Second in my somewhat badly thought out Unnatural Products series is dodecahedrane. To be honest, this compound is actually pretty much the main reason for this series. I only found out it had been made at the start of the year (even though Paquette did it way back in 1983), and it blew my mind. I mean look at it – where do you even start? It took me a full hour to draw the damn thing in Chemdraw, and I still can't get it to look right on paper. As Paquette himself said in an abstract ‘the aesthetically pleasing symmetry of the dodecahedral framework was clearly apparent’. In common with the syntheses of the other two compounds above the route involved both Diels-Alder reactions and lots of photochemistry. Let’s take a look.
The group began with the dimerization of sodium cyclopentadienide, not isolating the product but reacting it straightaway with dimethyl acetylenedicarboxylate in an unusual double Diels-Alder reaction that gave a mixture of two products, slightly in favour of the desired one. Although the yield is a little low, if we redraw the product (which is non-trivial) we see that actually this one step forms four rings - a third of the target!
Despite this early success, far more effort was required to introduce the next two rings. Hydrolysis of both esters, followed by iodolactonisation, formed two α-iodolactones. Methanolysis of these, Jones oxidation of the alcohols released, and dehalogenation gave the diketone shown. Although it seems strange that five steps were required to obtain this compound from the diene at the start of the scheme, hydroboration and oxomercuration approaches were unsuccessful, and although the route is a bit lengthy, the yields obtained were all excellent. Next, a double Corey-Chaykovsky epoxidation with the unusual cyclopropane ylide shown, followed by rearrangment of the resulting spiroepoxides under acidic conditions give the bis(spirocyclobutanone). Finally, Baeyer-Villiger oxidation and another acid catalysed rearrangement with Eaton’s reagent gave the corresponding cyclopentenone.
Both olefins and both ketones were then reduced (the latter with unsurprisingly high stereoselectivity) to give, after a touch of acid, the bis(lactone). Treatment of this with methanolic hydrogen chloride then gave the dichlorodiester. Here, the group changed tactics and stopped working on both sides at the same time, desymmetrising this compound by treatment with lithium in THF-liquid ammonia. By quenching with chloromethylphenyl ether the group also introduced a blocking group next to the remaining ester to facilitate the subsequent steps. Next came the first of three (!) Norrish reactions used in the synthesis, forming the eighth ring in fantastic yield.
The Norrish reaction unfortunately left behind an extra hydroxyl, which was eliminated under acid conditions, and the resulting double bond reduced out with a little diimide. Conversion of the ester to the aldehyde set the stage for a second Norrish cyclisation that unfortunately occurred in much lower yield than the first. Four steps were then used to remove the blocking group, namely Birch reduction of the phenyl ring and acid hydrolysis of the resulting enol ether, followed by oxidation of the liberated alcohol to the aldehyde and retro-Claisen condensation. A third and final Norrish reaction was used install the next ring, in quantitative yield, and the leftover alcohol was removed as before. The final reaction in the sequence, which in modern grant-speak would probably be a double CH-activation, is like nothing I’ve seen before in a synthesis. Although formally just a dehydrogenation, something palladium is well known to do, it's not commonly used to form C-C bonds in this way and it’s a pretty audacious last step. 23 equivalents of palladium sounds like a lot, but as the group only managed to make a few milligrams of product I guess it didn’t cost them too much. Sounds a bit like a natural product synthesis doesn't it?
Addenda and Miscellanea
1. The original paper is J. Am. Chem. Soc., 1982, 104, 4503. For more references and a more detailed scheme there's also a synarchive page, which unfortunately I didn't find until after I had written this. The first attempt at a dodecahedrane synthesis was that of R. B. Woodward, ahead of his time as ever, way back in 1964. He synthesised Triquinacene, which he had hoped to dimerise to give the target, but was unsuccessful. Several other groups, including Eaton's and Pettit's pursued dodecahedrane in the intervening years.
2. I’ve seen one or two Norrish reactions, but never to form C-C bonds. With everything held so close together and no other functionality, little can go wrong and the yield is excellent in two out of the three performed.
3. The group prepared this the oldschool way, by oxidation of hydrazine with hydrogen peroxide, a reaction also used in the engine of the first fighter jet.
4. Some of you might be wondering why you’d do a dehydrogenation under a hydrogen atmosphere. The reason is that in a model methyl dodecahedrane system it was found that simply using palladium-on-carbon resulting in over-dehydrogenation, where further reaction with the methine CHs in the product occurred. This reaction could be suppressed by applying the correct pressure of hydrogen. Clever.