These transition state calculations look good; both the NEB and dimer calculation.
The dimer, however, found a significantly lower energy saddle than in the assumed mechanism in your NEB calculation. This is a positive aspect of the dimer method - it can show you saddles that you did not anticipate. To understand the reaction mechanism, make a small displacement in a positive and negative directly along the negative mode at the saddle from your dimer calculation, and then minimize to find the initial and final states. We have a small script called 'dimmins.pl' to set up these minimization calculations.
Transition state calculation for wealy bound molecules
Moderator: moderators
-
- Posts: 28
- Joined: Fri Aug 03, 2018 7:06 am
Re: Transition state calculation for weakly bound molecules
Dear Prof.Graeme,
Extending my gratitude for your kind support,I have gone for both ways to confirm the reaction mechanism.
Initially,I confirmed that I got the right saddle point by dynamic matrix which resulted in only 1 imaginary vibrations @848.34 cm-1 .Then from that T.S.:
(a)Dimer:Using dimmins.pl with 0.001 displacement and try to minimize both initial and final states in the mins subfolder.
However,I have doubt about how much displacement is appropriate in that case,does it affect the final result?
Another concern about the path from initial state to saddle point (min1) and if that will consider the dissociative adsorption step (CH4-->CH3)?
(b)cNEB:Can you please confirm that I got exactly what you explained before?
1-Minimize the intermediate T.S. (04 structure which has been confirmed as saddle point)
In both below cases I go for QM until force <0.1 then switch to IOPT=1 with default setting (POTIM=0)
(1)CH4-->CH3+H (intermediate):In progress and I use 6 images in between.
(2)CH3+H(intermediate)-->CH3+H(chosen final state) :In progress and I used 6 images in between too.
Thank you
Regards,
Omran
Extending my gratitude for your kind support,I have gone for both ways to confirm the reaction mechanism.
Initially,I confirmed that I got the right saddle point by dynamic matrix which resulted in only 1 imaginary vibrations @848.34 cm-1 .Then from that T.S.:
(a)Dimer:Using dimmins.pl with 0.001 displacement and try to minimize both initial and final states in the mins subfolder.
However,I have doubt about how much displacement is appropriate in that case,does it affect the final result?
Another concern about the path from initial state to saddle point (min1) and if that will consider the dissociative adsorption step (CH4-->CH3)?
(b)cNEB:Can you please confirm that I got exactly what you explained before?
1-Minimize the intermediate T.S. (04 structure which has been confirmed as saddle point)
In both below cases I go for QM until force <0.1 then switch to IOPT=1 with default setting (POTIM=0)
(1)CH4-->CH3+H (intermediate):In progress and I use 6 images in between.
(2)CH3+H(intermediate)-->CH3+H(chosen final state) :In progress and I used 6 images in between too.
Thank you
Regards,
Omran
Re: Transition state calculation for wealy bound molecules
Omran,
The "negative" barrier is likely because what you have shown is not an elementary step. As Prof. Henkelman pointed out, your initial and final states do not involve H-abstraction at the same proposed active site. What I imagine happened is that the transition state you found is best represented as a diffusive barrier, which is why it appears "negative" -- the highly exothermic C-H activation has already taken place prior to the transition state you observed. Refer to Fig. 5 in Ref. [1] below.
To clarify this, look at the snapshots below. It looks like you were anticipating the H atom to get adsorbed between Ni3, Ni6, and Cu3. However, in the final state, it is adsorbed between Ni3, Ni4, and Cu3 instead. You can see from this set of images that the process involves both H-abstraction and subsequent diffusion of the H* species. I would get a more appropriate final state and then rerun your set of calculations, unfortunately starting from the coarse CI-NEB calculation again. If you use dimmins.pl on your dimer calculated TS, I am fairly confident you would not get back your initial and final states that you propose, indicating that the TS does not directly connect your proposed local minima. Also, on a potentially related note -- is there a more stable initial state with CH4 adsorbed closer to the surface near the proposed active site? It seems a bit far away from the surface, but perhaps it's just very weakly physisorbed to the metal surface.
[attachment=0]1.png[/attachment]
[attachment=1]2.png[/attachment]
The following two papers may help clarify what I'm talking about:
[1] https://pubs.rsc.org/en/content/article ... c8cp03191f
[2] https://pubs.rsc.org/en/content/article ... c6cp08003k
The "negative" barrier is likely because what you have shown is not an elementary step. As Prof. Henkelman pointed out, your initial and final states do not involve H-abstraction at the same proposed active site. What I imagine happened is that the transition state you found is best represented as a diffusive barrier, which is why it appears "negative" -- the highly exothermic C-H activation has already taken place prior to the transition state you observed. Refer to Fig. 5 in Ref. [1] below.
To clarify this, look at the snapshots below. It looks like you were anticipating the H atom to get adsorbed between Ni3, Ni6, and Cu3. However, in the final state, it is adsorbed between Ni3, Ni4, and Cu3 instead. You can see from this set of images that the process involves both H-abstraction and subsequent diffusion of the H* species. I would get a more appropriate final state and then rerun your set of calculations, unfortunately starting from the coarse CI-NEB calculation again. If you use dimmins.pl on your dimer calculated TS, I am fairly confident you would not get back your initial and final states that you propose, indicating that the TS does not directly connect your proposed local minima. Also, on a potentially related note -- is there a more stable initial state with CH4 adsorbed closer to the surface near the proposed active site? It seems a bit far away from the surface, but perhaps it's just very weakly physisorbed to the metal surface.
[attachment=0]1.png[/attachment]
[attachment=1]2.png[/attachment]
The following two papers may help clarify what I'm talking about:
[1] https://pubs.rsc.org/en/content/article ... c8cp03191f
[2] https://pubs.rsc.org/en/content/article ... c6cp08003k
- Attachments
-
- 1.png (338.44 KiB) Viewed 14530 times
-
- 2.png (337.58 KiB) Viewed 14530 times
-
- Posts: 28
- Joined: Fri Aug 03, 2018 7:06 am
Re: Transition state calculation for weakly bound molecules
Hi Andrew,
Thank you very much for your insights and detail.I have already tried dimmin.pl with the 0.1 displacement and with now change in the result still same with "negative" barrier .I will give another chance with 0.2/0.3 displacements to see if any difference.
Meanwhile,I will try a new starting point and look for more appropriate final state , repeat cNEB again.In fact,I believe that the problem is more related to initial state rather than final state.Unfortunately,this is almost closer initial state to surface as it is very weakly physical adsorption (-.024) as you anticipated.
Thank you very much for your insights and detail.I have already tried dimmin.pl with the 0.1 displacement and with now change in the result still same with "negative" barrier .I will give another chance with 0.2/0.3 displacements to see if any difference.
Meanwhile,I will try a new starting point and look for more appropriate final state , repeat cNEB again.In fact,I believe that the problem is more related to initial state rather than final state.Unfortunately,this is almost closer initial state to surface as it is very weakly physical adsorption (-.024) as you anticipated.