By reading the book " Practical HPLC Method Development " (2nd Edition, page 654, Lioyd R. Snyder etc.), I found the normalized peak area method can be used in calculation of impurity or degradation products in a purified compound even when the main peak is a flat-top peak (see Figure 14.6 on page 654). I was told that whenever the peak is flat-top, the peak area integration is not correct. If that is true, the normalized peak area method should not be used when main peak is flot-top. I am confused by this. Can anyone help to explain it?

Is the peak actually out of the detector range or is it just off the displayed scale but still within the detector range?

It does not give detail information in the book. If the peak is out of detector range but not overload column, can the normalized peak area still be used ? Also, I would like to know the general method to qualitatively test the trace level impurity (e.g. 0.2%): Normalized peak area, LOD or other method?? and which method for limit test of trace impurity is concidered acceptable by FDA ?

Thanks in advance !!!

If the peak is out of the detector range (reached saturation), I am not aware of any way to use peak areas or heights for normalized quantitations as the value you would have for the out of range peak is meaningless. My guess is the display in the book was blown up to better show the smaller peak and the larger peak had not reached saturation.
If you do not know the identity of the impurity, you would have to make assumptions to get a "qualitative" result. One method is the normalized peak area where you assume all peaks have the same detector response and you have a wide linear range for calibration.

Another method would be to make a standard or standard curve of your active material until the peak area for it is approximately the same as or brackets the impurity area in your tested sample. You would then use single point calibration or a calibration curve to calculate a concentration for the impurity in your sample. In this case you are assuming identical detector response for the two compounds but are not assuming a wide linear range for calibration.

The problem is that identical response for two compounds is not always true. For UV detectors, this response can be significantly different. Don't know if FDA accepts either of these "methods" or if they have another they prefer. They may even expect the impurity to be identified. Maybe someone with experience dealing with the FDA can help you.

You have to inject a large amount of sample, in order to be able to integrate small peaks of impurities, and the main peak is out of detector's linear range. Inject also a diluted sample (e.g. 100 times). From the area of the main peak of diluted sample, you can calculate the correct area for the main peak of undiluted sample. From here you can calculate the area percent of each impurity. Don't know about FDA, but this method is used by USP in some cases.