NWchem vs Gaussian counterpoise BSSE calculation

[Dan Fowles]

Hi,

Im trying to recreate the energy results from a Gaussian energy calculation in NWchem. This calculation involves a dimer system, where I am trying to determine the intermolecular potential energy between the two monomers with the counterpoise method.

Gaussian - 

%chk=checkpoint.chk

%nprocshared=8

#p mp2/aug-cc-pvtz counterpoise=2

acetonitrile conformer a cp

0 1 0 1 0 1

 C(Fragment=1)      0.49790943    1.37034675    0.00000000

 C(Fragment=1)      0.00000000    0.00000000    0.00000000

 H(Fragment=1)     -1.08632000    0.00000000    0.00000000

 H(Fragment=1)      0.34492955   -0.52005940    0.88925802

 H(Fragment=1)      0.34470287   -0.52045046   -0.88911712

 N(Fragment=1)      0.89712487    2.46906783    0.00000000

C(Fragment=2) 6.458 0.00000000 0.00000000

C(Fragment=2) 5.0 0.00000000 0.00000000

H(Fragment=2) 4.62902 -1.02100776 0.00000000

H(Fragment=2) 4.61704 0.50139092 0.88440095

H(Fragment=2) 4.61704 0.50139092 -0.88440095

N(Fragment=2) 7.627 0.00000000 0.00000000

NWchem -

 start acetonitrile

 title acetonitrile

 geometry 

 C      0.49790943    1.37034675    0.00000000

 C      0.00000000    0.00000000    0.00000000

 H     -1.08632000    0.00000000    0.00000000

 H      0.34492955   -0.52005940    0.88925802

 H      0.34470287   -0.52045046   -0.88911712

 N      0.89712487    2.46906783    0.00000000

 C 6.458 0.00000000 0.00000000

 C 5.0 0.00000000 0.00000000

 H 4.62902 -1.02100776 0.00000000

 H 4.61704 0.50139092 0.88440095

 H 4.61704 0.50139092 -0.88440095

 N 7.627 0.00000000 0.00000000

 end  

 basis "ao basis" spherical  

  h   library   aug-cc-pvtz 

  c   library   aug-cc-pvtz

  n  library   aug-cc-pvtz  

 end

 bsse

  mon aceto1 1 2 3 4 5 6

  mon aceto2 7 8 9 10 11 12

 end

 mp2

 end

 task mp2 energy

Gaussian uncorrected energy = -264.996279530975

Gaussian corrected energy = -264.996799302691

NWchem uncorrected energy = -265.100208375963

NWchem corrected energy = -265.100208375962

There is a significant difference in BSSE error energy between both calculations.

I would appreciate if anyone could tell me what I have missed to achieve the same energy results in both Gaussian and NWchem, as I an hoping to use NWchem more fully in the future.

Thanks

[Edoardo Aprà]

Since your MP2 input section is empty, NWChem will freeze orbitals by default https://nwchemgit.github.io/MP2.html#freeze-freezing-orbitals

You will need to add the following lines in the MP2 input section to freeze orbitals.

mp2
  freeze atomic
end

[Dan Fowles]

Thanks for the suggestion, that fixed the difference in MP2 calculated energy between NWchem and Gaussian.

However, there is still a big difference in BSEE energy between the two calculations:

Gaussian uncorrected energy = -264.996279530975

Gaussian corrected energy = -264.996799302691

Gaussian BSSE energy = 0.000169759016

NWchem uncorrected energy = -264.996922005765

NWchem corrected energy = -264.996922005765

NWchem BSSE energy = -0.000000000001

Any ideas on how to correct for this?

Thanks

[Edoardo Aprà]

Your input file is missing the definition of the ghost (aka bq) functions  https://nwchemgit.github.io/TASK.html
basis "ao basis" spherical  

  H library aug-cc-pvtz 

  C library aug-cc-pvtz

  N library aug-cc-pvtz

  bqH  library  H aug-cc-pvtz 

  bqC  library  C aug-cc-pvtz

  bqN  library   N aug-cc-pvtz  

 end

[Dan Fowles]

Thanks for the suggestion, the two results are getting closer!

There is however still a difference in BSSE energy between the two. See below the updated nwchem input:

start acetonitrile

title acetonitrile

geometry 

C      0.49790943    1.37034675    0.00000000

C      0.00000000    0.00000000    0.00000000

H     -1.08632000    0.00000000    0.00000000

H      0.34492955   -0.52005940    0.88925802

H      0.34470287   -0.52045046   -0.88911712

N      0.89712487    2.46906783    0.00000000

C      6.458         0.00000000    0.00000000

C      5.0           0.00000000    0.00000000

H      4.62902      -1.02100776    0.00000000

H      4.61704       0.50139092    0.88440095

H      4.61704       0.50139092   -0.88440095

N      7.627         0.00000000    0.00000000

end  

basis "ao basis" spherical 

 H   library   aug-cc-pvtz 

 C   library   aug-cc-pvtz

 N   library   aug-cc-pvtz  

 bqH library H aug-cc-pvtz 

 bqC library C aug-cc-pvtz

 bqN library N aug-cc-pvtz 

end

bsse

 mon aceto1 1 2 3 4 5 6

 mon aceto2 7 8 9 10 11 12

end

mp2

 freeze atomic

end

scf

 direct

end

task mp2 energy

Comparing these NWchem results to Gaussian:

Gaussian uncorrected energy = -264.996279530975

Gaussian corrected energy = -264.996799302691

Gaussian BSSE energy = 0.000169759016

NWchem uncorrected energy = -264.996922005764

NWchem corrected energy = -264.996799390166

NWchem BSSE energy = 0.000122615598

There is still a difference in BSSE of 0.25 kcal/mol between both methods. Do you have any other suggestions for improving these results? Ive included an SCF command which has made the above improvements, but I am not sure where to go next.

Thanks in advance.