Fast all the elements uncalibrated, but zinc calibrate

Hi everyone,

I measured some methanol samples, with ethanol diluted. Most of the elements are uncalibrated, but zinc is calibrated.

I have two questions:

1. I thought it is because of the transportation of sample. But when the sample was not stable transported, why did I get the calibrated zinc? What could be the problem here?

2. When I want to do the elements screening, to measure many elements together, is it the reason of wavelength interference? If I did not choose so many elements, is there still wavelength interferece?

Thank you very much!

Best regards


  • Hello ,

    Great application question here.

    1) What are the elements you have in your worksheet, and the calibration range?

    Did you check if removing the some points (outliers) you can get the calibration? Sometimes you have contamination or poor counts for initial points for other elements and another ones are OK (such as Zn).

    2) No, this is not the problem, you can measure as many elements you want in the same worksheet without any issues. The challenge might be choosing the correct wavelength for each element since you have a mixture of analytes and you might get spectral interferences. As a tip, when choosing all elements and wavelengths, go through each element to see if there are interferences close to the emission WL you chose. If so, I recommend choosing another WL or using FACT to deconvolute the spectral interference.

    Did you try to run an IntelliQuant to use is a screening tool? You have a 5800 right?

  • Hi Rodolfo,

    Thank you very much for your reply. Yes. We have 5800. I did not run an InterlliQuant. I will try it. I am a beginner of ICPOES. Many tools I have not used.

    I have only Fe, Cu, Al and Zn. Standard is 1, 2, 5, 10, 50 ppm. Is this what you mean the calibration range?



  • Hey Sophia,

    No worries, each day we learn a little bit more - takes time indeed.

    You can run IntelliQuant within your worksheet - you have to check the "IntelliQuant" in "Configuration" in the worksheet. Each sample measured will be scanned for all possible elements, and you get a semi-quantitative result, plus insights of possible interferences.

    That's right for the calibration range. You are doing it axially or radially? In axial mode you will have more signal and better detection limits, and in radial mode you can get a better linear dynamic range.

    Are you mixing monoelemental standards to make an intermediary solution? Are they ICP grade? Or are you adding them separately?

    I'm asking this because if the standards are not ICP grade, you can bring contamination to your calibration curve. If you make a multielemental solution and then transfer them to your volumetric flasks, then the transfer would not be the issue, since Zn is OK. If doing it separately, you can have a problem.

    Have you tried adding two or three wavelengths of each element, so you could check if the calibration curve profile is the same (I mean, the points have the same position in different WLs)? This is a good option to isolate the problem.

    What are the calibration criteria (maximum error, R...)?

    If you want, share some images of those elements whose calibration curves are not OK.

  • Hi Rodolfo,

    we used axial mode for the measurements. 

    Standard is ICP grade and also a multi-element standard. I have previously used the other multi-element standard in mineral oil for organic samples. There was no problem. Should I put these water standards in the refrigerator? I just keep it in a cabinet for chemicals in the room temperature.

    Here are the results of 4 wavelength of aluminum and the condition for the measurements. I thought maybe I did not dilute the standard properly. But most of the elements are calibrated. I am not sure where is the problem. Could you please help me? Thank you so much!

    Best regards


  • Hey Sophia,

    I can see that the reason that you cannot get the calibration for Al is because the error for some points are too high. Curiously, it works for 2 wavelengths (WL). Make sure you drag the red line to the peak of the spectra (I can see that they are a bit aside).

    The 237.312 nm Al line seems to have an interference. I would not consider this WL for your analysis.

    You can store the standards in the cabinet, no problem. Sometimes people store them in the refrigerator, but honestly I don't think it is necessary (they come to you in room temperature, so it is not critical).

    I think the most reliable WL for this case is 394.401 nm, you can see that you have good percentage recovery for a 2 ppm standard. Keep in mind that Al suffers a lot from contamination, and you should track all the preparation process to avoid any source of Al (for example, decontamination baths, any metal apparatus used to clean the glassware, etc). Overall, you can see that the high error is mainly occurring in the lower concentrations, which can be related to contamination (even with the 394.401 nm WL).

    Hope this helps!


  • Hi Rodolfo,

    thank you for your reply. It helps me a lot!

    For me it is very strange, two WLs of Al are calibrated, two WLs are uncalibrated.  I think because I did not prepare the standard solution very well. I will prepare the standard again and measure again.

    Are there maybe some guidelines for the preparation of standard solutions? Like maybe with 5 different concentrations better than 3 concentrations?


  • Great Sophia, re-prepare your standards so you can assay this with more data.

    Regarding the guidelines, this is the "million dollars" question. Some labs run just the blank and a standard. Others run 10 standards. It depends on your quality protocol.

    Overall, 5 points plus the blank is unofficially considered the standard procedure. From the perspective of statistics, more points might lead you to a better regression model, but we know that working with "infinite" calibration points is not possible (from the economical perspective). When you go through the validation process of the method, you can work with a reasonable quantity of points (again, normally 5) to achieve a good performance (lower uncertainty) without spending too much time and money.

    Let's say you need to quantify the minimum possible concentration of Cd in a water sample, to comply to regulatory documents. If the stated limit is 1.0 ppb, and the quantification limit of the 5800 is 0.2 ppb, you can make 5 calibration points around 1.0 ppb (for example, 0.5; 1.0; 2.0; 5.0 and 10 ppb. By doing this, you have more points around 1.0 ppb (which makes the regression more effective around this concentration) plus get an extension if your sample goes above the stated limit.

    Another case: you have a sample that you know the Ca concentration will be around 20 ppm. You know the 5800 can achieve very low concentrations of Ca (guessing 0.1 ppb), but there is no need to run this sample on a calibration curve just as Cd mentioned above. You can do a calibration curve from 10 to 50 ppm in radial mode in a less sensible Ca emission WL, adding closer points to 20 ppm. Got it?

    Knowing the sample and the analyte concentration is the key to elaborate an assertive calibration curve. In the case of not knowing the analyte concentration, you can work with more points OR run the IntelliQuant to screen the sample and know if it is the case of running on lower or higher concentration calibration curves.

    Think of the 5800 as a versatile tool to combine the sample introduction system, optical view and different wavelengths to work with ppb or % levels of concentration. Some labs run worksheets with two or more conditions to do ppb and % levels at the same run, depending on your demand.

    I have some articles on LinkedIn explaining these tools, but they are in Portuguese. You can check if you can translate them using LinkedIn or Google Translator, might help Slight smile

    Best regards,

Was this helpful?