Bader Charge analysis

[Chaure]

Dear Users,
I have been trying to calculate charge transfer using bader program so many times but could not get success. I keep on increasing FFT grid NG(X,Y,Z)F values and run SCF calculation each time in order to achieve fine grid convergence. In this way i have wasted my time doing so many calculations and looking for the results. When i sum AECCAR0 and AECCAr2 to find total CHGCAR_sum for badar analaysis, i never get total number of electrons in ACF.dat file equal to that of total pseudopotential valence electrons. I really, want to do it correctly. If i calculate charge transfer using valence charge density only, how accurate it will be. Please help me to calculate charge transfer correctly using bader program.
Thanks

[graeme]

It is very important to do the Bader partitioning based upon the total charge and the integration based upon the valence charge.

From the Bader analysis webpage:

By adding the LAECHG=.TRUE. to the INCAR file, the core charge is written to AECCAR0 and the valance charge to AECCAR2. These two charge density files can be summed using the chgsum.pl script:

chgsum.pl AECCAR0 AECCAR2

The total charge will be written to CHGCAR_sum.
The bader analysis can then be done on this total charge density file:

bader CHGCAR -ref CHGCAR_sum

One finally note is that you need a fine fft grid to accurately reproduce the correct total core charge. It is essential to do a few calculations, increasing NG(X,Y,Z)F until the total charge is correct.

When you analyze the CHGCAR, you will always get the total number of valence electrons. If for some reason you are worried about getting charge transfer from core electrons, then you can't use a pseudopotential / PAW framework. You can always test this by including semi-core electrons as valence with the _sv or _pv potentials.

[Chaure]

Dear Professor,
Thank you so much for your quick reply. So I will do SC calculation using _sv or _pv potential. Is it that the correct way of Choosing NG(X, Y, Z)F value is to take from optimized OUTCAR or CHGCAR. I will definitely do few calculations for Valence charge integration.Is it that each time I need to run bader program and check the total number of electrons given by ZVAL to be equal to the total electrons written in ACF.dat ? I want to be in touch with you until I get success in calculating charge transfer.
Sincerely,

[graeme]

I suggest using the default values of NG(X,Y,Z)F and then increase them by 50% to check for convergence. If you are integrating the valence charge, you will not need a super-fine grid. The total number of electrons should be precisely correct, independent of the grid density - this is a property of the CHGCAR file - so checking for convergence is really only for the partial charges. You can also use the "-b weight" method for faster convergence as a function of grid density.

[Chaure]

Dear Professor,
Using Valence Charge density only , I don't get the value of charge transfer as accurate as found in literature. So, I will let you know how I will go doing with fine grid convergence.
Sincerely,

[Chaure]

Dear Professor,
Some of the elements like Sb, Zn do not have _sv or _pv potential but have _sv_GW or _pv_GW. In that case , can we use _GW potential to calculate charge transfer.
Sincerely,
DT

[graeme]

I would not recommend it. I would just use the standard Sb and Zn potentials. Zn, in particular, has plenty of valence electrons.

[skm]

Dr Graeme,

As i need to understand more, about your part reply above, as i copied below: ((this is to check whether my procedure is correct or not) )

".....One finally note is that you need a fine fft grid to accurately reproduce the correct total core charge. It is essential to do a few calculations, increasing NG(X,Y,Z)F until the total charge is correct. "

what do you mean by this? and in actual check where and what you are expecting to check?

As in my case i always get total number of electrons from OUTCAR, ACF.dat file (after doing 'bader CHGCAR -ref CHGCAR_sum) are same. Is this not what is required?

Regards

[graeme]

If you are analyzing the CHGCAR file, the sum of the valence charge will always be correct. Increasing the fft grid density can be used to check convergence of the Bader charges on each atom.

[Chaure]

Dear Professor,
I want to do convergence of bader charge on each atom by increasing fft. Here also, I have never achieved total electrons in ACF.dat file equal to total ZVAL for all atoms though I keep on increasing FFT in each step by 50%.
Sincerely,

[Chaure]

Dear Professor,
I am able to get equal number of electrons in ACF.dat and total ZVAL even with default FFT, I mean without increasing its value. Finally, I want to understand that how and when we could be confirmed that our FFT values are converged? If the default values are giving correct charge transfer, do we still need to increase its value and do further calculation ?
Sincerely,

[graeme]

If you are using a command like: bader CHGCAR -ref CHGCAR_sum

then you should always get the correct total valence charge. Double check this with the -vac off flag.

If you do not see the total correct charge in the ACF.dat file, and you are willing to post the files here, I can take a look at it.

[Chaure]

Dear Professor,
I attached my files here. I have issued bader -b weight CHGCAR -ref CHGCAR_sum.Please let me know if i need to do further FFT grid convergence calculation or not. I am not able to attach files here. So i sent your email. Later you can paste my files here.
Sincerely,

[graeme]

In my test I get the following output. Both the analysis and the ACF.dat report a total number of valence electrons of 136, which is expected for your system (12*8 for Zn and 5*8 for Sb).


-----------

bader CHGCAR -ref CHGCAR_sum

GRID BASED BADER ANALYSIS (Version 1.03 11/13/17)

OPEN ... CHGCAR
VASP5-STYLE INPUT FILE
DENSITY-GRID: 96 x 120 x 128
CLOSE ... CHGCAR
RUN TIME: 0.49 SECONDS

OPEN ... CHGCAR_sum
VASP5-STYLE INPUT FILE
DENSITY-GRID: 96 x 120 x 128
CLOSE ... CHGCAR_sum
RUN TIME: 0.48 SECONDS

CALCULATING BADER CHARGE DISTRIBUTION
0 10 25 50 75 100
PERCENT DONE: **********************

REFINING AUTOMATICALLY
ITERATION: 1
EDGE POINTS: 1215984
REASSIGNED POINTS: 68978

RUN TIME: 3.60 SECONDS

CALCULATING MINIMUM DISTANCES TO ATOMS
0 10 25 50 75 100
PERCENT DONE: **********************
RUN TIME: 0.22 SECONDS

WRITING BADER ATOMIC CHARGES TO ACF.dat
WRITING BADER VOLUME CHARGES TO BCF.dat

NUMBER OF BADER MAXIMA FOUND: 38684
SIGNIFICANT MAXIMA FOUND: 10921
VACUUM CHARGE: 0.0000
NUMBER OF ELECTRONS: 136.00000

-----------
cat ACF.dat
# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 2.886798 0.832153 7.178721 12.143279 1.022100 23.128599
2 3.396977 6.987807 1.046296 12.143333 1.022100 23.128873
3 0.255090 6.987807 3.066212 12.145062 0.979805 23.080633
4 6.028685 0.832153 5.158805 12.145225 0.979805 23.081455
5 3.396977 4.742133 5.158805 12.167561 1.022100 23.186707
6 2.886798 3.077827 3.066212 12.167515 1.022100 23.186707
7 6.028685 3.077827 1.046296 12.189509 1.005420 23.682543
8 0.255090 4.742133 7.178721 12.189454 1.005420 23.682269
9 0.890713 0.648736 0.899899 4.851099 1.024039 27.454899
10 5.393062 7.171224 7.325118 4.851595 1.024039 27.456269
11 2.251175 7.171224 5.012408 4.854683 1.002889 27.262759
12 4.032600 0.648736 3.212609 4.854190 1.002889 27.261388
13 5.393062 4.558716 3.212609 4.830665 1.002889 27.196154
14 0.890713 3.261244 5.012408 4.829971 1.002889 27.194235
15 4.032600 3.261244 7.325118 4.818680 1.054414 27.093095
16 2.251175 4.558716 0.899899 4.818178 1.054414 27.091450
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0000
VACUUM VOLUME: 0.0000
NUMBER OF ELECTRONS: 136.0000
----------

[Chaure]

Dear Professor,
So in this case I don't need further calculations increasing FFT for convergence, right ?