LOQ for unknown impurities

[htremblay]

Hi,
I am struggeling with how to determine the LOQ for impurities in my GC method. The method is used to evaluate the purity of one of our raw materials, and results are reported as % peak area for the main compound, as well as for impurities. The impurities are all unknown impurities. How should I proceed to determine the LOQ?
Thank for your help!

[grzesiek]

have you read anything about it already? do you have any knowledge concerning the subject?

[htremblay]

I did read about it and have some knowledge of the subject but could not find anything specific to this case, where I am only looking at area % and cannot control the impurity levels. I have access to a few lots of material with specific amounts of impurities, and cannot prepare samples with devcreasing concentrations (vs main peak) to either evaluate linearity and/or determine the LOQ experimentally.

[krickos]

Hi

Some general comments, i do not know what field you work in but...

As the last resort when everything else fails I suppose you can resort to diluting the main peak to check linearity and estimate LOQ.

But some effort should perhaps be made first.

For "simple" raw materials a GC/MS run can be enough to identify your unknows.

Have you spent time with a process engineer/chemist and discussed potential impurities?
How critical are the impurities identity/limits with regard to the process?

Position isomers of raw materials (and for instance the corresponding alkohol or chloride of an aldehyde) may be commersial availeble.

If you know the synthetic route of the raw material it can give clues to potential impurities or even a recept to make some yourselves if needed.

This might be over the top for your applications but at least a few ideas.

[htremblay]

I already tried dilution of the main peak. It gave me the LOQ as absolute area units, but it is hard to correlate to % area impurity as the area of the main peak varies between lots and analyses...

As for impurity identification, we already tried GC-MS (without success for the impurity of interest). It is not critical for us to identify it, and we already know how much of it we can tolerate in our process.

Thank you!

[grzesiek]

"It is not critical for us to identify it, and we already know how much of it we can tolerate in our process. " - you don't know what it is but you know how much of it you can have it in your process - how do you control it then in the process?

"I already tried dilution of the main peak. It gave me the LOQ as absolute area units, but it is hard to correlate to % area impurity as the area of the main peak varies between lots and analyses... " - how much does it vary? and why?

[htremblay]

Hi grzesiek,

"you don't know what it is but you know how much of it you can have it in your process - how do you control it then in the process?"

We test the raw material before using it in our process, and we have a maximum acceptable % area for the peak that we know interferes with the process (we know that it is a cross-linker of some sort but don't know its exact structure).

"how much does it vary? and why?"

The specification for the main peak is 89 - 100 % area, so it can vary quite a bit depending on the lot used.

Right now, I am leaning towards defining our LOQ as the lowest %area obtained that consistently results in a %RSD of less than 10...

[grzesiek]

ah ok it's in the raw material

"The specification for the main peak is 89 - 100 % area, so it can vary quite a bit depending on the lot used. " - it is purity of the raw material if I understand correctly?

"I am leaning towards defining our LOQ as the lowest %area obtained that consistently results in a %RSD of less than 10..." - few things here, LOQ is lowest concentration that you can quantify, so it is not %area, you can then calculate it as %area of the main peak

how do you measure purity? lets say you have 95% - is it for you normalisation result or what? do you have linear response all the way?

[htremblay]

You are correct, we use the same assay to confirm the purity of the raw material, as well as to make sure that there is less than 0.8 % of the impurity known to interfere with the process.

We report all the results as percent area, both for the main peak and impurities. I know that the response is linear for the product (10 to 100 %) and for the impurities (in a limited range, evaluated by mixing the lots of product having the most and least amounts of impurities).

Part of my problem is due to the fact that we do everything as % area, and never measure any concentrations. I was able to evaluate accuracy on the main peak by diluting the product with a known related compound (and follow the decrease in percent area), but I cannot really do it for the impurities...

[grzesiek]

I'm not sure about the normalisation method you use, at least main peak should be assayed using some reference standard so you know that at least you see everything that you think you see in your sample

"I know that the response is linear for the product (10 to 100 %) and for the impurities" - please could you post or email results so this discussion can be more productive