I'm attempting to use Q-Chem 5.0 to replicate CHESHIRE-esque http://cheshirenmr.info/ ab initio calculations of chemical shift values. Unfortunately Gaussian seems to be the predominant tool in this space, and translating from Gaussian16 to Q-Chem is quite a challenge with respect to the solvent model. Currently, just to debug, I am
- Performing geometry optimization in Gaussian
B3LYP/6-31+G(d,p) (gas phase)
- Using those geometries in both Gaussian
mPW1PW91/6-311+G(2d,p) (giao, scrf)and Q-Chem. For Gaussian, the solvent setting is
For step #2, both agree in the gas phase (to within 3 digits!) but in solvent I get very different answers (Q-Chem is off by ~3ppm in 13C vs Gaussian's 2.1ppm for my dataset.) I think this suggests that I am using the Q-Chem solvent model wrong, which is considerably more complex than what Gaussian seems to expose.
For a start, I have been using:
$rem BASIS 6-311G** METHOD mPW1PW91 JOBTYPE NMR SOLVENT_METHOD smd $end $smx solvent trichloromethane $end
in my solvent settings for Q-chem, with the above results. Can anyone recommend a set of Q-chem solvent settings that might yield more accurate chemical shift values and approximate whatever Gaussian's solvent model is doing? I'm also totally open to a different set of best-practices for Q-chem.