Dear colleagues,
I have a question about TD-DFT calculations of absorption spectra. A similar question was posted before (https://www.researchgate.net/post/I_am_trying_to_calculate_absorption_spectra_using_tddft_but_it_is_giving_lambdamax320nm_but_exp_410_nm), but without a reply that would work for me.
I use Firefly (formerly PC/Gamess) for TD-DFT calculations of organic molecules, this program although maybe not the most popular, implements many functionals and features, and it is free for researchers. For one molecule, I get excitation energies like that
------------------------------------------------------------------------------
TDDFT EXCITATION ENERGIES
STATE HARTREE EV KCAL/MOL CM-1 NANOMETERS OSC. STR.
------------------------------------------------------------------------------
1B1 0.1071552650 2.9158 67.2410 23517.86 425.21 0.0487557
1A1 0.1152841696 3.1370 72.3419 25301.95 395.23 0.0169178
1B1 0.1243385284 3.3834 78.0236 27289.15 366.45 0.4694982
1A1 0.1372843976 3.7357 86.1473 30130.44 331.89 0.0040060
so the most intense transition as calculated is at 366 nm. Actual value from experiment is 487 nm. This is measured in a gas phase, so solvation is not an issue.
Changes to optimization methods (basis sets, MP2 vs DFT with different functionals), and choice of different DFT functionals for TDDFT gives a difference of a few nm, but all results are some 100 nm blue of the experimental value.
Looks like something is fundamentally wrong with my calculation, but I cannot figure out what, could anyone please give me ideas?
That's the same issue as in this discussion:
https://www.researchgate.net/post/Can_anyone_help_me_to_find_a_suitable_basis_set_for_energy_calculation_of_Iodine_molecule
Here's the deal: TDDFT is not a chemically precise method so there is no "doing the calculation right" that would result in the exact spectrum. What it can be used for: calculating relative shifts (e.g. substituend effects...) to get an idea how a spectrum would change in comparison with a known one.
Jürgen Weippert, thank you for your input, and for referencing a useful discussion. I am in synthetic chemistry and dyes, being pretty ignorant in quantum chemistry. I did some thermodynamic calculations for strained cyclic olefins before, and was very impressed how close my calculated values of hydrogenation enthalpies matched literature data. This was calculated with little knowledge and not much effort after some experiments with calculation methods and basis sets. So the situation is different for UV spectra. We have spectra for some volatile molecules in gas phase, and I expected TDDFT to give absorption maxima within say 10 nm.
If having calculated absorptions 120 nm away is still publishable, I could probably make a relative comparison between different compounds in my calculations (hope they reproduce the trends), although matching absolute value of course would be better...
Yes, hydrogenation enthalpies are much easier to get accurate because they don't involve excited states. TDDFT quite often gets more praise than it probably deserves but at a certain (rather small) molecule size you don't have any other chance.
I found that the DFT functional plays an important role in stimulating the UV Vis spectra. Try one with good Grimms dispersion like B97D3. Also you need to use a good basis set.
For transition metal complexes, playing with the effective potential could help too
Ismail Badran thank you for your reply. How close you think the absorption maxima should be to experiment for usual organic molecules, when the functionals and basis sets are good?
Thank you , I will say within +- 20 nm, the closer it is the better.
Also you need to use salvation and choose the accurate solvent.
Also try to probe more transition states, like I use N=40 or even more.
- Badges
- Science topic