CDFT for dimer calculation


Just Got Here
Hello,

I'm trying to get the IP and EA energy of TCNQ dimer by using CDFT. So I impose one charge on one of the TCNQ molecule. But the energy doesn't converge in this situation. I have tried using different basis sets and different pop for CDFT, the problem still exists. Please can anyone give me any idea?

Here is my input:


start cdft
charge 1.0
permanent_dir ./perm
scratch_dir ./tmp

geometry units angstroms
C 8.75682754 8.51380000 4.86094765
C 8.77878471 9.25420000 3.62694867
C 8.78408467 10.61560000 3.62794891
C 8.76602745 11.35420000 4.86294808
C 8.75207028 10.61380000 6.09694743
C 8.74847033 9.25220000 6.09594727
C 8.75962741 12.75440000 4.86394778
C 8.77408333 13.49980000 3.65669845
C 8.73457140 13.49740000 6.07294662
N 8.71202488 14.09880000 7.07694558
N 8.78632979 14.10320000 2.65394902
C 8.74142761 7.11360000 4.85944694
C 8.71977166 6.36860000 6.06719594
C 8.74258361 6.37020000 3.65069699
N 8.70182521 5.76540000 7.06994511
N 8.74213011 5.76820000 2.64694697
H 8.79012853 8.70380000 2.68119919
H 8.79982843 11.16740000 2.68319964
H 8.74322647 11.16400000 7.04269702
H 8.73692656 8.70040000 7.04069673
C 11.75682754 8.51380000 4.86094765
C 11.77878471 9.25420000 3.62694867
C 11.78408467 10.61560000 3.62794891
C 11.76602745 11.35420000 4.86294808
C 11.75207028 10.61380000 6.09694743
C 11.74847033 9.25220000 6.09594727
C 11.75962741 12.75440000 4.86394778
C 11.77408333 13.49980000 3.65669845
C 11.73457140 13.49740000 6.07294662
N 11.71202488 14.09880000 7.07694558
N 11.78632979 14.10320000 2.65394902
C 11.74142761 7.11360000 4.85944694
C 11.71977166 6.36860000 6.06719594
C 11.74258361 6.37020000 3.65069699
N 11.70182521 5.76540000 7.06994511
N 11.74213011 5.76820000 2.64694697
H 11.79012853 8.70380000 2.68119919
H 11.79982843 11.16740000 2.68319964
H 11.74322647 11.16400000 7.04269702
H 11.73692656 8.70040000 7.04069673
end

basis
* library dzp
end
dft
xc xpbe96 cpbe96
direct
iterations 500
grid file
mult 2
convergence nolevelshifting damp 90
cdft 1 20 charge 1.0
end
task dft


Best,
Xiaoming Wang

Just Got Here
Hi Xiaoming,

A key thing that I often forget when I revisit CDFT is that you need a sufficiently diffuse basis set for the Lagrange multipliers to be able to act.

I dug up a couple of examples for CDFT from a few years ago. I would imagine adding a load of diffuse functions to the above will get it to work. More complex ways of getting it to converge are to do calculate the electronic structure on the fragments in the correct charge state separately, and then combine these as the guess for CDFT. These are presented for a Ethylene molecular pair, but I used the same scripts for dye molecules + fullerene/oligomer pairs.

https://github.com/jarvist/NWCHEM-examples/tree/master/CDFT-Ethylene


Forum >> NWChem's corner >> Running NWChem