Sciencemadness Discussion Board
Not logged in [Login ]
Go To Bottom

Printable Version  
Author: Subject: ICP-OES calibration curves
aeacfm
Hazard to Others
***




Posts: 129
Registered: 24-7-2010
Member Is Offline

Mood: chem

[*] posted on 21-7-2011 at 02:26
ICP-OES calibration curves


i am a bit confused about calibration curves using ICP-OES .....
in some samples (mainly low salt samples) we use linear with or without intercept , other samples use non-linear calibration curve !!!!!!!!!
when to linear and when to use nonlinear






View user's profile View All Posts By User
DDTea
National Hazard
****




Posts: 940
Registered: 25-2-2003
Location: Freedomland
Member Is Offline

Mood: Degenerate

[*] posted on 22-7-2011 at 20:17


As a rule of thumb, use a linear calibration curve. It is the easiest to setup because it follows directly from Beer's Law. Only move on to non-linear calibrations when the linear curve is clearly not giving accurate results (which you test by reference materials).

A lot of things can cause this to happen, but you'll mostly run into problems regarding non-linear responses when you are analyzing complex matrices and measuring a few dozen wavelengths. It's simply an issue of probability: the more stuff in your sample, the more complex its emission spectrum will be. At some point, you will encounter a spectral interference, which can dramatically affect the response of one of the wavelengths you're analyzing.

Spectral interferences and ways to handle them are a topic that's bigger than I'm willing to write about in a single post, but you should definitely read about them if you're working with ICP-OES. A good example that I can think of is the sodium D lines (~589 nm). They emit VERY strongly and can easily saturate your detector. It will also bury any peaks in its tail, and those smaller peaks will not be properly measured.

That's where inter-element correction factors and chemometrics come in. Consider two wavelengths of iron: 495.761 nm and 466.814 nm, and suppose that you have another peak interfering with the line at 466.814 nm. You can measure the ratios of Fe(495.761) to Fe(466.814) and thus indirectly measure the peak at 466.814.

In reality, it's more complicated than that, but I hope that helps you understand why non-linear calibration curves come into play!




"In the end the proud scientist or philosopher who cannot be bothered to make his thought accessible has no choice but to retire to the heights in which dwell the Great Misunderstood and the Great Ignored, there to rail in Olympic superiority at the folly of mankind." - Reginald Kapp.
View user's profile View All Posts By User
smuv
National Hazard
****




Posts: 842
Registered: 2-5-2007
Member Is Offline

Mood: Jingoistic

[*] posted on 22-7-2011 at 22:10


I have some experience developing ICP-AES methods. The norm of the company I was working for, was to use the cal curve that works best with the matrix. It is very important with ICP-AES not only to test the method with the calibration standards but also make sure that it works with a process sample and also maintains good readings after 1:2, 1:4 etc dilutions of the process sample.

That said, I got pretty frustrated with ICP-AES, the whole process felt like a random guessing game. I really wonder how robust the methods which passed the companies internal standards actually were. Good thing is ICP-AES is kind of a niche instrument, if you have experience with it you can be pretty desirable to some companies.

[Edited on 7-23-2011 by smuv]




"Titanium tetrachloride…You sly temptress." --Walter Bishop
View user's profile View All Posts By User
aeacfm
Hazard to Others
***




Posts: 129
Registered: 24-7-2010
Member Is Offline

Mood: chem

[*] posted on 24-7-2011 at 08:10


i want to thank you very much for your very helpfull post



Quote: Originally posted by DDTea  
As a rule of thumb, use a linear calibration curve. It is the easiest to setup because it follows directly from Beer's Law. Only move on to non-linear calibrations when the linear curve is clearly not giving accurate results (which you test by reference materials).

!

thats what i thought 1st




Quote:

A lot of things can cause this to happen, but you'll mostly run into problems regarding non-linear responses when you are analyzing complex matrices and measuring a few dozen wavelengths. It's simply an issue of probability: the more stuff in your sample, the more complex its emission spectrum will be. At some point, you will encounter a spectral interference, which can dramatically affect the response of one of the wavelengths you're analyzing

thats what i am dealing with




Quote:

Spectral interferences and ways to handle them are a topic that's bigger than I'm willing to write about in a single post, but you should definitely read about them if you're working with ICP-OES. A good example that I can think of is the sodium D lines (~589 nm). They emit VERY strongly and can easily saturate your detector. It will also bury any peaks in its tail, and those smaller peaks will not be properly measured

does this mean the sensitivity , or its not ? could you please explain a little bit




Quote:

Consider two wavelengths of iron: 495.761 nm and 466.814 nm, and suppose that you have another peak interfering with the line at 466.814 nm. You can measure the ratios of Fe(495.761) to Fe(466.814) and thus indirectly measure the peak at 466.814

cant get it?!!!


[Edited on 24-7-2011 by aeacfm]






View user's profile View All Posts By User
DDTea
National Hazard
****




Posts: 940
Registered: 25-2-2003
Location: Freedomland
Member Is Offline

Mood: Degenerate

[*] posted on 25-7-2011 at 16:34


The way you handle the specific problem is going to be highly customized to the particular sample you're working on, so there isn't a lot of advice that I can give you there. If one line is causing you a lot of problems, it's better to simply choose a different line: it's usually easier to find a different wavelength than to make a troublesome wavelength work. For each element of interest, start by "screening" at least 5 different wavelengths. Chances are, at least one of them will be usable.

There are other steps that you can take. Careful selection of background points is essential. If you're performing measurements by peak areas, try performing them by peak heights instead.

Regarding the interelement correction factors: if at a certain temperature, an element is emitting several wavelengths of light, you can find the ratios of the intensities between two lines of a particular element. That ratio can then be used in your calculation. This is one way to handle an interference, but it may not be the best.

If you want to get really tricky, you may have to adjust the plasma itself. Changing the power, nebulizer flow rate, and auxiliary gas flow can alter the plasma temperature which, in turn, changes the intensities of certain wavelengths. So in principle, you can optimize the plasma for a measuring a certain set of wavelengths. However, I wouldn't recommend this approach unless you have a lot of time to fine-tune your method, are comfortable with trial and error, and don't mind working with Lagrange Multipliers in at least 3 dimensions :)

Sometimes, though, ICP-OES just isn't the best technique to use. For example, the emission spectrum of copper can be absolutely horrible. We use a DCP (5000 K plasma vs. 10 000 K for an ICP) for a few elements; unfortunately, DCP's are becoming harder to find these days. You may have to use a technique like atomic absorption (especially for alkali metals) or even classical wet chemical tests.

Check out the book "ICP Emission Spectrometry: A Practical Guide " by Joachim Nölte--it's short and user-friendly. Keep it by your instrument!




"In the end the proud scientist or philosopher who cannot be bothered to make his thought accessible has no choice but to retire to the heights in which dwell the Great Misunderstood and the Great Ignored, there to rail in Olympic superiority at the folly of mankind." - Reginald Kapp.
View user's profile View All Posts By User

  Go To Top