I have optimized molecular geometry as the lowest excited singlet state at the TDDFT/RPA level of theory, using the D22 symmetry group (D2 for the largest abelian counterpart). However, the subsequent frequency calculations (following automatically after the geometry optimization) used reduced symmetry (C2v, or - abelian - C2) and failed due to unconverged SCF.
Is this symmetry change a regular behavior of QChem? Some sort of algorithm-related symmetry restrictions on the excited state frequency analysis?
Or prehaps it may be related to the rather strange message that appeared after the first SCF (and repeated at every step of the geometry optimization) - that the program was unable to determine the symmetry of several (occupied!) orbitals. Again, other packages worked fine in this capacity.
Has any of you come across a similar behavior. What can be done to avoid, or at least circumvent the problem?
My very best regards
Marcin
This is hard to diagnose without an input file.
Certainly. It is rather long, I’m afraid, but I am pasting it anyway.
$molecule
0 1
C -0.8455806 0.8455806 -5.7859466
C 0.0000000 0.0000000 -5.0766735
C 0.8455806 -0.8455806 -5.7859466
C 0.8462900 -0.8462900 -7.1708921
C 0.0000000 0.0000000 -7.8674392
C -0.8462900 0.8462900 -7.1708921
C -0.0000000 0.0000000 -3.5884304
C -0.8637392 -0.8637392 -2.8913192
C -0.8638212 -0.8638212 -1.4447478
C -0.0000000 0.0000000 -0.7441180
C 0.8638212 0.8638212 -1.4447478
C 0.8637392 0.8637392 -2.8913192
C -1.7347621 -1.7347621 -3.5572492
C -2.5932985 -2.5932985 -2.8814254
C -2.5928770 -2.5928770 -1.4519138
C -1.7322499 -1.7322499 -0.7780964
C 1.7322499 1.7322499 -0.7780964
C 2.5928770 2.5928770 -1.4519138
C 2.5932985 2.5932985 -2.8814254
C 1.7347621 1.7347621 -3.5572492
C -3.4745933 -3.4745933 -3.5620462
C -4.3036720 -4.3036720 -2.8723494
C -4.3027591 -4.3027591 -1.4570259
C -3.4721816 -3.4721816 -0.7695776
C 3.4721816 3.4721816 -0.7695776
C 4.3027591 4.3027591 -1.4570259
C 4.3036720 4.3036720 -2.8723494
C 3.4745933 3.4745933 -3.5620462
C -0.0000000 -0.0000000 0.7441180
C -0.8638212 0.8638212 1.4447478
C -0.8637392 0.8637392 2.8913192
C -0.0000000 -0.0000000 3.5884304
C 0.8637392 -0.8637392 2.8913192
C 0.8638212 -0.8638212 1.4447478
C -1.7322499 1.7322499 0.7780964
C -2.5928770 2.5928770 1.4519138
C -2.5932985 2.5932985 2.8814254
C -1.7347621 1.7347621 3.5572492
C 1.7347621 -1.7347621 3.5572492
C 2.5932985 -2.5932985 2.8814254
C 2.5928770 -2.5928770 1.4519138
C 1.7322499 -1.7322499 0.7780964
C -3.4721816 3.4721816 0.7695776
C -4.3027591 4.3027591 1.4570259
C -4.3036720 4.3036720 2.8723494
C -3.4745933 3.4745933 3.5620462
C 3.4745933 -3.4745933 3.5620462
C 4.3036720 -4.3036720 2.8723494
C 4.3027591 -4.3027591 1.4570259
C 3.4721816 -3.4721816 0.7695776
C 0.0000000 0.0000000 5.0766735
C -0.8455806 -0.8455806 5.7859466
C -0.8462900 -0.8462900 7.1708921
C 0.0000000 0.0000000 7.8674392
C 0.8462900 0.8462900 7.1708921
C 0.8455806 0.8455806 5.7859466
H -1.5077484 -1.5077484 5.2443726
H 1.5077484 1.5077484 5.2443726
H 0.0000000 0.0000000 8.9487185
H 1.5077484 -1.5077484 -5.2443726
H 0.0000000 0.0000000 -8.9487185
H -1.5077484 1.5077484 -5.2443726
H 1.7371434 1.7371434 0.3026747
H 1.7428412 1.7428412 -4.6376215
H 3.4697442 3.4697442 0.3125111
H 4.9678763 4.9678763 -0.9239001
H 4.9696218 4.9696218 -3.4034850
H 3.4736502 3.4736502 -4.6442110
H -1.7371434 -1.7371434 0.3026747
H -1.7428412 -1.7428412 -4.6376215
H -3.4697442 -3.4697442 0.3125111
H -4.9678763 -4.9678763 -0.9239001
H -4.9696218 -4.9696218 -3.4034850
H -3.4736502 -3.4736502 -4.6442110
H -1.7371434 1.7371434 -0.3026747
H -1.7428412 1.7428412 4.6376215
H -3.4697442 3.4697442 -0.3125111
H -4.9678763 4.9678763 0.9239001
H -4.9696218 4.9696218 3.4034850
H -3.4736502 3.4736502 4.6442110
H 1.7371434 -1.7371434 -0.3026747
H 1.7428412 -1.7428412 4.6376215
H 3.4697442 -3.4697442 -0.3125111
H 4.9678763 -4.9678763 0.9239001
H 4.9696218 -4.9696218 3.4034850
H 3.4736502 -3.4736502 4.6442110
H 1.5105954 1.5105954 7.7066737
H -1.5105954 1.5105954 -7.7066737
H 1.5105954 -1.5105954 -7.7066737
H -1.5105954 -1.5105954 7.7066737
$end
$rem
jobtype opt
exchange cam-b3lyp
basis def2-TZVPP
CIS_N_ROOTS 2
CIS_SINGLETS true
CIS_TRIPLETS false
CIS_STATE_DERIV 1 Lowest TDDFT state
RPA true
SCF_CONVERGENCE 9
XC_GRID 3
mem_static 2000
mem_total 180000
$end
@@@
$molecule
read
$end
$rem
jobtype freq
exchange cam-b3lyp
basis def2-TZVPP
CIS_N_ROOTS 1
CIS_SINGLETS true
CIS_TRIPLETS false
CIS_STATE_DERIV 1 Lowest TDDFT state
RPA true
SCF_CONVERGENCE 9
XC_GRID 3
mem_static 2000
mem_total 180000
$end
In fact, there is a good reason for the program to “want” to go from the D2d to the c2v symmetry, as the excitation may like to be located on one of the (pentacene) chromophores, that are symmetry related in the ground state. Nonetheless, I would rather expect the frequency analysis to continue with the initial symmetry and produce an imaginary frequency normal-mode, that would indicate the high symmetry structure corresponds to a saddle-point between two basins, each one for a diabatic state with broken symmetry.
So far, however, I have never come across a program that would enforce such a solution before the frequency job is done.
Also, there is still the misterious orbitals without appropriate representation assigned…
Looking at the molecular structure I can see what you mean, if the D2d is an unstable symmetry-breaking point between two symmetry-equivalent C2v minima (basically, a Jahn-Teller problem). That’s a challenging enough electronic structure problem that my gut feeling is that it may be too much to ask to land on that D2d point and get the imaginary frequency; you might be better off starting from the C2v minimum figuring out which normal mode connects to the symmetry-equivalent minimum, and just mapping out the 1d potential point-by-point.
I’ll run this input and if I find anything worth reporting I will do so.
The potential energy scan (relaxed) is a good idea to figure out, what is going on there physically, and how high is the D2d-symmetry barrier separating the two C2v-symmetry basins. This is crucial in order to determine what is the actual dominating conformation (if the barrier is small enough, the molecule can still be planar rather than distorted, even if the PES minimum corresponds to the latter conformation).
I have already found out the key normal mode, whiah has the irrational frequency, using Turbomole.
That package, however, having converged to the stationary point, allows for the harmonic analysis at this point without symmetry breaking. I wonder what happens with QChem that it spontaneously breaks symmetry when geometry optimization job is followed by the frequency one. This is actually the first time I have enountered such a behaviour, and I have dealt with a number of quantum chemistry packages.
Marcin
Are you confident that you are obtaining the same SCF solution with all of the codes? This feels like a problem where a somewhat different guess might easily converge to a different solution.