Chromatography Forum

I measure free and total glycerin in finished biodiesel using method ASTM D6584. I have an Agilent 6890N GC with an Agilent autosampler that injects a 1 uL sample (cool-on injection) onto a DB-5 HT capillary column. The detection is done by FID. I use 1,2,4-butanetriol as an internal std for free glycerin and tricaprin as an internal std for mono-, di-, and triglycerides.

About two weeks ago I began noticing that our QC samples were having trouble passing because everything was coming out at the high end of the std deviations (I use ASTM round robin biodiesel samples for QCs). Thinking that it was the fact that the column was about at the end of its lifetime, I decided to change the column out and replace it with an identical model from the same company. This did not fix the problem and I soon realized that the jet has a plug in it. I replaced the jet with a new one and thought everything would be fine.

I then calibrated the new column like usual with purchased std mixes and purchased internal stds. The calibration seemed to have worked fine with R-Squared values above 0.99 on the curves for each component. However, I began noticing that the peak areas of all components (especially the internal stds) were considerably lower than usual (about 880 for tricaprin compared to the usual area of about 1300 or so). This is in turn increasing our test results. I had a sample sent to two independent labs and I ran it as well to confirm that something was wrong. The two ind labs got similar (and normal) results for everything, but the numbers I got for results for each component were about 20% higher than the other labs' results.

I don't know if maybe I replaced a part (like the jet) incorrectly or what. I reinstalled the original column today and the peak area of the tricaprin std was down to about 850 or so from it's original area before removing of about 1300. I thought it was just a bad column after I put the new one in and saw the lower areas but now it is leading me to believe it is a problem with the GC at the inlet or the detector side.

Any help would be greatly appreciated!

If the peak areas of all components are reduced, and the areas of internal standards are reduced more than the analytes then I strongly suspect that something has gone wrong with the preparation of samples and standards. Anything wrong with the GC should affect all compounds equally, especially with cool on column injection that does not have the molecular weight discrimination seen with vapourizing inlets.

Peter

If the peak areas of all components are reduced, and the areas of internal standards are reduced more than the analytes then I strongly suspect that something has gone wrong with the preparation of samples and standards. Anything wrong with the GC should affect all compounds equally, especially with cool on column injection that does not have the molecular weight discrimination seen with vapourizing inlets

I just calculated fractions of what the peaks used to be in terms of area and compared them to the area of the same peaks using the same column now. In other words, I ran the same sample before and after the problems began using the same column and the same internal stds. It seems as though all peaks are roughly 70% of what they used to be in terms of area. Maybe that is another clue?

If the peak areas of all components are reduced, and the areas of internal standards are reduced more than the analytes then I strongly suspect that something has gone wrong with the preparation of samples and standards. Anything wrong with the GC should affect all compounds equally, especially with cool on column injection that does not have the molecular weight discrimination seen with vapourizing inlets

I just calculated fractions of what the peaks used to be in terms of area and compared them to the area of the same peaks using the same column now. In other words, I ran the same sample before and after the problems began using the same column and the same internal stds. It seems as though all peaks are roughly 70% of what they used to be in terms of area. Maybe that is another clue?

It is indeed a clue - and different to what you initially posted, A uniform decrease across all the peaks points to an instrument problem rather than sample prep, except for a final dilution after the standards have been added.

Try with another syringe. Check (again ?) the insertion of the column into the inlet and the detector. Check the flame gas flow rates - with an independent flow meter not the meters built into the GC. Check for leaks using a leak seeker.

Peter