isobestico
commented
1 year ago Hi,
It has been a while since I used SLABCC.
From what I remember, in one of the recent updates, certain security measures were implemented to ensure highly reliable results. When encountering a 'too delocalized charge model', the code will now fail and stop without producing any results.
You may have already seen the warning message.
This indicates that using a Gaussian Charge Fitted Model is not suitable for obtaining a reasonable approximation of your charge model, and consequently, the final energy correction result might be incorrect.
However, in some previous versions of SLBACC, it was possible to obtain results even with these 'bad/untrusted' charge models, and they were not entirely unfavorable despite their delocalization
-
as I remembered when I used/tested.
Nevertheless, if your charge model is too delocalized, you can try adjusting the description of your systems (charged and neutral systems).
Here are a few suggestions (my personal view):
(1) Increase the length of the lattice vectors in all directions for bulk systems (something around 20 Angstroms ); (2) Increase the length of the lateral periodic lattice vectors while keeping the vacuum constant (~ 20 A ngstroms in the periodic direction, for the vacuum 15A - 20A should be enough and should be kept constant) ; (3) For slab systems, it may be necessary to increase the thickness of the material by adding more layers of atoms to your slab model (again around 15A - 20A); (4) Experiment with hybrid functionals such as B3LYP or HSE. Hybrid functionals tend to provide more localized solutions for charged systems; (5) If your system contains 'd-elements,' consider applying some Hubbard correction. This could lead to a more localized solution; (6) You can try fitting more than one Gaussian model to your system. SLABCC allows the use of multiple Gaussian fitting models, which can be helpful for moderately delocalized systems. Please, check the manual carefully for multiple Gaussian fitting;
These are some techniques that might assist you in finding a more localized model to use with energy charge corrections software like SLABCC.
Dear Sir or madam:
I always got the error as below:
' The model extra charge seems to be (at least partially) delocalized. The current charge correction method is not suitable for the delocalized charges. [warning] The optimized position for the extra charge is significantly different from the initial value. Please make sure that the final position of the extra charge have been estimated correctly! Increasing the calculation grid size did not decrease the discretization error. Most probably the model charge is fairly delocalized!'
Could you please let me know how to solve this problem ?