Cerium ECP with 46 core electrons

Started by gpetrova, September 21, 2010, 04:26:05 PM

Previous topic - Next topic

gpetrova

Hi everybody! :)

I have a problem with running DFT calculations of cerium-containing system with ECP with 46 core electrons for Ce atoms. Since no ehtdata are included in the TURBOMOLE/basen/ce file for the corresponding basis set (ce ecp-46-mwb), the eht run for the initial guess could not be performed. Is there a basis set file for cerium containing the necessary ehtdata?

Regards,

Galina

Arnim

Hello Galina,

there is no ehtdata available, because there is no optimized basis set for this ECP.

As a workaround, choose * in define in
OCCUPATION NUMBER & MOLECULAR ORBITAL DEFINITION MENU
and run 'hcore' after define. Then you start with a Hamilton core guess.

Maybe run a stability analysis afterwards to checks, if you have converged
to the electronic minimum. And fermi smearing often helps to find the correct
occupation during the SCF cycle.

Best
Arnim

gpetrova

Dear Arnim,

Thank you for your prompt answer and the explanation.
Is it possible to generate the ehtdata with "$make ehtao" option and just add it to the available basis set file for Ce?

The problem with running hcore is that the system is with open shell (4 unpaired electrons at least) and since I am running just a single point MP2 calculation, I prefer to predefine the spin state of the system.

Regards,

Galina

Arnim

Hi Galina,

that is possible, but you have to spend some time in define menus and
check afterwards carefully on the atom or on small model systems, if it works correctly.

Put an Ce atom in coord, use i symmetry, then choose mkeht in define.
Choose y at 'DO YOU WANT THE DEFAULT OCCUPATION ASSIGNMENT FOR ATOMS ?' and
move carefully through the menus. If you don't know the Roothaan parameter try 'val'.

A data group $ehtao will be written to control. This can be added in $TURBODIR/basen/ce.
The data group
$ehtdef    definitions=  1
has to be adjusted here also to
$ehtdef    definitions=  2
with the entries from control.

This is in principle the way, but I don't know what problem might occur on the way.
Hope it helps,
Arnim


gpetrova

Hi Arnim,

I will try this option.

Regards,

Galina