FID and expected peak areas around LOQ levels

[Aussiemosis]

Hi guys I am investigating how to demonstrate our planned LOQ of 0.02 %. By the way what is a typical LOQ for a GC/FID impurity analysis of an isocyanate?

Wanting to show >99% purity we are going with area normalisation because every one of the impurities is an unknown compound. No way to check relative responses.

So I have an area of 3,500,000 for the main peak. Then I made a 0.01 % dilution to show suitable signal to noise. But the peak has an area of around 700 which corresponds to 0.02 %, double the size I expected.

A 0.02 % dilution has a peak area of a little over 1400. Am I being unreasonable to expect a peak area closer to 350 for the 0.01 % dilution?

Integration influence from the baseline at these levels is minimal.

Thanks.
Daniel.

[bisnettrj2]

A couple things...

How did you do the dilution?

What does a 1:10 dilution look like - 350,000 area or 700,000 area?

My thought is that you might have reached the upper end of the linearity of the detector with the first injection, and you're seeing a linear response at the diluted levels. Or you made an error in doing the dilution.

[Peter Apps]

Another issue may be injector discrimnation. Depending on what the target analyte and the solvent are, changing the concentration of one component by four orders of magnitude might have some dramtic effects on vapour pressures and vapour volumes.

The only safe way to do this would be cold on column injection, or PTV at a stretch.

Peter

[Peter Apps]

Another issue may be injector discrimnation. Depending on what the target analyte and the solvent are, changing the concentration of one component by four orders of magnitude might have some dramtic effects on vapour pressures and vapour volumes.

The only safe way to do this would be cold on column injection, or PTV at a stretch.

Peter

[AICMM]

Aussiemosis,

First, what is your split ratio? Very important to try and achieve any sort of near-linearity with neat samples. Second, look at D5501 as a good example of this problem (impurities in ethanol.) Third, assume your low level response is double do you come anywhere near the 1% mark with impurities? If not, does the high bias negatively affect you?

Best regards.