I'll begin by saying help! I've used diffusion pumps only a little and I'm sure someone else here can say more and better than I am now, so please do.
I am going to assume you want to work at plasma pressures here -- say nothing below e-3 millibar. You want purity, so you want to be able to have a base pressure of 2-3 orders of magnitude below that to get all contaminants down to less than a percent or so. For this, a diffusion pump will do. There are some problems and issues, but they work and are reliable. For a small surface area, nearly sealed off system, most diffusion pumps are actually "too good" and need some kind of throttling to keep the gas flow at the desired pressure low enough that you're not wasting too much process gas. They also have a nasty tendency to backstream oil into your apparatus, so you have to apply them more carefully than you do a turbo. Further, an inrush accident can dump your nice, expensive diff pump oil into the backing pump, which costs you having to change the oil in both -- diff pump oil isn't a good lubricant in most cases.
There are two effective ways to throttle a diff pump. One is simply by putting a valve between it and the tank -- good move as this also cuts backstreaming down. The other is to heat it less, via a variac or lamp dimmer in the heater circuit which helps as most valves are a little twitchy near the fully closed position you're going to probably wind up using without some heat reduction.
Most surplus diffusion pumps are simply too big for the small science worker. You want probably 3" throat or less for a system that fits in a small lab, and if you can find one, an air cooled one is by far more convenient than a water cooled one. If you're not finding what you want surplus, keep looking as having the wrong pump for the job is more hassle and more expense by a lot.
Backstreaming of oil vapor into the tank can be a real problem but this depends more on what you are doing there. If you are making low energy charged particle optics, it's pure poison.
If on the other hand you're running a fusor, not so much -- it decomposes into carbon and hydrogen, with the carbon getting on the hot stuff, but not affecting how things work much
at the high currents and voltages involved. The ways to stop it are various traps, a "cold top" which is kept at room temperature or below, and by using a low back-streaming oil. Also a simply labyrinth trap will help lots if you remember to keep cleaning it out, which can be tough with some oils that cling tightly and don't respond to the normal solvents well. Nevertheless, you have to clean them too, as they "creep" along surfaces until they find a place to make trouble.
A further issue is that most oils if exposed to air when hot, either catch on fire or are decomposed into lighter fractions, ruining the oil, so with any diff pump system you'll be wanting to have a few more valves, or have to be really patient in letting things cool off before allowing air into the diff pump. Here's a picture of an old style experimental system that is valid today, except for one thing - most newer pumps are already multistage, so you'll only need one of them. Note that with this system, you can let your main tank up to STP while keeping good vacuum in the diffusion pumps(s) and leaving them hot. You can then use the forepump to rough pump the chamber, open all the right valves, and get going again quickly. There's another little trick in this picture I recommend highly, and that's having a discharge tube in the foreline someplace as a poor mans mass spectrometer. With a little hand held optical spectrometer and some practice, you can find out quickly what your gas composition looks like on a rough scale -- often all you need to know. Also helps in leak hunting when you are doing that. You simply use a leak probe gas or fluid that gives a different color glow discharge. Click the pic for a readable version.
The fusor we built for Bill Fain here uses almost exactly this design, it's a good one, except we used some good ball valves instead of the hose clamps for those parts. This really isn't rocket science, and the system works great.
Many agonize over the choice of diff pump oil. We kept it simple, just using Lesker's Diffoil-20 for both the diff pump and the backing pump, as it's rated for both. We never did put a really good ion gage on that system, but it gets well below where a fusor runs, and seems to have no troubles due to the cheap oil we used. His system is water cooled, which due to a lack of plumbing in my lab is done via a 6 gallon bucket of water, a fountain pump, and a bunch of "blue ice" so we can run long enough for it to be useful. It has a cold-top trap which sees the water while it's the coldest. We don't have evidence that backstreaming oil is a problem, but we are careful running it. This diff pump is also on a lamp dimmer. We run it full blast for about 10 minutes at initial pumpdown, then set it to about 1/3 power, which it cools down to in another 10 minutes or so and becomes stable there. We recently acquired a smaller air cooled diff pump and will be retrofitting that one to get rid of the inconvenient water system and save that space on the cart.
There are more expensive oils out there for special purposes and a lot of the fusor crowd swear by the DC-705 series, currently listed at $315 for half a liter at Lesker. It must work as they are getting good results. At the extreme you get into some very good but also very expensive oils that need no traps and backstream very little, for things like table top electron microscopes -- a very picky application compared to plasma kinds of things, and there is no need for that for most people. Lucky as it can cost half a grand for one fill of that stuff shown here. Unless you need that and know why, you don't need that. Things like low current low voltage electron microscopes or quadrupole mass spectrometers are really trashed by a thin layer of insulating oil on the electrodes. Glow discharge electrodes couldn't care less, they just burn it off.
With anything other than the cheap Diffoil-20, you have to begin to worry about cross contamination of oils. If you get too much diff pump oil in the mech pump, it will wear out quick. If you contaminate the other way, then your diff pump will stop working well and backstream more. Did I mention that it's a real PAIN to clean oil contamination off things, particularly silicone oil? A quick price comparison and hassle comparison will show most why we landed with what we did. But I am also sure that someone else knows more than I. I've run Bill's system, which I built, and some systems way back when I was in college (early '70s) and that's about what my experience is. We have seen many of the more successful plasma workers using the DC class pump oils in the air cooled pumps with great success. We here are lucky enough to have a couple of oil free turbo systems, and have basically never looked back. Sadly, the good turbo-drag pumps aren't hitting surplus yet, so if you're on a budget, this is probably what you'll wind up doing -- a mech pump/diff pump system.