Methodology for calculating J-couplings

Clicked A Few Times

7:42:08 AM PDT - Thu, Apr 21st 2016

Hi, I just wanted to confirm whether the implementation of indirect spin-spin couplings (J-couplings) is based on CPDFT in nwchem. If so, the methodology should be the same as in Gaussian and I should get identical results. However, when I calculate the J-coupling constant for H2, the FC contribution comes out slightly different in Gaussian compared to nwchem. I'm using the following input:

start h2_nmr
geometry units angstroms

 H       3.18205 0.0 0.0

 H       3.78205 0.0 0.0

end
basis

 H library cc-pvdz

end
dft

 xc pw91lda

 grid lebedev

 tolerances tight

 odft

end
property

 spinspin 1 1 2

end
task dft property

Even though the implementations differ a little (nwchem has advanced support for relativistic calculations), I should still get the same results for light elements. Could I be missing something essential in the input?

Forum Regular

7:30:53 AM PDT - Tue, May 3rd 2016
Hi Raul, Our implementation appears to be working correctly. Can you try comparing with pure Hartree-Fock ? Thanks. Best, -Niri niri.govind@pnnl.gov

Just Got Here

6:55:00 AM PDT - Fri, Aug 3rd 2018
spin spin coupling
Dear NWChem users I have managed to compute spin spin couplings (and in good agreement with the experiment!) when the atom pair under investigation is H-H for several standard neutral organic molecules. Just an open shell calculation and using the spinspin keyword in the property section as described in Documentation : SPINSPIN [<integer> number_of_pairs <integer> pair_list] However, when i tried a C-H coupling instead the calculation always stops during the iterative solution of linear equations and with no error.... I know that C12, the most abundant, is not NMR active... so i define isotope C13 by specifying the keyword mass, followed by 13 after the line of the Cartesian coordinate of this Carbon atom under study. Same outcome for C12 and C13. The calculation always stops during the itterative solution of linear equations and with no error.... Has anyone been able to run successfully a C-H coupling prediction of any standard organic neutral organic molecule? If so, please share the input how to do so. Thank you very much in advance Inaki

Forum Vet

9:37:33 AM PDT - Fri, Aug 3rd 2018
Dear Inaki Could you please provide an input file reproducing the bug you are reporting? Thanks

Just Got Here

1:39:45 AM PDT - Mon, Aug 6th 2018

Thank you very much for answering me.

Firstly, i am using version 6.5. I have requested to our system admin to update the software but still pending...

It may be fixed in a newer version. I cannot test that, sorry.

After lots of trial and error i noticed that i can run the coupling calculation but only 1 by 1. In other words, i can not compute 2 C-H couplings in the same calculation.

Please find below the input of the calculation that failed:

scratch_dir /scratch

geometry units angstroms noautosym noautoz

 C        0.04907        0.82213       -0.65775 mass 13.003355

 C       -1.15981        1.28999       -1.16070

 H       -1.20486        2.13123       -1.84724

 N       -2.30970        0.69954       -0.80166

 C       -2.24978       -0.34780        0.04068

 H       -3.19802       -0.80119        0.31094

 N       -1.15919       -0.90083        0.59434

 C       -0.04509       -0.28180        0.22092 mass 13.003355

 N        1.37198        1.20191       -0.81802

 N        1.24761       -0.57303        0.58134

 C        1.66725       -1.61610        1.49685

 H        1.12804       -2.53639        1.26672

 H        1.44762       -1.32294        2.52807

 H        2.73910       -1.77789        1.37965

 C        2.03623        0.35215       -0.07465

 H        3.11035        0.33755        0.05285

end

basis

* library 6-31G**

end

scf

uhf

end

property

spinspin 2 16 1 16 8

end
task scf property

Please let me know if you get the same unfinished run without error.

If you compute only 1 C-H coupling, (either spinspin 1 16 1 or spinspin 1 16 8) the calculation is fine

Many thanks in advance

Inaki

Forum >> NWChem's corner >> NWChem functionality