Correlating data from two GC's with big diff in baseline pA
Posted: Mon Jul 21, 2014 5:58 pm
Hi,
Working on validation of a headspace method where linearity was established on one instrument and accuracy done on another because instrument availablility was an issue.
My problem is that when using the linear curve established on system 1 to calculate the theoretical peak area response at 50, 100 and 200% of nominal on system 2, it results in low recoveries <80%. The 1-point calibration at nominal that I ran with my accuracy run indicates that I should be getting results close to 100% across the accuracy range tested (these are spiked samples using the same standard used for SS and bracketing).
One thing I noticed as a potential source of the discrepancy is that the baseline pA difference between the two instruments is quite large. System 1 was around 7 pA and system 2 was around 21 pA (this is on an Agilent 7980A).
My theory is that the large difference in baseline resulted in a lower sensitivity for system 2 relative to system 1 and thus the lower peak area response. In effect, I would have had to establish another linear curve on system 2 and use that if I wanted to calculate theoretical peak area response on that system.
If this is the case, what can I do to salvage the accuracy run apart from running it again with a multi point calibration? Can I use a correction factor taking the average SS peak area response and dividing it by the theoretical peak area response obtained using the linear curve data?
Working on validation of a headspace method where linearity was established on one instrument and accuracy done on another because instrument availablility was an issue.
My problem is that when using the linear curve established on system 1 to calculate the theoretical peak area response at 50, 100 and 200% of nominal on system 2, it results in low recoveries <80%. The 1-point calibration at nominal that I ran with my accuracy run indicates that I should be getting results close to 100% across the accuracy range tested (these are spiked samples using the same standard used for SS and bracketing).
One thing I noticed as a potential source of the discrepancy is that the baseline pA difference between the two instruments is quite large. System 1 was around 7 pA and system 2 was around 21 pA (this is on an Agilent 7980A).
My theory is that the large difference in baseline resulted in a lower sensitivity for system 2 relative to system 1 and thus the lower peak area response. In effect, I would have had to establish another linear curve on system 2 and use that if I wanted to calculate theoretical peak area response on that system.
If this is the case, what can I do to salvage the accuracy run apart from running it again with a multi point calibration? Can I use a correction factor taking the average SS peak area response and dividing it by the theoretical peak area response obtained using the linear curve data?