Hi, all.
I'm doing calculation of a dissociation curve of singlet lithium dimer using UHF module.
The point is that the spin multiplicity would be staying at singlet state so that one alpha electron is localized on one lithium atom, one beta electron on the other.
However, when I did a simple UHF calculation without any additional directives, the wavefunction remained symmetric for both alpha and beta electrons until dissociation limit.
So I divided wavefunction into fragments to make a guess orbital that is localized on each atom.
Here is my input.
geometry lidimer noautosym units angstrom
Li 8.00 0.00 0.00
Li 0.00 0.00 0.00
END
geometry li1 noautosym units angstrom
Li 8.00 0.00 0.00
END
geometry li2 noautosym units angstrom
Li 0.00 0.00 0.00
END
BASIS spherical
Li library def2-qzvpp
end
SET geometry li1
SCF
DOUBLET
UHF
ADAPT OFF
SYM OFF
VECTORS INPUT atomic OUTPUT li1.movecs
END
TASK SCF
SET geometry li2
SCF
DOUBLET
UHF
ADAPT OFF
SYM OFF
VECTORS INPUT atomic OUTPUT li2.movecs
END
TASK SCF
SET geometry lidimer
SCF
UHF
SINGLET
ADAPT OFF
SYM OFF
VECTORS INPUT fragment li1.movecs li2.movecs OUTPUT lidimer.movecs
END
task scf
However, this calculation causes an error that spin multiplicity is not correct in the dimer.
Here's my error message.
------------------------------------------------------------------------
movecs_fragment: open shell mismatch 3
------------------------------------------------------------------------
I think the program innately assigned the sum of doublet should always be triplet, but I think it should be corrected because there are counterexamples like my calculation..
Do you have any idea to do this calculation successfully?
Thanks in advance.
|