Chromatography Forum

Hi everybody,

I'm working on the Residual Solvents method. My diluent is Methanol. One of my analyte is Ethanol (specification is NMT 5000 ppm). Methanol has small amount of ethanol (about 50 ppm, S/N ratio is about 30). How to determine LOQ for Ethanol in this case?

Thanks

Well, first you have to determine the variation of ethanol in the methanol. How much variance does a sample show in its ethanol content?

If you measure say: 1000 counts for ethanol in your methanol solution, do you get a variance of 2%, plus or minus 40 counts? Let's suppose:

960, 1000, 1040, 988, 1037, 1001, 990, 1030 counts for 8 injections.

Make up a solution that would add 50 counts of ethanol to your methanol. Does the mean of several injections increase by 50 counts or not?

You have to state how large an increase of ethanol in your methanol allows you to see a 95% probability that the ethanol peak is larger than your blank.

That amount will be your limit of detection. It very well may be that you should state 50 ppm is your LOD if you can see an increase of the ethanol peak when its concentration is doubled WITH CERTAINTY.

Too bad you can't replace methanol as your solvent, or you can't get higher purity methanol.

What happens if the next lot of methanol has a larger amount of ethanol in it?

best wishes,

Rod

Rod,
Thanks, for your reply.
I have other five components (one of them is DMSO). So it is not easy to replace methanol with another solvent.
It is always possible to sustract ethanol amount in the Blank from actual result.
I just would like to know, if there is formal way of calculating LOQ level in my case. As per our SOP, solution with S/N ratio 10 considered as LOQ.
My blank has S/N ratio about 30. I'll prepare solution with S/N ratio about 40, inject it 6 times and show that area of ethanol corrected by area in blank is reproducible. Could I use this solution as LOQ?

Thanks,
Mark

You have a plan.

"I have other five components (one of them is DMSO). So it is not easy to replace methanol with another solvent."

Not easy. True.

How about propylene glycol as a solvent?

But what happens when the next lot of methanol has ethanol with a S/N of 200? 2000?

Then what?

You tell me.

Rod

Dear Rod,

I ask this question, because I felt its common problem during Residual Solvents method validation. You see bunch of methods like this: Methanol in DMSO or DMF, Benzene in Toluene, Acetaldehyde in Methanol, etc. Nothing wrong with these methods. The problem is only that solvents may have small traces of analytes. You can easily calculate amount of your peak of interest by subtracting area count in blank. This is not a problem. My question was, how other chemists calculate Limit of Detection for these kind of test during formal validation.
I don't understand, why you are criticizing my method, instead answering my question? It's always easy to criticize somebody's method, especially if we don't know some important details about our sample solubility, recovery, etc.

Sincerely,
Mark

I am only being as critical as anyone in a regulatory agency would be. I have dealt with such and you want to have a VERY GOOD defense for your decisions. Better me than HE.

Ultimately, this is why you are doing the validation, no? I am trying to help you, not kick you.

Having a known interfering peak in your dissolution solvent will be seen as a problem to any claims of LOD you may make. It is not a recommended practice in every place I have worked, Let me correct that statement. It would NOT be accepted, period. Another technical solution would be demanded.

But to answer your question. You will have to claim a LOD level which when added to your solvent will show an unquestionably larger peak to your blank.

Good luck with your validation.

Rod