I am performing some TD-DFT calculations to reproduce the UV-Vis spectrum of a molecule using Gaussian. After optimizing the structure, I performed a TD-DFT calculation with 150 states and I tried different basis sets and functionals (such as B3LYP, CAMB3LYP, M062X).

However, in all cases, I fail to reproduce the spectra (in terms of numbers of peaks, not only the position) and it seems very far from the experiments.

I wouldn't expect the solvent to change the resolution of the peaks. Does anyone have some tips on how to enhance these types of calculations?

  • 2
    $\begingroup$ Without any specifics about the problem, it is hard to give any advice. First, I would recommend reviewing the literature about the given material family to set expectations. The number of peaks is a vague characteristic of a spectrum, often miscalculated by experimentalists, and it is not clear how much the calculated energy is off. In your calculation, there might be effects missing (Frank-Condon, excitons, etc), or you did not consider conformers, or chemical species in your calculations. $\endgroup$
    – Greg
    Jun 9 at 10:24
  • $\begingroup$ The calculation considers only light atoms (C,B,N,H) and in most of the literature the 6-311*g or 6-311*g+ is used to obtain comparable UV-Vis spectra to experiments. I also used a lot of states to ensure convergence and calculated the TD-DFT with: # td(nstates=150) b3lyp/6-311g* $\endgroup$
    – Laura
    Jun 9 at 15:35
  • $\begingroup$ @Laura I gave my +1 long ago, but since it has been almost 6 months, is there any update you can provide us? Are you still urgently or actively in need of an answer to this question? Please let us know! Also, Greg might not have seen your response to his comment because you didn't use the @ character to ping him. $\endgroup$ Dec 1 at 22:04
  • $\begingroup$ Unfortunately, I didn't find an answer. However, I don't need one urgently. $\endgroup$
    – Laura
    Dec 2 at 18:01


Browse other questions tagged .