In my experience, most organic and medicinal chemists are always ready to hear about the latest results in two branches of the science: things that explode and things with weird smells. Maybe we are in our way “singularly in touch with the primitive promptings of humanity”, as Captain Grimes says in Decline and Fall (although let it be noted that he had something very different in mind). So let’s do a bit of solid, 100% odoriferous chemistry today as a brief change of pace from vaccine work. Ever heard of vetiver? It’s a tropical grass, very large and sturdy, with an extremely deep, dense root system that’s led to the plant’s use in soil stabilization. But those roots have long been used for something else entirely: perfume.
There are a lot of plants in that category – not just flowers, but thing like sandalwood and patchouli. The problem is that in many cases it’s just not possible to grow enough of some of these to meet the fragrance industry’s needs. True sandalwood, for example, has been ferociously overharvested over the decades and is now a protected species. This C&E News article is a nice look at the problem, and at the balance between supply chain needs, cost of ingredients and many consumer’s stated preference for “all-natural” formulations. This can be an opportunity for organic synthesis to step in with larger-scale production, either from other chemical feedstocks or from natural products that can be harvested from more easily grown plants. Medicinal chemists will immediately recall the situation with the natural product taxol, whose use in chemotherapy threatened to make the Pacific yew extinct at one point (the inner bark of the tree had to be stripped). A semisynthetic route was developed at Bristol-Myers Squibb from a related intermediate that could be harvested from pine needles, and now there is even commercial production from plant cell culture.
Now, you don’t hear too much demand from oncologists for all-natural paclitaxel, but all of these techniques do raise questions about what can be called natural and what can’t. It should go without saying that (for example) pure vanillin made in an industrial reactor is exactly the same as pure vanillin isolated from the tropical bean. But it should also go without saying that the extract from the plant will likely contain all sorts of other constituents (some of them at very low levels) that add depth and complexity to an aroma or flavor. Isoamyl acetate, for example, is by far the dominant note in the smell and taste of a banana. But by itself, it smells like banana candy more than it smells like a fresh banana, thanks to all the other components. (Note: I’ve heard the story about how isoamyl acetate was marketed to mimic an earlier variety of banana, but that doesn’t quite hold up). Similarly, ethyl methylphenylglycidate is a big part of the smell of a strawberry, but definitely comes across as artificial and cheap-candy-like on its own.
Most natural extracts have been investigated to the point that we know their “principle”, as the perfume industry calls it. Instead of trying to source original sandalwood oil (not easy on a commercial scale) or deal with the various other oils from related trees and shrubs, you can go get some santalols made by fermentation processes, or buy some sandalore or brahmanol, or try “Dreamwood“, a macrocyclic lactone in a totally different structural class that also recreates much of the effect. Vetiver, though, has been a tough one.
Vetiver itself is a lot easier to cultivate than sandalwood, but harvesting and distilling tons of vetiver root is still something that might be improved on just a bit. But despite decades of effort, no one has quite pinned down which of the hundreds of compounds in vetiver oil really carries the bulk of its distinctive character. Until now, apparently. This new paper seems to have tracked it down, and it took a fair amount of synthetic chemistry to confirm it.
You can see some of those components above (graphic from Angewandte Chemie/Wiley). And you’ll notice that some of these things are pretty easy to smell, like beta-vetivone with a threshold of 0.3 nanograms/liter of air. But it smells more like grapefruit than it does vetiver oil itself, so that can’t be the principle (rather, it’s one of the “top notes” of the overall scent). You may be surprised to see geosmin in there – now that’s a compound with a low detection threshold, and you’ve certainly smelled it, whether you know it or not. It’s the smell of fresh dirt, or the smell you get as a rainstorm starts to wet the ground. There’s a tiny bit of that in the overall vetiver profile, too. But there’s a unique-to-vetiver part of that profile that people have been trying to track down all these years. As the paper details, khusimone (lower left) has long been thought to be that principle, but although its in the right ballpark, its odor detection threshold is just too high for it to be the real candidate.
The paper is a joint effort from the Max Planck Institute in Mülheim and from Givaudan (a big player in the fragrance industry), and it mentions an effort at the latter company to carefully distill 800g of vetiver oil through a Sulzer column and then a Spaltrohr. Even a lot of synthetic organic chemists will be saying “Through a what and a which?” to that one. Sulzer is a Swiss company that makes all sorts of fluid-handling equipment, including many high-end distillation column systems. A look through their material will take you into a whole new world of chemical engineering for such hardware, if you’ve never had to worry about such things. (I haven’t myself – the most hardcore distillation I’ve ever done was with a spinning band column back in 1985 or so, and that was a one time event!) A Spaltrohr distillation (the second step in the Givaudan work) is a tube-within-a-tube column with only a very small separation between the tube surfaces. It’s typically used for small scale vacuum distillation work, although I don’t think “typical” is the right word for what’s a rather uncommon piece of equipment. You’ve got to take those things up very slowly, but you can get some pretty amazing separations if you have the patience.
Now, if you put vetiver oil into a gas chromatograph system, you get hundreds of peaks for different compounds, so isolating things through fractional distillation is not something you’d undertake lightly. A mere 155 of those peaks, by the way, have ever been firmly identified. But you can narrow things down with an interesting GC accessory called a “nose port”, which is exactly what it sounds like: you sniff the separated eluent from the column and use the human nose (which can be a startlingly good detector) to figure out where your compounds of interest lies. Not recommended for ordinary use in the lab, but for flavor-and-fragance work it’s just the thing. That showed a key region eluting at between 45 and 46 minutes with the sought-after aroma, but there were still a number of different peaks in there. And the problem with all research on olfaction is that trace constituents, things that you would hardly even notice in such a forest of GC peaks, can have enormous impacts on smell.
So the only way to be sure was to synthesize some of these things, and that’s what the team in Mülheim did. An earlier attempt to deconvolute both vetiver oil and the vetiver oil literature, which was hardly in better shape due to such trace-component problems, suggested the ziza‐6(13)en‐3‐on skeleton compounds as candidates, but no one had ever prepared these in enough quantity to see if they came at the right spot in the GC smell traces. The Givaudan work agreed with this, pointing especially to compound 10 in the graphic above as a likely winner. Chemistry time! The synthesis (which gives three stereoisomers that are then separated) is about ten steps from cyclopentenone, and involves a tricky intramolecular Pauson-Khand for ring formation. The double bond left from the starting material turned out to be a real pain to reduce selectively in the presence of the exo-alkene left over from the Pauson-Khand, and as you can see from the structure above, that latter one needs to be in the final product. You can tell that this is not what the team was looking to be fighting ten steps into the synthesis, but they got around the problem by epoxidizing the exo-alkene. Hydrogenation then knocks down the enone double bond while converting the epoxide into a hydroxymethyl, which is then eliminated.
And sure enough, compound 10 had a unique odor that hit the elusive “transparent woody-amber” note of vetiver. It’s at least 150 times more detectable to the nose than khusimone, down to 0.03 ng/l of air or below. The authors noted that its scent is rather similar to arborone, which is the active component of a well-known fragrance ingredient called Iso E Super (and other other names as various mixtures of stereoisomers). Those compounds don’t look a heck of a lot alike at first glance, but are quite similar stereoelectronically in a 3-D overlay, which makes sense. Iso E Super has been described as “remarkably pleasant”, and apparently compound 10 is as well. You can bet that the fragrance chemists are modeling the two of them as we speak to figure out some other interesting molecules that might share the same property!