[CP2K-user] Is there a need to employ DFT+U for the first-row transition metal ions dissolved in aqueous/nonaqueos solutions?

[Vyacheslav Bryantsev]

I am writing to check if there is a consensus on using GGA DFT+U for the 
first-row transition metal ions in solution.
In some papers researches do include such corrections, but most often for 
simulations of solutions the DFT_U correction is not applied.
The problem of course, is in the description of the d states that are 
overdelocalzed using GGA. 
I have case of a molten salt containing MgCl2-KCl with 5% CrCl3.

If no DFT_U correction is applied, I have the following HOMO - LUMO gap and 
Hirshfeld spin moments
HOMO - LUMO gap [eV] :    0.032428
*HOMO - LUMO gap [eV] :    0.001084*
Hirshfeld spin moments on six Cr3+
2.817
2.774
2.814
-2.759
-2.859
-2.855

With U_MINUS_J  of 3.36 eV, I have
HOMO - LUMO gap [eV] :    0.067988
HOMO - LUMO gap [eV] :    0.049083
Hirshfeld spin moments on six Cr3+
2.965
2.940
3.010
-2.921
-3.008
-3.000

Clearly, the HOMO - LUMO gap is larger with DFT+U and the spin moment is 
closer to +/-3.0.
Now, is based on this, DFT+U is preferred or there are arguments against it?
Could also one elaborate on the choice of U_MINUS_J for species in solution?
I have taken U_MINUS_J based on the solid-state calculations for Cr2O3 used 
with the projector augmented wave method. 
I saw somewhere that U_MINUS_J is not transferable from VASP to CP2K. If 
this is the case, what would be the best choice of U_MINUS_J.

Thank you,
Slava



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190425/40a0d28b/attachment.htm>