Question: Where can I find "Gaussian/Lorentzian mix"

[kevinsmia1939]

Hi,
I have been reading about XPS fitting. XPS fitting paper suggest to keep glmix in certain range for different material.

glmix = lhm/lhm+ghm

Compounds: glmix below 10% for Voigt functions and M = 0.5 for pseudo-Voigt product functions and M = 0.2 for fpseudo-Voigt sum functions.
Metals and elements: glmix above 60% for Voigt functions and M above 0.8 for pseudo-Voigt functions and M above 0.6 for pseudo-Voigt sum functions.

Ref
https://pubs.aip.org/avs/jva/article/38/6/061203/1023652/Practical-guide-for-curve-fitting-in-x-ray

Where can I find this value in LG4X? Thanks.

[Julian-Hochhaus]

I haven't yet incorporated a glmix parameter into my implementation, but I'm open to being convinced of its usefulness. However, I am aware of the G. Major guide, but I don't agree that working with a strict glmix value provides significant value.

Typically, I have followed this rule of thumb in my work:
The Gaussian distribution serves as an approximation of the convolution of three distinct Gaussian broadening functions that play a role in the photoemission process:

1. Broadening caused by the excitation source.
2. Broadening resulting from thermal broadening and vibration modes (phonon broadening, depending on the material).
3. Broadening introduced by the analyzer/spectrometer.

These conditions tend to remain relatively constant during measurement and are influenced by factors such as the experimental setup and temperature. On the other hand, Lorentzian broadening carries physical significance as it represents the intrinsic lifetime broadening of the core-level hole state. Only Gaussian contribution due to sample is the phonon broadening.

Consequently, when constructing a fitting model, it is essential to consider the convolution of a model that captures the physics of the sample and a Gaussian function that accounts for spectrometer, light source, and thermal broadening. Therefore, it is advisable to avoid using simple Lorentzian or Gaussian models and instead opt for convoluted models such as the Convoluted Doniach-Sunjic with Gaussian or VoigtModel.

Given that Gaussian broadening is predominantly influenced by the experimental setup, it is possible to fit data recorded from different samples under similar experimental conditions using the same Gaussian broadening function. Furthermore, if you have information about the broadening caused by your light source and details about your spectrometer's characteristics, such as resolution for different pass energies and transmission function, you can calculate the contributions of these factors to the overall Gaussian broadening. Additionally, if you have knowledge of the sample's temperature during the measurement, you can calculate nearly the complete Gaussian broadening. One contribution remains Gaussian, which is the phonon broadening depending on the sample's material.

In cases where you don't have access to all the necessary information, you can still acquire many of these details. Measuring the Fermi edge provides you with the Gaussian broadening associated with the specific kinetic energy of the Fermi electrons. If you have access to a dual-anode X-ray tube or even a synchrotron, you should be able to identify the Gaussian broadening of the exciting light source and spectrometer.

The glmix parameter offers 'rule-of-thumb' values for different materials. However, its usefulness as a guide is limited to cases where phonon broadening dominates over other Gaussian broadenings. It's important to note that the glmix parameter does not provide significant reasoning or benefits beyond serving as a relatively good starting parameter. (At least that's my experience and opinion :D)

In summary, rather than attempting to make sense of the glmix parameter, I would recommend focusing on identifying the contributions of spectrometer broadening and excitation source broadening through the Gaussian broadening.

Feel free to convince me or make a contribution, it should be relative easy to introduce an additional parameter, which is used for all models used in a fit and limits gaussian to lorentzian fwhm or vice versa using the expr parameter of the lmfit Parameter class.

[kevinsmia1939]

Thank you for very in-depth answers. This goes pretty much over my head, XPS is very new to me. Thank you for your time.