Dear Verma,

If you use Opt Freq option, it will give you both energy and ZPVE values. You dont need to do separate calculation for SPE as well Freq calculation.

Dear Sai

Correct me if I'm wrong.

I know opt freq calculation and we calculate SPE for better prediction of energy. Generally we do opt+freq at medium level of theory and calculate the SPE relatively at large level of theory to predict an accurate energy. I also know that freq calculation gives ZPVE values and this ZPVE value should be added to the optimization energy. But my question was that: 

Do we require to add the ZPVE values to SPE values also ? or not?

Since u r considering the higher level theory for your molecule, you need to do freq calculation at the same higher level theory and ZPVE value needs to be added to the SPE value (which you calculated at high-level theory).

ZPE, as it is calculated by Gaussian by obtaining normal modes from quadratic expansion around the stationary point, is only physically meaningfull for stationary points (minima and transition states). There are methods to obtain vibrational corrections for each point on the potetnial energy surface, but these are quite complicated, very rarely used and only for small systems (and Gaussian can't do that - if you're interested, see e.g. http://dx.doi.org/10.1063/1.3052076).

So to answer the questions: ZPE should not be added to SPE, unless it's a stationary point (zero or one imaginary frequency).

Thank you all and....

Dear Bartosz

Thanks for reply.

But I'd like to ask you some extra things:

1) If I want to add the ZPE to SPE for stationary points then I should have ZPEs at the SPE level of theory but some high level of theories don't offer freq calculations. Ideally freq analysis should be done on same level of theory. Isn't it? So in this case what should I do?

2) I have done few calculations without adding ZPEs to SPE (@ CCSD) and I have asked to the Gaussian guys, they told that it's upto you whether you want to add or not but you'll have to be consistent and mention that ZPEs are not added in your article. What do you think of this?

First, let me clarify my previous post: to be consistent you either add ZPE to all energies, or to none of them. From practical point of view, if you calculate energy barriers and get stationary points (reactants, ts, products) you should add ZPE to all of them. But if you do PES scan with a lot of points, you should not add ZPE to any of them to be consistent.

To answer the questions:

1. You can add ZPE at a different level of theory, which is not perfect, but commonly used. In the case of CCSD the most common way is to add ZPE calculated at MP2/large basis-set, see e.g. http://www.rsc.org/suppdata/cp/b8/b823358f/b823358f.pdf. I rarely do such high-level calculation, but are commonly reporting energies as E(M06/6-311++G**) + ZPE(M06/6-31G**), so getting energies and ZPE's at different basis set.

2. You definitely need to be consistent and mention all the details in the manuscript. 

The thing with ZPE's is that for e.g. products and substrates of the same system they're usually quite similar, since you have the same number of identical atoms (just connected in a different way). So I can see how omitting them is justified. I, personally, would however do them when preparing paper-quality data, just to make the reviewers happy. If you can afford CCSD, you can definitely afford a single-point freq calculation with MP2 to get them.