You have to pick a same reference system for both methods. If you are looking at conformers, choose the ground state conformer and run all your various conformers against it, and report everything as relative energies. Remember that absolute energies don't mean anything in computational chemistry 

Yes that's correct, I am only using relative energies as my basis of comparison. I would like to compare the relative energies with the thermal corrections.

Hi Patrick,

From your comment, I think what you are looking for is the zero-point and enthalpy corrections for MP2(singlepoint)/WB97XD(geometry) or MP2//WB97XD.

Instead of running a single point calculation using MP2, you can run a frequency calculation which not only gives you the electronic energy for the single point but also the frequency analysis followed by zero-point corrections on the same level of theory that you calculate the single point energy.

Although, you might say that running a frequency calculation on a geometry that is not optimized with the MP2 is not realistic. However, think of the assumptions that the frequency analysis are based on. For instance, from Gaussian website:

"One of the most important approximations to be aware of throughout this analysis is that all the equations assume non-interacting particles and therefore apply only to an ideal gas. This limitation will introduce some error, depending on the extent that any system being studied is non-ideal. Further, for the electronic contributions, it is assumed that the first and higher excited states are entirely inaccessible. This approximation is generally not troublesome, but can introduce some error for systems with low lying electronic excited states."

http://www.gaussian.com/g_whitepap/thermo.htm

So, the frequency analysis are not very accurate themselves. It depends on your system and how important those corrections are to you.

You can also compare the relative pure electronic energies to those relative corrected energies to make sure that these corrections does not affect the trends.

Hope this is helpful

Fatemeh

Hi Patrick,

common practice is to apply total dG corrections from the level of theory at which the molecule was optimised, here DFT. Such approach is based on observation that unless your two methods would produce terribly different geometries, total correction will be similar. In principle you can follow the idea proposed by Fatemeh but here you should be careful as low frequencies might appear (and I expect them along with a set of imaginary frequencies that are ignored in dG corrections). In such cases within rigid-rotor harmonic oscillator approximation the vibrational enthalpy approaches a constant factor of RT while the vibrational entropy goes towards infinity with decreasing frequency. So my sugestion would be to stay at DFT correction or do full MP2 optimisation that should not be terribly expensive with modern codes that emply RI approximation.

Cheers,

Adam

As Adam mentioned, it is always good to make sure that there are no imaginary frequencies in a geometry optimized using another method. I remember that for my system there were no imaginary frequencies for MP2 calculations on B3LYP geometries. 

Bests

Fatemeh

Such case, no imaginary frequencies for MP2 calculations on B3LYP geometries, would simply mean that MP2 minimum and B3LYP minimum are (almost) identical (within given precision). However, quick look into frequency scaling database:

http://comp.chem.umn.edu/freqscale/version3b1.htm

says that MP2 scaling factors are about 0.1-0.2 smaller than corresponding B3LYP frequencies. That simply says that, statistically, unscalled MP2 corrections will be almost negligibly larger than B3LYP.

Given the computational expense of running an MP2 optimisation calculation over a medium-to-large sized compound, would it prove acceptable to run a single point energy calculation using MP2 with a smaller basis set, 6-31+g(d,p), over a structure previously optimised using wb97xd/6-311++g(2df,2p). From my understanding, it seems like most SPE calculations are performed with a higher basis set than that used for the optimisation.