Determinining the purity of solvents

[AnaChem1983]

Hi all,

I have a question regarding purity determinations of organic solvents by GC. To be specific, what is the best method for quantifying the purity.

Based on what I have done and, indeed, been taught in the past I have always thought that the area % method is the best way of assessing the solvents purity. Particularly given that the identity of many of the impurity peaks are unknown. However it has recently been suggested to me that if the identity of a couple of the impurities can be elucidated it is acceptable to calibrate these compounds and, by using the % normalisation method, adjust the response factors of these peaks. All uncalibrated peaks will have a response factor of 1.

This latter method does not seem an acceptable way of proceeding to my mind. However I was wondering if anyone on here had experience of assessing purity this way. If so could you provide any explanations as to why this is acceptable? What would you say is the best way of assessing an organic solvent's purity by GC?

Many thanks,

James

[chromatographer1]

If you truly do not know the identity of the impurity peaks then any purity claim is meaningless.

Most impurities of commercial organic solvents ARE well known. Manufacturers determine the actual impurities and their actual amounts.

For example ethanol may have methanol, isopropanol, acetaldehyde, ethyl acetate, and acid acid. Let us not forget the most obvious, water.

Hexane may have alkane homologs, C6 isoparaffins, cyclohexane, MCP, and benzene.

toluene has C6-C8 hydrocarbons, benzene, ethyl benzene, and xylenes.

You can report an area percent value by FID but this ignores water and other non-responsive analytes to an FID. Many may choose this route to give some information about the purity of a solvent, but it hardly is conclusive.

So your question is really about reporting the purity of solvents, not an actual determination of purity of solvents. Document whatever route you choose to do and report a number.

best wishes,

Rod

[Don_Hilton]

GC also fails to give a response for non-volatiles in the solvent. If you look at the specifications for the solvent, you will see an entry for residue after evaporation.

For the useage of a solvent, it may be more imortant to know the condentration of impurities that would cause a problem. In toluene, the presence of 0.01% xylene may not be a problem to you. But if this 99.99% pure toluene has been exposed to air and UV light from the fluorescent tubes in the room, it can form peroxides. The peroxide that forms as the result of free radical oxidation is very small -- but if you are doing trace analysis of an easily oxidized compund, it has the potential to affect your results.

[Johnny Rod]

The first question for me is: why? If you are QCing batches of raw materials that come with a C of A then area% coupled with a water determination should be enough for you. Once you get into quantifying specific analytes you are looking at a bit more analytical work, so you would need to be sure it's worth the bother.

[AnaChem1983]

Cheers for your answers so far folks.

I've been asking questions within my own company about why we proceed in the manner chosen but can't seem to get any sort of useful answer.

I'm aware that with these things there is more than one way of doing it. As Rod has pointed out unless all peaks are known the reported value is not really a conclusive answer.

Thank you all again for your time and consideration.

James