Sample directory: adf/e_QMMM_Surface/
This is an example of a Ziegler-Natta type catalytic system: a TiCl complex embedded in a MgCl surface with two organic substrates also attached to the surface. To make the computation faster, the QMMM approach is applied. The QM part includes only the active site and a piece of the MgCl surface.
The computation is formally a geometry optimization, but to keep the sample doable in a reasonable time the sample performs only one geometry update step. In the optimization, all of the MgCl surface atoms are frozen.
The standard force field has been modified to accommodate this calculation. The modified force field file is part of the sample run script. In this modified file, bonds are defined between Mg-Cl atoms in the MM connection table. This results in some torsions where the atoms are collinear. To rectify this problem, the torsional potentials for these atoms are set to potential type '0' (no potential).
There are no capping atoms mediating the bonds between the QM and MM regions because the boundary goes through the MgCl surface, which is ionically bound.
cat << eor > champ_de_force.ff YBYL/TRIPOS FORCE FIELD FILE FOR ADF QM/MM MODIFIED WITH UFF1.01 FOR Si Mg Ti Cl L. Petitjean 15.11.1999 *************************************************************************
(Most of the contents of the modified force field file is omitted here. You quickly get the difference with the standard sybyl force field file in the ADF database by running a UNIX diff on the two files.
==================================== eor $ADFBIN/adf << eor Title ADF-QMMM in a surface study NoPrint SFO, Frag, Functions MaxMemoryUsage 20 ! keywords for calculation methods and optimization XC LDA VWN GGA BLYP End Geometry Optim Cartesian Selected Iterations 1 HessUpd BFGS Converge e=1e-4 grad=1e-3 rad=1e-2 Step rad=0.15 END
The 'Iterations 1' subkey specification in the Geometry block specifies that only one step in the optimization is carried out.
Integration 3.0 3.0
SCF
Iterations 250
Converge 1E-6 1E-6
Mixing 0.2
DIIS N=10 OK=0.5 cyc=5 CX=5.0 BFAC=0
End
! keywords for molecule specification
Charge 0 0
Atoms Cartesian
1 Mg x1 y1 z1
(all other atoms in the Atoms block omitted here)
End GeoVar x1=.00000 F y1=.00000 F z1=.00000 F x2=.00000 F y2=1.72129 F z2=1.82068 F x3=.00000 F y3=.00000 F z3=-3.64100 F x4=.00000 F y4=-1.72130 F z4=-1.82068 F x5=.00000 F y5=1.72130 F z5=-1.82032 F x6=.00000 F y6=1.72130 F z6=-5.46132 F x7=2.53903 y7=.03004 z7=-3.50645 x8=2.50628 y8=-.07048 z8=-.10022 x9=2.63009 y9=3.50093 z9=-3.02634 ...
Many of the coordinates have a 'F' after their initial value specification under Geovar, indicating that these coordinates will be kept frozen during optimization.
The remaining initial value specifications are omitted here.
END
QMMM
OPTIMIZE
MAX_STEPS 3000
MAX_GRADIENT 0.01
METHOD BFGS
PRINT_CYCLES 100
SUBEND
FORCE_FIELD_FILE champ_de_force.ff
The local file 'champ_de_force.ff' is used as force field file. Of course, this is the file we've just set up in the run script.
OUTPUT_LEVEL=1
WARNING_LEVEL=1
ELSTAT_COUPLING_MODEL=1
MM_CONNECTION_TABLE
1 Mg QM 2 4 5 8 58 60
...
Contents of the MM_Connection_Table block is omitted.
SUBEND
CHARGES
1 .957
2 -.608
3 1.017
4 -.411
5 -.561
...
Initial charges are specified for (all) the atoms. Whether or not the charges on the QM (and LI) atoms are used depends on the type of electrostatic coupling between the QM and MM system. See the QMMM manual for details.
SUBEND END Fragments Ti t21.Ti Cl t21.Cl Mg t21.Mg C t21.C H t21.H End End Input eor
