GC Performance drop and ISTD calibration procedure

[fangs]

Two months ago I prepared solutions of benzene in methanol at different concentrations ranging from 25mg/L up to 400mg/L and I used ethyl acetate in methanol at 500mg/L as my internal standard. Back then I did 3 injections of each benzene solution with my internal standard to get a calibration curve with the ratio of area of benzene peak to area of ethyl acetate peak as my y-axis and benzene concentration as my x-axis. The curve had a good linear fitting with R^2=0.993.

I had to make the same calibration curve a couple of days ago and I prepared fresh solutions of benzene and ethyl acetate and ran the injections, except I did two injections this time per solution rather than three. All the GC settings are the same, I used the same chemicals as before and also the amount of chemicals I used was the same as before. However, this time the calibration curve isn't as good this time around, the R^2 value is lower and it's 0.971.
What caused the change in the calibration curve fitting, how can I check what is the culprit?

My second part of the question is: what is the typical procedure when making internal standard calibrations in terms of number of injections? And is it proper to use my calibration curve from 2 months back?

Thank you.

[Don_Hilton]

As you can see from your current results, the response of an instrument changes over time - so no, you can not use the calibration curve from two months back. What can cause the curve to change? Many things.

First steps are to be sure that proper inlet and column maintanance are performed. And, if you used an old solution containing methanol and ethyl acetate, look for the methyl acetate peak! You will get transesterification with this mixture.

Ethyl acetate would not be my first choice as an internal standard, by the way. Ideally you would like something chemically similar - My first thoughts would be either furan or toluene -- fairly nonreactive aromatic compounds.

[chromatographer1]

I agree with Don.

I would use another solvent instead of methanol to make my std solutions. Pentane or acetonitrile or DCM ? ? ?

Toluene as an internal std would be good idea. I would not use an ester in the presence of an excess of alcohol with water present.

Benzene is very volatile in a polar solvent such as methanol or water. Losses can be quick and unexpected.

best wishes,

Rod

[fangs]

I have a further question regarding ISTD calibration curves.

I've been doing 3 injections per concentration of analyte with a fixed amount of ISTD. And for my calibration curve I'm plotting ratio of area of analyte to area of ISTD on the y-axis versus the concentration of analyte on the x-axis.

So when reporting the calibration curve do I simply plot the ratio of areas for all the trails versus the corresponding analyte concentration or do I have to take an average of the ratio of areas versus the concentration?

[Don_Hilton]

I prefer a regression throught all points collected. If you use a least squares regression, it makes some assumptions about variability. A simple regression can be computed in a spreadsheet - or even on some hand-held calculators. Weigted regressions are a bit more complex, but may be found in many chromatographic data systems. There are arguments for each type of regression. (And I believe yoyu can find discussion multiple places on this site.)