I really enjoyed this paper, because it goes into detail on a technique that organic bench chemists the world over have all used at some point: “salting out”. I’ll go into some background for the nonchemists for a few paragraphs and then return to the paper itself, which all working organic chemists should have a look at (and they can, because it’s open access).
In the lab, we spend a good amount of time dealing with what looks like salad-dressing mixtures – layers of water and some other solvent. It’s not usually olive oil, as with Italian dressing, but it might as well be, chemically. The key point is that you have two liquids that don’t really mix with each other. Given time, they’ll settle out into top and bottom layers. Water, being as dense as it is, usually shows up on the bottom, but if you use dichloromethane, your water layer will be on the top, because that one’s even denser.
The point of having two layers is that different substances will dissolve in each of them. That, in fact, is where the famous-to-medicinal-chemists concept of “logP” comes from. Experimentally, it comes from shaking up a measured amount of a compound with equal volumes of water and octanol (a completely water-immiscible solvent) and seeing how much of it ends up in each layer. A compound with a logP of 3 has a 1000:1 ratio of going into the octanol layer versus the water layer; a logP of 1 is a 10:1 ratio, logP of -2 means it went 1:100 towards the water layer, and so on. To stick with salad dressing, analysis would show that any salt present is virtually 100% in the bottom (aqueous) layer, because table salt just flat doesn’t dissolve in olive oil. Meanwhile, some of the aromatic compounds that are found in the oregano, basil, and black pepper will have extracted almost entirely into the oil layer, because they have a lot of hydrocarbon character to their structures and don’t go into water very well at all.
And that’s why we chemists like the two-solvent trick, because such extractions let us quickly separate mixtures of compounds. The bulk of the compounds that medicinal chemists make, for example, have logP values that indicate that they’re far happier dissolving in octanol rather than water. And if you use a solvent that’s even less greasy than octanol but still won’t quite mix with water (ethyl acetate is a classic for this purpose), then the ratio is even higher. Anything with a logP of 3 is going to go into ethyl acetate over water much more than merely 1000:1 (update: or not! There’s experimental evidence that ethyl acetate is about the same as octanol for this purpose, although the data in this paper would suggest that ethyl acetate is indeed a better solvent for some compounds), and a logP of 3 is considered a perfectly reasonable level of “greasiness” for most drug compounds. So a quick “aqueous workup”, as the phrase goes, will clear out all the water-soluble stuff while leaving behind whatever’s in the organic-solvent layer – that’s where the classic lab item, the separatory funnel, comes in. It’s a convenient way to shake up the two layers, let them settle out again, and drain them off into separate flasks. Two or three rounds of that with fresh ethyl acetate each time and you’re ready to move on.
Now, salting out. That is a trick for those times that you’ve made something whose logP is heading down into the more polar range. If your desired substance shows some tendency to go into water, a quick sep-funnel workup may lose you some material, or it may also form a thick, cloudy emulsion that takes hours (days, years) to settle out again. A truly nasty emulsion is basically a milkshake in your sep funnel and is sometimes nearly that thick. Sometimes you have a compound or crude mixture that’s especially prone to that sort of thing, and that’s especially enjoyable if both layers are pitch-black. You’ll occasionally see organic chemists searching for a flashlight to shine through their sep funnels to see if they can spot two layers forming or not in such situations.
One way to deal with these problems is to add some salt to the aqueous layer. The basic principle behind this is pretty straightforward: hardly anything dissolves better in water than a classic ionic solid (like good ol’ table salt, sodium chloride). The interaction between the ions and the water molecules are so strong, in fact, that if you give the water a chance to form up around the salt components, it may well release its hold on your desired compound, which is probably not quite so desirable in comparison. When this works well, it can be downright dramatic – flooosh, the emulsion unravels in front of your eyes, or you actually see small liquid bubbles of your desired product forming up and rising out of the salty aqueous layer to the organic solvent above. Good times!
But the classic “throw some salt in there” technique is only the beginning. What’s so informative about the paper linked above is that it brings together a lot of lab lore about the different kinds of salting-out effects. There are many details and complications to that simple mechanistic picture in the paragraph above, and the paper is all about making use of just those factors. The authors, from the well-respected Merck process chemistry department, were working on a rather water-soluble nucleoside drug and needed better extraction procedures:
A cursory inspection of the organic and process chemistry primary and reference literature revealed a surprising absence of detailed information regarding the topic of salting-out extraction. However, a more thorough search through older literature and in journals considered out of field to organic chemists actually revealed a wealth of useful information on salting-out that is currently underappreciated.
That it is, and it’s a great service to see it brought together into one place. These problems are especially acute on scale, because classic extraction procedures can take you into solvent volumes that cause difficulties in cost, time, handling, and disposal. (Remember, all that solvent that you used in the extraction is going to have to be evaporated at some point!) Table 2 in the paper is a glorious sight, because that’s where the group took their nucleoside and partitioned it between the same polar organic solvent mixture and aqueous salt solutions with eighty-six different salts. If the prospect of assembling those data doesn’t excite you, then you probably don’t have what it takes to be a good process chemist. Here’s the take-home:
Although relative salting effects were only established for a single compound, our findings are expected to be general based on parallel trends to the extensive solubility studies reported by others and the generality of the Hofmeister series. In view of the frequency that chemists encounter difficult extractions, we wholeheartedly recommend expanded use of salting-out liquid−liquid extraction, particularly with salts that are consistent with green chemistry principles. At least in the context of process development, Na2SO4 has been underutilized in salting-out extractions, but we highly recommend increased use based on cost, efficiency, and chemical inertness considerations. In a broader context, and in view of the extensive literature available, we recommend testing a particular set of salts in addition to NaCl when significant aqueous losses are encountered during a workup: K3PO4, K4P2O7 (potassium pyrophosphate), K2HPO4, NaH2PO4, Na2FPO3, K2CO3, NaOH, (NH4)2SO4, Na2SO4 (at 30−40 °C to increase the amount that can dissolve in water), Na3-citrate, NaK-tartrate, and Na2-malonate. These salts should always be tested at high concentrations, and interpretation should account for acid− base equilibria, solute stability, and any potentially interfering ions in the mixture.
(Link in the above added by me). In the Merck case, good ol’ sodium sulfate, used warm as mentioned, turned out to do the trick, providing product that crystallized cleanly on concentrating at scale. The group was already using 2-methyl-THF/dimethoxyethane as the organic layer, which the organic chemists in the crowd will immediately recognize as pushing it already, but it was clearly a great relief to be able to move to a clean extraction procedure as opposed to multiple passes and compound loss.
The paper is excellent work, and goes into interesting details that I haven’t mentioned here (such as what happens when you start varying the ratios of the two organic solvents, etc.) It’s pure 200-proof process chemistry, but it’s also immediately applicable to bench-scale synthesis around the world, since it recommends cheap, easily available salts that can improve aqueous recoveries immediately. Read and heed, fellow chemists!