[NEB] Energy barrier depends on number of images

[Soon Ho Kwon]

Hi.

I'm calculating the energy barrier of CO oxidation.
Although I found some posts about the appropriate number of images (3~5 or 4, 6, 8), I have a question.
My calculation shows that the energy barrier depends on the number of images. (I attached a graph about it)

I'm wondering that 4 images are enough or 9 to 10 images are required.
or is there any criteria on determination of number of images? (like, 0.01eV/image)

thanks.

[graeme]

I suggest minimizing the intermediate configuration and run two separate bands of 3-5 images for each elementary step.

[Soon Ho Kwon]

[quote="graeme"]I suggest minimizing the intermediate configuration and run two separate bands of 3-5 images for each elementary step.[/quote]


Thanks for the quick reply.

But did you understood that the attached figure (and my question) is about energy barrier vs. reaction coordinate?
Actually the reaction is quite simple. O-O-C-O* -> O* + CO2.
So there is a only one energy barrier (one transition state).
The question was about energy barrier depending on "number of images" in the simple reaction.

Or did I misunderstand...? If it is, can you explain more in detail...?
Thanks.

[graeme]

I'm looking at the energy pathway that you posted. You may view this as a single mechanism, but there is clearly an intermediate minimum along the path. I suggest minimizing that image (the 5th point in your plot) to see what minimum that corresponds to. Then, calculate separate paths for each elementary step.

[Soon Ho Kwon]

[quote="graeme"]I'm looking at the energy pathway that you posted. You may view this as a single mechanism, but there is clearly an intermediate minimum along the path. I suggest minimizing that image (the 5th point in your plot) to see what minimum that corresponds to. Then, calculate separate paths for each elementary step.[/quote]


Thank you for the response. And I'm sorry that I questioned vaguely.

The x-axis in the figure is the number of images used in INCAR file. So the figure I attached is not energy pathway. when you closely look at the y-axis, it is energy "barrier", not energy.
So, what I did is calculating energy barrier of same reaction with varying IMAGE-tag in INCAR.

Now, I can question again.
I wonder whether 4 images are enough or 9 to 10 images are required in my calculation.
or is there any criteria on determination of number of images in INCAR (IMAGE-tag)?

thanks.

[graeme]

Oh I see. You should use the climbing image method so that a single image will converge right to the saddle.

[Soon Ho Kwon]

[quote="graeme"]Oh I see. You should use the climbing image method so that a single image will converge right to the saddle.[/quote]


Dear Dr. Henkelman,

Thanks for your response, again.

The thing is that I already used the climbing image method as you can see the OUTCAR file.

OPT: Using VASP Quick-Min optimizer
CHAIN: Read ICHAIN 0
CHAIN: Running the NEB
NEB: SPRING -5.000000
NEB: LCLIMB T
NEB: LTANGENTOLD F
NEB: LDNEB F
NEB: LDNEBORG F
NEB: LNEBCELL F
NEB: EFIRST 0.000000
NEB: ELAST 0.000000

So, I was kind of confused that if the clmbing image method can find the saddle point, why such results can occur.

[graeme]

If you could post two of these calculations which give different barriers, I should be able to see what's going on.

[berna_akgenc]

Dear Soon Ho Kwan and all,

I have same problem like you, I would like to share my results as well. I tried 3 different number of images; 5,10,20, respectively. If you focus my graphs, you can understand so easily.

I am confused that is there any limitation of number of images?

Many Thanks!

[graeme]

When you have a set of connected minima and saddles you are finished. By connected I mean that a steepest descent path along the negative mode (in both directions) from each saddles gives you a connected path from your initial state to final state. In your case, it appears that 5 is too few and 10 is sufficient, but the minimization check that I mentioned would help you be confident of that.