[aeacfm]

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

[JGK]

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

For any quantitative analytical method, you should test for the most appropriate calibration model at the development (pre-validation) stage.

The calibration model selected will be the one which providedes the most accurate quantitiation of known sollutions over the desired concenrtration range. This is the tested rigorously during the validation procedure.

[aeacfm]

For any quantitative analytical method, you should test for the most appropriate calibration model at the development (pre-validation) stage.

i know will be the first one to answer ..
could you please explain a little bit

[JGK]

As part of any development work you should perform measurements on standard solutions of known concentration. From this data you can perform a calibration. there are many different ways (models) to perform the calibration calculations examples may include (but are not limited to) linear unweghted or weighted and quadratic.

Using a selection of calibrations, the accuracy of measuement of further standard solutions can be tested. The choice of calibration should be the model which provides the best accuracy. You would then fully validate the method using this calibration model

[aeacfm]

what i understood is that:
- prepare stds
- make different calibration curves models  using these stds
- prepare validation or verification stds
- measure the verification stds with all the calibration models
- the calibration model which gives the accurate value of the verification stds is that i should work with
Am i right ?

[JGK]

what i understood is that:
- prepare stds
- make different calibration curves models  using these stds
- prepare validation or verification stds
- measure the verification stds with all the calibration models
- the calibration model which gives the accurate value of the verification stds is that i should work with
Am i right ?

Basically yes. However, after deciding on the calibration model you should formally validate the method by repeating the same exercise multiple times (at least 3) using fresh standards each time to ensure the method can be reproduced to an acceptable standard.

[aeacfm]

Basically yes. However, after deciding on the calibration model you should formally validate the method by repeating the same exercise multiple times (at least 3) using fresh standards each time to ensure the method can be reproduced to an acceptable standard.

yes i got it , but no other way can effect this with the trial steps , i mean that good selection of wave lengths that of moderate intenisity or of high intensity , of spectral interfernce . others which can shorten the trials of calibration models

[JGK]

Basically yes. However, after deciding on the calibration model you should formally validate the method by repeating the same exercise multiple times (at least 3) using fresh standards each time to ensure the method can be reproduced to an acceptable standard.

yes i got it , but no other way can effect this with the trial steps , i mean that good selection of wave lengths that of moderate intenisity or of high intensity , of spectral interfernce . others which can shorten the trials of calibration models

I would expect that selection of wavelengths assessment of spectral interference etc would be determined before you get to the calibration selection stage.

You should always proceed stepwise rather than using a "shotgun" approach. If you experience problems having adjusted multiple variables, isolating the cause becomes very hard