Inconsistent GC MS VOC Performance

[wtlljf]

My lab has a Perkin Elmer Clarus 500 GC MS that we use almost exclusively for VOC panel analysis. It is relatively new (purchased 2-3 years ago) and was programmed by a Perkin Elmer specialist to give us the 'best recoveries' for our particular analysis.

The problem is it provides very inconsistent analyses. Day one it will work fine, day two it won't provide even a relatively presentable continuing calibration curve, and then on day three (without ever having changed a thing) it will work fine again.

Also, even within a 'good' run, there are inconsistencies between duplicates. If 100 mLs of standard is mixed and two separate 40 mL aliquots poured, one will be higher than the known concentration of the standard, and one will be lower.

I should also mention that sometimes, in known proficiency samples with 21 analytes, it's generally 16-18 that are recovered in the desired range, while the remaining few are difficult. From this, I would suspect that it's NOT an integral problem with the preparation or analytical method or else the analysis would be off on all counts.

The Perkin Elmer tech is supposedly going though hours of data to try and find a problem...any suggestions from the group on what might cause this kind of rampant inconsistency?

[AICMM]

wtlljf,

You don't say if this is just a certain class of compounds (like the gases or ketones) or if it is more random than that. You also don't say if you are using a purge-and-trap or a headspace autosampler or direct injections.

If you are using a purge-and-trap my first objective would be to eliminate that as your source of variability with some split injections by hand of one of your standards. Depending on what that reveals I would take it from there.

Best regards.