As many of you will have noticed in the comments to the previous post (which was thoroughly hi-jacked), an intriguing paper has been published in JACS by Xinbo Wang, Bo Zhang and David Zhigang Wang.  In this, they suggest it is possible to oxidise benzylic alcohols to the corresponding ketones using sodium hydride (amongst other chemistry).  Given that sodium hydride is, well, a hydride – this is quite something.  Does it work?  Hard to say without giving it a go, so I am.

We had this gear in the lab, so I’m giving it a go.

10.30 – Took the dried flask out from the oven, flushed it with nitrogen and vacuum.  Added (dry) THF, and cooled with ice water.  Also, found the big ice scoop I’ve been looking for ages on the biology floor.  Thieving bastards…

10.40 – Added the alcohol (626mg, 4 mmols), left it for a couple of minutes and then stuck in the sodium hydride (60% dispersion in oil, 320 mg, 8 mmols) in one go.  Cue fizzing.

10.50 – Warmed to RT.  Let’s wait.

12.00 – ‘Sgone orange:

13.20 – Okay, I couldn’t wait any longer, so I LCMSed the reaction mixture.  This is what I got:

[Click on the picture to get a larger image]

Looks like two main peaks, with the first (3) being more polar and having a negative mass ion corresponding to the starting material.  The other peak (4) doesn’t ionize too well, so it’s hard to conclude anything so far.  I’ve just got to sit on my hands until tomorrow morning…

16.30 – Okay, I’ve done as suggested in the comments and LCMSed both the product (Aldrich) and SM (Acros) against the reaction mixture.  Here’s the SM: (contaminated with what looks like a little  product – it’s an old sample…)

And the product:

And the reaction mixture:

You can see that the large peak in the reaction mixture appears to be product…

I’ll wait for the NMR before I confirm it, but it looks like we’re in need of a new mechanism.

11.00, Thursday – The results of the NMR are in.

I quenched the reaction via dropwise addition of saturated ammonium chloride solution.  No fizzing, really – just a few bubbles, so I reckon that the NaH is pretty dead.  I then pipetted a small amount of the organic layer off, and vacced it down in the completely awesome Biotage v10. This gave me a orange oil which I took up in CDCl3 and NMRed.  Here’s the NMR of the SM:

And here’s the product mixture:

Looking in the aromatic region, in the SM all four protons have coalesced into one peak, integrating nicely.  In the product mixture, this is still evident, along with a roofed pair of doublets, corresponding to product.  (I haven’t got an NMR of it, but look on the Sigma-Aldrich website).  Doing the maths, this corresponds to a 15% yield.  Quite a long way from the 85% quoted in the paper.

One thing worth pointing out is that I did the reaction on a 4 mmol scale.  As is often noted, increasing scale often negates problems with trace water due to the macroscopic effect.  This is also likely to apply to trace oxygen (or other oxidants).  This leads me to believe that the reaction is as suspected – oxidation of the alkoxide by some trace oxidant.  What this is, I don’t know.  If anyone else feels like following up on this, perhaps some inorganic chemists could try doing this chemistry in ultra-dry/deoxygenated environments.

18.30 – Dénouement
Sounds like lots of other chemists have been trying this thing – and it’s hard not to conclude that keeping the reaction ultra-anaerobic prevents oxidation. It’s also true that the starting materials – benzylic alcohols – are exceptionally easy to oxidise.  However, this could still be a useful prep – especially the amide formations.  I don’t have the time to try that chemistry just now, but I’d love to see a similar series of reaction (aerobic and anaerobic).  That brings us to the problem with this paper – a lack of (application of the) Scientific Method.  THings they should have done:

1. Tested the reaction under various atmospheric conditions – i. Normal atmosphere; ii. Nitrogen; iii. Nitrogen (degassed); iv. Argon; Oxygen

2. Tried various starting materials, including homobenxylic.

3. Perhaps tried to trap intermediates.

Thanks to everyone who has contributed to this discussion, and also to the folks at Arrow for allowing me to spend a little time on this (though I didn’t spend that long – honest).