Home > Community > Frequency Calculation in Gaussian?
Upvote

25

Downvote
+ Gaussian
+ Thermochemistry
Posted by
Marlon Izzo

Frequency Calculation in Gaussian?

Chris Conrad  Follow

Of course you can, but probably you will find one or more negative eigenvalue(s) for the frequencies. They have no physical meaning, but the number of them may tell you where your structure is on the PES. One negative eigenvalue means a first order saddle point, two negative eigenvalues mean a second order saddle point and so on

More

Upvote

VOTE

Downvote
Brian Gregory  Follow

Hi,
Yes, it is possible. But keep in mind that you are using the harmonic approximation, so you expand the Taylor serie in the minimum, where the first derivative of the potential is zero.
Try this: if you do a frequency calculation of the H2 molecule in the minimum (optimized geometry) and another with different H-H distance (where the first derivative is not zero) you will obtain different vibrational levels (which would depend only on the potential -> wrong result). Since the thermochemistry depends on the vibrational partition function, a bad vibrational spectrum will give you bad thermochemical values.
So I would not rely on a frequency calculation if the structure is not optimized... but you can compare the electronic energies.
Hope it helps.
Fernando

More

Upvote

VOTE

Downvote
Dmitry Dementiev  Follow

Of course you can, but probably you will find one or more negative eigenvalue(s) for the frequencies. They have no physical meaning, but the number of them may tell you where your structure is on the PES. One negative eigenvalue means a first order saddle point, two negative eigenvalues mean a second order saddle point and so on

More

Upvote

VOTE

Downvote
Brian White  Follow

It's not a good idea to do a freq calculation at a non-optimized geometry. Gaussian may compute the matrix of 2nd derivatives correctly, but the algorithm for converting these to vibrational frequencies relies on the 1st derivatives (forces on atoms) being zero.  So whatever frequencies you get at a non-stationary point are pretty much garbage.
The sign of the frequencies has nothing to do with the quality of the frequencies.  Instead, 1 negative frequency implies a transition state, no negative frequencies implies a minimum (optimized geometry), and more than 1 negative frequency implies some other type of stationary point which is usually not of chemical interest.
What you CAN do is copy the checkpoint file (filename.chk) and do an opt freq job on the copy, possibly at a lower or higher level of theory than the original.  The copy will have a changed geometry of course, but you still have the original checkpoint file with the original geometry.

More

Upvote

VOTE

Downvote
Bunny97  Follow

i optimised a structure to its saddle point. using the last point structure from the log file. can i do freq calculation to find the negative frequency? will it work? what other ways are possible to find negative frequency from the log file of an optimisation log file? @david shobe @Fernando Aguilar-Galindo

More

Upvote

VOTE

Downvote
David Cunningham  Follow
Upvote

VOTE

Downvote
Chief Thunder  Follow

Hi,
Yes, it is possible. But keep in mind that you are using the harmonic approximation, so you expand the Taylor serie in the minimum, where the first derivative of the potential is zero.
Try this: if you do a frequency calculation of the H2 molecule in the minimum (optimized geometry) and another with different H-H distance (where the first derivative is not zero) you will obtain different vibrational levels (which would depend only on the potential -> wrong result). Since the thermochemistry depends on the vibrational partition function, a bad vibrational spectrum will give you bad thermochemical values.
So I would not rely on a frequency calculation if the structure is not optimized... but you can compare the electronic energies.
Hope it helps.
Fernando

More

Upvote

VOTE

Downvote