CC C6 (vdW) coefficients

[LucasVisscher]

Dear Dalton community,

I'd like to use the CC response code to get van-der-Waals dispersion coefficients for atoms. I can see I can do this in the **RESPONSE section, but this only allows HF, DFT and CI. For consistency with other calculations, I would like to do this with the separate CC response module. Can this one handle imaginary frequencies (which is needed for this) ? And if, so which would be the keywords to look up ?

thanks for your help,

Luuk

[sonco]

Dear Luuk,

we already talked about it directly, but I am posting this for the benefit of the Dalton community.

One way would be to use the same strategy as used in the analytic implementation of the dispersion coefficients of the
polarisability (J. Chem. Phys., Vol. 107, 1997, 10592).
The Cauchy moments are obtained adding the

.DISPCF
10 (or whatever order you want)

to the *CCLR section relative to a polarisability calculation.

Another way could be to use the Lanczos implementation of the complex polarisability presented in
dx.doi.org/10.1021/ct200919e | J. Chem. Theory Comput. 2012, 8, 1616−1628.

All the best
Sonia

[yixin]

Dear Sonia,

Thank you for posting this. I am wondering if Dalton can calculate imaginary part of polarizability with respect to photon energy.

I want to plot this spectrum in order to compare with other methods. It seems that Cauchy moments can only provide correct imaginary spectrum below the first pole.
In this paper "Lanczos implementation of the complex polarisability", the approximation like a plasmon-pole is used to get the spectrum (Fig.7).

I am wondering if there is a sort of "ab initio" way so that I can calculate this imaginary response for each frequency.

Best,
Yixin

[sonco]

Dear Yixin,
To be 100% sure I understood you:
Do you want to compute the absorption spectrum directly from the imaginary polarizability computed at different (real) frequencies?
At which level of theory?
All the best
Sonia

[yixin]

Dear Sonia,

Thank you for your reply. I don't need to calculate the absorption. What I need is to calculate the polarizability (\alpha) with respect to imaginary frequency- \alpha( i \omega). I am wondering if I can use Dalton to calculate this spectrum in order to compare with spectrums calculated by other methods (like using a model for polarizability).

Now, I only know how to calculate the polarizability with respect to real frequency \alpha( \omega ). But this transformation between real frequency and imaginary frequency seems to be non-trivial.

Since I want to compare with other methods, it will be perfect to have this spectrum at relatively high accuracy level, like CCSD. But I am not sure if this is possible right now.

Best,
Yixin

[Joanna]

Dear Yixin,

what you are asking for is available 'explicitly' at HF, DFT and MCSCF level of theory.
More details you can find in the manual, chapter about **RESPONSE, section *ABSORP

All the best,
Joanna

[tteodoro]

Dear Dalton experts,

The C6 coefficient for the Ne2 dimer that I get from the following output is totally different from what is found on literature (e.g., around 6.4 from Hattig, Christiansen, and Jo”rgensen, J. Chem. Phys, 107, 1997). I tried different CC methods and basis sets, results are similar.

What is missing?


Content of the .dal input file
----------------------------------

**DALTON INPUT
.RUN WAVE FUNCTIONS
**INTEGRALS
.DIPLEN
**WAVE FUNCTIONS
.CC
*CC INPUT
.CCSD
*CCLR
.OPERATOR
ZDIPLEN ZDIPLEN
.DISPCF
10
**END OF DALTON INPUT


Content of the .mol file
----------------------------

BASIS
aug-cc-pCV5Z
Neonium dimer
using the aug-cc-pCV5Z basis
Atomtypes=1
Charge=10.0 Atoms=2
Ne 0.0 0.0 0.000
Ne 0.0 0.0 6.630


Output:


A operator B operator n D_AB
----------------------------------------------------------------------------------------

ZDIPLEN ZDIPLEN -4 41248.6877579
-2 20.0385938177
0 5.38230818615
2 5.63794974443
4 9.46851069558
6 19.6218654445
8 44.4048295907
10 103.959739247

[michaljz]

hello,
as I understand, the recent versions of Dalton work more or less automatically with the proper
input data, providing C6 and/or, as a byproduct, the needed imaginary frequency polarisabilities,
but some comments on the old techniques that apparently have been forgotten may be useful:

if you want the coefficient for the Ne-Ne interaction, you need to run the calculation for the
Ne atom - not for the dimer at any geometry (that, I believe, has to be so even now...).
In the output shown you get the coefficients of the Cauchy expansion for the atom, which enables
you to compute the real and imaginary polarisabilities at any frequency (but note: you need to use some
analytic continuation of the series, e.g. Pade approximants; it is only asymptotically convergent
beyond the first pole). These enable you to get via the following integration the C6. If you want
the C6 coefficient for e.g. He-Ne interaction you integrate the product of He and Ne imaginary
polarisabilities etc.

Moreover (commenting on some older questions) the same technique may be applied to get many
other interaction coefficients, e.g for quadrupolar interactions or interaction-induced hyperpolarisabilities;
I am not sure that these are so easily available in a more automatic manner.

regards,
Michal

[sonco]

Dear Tiago,

Michal is right, and I apologize if my message was misunderstood.
As I wrote, one can use the same _strategy_ described in J. Chem. Phys., Vol. 107, 1997, 10592,
i.e. compute the Cauchy moments for the atom(s), apply Pade approximants and integrate.

Regards
Sonia

[tteodoro]

Dear Sonia and Michal,

Thank you for the clarification.

Best wishes,

Tiago