Need advice on bulk metal system calculations

[Stephen]

Hi All, 

I was wondering if anyone has tips for computing bulk metallic system in CP2K (specifically for bulk Cu and bulk Ni system). I need the electronic energy for bulk Cu and bulk Ni for some modeling on NP thermodynamic stability. 

I've optimized the unit cell parameters for both Cu and Ni systems by varying the unit cell dimensions found on Materials Project with the same level of theory as my CP2K on metal-organic frameworks. When I try to compute different metallic properties though such as atomization and bulk Cu + O2 (g) --> bulk CuO my energies are way off.. 

From Web Elements for Ni, the atomization energy was determined:

bulk Cu (s) --> Cu(0) (g) w/ deltaH=338 kJ/mol

Now using DFT, the follow atomisation energy was determined to be:

bulk Cu (s) --> Cu(0) (g) w/ deltaE=1658.2 kJ/mol

Either my bulk system is too stable, or my atomized Cu is too unstable. Here is my DFT section for the bulk Cu calculations. 

   &DFT

     BASIS_SET_FILE_NAME BASIS_file

     POTENTIAL_FILE_NAME POTENTIALS_file

     UKS  T 

     MULTIPLICITY  2 

     CHARGE  0

     &SCF

       MAX_SCF  1000

       EPS_SCF     9.9999999999999995E-07

       SCF_GUESS  ATOMIC

       &OT  T

         MINIMIZER  CG

         PRECONDITIONER  FULL_ALL

         ENERGY_GAP     1.0000000000000000E-03

       &END OT

       &OUTER_SCF  T

         EPS_SCF     9.9999999999999995E-07

         MAX_SCF  50

       &END OUTER_SCF

     &END SCF

     &QS

       EPS_DEFAULT     1.0000000000000000E-10

       METHOD  GPW

     &END QS

     &MGRID

       NGRIDS  5

       CUTOFF     3.6000000000000000E+02

       REL_CUTOFF     8.0000000000000000E+01

     &END MGRID

     &XC

       DENSITY_CUTOFF     1.0000000000000000E-10

       GRADIENT_CUTOFF     1.0000000000000000E-10

       TAU_CUTOFF     1.0000000000000000E-10

       &XC_FUNCTIONAL  NO_SHORTCUT

         &PBE  T

         &END PBE

       &END XC_FUNCTIONAL

       &VDW_POTENTIAL

         POTENTIAL_TYPE  PAIR_POTENTIAL

         &PAIR_POTENTIAL

           TYPE  DFTD3(BJ)

           PARAMETER_FILE_NAME dftd3.dat

           REFERENCE_FUNCTIONAL PBE

           CALCULATE_C9_TERM  F

         &END PAIR_POTENTIAL

       &END VDW_POTENTIAL

     &END XC

   &END DFT

Sample .inp and .out files are attached. 

Any feedback would be much appreciated! Thanks for taking a look at my issues. 

With much appreciation, 

Stephen 

[Krack Matthias (PSI)]

Hi Stephen

 

For the description of metallic bulk systems, you will need Brillouin zone sampling with an appropriate k point mesh. It seems this is missing in your input.

 

HTH

 

Matthias

--
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/a3f838aa-94cb-4998-b501-09656b8efd2en%40googlegroups.com.

[Stephen Vicchio]

Hi Matthias, 

Thanks for your input! Using the CP2K example 'Getting the band structure of graphene' (https://www.cp2k.org/exercises:2016_uzh_cmest:band_structure_calculation), I was able to add the KPOINT section into my current input shown above. I've been altering the parameters and playing around with the input file, but I'm having issues troubleshooting why my new input isn't working properly.  The CP2K example was able to run, but my job isn't which makes me think there is an issue with some of the other flags I've set in my .inp file for bulk Cu. 

Below you can find the new .inp file that I've been using. Thanks for taking a look! 

With much appreciation, 

Stephen 

&GLOBAL
   PRINT_LEVEL  MEDIUM
   PROJECT_NAME bulk-Cu-CELL_DIMA
   RUN_TYPE  ENERGY_FORCE
   WALLTIME  23:40:00
&END GLOBAL
 &FORCE_EVAL
   METHOD  QS
   STRESS_TENSOR  ANALYTICAL

   &DFT
     BASIS_SET_FILE_NAME BASIS_file
     POTENTIAL_FILE_NAME POTENTIALS_file
     UKS  T
     MULTIPLICITY  2
     CHARGE  0
     &SCF
       MAX_SCF  1000
       EPS_SCF     9.9999999999999995E-07
       SCF_GUESS  ATOMIC
       &OT  T
         MINIMIZER  CG
         PRECONDITIONER  FULL_ALL
         ENERGY_GAP     1.0000000000000000E-03
       &END OT
       &OUTER_SCF  T
         EPS_SCF     9.9999999999999995E-07
         MAX_SCF  50
       &END OUTER_SCF
     &END SCF
     &QS
       EPS_DEFAULT     1.0000000000000000E-10
       METHOD  GPW

       EXTRAPOLATION USE_GUESS

     &END QS
     &MGRID
       NGRIDS  5
       CUTOFF     3.6000000000000000E+02
       REL_CUTOFF     8.0000000000000000E+01
     &END MGRID
     &XC
       DENSITY_CUTOFF     1.0000000000000000E-10
       GRADIENT_CUTOFF     1.0000000000000000E-10
       TAU_CUTOFF     1.0000000000000000E-10
       &XC_FUNCTIONAL  NO_SHORTCUT
         &PBE  T
         &END PBE
       &END XC_FUNCTIONAL
       &VDW_POTENTIAL
         POTENTIAL_TYPE  PAIR_POTENTIAL
         &PAIR_POTENTIAL
           TYPE  DFTD3(BJ)
           PARAMETER_FILE_NAME dftd3.dat
           REFERENCE_FUNCTIONAL PBE
           CALCULATE_C9_TERM  F
         &END PAIR_POTENTIAL
       &END VDW_POTENTIAL
     &END XC

     &KPOINTS
       SCHEME MONKHORST-PACK 1 1 1
       SYMMETRY OFF
       WAVEFUNCTIONS REAL
       FULL_GRID .TRUE.
       PARALLEL_GROUP_SIZE  0    
     &END KPOINTS
   &END DFT
   &SUBSYS
     &CELL
       ABC  2.24 2.24 2.24
       ALPHA_BETA_GAMMA 60 60 60
       MULTIPLE_UNIT_CELL  1 1 1
     &END CELL
     &TOPOLOGY
       COORD_FILE_FORMAT CIF
       COORD_FILE_NAME bulk-Cu.cif
       NUMBER_OF_ATOMS  1
       MULTIPLE_UNIT_CELL  1 1 1
     &END TOPOLOGY
     &KIND Cu
       BASIS_SET DZVP-MOLOPT-SR-GTH-q11
       POTENTIAL GTH-PBE-q11
     &END KIND
   &END SUBSYS
   &PRINT
     &FORCES  ON
     &END FORCES
   &END PRINT
 &END FORCE_EVAL

You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/g7wXQbqxfQs/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/734dfa8497b94d2fb178deb4655bfee5%40psi.ch.

--

Stephen P. Vicchio

PhD Candidate | Getman Research Group

Dept. of Education GAANN Fellow

Department of Chemical & Biomolecular Engineering

105 Earle Hall | Clemson, SC 29634-0909

svi...@g.clemson.edu

[Marcella Iannuzzi]

Dear Stephen

Are you sure about 

 ALPHA_BETA_GAMMA 60 60 60 

Best regards

Marcella

[Stephen Vicchio]

From materials project, the .cif file I'm using for bulk Cu has unit cell dimensions of 2.24 2.24 2.24 and 60 60 60. Do you think that might explain what's happening? The error message I'm receiving isn't located in the out file from CP2K, but the out file from the cluster:

Error termination. Backtrace:
At line 470 of file /tmp/spack/spack-stage/spack-stage-cp2k-7.1-h7ky2rvmv3fvzjrtpbqv2rr3fie3l6zy/spack-src/src/cell_methods.F
Fortran runtime error: Bad real number in item 1 of list input

The example scripts run fines, but it my current .inp that is giving me issues. 

Stephen 

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/b3ea4a80-54af-4ac4-b5e7-a4922f4762dan%40googlegroups.com.

[fabia...@gmail.com]

Dear Stephan,

With k-points enabled you have to use diagonalization instead of OT (as is done in the example); also, do not use WAVEFUNCTIONS REAL.

Best,

Fabian

[Krack Matthias (PSI)]

Hi Stephen

 

For bulk Cu you will need to apply

-          k point sampling (e.g. MP 8 8 8)

-          diagonalization (with Broyden mixing)

-          smearing (e.g. T(elec) = 2000 K)

 

Moreover, you should use the conventional (cubic) unit cell with 4 Cu atoms instead of the primitive one with one Cu atom. CP2K is slower for non-orthorhombic cells.

Plane wave (PW) codes are fast for small unit cells as they scale with the size of the cell which has to be filled with PWs. That’s not the case for CP2K using atomic basis functions which requires to calculate the interactions with the image Cu atoms in neighboring cells. So it can make sense to use multiple unit cells, e.g. a 2x2x2 supercell, and to reduce the k point mesh correspondingly (e.g. from 8x8x8 to 4x4x4).

 

Below you will find a CP2K input for bulk Cu in which I have considered the suggestions above. No guarantee, of course, that it will work properly.

 

Matthias

 

&GLOBAL

PRINT_LEVEL low

PROJECT_NAME Cu

RUN_TYPE cell_opt

&END GLOBAL

 

&MOTION

&CELL_OPT

  EXTERNAL_PRESSURE [bar] 1.0

  MAX_DR 0.001

  MAX_FORCE 0.0001

  MAX_ITER 400

  OPTIMIZER BFGS

  PRESSURE_TOLERANCE [bar] 10.0

  RMS_DR 0.0003

  RMS_FORCE 0.00003

  TYPE direct_cell_opt

  &BFGS

   TRUST_RADIUS 0.1

   USE_MODEL_HESSIAN off

   USE_RAT_FUN_OPT on

  &END BFGS

&END CELL_OPT

&END MOTION

 

&FORCE_EVAL

METHOD QS

STRESS_TENSOR analytical

&DFT

  BASIS_SET_FILE_NAME BASIS_MOLOPT

  POTENTIAL_FILE_NAME GTH_POTENTIALS

  &KPOINTS

   SCHEME MONKHORST-PACK 2 2 2

   FULL_GRID yes

   SYMMETRY yes

   VERBOSE yes

   PARALLEL_GROUP_SIZE -1

  &END KPOINTS

  &MGRID

   NGRIDS 5

   CUTOFF 400.0

   REL_CUTOFF 60.0

  &END MGRID

  &QS

   EPS_DEFAULT 1.0E-12

   EXTRAPOLATION use_prev_p

  &END QS

  &SCF

   ADDED_MOS 60

   EPS_SCF 1.0E-8

   MAX_SCF 300

   SCF_GUESS restart

   &DIAGONALIZATION yes

    ALGORITHM STANDARD

   &END DIAGONALIZATION

   &MIXING yes

    ALPHA 0.4

    BETA 1.0

    METHOD broyden_mixing

    NBROYDEN 8

   &END MIXING

   &SMEAR on

    METHOD FERMI_DIRAC

    ELECTRONIC_TEMPERATURE [K] 2000.0

   &END SMEAR

  &END SCF

  &XC

   &XC_FUNCTIONAL PBE

   &END XC_FUNCTIONAL

   &VDW_POTENTIAL

    POTENTIAL_TYPE pair_potential

    &PAIR_POTENTIAL

     TYPE DFTD3(BJ)

     PARAMETER_FILE_NAME dftd3.dat

     REFERENCE_FUNCTIONAL PBE

    &END PAIR_POTENTIAL

   &END VDW_POTENTIAL

  &END XC

&END DFT

&SUBSYS

  &CELL

   ABC 3.62 3.62 3.62

   MULTIPLE_UNIT_CELL 2 2 2

  &END CELL

  &COORD

   SCALED

   Cu    0    0    0

   Cu    0  1/2  1/2

   Cu  1/2    0  1/2

   Cu  1/2  1/2    0

  &END COORD

  &KIND Cu

   BASIS_SET DZVP-MOLOPT-SR-GTH-q11

   POTENTIAL GTH-PBE-q11

  &END KIND

  &TOPOLOGY

   MULTIPLE_UNIT_CELL 2 2 2

  &END TOPOLOGY

&END SUBSYS

&END FORCE_EVAL

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAHf%3DnNaw5e%2BioxzubJD7RNmia8OjAoekUOZYx_1bi-kvh4tSuQ%40mail.gmail.com.

[Brendan Smith]

Hi Matthias,

I have a separate question regarding the 2x2x2 super cell. What if I make a 2x2x2 super cell manually and use the super cell atomic coordinates and cell parameters in the input file instead of the multiple unit cell 2x2x2 option. Would it be the same as using the single conventional unit cell but with the multiple unit cell 2x2x2 option turned off?

Kind regards,

Brendan Smith

SUNY Buffalo

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/75e08d6442354e1d9a1393a676b0aac2%40psi.ch.

[Stephen Vicchio]

Hi Matthias, 

Thank you so much for the information! From your suggestions, I was able to set up the input file for my bulk Cu metal systems. For other metal systems, do you recommend a similar set of parameters? Specifically, the mixing, smearing, and kpoints? My previous CP2K calculations were on metal-organic frameworks so my computational parameters were set up for these porous structures. 

With much appreciation, 

Stephen 

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/75e08d6442354e1d9a1393a676b0aac2%40psi.ch.

[Krack Matthias (PSI)]

Hi Brendan

 

That should be the same, since the MULTIPLE_UNIT_CELL keyword just replicates the unit cell internally and the atomic coordinates printed by CP2K are for the supercell. This keyword facilitates the input of supercells. CP2K can also be used in this way to create supercells, though this might be easier with other tools.

 

Best regards

 

Matthias

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/4A2812B9-3B08-4EB5-AF76-BACB63671304%40buffalo.edu.

[Krack Matthias (PSI)]

Hi Stephen

 

Such a set up should work for bulk metal systems in general. Just check the convergence of the results with respect to the k point mesh size. You can reduce the mixing parameter ALPHA if convergence problems occur.

 

Best regards

 

Matthias

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAHf%3DnNbo-MzBPoYkzS3MJimgMbQDjnawmkz3%3DUJZRX1zbiJB0w%40mail.gmail.com.