# Gaussian: Warning - assumption of classical behavior for rotation may cause significant error

I am playing around with a prostaglandin molecule (23 heavy atoms) in Gaussian09. My goal is to calculate the vertical ionization potential and electron affinity for the molecule. I already managed to converge the geometry of a neutral molecule (with the invaluable help of @Wildcat and Greg at this forum) using Opt=Tight and Integral(Grid=Ultrafine) as well as Opt(Maxcycle=1000) and SCF(Maxcycle=1000). In the log file I got the long-awaited line "Stationary point found" :) However there are also two warning messages there that make me worry a bit:

1. Warning -- assumption of classical behavior for rotation may cause significant error

2. Warning -- explicit consideration of 37 degrees of freedom as vibrations may cause significant error

Now, the zero-point energy that I am interested in has been reported as:

SCF Done: E(RB3LYP) = -1004.62842050 A.U. after 14 cycles

However, right below the warning messages I found something like:

Zero-point correction= 0.442805 (Hartree/Particle)

Sum of electronic and zero-point Energies= -1004.185615

Now, my specific questions are:

1. Should I worry too much about these warnings if my resultant Ionization Potentials (IP) and Electron Affinities (EA) look reasonable?

2. Should I do my IP and EA calculations taking the Zero-point correction (see above) into account and use the value of -1004.185615 as my "true" zero-point energy?

3. What can I do to eliminate the warning messages mentioned above? I read something about a keyword "freq=HinderedRotor" that could be used in such a case.

I am eager to know how to interpret this whole mess and how to proceed. Thank you for any help you may offer!

• Which basis set are you using? What do you mean by "true"? Maybe you can retain only the electronic contribution – user1420303 Aug 30 '16 at 11:40
• # freq b3lyp/6-31g(d) geom=connectivity scf(maxcycle=1000) integral(grid=ultrafine) – Ricardo Moreno Aug 30 '16 at 12:30
• Well, I mean "true" as closer to real zero-point energy that I want to assess for IP and EA calculation. Sorry if I cannot make myself clear enough. I am new to the field of computational chemistry. – Ricardo Moreno Aug 30 '16 at 12:33
• Given the inaccuracy of the method, I would go for neglecting the vibrational rotational parts. – user1420303 Aug 30 '16 at 12:34
• It is ok. I find hard to justify the usage of these contribution to energy if there are few confidente of their accuracy. Most of they will cancel out. – user1420303 Aug 30 '16 at 12:38