About CI-NEB calculation

[atash]

Hi All,
I want to calculate the energy barrier for the decomposition of an organic molecule on copper surface. After 200 ionic steps, my calculation seems doesn't converge.
here is my nebef.dat file

0 0.018009 -667.950900 0.000000
1 0.177444 -669.826200 -1.875300
2 0.111973 -669.753900 -1.803000
3 0.126540 -669.655900 -1.705000
4 0.187862 -669.548500 -1.597600
5 0.165763 -669.426400 -1.475500
6 0.111985 -669.383500 -1.432600
7 0.182682 -669.367500 -1.416600
8 0.262777 -669.382600 -1.431700
9 0.290907 -669.519300 -1.568400
10 0.213035 -669.668800 -1.717900
11 0.197178 -669.795900 -1.845000
12 0.093725 -669.868200 -1.917300
13 0.000000 -667.945200 0.005700

it seems that there is some problem with connecting initial and final state. I have some questions.

(1) Since the molecule is quite complicated, my reaction coordinates do not correspond to the single atom moving. Here admin suggested that The path should correspond to a single atom moving and cannot be rearrangement of several the atoms, did I understand admins reply correctly here. http://theory.cm.utexas.edu/henkelman/f ... ?f=2&t=655

(2) So what is the correct way of generating images? I explain my procedure: first, I optimized adsorption of the molecule on copper surface. Then, to obtain the final state, I stretched the bond between two atoms in the molecule and optimized this structure. However, this resulted in the rearrangement of other atoms in the molecule.
Should I run single point calculation or do geometry optimization when I prepare the final image?

here is my INCAR file
System = NEB
PREC= LOW
ISYM= 0
ICHAIN=0
SPRING=-5
IMAGES=12
POTIM=0.1
LCLIMB = .TRUE.
NSW = 500
ENCUT=450
EDIFFG = -0.05
EDIFF = 0.0001
IBRION = 1
IALGO = 48
LSCALAPACK = .FALSE.
LREAL=Auto
LWAVE = .FALSE.
LCHARG = .FALSE.
NPAR=24

(3) is anything wrong with my INCAR file that caused the energy doesn't converge?

I appreciate your advice. Thanks for your help.

Regards,
Yasheng

[graeme]

The most significant problem is that it seems that you have used different settings for your endpoints as compared to your NEB calculation. There is a difference of about -1.9 eV between the two calculations. Compare the INCARs to find the difference, and the repeat your NEB calculation using the same settings.