Area % Discrepancies

[cawarhur]

Hello again everyone, I'll dive right in:

I am doing method development for a 105m Rtx-5 column starting from a method for a 30m HP-5 column to analyze the same kinds of samples (DCPD streams, C5's-C20's). We have GC's set up with each column and to validate the method that I developed for the 105m, we're looking for consistent area %'s between each system for a couple specific peaks (e.g. if peaks A and B are 49.06% and 50.04% respectively on the 30m, we're looking for those numbers on the 105m...within reason, whatever that is).

Thus far, the results of our preliminary tests have not been good and in general, have been off by a full 1-2% per peak comparing the runs on the different columns. It seems that some peaks have higher area%'s than others and some are pretty close.

My question is what can cause this discrepancy and what is an acceptable difference between columns? Our samples are pretty complex, so I'm not keen on the idea of running internal standards, but I'm thinking I might have to...what are my other options?

Thanks again, ChromForum Community.

[chromatographer1]

You have samples that have a wide range of volatility.

You are not injecting directly but are injecting with a splitter (assumed).

Due to pressure differences in keeping the same flow through the two different lengths of column the injections are being conducted under conditions that are the same, thus using a splitter can change the manner in which the sample is introduced onto the column (this is called discrimination of the splitter).

Even when installing the same column twice in an splitter injector one can get different discrimination with the same sample.

This also happens when you change the manner of injecting a sample, slow - fast - large - small.

Don't expect consistency where none should exist except under lucky circumstances.

Adding an internal std should not help.

Injecting a much smaller amount directly will help.

Or taking actions to minimize the effects of injector discrimination will help.

Remember that your sample is not a stable sample and can undergo reactions when heated within the injector as well.

GOOD LUCK.

Rod