if calibration curve is not linear, can I still use it?

[zimanli]

I noticed people were talking about quadratic calibration curves on non-polar column. I had the same problem. I tried to play with inlet and interface temperature and even ramp temperature, but was not able to get a linear calibration curve. If we can not conveniently change to another column, can I use the quadratic curve to quantitate my real sample?

thanks a lot,
ziman

[CE Instruments]

By definition calibration curve
Ideally you should have a straight line however it is not unusual to get non linearity at the ends of the line. If the calibration is actually a curve then you should make sure you have sufficient points that you are confident that any compounds you quantify are accurately defined by the calibration curve. With a straight line you can get away with just three calibration points with a true curve you will need a lot more.

[Peter Apps]

You can use any shape of calibration that you want, as long as you have enough points - the sharper the curve the more points you need.

BUT - can you be sure that the mass - signal relationship that you get from the standards that you use to construct the curve will be the same for the analyte in a real sample. In this particular case the assumption that they will be the same might not be a robust one because there is a good chance that the adsorption of analyte that is causing the deviation from linearity with the standards will not be same when you run samples.

You need to spike samples with analyte and determine recovery (using the quadratic curve calibration) at high, medium and low concentrations. If it stays the same you are OK.

Peter

[zimanli]

thank you for your reply.
My curve is not non linear at the end, but at the beginning, then when the concentration is bigger, it is more linear. slope of the tangent is increasing instead of decreasing according to increasing concentration. Tt is flat at low concentration and steep at high concentration. Is that due to the active adsorption spots in my non-polar column?

thanks a lot for your time
ziman