TL;DR: A liquid is circulated between a high vacuum chamber, where it is chilled to dissolve gas molecules, and a low vacuum chamber, where it is heated to drive the molecules out. This should result in the high vacuum chamber being highly evacuated.
(This has been on my list for years, but [Frankx]'s closely related CO2 direct air capture idea [link] reminded me to write it up and post it.)
This is a type of high vacuum pump intended to lower costs by not requiring much specialized hardware. It does require a specialized liquid medium, though, which might be expensive. Anyway, it consists of a circuit for pumping this liquid (the solvent) around, where the liquid is exposed to the surroundings in two places: inside the high vacuum chamber and inside the low vacuum chamber. The low vacuum chamber is evacuated by a conventional backing pump like a sliding-vane vacuum pump (which is already inexpensive, and can easily be bought used). The differential solubility pump is therefore a substitute for other high vacuum pumps like the turbomolecular pump and the diffusion pump.
Anyway, as I said above, we (the general we) have (hypothetically) a low vacuum chamber, and a high vacuum chamber. Initially, both are at low vacuum. The differential solubility pump has a part of itself in each of the two chambers; these parts are largely similar or even identical and consist of trays for the solvent to have a high surface area exposed to the chamber, as well as heat exchangers or phase change condenser/expander units. The two trays are connected to each other by two pipes, so that the solvent can flow continuously through both trays in a loop. The pipes each have a positive- displacement hydraulic pump/motor; these are geared to each other and driven together by a motor. This ensures that whenever solvent goes one way through one pipe, an equal amount of solvent must go the other way through the other pipe. This prevents the solvent being sucked into one chamber that has a lower air pressure in it. (The requirements on that system are not great, though, because there's not much difference in air pressure between low and high vacuum, even though there is a large difference in vacuum. This is because pressure and vacuum are inversely related, and inverse relationships are nonlinear.)
The heat exchangers (in that version) are connected to a heat pump, or the condenser and expander (in that version) are connected to a refrigerant compressor to make a heat pump out of them. This is to make the solvent in the tray in the high vacuum chamber cold, and the solvent in the tray in the low vacuum chamber hot. (Additional solvent-to-solvent heat exchangers may be added to the solvent pipes at the inlet/outlet of each tray, to conserve heat/cold in the tray and thus increase efficiency.) The result of this is that gas molecules that hit the cold solvent in the high vacuum chamber are likely to be dissolved in it, and gas molecules dissolved in the solvent are driven out of it in the low vacuum chamber. With the solvent being continuously pumped between the two chambers, this results in transport of gas molecules from the high vacuum chamber to the low vacuum chamber, producing the high vacuum.
To save space, this pump could also be made in a form that doesn't need a low vacuum chamber, just a backing pump, by substituting a (coiled?) long pipe where the solvent and low vacuum can come into contact with each other for the low vacuum chamber and tray.
The solvent has to be some liquid with an extremely low vapor pressure, and that can dissolve lots of different gas molecules. Vacuum pump oils meet the first requirement, but I don't know how well they dissolve gases. In sliding-vane pumps, it causes problems when the oil dissolves process gas, which is why a sliding-vane vacuum pump will usually have a "gas ballast valve" that just lets in a bit of air to flush out the dissolved process gas (at the expense of achieving a worse vacuum on the suction side), which implies that they do dissolve at least some gases at least somewhat. Alternatively, an ionic liquid may be applicable. Either way, though, this pump will probably not pump all gases equally well, which is a limitation that applies to the current options as well—e.g. virtually all vacuum pumps are bad at evacuating helium.
If you can find a liquid solvent that will dissolve the gases you want to evacuate from your high vacuum chamber, but won't dissolve some other gas, you might be able to use that gas as an eluent to carry away the evaporated process gases from the warm side. In that case, you wouldn't need a low vacuum backing pump at all, just a supply of eluent gas free of any gases soluble in the solvent.
69/480 [between 2017-01-08 and late April 2017]-- notexactly, Oct 02 2019 CO2 direct air capture by [Frankx]. Closely related, but not inspiratory for this, because I had this idea two and a half years ago. It just reminded me that it was on my list. [notexactly, Oct 02 2019] It's an interesting idea, but as you point out the critical factor is the solvent.
It needs to have extremely low volatility (like diff pump oil) in high vacuum, yet have an affinity for gases; that suggests a polar molecule.
How well would this pump noble gases ?
What's the advantage over a sorbtion pump, or a sputter ion pump ? Ion pumps are pretty compact, have no moving parts, and deliver high vacuum.-- 8th of 7, Oct 02 2019 It would pump noble gases more or less well depending on their solubilities, and temperature coefficients of solubility, in your chosen solvent. I don't know if there would actually be any solvent good for noble gases, though.
You could use a sorption pump as a roughing pump before turning this one on (though you couldn't use it as the backing pump to continuously evacuate the low vacuum chamber, because it will saturate). I expect this will achieve better ultimate vacuum than a sorption pump.
Versus an ion pump, I expect you won't need as high a vacuum before you turn it on, and it's continuously disposing of the process gas molecules rather than just trapping them in itself, so it should never saturate in an absolute sense. But I expect it won't produce as much ultimate vacuum either. I think you could use it as a backing/roughing pump for an ion pump. Wikipedia says of ion pumps:
// Both the pumping rate and capacity of such capture methods are dependent on the specific gas species being collected and the cathode material absorbing it. Some species, such as carbon monoxide, will chemically bind to the surface of a cathode material. Others, such as hydrogen, will diffuse into the metallic structure. In the former example, the pump rate can drop as the cathode material becomes coated. In the latter, the rate remains fixed by the rate at which the hydrogen diffuses. //
In this pump, the solvent doesn't have a constant surface to get saturated by CO. Of course, its volume can still get saturated if the gas can't be driven out fast enough on the exhaust side, but this saturation will take a lot longer due to volumes being bigger than surfaces, which could make the difference when pumping down or taking up a transient release of gas from the experiment. On the other hand, hydrogen should diffuse through it (and out of it on the exhaust side) much faster than through a solid metal, though I hope not so fast that it can't be effectively pumped from one tray to the other.-- notexactly, Oct 03 2019 random, halfbakery