I have used the linear responce method to calculate Hubbard's U in VASP as indicated by VASP tutorals in their website. I give the url:
https://www.vasp.at/wiki/index.php/Calculate_U_for_LSDA%2BU
However my system is a zeolite and the transition metals I work with are not located in equivalent locations. One that has worked with zeolites knows that we identify various sites when we are are talking about metals in zeolites.
What I have not found in the bibliography and seems intimidating is to work with, is implementation of different values of U for the same kind of atoms in the system.
I should also add that sometimes the values of U are similar for different sites (regarding cationic locations). But in some other cases I may get different values of U even for the same kind of atoms in the same category of sites, which really seems not intuitive.
Dear Giannis, sorry to see that your interesting technical question has not yet received any expert answers. As a synthetic inorganic chemist I'm absolutely not qualified to directly answer your question. All I can do is suggest to you the following potentially useful manuscript which mght help you in your analysis:
First-principles Hubbard U approach for small molecule binding in metal-organic frameworks
http://www.cchem.berkeley.edu/co2efrc/downloads/pdfs/Mann16.pdf
(please see attached pdf file)
I would like to mention that three of the authors have RG profiles. Thus it might be worth a try to contact one of them via RG and ask your question directly. They are the experts.
Good luck with your work!
Thanks a lot dear Frank. The article you sent me is very interesting and can really help understanding someone how U is used to optimize the properties of your system. However they don't address the question of different U values for the same kind of atoms in the same system. For example imagine that a zeolite cell has two Co atoms and, implementing the linear responce method, one gets a U of 5 and the other of 3.5 eV. Is it acceptable to seperate the Co atoms in two groups, or I should take the mean value to carry out the calculations?
Dear Giannis, thanks for our kind response and question. As I said, we work in synthetic inorganic and organometallic chemistry. Thus I'm absolutely not a specialist enough to answer your questions in detail. My suggestion would be to contact the authors of this or related papers directly via RG. You can also serarch RG for relevant terms and then click on "Researchers". That way (if you are lucky) you can perhaps identify other specialists in this field.
Also please check if some papers in the following list of search results are relevant for you:
https://www.researchgate.net/search.Search.html?type=publication&query=Performance%20of%20DFT+U%20Approaches%20in%20the%20Study%20of%20Catalytic%20Materials
Good luck and best wishes, Frank Edelmann
Hello,
It's not obviously wrong to use different U values for different sites of the same atom, although I wouldn't bother doing it unless the calculated U's are quite different, such as more than 0.2 eV. I believe it can be done by splitting the single species into two or more consecutive POSCAR entries, with a matching POTCAR just as you did for the linear response calculation. Then just split the LDAU and other INCAR entries in the same way.
Hello Steven,
Thanks a lot for your answer! Yes the difference is far more than 0.2. To be honest I seperate the atoms when the difference is more than 0.5 eV. The problem is that there is not an intuitive way to explain this difference and I have never read an article where they do this kind of seperation. The property that I calculate is adsorption energy so the effect is actually not very significant. But I don't want to publish something that has a methodological error that maybe I am not aware of.
Check https://aip.scitation.org/doi/am-pdf/10.1063/1.4947240
Check https://aip.scitation.org/doi/full/10.1063/1.3660353?mimetype=application%2Fpdf&crawler=true
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms