Fill the cold trap slowly with liquid Nitrogen. The LN2 will boil like mad initially as it hits the warm surfaces inside the cold trap. Be patient and just partially fill the trap initially to allow it to cool slowly.
The important thing to remember at this point is to never let a hot diffusion pump see high or atmospheric pressure, a situation that will crack many diffusion pump oils and create a horrible mess. If you break it you get to fix it and it may not be pretty!
Close the foreline valve to the diffusion pump. Make sure the vent valve to the vacuum chamber is closed, then open the vacuum chamber roughing line.
The rotary pump will get noisy again, the noise should subside in a few minutes and the pressure on thermocouple gauge 2 should begin to drop. Keep an eye also on the foreline pressure. If pressure rises above about .15 millibarrs (200 millitorr) you will have to pause by closing the chamber roughing valve, opening the foreline valve, and re-pumping the foreline. Close the foreline valve and re-open the chamber roughing valve when everything looks ok. Skip the next paragraph unless there is a problem.
If you can't get the chamber down to .15 mbar (200 millitorr) or less in about fifteen minutes there is most likely a leak or gross outgassing in the vacuum chamber. Close the roughing valve, re-open the foreline valve, and arrange to leak test or clean the vacuum chamber. Do not pass GO, STOP HERE.
Rough pump the vacuum chamber down to about .15 mbar (200 millitorr), then close the roughing valve. Note that there is no great advantage in going much below this pressure as the oil backstreaming rate for most mechanical pumps gets progressively worse below the viscous flow region and can cause some system contamination. One school of thought also considers it a Good Thing to always start a diffusion pump with the chamber pressure slightly higher than the foreline pressure to reduce diffusion pump backstreaming at start-up.
Open the foreline valve then open the high vacuum (a.k.a. baffle or gate) valve. Wait for a few minutes then start the ion gauge. If ion gauge pressure is above 1.3 X 10e-05 mbar (1X10e-05 torr) a good practice is to switch the gauge off and just make spot checks until below 1.3e-05 mbar to prevent oxidation of the hot gauge elements. Gauge outgassing should always be done at 1.3e-05 mbar or less with pauses to let the evolved gases be pumped away.
Congratulations, you're in business! Have patience. Reaching ultimate pressure may take several hours.
Top up the liquid Nitrogen trap, check it occasionally, and proceed with whatever it was that you wanted to do in vacuum.
Warning: some systems may use argon as a backfill gas, your thermal heat loss (e.g. thermocouple, Pirani) gauges will indicate much lower pressures than actual pressures, so please read and use the gauge manufacturer's correction charts.
Running systems continuously to prevent the ingestion of water vapour and to prevent corrosion of oil-sealed rotary pumps is good vacuum practice and should be done if possible. Some means of keeping cold traps filled should be used on an untended system.
Backstreaming from an untrapped diffusion pump will contaminate your vacuum chamber in many cases, so it's best to close the high-vacuum valve, shut the pump heater off, and just keep the rough pump on the diffusion pump foreline. Alternatively a good water cooled chevron baffle and a diffusion pump fluid selected for low backstreaming may allow continuous operation with no liquid nitrogen. We've had good success with polyphenyl ether but it is very expensive.
Close all the valves on the vacuum system and shut off heater power to the diffusion pump.
Whether or not to switch an unused pump off will be your decision. Leaving the pump running continuously is the preferred option since any water vapour, etc. that has been pumped will remain in the pump and cause instant corrosion. If your application is relatively clean, as most of ours are, then switch off. For dirtier applications, an oil change is recommended before storage.
Techniques shown here are generally used for the operation of diffusion pumped systems at this site. Your mileage may vary.