Chromatography Forum

Hello to all,
I'm new in Ion Chromatography. I have a problem about the building of anionic calibration curve by Chromeleon 6.7. I need a linear and quadratic curve with external standards.
My operations are: Preparation of 3 known concentrations from stock solution. The stock standard solution (ThermoFisher) contains 7 anion (Fluoride 20mg/L, Bromide 100mg/L, Chloride 100mg/L, Nitrite 100mg/L, Nitrate 100mg/L, Phosphate 200mg/L and Sulfate 100mg/L) in 100ml bottle. So, I preparare 3 dilutions from this standard solution 1:1, 1:5, 1:10 (500um stock + 500um water; 1ml stock + 4 ml water; 1ml stock + 9ml water). After that, I inject to obtain the runs, I create my Peak Table where I name the seven peaks and I add amounts in the three coloumns. The integration is set to the disabilitation of Negative Peak. At this level, I obtain my curve where only the 1:1 solution is on the line, but the other two are outliers, why? If it's useful to know, the eluent is 9mM Na2CO3, the coloumn is AS9-HC and the suppressor is SRS 300 (4mm).
Can someone help me to understand my mistakes? Please, thank you!

Hi there,
A couple of questions

What is the injection volume?

Have you tried fitting the calibrations to linear with offset?

Do you see any analyte peaks in the blank?

Yours
Kevin

The inject volume is 25um. In effect, I have never tried with offset and I notice some small peaks in the blank run. If you have other information, I'm very pleased to learn them.
Thank you very much!
Marianna

Hi again,
A couple more questions.
Are you using an auto sampler to inject the 25ul. If so can you let me know the model number.
Could you email me PDFs of the chromatograms showing the peak and the peak areas for each standard and the blank. If you have done repeat injections could you email me those as well.

My email is ksoulsbu at bcit.ca

Thanks
Kevin

A quadratic fit with only three points would not be very accurate as you can not know if the points are linear but just a little off the line or are truly quadratic. Most methods will say you need at least six points for quadratic to be a good model, though five will usually work.