Bader charges for Ge-Ge Dimers on Ge(2x1) surface

[Eugene]

I'm trying to get correct Bader charges for Ge-Ge dimers on Ge(2x1) surface reconstruction. The slab has 2 identical (2x1) surfaces with 4 dimers per surface. The slab has 64 atoms (with no H passivation). If Ge bulk atom has Bader charge 4, the upper atom in the dimer should have Bader charge around 5 and lower atom in the dimer should have Bader charge around 3. I persistently get different values: upper Ge atom in the dimer has charge around 3.92 |e| and lower Ge atom in the dimer has charge around 3.96 |e|. Values look wrong and relation between upper and lower atom charges looks wrong. I use VASP. Below is one of my INCAR files. In fact, I repeated this simulation with various parameters: a) spin polarized and no-spin; b) various PP's: Ge, Ge_d, Ge_h; c) PREC=Medium and PREC=High; d) KPOINTS were GAMMA 3x3x1 and GAMMA 7x7x1 e) with Bader core correction and without it; f) ISMEAR=0 and ISMEAR=-5. All these simulations returned similar values which look wrong. Please let me know if you know how to fix it. We use the latest "bader" code. Thank you.

# SYSTEM NAME
SYSTEM = Ge 2x1 reconstruction
# OUTPUT AND START
NWRITE=2
ISTART=1
# ELECTRONIC CONVERGENCE
NELM=100
ISMEAR=0
SIGMA=0.2
EDIFF=1E-6
PREC=High
LREAL=Auto
# TYPE OF RUN
IBRION=-1
NSW=0
LAECHG=.TRUE.
# SPIN AND FUNCTIONAL
ISPIN=2
GGA=PE

[graeme]

Charges around 4 sound reasonable; perhaps your intuition about the charge distribution is wrong?