LOD/LOQ varian galaxie

[bublle]

how to determine LOD/LOQ in varian galaxie system?If you run a blank run and get a sinal to noise ratio at certain minute where your peak would elute-do you use that hight and put it in a calibration equation for that component (x3 and x10)?
How to get signal to noise area instead of height?
I measure area in calibration and so far I got only peak height from my run and thats not the same.

Does anyone work with that software?Is there a possibility to make it automatically?
How to resolve that?

Thank you in advance

[gpronger]

Is you analysis for a specific protocol (EPA, FDA, etc)?

The EPA recognizes two fundamentally different analytical techniques; those that have a "zero" signal level (typically spectroscopic techniques such as ICP) and those that do not (GC and GCMS).

For the former, you can measure background signal and use that to calculate a standard deviation which you can then relate to whatever signal to noise is appropriate (I've seen 3X and 10X used).

For the latter, what is done (for EPA work until they finish the re-work of their MDL protocol) is you run a minimum of 7 spiked samples (through whatever prep may be utilized) spiked at what near what you believe the lowest level you can detect. You quantitate these runs, and calculate the standard deviation for each analyte. You then multiply the standard deviation by the "Student-t" value of the analyses. The protocol can be found at:

http://ecfr.gpoaccess.gov/cgi/t/text/te ... .2&idno=40

Greg

[bublle]

ISO standard is the answer!
I would like to avoid diluting sample to the lowest concentration visible by gc and calculate it from signal to noise ratio.
there should be some way of reporting signal to noise ratio in area units
and not in height units.
if it would be so, the rest is easy

[gpronger]

In searching the internet I do not see a specific protocol for calculating the SNR for area. So, I would create an SOP documenting how my lab was doing it. My recommendation would be, using the same integration thresholds, let the system integrate a blank. Calculate an average noise area in a window around the RT of the peak I'm working on (say =/-0.5 min). Use that as my noise. You tn need to "lock-down" the threshold to that level for your analyses going forward.Greg

[Peter Apps]

If you calculate LOD or LOQ as simple multiples of the baseline noise at the expected retention time you get a hopelessley optimistic figure for the performance of the method as a whole, because it does not take into account any of the variability in sample prep or calibration. The EPA approach gives by far the most robust and "honest" estimate for the performance of the whole method.

Peter