Dimer calculation with one frozen molecule

Aarzoo · July 11, 2024, 2:23pm

Hi
I am doing a dimer calculation by freezing one molecule using the 6.1 version. I did S1 optimization but I got imaginary frequency in that. So I put this input taking previous optimized coordinates. In the 2 step when the molecule is optimizing, the frozen molecule is also changing its angles but I do not want that. I already gave a fixed coordinates command for the frozen molecule. So, why is this happening?

$molecule
0 3
Dimer system consists of 72 atoms
$end

$rem
BASIS = cc-pVDZ
EXCHANGE = omegaB97XD
JOB_TYPE = frequency
SCF_CONVERGENCE = 10
MAX_SCF_CYCLES = 400
MAX_CIS_CYCLES = 400
UNRESTRICTED = true
SPIN_FLIP = 1
CIS_N_ROOTS = 4
THRESH = 14
SYMMETRY_IGNORE = true
SYMMETRY = false
CIS_STATE_DERIVATIVE = 3
SOLVENT_METHOD = PCM
$end

$solvent
DIELECTRIC 32.613000
OPTICALDIELECTRIC 1.765709
$end

FIXED
37:72 XYZ
ENDFIXED
$end

@@@

$molecule
read
$end

$rem
BASIS = cc-pVDZ
EXCHANGE = omegaB97XD
JOB_TYPE = optimization
SCF_CONVERGENCE = 10
MAX_SCF_CYCLES = 400
MAX_CIS_CYCLES = 400
UNRESTRICTED = true
SPIN_FLIP = 1
CIS_N_ROOTS = 4
THRESH = 14
SYMMETRY_IGNORE = true
STS_MOM = true
SYMMETRY = false
CIS_STATE_DERIVATIVE = 3
SOLVENT_METHOD = PCM
GEOM_OPT_MAX_CYCLES = 300
$end

$solvent
DIELECTRIC 32.613000
OPTICALDIELECTRIC 1.765709
$end

FIXED
37:72 XYZ
ENDFIXED
$end

@@@

$molecule
read
$end

$rem
BASIS = cc-pVDZ
EXCHANGE = omegaB97XD
JOB_TYPE = frequency
SCF_CONVERGENCE = 10
MAX_SCF_CYCLES = 400
MAX_CIS_CYCLES = 400
UNRESTRICTED = true
SPIN_FLIP = 1
CIS_N_ROOTS = 4
THRESH = 14
SYMMETRY_IGNORE = true
SYMMETRY = false
CIS_STATE_DERIVATIVE = 3
SOLVENT_METHOD = PCM
$end

$solvent
DIELECTRIC 32.613000
OPTICALDIELECTRIC 1.765709
$end

FIXED
37:72 XYZ
ENDFIXED
$end

kaushik · July 11, 2024, 5:26pm

Can you please post a full minimal opt job that shows this behavior, so we can double check? Also, “$opt” string is missing before FIXED.

jherbert · July 11, 2024, 8:27pm

It looks like there are typos in your input file (no $opt before the FIXED). Check that first. In general, you should expect that if you fix some atoms then you are going to get imaginary frequencies because you have not relaxed all degrees of freedom.

If the problem persists, I suggest that you experiment with a small example (e.g., water dimer) to see if you can figure out the input format.

Aarzoo · July 12, 2024, 5:18am

Dear @jherbert and @kaushik
Thanks and I also noticed, in this input I forgot to give $opt
But I want to ask one question, using $opt before fixed I optimized geometry at S0 and S1 states but I am not getting structural change in the relaxing molecule.

Input for S0-opt
$molecule
0 3
dimer coordinates (72 atoms)
$end

$rem
BASIS = cc-pVDZ
EXCHANGE = omegaB97XD
JOB_TYPE = OPTIMIZATION
SCF_CONVERGENCE = 8
MAX_SCF_CYCLES = 700
MAX_CIS_CYCLES = 700
UNRESTRICTED = true
SPIN_FLIP = 1
CIS_N_ROOTS = 4
THRESH = 14
SYMMETRY_IGNORE = true
SYMMETRY = false
CIS_STATE_DERIVATIVE = 1
SOLVENT_METHOD = PCM
GEOM_OPT_MAX_CYCLES = 200
$end

$solvent
DIELECTRIC 32.613000
OPTICALDIELECTRIC 1.765709
$end

$opt
FIXED
37:72 XYZ
ENDFIXED
$end

end of the output is like this:
Maximum Tolerance Cnvgd?
Gradient 0.000253 0.000300 YES
Displacement 0.000251 0.001200 YES
Energy change ********* 0.000001 NO

Final energy is -1798.28778575099

** OPTIMIZATION CONVERGED **

Input for S1-opt:
$molecule
0 3
dimer coordinates (72 atoms)
$end

$rem
BASIS = cc-pVDZ
EXCHANGE = omegaB97XD
JOB_TYPE = OPTIMIZATION
SCF_CONVERGENCE = 8
MAX_SCF_CYCLES = 700
MAX_CIS_CYCLES = 700
UNRESTRICTED = true
SPIN_FLIP = 1
CIS_N_ROOTS = 4
THRESH = 14
SYMMETRY_IGNORE = true
SYMMETRY = false
CIS_STATE_DERIVATIVE = 3
SOLVENT_METHOD = PCM
GEOM_OPT_MAX_CYCLES = 300
$end

$solvent
DIELECTRIC 32.613000
OPTICALDIELECTRIC 1.765709
$end

$opt
FIXED
37:72 XYZ
ENDFIXED
$end
and in output:

                         Maximum     Tolerance    Cnvgd?
     Gradient           0.000252      0.000300     YES
     Displacement       0.000916      0.001200     YES
     Energy change     -0.000002      0.000001      NO

Final energy is -1798.15487634121

** OPTIMIZATION CONVERGED **

FC energy is -1798.15486587 and from the energies, it is showing that at the S1 state molecule is not relaxing.
And from the output file It shows converged.

kaushik · July 12, 2024, 6:15am

You mean you do not see any difference between geometries optimized for S0 and S1 spin-flip states, each calculated using the same initial geometry and $opt section specifying the FIXED atoms, right? What happens if you use the S0-optimized geometry as input for the S1 geometry optimization? Does the optimization finish within a single step? Also, what do you mean by “FC” energy?

Aarzoo · July 12, 2024, 6:22am

Firstly, I optimized at S0 state SF state, and for S1-opt input, I used S0 optimized coordinates. The optimization for S1-state finished after 5 optimization cycles.
FC energy is Franck-Condon energy.

kaushik · July 12, 2024, 6:41am

What is the nature of the S1 state? Is it a localized spin-flip transition on one of the monomers? Which one?

You may also try to optimize S1 using the same initial geometry that you used in the S0 optimization perhaps and see what happens. I think your input setup should be fine as such; my calculation with a small test dimer system with your settings seems to suggest that.

Aarzoo · July 12, 2024, 10:00am

I checked the HOMO-LUMO excitations at the S1 state, HOMO is localized at the frozen monomer and LUMO is localized at the relaxing monomer.

jherbert · July 12, 2024, 2:02pm

I feel like you (and we) are probably missing something very simply. Why don’t you try similar options with water dimer, freezing one monomer, HF/STO-3G so that it runs fast. See if you get changes in the geometry or not. Otherwise this is hard to diagnose without a complete input file.

Aarzoo · July 12, 2024, 2:42pm

I tried with the same inputs for another dimer molecule and I am not facing any issue. For another molecule, I am getting changes in geometry, and at the S1 state, it is relaxing. That dimer consists of (108 atoms).

jherbert · July 12, 2024, 4:18pm

Then it is probably the case that your system is so highly constrained that you are locked into something where the energy can’t change. Optimization is considered converged when the gradient criterion is met along with one of the other two criteria.

jherbert · July 13, 2024, 1:47pm

If it ends with the same message (i.e., successful optimization with gradient and displacement criteria met, and energy change small) then it means you’ve reached a minimum so far as the optimizer can tell. If you think the structure should be something different, it probably suggests that it is too highly constrained.