B.R.S.M. The road to Tet. Lett. Is paved with good intentions


On The Synthesis Of Ketones

A bit of a lack of exciting syntheses so far this week, so here's some methodology and random reflections and recollections.

I don't mind that we don't get told the whole truth as undergraduates, because most of us can't handle the truth (well, not all of it). I appreciate that trying to convey even the basic concepts of organic synthesis to a large room full of people of mixed abilities, attention spans and interest levels in a reasonable amount of time is hard. I realise that only a tiny percentage of students on any given organic chemistry course will ever pursue the subject to a level where the simplifications they're taught in their first few years cause them much trouble.

One of the earliest things I remember from undergraduate lectures on carbonyl chemistry is being told that Grignard reagents don't add to carboxylic acids, and that ketones (or tertiary alcohols) can't be made this way. The reason for this is simple - Grignards, like most nucleophilic organometallic reagents, are also strong bases so they deprotonate the acid and are then unable to attack the resulting anion. This property of carboxylic acids can be useful as it can be used to protect them from harm during a synthetic sequence (and is one of the reasons that carboxylic acids are just about the only carbonyl group to survive the Birch reduction).

I'd gone on to assume that as carboxylic acids don't react with Grignards that they'd also be inert to all other organometallic reagents - organolithiums, cuprates etc. for exactly the same reason. That's what we organic chemists do, right? Rather than memorise everything we try and extrapolate reactivities of similar looking reagents. Well, last week I learned that actually Grignards are more of an exception than a general case, and that actually both organolithium reagents and cuprates can be used to convert carboxylic acids to ketones! It's not the most general reaction, and obviously the substrate scope is a bit narrow (so don't go throwing away your Weinreb amides yet), but it actually works pretty darn well in some cases. To give you an idea of a representative procedure, here's a an organic syntheses prep for cyclohexyl methyl ketone.

A brief poll of the people nearest me in the lab today concluded that nobody I work with knew you could do this. Ah well, we're here to learn. I only found out from reading a recent Org. Lett. ASAP article which introduces the use of R2CuLi•LiCN type cyanocuprates for the same reaction. Obviously these are considerably milder than, say, methyllithium and are great for substrates with racemisable centres or arylhalides which are rather prone to exchange or other side reactions. The copper analogues of the tetrahedral intermediate shown above are also a great deal more stable lasting all the way through to aqueous workup without fragmenting or misbehaving. A useful extension of an old, but not so well known reaction.


While considering the harshness of this method I began to wonder what the mildest way to convert a sensitive carboxylic acid to the corresponding ketone is. I remember that Stork and coworkers did have this problem when attempting to transform the rather fragile acid below into its corresponding methyl ketone en route to the cedrenes,[1] and used the following rather creative organometallic free sequence. First the sodium salt of the carboxylic acid was converted to the acid chloride. This was then reacted with diazomethane on an unhealthy scale to give the corresponding α-diazoketone. Treatement with dry HCl in ether gave the α-chloroketone that could finally be reduced using zinc in acetic acid to yield the required methyl ketone. Phew! Although quite a long sequence the overall yield was excellent.

Everything else the group tried, including reacting the acid chloride with dimethylcadmium (eek!) proved unsuccessful due to rearrangements occuring, presumably catalysed by the Lewis acidic metal salts. Anyone else know of a longer (or milder, or more interesting) sequence used in the literature?


1. The full paper can be found in J. Am. Chem. Soc., 1961, 83, 3114. The original communication was some half a decade earlier. This is a brilliant paper, way ahead of its time with great experimental, discussion and mechanistic rational throughout. And they even had to assign the natural product stereochemistry before they started. Epic.



Comments (3) Trackbacks (0)
  1. Nice article! I actually was going to review that R2CuLi paper. While I did know about the R-Li + COOH —> R=OR, I knew it wasn’t the best reaction. The guy I used to work for told me about how certain substrates work awesome and others don’t work at all (usually the ones you really want to work!). In fact you don’t even need to use LiH (you can use 2.1 equiv of the organolithium though it saves you some money if you pre-deprotonate the acid). If your interested in doing COOH to a COMe conversion, I’d say do a darkin west conversion ( http://pubs.acs.org/doi/abs/10.1021/jo0607966 ) That’s a personal fav of mine and its pretty straight forward, being performed under somewhat mild conditions.

  2. Even with Grignards one has to be careful: I remember doing reaction of PheMgCl (excess) with m-benzoylbenzoic acid, to prepare m-carboxy-substituted trityl alcohol as a linker for solid phase peptide synthesis. One had to be very careful with the reaction temperature control otherwise there was lots products from the reaction on carboxylate.

    Dakin-West is great (for making stuff like phenylacetones, and also for trifluoromethyl ketones) but you will scramble the stereochemistry alpha to carbonyl. I think Weinreb amide + MeLi is the most reliable way to go, of course all other carbonyls have to be protected.

  3. I made cyclopentyl butyl ketone from cyclopentyl-MgBr and butyryl chloride (via Chem. Pharm. Bull, 1989, 37, 958). It actually worked better than the Weinreb route in my hands.

Leave a comment

No trackbacks yet.