Peak purity

[mk12]

I did my forced degradation study and I want to know how to calculate peak purity using chem station software. To pass purity angle to should lower than purity threshold am I right?

I am repeating the study which was done previously somewhere else there peak purity passed but mine is not passing? we used the same API.
can I enter the threshold value manully or it should be autocalculated. I did try to enter manully to make it pass but its not helping..

Please guide is i am doing something wrong..really appreciate your help

[mbicking]

What version of ChemStation are you using? The Peak Purity algorithm has changed from the earliest versions.

I can only discuss the "newer" version.
1. The system obtains spectra across the entire peak. Then it calculates an "average" spectrum, and compares each individual spectrum with the average.
2. The differences between spectra and the average are compared to statistical calculations. If the differences are larger than expected for noise only, and more than three adjacent spectra show these differences, the peak is considered impure.
3. The "threshold curve" shows the statistical limit for differences. The "similarity curve" shows the actual values. When the differences are too great, the simlarity curve rises above the threshold curve.

The problem with Peak Purity is that it is very sensitive to the settings. I can change a pure peak to impure, or an impure peak to pure, if I manipulate the settings. And the system will label a peak as impure when the noise levels are very low, even if it is a good peak.

The best method is to use a fixed threshold of 990. Any peak with a value larger than this limit should be considered pure.

But remember, this is not a perfect test. Two different compounds with similar spectra may not be detected.

And since this method is based on statistics, the rules of hypothesis testing apply: "You can never prove that a peak is pure. You can only say that it has not been shown to be impure."

[mk12]

I am using chemistation A09:01(1206) vesion. If I use "include thershold curve " it does not pass but I i exclude it its passes? I know is one of the most important option but if i still exclude it than my results are invalid? or I can still use peakpurity without threshold curve

[mbicking]

Where do you see the "Include threshold curve" option? In which settings window?

[mk12]

I saw that in Purity option window. Its under Draw &use simliarity curve option?

[mbicking]

I do not see that purity option in my version. You may be using an older version of peak purity. In my version, the options are: Calculate Threshold and Fixed Threshold. The option you are asking about is probably the Calculate Threshold in my version.

This is the choice that uses the noise in the chromatogram to set the threshold, rather than using a fixed value. If this is the setting that you used before, then you should continue to use it. I do not know why the peaks failed in the last experiment. As I mentioned in an earlier reply, sometimes the ChemStation will label a peak as impure when we actually believe it is pure. You would have to look more carefully at the information.

For example, do you see values for calculated threshold and actual peak purity? If so, please tell me what those values are.

[mk12]

Yeah I think we are using old version of software Rev A:0901(1206). Threshold value there is autocalculated. Fixed value is 990. even if i use fixed value and play with it it still does not pass. I read manual it says its imp to use "include threshold curve" option. I will check my value and will let you know ASAP.

Thanks a lot for all your help

[mbicking]

In your software version, there are two different Peak Purity programs available. The first time you use Peak Purity, you get a question box, asking if you want to upgrade to the newer version. If you select No, then you will be using the old version, and there is no way to upgrade this method.

Try creating a new method (Def_LC.M) and see if this message appears when you try to access Peak Purity. If it does, upgrade and repeat the tests on your "good" and "bad" peaks, then report the results here.

[nimbi63]

Also look at the comparision spectra (overlayed). Many times you will see the spectra look perfectly overlayed except for the very low UV region where you are approaching UV cutoff of the solvent. Also look at the integration results, if there is a lot of baseline before or after the peak, you may be using the spectra of the solvent to compare to the spectra of the sample compound. I've had some where I had to change the range from 190-600nm to 220-600 because my solvent UV cutoff was 212.