[Wien] vacuum thickness

[Stefaan Cottenier]

> Here I¡¯m dealing with a phase boundary question. For the
> following struct I post, when perform initialization, sgroup always produces
> another struct,

If I use the case.struct which you sent, sgroup does not change it. 

> which seems neither z=0 or z=0.5 will be the center of the
> vacuum. 

The structure you sent is a 5-layer Ni(001) slab, with in one of the two 
surfaces 50% of Ni replaced by Al. This surface layer is at z=2/3 (with the 
two Ni-layers above it belonging to the material, and the region below it 
vacuum), while the pure Ni surface layer is at z=1/3 (above it vacuum, below 
it material). The centre of the vacuum is at z=0.5.

Two remarks:

1) You will have more symmetry in the structure (and hence less k-points = 
faster calculation) if you treat both surfaces in the same way (= put also 50% 
of Al in the other surface layer). You will even keep inversion symmetry then, 
which allows to use the real version of the code = faster, and less memory 
needed.

2) Your Ni slab consists of 5 layers only, which is the very minimum, and 
almost always not thick enough for sufficiently accurate surface calculations. 
My experience is that at least 7 layers are needed. Moreover, your current 
vacuum is very small: 6.7 au (= 1 Ni lattice constant). You'll probably need a 
few times this value.

Stefaan