Comparing ICP Vendors And OES And MS Methods

[Dan_P]

@AlbertT,

@TomTheSeagull recently shared a link to an ICP study on Serious Reefs comparing seven vendor results for the same water sample

https://www.patreon.com/SeriousReefs/posts/reef-icp-test-ep-168983578?collection=2301031

Luckily, there was a link to their data (2-MS, 5-OES) which I reanalyzed. Note that I am not including nitrate, phosphate, alkalinity, or salinity in the analysis.

I first compared ATS to Oceamo MS results. The Log-Log plot shows a near perfect match of the results with a bit more variation at the lowest concentration. This was fortuitous because it gave me a benchmark to compare the OES results.

For the OES data analysis, I plotted the relative percent difference between the OES and averaged MS results ((OES - average MS) / average MS). The plot with elements ordered by increasing averaged MS concentration shows the OES results gradually deviating from MS results with decreasing concentration until silicon where variability explodes. Up to this point, the OES results are generally within 5-10% of the MS results. The negative 100% points are instances where the element was not detected.

The next relative percent difference plot has a change in X-axis from elements to concentration (μM), showing that the rapid increase in variability after lithium corresponds to concentrations below 10 μM. Note that Oceamo but not ATI OES maintains low variability to about 0.1 μM, corresponding to F in the previous plot.

What this analysis suggests is that an MS analysis is required for accurate trace element concentration below 10 μΜ (previously I proposed a 10 ppb limit) with the possible exception that Oceamo OES may provide useful information down to 0.1 μM.[/USER]

 

[Randy Holmes-Farley Verified Community Expert]

Thanks for the interesting analysis!

 

[JonasRoman]

@AlbertT,

@TomTheSeagull recently shared a link to an ICP study on Serious Reefs comparing seven vendor results for the same water sample

https://www.patreon.com/SeriousReefs/posts/reef-icp-test-ep-168983578?collection=2301031

Luckily, there was a link to their data (2-MS, 5-OES) which I reanalyzed. Note that I am not including nitrate, phosphate, alkalinity, or salinity in the analysis.

I first compared ATS to Oceamo MS results. The Log-Log plot shows a near perfect match of the results with a bit more variation at the lowest concentration. This was fortuitous because it gave me a benchmark to compare the OES results.

For the OES data analysis, I plotted the relative percent difference between the OES and averaged MS results ((OES - average MS) / average MS). The plot with elements ordered by increasing averaged MS concentration shows the OES results gradually deviating from MS results with decreasing concentration until silicon where variability explodes. Up to this point, the OES results are generally within 5-10% of the MS results. The negative 100% points are instances where the element was not detected.

The next relative percent difference plot has a change in X-axis from elements to concentration (μM), showing that the rapid increase in variability after lithium corresponds to concentrations below 10 μM. Note that Oceamo but not ATI OES maintains low variability to about 0.1 μM, corresponding to F in the previous plot.

What this analysis suggests is that an MS analysis is required for accurate trace element concentration below 10 μΜ (previously I proposed a 10 ppb limit) with the possible exception that Oceamo OES may provide useful information down to 0.1 μM.[/USER]

@Dan_P Impressive work, and a good reminder that ICP MS is needed to get accurate measurement of some traces. When I read about this a common info is that OES goes down to ppb and MS down to ppt. So all traces below 1 ppb is not reliable to measure with only OES. Do you agree?

Jonas

 

[Dan_P]

When I read about this a common info is that OES goes down to ppb and MS down to ppt. So all traces below 1 ppb is not reliable to measure with only OES. Do you agree?

Jonas

Yes, your statement is a good provisional conclusion.

Unlike the excellent agreement between vendors using MS detection, agreement between vendors using OES detection has always been poor below 10 ppb or to be generous 1 ppb. It is this high variability that leads me to suspect that 10 ppb might be close to the limit of detection for the method.

I should point out that In this study Oceamo has managed to provide results in reasonable agreement with MS detection down to 0.1 ppm. We would need more examples though before declaring Oceamo “best in class” for OES detection.

 

[Christoph]

Hello,

thanks for the analysis.

It is very difficult to give a general "cutoff concentration" for when ICP-MS should be used instead of ICP-OES. The achievable detection limits depend strongly on the individual element, the emission line used, the instrument, and of course the sample matrix.

Some elements can be measured extremely sensitively by ICP-OES. A prime example is manganese: it has several very strong emission lines that allow reliable measurement even at quite low concentrations.

Other elements, such as rubidium, iodine, and several others, have much less favorable emission lines and/or suffer from spectral interferences. Consequently, their detection limits by ICP-OES are considerably higher, and ICP-MS becomes advantageous at much higher concentrations than it does for elements such as manganese.

So rather than having one general concentration cutoff between ICP-OES and ICP-MS, this really needs to be considered individually for each element and matrix.

All the best,
Christoph

 

[Dan_P]

Hello,

thanks for the analysis.

It is very difficult to give a general "cutoff concentration" for when ICP-MS should be used instead of ICP-OES. The achievable detection limits depend strongly on the individual element, the emission line used, the instrument, and of course the sample matrix.

Some elements can be measured extremely sensitively by ICP-OES. A prime example is manganese: it has several very strong emission lines that allow reliable measurement even at quite low concentrations.

Other elements, such as rubidium, iodine, and several others, have much less favorable emission lines and/or suffer from spectral interferences. Consequently, their detection limits by ICP-OES are considerably higher, and ICP-MS becomes advantageous at much higher concentrations than it does for elements such as manganese.

So rather than having one general concentration cutoff between ICP-OES and ICP-MS, this really needs to be considered individually for each element and matrix.

All the best,
Christoph

Thanks Christoph for the clarification. I seem to have oversimplified things.

Dan