The ABINIT Users Mailing List ( CLOSED )

[Fred Nastos <nastos@physics.utoronto.ca>]

On February 12, 2003 03:47 pm, rhxie@nist.gov wrote:
> (I attached the input files: AlAs2.in and AlAs2.files). For this case,
> I noticed that the gap is zero at  Gamma point (see gap.ps attached).
> This is not true. Pls help me figure out what happens. Thanks.

Short answer:  Specify the primitive lattice translations, or be
more careful about your acell value, your xred values, or you
choice of reciprocal lattice points in the bandstructure.

Long answer:  When I use your input file I reproduce the same band
structure you provided.  I modified your file AlAs.in slightly
and got a very reasonable bandstructure, in that it looks like the
published ones.  I am using ABINITv3.4.3.

Instead of
angdeg  60.0 60.0 60.0

I used
rprim   0 .5 .5
       .5  0 .5
       .5 .5  0

since AlAs is a zincblende semiconductor.

The bandstructure I get then has an indirect gap at Gamma-X of
about 1.19 eV (the experimental value is about 2.1 eV, but that
is to be expected).  My energy separation at Gamma is about 1.95
eV.

More details:

In my output file corresponding to your input file I find the lines
     rprim    5.7735026919E-01  0.0000000000E+00  8.1649658093E-01
             -2.8867513459E-01  5.0000000000E-01  8.1649658093E-01
             -2.8867513459E-01 -5.0000000000E-01  8.1649658093E-01

Which do not look like the usual primitive lattice vectors for a
zincblende structure (e.g. Ashcroft&Mermin).  Of course, they
are still acceptable,provided everything else is defined properly.
The online help, under angdeg, says
### begin quote ####
If the three angles are equal within 1.0d-12, the three primitive vectors
are chosen so that the trigonal symmetry that exchange them is along the
z cartesian axis : 
R1=( a  ,           0,c)
R2=(-a/2, sqrt(3)/2*a,c)
R3=(-a/2,-sqrt(3)/2*a,c)
where a^2+c^2=1.0d0 
#### end quote ####

Many quantities are defined in terms of rprim.  I think your choices
for acell, xred, and the bandstructure high-symmetry points are not
necessarily correct given the automatically generated rprim.

--
Fred Nastos
Department of Physics
University of Toronto
Toronto, Canada