I have a fundamental question about the definition of "stretching mode" obtained from molecular dynamics simulations. Is it the fourier transform of autocorrelation of bond lengths over a time period of simulation or is it the fourier transform of the velocities of the two atoms I am interested in? How does the two vary in terms of degrees of freedom? Any textbook references are highly appreciated. I am looking at rutile-TiO2 test system and interested in Ti-O bond stretches, Ti here is octahedrally bonded.
1 Answer
$\begingroup$
$\endgroup$
1
I have a fundamental question about the definition of "stretching mode" obtained from molecular dynamics simulations. Is it the fourier transform of autocorrelation of bond lengths over a time period of simulation or is it the fourier transform of the velocities of the two atoms I am interested in?
The vibrational density of states (absolute square of the Fourier-transformed velocities) will give you a power spectrum of all phonons that appear in your simulation. You will see a lot of peaks and it will not be obvious which ones are related to which mode.
The Fourier transform of the autocorrelation function of the bond lengths will likely result in a similar spectrum, maybe with fewer peaks, but it will be still difficult to learn something about a single mode. The reason is that you are interested in Ti-O bond streching modes, but in fact there will be multiple modes, that stretch some of the Ti-O bonds.
What you might actually be looking for normal mode analysis, which can also be done with many MD codes.
-
$\begingroup$ thank you very much for the explanation $\endgroup$ Commented Nov 12, 2019 at 18:20