Molecule fall apart

Questions
#1

I’m a new Q-Chem user. I am doing an optimisation of the first singlet excited state. I have already optimised and checked that there is no negative frequency for ground state. However, when I tried to optimise the singlet excited state, the molecule fell apart and was not converge with an error message “Q-Chem fatal error occurred in module setman/rpa_dia.C, line 59: Unable to diagonalize A-B”. Would someone be able to suggest what I could do to fix this issue? Thank you very much

Here is my input file:

$molecule
0 1
XXX
$end

$rem
JOB_TYPE opt
EXCHANGE wb97xd
BASIS def2-svp
CIS_N_ROOTS 5
CIS_STATE_DERIVATIVE  1
RPA 1
CIS_SINGLETS 1
CIS_TRIPLETS 0
CIS_RELAXED_DENSITY TRUE
SOLVENT_METHOD PCM
GUI 2
$end

$pcm
Theory IEFPCM
$end

$solvent
Dielectric 35.688000
OpticalDielectric 1.806874
$end
#2

It could be that the excited state in question is dissociative; you should plot the energies along the optimization trajectory. That error message is usually a sign of a triplet instability, which is not that surprising if you have a highly stretched bond. (Use RPA = 0 to avoid this, which invokes the Tamm-Dancoff approximation. That is the default setting.)

#3

Thank you very much for your suggestion. I forgot to mention that I also found something similar in the previous post. I did try using RPA = 0 as well, however, it still felt apart into small fragments. Do you think if I can do the same approach but use different level of theory in case of TDA is not good enough for my system? Thanks again

#4

That suggests that the surface is simply dissociative. It’s probably worth stepping through the geometries to follow what is happening along the path of the optimization. You are using TDDFT and Tamm-Dancoff approximation is more-or-less required for photochemical problems due to the appearance of numerous triplet instabilities away from the ground-state geometry. It should not be regarded as a “lesser” level of theory. See here for more information:
https://doi.org/10.1016/B978-0-323-91738-4.00005-1

1 Like