CP2K energy very different from Gaussian03

[Megha Anand]

Dear All,

I am pretty new to CP2K and still learning the basics of the program. I want to get the energy of the energy of the reaction: Fe(g) + S(g) -> FeS(g). To begin with, I simply did single point calculation of iron atom using Gaussian03 and CP2K. Browsing through the previous posts, I figured out that the ECP basis sets used by most quantum chemists cannot be used in CP2K. Even if I use different level of theory and basis sets across the two programs, I would anticipate roughly similar energies if not exactly same. I used basis set B3LYP/SDD in Gaussian03 and SZV-MOLOPT-SR-GTH in CP2K. The energies I get from Gaussian is: -123.85372 while in CP2K: -19.96605887771384. Even if the two programs results in two spin states I would not expect such large different. 

I do not know what I am doing wrong. Any help will be greatly appreciated. I have included the input below. Apart from these, I have few more general questions:

(a) Is there any recommended combination of pseudopotential and basis set for complex compounds containing metal iron. There are also: carbon, oxygen, sulfur, and nitrogen in the system.

(b) Should I use GPW or do all electron calculation using GAPW. I would have used ECP basis set, if I had to use Gaussian03. But I am not sure what is right way proceed with CP2K.

(c) Any suggestion for the functional available in CP2K for studying iron complexes of high spin multiplicity. 

(d) I saw ECP_POTENTIALS file on CP2K Github, is it possible to use ECPs like SDD, LANL2DZ, etc. in CP2K.

Here is my CP2K input:

&GLOBAL

  PROJECT Fe

  RUN_TYPE ENERGY

  PRINT_LEVEL MEDIUM

  WALLTIME 86400

&END GLOBAL

&FORCE_EVAL

  METHOD Quickstep

  &DFT

    BASIS_SET_FILE_NAME /share/apps/cp2k/cp2k/tests/QS/BASIS_MOLOPT

    POTENTIAL_FILE_NAME /share/apps/cp2k/cp2k/tests/QS/GTH_POTENTIALS

    CHARGE 0

    MULTIPLICITY 5

    UKS T

    &MGRID

      CUTOFF 400

      REL_CUTOFF 60

    &END MGRID

    &QS

      METHOD GPW

      EPS_DEFAULT 1.0E-12

    &END QS

    &SCF

      MAX_SCF 500

      &OUTER_SCF

        EPS_SCF 1.0E-6

        MAX_SCF 60

      &END OUTER_SCF

    &END SCF

    &POISSON

      PERIODIC NONE

      PSOLVER WAVELET

    &END POISSON

    &XC

      &XC_FUNCTIONAL PADE

      &END XC_FUNCTIONAL

    &END XC

  &END DFT

  &SUBSYS

    &CELL

      ABC 7.00 7.00 7.00

      PERIODIC NONE

    &END CELL

    &COORD

      Fe    0.0000    0.0000    0.0000

    &END COORD

    &KIND Fe

      BASIS_SET SZV-MOLOPT-SR-GTH

      POTENTIAL GTH-PADE-q8

    &END KIND

  &END SUBSYS

&END FORCE_EVAL

Thanks,

Megha

[David T]

Hi Megha

someone more expert than me in CP2K will answer at some point.
In the meantime I would say that you should at least use the same level of theory (for instance B3LYP in both or pade in both, although I am not sure that is available in gaussian).

Best
Davide

[Matt W]

Hi,

you certainly should get a similar number for the observable reaction energy Fe + S -> FeS. 

But the energies of the individual species will be different. Potentially totally different if using different pseudopotentials. So just looking at the Fe atom energy is not a good idea.

On an actual CP2K note your basis set is too small - change SZV to DZVP for something OK. 

As David T says, best to compare like with like where possible, PADE functional in CP2K is LDA, so probably like SVWN keywords for exchange correlation in Gaussian for comparison.

Matt

[Megha Anand]

Thank you Matt and Davide! I will try your suggestions and hope to get more from the experts.

Best regards,

Megha