Electronic structure of s-indacene

Question

Structure shown in (a) is the regular $s$-indacene molecule, and (b) is a structure that i generated such that the central six-member ring remains similar to benzene. It is well-known in literature that the electronic structure of $s$-indacene is contested, with multiple groups suggesting a fully-delocalized $D_{2h}$ symmetry, and others suggesting a bond-localized $C_{2h}$ symmetry for the molecule. I recently found a paper contesting the use of B3LYP for core-expanded systems (like $s$-indacene). My question concerns the probable existence of the structure represented in (b). Will that be the reason why some groups got a $D_{2h}$ minima?

Further context:

I am interested in studying the electronic properties of periodic systems that can be generated from $s$-indacene. As a first step, I tried optimizing the geometry of the molecule with the fully delocalized $D_{2h}$ symmetry. I observed that this structure is a first-order saddle-point on the potential energy surface with an imaginary mode of 354.52 cm$^{-1}$ converting the $D_{2h}$ structure to $C_{2h}$ symmetry. However, when i tried optimizing the molecule with a triplet ground state, I observed that it was a minimum on the potential energy landscape, but the total energy was ~ 0.57 eV higher than the singlet $C_{2h}$ state. Now, If i try optimizing the molecule with a triplet starting guess and force it to a singlet state, I can get a new singlet minima with $D_{2h}$ symmetry, which is ~ 0.03 eV lower than the $C_{2h}$ symmetry. This led me to think that the structure represented in (b) might be (at least partially) responsible for the observed $D_{2h}$ symmetry state. This state should have two electrons, which are spatially separated but still remaining as $\alpha$ and $\beta$ spins.

References

1. s-Indacene: A Delocalized, Formally Antiaromatic 12 π Electron System
2. s-Indacene, a quasi-delocalized molecule with mixed aromatic and anti-aromatic character

Answer 1

Never trust a result obtained with a single density functional. You should perform calculations also with other types of functionals and see if you get a similar answer.

The molecule in question is not very big, so you could also tackle the question about the geometry with high-level ab initio methods. A good sanity check would be to evaluate CCSD(T) single-point energies at your DFT optimized geometries to assess whether the state ordering flips. You can even use approximate CCSD(T) methods for this check, such as pair-localized or local natural orbital truncations.