Ca Measurement in Lithium Samples - Standard Curve Contamination from Matrix Matching - Blank Subtraction - Agilent 5110 ICP-OES w/ ICP Expert version


We're using Agilent articles 5991-9507EN, 5994-5561EN, and 5994-5149EN to develop a method for the analysis of calcium and other metals in lithium carbonate. We prepare high TDS samples by the dissolution of Li2CO3 with concentrated nitric acid then dilution with 2% nitric acid. 

To correct for the High-TDS samples, we matrix match by addition an aliquot of high purity lithium solution to each one of our calibration samples to build a standard curve of 0.1ppm, 0.5ppm, 1.0ppm for all of our elements of interest. Additionally, we are running everything with a continuously supplied internal standard (Y (371.029 nm)). We also are using the QC portion of the software to run QC Spikes to verify the accuracy of the method. Our QC spikes are at 0.1ppm. 

The issue we are running into is that the lithium solution we use to matrix match has a similar amount of calcium as some of our samples, so measurements on the low-end of the concentration curve become innacurate. Our QC spike shows concentrations of 0.3 ppm in the solution instead of the 0.1ppm it was spiked with (All other elements, including Na, K, etc, have reasonable and acceptable spike recoveries). I've remade the standard curve several times as well as remade the sample dilutions to be certain that it isn't contamination, and I don't think that it is. 

I have not found anywhere in the software where I can subtract the blank from the calibration curve. The QCBLK function appears to only subtract the blank from the samples, when it is my standard curve that is contaminated. There is no way to avoid this contamination in the standard curve.

Is it possible in the ICP expert software to subtract the blank from the calibration, not just to not include the calibration? 
Is there a way to do this analysis that isn't the method of standard additions? 

(Screenshot of current analysis, with samples and QCSPK samples)

iScreenshot of Analysis  

  • Are Run #10 and Run#11 Lithium samples that you spiked with 0.1ppm Ca?  Or are they DI water with some nitric acid only that you then spiked?    When you ran Run#10, it came out to be"0", which means it has the same amount of Ca as your calibration blank. However, when you ran Run#11, it came out as 0.1ppm which means it has 0.1ppm more Ca than the calibration blank.  Ca can also be a contaminant in DI water or in nitric acid. 

  • Runs #10 and #11 are lithium samples.  the "QC Spike RUN#10" and "QC Spike Run#11" are the spiked samples. The QC Spikes were made with 0.1ppm Ca, but upon analysis they show 0.21 for the spiked sample of run#10 and 0.31 for the spike sample of run#11. 

  • I would recommend you confirm the source of your Ca contamination. Use a second wavelength of Ca for added confirmation, use only Type I DI water, high-purity nitric acid, and pre-rinse your sample tubes. You can run a tube of only DI water, and look at the spectrum to see if there is a peak of Ca there. That will tell you if Ca is in the DI water. If not, then run a tube of 2% nitric and see if a peak of Ca is there. If so, that will tell you it's coming from the nitric acid.  You can also remove the peristaltic pump tubing from the solution, and run "air" and see if it's in your sample introduction system. 

    If the Ca is coming from the Li2CO3, there is no blank subtraction option in the software. You would need to quantify how much Ca is in the Li2CO3 by performing an aqueous external Ca calibration with an internal standard and Ca standards (without any Li2CO3). Then dilute the Li2CO3 "blank" and run it as a sample to quantify the Ca. Then you can subtract that value from your matrix-matched sample results. The reason you need to dilute the Li2CO3 "blank" is because Li in high concentrations causes a severe matrix suppression. So you need to diulte to get an accurate Ca value. Again, use 2 wavelengths for added confirmation. 

Was this helpful?