Bader volume to RWIGS in VASP
Posted: Wed Jun 28, 2006 9:13 am
Dear Bader users,
Graeme think it might be worthwile to post the discussion we had on a topic of correlating the 'min. dist. to atoms' from the Bader analysis to the RWIGS values in VASP.
Here it is (in the forward timing order).
------------------------------------------------------------
On Jun 28, 2006, at 12:14 AM, Yi-Ren Tzeng wrote:
Dear Prof. Henkelman,
I was using your bader code to set the rwigs values in the VASP. On the TiO2 surface, everything worked perfectly. However, there seems to exist an unresolved issue for Si And H, to my own experience and according to bader forum.
To get a better understanding, I tried to repeat the bader analysis in the paper of 'Computational Materials Science 36 (2006) 354–360'. The first one I tried is H2O, which I only got total of 8 electrons from bader analysis, with the density of core electrons [O(1s)?] being deleted from the total density.
I wondered if you could check if I have done something wrong, or have missed some keywords in the (Gaussian) calculation to generate the density file.
Below is the input files of H2O and the commands used to generate the density.
command line:
-------
formchk h2o.chk h2o.fchk
cubegen 0 density=scf h2o.fchk h2o.cube 150 h
---------
H2O.com
----------------------------------------
%chk=h2o
%mem=100MW
%nproc=2
#opt mp2/aug-cc-pvdz pop=full
title Card Required
0 1
O 0.000000 0.276598 0.000000
H 0.763578 0.861701 0.000000
H -0.763578 0.861701 0.000000
------------------------------------------
The final goal is to get the min-dist values for Si and H from bader analysis of gaussian file and used them as the rwigs values in VASP.
BTW, I tried to register to bader forum, but the activation email never come. (phtobias: yrt.subscribe@gmail.com)
I will appreciate your kind help.
Yi-Ren
On Jun 28, 2006, at 1:43 PM, Graeme Henkelman wrote:
Yi-Ren,
I don't know much about running Gaussian. The H2O charge density file was generated by a graduate
student familiar with Gaussian using input file below. This was generated quite a while ago, so the
tag to generate the full density, including the core, might have changed. But this was the input file used
to generate the H2O density file on our bader webpage. If this works for you, or if you learn how to get
the core charges in Gaussian, let me know.
How does it make any sense to use the minimum distance between the Bader surface at the atomic
centers as RWIGS in vasp?
- GH
-------------
%Mem=28000000
%Chk=w.13.hf.chk
#P HF/aug-cc-pVDZ SCF=(Tight) Guess=(Read) Density=(Current) NoSymm
# Cube=(Full,Density,Cards) IOp(6/48=20)
H2O Density Calculation.
0 1
H 0.000000 0.756950327 0.520784245
O 0.000000 0.000000000 -0.065098030
H 0.000000 -0.756950327 0.520784245
H2O_HF_aug-cc-pVDZ.den
1 -4.00 -4.00 -4.00
257 0.03125 0.00 0.00
257 0.00 0.03125 0.00
257 0.00 0.00 0.03125
On Jun 28, 2006, at 2:43 AM, Yi-Ren Tzeng wrote:
Dear Prof. Henkelman,
Thank you very much for your kind help in gaussian input file, for activating the account and for questioning my usage of bader-rwigs in the VASP.
First, the input file worked perfectly. I will send you a revised input file in the g98 and g03 format. Secondly, I can login into the bader forum now. (Should I post our discussion to the forum?)
In the followings, I will try to explain my reasoning of correlating the the bader volume to the rwigs in VASP. (Please excuse me for the poor English.) First of all, I assumed the 'min-dist to atoms' is the min-dist of bader volume (please correct me, if I am wrong). Furthermore, I approximate the bader volume to be a sphere, which becomes the rwigs in VASP. As you know, to assign rwigs in a VASP DOS calculation is quite random. This randomness makes the otherwise very useful population (charge) analysis, in principle, meaningless.
The system we study is TiO2, an ionic semiconductor. In the past, by playing around with rwigs, I never got the charge right (O:-1, Ti:+2, J. Phys. Chem. B 2004, 108, 7844-7853, thanks to 'aarondesk' for posting the ref. in the bader forum). When I put the max. of baderized-rwigs (O:0.75, Ti:1.7 angstrom, obtained from 'bader CHGCAR') back to VASP, I got a fairly reasonable charge (O:between -1.0 and -1.4, Ti:+3).
Given this successful bader way, I am quite tempted to get a more accurate charge analysis by replacing the quite vague 'max. of min-dist' with each atom's own baderized-rwigs.
The only foreseen obstacle is the rwigs of Si and H, of which I might have to settle down with the gaussian version.
Thanks again for your kind and prompt message.
All the best,
Yi-Ren
On Jun 28, 2006, at 4:28 PM, Graeme Henkelman wrote:
Great, good to know that worked. Yes, can you please repost this to the forum. I want to move to the forum so that my students can help with answering question (and learn from comments like yours), and so that other people, who might have similar questions, can read the discussion and get some help. Cheers, Graeme
Dear Prof. Henkelman,
Below please find the h2o.com input file for generating all electron density cube file. In principle it is almost the same as the one you show me, except I replaced the HF with MP2, which gave almost the same (HF/MP2) bader results.
I will post our discussion to the bader forum. Thanks for the kind help. Hope I can get a proper Si and H rwigs by using your excellent bader code.
yi-ren
---- h2o.com starts from here ----
%chk=h2o
%mem=100MW
%nproc=2
# opt mp2/aug-cc-pVDZ pop=full
this first optimizes the H2O geometry and then generate the density cube file (in link1)
0 1
O 0.000000 0.269652 0.000000
H 0.760508 0.865174 0.000000
H -0.760508 0.865174 0.000000
--link1--
%chk=h2o
# mp2/aug-cc-pVDZ geom=allcheck guess=(read,only) cube=(full,100,density)
h2o.cube
---- h2o.com ends here ----
Below is the output from bader.
------------------------------------------------------------------------------------------
# X Y Z VORONOI BADER % MIN DIST
------------------------------------------------------------------------------------------
1 4.9707 4.9707 4.9707 7.2920 9.1735 92.0113 1.3658
2 4.9707 6.4079 3.8454 1.3390 0.3982 3.9943 0.2166
3 4.9707 3.5336 3.8454 1.3390 0.3982 3.9943 0.2166
NUMBER OF BADER MAXIMA FOUND: 3
SIGNIFICANT MAXIMA FOUND: 3
NUMBER OF ELECTRONS: 9.97000
------------------------------------------------------------------------------------------
This compares the charge analysis of current result to that in your paper 'Comput.Mat.Sci.36(2006)354':
-------------------------
paper current
-------------------------
O: -1.16 -1.17
H: +0.58 +0.60
-------------------------
Graeme think it might be worthwile to post the discussion we had on a topic of correlating the 'min. dist. to atoms' from the Bader analysis to the RWIGS values in VASP.
Here it is (in the forward timing order).
------------------------------------------------------------
On Jun 28, 2006, at 12:14 AM, Yi-Ren Tzeng wrote:
Dear Prof. Henkelman,
I was using your bader code to set the rwigs values in the VASP. On the TiO2 surface, everything worked perfectly. However, there seems to exist an unresolved issue for Si And H, to my own experience and according to bader forum.
To get a better understanding, I tried to repeat the bader analysis in the paper of 'Computational Materials Science 36 (2006) 354–360'. The first one I tried is H2O, which I only got total of 8 electrons from bader analysis, with the density of core electrons [O(1s)?] being deleted from the total density.
I wondered if you could check if I have done something wrong, or have missed some keywords in the (Gaussian) calculation to generate the density file.
Below is the input files of H2O and the commands used to generate the density.
command line:
-------
formchk h2o.chk h2o.fchk
cubegen 0 density=scf h2o.fchk h2o.cube 150 h
---------
H2O.com
----------------------------------------
%chk=h2o
%mem=100MW
%nproc=2
#opt mp2/aug-cc-pvdz pop=full
title Card Required
0 1
O 0.000000 0.276598 0.000000
H 0.763578 0.861701 0.000000
H -0.763578 0.861701 0.000000
------------------------------------------
The final goal is to get the min-dist values for Si and H from bader analysis of gaussian file and used them as the rwigs values in VASP.
BTW, I tried to register to bader forum, but the activation email never come. (phtobias: yrt.subscribe@gmail.com)
I will appreciate your kind help.
Yi-Ren
On Jun 28, 2006, at 1:43 PM, Graeme Henkelman wrote:
Yi-Ren,
I don't know much about running Gaussian. The H2O charge density file was generated by a graduate
student familiar with Gaussian using input file below. This was generated quite a while ago, so the
tag to generate the full density, including the core, might have changed. But this was the input file used
to generate the H2O density file on our bader webpage. If this works for you, or if you learn how to get
the core charges in Gaussian, let me know.
How does it make any sense to use the minimum distance between the Bader surface at the atomic
centers as RWIGS in vasp?
- GH
-------------
%Mem=28000000
%Chk=w.13.hf.chk
#P HF/aug-cc-pVDZ SCF=(Tight) Guess=(Read) Density=(Current) NoSymm
# Cube=(Full,Density,Cards) IOp(6/48=20)
H2O Density Calculation.
0 1
H 0.000000 0.756950327 0.520784245
O 0.000000 0.000000000 -0.065098030
H 0.000000 -0.756950327 0.520784245
H2O_HF_aug-cc-pVDZ.den
1 -4.00 -4.00 -4.00
257 0.03125 0.00 0.00
257 0.00 0.03125 0.00
257 0.00 0.00 0.03125
On Jun 28, 2006, at 2:43 AM, Yi-Ren Tzeng wrote:
Dear Prof. Henkelman,
Thank you very much for your kind help in gaussian input file, for activating the account and for questioning my usage of bader-rwigs in the VASP.
First, the input file worked perfectly. I will send you a revised input file in the g98 and g03 format. Secondly, I can login into the bader forum now. (Should I post our discussion to the forum?)
In the followings, I will try to explain my reasoning of correlating the the bader volume to the rwigs in VASP. (Please excuse me for the poor English.) First of all, I assumed the 'min-dist to atoms' is the min-dist of bader volume (please correct me, if I am wrong). Furthermore, I approximate the bader volume to be a sphere, which becomes the rwigs in VASP. As you know, to assign rwigs in a VASP DOS calculation is quite random. This randomness makes the otherwise very useful population (charge) analysis, in principle, meaningless.
The system we study is TiO2, an ionic semiconductor. In the past, by playing around with rwigs, I never got the charge right (O:-1, Ti:+2, J. Phys. Chem. B 2004, 108, 7844-7853, thanks to 'aarondesk' for posting the ref. in the bader forum). When I put the max. of baderized-rwigs (O:0.75, Ti:1.7 angstrom, obtained from 'bader CHGCAR') back to VASP, I got a fairly reasonable charge (O:between -1.0 and -1.4, Ti:+3).
Given this successful bader way, I am quite tempted to get a more accurate charge analysis by replacing the quite vague 'max. of min-dist' with each atom's own baderized-rwigs.
The only foreseen obstacle is the rwigs of Si and H, of which I might have to settle down with the gaussian version.
Thanks again for your kind and prompt message.
All the best,
Yi-Ren
On Jun 28, 2006, at 4:28 PM, Graeme Henkelman wrote:
Great, good to know that worked. Yes, can you please repost this to the forum. I want to move to the forum so that my students can help with answering question (and learn from comments like yours), and so that other people, who might have similar questions, can read the discussion and get some help. Cheers, Graeme
Dear Prof. Henkelman,
Below please find the h2o.com input file for generating all electron density cube file. In principle it is almost the same as the one you show me, except I replaced the HF with MP2, which gave almost the same (HF/MP2) bader results.
I will post our discussion to the bader forum. Thanks for the kind help. Hope I can get a proper Si and H rwigs by using your excellent bader code.
yi-ren
---- h2o.com starts from here ----
%chk=h2o
%mem=100MW
%nproc=2
# opt mp2/aug-cc-pVDZ pop=full
this first optimizes the H2O geometry and then generate the density cube file (in link1)
0 1
O 0.000000 0.269652 0.000000
H 0.760508 0.865174 0.000000
H -0.760508 0.865174 0.000000
--link1--
%chk=h2o
# mp2/aug-cc-pVDZ geom=allcheck guess=(read,only) cube=(full,100,density)
h2o.cube
---- h2o.com ends here ----
Below is the output from bader.
------------------------------------------------------------------------------------------
# X Y Z VORONOI BADER % MIN DIST
------------------------------------------------------------------------------------------
1 4.9707 4.9707 4.9707 7.2920 9.1735 92.0113 1.3658
2 4.9707 6.4079 3.8454 1.3390 0.3982 3.9943 0.2166
3 4.9707 3.5336 3.8454 1.3390 0.3982 3.9943 0.2166
NUMBER OF BADER MAXIMA FOUND: 3
SIGNIFICANT MAXIMA FOUND: 3
NUMBER OF ELECTRONS: 9.97000
------------------------------------------------------------------------------------------
This compares the charge analysis of current result to that in your paper 'Comput.Mat.Sci.36(2006)354':
-------------------------
paper current
-------------------------
O: -1.16 -1.17
H: +0.58 +0.60
-------------------------