Chromatography Forum

I analyzed fatty acid methyl esters by GC for over 4 decades at work. Early on, we used DEGS packed columns, but we changed early-on to cyanopropyl silicone packed columns, and later to wide-bore capillary columns, also with same SP-2330 or similar phases. Yes, retention times were not always "spot-on" every day we set up this assay, but we just modified the retention time windows in Chemstation, even after the sequence was completed.

The Chemstation software could provide equivalent weight, acid value, etc.

Correct, retentions times may vary day to day or batch to batch even. However, our retentions times are vary stable even day to day except for the first 10-20 injections after the instrument has sat for ~1day or more. The instrument can be on or off, this does not effect the outcome. If we keep the instrument running we can inject 300+ samples and the retention time does not shift after it becomes stable after the 10-20 injections. We can sometimes even utilize the same integration method for a week or more, but this only happens once the retention times have stabilized.

We are trying to figure out why this is as right now we put a Band-Aid on this issue and just hand integrate the first 10-20 injections and keep modifying the retention times until things are stable. After this our method takes over and integrates the samples. Normally we utilize 96-well plates and the time it takes to do the first 10-20 samples can be near the same time it takes to do the other 76-86 samples. We would like to get away from having to do this if it is at all possible.

Thanks for all the additional information. It really helps at least understand why you want to resolve the issue. We also do a lot of FAME work in 96-well blocks in high-through-put screening. It's not uncommon for us to 350-400 FAME samples per 24 hours (but not on a cyanopropyl siloxane column). So very short cycle times with narrow RT windows. I can really appreciate the frustration of chasing retention times around process large data sets. Analysts here were wasting lots of time doing this. Our solution, which won't help you, is we moved from an LTM, to a custom length column where we were nearly able to match cycle time and performance of the LTM. Drifting RTs went away.