UT theoretical chemistry code forum

Dear Dr.Graeme,
For the reaction of H+H-->H2 over Ni-based catalyst,I have successfully performed cNEB followed by Dimer to get the T.S. with calculation converged to the given setting of -0.01eV. Trying to use dynamic matrix to find imaginary frequecy/ZPE ,multiple imaginarg frequencies are persistent.I have tried to change DISPLACECAR typical displacement.The same problem happened when I tried to perform the vibrational analysis with IBRION=5.
Can you advice please?
Please find attached calculation results

Regards,
A.Omran

You need more images in your NEB calculation to make sure that you have a true saddle point.

Thanks for your reply.
Here is the calculation details
First ,I have started with 6 images using QM and get
0 0.007584 -183.098500 0.000000
1 0.014441 -182.642100 0.456400
2 0.018104 -182.490600 0.607900
3 0.078443 -182.197500 0.901000
4 0.098946 -182.266200 0.832300
5 0.088267 -182.319200 0.779300
6 0.019988 -182.358600 0.739900
7 0.000000 -182.528900 0.569600
then ,i have switched to dimer method to get saddle point until the calculation was converged at -0.01 eV.I attached results for those steps..Still I have to increase it more?

If you look at the trajectory of your dimer calculation, you can see that it moved a long way from the initial position near the climbing image; that geometry does not correspond to the dissociate adsorption mechanism that you are trying to find. The challenge that you are facing is that you have a long pathway which likely involves multiple reaction and diffusion steps. Your NEB calculation may be able to resolve the saddle for dissociate; try continuing that calculation and see if you can get the force closer to 0.01 eV/Ang. You can try another dimer calculation when it is closer to the saddle, but keep and eye on it and abort if the curvature goes positive and it then goes to find some other saddle.

I expect that if you used even more images for your NEB calculation, you would see a more complex reaction mechanism involving intermediate minima, such as you can see in http://henkelmanlab.org/pubs/henkelman06_044706.pdf Figs. 2 and 3. I know that your reaction does not involve subsurface hydrogen, but I expect that you will see a reaction in which H2 dissociates over a single surface atom and then the H atoms will have to diffuse to your specified final state.

Dear Dr.Graeme,
Thank you for advice.
-Totally agree with your suggestion,I set up NEB calculation to continue further to less than -0.05 eV using FIRE.

-Thanks for the tip about stopping reaction if Dimer cross from -ve to +ve curvature .In fact, I have just faced this problem with CH4 dissociation when I use dimer and calculation ran away from expected barrier.Now I do same as with H2 dissociation

Regards,
Omran

Dear Dr.Graeme,

I have continued with neb calculation more close to 0.01 eV Force as follow
0 0.007584 -183.098500 0.000000
1 0.006079 -182.749800 0.348700
2 0.003504 -182.516000 0.582500
3 0.001 476 -182.209900 0.888600
4 0.017795 -182.319100 0.779400
5 0.027094 -182.352200 0.746300
6 0.034287 -182.364400 0.734100
7 0.000000 -182.528900 0.569600

it looks like saddle point is somewhere between Image 2-3,so I switch to dimer.As expected, the dimer calculation did not take time and converge within only 70 step indicating that it was close to T.S.

However,again using both dynamic matrix and vibrational frequencies,the problem of multiple imaginary frequencies persists.Any advice?

I think that you just need to converge more precisely to the saddle. In the dim01 calculation, I reduced the force tolerance to 0.002 eV/Ang. You can see that the geometry moved significantly upon convergence.

Then using the same settings as you, I see a single negative mode at the saddle, in the 0.001a.tar.gz calculation:

838.311276 cm^{-1} ... 1
161.540670 cm^{-1} ... 0
246.541685 cm^{-1} ... 0
353.760720 cm^{-1} ... 0
1631.219923 cm^{-1} ... 0
1880.268508 cm^{-1} ... 0

Dear Dr.Graeme ,

Crystal clear! ..Thank you very much for your help.

Regards,
A.Omran