UT theoretical chemistry code forum

Hi, Dear all:
We run a cNEB with low accuracy to get a initial guess of TS and then to using Dimer for a more accurate TS.
The question is the cNEB is very quikly ended, but the energies of images are increased from begain to end.

any suggestion to deal with this case ?

thank you!

If you include the endpoints, you will see a small barrier near the final state:

0 0.093491 -98.056200 0.000000
1 0.027829 -97.979800 0.076400
2 0.040481 -97.834500 0.221700
3 0.067300 -97.747100 0.309100
4 0.092867 -97.651800 0.404400
5 0.099485 -97.561500 0.494700
6 0.099307 -97.583000 0.473200

Thank you, Professor!
that is realy my results.
1. we increat the K points and the convegence level basd on this running, and regrettably find the 5 image goes down and its energy is below the FS
2.we use the 5th image as an initial guess of TS, runing a improved dimer, finally it is also lower in energy than FS

this is realy annoyance

You need to have all of your frozen atoms fixed in the same positions in the initial and final state. If you run nebmake.pl, they will also be frozen in the same position in each image, which is also required.

thank you for the reply!
the example is curtiled, the original systen includes 4 layered atoms in surface, the top two layer atoms is optimized with the absorbate, so there is a little difference of the top two layer atom positions in IS and FS, to have a good initial guess quickly, we removed the bottom two layer atoms in the surface and fixed the optimized top two layers to run the cNEB as that in the example.
we hope this will approximately include the surface change effect in the reaction process, is this really becoming the cause of the problem ?

Very likely. Generate a new path with all of the frozen atoms in exactly the same positions in the initial and final states, and then also for the NEB calculation.

thank you, professor!
we'll go on fighting.....

onward!