I recently performed a sodium dichromate
oxidation of 2-butanol to 2-butanone
For reasons soon to be mentioned, I thought my attempt at this reaction had failed miserably, and so I decided to accept and forget. But a few days later I examined my experimental data - where I expected an alcohol proton in my NMR, there was none to be found!
The oxidation was performed via fractional distillation, with 2-butanone expected to distill in the 77-85°C range. I used 2.5mL butanol as my limiting reagent, and anticipated a modest ~2mL crude distillate.
I collected less than 0.5mL. Perhaps my setup was just very low-yielding? I ran an IR and submitted an NMR sample. The IR suggested the collected “product” was almost entirely 2-butanol. I decided to run the experiment again, double-checking each step of the procedure.
Again, less than 0.5mL product was collected. I sighed, and st-
What the hell? Gas was evolving from my apparatus! Turns out, the distillation head had a small hole at the bottom of its thermometer jointLikely from someone attaching a thermometer too violently.. This explained the lack of product, but this time I could not run the reaction again - less than an hour remained in the lab period, there were no spare distillation heads, and I would still have to purify my product and collecting a melting point.
I cleaned up and headed out, forgetting to retract my bad NMR sample.
A couple days later, I received the NMR spectra for my submitted product, which I knew was actually 2-butanol. And that’s when it got interesting.
Do you see it? Almost every hydrogen is accounted for - the signals at
1.06ppm correspond to the molecule’s methyl groups, at
1.70ppm is the hydrogen adjacent
the alcoholThis proton should actually be further downfield.
, and at
2.45ppm are the protons of the third carbonThe minute rise at
7.27 is the solvent, CDCl3. -
but the alcohol proton is missing!
So where did it go? An argument could be made that the NMR sample was
impure, producing a bad spectra. While this may be true, even in a poor
spectrum one can expect the protons of the sample to have at least a small
For completeness, here is a spectrum of pure 2-butanol..
In fact, there is a far more interesting solution this puzzle.
The alcohol group of 2-butanol is susceptible to hydrogen bonding, but hydrogen bonds are dynamic and constantly changing. This produces a fairly large variability in electron density around the alcohol proton, leading to a broad range of potential chemical shiftsThis type of behavior extends to most functional groups exhibiting hydrogen or other dynamic bonding, including amines. (roughly between 2 and 5 ppm).
It may seems crazy, but this suggests that the alcohol proton is really just
hidden amongst the doublet or triplet at
At least, that’s my best theoryIf you know of a better
one, please let me know!,
and it absolutely blows my mind.