Rhombohedral structure optimization (CELL_OPT)

[Tomoki Ogata]

Hi there everyone,

First time posting here, thank you in advance! 

I am trying to do a structural optimization of FAU zeolite, which has a rhombohedral unit cell (a = b = c = 17.9 A, alpha = beta = gamma = 60 degrees). 

I tried running the input file attached, but unfortunately, the crystallinity of the structure collapsed. 

I believe that the reason is related to the line in my input file "SYMMETRY RHOMBOHEDRAL", since the structure is preserved if this line is excluded. However, without SYMMETRY RHOMBOHEDRAL, the a, b and c parameters of the cell deviate from 17.9 A. I looked through the CP2K manual, but was unfortunately unable to find any information which might help solve the problem.

A part of my input file which is attached:

&CELL

    &CELL_REF

      A 12.6592017 12.6592017 0.0000000

      B 12.6592017 0.0000000 12.6592017

      C 0.0000000 12.6592017 12.6592017

      MULTIPLE_UNIT_CELL 1 1 1

      ALPHA_BETA_GAMMA 60 60 60

      SYMMETRY RHOMBOHEDRAL

    &END CELL_REF

    A 12.6592017 12.6592017 0.0000000

    B 12.6592017 0.0000000 12.6592017

    C 0.0000000 12.6592017 12.6592017

    MULTIPLE_UNIT_CELL 1 1 1

    ALPHA_BETA_GAMMA 60 60 60

    SYMMETRY RHOMBOHEDRAL

&END CELL

Would anyone happen to know of a way to do a structural optimisation for a rhombohedral system while preserving the a, b and c parameters and without the structure falling apart?

Thank you in advance,

Tomoki

[Matt Watkins]

Hello,

I think your structure / periodic boundaries are incorrect. When I run your input I see

"

 *** WARNING in particle_methods.F:684 :: The distance between the atoms  ***
 *** 78 and 89 is only 0.224 angstrom and thus smaller than the threshold ***
 *** of 0.500 angstrom

"
this is going to lead to massive force and explosions.

Making your structure into a cubic conventional cell might make life easier.

Matt

[Matt Watkins]

Hello, odd you don't see the same message.

I am using the dev version (10.0)

You could also ask cp2k to dump coordinates + cell to pdb or similar format and see if they are what you expect.

Matt

Dear Matthew,

Thank you very much for your prompt reply.

Unusually, I did not observe any such warnings, in fact there were none in the output file...

The version of CP2K I am using is 8.1.0, could that issue have perhaps been due to a version difference?

In the meantime, I will give it a try with a cubic conventional cell and update you.

Kind regards,

Tomoki

[Krack Matthias (PSI)]

Hi

 

the interatomic distance check for systems with less than 2000 atoms is a default task only since CP2K version 9.1. For earlier versions, you have to request that check explicitly using this keyword.

 

Best

 

Matthias

 

I believe that the reason is related to the line in my input file "SYMMETRY RHOMBOHEDRAL", since the structure is preserved if this line is excluded. However, without SYMMETRY RHOMBOHEDRAL, the a, b and c parameters of the cell deviate from 17.9 A. I looked through the CP2K manual, but was unfortunately unable to find any information which might help solve the problem.

 

A part of my input file which is attached:

&CELL

    &CELL_REF

      A 12.6592017 12.6592017 0.0000000

      B 12.6592017 0.0000000 12.6592017

      C 0.0000000 12.6592017 12.6592017

      MULTIPLE_UNIT_CELL 1 1 1

      ALPHA_BETA_GAMMA 60 60 60

      SYMMETRY RHOMBOHEDRAL

    &END CELL_REF

    A 12.6592017 12.6592017 0.0000000

    B 12.6592017 0.0000000 12.6592017

    C 0.0000000 12.6592017 12.6592017

    MULTIPLE_UNIT_CELL 1 1 1

    ALPHA_BETA_GAMMA 60 60 60

    SYMMETRY RHOMBOHEDRAL

&END CELL

 

Would anyone happen to know of a way to do a structural optimisation for a rhombohedral system while preserving the a, b and c parameters and without the structure falling apart?

 

Thank you in advance,

Tomoki

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/42a6704b-cdfd-4ebf-83e0-ec1aba5e6a28n%40googlegroups.com.

[Matt Watkins]

Mystery solved. 

Thanks, Matthias.