How is int. width of voigt profile calculated in Fityk ?

[Mouritz Svenson]

Hello 

I have been doing deconvolution of Raman spectra using voigt profiles in Fityk.

I have had some weird FWHM of the voigt profiles. As I can see from this forum, this has also hapened to other users, when voigt profiles were modeled with negative gwidth and shape parameters. However, these paramester can be set as postive, they can be locked in the deconvolution, and the same fit will result, however with a higher FWHM.

Before I realised this, I tried to compare the trend of int.width (also provided in the results of the voigt function) with the FWHM, to see if they basically described the same features. I found that for some voigt peaks, the int. width and FWHM followed the same trends over different samples, whereas for some voigt peaks they did not.

This lead me to wonder how the int. width of voigt peaks are calculated in Fityk. I looked through the manual but couldn´t find an answer there. 

So my question is simply: How is the int. width of the voigt profile calculated in Fityk ?

[Marcin Wojdyr]

Hi Mouritz,
as the manual says:
"FWHM is estimated using an approximation called modified Whiting
(Olivero and Longbothum, 1977, JQSRT 17, 233)",
the formula is also there, see http://fityk.nieto.pl/model.html

but if it gives weird results let me know, I'll look into it

Marcin

> --
> http://groups.google.com/group/fityk-users
> ---
> You received this message because you are subscribed to the Google Groups
> "fityk-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to fityk-users...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Mouritz Svenson]

Hello Marcyn

Thanks for the fast reply ! And thanks for making Fityk available and answering questions!

I am a afraid that my question might have been misunderstood though. When using the voigt profile for deconvolution, various parameters are provided in the results of the voigt line. Amongst them are FWHM of the laurentzian and Gaussian components. However besides those, there is another separate parameter called "int. width", which does not have the same value as neither the FWHM of the Laurenzian, Gaussian or voigt profile.

This is what makes me curious. For most of the peaks I deconvolute, int. width and FWHM of the voigt peaks follow the same trend, but for a few peaks they don´t. Therefore I was curious about how this other description of peak width (int. width) was calculated, and how/whether it was a better/worse description of peak width.

[Marcin Wojdyr]

The integral width is the width of a rectangle with the same height
and the same area as the peak.
I.e. it's area divided by height.
Hmm, I've just googled "integral width" and the results shown seem to
be all from the x-ray diffraction literature, so maybe this term is
used only in the x-ray powder diffraction community.

Marcin

[Mouritz Svenson]

Ahhhh "integral width". I thought that "int. width" stood for "internal width", but I see. 

It seems like a peculiar way to describe the width of a peak though. But okay, thanks a lot for the answer.

I will stick to describing the peaks by FWHM, but it is nice to know why int.width and FWHM gave so different results.

By the way, is their any special way that Fityk should be cited if referenced to in a scientific paper ? (I have never referred to software before).

[Marcin Wojdyr]

thanks for asking, you can cite:
J. Appl. Cryst. 43, 1126-1128 (2010)
http://scripts.iucr.org/cgi-bin/paper?S0021889810030499

Marcin