#!/bin/sh
# As an example of a non-self-consistent Green's function calculation, we will
# look at the density of states (DOS) and transmission of an infinite 1D chain
# of Aluminum atoms.
# First we need to perform a single-point calculation with ADF on a principal
# layer, consisting, in this case, of four atoms. Since bulk Aluminum has an FCC
# structure with a lattice constant of 4.05 Angstrom, the nearest neighbor
# distance is approximately 2.83 Angstrom. green requires SYMMETRY NOSYM, so we
# have the following input file for the principal layer:
$ADFBIN/adf <<eor
TITLE Principal layer
ATOMS
Al -4.290000 0.000000 0.000000
Al -1.430000 0.000000 0.000000
Al 1.430000 0.000000 0.000000
Al 4.290000 0.000000 0.000000
END
SYMMETRY NOSYM
BASIS
Type DZP
Core Large
CreateOutput None
END
SCF
Converge 1.0e-8
END
XC
LDA SCF VWN
END
eor
mv TAPE21 layer.t21
# The bulk contact geometry consists of three principal layers:
$ADFBIN/adf <<eor
TITLE Bulk
ATOMS
Al -15.730000 0.000000 0.000000 f=left
Al -12.870000 0.000000 0.000000 f=left
Al -10.010000 0.000000 0.000000 f=left
Al -7.150000 0.000000 0.000000 f=left
Al -4.290000 0.000000 0.000000 f=center
Al -1.430000 0.000000 0.000000 f=center
Al 1.430000 0.000000 0.000000 f=center
Al 4.290000 0.000000 0.000000 f=center
Al 7.150000 0.000000 0.000000 f=right
Al 10.010000 0.000000 0.000000 f=right
Al 12.870000 0.000000 0.000000 f=right
Al 15.730000 0.000000 0.000000 f=right
END
SYMMETRY NOSYM
FRAGMENTS
left layer.t21
center layer.t21
right layer.t21
END
XC
LDA SCF VWN
END
SCF
Converge 1.0e-10
AccelerationMethod LISTi
END
eor
mv TAPE21 bulk.t21
# Notice that we have increased the number of SCF iterations. The combination of
# SYMMETRY NOSYM with a 1D chain of metal atoms generally leads to convergence
# problems. This is the main reason why the principal layer consists of only
# four atoms. Fortunately, for larger 3D contacts, the convergence is generally
# better.
# From the bulk TAPE21 file green can calculate the self-energies of the left
# and right contacts. As discussed in the introduction, the self-energy of the
# left contact needs the center and right fragments of the bulk calculation, and
# the self-energy of the right contact needs the center and left fragments.
# Since we need a self-energy matrix for every energy for which we want to
# calculate the DOS and transmission, already here we have to specify the energy
# range. We take 1000 points between -0.4 and 0 Hartree.
$ADFBIN/green <<eor
SURFACE bulk.t21
FRAGMENTS center right
END
EPS -0.4 0 1000
ETA 1e-6
eor
mv SURFACE left.kf
$ADFBIN/green <<eor
SURFACE bulk.t21
FRAGMENTS center left
END
EPS -0.4 0 1000
ETA 1e-6
eor
mv SURFACE right.kf
# Since we want to calculate the DOS and transmission of bare aluminum, we can
# reuse the bulk.t21 file for the extended molecule. We couple the left self-
# energy to the 'left' fragment and the right self-energy to the 'right'
# fragment in bulk.t21. Since we performed restricted ADF calculations, there is
# no difference between spin-A and spin-B and we can omit spin-B from the
# calculation.
$ADFBIN/green <<eor
DOS bulk.t21
TRANS bulk.t21
EPS -0.4 0 1000
ETA 1e-6
LEFT left.kf
FRAGMENT left
END
RIGHT right.kf
FRAGMENT right
END
NOSAVE DOS_B, TRANS_B
eor
# As would be expected for a 1D system, the DOS shows Van Hove singularities at
# the band edges. Apart from oscillations due to the finite size of the system
# in ADF, the transmission only reaches integer values. Between approximately
# -0.35 and -0.15 Hartree, only the sigma channel contributes to the
# transmission. Above -0.15 Hartree also the two pi channels start to
# contribute.
echo ""
echo "Contents of DOS_A:"
cat DOS_A
echo "END"
echo ""
echo "Contents of TRANS_A:"
cat TRANS_A
echo "END"