Hi,
I just want to verify if these are the correct steps to Bader Charge Analysis:
Steps:
1. Relax the system to a converged geometry.
2. Using the relaxed geometry, perform static calculation using the following tags:
LAECHG=.TRUE.
3. Sum the output charge files AECCAR0 AECCAR2 using the the chgsum.pl script. Output the sum to CHGCAR file.
4. Run the Bader charge analysis to CHGCAR output from step#3.
5. Check if converged and vary NGXF, NGYF, NGZF accordingly (Iterate from step#2 to step#5 until convergence is achieved.)
Question:
1. For step #2, do I need to set LASPH=.TRUE.? And should I just use the default ICHARG=0?
2. For step#3, what output value should I be checking to check for convergence?
Thanks,
Shaira
Mr. Narido,
I do not know in VASP how you should compute the electron densities technically, but in any case the step 1 is a correct one. However, you could check in VASP for computation of the Laplacian densities. In 'Atoms in molecules', this is the second approach available.
LASPH=.TRUE. is not necessary in most of the case. You may want to refer to http://theory.cm.utexas.edu/henkelman/code/bader/ for detailed instructions
Hi All, For item#2, I realized that increasing the FFT mesh density by setting the NGXF, NGYF, and NGXF can cause the charge per atom to converge to a certain value. Basically, we get the charge value per atom from the ACF.dat file per atom. Then we compare the difference of the charges before and after FFT mesh increase (e.g. 3*FFT - 2*FFT). I tried up to four times the default NGxF settings; at 5 times the settings, the CPU cannot handle the memory requirements.
Also, ADDGRID=.TRUE. can give better or similar effect with quadrupling the original FFT grid. So, leaving the NGxF settings as is and setting ADDGRID=.TRUE. should be enough for Bader analysis, at least for the system I studied.
Hope this helps.
Hi,
Method you described is not optimal. For best results use the following steps:
1) Get your desired geometry, either through geometry relaxation or experiments or any other method of your choice.
2) Prepare an INCAR file with the following keywords: NSW = 0 (don't change geometry), Prec = Accurate (use accurate integration grids), LAECHG = .TRUE. (generate the AECCAR0 and AECCAR2 files, LCHARG = .TRUE. (generate the CHGCAR file), and ENCUT = value equal to or larger than 400 eV.
3) Sum the output charge files AECCAR0 AECCAR2 using the the chgsum.pl script to generate the CHGCAR_sum file.
4) Run the Bader charge analysis program using the following command:
bader CHGCAR -ref CHGCAR_sum
As described on http://theory.cm.utexas.edu/henkelman/code/bader/ this procedure uses all electrons (CHGCAR_sum) to draw the compartments, then integrates the valence pseudodensity (CHGCAR) over these compartments to get the valence electron populations. This will avoid any need to iterate between steps # 2 & 5 in your procedure and will avoid the need to use such a fine integration grid or to add extra spheres the procedure in your original post requires.
5) Use this equation to compute the Bader net atomic charges:
z - num_frozen_core - Bader population = Bader net atomic charge
where z is the atomic number, num_frozen_core is the number of frozen core electrons for that atom in the PAW potential you used, and Bader population is the number printed in the ACF.dat file.
However, that involves some manual work. An easier approach is to use the free Chargemol program (http://ddec.sourceforge.net) to compute the DDEC6 net atomic charges and bond orders. You follow steps # 1 and 2 above as before. However, then you have the following remaining step:
3) Place the AECCAR0, AECCAR2, CHGCAR, and POTCAR files into a directory along with the job_control.txt file. (See the Chargemol program manual at the above link for details on how to set up the job_control.txt file.) Then run the Chargemol program to get the net atomic charges, atomic spin moments (for magnetic materials), atomic dipoles and quadrupoles, bond orders, overlap populations, and other atomic properties printed in easy-to-read text files.
The DDEC6 methodology is described in the following publications:
(1) T. A. Manz and N. Gabaldon Limas, “Introducing DDEC6 atomic population analysis: part 1. Charge partitioning theory and methodology,” RSC Advances, 6 (2016) 47771-47801 (http://dx.doi.org/10.1039/C6RA04656H).
(2) N. Gabaldon Limas and T. A. Manz, “Introducing DDEC6 atomic population analysis: part 2. Computed results for a wide range of periodic and nonperiodic materials,” RSC Advances, 6 (2016) 45727-45747 (http://dx.doi.org/10.1039/C6RA05507A).
(3) T. A. Manz, "Introducing DDEC6 atomic population analysis: part 3. Comprehensive method to compute bond orders," RSC Advances, 7 (2017) 45552-45581 (http://dx.doi.org/10.1039/C7RA07400J ). (open access)
Dear Dr. Manz,
First congratulation for launching the Chargemol program !
I am little confused in calculating bader charge. Can you clarify my doubt according to above method finally I have ACF.dat file in which charge of each atom is written. Now how will calculate bader net charge above mentioned formula
z - num_frozen_core - Bader population = Bader net atomic charge
(1) z,
I got atomic number for particular atom
(2) num_frozen_core,
I am not getting as you wrote this is the number of frozen core electrons for that atom in the PAW potential this means if for example W has ZVAL=12 electron in PAW potential and now remaining 62 electrons will be num_frozen_core, (because W has 74 atomic number)
(3) Bader population:
From ACF.dat what we will take charge on particular atom.
Please clarify my doubts !
I, also want to try your program but it will take time to me to learn. Here I want to ask how much time your code takes for calculating bader net charge.
Thank You
Shilendra
Hi Shilendra,
answer to 1st question: You can simplify the above formula using
atomic_number - num_frozen_core = ZVAL value printed in VASP POTCAR
which gives
Bader net atomic charge = ZVAL - Bader population
where Bader population is printed in ACF.dat.
answer to 2nd question: Here I want to ask how much time your code takes for calculating bader net charge?
The Chargemol program computes DDEC6 not Bader results. The DDEC6 net atomic charges are optimized to simultaneously describe chemical states (i.e., charge transfer between atoms in materials) and to approximately reproduce the electrostatic potential surrounding materials (which makes them great for constructing force-fields). It takes about 10 cpu seconds per atom to perform the DDEC6 analysis. So if you have a material containing 100 atoms, it will take ~100 seconds to complete the calculation if run in parallel on 10 processors, or ~17 minutes if you run in serial mode on a single processor. You get lots of outputs in addition to the net atomic charges. The outputs include net atomic charges, atomic dipoles and quadrupoles, atomic spin moments (for magnetic materials), bond orders, etc.
Dear Manz,
I tried your serial code it is giving me same bader charge as I calculated from bader code and other properties also. It took almost 20-30 minutes for my 128 atoms magnetic cell.
Is there any output from your chargemol code which we can visualize and by which software ?
Thank You for developing such nice code !
Shilendra
Hi Shilendra,
> Is there any output from your chargemol code which we can visualize and by which software ?
Yes, the net atomic charges, atomic spin moments, sum of bond order (SBO) for each atom, etc. can be read and displayed by the Jmol program, which is a free molecular visualization program. You can label or color the atoms by these properties. Please see 'Part D: Reading and Visualizing the Chargemol Output Files' in the instructions.pdf file in the Chargemol program folder instructions and examples of visualizing the code outputs with Jmol.
> it is giving me same bader charge as I calculated from bader code and other properties also
Well, the properties computed by the Chargemol program are the DDEC6 values, not the Bader values. The DDEC6 values have lots of nice properties as described in these papers: http://dx.doi.org/10.1039/c7ra07400j (open access), http://dx.doi.org/10.1039/c6ra04656h, http://dx.doi.org/10.1039/c6ra05507a
Glad you found the program useful!
Tom
Thomas A. Manz
hi,
read the procedure you explained. i have posted my query at another question raised by different member.
but here i wish to ask same, if you correct me if my procedure is wrong, please.
for my hybrid system i.e. say Material A and B. that is B placed on surface of A.
1. Relaxed the mixed system (i.e. A+B).
2. then run the resulted system with another INCAR file to generate the AECCAR0 and AECCAR2 files
3) Run chgsum.pl script to generate the CHGCAR_sum file.
4) Run the Bader charge analysis using :
bader CHGCAR -ref CHGCAR_sum
5) now if i want to check whether the material B accumulated charge or released charge, i did as below:-
a) subtracted the value of CHARGE in ACF.dat file from ZVAL in POTCAR file of that element. if this subtraction resulted in +ve value then it indicates Charge is depleted and if -ve, then charge accumulated.
b) after done this for all atoms sum up the results and find net +ve or -ve value for material B, and then decide accumulation or depletion.
does this procedure correct? any comment?
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science