How We Use ICP-OES Results Of Unknown Accuracy And Precision

[A_Blind_Reefer]

This thread is for the general discussion of the Article How We Use ICP-OES Results Of Unknown Accuracy And Precision. Please add to the discussion here.

Interesting read. I appreciate the time and effort you all put into this, thank you. Wish I could read the charts, but I’m not able to with my vision loss. I did listen to the whole article though. I have always wondered about accuracy, especially taking i account for all the variables involved.

 

[rtparty]

All that went way over my head. I need the "Results for Dummies" breakdown. Just tell me which vendor was best overall.

With that said, I use the Reef Moonshiners method and use ATI for all ICP testing. I can say their results are usually spot on with my home test kits almost every time I send in a sample. I usually test for the big 5... alkalinity, calcium, magnesium, nitrate and phosphate. The only one that ever varies from my testing it phosphate yet ATI's results are definitely more accurate IMO. For example, my result showed something like .15 but ATI showed .01. The ostreopsis dinos in my tank at the time tell me the .01 was the more believable result. I believe the Hanna meters struggle if there is silica in the water? Which my tank runs high in for some reason.

On a more cycnical note, I wouldn't trust anything from Fritz or ICP Analysis. Been around long enough and seen far too many inconsistencies and poor practices from both to truly believe anything either says. Sorry, not sorry.

In the end, I'm going on 15 months of Reef Moonshiners and have never had a tank flourish like this. Andre has found something and the proof is in the pudding. There are thousands of members in his group running it with success. I don't need any testing or science to tell me what my eyes are seeing. But (and it's a big but) I never make a crazy change because one element came back with a reading that doesn't make sense or follow the trend. I'll test again in a month and see what it says.

 

[Rick Mathew]

Interesting read. I appreciate the time and effort you all put into this, thank you. Wish I could read the charts, but I’m not able to with my vision loss. I did listen to the whole article though. I have always wondered about accuracy, especially taking i account for all the variables involved.

Thank you very much. If it would help to send you a hard copy of the Article I would be happy to do so.

Rick

 

[Lasse]

But (and it's a big but) I never make a crazy change because one element came back with a reading that doesn't make sense or follow the trend. I'll test again in a month and see what it says.

Its the well known - not panic tactic - I whish more people could learn that

Lasse, what trace elements are you monitoring? Have noticed any of the elements depleting faster than others?

I monitoring all that I get from the ICP analyzes but I only dose Strontium, Boron, Iron, Iodine, Rubidium, Vanadium and Manganese on a regular base. Vanadium seems to disappear very much in my aquarium compared with others.

When I get back an ICP - I valuate the the advice I get and normally dose what´s is recommended but stay normally away from zink, tin and nickel. I´m low in Fluoride but have try to rise it slowly

Here is the ICP-MS test I have done. Note - it is nearly 5 years since last WC.

Sincerely Lasse

 

[taricha]

Based on what you found though, making corrections based on the ICP results seems a bit of a gamble or shall we say a leap of faith.

Well, if you were hoping to dial in your molybdenum, you're in luck. Molybdenum could be distinguished between 0 and 1 ppb for all for vendors we checked. Barium and Nickel were almost as good for some (but not all 4) vendors. However, if you were hoping to dial in your trace metals to low single digit ppb levels, you just don't get to do that. sorry!

 

[taricha]

All that went way over my head. I need the "Results for Dummies" breakdown. Just tell me which vendor was best overall.

Not really that kind of article, heh.
I would characterize the results to say 3 of the 4 vendors were similar enough, that I can't be confident the differences between them in numerical assessments are more than just noise in the data we analyzed.

Here is the ICP-MS test I have done.

I'm so glad ICP-MS is outside the scope of this article!
Oceamo is a smaller volume operation that the 4 we tested also, so it's difficult to decide which of our sources of variation I think would and wouldn't apply to them similarly.

 

[A_Blind_Reefer]

Thank you very much. If it would help to send you a hard copy of the Article I would be happy to do so.

Rick

Thank you for offering to do that, you are awesome! It’s not necessary, but I very much appreciate the offer. Again, thank you guys for all the work. Great job!

 

[MBruun]

@Lasse I just ordered an ICP test from OCEAMO to try it out.
The last two I used was from ATI but I found that the target / recommended values was calculatd based on the measured salinity (I asked them) but I'm certain that ATI measure it wrong. I have a lab-grade aerometer (flouting hydrometer) and a refractometer.
Do you know if OCEAMO also calculate the target vales based on the salinity ?

Best regards
Morten

 

[Lasse]

Do you know if OCEAMO also calculate the target vales based on the salinity

For me it looks that way @Christoph can confirm

Sincerely Lasse

 

[Christoph]

Hello @MBruun,

thanks for ordering Oceamo sampling kits!

we measure salinity using a conductivity probe. Target values for macroelements are correlated to salinity, target values for trace elements are not.

Let me know if youre having any questions!

Best regards,
Christoph

@original posters: Great work!!

 

[Randy Holmes-Farley]

Hi Rick,

I'm working my way through the article and think this sort of work is very important and useful. Thanks to you and others who helped!

Edit: I had some questions, but I decided to hold them until i finished the article since they may be answered later.

 

[Rick Mathew]

Hi Rick,

I'm working my way through the article and think this sort of work is very important and useful. Thanks to you and others who helped!

Edit: I had some questions, but I decided to hold them until i finished the article since they may be answered later.

Thank you very much...Will be happy to answer any question...Make your day a great one

Rick

 

[Randy Holmes-Farley]

OK, thanks again.

A few questions.

How was chart 4 generated? Where did the high and low values come from? The salinity also seems pretty high in it.

I'm not fully sure how the different tests fit together.

Let's look at iodine. ICP-A has a red code at 1 ppm, so presumably it had an iodine reading on at least one sample that was close to 1 ppm? But then in chart 5, ICP-A values for iodine are shown as having a bit better than moderate precision?

 

[taricha]

I'm not fully sure how the different tests fit together.

Let's look at iodine. ICP-A has a red code at 1 ppm, so presumably it had an iodine reading on at least one sample that was close to 1 ppm? But then in chart 5, ICP-A values for iodine are shown as having a bit better than moderate precision?

Here's some details that adds context for how these results got generated. This was from samples spiked with 4 different levels of trace elements and sent to 4 vendors.

(Red represents Rick's method based on Red Sea chemistry with +-20% error bars for scale )

It's difficult to look at this data and say ICP-A was meaningfully measuring Iodine at all in these 4 (trace-spiked) samples.
So even a reading of 1000ppb Iodine from ICP-A does not assure you that it's "real". It could be - and was - only around a tenth of that value. Hence Red square on CHART 3 for ICP-A even at 1000ppb I.

And this wasn't a one-off. Another data set that did a set of dilutions between I.O. and Brine also contained data where Iodine was ~120ppb in the I.O. and ~60ppb in the Brine, but mixing them together generated a value of ~330 ppb. so a +200ppb inconsistency that was +200% high.

Neither of these spiked element studies lend itself to the Gage R&R analysis because they don't contain any samples expected to be the same for the spiked elements, which is where the R&R analysis is a better fit. So they aren't included in that analysis.

So the analyses were fed a lot of data, but not every single data set.

 

[Randy Holmes-Farley]

Here's some details that adds context for how these results got generated. This was from samples spiked with 4 different levels of trace elements and sent to 4 vendors.

(Red represents Rick's method based on Red Sea chemistry with +-20% error bars for scale )

It's difficult to look at this data and say ICP-A was meaningfully measuring Iodine at all in these 4 (trace-spiked) samples.
So even a reading of 1000ppb Iodine from ICP-A does not assure you that it's "real". It could be - and was - only around a tenth of that value. Hence Red square on CHART 3 for ICP-A even at 1000ppb I.

And this wasn't a one-off. Another data set that did a set of dilutions between I.O. and spiked Brine also contained data where Iodine was ~120ppb in the I.O. and ~60ppb in the Brine, but mixing them together generated a value of ~330 ppb. so a +200ppb inconsistency that was +200% high.

Neither of these spiked element studies lend itself to the Gage R&R analysis because they don't contain any samples expected to be the same for the spiked elements, which is where the R&R analysis is a better fit. So they aren't included in that analysis.

So the analyses were fed a lot of data, but not every single data set.

Thanks, taricha.

Perhaps that points up limitations to the interpretations (presumably because the total number of samples was not really very large), since the ICP-A result from the reproducibility test said it was useless while the result from the precision test indicated it was pretty good. And it's not apparently a difference of reproducible vs precision, but rather a difference that a big outlier fell in one test and not in the other.

 

[Randy Holmes-Farley]

On the other question, how was Chart 4 generated?

 

[Rick Mathew]

OK, thanks again.

A few questions.

How was chart 4 generated? Where did the high and low values come from? The salinity also seems pretty high in it.

I'm not fully sure how the different tests fit together.

Let's look at iodine. ICP-A has a red code at 1 ppm, so presumably it had an iodine reading on at least one sample that was close to 1 ppm? But then in chart 5, ICP-A values for iodine are shown as having a bit better than moderate precision?

Chart # 4 was generated from the Vendors "recommended" highs and lows most of them came from Fauna Marin data.

 

[Randy Holmes-Farley]

Chart # 4 was generated from the Vendors "recommended" highs and lows most of them came from Fauna Marin data.

OK. I was mostly thrown by the magnesium being so high. The low end of the magnesium range is 100 ppm above 35 ppt seawater. Perhaps that's just a Fauna Marin preference, but I'd rather see magnesium below the low end at 1300 ppm than within the range at 1500 ppm.

 

[Dan_P]

Not really that kind of article, heh.
I would characterize the results to say 3 of the 4 vendors were similar enough, that I can't be confident the differences between them in numerical assessments are more than just noise in the data we analyzed.


I'm so glad ICP-MS is outside the scope of this article!
Oceamo is a smaller volume operation that the 4 we tested also, so it's difficult to decide which of our sources of variation I think would and wouldn't apply to them similarly.

A second opinion about ICP-MS and some thoughts.

There are two parts to this method: ICP and MS. Ditto ICP-OES: ICP and OES. I assume the variability we detect in results comes from uncontrolled shipping conditions and the ICP part of the method. MS improves detection limits over OES, but that does not matter if the ICP part of the analysis is a major contributor to noisy data.

With that said, I will recall Randy’s observation which sort of goes like “you cannot expect a first class ICP-X analysis for 50 USD”, hence our desire to make the best of what we get.

 

[Lasse]

I assume the variability we detect in results comes from uncontrolled shipping conditions and the ICP part of the method.

How are you thinking now - did you not send in all samples to the same vendor at the same time?

Sincerely Lasse