I tried b3lyp/6-31g(d) which is approx. 100nm red shifted from experimental UV-vis absorption. cam-b3lyp/6-31g(d) is about 100nm blue shifted.wb97xd/6-31g(d) is further blue shifted to 150-200 nm. Can we guess some basis set based on these data or there is only hit and trial base method only?
B3LYP/6-31+G** geometry optimization in solvent (PCM) + TDDFT computations at PBE1PBE/6-311+G(2d,p) level of theory (+PCM!)....
The main problem with DFT (and to an extent TD-DFT) is that there is no systematic way to calculate almost any property in different substrates. I mean, a functional which gives a nearly perfect fit with the experimental result for a given molecule could be a disaster for a different one. And (answering your question) the relation between the basis set and the result is not as critical as the relation between the functional and the result.
So yes, you have to do some benchmark to look for the best functional for your systems. I agree with Silvia in the method for the optimization (you can even optimize with no solvent model if your system is very large and the solvent is non polar), for the TD-DFT calculations use at least a triple-zeta basis set (I particularly like the aug-cc-pVTZ, but a split-valence set with polarization and diffuse functions may work well too) and test as many functionals as you can. Hybrid functionals (B3LYP, the M06 family...), some long-range corrected (like CAM-B3LYP or LC-wPBE)... look in the bibliography for others (in my case BP86 works flawlessly, but remember what I said in first paragraph!)
Hope you find it helpful
What's the purpose of the computation?
If you just want to reproduce a number, that you got from an experiment try as many functional and basis set combinations as you like. There will surely be something that fits more or less. However, the scientific value gained is close to zero.
There are a ton of different obstacles, that you have to face when you try to model experimental spectroscopic data especially if the experiment is conducted in solution. Many obstacles means many error sources, consequently this topic is the perfect playground for error cancellation.
So, to give any meaningful answer to your question, you need to state the purpose of your computation. What insight are you hoping for?
- Badges
- Science method
- Similar topics
- Methodology
- Absorption Spectroscopy
- UV-Visible Spectroscopy