Someone at Kurt J. Lesker Company who, for many years, earned his living elsewhere detecting leaks once remarked, “There are thousands of large leaks, hundreds of smaller leaks, tens of very small leaks, the occasional leak in the 10-8 torr/liter/sec range, and only one I've ever seen at 10-9 torr/liter/sec.” To that list our leak detector operator recently added a number in the 10-10 atm.cc/sec range. Each was in a thin stainless sheet “cup” destined for electrical feedthrough weld-lip mounts. Apparently, the vendor's die (used to press these cups) had “stretched” the metal a little too far in a specific area. The cups suddenly changed their destination to the recycled stainless bin. But even with such a blip in the statistics, if you have a leak, chances are good it will be large.
Another observation about detecting leaks in existing vacuum systems (or a system you are assembling), ranks “places most likely to leak” with the most likely first:
The general principles of vacuum leak detection are simple. You spray the atmosphere side of the leak with a gas or liquid that gives a different response to air on your “detection device.“ The gas (or vapor) diffuses through the leak, displacing the air, and the output of the detection device changes.
In practice, it is far from easy. Initially, all you know is the pressure is not as low as you expected or wanted.
In addition to a leak, that condition might arise from:
Even when you are certain the system leaks, other considerations complicate leak detection:
One detection device, the residual gas analyzer (RGA), overcomes most of these difficulties and gives an immediate indication of the leak's cause, no matter what its source. Unfortunately, most people feel (unjustly) overwhelmed by the prospect of interpreting an RGA's results, and would rather spend many (expensive) hours tracing leaks than install an RGA. For those people, we offer this basic guide to leak detection.
Fluids & Detectors
|Air||Ear or Stethoscope|
If the vacuum chamber has a good mechanical pump attached yet will not go below a few tens of torr, you should be able to hear a high pitched whistle associated from air rushing through the leak (Honest! That's no joke). The most likely place for a leak this size is a missing/chopped O-ring or a feedthrough insulator that has snapped. In one famous NASA example, a huge space simulation chamber with numerous 35-inch diameter diffusion pumps would not go below 10-4 torr range. The prospect of doing a helium leak check on a chamber perhaps 80-foot long an 12-foot diameter was a bit daunting. The noise level in this building was pretty high but an operator thought he heard a high pitched whistle that was different than on previous pumpdowns. So he wandered around this monster turning his head this way and that to get the “binaural”location effect. It worked— and he quickly honed in on a 1-and-a-half-inch port that had no flange!
For those wondering how a huge chamber, no matter how much pumping speed was attached, could reach 10-4 torr with a missing flange, look up choke flow sometime.
Leaks large enough to stop you switching on the ion gauge may sometimes be detected by squirting acetone from a plastic squash bottle on suspected parts and joints. Some question the wisdom of soaking O-rings in acetone, expecting them to dissolve but that's unlikely. More to the point, others wonder about the vapor permeating through the elastomer for the next decade! And they're right, but if this is the only leak checking equipment you have.
If the pressure reaches the millitorr range on a thermocouple or Pirani gauge, try helium as the spray gas. Often the question is asked, “When I find the leak, which way will the gauge move?” My usual answer is: “as long as it moves, who cares?” But, for Pirani and thermocouple gauges you can work it out. Helium and air have about the same relative response (see O'Hanlon A User's Guide to Vacuum Technology pages 86 and 87). Since helium diffuses faster than air, the pressure inside the chamber will be higher when helium is the dominate gas at the leak site. That is, more gas molecules will hit the sensing wire and remove heat. The indicated pressure should rise.
If you look carefully at O'Hanlon's diagrams, you will see that the difference between hydrogen's and air's response is much larger than helium/air (the indicated pressure being much higher than the actual pressure). Also hydrogen diffuses through a leak even faster than helium. I am not suggesting you spray hydrogen into your atmosphere, rather merely pointing to a scientific fact.
If you can reach the 10-4 torr range and get the ion gauge filament to stay on, then do so. The relative sensitivity of helium and air is large (see O'Hanlon page 92) but the wrong way round. That is, while helium diffuses faster (and therefore gives a higher chamber pressure than air), for identical pressures of helium and air, helium give less the 20 percent of the signal air gives. But, typically the gauge reading will change when helium hits the leak. (I'm just not into predicting the direction of change.)
A leak detector (LD) is a self-contained mass spectrometer tuned to the mass of helium. That is, an LD ignores air and other gases which have different atomic/molecular weights. The total pressure in the component under test is not of great importance since the LD has its own pumping system. But typically it helps if you can pull the component into the 10-4 torr range.
LDs are used extensively by vacuum manufacturers, but less often by system users. To test a chamber after welding, all ports are temporarily sealed with flanges, a flex hose is connected from chamber to the LD's inlet valve, and a rough pump used to reduce the chamber below a few hundred millitorr. The LD's inlet valve is cracked and the LD's pressure maintain within its working range. The outside of the chamber is first flooded with helium and, if the LD responds indicating a leak, then carefully check at every weld until the leak is located.
Some LDs have an audible alarm that changes pitch with varying helium level. This is a great idea when leak checking the rear of a large chamber. Leak rates down to 10-9 torr.liter/sec range are fairly easily measured. A few commercial models have sensitivities of 10-11 torr.liter/sec. But measurement in this range is not without problems including: noise on the signal, background helium, and, of course, one's experience.
Equipment & Methods
Helium Supply: You need a helium cylinder, high-pressure regulator, simple flow regulator (for milliliters/second flows), a long, small bore rubber tube that attaches to the flow regulator, and a piece of very small bore tubing (1/16-inch OD), which acts as the spray nozzle. If you are concerned about the world's resources, helium can be conserved (and over-spray reduced) by substituting a proper helium probe with valve in place of the 1/16-inch tube.
Helium Tent: Sometimes the system is so large that spraying helium on every part when you only suspect a leak is a huge inconvience. The helium tent method lets you check everything at once and, at least, confirm that a leak exists. Cover the entire equipment with large sheets of plastic film. The cheap, thin polythene painter's drop-cloths from a hardware store may be ideal. Drape the sheets so they touch the floor around the equipment and if you need two or more sheets, ensure the overlap is large. Then blast helium into the underside of the tent and watch the detection devices. Clearly, this is a very wasteful of a wonderful and useful resource (definitely non-renewable, unless you're into mega-giga-tera watt power generation using nuclear reactors).