I'm doing calculations using Gaussian DFT method. My molecules contain C, O and Cu. I read some papers, they use split basis set for this kind of molecule, and very often they use LANL2DZ for Cu and 6-31G* for others.
Regardless of computational time, the split basis set LANL2DZ + 6-31G* vs LANL2DZ ONLY to all atoms, which calculation should offer a better result?