GPC Peak Integration

[LC101]

I was talking with a colleague and they were describing how they integrate GPC peaks. They indicated they truncate the peak at the front and back by a certain percentage. It gives better repeatability. Has anyone heard of this method of integration. The manuals I have looked at do not deal with this. Thanks, LC101

[HW Mueller]

Fudge!

[mbicking]

Double fudge, topped with [expletive deleted]!

We integrate to determine the size (area or height) of a peak, not to make sure that the peak has a certain size. There is no reason to reject part of the peak unless you have a shoulder or other unresolved component (visibly) present. And even then, your errors could be very large. In GPC, this approach has even less validity, because you really need to know the molecules present at both large and small extremes, since these have an impact on the MWD parameters.

Rather than trying to force the data into a pre-determined conclusion ("fudging" is a nice way of saying "cheating"), they should be looking at why they have repeatability problems. Is it actually a chromatography problem, or is there a problem with the sample(s)?

[ivanvins]

The GPC results are extremely dependent, how the chromatograms are integrated and are prone to great errors. The integration algorithm in the used software may not be able to detect the peak starts/ends correctly - the peak shapes are too far from ideal. In my experience, the (subjective) manual corrections are often required and a deep knowledge of the system performance is necessary to discriminate baseline drifts or disturbances from a real presence of traces of a high or low molecular weight polymer.

The described approach may be quite legitimate (if performed consistently and used for the standards too) - in fact the effect may be similar to setting a higher threshold or slope for a peak detection.

[HW Mueller]

If one calls results "rough estimations" one could grudgingly agree to opinionating as desccribed. Now at least for molecular weight determinations there are light scattering detectors to alleviate problems. Maybe a different method could take care of quantitation problems.
Another example of what one may do if confronted with such severe problems is what I had to do when a light scattering, etc., detector was denied for our lab: The project was scuttled.

[wtgrey]

I would recommend a two-prong approach. First - the use of an absolute Mw detector (Light Scattering) to provide accurate Mw data for the entire distribution, and (2) improvement of the chromatography so that you can better resolve your sample and determine the Mwd of individual peaks.

Regards,

Todd

[LC101]

1st, Thank you for all of the comments.
More discussion with my colleague. He uses the Waters Millenium software and uses the "threshold" option at 1/10,000 of the total area of the peak. I'm not sure of any more of the details since I do not use the Waters software for analysis.
I used the word "truncate" to describe this. My thought is that this is still truncating. Yes?

Also, the material analyzed is polycarbonate. In the chromatogram using Waters HR5, HR4, HR3 and HR2 columns, the end of the peak does not come down to baseline due to additives, etc.
For myself, I have always visually, but consistently, determine the end integration cutoff. Usually at a valley point, right before an additive peak begins to rise. Is this acceptable, or could someone recommend something else.

[Koen Hollebekkers]

There are norms that define how to integrate GPC peaks (DIN, ISO). Rejecting certain parts is in general a bad idea. I would rather use blanks to define which are solvent peaks and which come from my sample