Question about dynamical matrix calculation

[jianmin]

Dear guys,

To learn how to get the prefactor of a reaction by a dynamical matrix calculation, I have built a model of H difusion on Pd (111) surface. After performing two separate dynamical calculations for both initial state and transition state, I issued commands dymmatrix.pl DISPLACECAR OUTCAR to get the two freq.dat files for dymprefactor.pl.

The freq.dat file from initial state looks like:
44.264242 cm^{-1} ... 1
799.984823 cm^{-1} ... 0
1012.129608 cm^{-1} ... 0

The freq.dat file from transition state looks like:
288.652000 cm^{-1} ... 1
0.000013 cm^{-1} ... 0
977.836375 cm^{-1} ... 0

Then I issued the command dymprefactor.pl [freq.dat of initial state] [freq.dat of transition state]. And I received a warning:
Initial state freq.dat has imaginary frequencies.

What should I do to solve this problem and get the value of prefactor?

Thanks a lot.

[graeme]

The minimum should have only positive modes. The negative mode has a small magnitude -- is it possible that you have no frozen atoms so that the system can translate? If yes, than you can remove the near-zero modes from both the minimum and saddle point freq.dat files and recalculate the prefactor.

If you have frozen atoms, or have not included all the atoms in your dynamical matrix calculation and you still have a negative mode in the minimum, there is a problem with that calculation. Make sure that you have accurate forces (ediff=1e-7) and systematically reduce the displacement size until you see convergence in the modes. Also, make sure that the minimum is well minimized, for example with forces on the order of 0.005 eV/Ang.

[jianmin]

Hi Prof. Graeme, I setup a Pd(111) slab with 12 atoms (p(2x2) by 3 layers) and freeze all 12 Pd atoms and only relax the H atom. Is that fine with the calculation?

[leon_qian]

Dear sir'
After producing DISPLACECAR, I tried to issue commands "dymmatrix.pl DISPLACECAR OUTCAR", but it failed to proceed, and I obtained an messege "Can't locate Math/MatrixReal.pm in @INC". What's wrong with my calculation?
Many Thanks!

[graeme]

Make sure that you download our entire package of scripts:
http://theory.cm.utexas.edu/vtsttools/scripts/
They work together.