[CPMD-list] Help with molecular crystals

[Juerg Hutter]

Hi

there are often problems with the specification of
coordinates and cell parameters in CPMD when taken
from other programs or resources. The reason is
that different definitions and standards are used.

Some tricks to be used:

1) specify the 2 molecules with the coordinates such that
   the molecules are not "split".
   The atoms specified have not to be in the unit cell.

2) specify the cell using cell vectors, although
   orthorhombic should be easy enough to avoid errors

3) use POINT GROUP AUTO
   if the program doesn't find the correct point group
   your input is not correct or the coordinates not accurate
   enough

4) The add bonds feature is only for the empirical Hessian.
   It has no effect on the DFT calculation. For your system
   it is probably best to use HESSIAN UNIT.

5) Check the input by displaying the system with a graphics
   program. It most of the time easy to see when the
   box and coordinates are not maching.

6) ODIIS resets usually mean that either your trying to
   optimize a strange system (e.g. short bonds)
   or you have an otherwise difficult to optimize system.
   If all is correct, use the PCG MINIMZE obtimizer.
   This optimizer is slower than ODIIS but more robust.

greetings

Juerg

----------------------------------------------------------
Juerg Hutter                   Phone : ++41 1 635 4491
Physical Chemistry Institute   FAX   : ++41 1 635 6838
University of Zurich           E-mail: hutter at pci.unizh.ch
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
----------------------------------------------------------


On Mon, 11 Nov 2002, mkosmows wrote:

> Dear CPMD community:
>
> I'm trying to do a geometry optimization of a
> molecular crystal (with 2 molecules per cell).  I
> entered the literature data into rpluto, filled the
> unit cell, and got rid of the atoms outside the unit
> cell.  So I have one whole molecule and a bunch of
> lonely atoms due to the way the molecules pack into
> the cell.
>
> I chose the cell symmetry based on the geometry of
> the cell (all angles 90 degrees, so orthorhombic) and
> the point group based on the point group of the space
> group.
>
> The results were rather bizarre.  There were many
> assumed bonds that made no sense and the geometry was
> very much distorted from observed.
>
> The next thing I tried was to manually CHANGE BONDS.
> This got rid of wrong bonds, but failed to add the
> right bonds that were not assumed.  These bonds are
> the bonds that end up being spread across the ends of
> the cell due to packing.  (In one cell these would be
> from two different molecules).
>
> I then tried to use two whole molecules even though
> they had atoms outside of the cell but got "Failure
> to symmetrise coordinates," leading to early
> termination.  If anyone knows how to properly set up
> molecular crystals I would appreciate hearing about
> it.
>
> I'm using a 40ry cutoff with the Goedecker
> pseudopotentials downloaded from CPMD.org.
>
> I am also getting several ODIIS insufficient
> progress, reset messages throughout each run.  I'm
> still trying to understand what the ODIIS is and any
> assistance in how to properly set the ODIIS variable
> would be nice!
>
> Thank's,
>
> Mark Kosmowski
>
> Chemistry Department
> Syracuse University
> mkosmows at syr.edu
>
> _______________________________________________
> CPMD-list mailing list
> CPMD-list at cpmd.org
> http://www.cpmd.org/mailman/listinfo/cpmd-list
>