LEAK DETECTION OF QUADRUPOLE GC/MS VACUUM SYSTEMS
"Nature abhorrers a vacuous experimenter: Campbell's Law (Apologies to Francois Rabelais)
A mass spectrometer is probably the best leak detector ever devised and people who have never worked with one have no idea of the meaning of the term "leak free". This can lead to exasperating problems but the instrument itself is also admirably suited to help you track down the leak.
WHY A VACUUM SYSTEM ANYWAY ?
It is worth reviewing why mass spectrometers are operated in a vacuum chamber.
Since a mass spectrometer separates ions by mass it would seem that unless you were interested in one of the normal components of air you would have no fundamental reason to worry about air at all. This is indeed true but air (or any other gas) also affects the distance an ion is likely to travel before it encounters a gas molecule and is lost (Mean Free Path). If this mean free path is not at least as long as your instrument you will have very little chance sorting or detecting your ions (See Table 1). This is the primary reason mass spectrometers are operated in vacuum. There are lots of secondary reasons:
--electron emitting filaments in the ionizer also have mean free path constraints and these are directly related to the service life of the filaments. (As you have no doubt found out if you have ever vented your system with the filaments on.)
--The ionization process can change with ionizer pressure yielding C.I. like fragmentation patterns instead of E.I. patterns.
--There are often high voltages employed in the operation of the mass spectrometer which are conveniently insulated by vacuum.
--The "background" of things always in the spectrum can be kept to a minimum.
--Most detectors also have mean free path constraints and some are not compatible with prolonged exposure to air.
Notice that it is primarily a matter of the density of molecules per unit volume and not the air itself that concerns us. The average distance an ion or molecule will travel before it hits another ion or molecule is called MEAN FREE PATH.
Table 1 Pressure: MM HG (Torr) Density: Molecules per cc Mean Free Path: CM 760 10+19 10-4 1 10+16 10-3 10-1 10+15 10-2 10-2 10+14 10-1 10-3 10+13 1 10-4 10+12 10+1 10-6 10+10 10+3 10-9 10+7 10+6 Also from table one we can see that at typical MS pressures any given gas molecule is much more likely to hit the vacuum chamber wall or enter the pump than to hit another molecule. This implies that there is very little "up stream" or "down stream" in the vacuum chamber. Just because a leak is closer to the pump than to the ionizer does not mean the leak will be pumped away "because it can't go up stream". Up stream and down stream apply only at pressures (above about 10-2 torr) where forces between molecules dominate the behavior of gases and result in so called viscous flow.
HOW MUCH AIR IS TOO MUCH?
As you can easily prove by looking at the mass spectrum between m/z 15 - 35 most analytical mass spectrometers have some air in them. The major components being Nitrogen, m/z 28 and Oxygen, m/z 32. You will also see water at m/z 18.
The ratio of these three peaks to one another are very instructive. The ratio of N2 to O2 in atmospheric air is about 4 to 1. If you see this ratio between the m/z 28 and m/z 32 peaks (E.I. mode) this means that the air in the vacuum system is being replenished from the atmosphere and you have a leak. This does not mean that you must drop everything and find the leak. Look at the m/z 18 peak (H2O). It should be the largest peak in this part of the spectrum. If the m/z 28 peak is about one half the intensity of the m/z 18 peak the leak is insignificant for the vast majority of analyses. Indeed you can have m/z 28 equal to m/z 18 and still do good E.I. work, though you will have problems with C.I. When m/z 28 is larger than m/z 18 it is time to start leak checking.
You should note that a vacuum system that has been open to atmosphere for several hours will adsorb a significant amount of water. This will show up in the spectrum when the system is first pumped down and may mask a small leak. Using a clean DRY inert gas to bring the system to atmospheric pressure will minimize this and is good vacuum practice in any case.
GAS CHROMATOGRAPH LEAKS
The first problem in finding your leak is to decide if it is somewhere in the vacuum system or whether the air is coming in with the GC carrier gas. A leak in the carrier gas plumbing will let air in as well as carrier gas out.
Fortunately there is a quick and easy way to prove the air is coming in with the carrier gas since the air peaks will react to changes in column flow. Get a display of the m/z 28 and m/z 32 peaks and increase the carrier gas flow or head pressure by about a factor of 2. If you have a leak in the injector area or septum not much will happen at first but after about a minute (depending on the column) these peaks will increase dramatically and then fall to a level below their starting point. This will be most dramatic with a capillary column coupled directly to the ionizer. What happens is this. The increase in flow rate pushes the slug of carrier gas already in the column into the ionizer at a faster rate so the peaks initially go UP. Meanwhile the higher pressure in the GC pneumatics, injector and column means more carrier gas leaks out but less air leaks in. Hence the peaks eventually come down to a level below there starting point. The vast majority of GC leaks are in the septum, injector, interface connections, or column connections.
If the leak is in the GC plumbing from the flow or pressure controller back to the tank the air peaks will increase with increased column flow but will not subsequently fall to a lower level. This area is not often disturbed and not prone to developing spontaneous new leaks. A leak here, in the high pressure plumbing, will generally make itself known by consuming large amounts of carrier gas. The best way to pinpoint a leak in this area is with some kind of liquid that can be applied to fittings to create bubbles. Isopropyl alcohol works well as does soapy water but be careful not to get liquids into the plumbing.
If you decide the leak is associated with the GC injectors or column the exact location can be pinpointed using a leak detecting probe gas as described for main vacuum system leaks. Remember that the mass spectrometer's response to a leak on the injector side of the column will be delayed by the transit time of the column. You can also, if the GC and ionizer are cold, disconnect the column from the MS, cap the exit of the column and pressurize the system. This will let you use the bubble leak detection technique.
Another sign that the air is coming in through the GC or the interface is that the pressure, measured by the ionization gauge, seems too low for the amount of air in the spectrum. If the m/z 28 peak is 100 times the size of the m/z 18 peak and the pressure is not much different from when things are tight, the air is being introduced directly into the ionization region. Since the ionizer was designed to give maximum response to material presented to it in this way the air shows up in the spectrum way out of proportion to its contribution to pressure. (This assumes that you pay attention to such 'unimportant' things as chamber pressure when things are working well. One of the many benefits of keeping an instrument log book is that you can look this kind of stuff up and be sure rather than guessing.)
MAIN VACUUM SYSTEM LEAKS
If neither of these tests indicts the GC or interface your leak is probably in the main vacuum system.
Here we can take a more direct approach. The essence of this technique is to try to introduce something into possible leak sites while looking at an m/z range where we would expect to see this 'something'. The optimum leak probe materials are gases with Helium and Hydrogen being the best. But Hydrogen is very flammable and Helium the usual carrier gas. As a practical matter Argon and CO2 are the best choice with an unlit propane torch as a backup. A lecture bottle of gas, a regulator, and a length of flexible tubing with a toggle valve at the end are invaluable. If you have access to dry ice you can put some in a polyethylene wash bottle. You can also use various liquid solvents but once they get into a leak, or worse an O-ring seal, they take what seems like forever to clear away.
Get a display of the mass range of interest 10 AMU wide with the peak in the center and about 1/2 the maximum display intensity. If you are using Argon as a leak probe gas sweep from 35 to 45 AMU and adjust the multiplier gain to give a peak 1/2 the total display height. The naturally occurring 1% Argon in the incoming air will make this a reasonable sized peak. Adjust your regulator to a few PSI and apply the gas, briefly, to possible leak sites. Watch the display as you do so. When you apply gas to the leak you will see a very dramatic increase in the peak height. If you have not hit upon the exact site of the leak but have come close to it you will probably see a slow (several minutes) rise in the peak height rather than a dramatic one.
Patience and a little common sense go a long way. Most leaks occur at frequently uncoupled joints immediately after they have been rejoined. The next most likely candidates are joints that are thermally stressed as in the GC or interface, particularly if new ferrules have recently been installed. Valves into the vacuum system are also suspect well as shafts or probes that may go through them. A welded in place electrical feed through or flange that has not been disturbed are very far down the list of likely leak sites.
If you can't pinpoint the leak using the gas probe technique the next step is to start capping things off. The idea is to reduce the vacuum system to its simplest configuration by capping off or removing as many probes, inlets, interfaces, and valves as possible. (You will of course need a collection of caps, plugs and blanks for the various types of plumbing in question.) You Then pump the system down and check the spectrum for air and, if all looks well, start putting things back in place one piece at a time. Checking the spectrum after you install each new piece. This sounds time consuming, and it can be, but it is better than having a useless system or one in which you have no confidence.
MASSIVE LEAKS
Occasionally you will develop a leak that is so large that you cannot get the system down to operating pressure or can only reach the ragged edge of the ionization gauges range (10-3 torr). This invariably happens when the system has just been put back together. You have probably forgotten to install a gasket, installed it wrong, not tightened a fitting, or not turned on all the pumps. (Anyone who cannot admit doing at least one of these things has not run a vacuum system very long.)
If you can get the ionization gauge running you can use it as a Helium leak detector. Reduce the GC carrier gas as much as possible. Apply Helium to likely leak sites while watching the ionization vacuum gauge readout. When you apply Helium to the leak you will see the pressure drop. This is because the ionization cross section of Helium is about 10% that of air so that the ionization gauge "lies" when it is reading the pressure of Helium.
FIXING THE LEAK
Once you have pinpointed your leak you must then decide how to fix it. In the case of compression fittings the first impulse is to simply tighten it further. If a modest amount of torque fixes the leak: fine. If this does not work however, resist the urge to simply continue tightening. No joint was designed to require a gorilla to tighten it. The problem is probably with a deformed or dirty sealing surface or improper parts. Ferrules made of soft materials and O-rings can become deformed or nicked. You are much better off dismantling the leaky joint and replacing parts as necessary than continually tightening, and eventually ruining, the fitting.
Flanges on the main vacuum chamber are usually sealed with copper or rubber gaskets. The problems here is almost always a gasket out of position or a flange bolted more tightly on one side than the other. If you decide to use a vacuum grease on rubber seals be very sparing. An ounce of grease should be a lifetime supply.
If you have pipe thread fittings on your system remember that you must use some type of thread sealant since these fittings will not seal at all with out it. "Teflon" tape is the best choice. Be sure to clean out any old tape or sealant before reassembly and to wrap the Teflon tape in the direction the fitting will be tightened in.
IN CONCLUSION
Your vacuum system may never become your best friend but there is no reason it should be your enemy either. At some point it will leak. By gathering materials and supplies ahead of time and approaching the problem with some subtlety rather than brute force, you can keep your down time to a minimum.
| - Home - | - Contact - | - Web Sites - | - MS Info - | - Book Reviews - | - Curiosities - |