I completed my master's thesis with XPS analysis from LG4X

[kevinsmia1939]

Yesterday I defended my master's thesis, and my supervisor has approved my thesis as complete. The XPS spectrum was analyzed with LG4X and plotted with Matplotlib after I exported the data.

The sad thing is that I will most likely butcher the Ti spectrum. And that we will order a license for CasaXPS to work with other researchers.

[Julian-Hochhaus]

Congratulations on successfully defending your master's thesis! It's a significant accomplishment, and your supervisor approving your work is a testament to your hard work and dedication!

Regarding your concerns with the Ti: The fitting of the p3/2 looks more or less fine, however, due to the dublett characteristic of the peak, you need to take care that the second peak, the p1/2 is correlated with the p3/2 peak. In each peak, the area ratios between the components should stay the same (i.e. ratio between the Ti2+ and TiC areas should be the same in both, the p1/2 and p3/2 peak). In addition, you need to identify the same components in both orbitals and the ratio between the FWHMs of the components identified should be roughly 1 between both orbitals.
I would suggest using the Doniach-Dublett Model, maybe with asymmetry fixed to zero.

Especially in the MX-0.5 spectra there is one really broad component in the p1/2 peak. Meanwhile, there are three relative small components in the p3/2 peak and only two small components in the p1/2 peak plus the broad one.

Concerning the MX-0.1: The TiC is rather small and the rightmost component. In the p1/2 the rightmost component then should be TiC with small FWHM too.

Looking at the spectrum, I am in addition not sure if the SOC-splitting is kept the same for all components identified.

Always feel free to ask, I am always happy to share knowledge and help!

Best of luck with your future endeavors in the field!

[kevinsmia1939]

Thanks for the input on the Ti spectrum. I am very thankful for your help.

The material in question is MXene, which contains TiC and some -F and -Cl on the surface.
From my understanding, if the material is conductive, I have to give it a bit of asymmetry. Is that right?
I use this as my guide to fit MXene: https://doi.org/10.1016/j.apsusc.2022.153442.
Which this paper mention that I should use Voigh line shape with tail, so I assume that skew Voigh would do the same.

I did not encounter the Doniach-Dublett model when I was searching for it when I tried to fit these Ti 2p peaks. Did you mean that I use the dublett to fit both 1/2 and 3/2 peaks?

[Julian-Hochhaus]

It is totally fine to use skewed voigt (behaves similar to a doniach line shape) the main point here is that you need to fit the two peaks of the dublett together and set relations between them.
For each component in the 3/2 peak you need to find one in the 1/2 peak. FWHM and area in relation to other components should stay the same. In addition, the SOC should be the same for all pairs of components.

To reach that, I usually use the Doniach-Sunjic Dublett in LG4X-V2. Setting asymmetry to 0 gives you a pair/dublett of voigt peaks, if you need asymmetry, simply allow it to vary free. Guessing from your data, it's roughly between 0.05 and 0.1.

Using some kind of dublett model for fitting does all the work for you with keeping peak heights relative and FWHMs the same as well as giving you access to a SOC Parameter which you then can use to set it to the same value for all different dublett components found.

[kevinsmia1939]

For Doniach-Sunjic Dublett, what does SOC stand for? And just to clarify, the height_ratio is the amplitude ratio; is that right, not the area ratio?

Also, it seems like I could not set asymmetry(gamma) to negative value which would create a tail toward the high binding energy value, I could only use positive value which create tail for low binding energy value. Do you think this should be change?

I also wonder if Voigh dublett exist?

[Julian-Hochhaus]

For Doniach-Sunjic Dublett, what does SOC stand for?

SOC is the abbreviation for spin-orbit coupling and is often used as a synonym for spin-orbit splitting which describes the difference in energy between the two orbitals of a dublett. In your case, for each component you identified in your Ti 2p 3/2 peak, you should find one with similar FWHM and relative area to the other components in the 1/2 peak with a difference in the binding energy of SOC~(5.8-6.0)eV (taking some average value from NIST, I am not an expert of the Ti 2p signal).

And just to clarify, the height_ratio is the amplitude ratio; is that right, not the area ratio?

Yep, the height ratio is the amplitude ratio NOT the area ratio. That's due to consistency. If I would only use models without asymmetry, I would rather define them in a way that the model area itself is normed and scaled by a factor, which would then enable me to set something like area ratios. But, as soon as we consider asymmetric peaks, the area becomes ill-defined and is therefore only approximated afterward limited to the measured data range (for peaks with high asymmetry, the asymmetric tail could become pretty long without converging close to zero and therefore significantly contributes to the peak area even far away from the actual peak.).

[Julian-Hochhaus]

Also, it seems like I could not set asymmetry(gamma) to negative value which would create a tail toward the high binding energy value, I could only use positive value which create tail for low binding energy value. Do you think this should be change?

I am pretty sure that due to the physics behind the asymmetric tail, an asymmetric tail could only occur on the left of the peak, at lower kinetic energies (btw. did you mix up low and high binding energies in your comment? positive asymmetry creates a tail on the left side of the peak, meanwhile, kinetic energy increases going from left to right in the usual representation of data and binding energy decreases going from left to right, therefore positive asymmetry values create tail on the low-kinetic energy side/high-binding energy side of a peak.)
As a short summary of the topic (and physics behind), I recommend reading the following two papers: XPS insights: Asymmetric peak shapes in XPS and A discussion of approaches for fitting asymmetric signals in X-ray photoelectron spectroscopy (XPS), noting the importance of Voigt-like peak shapes.

I also wonder if Voigh dublett exist?

It exist in general, yes. In LG4X-V2, using the Doniach-Sunjic Dublett with asymmetry fixed to 0 you get a Voigt-Dublett.
Please refer to Eq. (5) and (6) and the following remarks after Eq. (6) in Moeini et al. and if you want to deep dive into theory, read the Doniach and Sunjic paper

[kevinsmia1939]

Hi,
Thank for the papers, that is a great help.

But I am a bit confuse, the tail can only be at the left (high BE) but the asymmetry(gamma) parameter only give the tail to the right, toward a low BE.

[Julian-Hochhaus]

well, I assume there is some kind of bug in evaluation mode with binding energies as x-data, judging from that some other values (area calculation, fwhm etc.) are quite off. Does it always happen using binding energy scale and evalutation mode? Probably need to check it.

[kevinsmia1939]

Hi, I think it is because binding energy is list from high to low which cause a negative value, I simply propose to use absolute value for FWMH and area in #72

[Julian-Hochhaus]

Yep I saw, I think the problem is caused by binding energies not being treated correctly in evaluation mode and I therefore would like to try to fix all the issues at once by rather fixing how binding energies are treated in evaluation mode, so catching the issue by its roots.

[kevinsmia1939]

Hi,
So me and my supervisor are trying to publish my master's thesis because we got an interesting result. I talk with another co-author who did the XPS for me, and they say that I just continue to use the program that I use to fit the XPS spectrum. So I will try to get LG4X into publication soon.
I only found one paper that uses LG4X: https://doi.org/10.1016/j.jallcom.2022.165208.
Maybe I will put you in acknowledgement for helping me or you want to be a co-author lmao.
I hope to add this to the list.

[Julian-Hochhaus]

If I got you right, you would like to reference the software LG4X-V2 in your publication?

Feel free to do so. There is actually a zenodo DOI which you can cite and which gives credit to the author of LG4X, @hidecode221b too, whose software gave me the idea to build upon to.

You can also use the CITATION.cuff included in this repo, whatever is more handy for you.

If I got you wrong, and you meant something else, please let me know.

[kevinsmia1939]

If I got you right, you would like to reference the software LG4X-V2 in your publication?

Feel free to do so. There is actually a zenodo DOI which you can cite and which gives credit to the author of LG4X, @hidecode221b too, whose software gave me the idea to build upon to.

You can also use the CITATION.cuff included in this repo, whatever is more handy for you.

If I got you wrong, and you meant something else, please let me know.

You got it right, but I use 10.5281/zenodo.7777422 as a reference to all the version, because I use multiple version as we update it. Anyway, thank you for the help.

[Julian-Hochhaus]

If I got you right, you would like to reference the software LG4X-V2 in your publication?
Feel free to do so. There is actually a zenodo DOI which you can cite and which gives credit to the author of LG4X, @hidecode221b too, whose software gave me the idea to build upon to.

You can also use the CITATION.cuff included in this repo, whatever is more handy for you.
If I got you wrong, and you meant something else, please let me know.

You got it right, but I use 10.5281/zenodo.7777422 as a reference to all the version, because I use multiple version as we update it. Anyway, thank you for the help.

Indeed, that's better, I am currently attending a conference and therefore only online via mobile and copied the wrong link.

For publication, using citation would I consider fair and would be totally "enough" for me, if you feel like you would like to acknowledge my contribution the scientific results, I would feel honored.

Honestly, without having contributed to the publication directly and without contributing to the actual scientific interpretation of the measured data directly, it would feel wrong to me to be named as a co-author.

But nevertheless, let me know if you need a second opinion especially regarding the XPS data, if my contribution and help is appreciated, feel free to ask! Looking forward to your results and publication! Good luck 🤞

[kevinsmia1939]

Don't worry, I was just joking about putting you as a co-author, but I think I can put you on acknowledgement. Enjoy the conference!

Also, I have some questions about XPS.
I have carbon paper, which is made of graphite. I tried to calculate the change in ratio of C=O to C-O in untreated and heat-treated carbon paper. However, if I use the area of the peak in C1s and O1s, they are different. C=O/C-O from O1s is 3% and 30%, while C=O/C-O from C1s is 10% and 40% for untreated and heat-treated carbon paper, respectively. Is this normal? I would expect them to have the same ratio. Could I fit it wrong?

[Julian-Hochhaus]

Does it work for the MX-0.1 and MX-0.5?

Are you always using same center and same FWHM?

Honestly, without any assumptions and limitations for the fit, I would say you can only confidently fit the C=O component in the O1s heat treated carbon paper, because all other C=O components are very small. From there, you can calculate the expected area for C=O in the C1s spectrum after heat treatment, using the atomic subshell cross sections of C1s and O1s.

As the C=O is really prominent in the O1s compared with C1s, I would then try if it is necessary (I don’t know if it is expected before heat treatment), to include C=O component with same FWHM and center as in the O1s after heat treatment into the fit for the spectrum recorded for the untreated CP O1s. And from there, again try the same in C1s. Once with letting the amplitude/area vary freely, and once check out how good it works with calculations you got from the cross section.

[Julian-Hochhaus]

Don't worry, I was just joking about putting you as a co-author, but I think I can put you on acknowledgement. Enjoy the conference!

Also, I have some questions about XPS. I have carbon paper, which is made of graphite. I tried to calculate the change in ratio of C=O to C-O in untreated and heat-treated carbon paper. However, if I use the area of the peak in C1s and O1s, they are different. C=O/C-O from O1s is 3% and 30%, while C=O/C-O from C1s is 10% and 40% for untreated and heat-treated carbon paper, respectively. Is this normal? I would expect them to have the same ratio. Could I fit it wrong?

Does it work for the MX-0.1 and MX-0.5?

Are you always using same center and same FWHM?

Honestly, without any assumptions and limitations for the fit, I would say you can only confidently fit the C=O component in the O1s heat treated carbon paper, because all other C=O components are very small. From there, you can calculate the expected area for C=O in the C1s spectrum after heat treatment, using the atomic subshell cross sections of C1s and O1s.

As the C=O is really prominent in the O1s compared with C1s, I would then try if it is necessary (I don’t know if it is expected before heat treatment), to include C=O component with same FWHM and center as in the O1s after heat treatment into the fit for the spectrum recorded for the untreated CP O1s. And from there, again try the same in C1s. Once with letting the amplitude/area vary freely, and once check out how good it works with calculations you got from the cross section.

@kevinsmia1939 how is it going, were you able to overcome your problems with the fitting of the data?

[kevinsmia1939]

Hi @Julian-Hochhaus
I send it to another co-author who is expert in XPS, they are looking at it. Hopefully, they will let me know this week.

[kevinsmia1939]

Hi @Julian-Hochhaus
Great news, I have published the research using LG4X-V2.
Sadly, it look like my supervisor remove my acknowledgement with you in it.

https://pubs.rsc.org/en/content/articlelanding/2024/ra/d4ra01380h

[kevinsmia1939]

Maybe we can make a list of publications that use LG4X-V2 on the github page to gain some more traction in the scientific community as well.

Here are example with Tomviz.
https://tomviz.org/publications/

[Julian-Hochhaus]

Hi @Julian-Hochhaus Great news, I have published the research using LG4X-V2. Sadly, it look like my supervisor remove my acknowledgement with you in it.

https://pubs.rsc.org/en/content/articlelanding/2024/ra/d4ra01380h

At least, the citation is there :) Great work!

[Julian-Hochhaus]

Maybe we can make a list of publications that use LG4X-V2 on the github page to gain some more traction in the scientific community as well.

Here are example with Tomviz. https://tomviz.org/publications/

If I find time, I will, great idea!

Edit: Already included some publications in readme, will surely extend the list as well as include it on the web page.
And one day I will have to work on the docs when I have time...