Organic ICP-OES Phosphorus FACT Modeling Application

Our lab is currently operating a 5000 series ICP-OES using an organic matrix (kerosene solvent) for detection of phosphorus (213.618) in a corn oil matrix. We currently apply the FACT correction for P with a dilution of an Agilent A21 metals standard (placed as analyte correction, and the FACT blank being our 0ppm standard from that day), but have found that Cu may be contributing negatively to the results of P under FACT correction, correcting our results to a value relative to the copper interference within the sample. We have not previously included Cu as an analyte, but we are now seeing that there may be spectral interference that the A21 sample exacerbates with the presence of Cu in the analyte correction matrix, effectively minimizing our results due to the lack of a copper peak in our samples.

The following images are from our P analyte under FACT correction. The 0ppm sample displayed was used for FACT blank. The top two graphs are examples of how we are currently applying FACT, and the bottom two display what is believed to be the correct FACT modeling application. We would like to know if the bottom two images are the correct application of FACT, or if the top two appear incorrect.

More information about this can be provided upon request, and any advice or considerations will be appreciated. Thanks for reading,

Kyle Bell

  • The Cu peak is 213.598nm and the P peak is 213.718nm. That means the Cu peak is 20 picometers to the left and the resolution of the instrument is very good so you can actually obtain baseline resolution of the 2 peaks with fitted background correction. Therefore you really do not need FACT, you can use fitted background correction. With FACT, it is dangerous to use a multielement standard for both the analyte and interferent models. You really need a single element P solution in kerosene for the analyte model, and a single element Cu solution for the interferent model. Here is a calibration curve for P 213.718nm using organometallic multielement standards in an organic matrix and Fitted background correction. 

  • Thank you for the response, Tina. When we began working on our method, we were recommended FACT due to matrix effects from the oxygenated nature of our oil. Do you think this is not a factor for this specific analyte at this wavelength? Also, at sub-ppm measurements, the P peak seems to drift upward in wavelength. Would FACT with the correct interferent/analyte solutions be better than fitted for sub-ppm measurements due to the nature of the curve? And one last question, I see the P curve you provided is a quadratic regression. Would a quadratic regression be a good fit for the range of 0.1 - 10 ppm, or is the linear regression model sufficient? I am not familiar with different regressions using ICP-OES, so any insight would be appreciated. 

    Thank you!
    Kyle Bell

  • When performing sub-ppm analyses in an organic matrix, yes the background structure of the ,carbon-based matrix can become an issue. FACT can be used to remove some of this background structure, but you need good models. So I would definitely recommend single element solutions for both your analyte and interferent models. You can easily change from fitted to FACT to see which one is better, and it may be wavelength dependent. If you have done a recent wavelength calibration, and the temperature and pressure in your lab is stable, then you should not see peaks "drifting". It could be that you have interference from background structure. Also, on the configuration tab of your worksheet, make sure "Drift correction" is OFF. This is very important in organic analyses because carbon emission wavelengths may interfere with the argon drift correction lines used for correction, and this may cause an apparent shift in wavelength position. 

    With respect to which regression fit is better, experiment with different ones to see which is better. I would expect 0.1 to 10ppm to be linear for most wavelengths, as long as you have "weighted fit" checked. I hope this helps! 

Was this helpful?