questions on LDA+U, LSDA, LSDA+U methods

[luke419]

I'd like to perform geometry optimization and single point energy for metallorganic compound in which copper is included.
I think that one of LSDA, LSDA+U, or LDA+U should be appropriate for this material because of copper rather than simple LDA method.

1. Can LDA+U describle polarized spin too?

2. How can we determine U and what is the commonly acceptable value of U for copper in copper oxide materials such as my system (copper is cooridnated by four oxygens and another copper)?
If U is determined somewhat arbitarily at first, how can we confirm the validation of the value of U even though it may be breifly, if ther is no experimental data?
Will it be some kind of iteration process, requiring the check of output results?

3. Copper has unpaired electron, then what method should I use among LSDA, LSDA+U, LDA+U?
Will LSDA be sufficient for the calculation of my system?

4. What should be different in input files in case of LSDA compared to LDA?
May I use the same potential file?
What should I add or change in INCAR file for LSDA compared to LDA?

5. Would you let me know detailed process for LDA+U in VASP?

6. Will the process of LSDA+U be similar to LDA+U generally?


Best regards,


Luke


<span class='smallblacktext'>[ Edited ]</span>

[apple]

hi,
The U values can be treated as tunable parameters chosen to reproduce known properties of the system of interest (empirical approach), or can be extracted from constrained DFT calculations. See the following reference for more details : Phys. Rev. B 75, 035109(2007) (It is for CeO compounds, but the same approach can be taken for other systems.
Also, see www.nsc.liu.se/lcsc2005/vasp/handson_excercises_4.ps (This reference shows how to implement the L(S)DA+U method. Also, please refer to the manual for more info about the method.
I am not familiar with your system, therefore, i cannot answer other questions.