search

Julian-Hochhaus / LG4X-V2

A graphical user interface of Python lmfit package was developed for standard X-ray photoemission spectroscopy (XPS) curve fitting analysis. The interface streamlines the fitting procedures for validating results and their consistency.
https://julian-hochhaus.github.io/LG4X-V2/
Other
20 stars 9 forks source link

Unsubtracted BG not shown, only subtracted BG shown #77

Closed kevinsmia1939 closed 1 year ago

kevinsmia1939 commented 1 year ago

Hi, I see that you have an update, so I give it a try on dev branch. I noticed that the BG line is flat when in fitting mode, which I assumed was the unsub BG. But the data is plotted against unsub BG, as you can see in the screenshot, which produces an artifact.

Evaluation mode Screenshot_20230627_170240

Fitting mode Screenshot_20230627_170250

kevinsmia1939 commented 1 year ago

So with some more testing with Voigh lineshape, it worked as expected. Fitting mode show unsubtracted BG. So I think there is something going on with the Doniach-Dublett lineshape.

Screenshot_20230627_171430

Julian-Hochhaus commented 1 year ago

Hmm, I am not yet sure if there truly is an error here. First of all, I assume that there might be something odd with the evaluation mode and fitting mode and that there might be a difference in what is plotted. However, as you only use the active Shirley, I do not expect to have a problem with the plotting here

However what I rather see is, that your doniach model actually results in a good agreement with the data but is a bad fit concerning physics. The right side of the spectrum only results in great agreement due to the peak being negative, meanwhile the background is way too high. This is caused by the high asymmetry you used. I would suggest to limit asymmetry to (-0.05, 0.0).

In addition, the coster kronig effect is usually quite small, I would expect less than 10% broadening of the higher bound peak. Judging from the spectrum, I would even suggest less than 3%. Therefore limit factor coster kronig to (0.97,1.00).

Julian-Hochhaus commented 1 year ago

Just as reminder for me:

Using binding energy scale: Checks need to be made that all limit concerning fct_coster_kronig, height_ratio etc. are set accordingly.

Update (30.06.23): All fine

Julian-Hochhaus commented 1 year ago

I am thinking of a second reason which might cause the differences between evaluation and fit mode. The Shirley Background is calculated differently in both modes currently (for the evaluation mode, the static shirley is used). At least, the static shirley should then be fixed to one iteration.

kevinsmia1939 commented 1 year ago

I am thinking of a second reason which might cause the differences between evaluation and fit mode. The Shirley Background is calculated differently in both modes currently (for the evaluation mode, the static shirley is used). At least, the static shirley should then be fixed to one iteration.

I see, why is evaluation mode use static Shirley? And about the physic behind asymmetry and kronig effect, I will need to dig deeper into it.

Julian-Hochhaus commented 1 year ago

I see, why is evaluation mode use static Shirley?

Lazy answer: Historical reasons :D. I never changed yet this part of code compared to the LG4X program. Will work on it today.