GreenD - 14-2-2014 at 18:13
So grad student I worked with shows me 9 peaks on an nmr spectrum that correlate to our product of interest. There is another peak that is some
impurity.
What he does is integrates the 9 peaks (lets say which coincide to 19 protons) he then normalizes so these 9 peaks integrate to 19 protons.
Then he adds the additional peak, which is normalized to have 1 proton (20 total, normalized, integration units)
He says the purity is 19/20 or 95%.
Is this bogus?
Chemosynthetic - 14-2-2014 at 19:32
I think that method sounds bogus, but the idea is correct. Sounds pretty similar in concept to something I have done before with one huge caveat; I
was sure what my peaks were. I believe you can back calculate concentrations and compare with regard to solvent, then get a purity ratio that way as
well. It has admittedly been an embarrassingly long time since I have done anything remotely like that, though, so check out a reference like this
and make up your own mind:
http://mestrelab.com/blog/article/purity-calculation/
Metacelsus - 14-2-2014 at 19:36
To determine purity, you first need to know how many protons the two compounds contain. Then you can calculate the molar percent of each.
The purity is probably not 95%. Could you provide the relevant spectra and structures of the product and impurity?
ziqquratu - 19-2-2014 at 20:28
The method is no good, for a couple of reasons.
As noted by the previous posters, you need to know how many protons are really represented by each peak in order to determine the purity. For example,
1mg of tetramethylsilane, (CH3)4Si, will give about twice the signal of 1mg of benzene, C6H6 - despite the
fact that they are both of similar molecular weight (88 vs 78 amu). Using your friend's method, you would thus conclude that the TMS-containing sample
contained twice as much impurity - even though it's the same by mass and, mole-for-mole, it actually contains slightly less! If you don't even know
what the impurity is, you can deduce nothing more than that an impurity IS present in your sample.
Secondly, when thinking about purity, always remember that 1H NMR spectroscopy can only detect compounds which have protons, and many
common impurities lack them. I had this particular argument with a mid-stage grad student in my former workplace, who insisted that NMR was enough to
determine purity. At a later date, he happened to see me looking at a spectrum and commented on how pure my sample was. I took the opportunity and
reached over to show him the sample - a thick brown sludge which, after distillation, was about 40% product by mass (the remainder was some
polybrominated scum that I never bothered to characterise). Long story short, don't trust NMR alone for purity determinations.
This was the same guy who once decided to check the purity of a bottle of trifluoroacetic anhydride by 1H NMR...
Nicodem - 20-2-2014 at 08:31
NMR can not be used to measure purity. Purity can only be measured by developing a method specific for each compound. You can actually develop an NMR
based method, but it will not be very sensitive and you would have to couple it with an assay method (which can also be NMR based).
Mind that even general purity determining methods like "chromatographic purity" (HPLC or GC area%) do not directly measure purity. As a consequence,
purity is a pretty abstract term and thus requires definition for each specific case.
NMR can however be used (and is commonly used) for assay determination. For this you need to add a weighted amount of a standard and then you can
calculate the assay (but only if you know the molar weight of your compound and you can assign at least one well separated NMR signal to the number of
hydrogen atoms - you don't need to know the exact structure of the compound).
unionised - 20-2-2014 at 12:49
"NMR can not be used to measure purity."
Oh yes it can.
But you need to be very careful.
And, as you say, the sensitivity sucks.
Also, if the original poster's sample were, for example, 50% tetrachloromethane, the NMR wouldn't see it at all.
Metacelsus - 20-2-2014 at 14:14
If the components of a mixture are known (and show up on NMR), their molar ratios can be calculated. However, this is totally useless with unknown
impurities.