Pre-Exponential factor calculation

[sapper1]

Dear admin, Proff. Graeme,
I was wondering how can we calculate pre-exponential factor after calculating activation barrier using CI-NEB?
thank you for your help in advance.
best

[kambiz]

It has been described in many kinetics books, such as http://www.amazon.com/Concepts-Modern-C ... 3527316728.

Cheers
Kambiz

[graeme]

There is some practical information here as well: http://theory.cm.utexas.edu/vtsttools/dynmat/

[hoy]

Dear Prof. Graeme, Admin,
I have few questions regarding calculations of a prefactor.

I am calculating a prefactor for bulk Ag (31 atoms + 1 vacancy) using VASP5.3.2+VTST3.0a.
According to the manual (http://theory.cm.utexas.edu/vtsttools/dynmat/) I should make a DISPLACECAR file for the initial state (and also for the transition state). Then I have to run VASP, compiled together with VTST, and get results in file freq.dat (using dymmatrix.pl). File freq.dat contains information about vibration frequencies. To simplify task, I have performed the calculations of vibrational modes of only one atom in the initial state. I have tried to change a lot of parameters in different combinations but the result of the calculations is always same: I get 3 vibrational frequencies: one is real and the two of thous are imaginary.

Could you please help me to find a solution of my problem? I compare my calculations with similar calculations of my colleague, who done such calculations before using VASP 4.6 and get correct results, but now have similar problem with VASP5.3.2+VTST3.0a.

I am attaching INCAR, POSCAR, KPOINTS, OUTCAR and DISPLACECAR files

Thank you.

[graeme]

I believe the main problem is that you have not set:
ISYM = 0
The imposed symmetry has prevented relaxation of atoms around your vacancy (you can see that they remain in crystal site), and gives you symmeterized forces in your dynamical matrix calculation. I'm re-running with ISYM=0 now to verify this.

Also, as a side note, it is a much better approach to start with inexpensive calculation settings and then look for convergence of the properties that you care about (e.g. normal mode frequencies) with respect to the calculation settings. So, for example, initial calculations with between a 2x2x2 and 4x4x4 k-point mesh makes much more sense than doing this debugging with a 12x12x12 mesh (which is 261 times more expensive than a 2x2x2 mesh!).

[graeme]

Yes, for the 2x2x2 mesh calculation, the modes look very reasonable (see below). The three zero modes are obviously translation.

--------

Num Displacements: 93
Total atoms: 31

1.917349 cm^{-1} ... 1
1.359982 cm^{-1} ... 1
1.358480 cm^{-1} ... 1
54.426643 cm^{-1} ... 0
54.435537 cm^{-1} ... 0
55.664824 cm^{-1} ... 0
56.617363 cm^{-1} ... 0
56.619824 cm^{-1} ... 0
63.505013 cm^{-1} ... 0
...
150.353407 cm^{-1} ... 0
157.485083 cm^{-1} ... 0
157.490534 cm^{-1} ... 0
157.507011 cm^{-1} ... 0
158.284391 cm^{-1} ... 0


--------

[graeme]

Turning off symmetry also gives reasonable modes for your single atom test (with the 12x12x12 kpoint mesh):

91.844691 cm^{-1} ... 0
106.024946 cm^{-1} ... 0
106.380726 cm^{-1} ... 0

[hoy]

Dear Prof. Graeme

Thank you for your answer and advises.
All calculations go good.

Thank you.