Frozen Phonon calculations for 3D periodic surface

I am doing some vibrational calculations on the hematite 0001 surface and not sure of the correct way to set these up. The system I have set up is a 3D periodic surface in a 2x2 supercell with defects on the surface. Originally by looking at the way the previous participant on this project had done his vibrational calculations, it seemed that the vibrations were set up to be calculated for only one face of the system (i.e. top or bottom face). When I finally got the rest of his files, I found that he had calculated most of the systems by moving the atoms on both surfaces (i.e. top and bottom faces). Looking into it further, the results came with the same number of normal modes regardless of which way the calculation was set up and judging from the .outmol files it seems that DMol knows to move the symmetry equivalent atoms. But when I calculate the gibbs energy of vibration for these systems, the results come out significantly different. As an example, for a 25% defective surface the previous students calculation found -2.006 eV/A and my calculation found -1.505 eV/A. I should note that while the two surfaces for these systems are not the exact same, they are stoichiometric (the difference comes in the arrangement of hydrogens on the terminating oxygens, his surface had water ligands mixed with hydroxyl ligands, my surface had only hydroxyl ligands, but this should not lead to such a significant change in gibbs energy should it?). Next I decided I would set up a vibrational calculation done for his surface but using my method of vibrational calculation to see if I could determine where the difference was coming from, but this resulted in the first 6 or 7 normal modes being negative. Why would his calculation result in real eigenvalues while mine done on the same structure resulted in negative eigenvalues?

My main question though is, which method is correct, or should it not matter that his calculations and mine are set up differently due to symmetry? Could such a large difference be due to the arrangement of waters and hydroxyls?