I use Gaussian 09 and Turbomole to do a same calculation (B3-LYP/6-311G) for a same molecule. But the results are not same, Turbomole shows SCF total energy = -772.16125945927, while Gaussian shows SCF Done E(UB3LYP) = -772.402241773, no UHF energy shows. I use UB3-LYP in Gaussian, while in Turbomole cannot specify that.
I use TmoleX to create the input file for Turbomole. It does not have UB3LYP or RB3LYP in the method part, I can only select DFT > B3LYP as my method. In previous part for molecular attributes, I can select multiplicity: UHF, then generate MOs.
The input file of Gaussian is very simple: UB3LYP/6-311g followed by cartesian coordinates of the molecule.

The energy is not the same in both of these cases. What causes the different results?


1 Answer 1


Answering your question from the title: no, SCF does not necessarily mean HF. Essentially the same self-consistent field (SCF) procedure is used to solve both the Hartree-Fock (HF) equations and the Kohn-Sham (KS) one, so for the case of DFT, SCF energy means not HF energy, but rather KS energy.

Regarding discrepancies in results obtained with different programs, there might be few reasons out there: different definitions of the B3LYP functional, different harmonics used for 6-311G basis sets (spherical vs. cartesian), different integration grids, different SCF convergence criteria, etc.

I checked the TURBOMOLE documentation and indeed found that to use the very same B3LYP functional that is used in Gaussian (with the VWN3 correlation) you need to use b3-lyp_Gaussian in TURBOMOLE, not b3-lyp. So, first of all, I suggest you try b3-lyp_Gaussian in TURBOMOLE instead of b3-lyp.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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