We make diffusion pumps – an antique but robust and effective route to high vacuum, especially in little and very large systems. Between the diff pump and a cryotrap, there’s a great deal of energy expended. A 20″ pump may use 15kW, with nearly all that power going into the cooling water, which is another story, since almost everybody is forced to use chillers. Cryo pumps also use a great deal of power, and, like diff pumps, can’t be turned on very quickly, so they’re generally left on all the time. They need a recirculating chiller to function as well. (Or, if air-cooled, the same burden thrown on the building’s HVAC – murderously expensive in clean rooms.) Turbomolecular pumps ramp up quickly, and use tens of watts instead of thousands, but in many applications, either cost or chemical considerations rule these out. Then, there’s liquid nitrogen, an increasingly expensive cryo trap filler.
For some applications, such as semiconductor fab or heat treating furnaces, there’s almost no choice: you use whatever technology is traditional and is known to work, regardless of energy consumption. For others, such as the R&D tools we make, there are more options. Before specifying a system, the user needs to look at the following:
a. How often, and how long, will the system be in use? There are many small systems that are actually in use at most a few hours a week, preparing a handful of samples. It would make no sense at all to have something – anything – running all the rest of the time, both from energy and wear considerations. Incidentally, it also doesn’t make sense to consider repair intervals or longevity – the system components may never wear out.
b. If the system is shut down most of the time, and is started to make a run, how long is a reasonable startup? A diff pump will hit stride in about half an hour, a cryo, two hours, a turbo, minutes.
c. From a contamination standpoint, is it ok to have a system not pumping? Forget semi device fab- there’s no question about diff pumps – you just can’t use them. In other applications, there will be startup transients that can have an effect on the process. Cautious researchers tend to leave everything running all the time, possibly because they never see the electricity bill as a budget item. Cold traps are the real issue in this question.
d. If a system needs cooling water, where does it come from? In the old days, this was tap water, wasted into the drain. States, institutions, and companies now have rules that do not allow this. You may have a building-wide chilled water supply that you just tap – at your peril if there’s much gunk in the lines. You may have to install a refrigerated chiller – a gross energy hog, and an expensive component to boot. We’ve found a couple of inexpensive, low-operating cost options that help in this area. By the way, if you’re cooling sputter magnetrons, there are several other vital considerations.
e. What, exactly, does your process require? We have maybe 4000 little diff-pumped systems out there, very few, if any, ever get changed to a different system, but it’s certain that some of these are inhibiting successful work because a mis-operated diff pump is just an oil coating mechanism. We don’t mention all of those mercury diff pumps still in use. If you need UHV, then an ion pump is not only worthwhile, but a very economical choice, especially since there’s no continuously running forepump.
Pumping and trapping systems are usually the largest factors in the energy budget of a vacuum system; process hardware, except in furnaces, uses very little. It pays to consider energy on the same page as initial cost and process needs whenever you have the luxury of designing or specifying a system.
Today’s Elephant Tip: In a multi-user environment, such as a university lab, consider the tradeoffs between user-training and equipment fragility. A turbo pump may be easiest to comprehend, and easiest to automate, but if there’s any chance for, e.g. a student, to vent one to air at speed, or drop some hardware down the throat, and you don’t have unlimited maintenance budget, consider something else. Along these lines, our customers will ask for rotary vane oil bath forepumps instead of dry pumps of the various kinds. Some of them even prefer belt-driven (!) rotary pumps as these are hard to kill, and easy to overhaul. In a semi fab, however, tossing a $75k dry pump every three months is just fine.
