HOMO-LUMO gap

[sca]

Hello,

When I did some DFT calculations for the adsorption of small metal clusters on a metal oxide surface, some of the optimized structures have HOMO higher in energy than LUMO. Even though I used the scforbitalshift option, this order did not change. Has anyone came across a similar situation? I would greatly appreciate any suggestion regarding the seriousness of this problem and how do I correct it!!! - Thanks!