This should not happen so I looked into CHGCAR. I did a integration over the whole supercell I used, and normalised the charge to number of atoms. I found that for less than 10 layers, the charge per atom is 1.5; while for more than 11 layers, the charge per atom is 4.0. The later one should be correct since it is InP.
I can make sure I only changed POSCAR. I only changed the number of atoms and vacuum thickness in POSCAR. INCAR, POTCAR and KPOINTS are kept identical. Vacuum thickness has also been tested and I found it is irrelevant.
POSCAR for 10 layers
Code: Select all
In??????P
4.242463011
1.4142135624????0???????0
0???????1???????0
0???????0???????10
10?10
cartesian
0???????0???????0
0.7071067812????0.5?????0.5
0???????0???????1
0.7071067812????0.5?????1.5
0???????0???????2
0.7071067812????0.5?????2.5
0???????0???????3
0.7071067812????0.5?????3.5
0???????0???????4
0.7071067812????0.5?????4.5
0.3535533906????0.5?????0
1.0606601718????0???????0.5
0.3535533906????0.5?????1
1.0606601718????0???????1.5
0.3535533906????0.5?????2
1.0606601718????0???????2.5
0.3535533906????0.5?????3
1.0606601718????0???????3.5
0.3535533906????0.5?????4
1.0606601718????0???????4.5
POSCAR for 11 layers
Code: Select all
In??????P
4.242463011
1.4142135624????0???????0
0???????1???????0
0???????0???????11
11?11
cartesian
0???????0???????0
0.7071067812????0.5?????0.5
0???????0???????1
0.7071067812????0.5?????1.5
0???????0???????2
0.7071067812????0.5?????2.5
0???????0???????3
0.7071067812????0.5?????3.5
0???????0???????4
0.7071067812????0.5?????4.5
0.3535533906????0.5?????0
1.0606601718????0???????0.5
0.3535533906????0.5?????1
1.0606601718????0???????1.5
0.3535533906????0.5?????2
1.0606601718????0???????2.5
0.3535533906????0.5?????3
1.0606601718????0???????3.5
0.3535533906????0.5?????4
1.0606601718????0???????4.5
0.3535533906????0.5?????5