Failure to converge geometry of ringed structure

Question

Hi I'm trying to optimise a molecule by using the b3lyp/6-31G(d) method , the calculus took more than 2 days and the energy never converged .

Here is my input code

%chk=S-21.chk

# opt freq  B3LYP/6-31G*  SCRF=(PCM,read,solvent=CH2Cl2)

S-21

0 1
 C                 -1.75068100    0.37734700    0.81263500
 C                 -2.62257500   -0.29227700   -0.14373800
 C                 -4.10723100   -0.09641000   -0.01731400
 C                 -4.61852400    0.73486000    1.04941400
 C                 -3.74681500    1.31785500    1.90876500
 C                 -2.31430800    1.13987700    1.79050600
 H                 -5.69934800    0.85871100    1.11964000
 H                 -4.10201200    1.94778900    2.72638200
 H                 -1.68650000    1.64617500    2.52642700
 N                 -4.80498300   -0.74379100   -0.93998600
 N                 -2.28519800   -1.08000700   -1.15378200
 S                 -3.70847600   -1.57391000   -1.94372300
 C                 -0.28701900    0.19538500    0.70641300
 C                  0.55775500    1.19398300    0.34091700
 C                  0.21097700   -1.14867100    0.99271500
 C                  2.05755400    0.97694800    0.39853600
 C                  1.50838800   -1.47640300    0.80224900
 H                 -0.52454700   -1.88180100    1.32139800
 C                  2.47232700   -0.47418800    0.32746700
 C                  0.10593100    2.52300100   -0.07746100
 C                  0.93014100    3.38020900   -0.71164500
 C                  2.31062000    3.02195300   -1.00934000
 C                  2.84013300    1.86879500   -0.54543500
 H                 -0.93897900    2.77149900    0.10783000
 H                  0.59157200    4.36097200   -1.04420600
 H                  2.87715100    3.70674700   -1.63618300
 H                  2.36087800    1.32019000    1.44439100
 C                  4.20468100    1.44278400   -0.87492700
 C                  4.61668300    0.17803800   -0.66911800
 C                  3.71092200   -0.82964300   -0.11389200
 H                  4.85713200    2.19281800   -1.32106100
 H                  5.62508700   -0.14589500   -0.92529400
 C                  4.14879800   -2.22396900   -0.04913700
 C                  1.99749300   -2.88270200    1.03166300
 C                  3.36272400   -3.17101600    0.48626400
 H                  5.13572700   -2.45364800   -0.44739700
 H                  1.27197900   -3.60363700    0.58826800
 H                  1.99321400   -3.09445600    2.12663400
 H                  3.67769000   -4.21084800    0.55589600

 radii=UAKS

                                                                                     

                                                  

Here is the errors that I got

Error termination request processed by link 9999.
 Error termination via Lnk1e

Information about geom opt convergence

     Item               Value     Threshold  Converged?
 Maximum Force            0.043390     0.000450     NO 
 RMS     Force            0.003584     0.000300     NO 
 Maximum Displacement     0.044852     0.001800     NO 
 RMS     Displacement     0.004028     0.001200     NO 
                                                     

Here is the output structure (the bond between C(16)-H(27) 16 got broken) I don't know how to fix these issues

Answer 1

The bond breaking is expected, because the C-H bond is much weaker than an average C-H bond due to extensive hyperconjugative effects. The C-H bond thus elongates to such a degree that GaussView does not think the atoms are bonded anymore.

The geometry non-convergence is likely an unrelated issue, probably due to some dihedral being too flexible. This may at least in part be due to that you forgot to add the dispersion correction (em=gd3bj). Without the dispersion correction, not only will some loose degrees of freedom (dihedrals, distances between non-covalently interacting atoms) be poorly described and lead to poor convergence, but also your results will be unreliable even if your calculation does converge. Thus, you may try adding the dispersion correction and continue the optimization, starting from the last geometry of your present job.

Another thing that may or may not have a negative influence on the reliability of your results, is that you have a large and not that conventional conjugated system, which may be susceptible to self-interaction error. Using a functional with higher HF exchange (e.g. M06-2X, wB97XD) may put you on the safe side, but note that M06-2X tends to worsen the geometry convergence due to its potential energy surface being relatively non-smooth.