What are some methods for modeling bulk phase infrared spectra?

Question

I am interested in predicting the infrared spectra of bulk phase molecules, it seems that AIMD (ab initio molecular dynamics) is the current best approach. I have found a tutorial on using CP2K to do exactly this, and I have seen a recent publication for computing condensed phase infrared spectra. I'm not entirely sure how different condensed phase and bulk phase spectra would be (I guess this could be a bonus question; What's the difference between bulk phase and condensed phase infrared spectra).
The mentioned above CP2K method seems to be very time consuming, I was wondering if there are other methods for doing these calculations?

mykd · Accepted Answer

Vibrational maps and density of states
In water AIMD community, to avoid excessively heavy calculations of molecular moments, researchers use velocity autocorrelation function (which converge much faster),  obtaining vibrational densities of state and use vibrational maps to scale the VDOS spectra appropriately. Many researchers also simply calculate vibrational/phonon DOS spectra as the partitioning of the total dipole moment into molecular dipole moments is not straightforward. VDOS spectra can also be used to compare and disentangle vibrational contributions from various molecular species in bulk phase.

Tyberius · Answer

I'm not an expert in MD/AIMD software so I can really only discuss methods rather than an actual workflow/software that can carry it out.
One approach that has been explored for accelerating AIMD and electronic structure methods in general is to combine calculations on fragments to approximate an electronic structure. John Herbert's Group is doing work in this avenue, though I think it's too early for an established workflow to exist.
Alternatively, if you are willing to sacrifice some accuracy for speed, you can do classical molecular dynamics. If you have a good force field for your system, you can just compute the overall dipole classically from the charges and distances, getting the IR spectrum from the Fourier transformed time correlation.
A third route would be to use classical MD to get snapshots and then obtain the IR from an extrapolation/Boltzmann weighting of electronic structure calculations on these snapshots. I know of this being done for UV/Vis and I have found at least one paper that takes this sort of approach.

What are some methods for modeling bulk phase infrared spectra?

2 Answers

Add your own answers!

Ask a Question