disc1
-RT * ln(k)
There's another possibility for the colorimetric kits. Another method of quantitative analysis is called standard addition.
In standard addition, the sample is first measured using some uncalibrated method. Then, a known amount of standard is added to an identical sample and it is measured by the same method. As long as the response is relatively linear or exponential, then a little algebra can tease out the concentration in the first sample.
Basically you have: (where f(x) is the response R for a given analyte concentration x)
f(Sample) = R1
f(Sample + Standard) = R2
or with a little algebra: (where the brackets denote concentrations of analyte)
[Sample] / [Sample + Standard] = R1 / R2
If any of you are math people like me, you see real quick that we have made some big assumptions here. We are assuming that f(x) is linear or that both measurements are within the linear range. Hopefully it will be close. If it is logarithmic or exponential, then we need some slightly different math.
This is not our first choice of methods when we are doing analysis, but sometimes it's all we've got. One example is when the sample matrix significantly affects the test and can't be replicated in a calibration curve. We do this with blood sometimes since we can't count on two different people's blood to be the same. In my line (mass spec) we label the standard with one or more isotope labels that allows us to tell the analyte from the standard and clear up the math. But I digress.
Another thing that can help is to use several standard additions over a range of values to create a sort of calibration curve. That math is a little more involved and there is the need to run several tests on the same sample to get a result. It might be worth looking into doing once or twice, but isn't something to do every time you test.
How can this help with a colorimetric test?
Since the result is so subjective with reading the card and all, then it allows you to run one extra test and reassure yourself a little.
For example, we have seen this thread a thousand times. Kit X reads 5ppm for nitrate and kit Y reads 50ppm. Which is correct?
If we add a known 10ppm from a standard solution, then we expect the result to change. If the real value is 50ppm, then we have only changed the concentration by 20%, so we would expect to see a 20% increase in the reading and get 6ppm from kitX and 60ppm from kitY.
If the real value is 5ppm, then we are changing the concentration by 200%. Since we are tripling the concentration, we expect to see a large change in the reading. KitX should read 15ppm, and kitY should read 150ppm or possibly off the charts.
The same sort of approach can be employed to answer the finding the right lighting conditions or which way to hold the tube questions. If you triple the concentration and the reading doesn't also triple, then you're doing it wrong.
Here's the rub. We are making the assumption that the concentration is in the testable range for both tests. That may not always be the case. So we have to be real careful with this method and think about what we are doing and the results we get.
We are also making some chemical assumptions that I'll not go into here. It has to do with any interference that is present in a small enough amount to throw off one measurement but not the other. I don't think we need to worry about it here.
I don't think we can use this to improve accuracy like we can with the titrations, but it does allow for a quick check when two kits don't read the same. At least for phosphate or nitrate tests. I don't want to talk about iodine or potassium until I have a chance to play with the chemistry. I know some lab methods, but haven't played with those kits.
In standard addition, the sample is first measured using some uncalibrated method. Then, a known amount of standard is added to an identical sample and it is measured by the same method. As long as the response is relatively linear or exponential, then a little algebra can tease out the concentration in the first sample.
Basically you have: (where f(x) is the response R for a given analyte concentration x)
f(Sample) = R1
f(Sample + Standard) = R2
or with a little algebra: (where the brackets denote concentrations of analyte)
[Sample] / [Sample + Standard] = R1 / R2
If any of you are math people like me, you see real quick that we have made some big assumptions here. We are assuming that f(x) is linear or that both measurements are within the linear range. Hopefully it will be close. If it is logarithmic or exponential, then we need some slightly different math.
This is not our first choice of methods when we are doing analysis, but sometimes it's all we've got. One example is when the sample matrix significantly affects the test and can't be replicated in a calibration curve. We do this with blood sometimes since we can't count on two different people's blood to be the same. In my line (mass spec) we label the standard with one or more isotope labels that allows us to tell the analyte from the standard and clear up the math. But I digress.
Another thing that can help is to use several standard additions over a range of values to create a sort of calibration curve. That math is a little more involved and there is the need to run several tests on the same sample to get a result. It might be worth looking into doing once or twice, but isn't something to do every time you test.
How can this help with a colorimetric test?
Since the result is so subjective with reading the card and all, then it allows you to run one extra test and reassure yourself a little.
For example, we have seen this thread a thousand times. Kit X reads 5ppm for nitrate and kit Y reads 50ppm. Which is correct?
If we add a known 10ppm from a standard solution, then we expect the result to change. If the real value is 50ppm, then we have only changed the concentration by 20%, so we would expect to see a 20% increase in the reading and get 6ppm from kitX and 60ppm from kitY.
If the real value is 5ppm, then we are changing the concentration by 200%. Since we are tripling the concentration, we expect to see a large change in the reading. KitX should read 15ppm, and kitY should read 150ppm or possibly off the charts.
The same sort of approach can be employed to answer the finding the right lighting conditions or which way to hold the tube questions. If you triple the concentration and the reading doesn't also triple, then you're doing it wrong.
Here's the rub. We are making the assumption that the concentration is in the testable range for both tests. That may not always be the case. So we have to be real careful with this method and think about what we are doing and the results we get.
We are also making some chemical assumptions that I'll not go into here. It has to do with any interference that is present in a small enough amount to throw off one measurement but not the other. I don't think we need to worry about it here.
I don't think we can use this to improve accuracy like we can with the titrations, but it does allow for a quick check when two kits don't read the same. At least for phosphate or nitrate tests. I don't want to talk about iodine or potassium until I have a chance to play with the chemistry. I know some lab methods, but haven't played with those kits.
] I was going to start a thread to see what others thought about the accuracy of R. Sea as compared to Salifert/Elos. I and another reefer ran both test kits 3 times in a row. We both got the same results every time... Using the high range for both: Salfert tested the nitrate between 10-25ppm; Red Sea was 4ppm. To me, that is a significant difference! The only conclusion we came to was for me to start Vodka dosing:headwalls:
