Correctness and repeatability of ICP-MS seawater measurements

[taricha]

Skeptic or not, how do you decide on the amount of trace elements to dose or whether you have an elevated pollutant when the data can be rather uncertain?

Certainly don't send samples to a bunch of ICP vendors. That way leads to madness

 

[Reefahholic]

Maybe other technical things matter more across these elements than just having an ICP-MS machine?

Absolutely

 

[Randy Holmes-Farley]

Absolutely

That was certainly my experience 20 years ago. At least when I did it, it was not plug and play.

Electronic Calcium Monitoring by Randy Holmes-Farley - Reefkeeping.com


For initial testing I chose to use as the "standard" a sample of artificial seawater that was mixed to an approximate salinity of S=35. I mixed a 44-gallon batch using Instant Ocean artificial salt mix and reverse osmosis/deionized (RO/DI) water to a conductivity of 52.7 mS/cm, and allowed it to settle for three weeks. I then proceeded to measure its calcium concentration by ICP-AES (inductively coupled plasma-atomic emission spectroscopy, an $80,000 analytical instrument. I was somewhat disappointed with my inability to use this sophisticated technique to get a precise answer. Despite taking five different samples and analyzing them at eight different emission wavelengths using two different calibration methods (five standard additions of known calcium concentrations to each sample, as well as comparison to a fixed 1000 ppm commercial calcium standard), I was unable to get consistent values. Some of the samples were acidified or filtered through submicron filter membranes to determine if solid materials were impacting the result (they were not). Overall, I took more than 200 measurements, each involving three replicate observations of the emission intensity. Nevertheless, the result was not very satisfying, with a substantial variation occurring between the different values. The average of every measurement taken was 336 ppm. With the uncertainty involved, however, I'd conclude that the true value was probably 340 ± 40 ppm. I also measured the same sample once with a Salifert brand test kit and got 330 ppm calcium.

 

[Reefahholic]

That was certainly my experience 20 years ago. At least when I did it, it was not plug and play.

Electronic Calcium Monitoring by Randy Holmes-Farley - Reefkeeping.com


For initial testing I chose to use as the "standard" a sample of artificial seawater that was mixed to an approximate salinity of S=35. I mixed a 44-gallon batch using Instant Ocean artificial salt mix and reverse osmosis/deionized (RO/DI) water to a conductivity of 52.7 mS/cm, and allowed it to settle for three weeks. I then proceeded to measure its calcium concentration by ICP-AES (inductively coupled plasma-atomic emission spectroscopy, an $80,000 analytical instrument. I was somewhat disappointed with my inability to use this sophisticated technique to get a precise answer. Despite taking five different samples and analyzing them at eight different emission wavelengths using two different calibration methods (five standard additions of known calcium concentrations to each sample, as well as comparison to a fixed 1000 ppm commercial calcium standard), I was unable to get consistent values. Some of the samples were acidified or filtered through submicron filter membranes to determine if solid materials were impacting the result (they were not). Overall, I took more than 200 measurements, each involving three replicate observations of the emission intensity. Nevertheless, the result was not very satisfying, with a substantial variation occurring between the different values. The average of every measurement taken was 336 ppm. With the uncertainty involved, however, I'd conclude that the true value was probably 340 ± 40 ppm. I also measured the same sample once with a Salifert brand test kit and got 330 ppm calcium.

No doubt it would be difficult. I couldn’t do it.

Christoph says…the machine itself is not the only definining factor for quality results. It’s the operator who develops and optimizes methods, based on chemical understanding also regarding the formation of polyatomic ions.

 

[Dan_P]

Certainly don't send samples to a bunch of ICP vendors. That way leads to madness

That explains the disorientation akin to walking through a house of mirrors I feel when I look at ICP data.

 

[Christoph]

But Oceamo (ICP MS) and Fauna Marin (ICP OES) are different types of test, correct? And can they be overlaid / compared if so? I don't know thus asking. Not suggesting you do this but grouping Oceamo and ICP-Analysis and then Fauna Marin, Triton, and Reef Labs may show something different.

With regards to skeptics there isn't anything wrong with that.

You noticed that fun nugget. ICP-A and Oceamo both are doing icp-ms test, so you would expect those two would hang together tighter than any other pair of vendors. This turns out not to be the case. FM was actually closer to Oceamo across these 100 ppb range and below than ICP-A was.
What does that suggest? I'm not sure. Maybe other technical things matter more across these elements than just having an ICP-MS machine?

Hi Everyone,

it is not surprising OES and MS data matches, when the concentration is measureable with both techniques, and the analyses are done right.

The technique (ICP-OES vs ICP-MS) just defines the possible limits. But many factors need to be considered, if those performance limits can actually be reached or not. For example the lower limit of detection is also limited by the purity of your blank ("0")-solution. If your blank contains more of a certain element then your sample (because of a not so pure environment or suboptimal handling or methods), the element will always be undetectable in the sample, regardless of the detection power of the instrument. This is why the surrounding and "howto" is so important.

To give you a visual comparison: Imagine a Formula 1 racing car (ICP-MS), and a solid Ford truck (ICP-OES). On a dirtroad the truck will actually perform better. Likely also a inexperienced driver will perform better driving the truck. Furthermore the formula 1 racing car is not very useful with the wrong tires attached. Its only on the racetrack with the correct tires and the experienced driver that the solid Ford truck doesnt stand a chance against the racing car ;-)

Also from the nice comparison @taricha posted, the advantage of ICP-MS can clearly be seen:



Its the very low-level elements that we can still measure with ICP-MS (with sufficient comfort), that are below the limit of detection for ICP-OES. Those elements include biologically essential elements (Se, Co, Cr, Cu, V,...). Only by measuring at those low levels, a dosing decision for those elements can be justified imo. Also very relevant is the "racing car factor" for pollutants such as Lead and other heavy metals: ICP-MS is especially sensitive for those "heavier" atoms while ICP-OES has usually much lower sensitivity.

Best regards,
Christoph

 

[Christoph]

Thanks for providing this data Christoph.

This is also a helpful clarification that you are using different materials for your calibration and this not circular re-measurement of the same material used to do the calibration.

Thanks @taricha for noticing this. It would be very poor practice to "validate" a calibration by measuring a calibration standard. The operator would not notice any issues, even if wrong by orders of magnitude.

That was certainly my experience 20 years ago. At least when I did it, it was not plug and play.

Randy, im very sure that 20 years ago ICP-OES was far more difficult to handle compared to modern spectrometers. Resolution on the optics got much better, so interferences are much less of a concern. Also operation became for sure much more stable, and i can imagine also the software must have been way more troublesome back then.

Since macroelements were also a topic in this thread, id like to show again some day-to-day stability data of remeasurement of the same solution. This time its not from a CRM, but from a control sample that we have established as SST (System suitability test) on our Cation IC. This sample is used as a "warmup" at the beginning of the sequence, and to also see if the system is in a suitable state for sample measurement. It is a in-house prepared sample that is made from puchased high quality single element standard solutions. This data is not coming from ICP, but from cation ion chromatography (cation IC), which we are using to measure those elements. Again i have not left out any datapoints from our last 23 measurement days,


Calcium %RSD over those 23 different measurement days: 0,75%


Magnesium %RSD over those 23 different measurement days: 0,68%



Potassium %RSD over those 23 different measurement days: 0,95%



Sodium %RSD over those 23 different measurement days: 0,93%

Again a correlation between all elements is visible without any statistics (@taricha) - the sample introduction is here the main factor, which affects all analytes.

I hoe you are finding this useful! Otherwise i think i have covered all questions from this thread so far. If more questions come to mind: Just let me know

Best regards from Austria, Christoph

 

[Reefahholic]

My ICP results are in. Would anybody from the other camp like to compare your last ICP results or current results publicly? Let me know. I want to see how far I’m lacking behind you guys.

 

[Christoph]

Hi everyone!

Because some pictures from our lab were requested here in this thread, we have made a small video to show you around:


All the best,
Christoph

 

[Dan_P]

Hello everyone,

jda is right, we should try to stay on topic here or it will be difficult to follow the thread. As mentioned before im usually on a small time budget, so i cannot answer all questions promtly- sorry for that. I will try answering the chronologically. Please let me know if i have missed anyone.



In my initial posting i have presented "hard data" towards consistency and correctness of our results, even in the low ppb (and sub ppb) range. Validation/checking of methodologies and calibrations using CRMs is something very established in the analytical community.

Regarding outside testing we have participated in such "ring testing rounds" regarding freshwater analyses. Unfortunately regarding seawater testing such modes of quality control are not offered (at least i am not aware of a provider). However we do have a mode of outside quality control currently under development, and will update the community once i am having news on this. For me it would also be much easier to have a certificate to show to all skeptics ;-)




This goes a bit into detail. ICP is a destructive method. The sample enters the plasma, everything gets evaporated, dissociated, excitated and ionized. So if you want to know the chemical form a certain element is in, you need to seperate the different forms before ICP-MS. This is usually done by coupling ion chromatography or HPLC systems to the ICP-MS, using the ICP-MS as element-selective detector. Those combined techniques are abbreviated IC-ICP-MS or HPLC-ICP-MS.

This is called speciation analysis. Typical examples for speciation analyses are

Cr(III)/Cr(VI) - not relevant in the reef tank
Iodide/Iodate - not really relevant in the reef tank
Bromide/Bromate - only in certain cases relevant for the reef tank

We do have the technical capabilities and equipment to perform those speciation analyses. However those runs are far from standard procedure and take quite a lot of analysis time. Cost will be in the range of several hundred USD per speciated element. Also you need to know beforehand which chemical forms you are looking for.

What elements are you especially interested to get speciated? For which elements do you think speciation matters for the reef environment? Feel free to contact me when you are having specific tasks/questions where we could help with speciation.



Thank you Dan! Unfortunatel you will most likely not like my answer to your question:

Asking for accuracy and precision for every individual element in each individual run is very easy - but to deliver this data in a meaningful way is unfortunately almost impossible.

The CRM/QK data i provided in my first post give a good indication of accuracy and precision (accuracy: how close are we to the specified concentration, precision: how high is the variability between results).

I could now take this data and say (just as example) for arsenic we have a %RSD of 6,5% (calculated from the standard deveation and mean value of above 18 data points). This 6,5% RSD would however not be valid in general for arsenic. If the concentration of arsenic in the sample would be higher, the %RSD would be lower (better statistics) - if the arsenic concentration would be lower, it would be the other way round. Also factors such as salinity have an impact, if a sample would have very high salinity, %RSD would generally be higher.

To provide precision data for each individual analyses it would be rquired to run every sample several times to allow for statistics. This is however very time consuming, and would thus also increase cost very significantly. It is impossible to give data on accuracy for each individual result, because i do not know the actual ("true") concentration. - Accuracy can only be checked with CRMs or other control samples (data shown in my initial post).

Data that i can offer is the replicate %RSDs: In ICP measurements (OES and MS) we are not measuring a single data point, but 3-5 replicates within a short time span. This replicates also allow to calculate a %RSD and thus give a hint towards precision. However the only error range that arises from this data is machine error (sample uptake, plasma fluctuations,...), and does not show any potential error that is arising from sample preparation or calibration. The CRM data (initial posting) does indicate errors from the whole analysis procedure. If you are interested in specific replicate %RSDs for your oceamo measurement, just let me know - i can look them up. Having this data on every analysis report would imo be overwhelming and not very useful.

I hope now it makes more sense to use things such as CRMs with repeated testing to assure quality in general.

All the best, Christoph

Christoph,

Validated analytical methods as used in the pharmaceutical industry and medical labs are capable of having their accuracy and precision established and described. Even the Hanna Checker manufacturer provides an indication of the accuracy of the results one can expect from the Checker. At this point, I feel that hobby ICP vendors while doing their best to calibrate their instruments are not using a validated method, ones in which the accuracy of the results are understood for every sample. I assume providing results from a validated method would increase the cost and since most customers just assume ICP are accurate, there is no business driver to do otherwise.