When processing gas laser tubes it is common practice to overfill the tube then reduce the pressure to the target value using a valve. A problem arises in a multi-gas environment (e.g. a mixture of helium and neon gases) since these gases are pumped at different rates by most pumps. In this example, a turbomolecular pump is used to adjust the pressure of a mixture of gases.
During the lab the manifold was mixed with an approximate 2:1 ratio of helium-to-neon as evident from the red trace in the RGA scan below. The manifold was initially filled to a total of 30 torr (approx. 20 torr helium, 10 torr neon). Analysis shows the actual partial pressures in the over-pressured manifold (as seen on an RGA which samples the manifold gases through a needle valve) are 2.65*10-6 torr of helium and 1.15*10-6 torr of neon for a ratio of 2.3:1. Note that this is actually the ratio of Helium:Neon-20 since the two pressures used were at 4 and 20 amu respectively. The actual ratio of total neon would require the addition of the partial pressures of neon-20 and neon-22.
Now the manifold pressure must be reduced by leaking gas from the manifold to the turbo pump (which is on the 'high vacuum' side of the system and at very low pressure). The high conductance valve is opened, gradually, and pressure is seen to decrease on the Baratron gauge on the manifold. After reduction of the total manifold pressure to 2.0 torr (required for the operation of the helium-neon laser tube attached to the manifold) the ratio of the component gases is quite different than what we started with as evident from the blue trace. The partial pressures are now 2.5*10-6 torr of helium and 3.0*10-7 torr of neon for a ratio of 8.3:1 ... precisely what is required for our laser tube (which, as you will have read in the background research for the lab, is somewhere between 7:1 and 10:1).
So what happened?? Opening the high-conductance valve connects, at least temporarily, the manifold directly to the turbomolecular pump which reduces pressure in the manifold however most pump systems remove gases at different rates depending on the gas employed. Logically, helium pumps much slower than neon since it is a smaller atom so the pump extracts neon more efficiently from the manifold. This, in essence, concentrates the proportion of helium in the manifold! Although selective pumping of different atoms accomplishes the desired effect, i.e. providing a 10:1 ratio, this is haphazard at best.
Although the rate of pumping (in L/s) is basically the same regardless of the gas pumped, the compression ratio (the ratio of intake to output pressure for a pump) varies wildly depending on the gas being pumped. A small turbomolecular pump, for example, might have a compression ratio of 4*108 for Nitrogen but only 630 for Hydrogen (figures quoted from Lesker's Tech Notes, below). Assuming the foreline had a partial-pressure of nitrogen of 1*10-1 torr, the partial pressure of nitrogen at the intake (our manifold) could be as low as 2.5*10-10 torr (a factor of 4*108 times lower)! However, if that same foreline had a hydrogen pressure of 1*10-1 torr the manifold would have a hydrogen partial pressure of 1.6*10-4 torr. The reason for the discrepancy in compression ratio is the velocity at which atoms of hydrogen and nitrogen travel ... light hydrogen atoms at room temperature can travel much faster than heavier nitrogen atoms ... so fast that a good number of them can pass backwards through the spinning rotor blades of a turbo pump, back into the manifold!
In our case, helium is the culprit. Opening the high-conductance valve floods the pump with helium and the partial pressure of this gas increases rapidly in the foreline (the gas mix is, after all, primarily helium). Since the compression ratio is less for helium than for heavier neon atoms, the pump effectively moves less helium atoms than neon atoms and so "preferentially" pumps neon changing the ratio of helium-to-neon in the manifold. For more information on turbomolecular pumps and applications follow this link to Lesker Tech Notes.
How can this be avoided (so that the gas mixture does not change)? Assuming we overfilled the tube (initially with the correct 7:1 or 10:1 ratio of He:Ne gases as required for the experiment) we could use the sampling needle valve to reduce pressures in the system instead of the high-conductance evacuate valve. This allows pressures at the pump head to stay in the 10-5 to 10-6 region keeping foreline partial pressures of helium low as well so that the actual odds of pumping a helium atom or neon atom entering the system through the tiny opening are more-or-less equal. This is, of course, a slow process since the needle valve is designed to be a controlled leak (it can take several minutes to reduce the manifold from 30 torr to 2 torr). Perhaps the best method would be to evacuate the manifold and CAREFULLY pressurize it to the exact pressure required the first time. For example, for a 2.0 torr 10:1 mix, neon would be released into the manifold first to a pressure of 0.2 torr then helium injected until the total pressure was 2.2 torr. This might be slightly higher than required so the sampling valve could then be used to reduce the pressure to the desired target value. Keeping the flow very low (by opening the valve only slightly) will help maintain the correct ratio. And, of course, the RGA should be used to verify the exact proportions of helium-to-neon by ratioing the pressures of the peaks at amu=4 and amu=20.