I want to simulate Raman spectra of benzene in Gaussian 09.

Calculated spectra for a single ring Benzene ($\ce{C6H6}$) in Gaussian differs about $\pu{40 cm^{-1}}$ for each peak, compared with experimental Raman's databases.

How can I solve this problem? Gaussian file created with GaussView6: MP2 config enter image description here

DFT config: enter image description here

Parameters: (for optimization) enter image description here

  • 3
    $\begingroup$ When discussion computational chemistry (and particularly when asking for advice), it's important to say which method(s) and basis sets you are using. Gaussian 09 is a software package with many very different quantum chemistry methods. $\endgroup$ – PJ R Dec 20 '18 at 19:42
  • 1
    $\begingroup$ Did you scale it? Did you compare it to gas phase data? $\endgroup$ – DSVA Dec 20 '18 at 19:57
  • $\begingroup$ @Persian_Gulf PJR is correct. The methods and basis set used are crucial for understanding what wrong with your results. If you could include your input script, that would be helpful in answering the questions. $\endgroup$ – Tyberius Dec 20 '18 at 19:57
  • $\begingroup$ @PJR I tried DFT and MP2. But it still is differs from experimental results. (MP2 is a little better). $\endgroup$ – Persian_Gulf Dec 20 '18 at 20:11
  • 1
    $\begingroup$ Liquid phase is possible at least as implicit solvent models via the scrf keyword. $\endgroup$ – pH13 - Yet another Philipp Dec 21 '18 at 1:59

Thank you for attaching your input file. The first place to start for improving these calculations is your choice of basis set. 6-31G is a rather small basis set and is likely a source of significant error here. If you're fond of using the Pople type basis then you would be better off using something like 6-31++G* so that you include polarization and diffuse functions. With your current basis I suspect that you are missing important polarization of the atoms upon distortion.

If you really wanted a very accurate answer I would recommend the following scheme. Optimize your geometry at the DFT level and then use that geometry as a starting place for a optimization and frequency calculation at a higher level of theory. For example use the cc-pvtz basis set (a correlation consistent basis set that's more efficient generally than the Pople type) and CCSD(T). This will provide a very accurate gas phase raman spectrum.

Edit: The only additional Gaussian specific parameter you should need is the Geom=Checkpoint command for the CCSD(T) optimization and frequency run. This will use your existing checkpoint file as a geometry starting point. The other keywords (basis and method) are exactly what they should be cc-pvdz and ccsd(t). For more information on basis keywords I recommend the users manual.

| improve this answer | |
  • $\begingroup$ Thanks a lot for your answer. You mean I set the optimization (DFT/ccpvtz) and for frequency CCSD. may I ask you to specify all parameters? $\endgroup$ – Persian_Gulf Dec 28 '18 at 9:29
  • $\begingroup$ I added the parameter for optimization below my question. $\endgroup$ – Persian_Gulf Dec 28 '18 at 9:35

Not the answer you're looking for? Browse other questions tagged or ask your own question.