SCAN functional gives inaccurate atomization energy for C2 dimer

62 views
Skip to first unread message

Tanvir ur Rahman Chowdhury

unread,
Sep 12, 2021, 11:34:31 AM9/12/21
to NWChem Forum
Hi,
I have been trying to compute the atomization energy of C2 dimer. The following two codes and then using (C_atom - 2* C2_dimer). It gives me an atomization energy of 5.22 eV where the reference value is 6.22 eV. I was wondering what might be the issue. Thanks in advance.
                                                                      C2 Dimer
echo
start c_mol_01

geometry
  C   0.0 0.0 0.0
  C   0.0 0.0 1.243
end

basis
 * library 6-311++G(3df,3pd)
end

dft
 odft
 mult 1
 grid huge
convergence energy 1d-8
 iterations 100
 xc scan
 sym off
 adapt off
end

task dft

                                                                  C atom
echo

start carbon_NS

geometry
  C   0.0 0.0 0.0
end

basis
* library 6-311++G(3df,3pd)
end

dft
 odft
 mult 3
 grid xfine
 convergence energy 1d-8
 xc scan
 iterations 100
 sym off
 adapt off
end

task dft


Edoardo Aprà

unread,
Sep 12, 2021, 1:57:11 PM9/12/21
to NWChem Forum
Could you provide a reference for the C2 SCAN atomization energy?

Tanvir ur Rahman Chowdhury

unread,
Sep 12, 2021, 2:35:03 PM9/12/21
to NWChem Forum
Hi Edoardo,
Sure. I am using the Computational Chemistry Comparison and Benchmark DataBase of NIST as a reference. As you could see in the link, the atomization energy of C2 dimer is about 600 kJ/mol or 6.22 eV. C2_Atomization
Thanks.
ref.PNG

Tanvir ur Rahman Chowdhury

unread,
Sep 12, 2021, 2:35:06 PM9/12/21
to NWChem Forum
The reference value is the experimental atomization energy, not the SCAN energy. Because we are comparing the computed SCAN  atomization energy to the experimental value. And, it is hard to believe that there will be a 1 eV difference between SCAN value and experimental value. I was wondering if it has to do with the converge or grid etc.
Thanks.

On Sunday, September 12, 2021 at 1:57:11 PM UTC-4 Edoardo Aprà wrote:

Edoardo Aprà

unread,
Sep 12, 2021, 3:19:49 PM9/12/21
to NWChem Forum
Not sure how present day DFT can deal with system such as C2 with a strong multireference character.
Anyhow, if you set the dimer multiplicity to triplet you get an atomization energy closer to the experimental value.
However, C2's ground state is not a triplet as discussed in the two references above.

Eric Bylaska

unread,
Sep 12, 2021, 4:35:40 PM9/12/21
to NWChem Forum
Generally ODFT gives energies that are too low for the C2 dimer, because it produces a (Dunlap) spin wave.  Maybe this is fixed with scan, but you should look at the spin density to make sure.
--
You received this message because you are subscribed to the Google Groups "NWChem Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nwchem-forum...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/nwchem-forum/6a8bf8d9-063c-447e-9af2-4bd87b717e19n%40googlegroups.com.

Eric Bylaska

unread,
Sep 13, 2021, 10:03:20 PM9/13/21
to NWChem Forum
Also if you're able to get a restricted singlet to converge, you’ll get a pretty good energy.  Note the lumo is lower than the homo so this can be tricky.
--
You received this message because you are subscribed to the Google Groups "NWChem Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nwchem-forum...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages