the total charge

[hlzya]

Hi,

I am using the VASP code to calculate the core charge and valence charge for a Ag-doped Ni surface, and then using the bader code to get the local valence charges of individual surface atoms. Now I have some problems to get the exact core charge. There are 35 Ni and 1 Ag atoms involved in the syetem, i.e., there should be 1027 electrons. The relationship between the calculated total electrons and FFT grid is listed below. I find it is very difficult to get the exact total electrons. I do use the PAW method and the POSCAR is also attached. Could someone tell me if I should further enlarge the FFT grid or something else is wrong with my calculation?

Thanks.

Yi-An
---------------------------
90X90X200 1755
100X100X240 1054
192X192X192 1099
---------------------------
POSCAR
------------------------------------------------------------------------
Ni
1.
7.4925034534526560 0.0000000000000000 0.0000000000000000
0.0000000000000000 7.4925034534526578 0.0000000000000000
0.0000000000000000 0.0000000000000000 17.2980000000000018
35 1
Selective dynamics
Direct
0.0006771212088093 0.0000000000000000 0.0979515878255582 T T T
0.0005039909122803 0.3328293424210551 0.0975054707159210 T T T
0.0005039909122803 0.6671706575789442 0.0975054707159210 T T T
0.3333333333333357 0.0000000000000000 0.0981987973797782 T T T
0.3333333333333357 0.3326562121245263 0.0979515878255582 T T T
0.3333333333333357 0.6673437878754732 0.0979515878255582 T T T
0.6659895454578544 0.0000000000000000 0.0979515878255582 T T T
0.6661626757543836 0.3328293424210551 0.0975054707159210 T T T
0.6661626757543836 0.6671706575789442 0.0975054707159210 T T T
-0.0086582381078587 0.0000000000000000 0.2978996140226131 T T T
0.0000823207919708 0.3332510125413647 0.2979591909898756 T T T
0.0000823207919708 0.6667489874586349 0.2979591909898756 T T T
0.3333333333333357 0.3419915714411944 0.2978996140226131 T T T
0.3333333333333357 0.6580084285588057 0.2978996140226131 T T T
0.6753249047745222 0.0000000000000000 0.2978996140226131 T T T
0.6665843458746931 0.3332510125413647 0.2979591909898756 T T T
0.6665843458746931 0.6667489874586349 0.2979591909898756 T T T
0.1666666666666643 0.1666666666666642 0.0000000000000000 F F F
0.1666666666666643 0.5000000000000000 0.0000000000000000 F F F
0.1666666666666643 0.8333333333333357 0.0000000000000000 F F F
0.5000000000000000 0.1666666666666642 0.0000000000000000 F F F
0.5000000000000000 0.5000000000000000 0.0000000000000000 F F F
0.5000000000000000 0.8333333333333357 0.0000000000000000 F F F
0.8333333333333357 0.1666666666666642 0.0000000000000000 F F F
0.8333333333333357 0.5000000000000000 0.0000000000000000 F F F
0.8333333333333357 0.8333333333333357 0.0000000000000000 F F F
0.1679407924229348 0.1653925409103937 0.2017142986793051 T T T
0.1654843872600231 0.5000000000000000 0.1993625847547351 T T T
0.1679407924229348 0.8346074590896061 0.2017142986793051 T T T
0.4987258742437288 0.1653925409103937 0.2017142986793051 T T T
0.5011822794066406 0.5000000000000000 0.1993625847547351 T T T
0.4987258742437288 0.8346074590896061 0.2017142986793051 T T T
0.8333333333333357 0.1678489460733053 0.1993625847547351 T T T
0.8333333333333357 0.5000000000000000 0.2001701920818864 T T T
0.8333333333333357 0.8321510539266946 0.1993625847547351 T T T
0.3333333333333357 0.0000000000000000 0.3272573632727211 T T T
-----------------------------------------------------------------------------------

[graeme]

You will not be able to accurately integrate the total charge density because of the cusps at the atomic centers. Instead, use the total charge to do the partitioning and the integration on the valance charge density. These values will converge with modest grid sizes. To do this, follow the instructions on this web page: http://theory.cm.utexas.edu/bader/vasp.php

[hlzya]

Hi,

I further increase the FFT grid to 200X200X480, and get the total electron number of 1039. I think it is an acceptable result and I can get the valence charge based on this CHGCAR_sum.

Thank you and happy new year.

Yi-An ZHU