after CINEB run,the dimmer calculation cannot converge

[hsy]

Dear Professor,

I came across a problem while running dimer simulation. I am working on systerm which the hydroxyl of chlorophenol will be dissosicated on the metal oxide surface. The TS was first pre-converged to 0.3 ev/A by CINEB. I used the IOPT= 3 because IOPT=7 cannot converge. And because of the short path, I set one image between initial and final state. I checked the vibration direction and genenrated MODECAR by frequency calculation. Then I have calculated the pre-convergent structure to 0.03 ev/A in the following cases.

1. I did the CINEB calculation by the pre-convered structure, which converged to 0.1 ev/A. But the vibration analysis suggested that the vibration direction of TS has chenged: there is no tendency for hydrogen-oxygen bonds to break in hydroxyl groups. And now I want to know that if I continues to converged to 0.03, is there any point in me going on?

2. I used the structure pre-convered to 0.3 ev/A (and 0.1 ev/A)by CINEB to run with dimmer. The dimmer calculation cannot converge. The curvature is negative at first and then positive, and I checked the TS and found the TS has deviated significantly from the desired configuration. I used DRotMAX = 20(up to 50) to ensure Torque meets the convergent standard(DFNMin = 0.01,DFNMax = 1.0), but it still cannot converge. So I terminated the calculation.

For dimmer calculation, I have tried IOPT=2 and IOPT =7, I wonder why they cannot converge and what to do to be convergent. Can you give me some advice?

DIMCAR as belows：
Step Force Torque Energy Curvature Angle
1 1.03528 8.80664 -555.15757 -0.50637 6.11466
1 1.03528 4.53340 -555.15757 -1.58506 11.71038
1 1.03528 3.42043 -555.15757 -2.26297 3.64732
1 1.03528 2.60246 -555.15757 -2.49241 2.36270
1 1.03528 1.83164 -555.15757 -2.62759 1.69373
1 1.03528 1.57873 -555.15757 -2.67645 1.62528
1 1.03528 1.54773 -555.15757 -2.75617 1.17217
1 1.03528 1.21353 -555.15757 -2.79228 1.40054
1 1.03528 1.34245 -555.15757 -2.76781 0.95123
1 1.03528 0.96405 -555.15757 -2.81187 1.13771
2 0.57564 1.24160 -555.16256 -2.56816 1.07760
2 0.57564 1.01769 -555.16256 -2.57800 0.97444
2 0.57564 0.90613 -555.16256 -2.56972 0.74319
3 0.62516 1.96372 -555.16042 -2.32654 3.09263
3 0.62516 1.33012 -555.16042 -2.42538 0.95705
3 0.62516 0.73166 -555.16042 -2.37577 1.07178
4 0.48841 2.03484 -555.15708 -1.95254 2.58389
4 0.48841 1.44278 -555.15708 -2.06666 1.15182
4 0.48841 0.88580 -555.15708 -1.97189 0.80841
5 0.57759 2.27395 -555.15371 -1.25871 3.39766
5 0.57759 1.47895 -555.15371 -1.40859 1.13494
5 0.57759 0.90305 -555.15371 -1.35006 1.18183
6 0.49345 4.05353 -555.14977 0.22317 5.15971
6 0.49345 2.16645 -555.14977 -0.17340 2.35441
6 0.49345 1.56284 -555.14977 -0.15490 1.29523
6 0.49345 1.26222 -555.14977 -0.30739 1.49071
6 0.49345 1.40191 -555.14977 -0.22043 1.08233
6 0.49345 1.10522 -555.14977 -0.39738 1.23071
6 0.49345 1.32884 -555.14977 -0.24675 1.04351
6 0.49345 1.09615 -555.14977 -0.44215 1.05773
6 0.49345 1.27518 -555.14977 -0.26632 1.03223
6 0.49345 1.18085 -555.14977 -0.48765 1.00961
6 0.49345 1.29159 -555.14977 -0.28919 1.04259
6 0.49345 1.25462 -555.14977 -0.52168 1.00704
6 0.49345 1.29733 -555.14977 -0.31283 1.03484
6 0.49345 1.28276 -555.14977 -0.53651 0.99769
6 0.49345 1.28538 -555.14977 -0.31682 1.01493
6 0.49345 1.28022 -555.14977 -0.54669 0.97799
6 0.49345 1.26821 -555.14977 -0.32152 0.99279
6 0.49345 1.26997 -555.14977 -0.55812 0.95973
6 0.49345 1.25468 -555.14977 -0.32685 0.97228
6 0.49345 1.25616 -555.14977 -0.56520 0.93979
6 0.49345 1.24007 -555.14977 -0.32992 0.95623
6 0.49345 1.24307 -555.14977 -0.58536 0.91928
6 0.49345 1.22749 -555.14977 -0.35158 0.94983
6 0.49345 1.24512 -555.14977 -0.59283 0.91708
6 0.49345 1.21753 -555.14977 -0.35325 0.93465
6 0.49345 1.23065 -555.14977 -0.59591 0.90179
6 0.49345 1.20536 -555.14977 -0.35279 0.92282
6 0.49345 1.21982 -555.14977 -0.59943 0.88970
6 0.49345 1.19349 -555.14977 -0.35323 0.91094
6 0.49345 1.20786 -555.14977 -0.60308 0.87969
7 0.40485 1.51113 -555.14793 0.03619 1.56766
7 0.40485 0.81780 -555.14793 0.01435 1.07493
8 1.41529 11.02374 -555.10140 3.82626 8.28236
8 1.41529 2.92484 -555.10140 2.21999 5.99622
8 1.41529 3.57828 -555.10140 1.97186 3.62519
8 1.41529 2.14794 -555.10140 1.80653 3.42953
8 1.41529 2.57967 -555.10140 1.63525 2.75356
8 1.41529 2.19717 -555.10140 1.55361 2.04578
8 1.41529 1.96835 -555.10140 1.45552 2.87763
8 1.41529 2.29634 -555.10140 1.40386 1.67855
8 1.41529 1.55500 -555.10140 1.29419 3.47400
8 1.41529 2.51936 -555.10140 1.24096 1.61816
8 1.41529 1.30197 -555.10140 1.14736 4.70519
8 1.41529 2.87725 -555.10140 1.06576 1.63383
8 1.41529 1.07677 -555.10140 0.96238 8.37171
8 1.41529 3.42107 -555.10140 0.81925 1.81906
8 1.41529 0.83440 -555.10140 0.69471 5.06695
9 1.59183 2.36575 -555.06070 0.66965 1.73657
9 1.59183 1.28377 -555.06070 0.58400 1.86370
9 1.59183 1.37500 -555.06070 0.54622 1.05380
9 1.59183 0.95381 -555.06070 0.52915 1.77333
10 1.69188 1.98744 -555.00353 0.56032 1.29322
10 1.69188 0.97089 -555.00353 0.52345 2.34731
11 1.82425 1.96738 -554.92889 0.55420 1.21945
11 1.82425 0.91983 -554.92889 0.48846 2.53677
12 2.02095 1.96734 -554.84043 0.53960 1.22204
12 2.02095 0.94625 -554.84043 0.47935 2.88912
13 2.27574 2.01246 -554.74200 0.48353 1.17438
13 2.27574 0.89594 -554.74200 0.44290 3.75723
14 2.61732 2.23424 -554.63906 0.37250 1.27631
14 2.61732 0.90503 -554.63906 0.32817 2.84233
15 2.82279 1.76566 -554.53778 0.28743 1.22465
15 2.82279 0.89116 -554.53778 0.23886 1.29287

I have attached the tar file of my calculations. I would be grateful if you could help me resolve this.

[graeme]

I'm very glad that you posted this calculation because it is such a classical example of what we call a shoulder on the potential energy surface. These can be hard to picture, but in this case, it is pretty clear because the reaction coordinate largely involves just H-transfer.

But first, I took your NEB and dimer calculations and redid them with a more conservative optimizer. What you can see in the attached calculations (neb_new and dimer_new) is that the force on the NEB climbing image reduces but was not able to reach convergence. In fact, the climbing image is just wandering in circles in a low-force region of configuration space. My dimer calculation did just what your's did - it looks like it's converging and then it shoots off into a completely different region of configuration space in search of a different saddle.

What explains these two results is that there is no saddle for your reaction. Instead, the energy becomes very flat when moving from reactant to product, and then drops down to the product state. To confirm the presence of this shoulder, I took your reactant configuration and reconverged it with a lower EDIFFG value. As you can see (min1), the force gets quite low until near the end of the calculation where the force starts to go up and the energy is dropping as it starts to head towards the product.

I killed all these calculations once I was convinced I understood the situation, but you can certainly continue the minimization from your reactant to your product and verify that it is downhill in energy along the path - there is no saddle point to find.

[hsy]

Dear Professor,

Thank you very much for your time on this! I will try your suggestion.
But I want to know If it's because of the low barrier or even no barrier that I can't find saddle point beteen reactant and product.

[graeme]

It is not a low barrier; there is no barrier. In the min1 calculation I posted, a careful relaxation of your initial state goes to the product state.

[hsy]

Thank you very much for your advice! I learned a lot about CINEB and dimer.