Evaluation of EDA-NOCV data

kbreitwieser · 2023-10-17 12:53:31

Hi,
I computed EDA-NOCV data via Orca and wanted to use Multiwfn to get the charge transfer between two fragments (see https://pubs.acs.org/doi/abs/10.1021/ac … et.7b00365 for reference).
I have a .molden file with all NOCV orbital pairs and cubes of the pairs and their differences (via the A²-B² function in Multiwfn).
What I am looking for is a way to partition the molecule via a Hirshfeld partioning and integrate over the difference of the orbital pairs in both fragments. I should then get information on the charge transfer between the fragments when they are combined. I thought this may be possible via the excited state analyisis (as in both cases effectively charge transfer is evaluated), but it does not seem to be possible (the calculation of interfragment charge transfer needs a TDDFT output file). Am I missing something?
Thanks,
Kevin

sobereva · 2023-10-21 17:43:19

In ETS-NOCV interface of Multiwfn, there is an option "14 Calculate composition of NOCV orbitals and pairs", which is able to derive fragment contribution to NOCV orbitals and pairs. See examples in Section 4.23.1 of Multiwfn manual for example.

kbreitwieser · 2023-10-27 15:16:57

Thanks for the reply. Indeed I can get the composition of NOCV orbitals and pairs as you said from Orca's wavefunctions.

I would however have a few additional questions:

1. In the decomposition I get the contributions of each atom to the two orbitals as well as "pair". It is clear that for the orbitals it is just the percentage to the total orbital and more or less adds up to 100%. However, I wonder how the pair contribution is determined. I see that it adds up to 0, so it probably is related to the total electron flow (which obviously has to end up at 0). However, I don't see where this comes from. I naively assumed that for the pair you more or less just take the difference of the contributions to both orbitals of the pair? That does not seem to be the case, though. Can I get the share of electrons moved to/from an atom by multiplying the pair% with the eigenvalue or would I have to do something else?

2. I noted that the sum over all atoms/orbitals... does not perfectly add up even at much lower threshholds then default (I set it to 0.005%), e.g., when decomposing for atoms, I get a total of 99.99% for one orbital and 99.92% for the other orbital. I then get 0.03% for the pair. Is there a way to get the percentages with more digits to prevent rounding errors and similar things? How good is the accuracy of the method in the first place?

Thanks for all,
Kevin

sobereva · 2023-10-28 02:11:50

1 Please search "The composition of NOCV pair is calculated as" in Multiwfn manual, you will find the formula. It noted that eigenvalue must be taken into account.

As an example, composition of NOCV pair 1 of CO-BH3 system:

Contribution of each atom to NOCV pair/orbitals:
  Atom         Orb.    1   Orb.   51   Pair    1
    1(C ):      23.99 %     65.82 %    -23.64 %
    2(O ):       0.28 %      2.94 %     -1.50 %
    3(B ):      69.40 %     30.30 %     22.10 %
    4(H ):       2.12 %      0.32 %      1.02 %
    5(H ):       2.11 %      0.31 %      1.01 %
    6(H ):       2.11 %      0.31 %      1.01 %

This output indicates that owing to the interaction characterized by NOCV pair 1, C loses 0.236 electron.

2 You need to modify source code and recompile. Please search "Contribution of each basis function to NOCV pair/orbitals" in ETS_NOCV.f90

Best regards,

Tian

Multiwfn forum

#1 2023-10-17 12:53:31

Evaluation of EDA-NOCV data

#2 2023-10-21 17:43:19

Re: Evaluation of EDA-NOCV data

#3 2023-10-27 15:16:57

Re: Evaluation of EDA-NOCV data

#4 2023-10-28 02:11:50

Re: Evaluation of EDA-NOCV data

Board footer