when my calibration passes through zero, r equal 0.87 while when it does not pass through 0, r=0.99.
My question is simple:is it an obligation that my standard cuve passes through the origin?
By Anonymous on Monday, May 3, 2004 - 08:33 am:
I do not force my equation through the origin, but I usually use 0,0 as one of the data pairs when calculating the line equation. I also make sure the standards I use bracket my sample concentrations.
By Greg on Monday, May 3, 2004 - 07:03 pm:
When you inject a concentration of 0 into your GC, you will not get a signal of 0, there will be signal attributed to noise present where your compound elutes. 3 times this noise is what is termed the "detection limit" and as such a cal curve may not go directly through the origin.
By Laurent on Tuesday, May 4, 2004 - 07:46 am:
I'll say it depends of the range you are working in.
If you have 5 points 100/200/500/800/1000 ppm forcing your curve trough (0,0) will not change significativly your rē
Do you use external standards ? this kind of problem generally appear when your preparation is too old (in the case of ES not IS).
Prepare fresh standards and reinject.
By Anonymous on Tuesday, May 4, 2004 - 10:30 am:
Actually, if your standard curves are based on amount vs. integrated peak area or height and you inject no compound, you will not get a peak area or height. Therefore 0,0 is a valid point if the investigator chooses to use it.
By greg on Tuesday, May 4, 2004 - 10:43 pm:
if you do inject "0" analyte, and you obtain a peak with "0' area and "0" height, how do you determine the detection limit according to the rule, 3 times signal:noise ration, as 3 times 0 = 0!!!???
Take a look at a chromatogram when you inject a blank and i will absolutely guarantee you there will be a peak present, albeit very very small, which is actually noise. This is the background/baseline, which does has noise attributed to it.
Sue one can choose to use 0,0 as a calibration point, but this is quite a nonsensical value.
By Anonymous on Wednesday, May 5, 2004 - 12:45 am:
Thank you all for your precious information
By Anonymous on Friday, May 7, 2004 - 12:11 pm:
A curve does not necessarily have to pass through zero, but your curve seems to have an ususually large offset. To go from an r2 of .99 to .87 is unusual. If I see that large an offset I will generally make up new standards and make sure all the glassware, etc. is clean. Don't forget to clean syringes thoroughly between injections, and go from low concentration to high. The only time I might expect to see that much offset is if I am using standards spiked into a matrix.
One other point, when calibrating the same volume should always be injected unless a solvent flush technique is used. If you inject nominal 1 and 2 microliter volumes using a 10 microliter syringe the real volume injected are approximately 1.5 and 2.5 microliters. This will have a definite effect on the intercept.
By M.Mandjikov on Thursday, July 1, 2004 - 11:30 am:
Our methods forbid us to force the calibration through zero. I was told that this is because 0,0 is not a "real" point. To improve my low points near my detection limit, I shoot a couple very low standands--below where I need to report and make "real" points near zero.