At first sight, pumping xenon seems like the ultimate in non-issues. First, apart from lighthouses, movie houses, Runway End Identification Lights (REILs), and strobes on top of some Pennsylvania school buses, there are not many applications requiring xenon pumping. Second, since it is a heavy, inert gas with very few reaction possibilities, any old pumping mechanism that works for argon is OK for xenon, right?. So, apart from xenon's huge price, what's your problem?
It turns out that compared to argon, xenon has some interesting properties that make pumping it, well — interesting.
|Atomic Weight (12C = 12.00)||39.95||131.3|
|Melting Point||83.8 K||164 K|
|Boiling Point||87.3 K||165.1 K|
|Enthalpy of Fusion||1.18 kJ.mol-1||2.30 kJ.mol-1|
|Enthalpy of Vaporization||6.5 kJ.mol-1||12.64 kJ.mol-1|
|Thermal Conductivity||0.01772 W.m-1.K-1||0.00565 W.m-1.K-1|
Turbo Pumps and Xenon
- Note xenon's atomic weight is over three times that of argon. From this we can predict two consequences: (a) the most probably atomic velocity of xenon will be much slower than argon. (b) in a turbo pump xenon's compression ratio will be higher than argon's. We can jump from (a) to (b) using a very simplistic argument. If xenon travels much slower than argon, then it can't flow backwards through the pump as successfully. The lumbering xenon atom won't clear the thickness of one blade set before the next blade swings around and swats it back into the higher pressure region.
- From published results for a turbo pump in Lafferty's Foundations of Vacuum Science and Technology, one can plot compression ratio (CR) against molecular (atomic) weight for normal gases and extrapolate to find xenon's CR is ~9 x 109 compared to argon which is 2 x 109.
That doesn't sound like a huge increase but applying a little (please excuse the bad language) thermodynamics leads to an interesting conclusion. The background, which you can ignore, is the equation for the Joule-Thompson effect for an ideal gas which says the final temperature (T2) in an adiabatic compression, is given by: T2/T1 = (p2/p1)(γ - 1)/γ
where p2 is the final pressure and γ is the ratio of the specific heats. Well, plugging in that ~4x higher final pressure means that xenon's final temperature is 40.3997 ( ~1.74) higher than argon's.
- Having a gas at almost twice the “normal” temperature in the turbo pump's final stages is not my idea of a barrel of laughs. And, without knowing why, I worry about that very low thermal conductivity too.
- But does all this really mean anything? Yes, indeed! I heard the story from a research facility that destroyed three turbos pumping xenon before they convinced the pump manufacturer that this might not be an “operator problem”. The manufacturer investigated, promptly blew up another pump, and became instant believers. I wasn't told the type of manufacturer or the damage. But the thermodynamics applies to any turbo and at 'typical' pump prices, why risk repeating this fiasco and being labeled by your colleagues (depending on your country of origin)-- jack-ass; wally; dummkopf; or a kangaroo short in the paddock?
Cryo Pumps and Xenon
- To re-quote one of my favorite cautions - a material's melting point tells you nothing about its vapor pressure. (If you find this difficult to believe, check out the vapor pressure curves for gallium/gadolinium, or aluminum/magnesium.) However, a material's boiling point speaks volumes. Xenon boils (i.e. has a vapor pressure of 760 torr) at 165.1 K. From the vapor pressure curves produced by Honig and Hook, 1960, RCA, at 80 K, xenon's VP is ~5 x 10-3 torr and at 40 K, xenon's VP is ~1 x 10-12 torr.
- These temperature/vapor pressure points are significant. The 'typical' cryo pump has two temperature regions, a 60 - 80 K shroud and louvers surrounding a 10 K “core”. If xenon's pressure in a cryo-pumped chamber exceeds ~5 x 10-3 torr, it will mostly condense and freeze on the louvers. Yes, some xenon will evaporate from the louvre's and freeze on the core structure but we get to that later.
- So, what' the harm in xenon condensing on the louvers? None at all, until you decide to stop the flow of xenon and pump the chamber to its base pressure. Assuming the base is below 5 x 10-3 torr, the louvers then act as a secondary outgassing source that may continue well into the middle of next week. No, the pump hasn't “exceeded its capacity” to pump xenon. The xenon is simply in the wrong place.
- One cryo pump manufacturer recommends selecting a louver temperature (which you can with their pumps) so xenon's VP (at that temperature) is higher than the process pressure. No xenon condenses on the louvers — only on the core. Pumping from process to base pressure is then swift.
- That's OK if the process pressure is, say, 1 x 10-4 torr since the louver temperature only needs to be slightly higher than 68 K and nothing will freeze out on it. But what if the process pressure is 1 torr? The appropriate louver/shroud temperature is then 104 K and one wonders if that is too high to maintain the core at ~10 K?
- Well, one comment is: these conditions can't exist. At 1 torr we would certainly have a throttle valve between process and pump, if only to reduce the otherwise enormous gas throughput. The pressure between valve and pump might easily be 1 x 10-4 torr which implies louvers slightly above 68 K are fine.
- However, what happens when the 1 torr process is finished and we want the chamber at base pressure. Well, we crank open the valve and — whoops! Xenon snow covers the louvers, the chamber pegs at 1 x 10-4 torr, and there's no sign of any cavalry riding hard to rescue you.
- Worse yet, look the enthalpies (formerly called latent heats) of fusion and vaporization in the table and see xenon's values are double argon's values. When the valve opens between chamber and cryo pump, suddenly 50 L of xenon at 1 torr is dumped into the pump. Let's say it takes 1 second to become snow (which isn't impossible). Popular cryo pumps accept a maximum argon throughput of 10 to 20 torr.L/sec. Here, we are dumping 50 torr.L/sec of xenon which, with the high enthalpies, means we are depositing between 5 and 10 times more watts than the pump's compressor can handle. The temperature will spiral out of control and all previously deposited gas will be released, which could be very interesting if the pump already has, say, 500 standard liters of argon in the core. Whoa, Nelly! Sounds like a chamber pressure of 10 atmospheres to me. Hope that burst disc is functioning.