cNEB Diverges in specific images

[RonenKru]

Hello Prof. Graeme,

I am trying to use cNEB for calculation of the energy barrier for catalysis of CO2 over MoS2 slab (On a specific side).
The reaction is as followed: MoS2 + H(ads) + CO2(g) ---> MoS2 + COOH (ads).
I am using IOPT=3 and 8 images, Attached cNEB files.
In this case it seems images 1-4 are diverging due to high forces on the X axis on CO2 molecule and maintains these high forces.
How would you suggest to approach this problem?

I am using these INCAR tags:
ISTART = 0
ICHARG = 2
SYSTEM = MoS2 slab
ENCUT = 500
ISMEAR = 0 ; SIGMA = 0.01
ALGO = Fast
EDIFF = 1E-6
EDIFFG = -0.1
NSW = 500
LREAL = A
ISIF = 2
PREC= Accurate
NPAR = 4
ISYM = 0

GGA = OR
LUSE_VDW = .TRUE.
AGGAC = 0.0000
LASPH = .TRUE.

LCHARG = FALSE
LWAVE = FALSE

ICHAIN = 0
LCLIMB = TRUE
IMAGES = 8
SPRING = -5
IBRION = 3
POTIM = 0

IOPT = 3
MAXMOVE = 0.2
TIMESTEP = 0.1

[graeme]

I used less computationally expensive settings and somewhat more conservative optimizer settings to reconverge both the endpoints and the neb. The primary barrier is 0.82 eV, but note that it looks like it will be a two-step process. The high spring constant for the CO2 molecule required a lower time step for the optimizer.

[RonenKru]

Thank you Professor!