.. _TUT\:sec\:SOC:
Computing relativistic effects in molecules
===========================================
|molcas| is intended for calculations on systems including all atoms of the
periodic table. This is only possible if relativistic effects can be added in a
way that is accurate and at the same time applies to all the methods used in
|molcas|, in particular the CASSCF and CASPT2 approaches. |molcas|
includes relativistic effects within the same wave function framework as used in
non-relativistic calculations. This has been possible by partitioning the
relativistic effects into two parts: the scalar relativistic effects and
spin--orbit coupling. This partitioning is based on the Douglas--Kroll (DK)
transformation of the relativistic Hamiltonian :cite:`Douglas:74,Hess:86`.
Scalar relativistic effects
---------------------------
The scalar relativistic effects are included by adding the corresponding terms
of the DK Hamiltonian to the one-electron integrals in Seward (use
the keyword :kword:`Douglas-Kroll`). This has no effect on the form of the wave
function and can be used with all |molcas| modules. Note however that it is
necessary to use a basis set with a corresponding relativistic contraction.
|molcas| provides the ANO-RCC basis set, which has been constructed using
the DK Hamiltonian. Use this basis set in your relativistic calculations. It has
the same accuracy as the non-relativistic ANO-L basis set. Scalar relativistic
effects become important already for atoms of the second row. With ANO type
basis sets it is actually preferred to use the DK Hamiltonian and ANO-RCC in all
your calculations.
Spin--orbit coupling (SOC)
--------------------------
In order to keep the structure of |molcas| as intact as possible, it was decided
to incorporate SOC as an *a posteriori* procedure which can be added after
a series of CASSCF calculations. The program :program:`RASSI` has been modified
to include the spin--orbit part of the DK Hamiltonian :cite:`Malmqvist:02a`. The
method is thus based on the concept of electronic states interacting via SOC.
In practice this means that one first performs a series of CASSCF calculations
in the electronic states one expects to interact via SOC. They are then used as
the basis states in the RASSI calculations. Dynamic electron correlation effects
can be added by a shift of the diagonal of the SOC Hamiltonian to energies
obtained in a CASPT2 or MRCI calculation. If MS-CASPT2 is used, a special
output file (:file:`JOBMIX`) is provided that is to be used as the input file
for RASSI. The procedure will below be illustrated in a calculation on the lower
excited states of the PbO molecule.
The SO Hamiltonian has been approximated by a one-electron effective
Hamiltonian :cite:`Hess:96`, which also avoids the calculation of multi-center
integrals (the Atomic Mean Field Approximation -- AMFI )
:cite:`Hess:96,Schimmelpfennig:96`.
The :math:`\ce{PbO}` molecule
-----------------------------
Results from a calculation of the potentials for the ground and lower excited
states of :math:`\ce{PbO}`, following the procedure outlined above, has recently been
published :cite:`Roos:03h`. The ground state of :math:`\ce{PbO}` dissociates to :math:`\ce{O}(^3P)` and
:math:`\ce{Pb}(^3P)`. However in the :math:`\ce{Pb}` atom there is strong SOC between the :math:`^3P`, :math:`^1D`,
and :math:`^1S` term of the (6s)\ |2|\(6p)\ |2| electronic configuration. All levels with
the :math:`\Omega` value :math:`O^+` arising from these terms will therefore contribute to
the ground state potential. The first task is therefore to construct the
electronic states that are obtained by coupling :math:`\ce{O}(^3P)` to any of the :math:`^3P`,
:math:`^1D`, and :math:`^1S` terms of :math:`\ce{Pb}`. In the table below we give the states. They have
been labeled both in linear symmetry and in :math:`C_2` symmetry, which is the
symmetry used in the calculation because it makes it possible to average over
degenerate components.
.. _tab\:so:
==== =============== ================================================================================ =============
Spin :math:`C_2` sym Labels in linear symmetry No. of states
==== =============== ================================================================================ =============
2 1 :math:`^5\Delta`, 2\ |x|\ :math:`^5\Sigma^+`, :math:`^5\Sigma^-` 5
2 2 2\ |x|\ :math:`^5\Pi` 4
1 1 3\ |x|\ :math:`^3\Delta`, 3\ |x|\ :math:`^3\Sigma^+`, 4\ |x|\ :math:`^3\Sigma^-` 13
1 2 6\ |x|\ :math:`^3\Pi`, :math:`^3\Phi` 14
0 1 :math:`^1\Delta`, 2\ |x|\ :math:`^1\Sigma^+`, :math:`^1\Sigma^-` 5
0 2 2\ |x|\ :math:`^1\Pi` 4
==== =============== ================================================================================ =============
The total number of states is 45. One thus has to perform 6 CASSCF (and
MS-CASPT2) calculations according to the spin and symmetries given in the table.
The RASSI-SO calculation will yield 134 levels with :math:`\Omega` ranging from 0 to
4. Only the lower of these levels will be accurate because of the limitations in
the selection of electronic states.
The active space used in these calculations is 6s,6p for :math:`\ce{Pb}` and 2p for :math:`\ce{O}`. This
is the natural choice and works well for all main group elements in most
molecules. The s-orbital should be active in groups IIa--Va, but may be left
inactive for the heavier atoms (groups VIa--VIIa). The ANO-RCC basis sets have
been constructed to include correlation of the semi-core electrons. For :math:`\ce{Pb}` they
are the 5d, which should then not be frozen in the :program:`CASPT2`
calculations. All other core electrons should be frozen, because there are no
basis functions to describe their correlation. Including them in the correlation
treatment may lead to large BSSE errors.
The input file for these calculations is quite lengthy, so we show here only one
set of :program:`CASSCF/CASPT2` calculations but the whole :program:`RASSI` input
for all six cases.
.. extractfile:: advanced/RASSI.PbO.input
&GATEWAY
Title= PbO
Coord= $CurrDir/PbO.xyz
Basis set
ANO-RCC-VQZP
Group= XY
AngMom
0.00 0.00 0.00
End of Input
&SEWARD
End of Input
&SCF
Title
PbO
Occupied
24 21
Iterations
20
Prorbitals
2 1.d+10
End of Input
&RASSCF
Title
PbO
Symmetry
1
Spin
5
CIROOT
5 5 1
nActEl
8 0 0
Inactive
23 18
Ras2
3 4
Lumorb
THRS
1.0e-8 1.0e-04 1.0e-04
Levshft
1.50
ITERation
200 50
CIMX
200
SDAV
500
End of Input
&CASPT2
Title
PbO
MAXITER
25
FROZEN
19 16
Focktype
G1
Multistate
5 1 2 3 4 5
Imaginary Shift
0.1
End of Input
>> COPY $Project.JobMix $CurrDir/JobMix.12
&RASSCF
Title
PbO
Symmetry
2
Spin
5
CIROOT
4 4 1
nActEl
8 0 0
Inactive
23 18
Ras2
3 4
Lumorb
THRS
1.0e-8 1.0e-04 1.0e-04
Levshft
1.50
ITERation
200 50
CIMX
200
SDAV
500
End of Input
&CASPT2
Title
PbO
MAXITER
25
FROZEN
19 16
Focktype
G1
Multistate
4 1 2 3 4
Imaginary Shift
0.1
>> COPY $Project.JobMix $CurrDir/JobMix.22
&RASSCF
Title
PbO
Symmetry
1
Spin
3
CIROOT
13 13 1
nActEl
8 0 0
Inactive
23 18
Ras2
3 4
Lumorb
THRS
1.0e-8 1.0e-04 1.0e-04
Levshft
1.50
ITERation
200 50
CIMX
200
SDAV
500
End of Input
&CASPT2
Title
PbO
MAXITER
25
FROZEN
19 16
Focktype
G1
Multistate
13 1 2 3 4 5 6 7 8 9 10 11 12 13
Imaginary Shift
0.1
End of Input
>> COPY $Project.JobMix $CurrDir/JobMix.11
&RASSCF
Title
PbO
Symmetry
2
Spin
3
CIROOT
14 14 1
nActEl
8 0 0
Inactive
23 18
Ras2
3 4
Lumorb
THRS
1.0e-8 1.0e-04 1.0e-04
Levshft
1.50
ITERation
200 50
CIMX
200
SDAV
500
End of Input
&CASPT2
Title
PbO
MAXITER
25
FROZEN
19 16
Focktype
G1
Multistate
14 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Imaginary Shift
0.1
>> COPY $Project.JobMix $CurrDir/JobMix.21
&RASSCF
Title
PbO
Symmetry
1
Spin
1
CIROOT
5 5 1
nActEl
8 0 0
Inactive
23 18
Ras2
3 4
Lumorb
THRS
1.0e-8 1.0e-04 1.0e-04
Levshft
1.50
ITERation
200 50
CIMX
200
SDAV
500
End of Input
&CASPT2
Title
PbO
MAXITER
25
FROZEN
19 16
Focktype
G1
Multistate
5 1 2 3 4 5
Imaginary Shift
0.1
End of Input
>> COPY $Project.JobMix $CurrDir/JobMix.10
&RASSCF
Title
PbO
Symmetry
2
Spin
1
CIROOT
4 4 1
nActEl
8 0 0
Inactive
23 18
Ras2
3 4
Lumorb
THRS
1.0e-8 1.0e-04 1.0e-04
Levshft
1.50
ITERation
200 50
CIMX
200
SDAV
500
End of Input
&CASPT2
Title
PbO
MAXITER
25
FROZEN
19 16
Focktype
G1
Multistate
4 1 2 3 4
Imaginary Shift
0.1
End of Input
>> COPY $Project.JobMix $CurrDir/JobMix.20
>> COPY $CurrDir/JobMix.12 JOB001
>> COPY $CurrDir/JobMix.11 JOB002
>> COPY $CurrDir/JobMix.21 JOB003
>> COPY $CurrDir/JobMix.10 JOB004
>> COPY $CurrDir/JobMix.22 JOB005
>> COPY $CurrDir/JobMix.20 JOB006
&RASSI
Nrof JobIphs
6 5 13 14 5 4 4
1 2 3 4 5
1 2 3 4 5 6 7 8 9 10 11 12 13
1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 2 3 4 5
1 2 3 4
1 2 3 4
Spin Orbit
Ejob
End of Input
In the above definitions of the JobMix files the labels correspond to
symmetry and spin. Thus :file:`JobMix.12` is for quintets (:math:`S=2`) in symmetry 1,
etc. The keyword :kword:`Ejob` ensures that the :program:`MS-CASPT2` energies
from the :file:`JobMix` files are used as the diagonal elements in the SO
Hamiltonian matrix. The output file of one such calculation is quite lengthy (6
:program:`CASSCF/MS-CASPT2` calculations and one :program:`RASSI`). Important
sections of the :program:`RASSI` output are the spin-free energies (look for the
word ``SPIN-FREE`` in the listing) and the SOC energies (found by looking for
``COMPLEX``). The complex SO wave functions are also given and can be used to
analyze the wave function. For linear molecules one wants to know the :math:`\Omega`
values of the different solutions. Here the computed transition moments can be
quite helpful (using the selection rules). It is important in a calculation of
many excited states, as the one above, to check for intruder state problems in
the :program:`CASPT2` results.
This example includes a large number of states, because the aim was to compute
full potential curves. If one is only interested in the properties near
equilibrium, one can safely reduce the number of states. For lighter atoms it is
often enough to include the spin-free states that are close in energy in the
calculation of the SOC. An intersystem crossing can usually be treated by
including only the two crossing states. The choice of basis states for the
:program:`RASSI` calculation depends on the strength of the SO interaction and
the energy separation between the states.
The above input is for one distance. The shell script loops over distances
according to: ::
Dist='50.0 10.0 8.00 7.00 6.00 5.50 5.00 4.40 4.20 4.00 3.90 3.80 3.75 3.70 3.65 3.60 3.55 3.50 3.40 3.30 3.10'
for R in $Dist
do
cat $CurrDir/template | sed -e "s/Dist/$R/" >$CurrDir/input
rm -rf $WorkDir
mkdir $WorkDir
cd $WorkDir
echo "R=$R" >>$CurrDir/energies
molcas $CurrDir/input >$CurrDir/out_$R
grep "Reference energy" $CurrDir/out_$R >>$CurrDir/energies
grep "Total energy" $CurrDir/out_$R >>$CurrDir/energies
grep "Reference weight" $CurrDir/out_$R >>$CurrDir/energies
done
Thus, the whole potential curves can be run as one job (provided that there are
no problems with intruder states, convergence, etc). Notice that the
:file:`JOBIPH` files for one distance are used as input (:file:`JOBOLD`) for the
next distance. The shell script collects all :program:`CASSCF` and
:program:`CASPT2` energies and reference weights in the file :file:`energies`.
We shall not give any detailed account of the results obtained in the
calculation of the properties of the :math:`\ce{PbO}` molecule. The reader is referred to the
original article for details :cite:`Roos:03h`. However it might be of interest to
know that the computed dissociation energy (:math:`D_0`) was 5.0 eV without SOC and
4.0 eV with (experiment is 3.83 eV). The properties at equilibrium are much less
affected by SOC: the bond distance is increased with 0.003 Å, the frequency is
decreased with 11 cm\ :math:`^{-1}`. The results have also been used to assign the 10
lowest excited levels.