Dear DMol3 users and developers,

I have a couple of questions regarding setting up and converging DFT calculations to a given spin state.

To be specific, I have performed geometry optimization (blyp functional + vpsr pseudopotntial) on the adsorption of hydroxyl (OH) on some organometallic molecules with one Co atom in the center. I have started calculations setting up several spin states (Spin "nr. of unpaired electrons", Occupation fermi) flags in the input, see the input below. However, whatever Spin value I set up it always converges to the same configuration with 2 unpaired electrons. This would be fine; although I tried to calculate singlet spin state by forcing calculation to be spin restricted (since the Spin flag appears to not work) and got configuration lower in energy by 0.6 eV that the one resulted from unrestricted calculations.

Therefore my questions are:

If Dmol3 converges to the lowest spin configurations is there any reason that could fail to find that this lowest energy singlet state as in this example?

Why Spin flag in input appears not to work?

I would be grateful for your comments and suggestion.

Thank you in advance,

Jakub

I am using Dmol3 v 6.0

Input file I use (for singlet state)

# Task parameters

Calculate                     optimize

Opt_energy_convergence        1.0000e-005

Opt_gradient_convergence      2.0000e-003 A

Opt_displacement_convergence  5.0000e-003 A

Opt_iterations                300

Opt_max_displacement          0.3000 A

Initial_hessian               improved

Symmetry                      off

# Electronic parameters

Spin_polarization             unrestricted

Spin                            0

Charge                        0

Basis                         dnp

basis_version                 basfile_v4.4

Pseudopotential               vpsr

Functional                    blyp

Aux_density                   octupole

Integration_grid              fine

Occupation                    fermi

Cutoff_Global                 4.5000 angstrom

Scf_density_convergence       1.0000e-005

Scf_charge_mixing             0.1000

Scf_spin_mixing               0.3000

Scf_iterations                343

Scf_diis                      6 kresse

# Print options

Print                         eigval_last_it

# Calculated properties

Plot                          spindensity

Mulliken_analysis             charge

Hirshfeld_analysis            charge

Frequency_analysis            off

Grid                          msbox  3 0.2500 0.2500 0.2500 3.0000

final occupancy from .occup file

  257  1.000000  1.000000     a   

     2  1.000000  0.000000     a   

     0