For your system with only Al2O3(0001) surface in the supercell, it may correspond to the ideal Al-terminated Al2O3(0001) surface. The ideal surface is usually unstable and the reconstruction could occur.
According to the atom stacking along the [0001] of Al2O3 [you take a look at Figure 1 in Phys. Rev. B 82, 155319(2010)], it may be unlikely to construct surface of Al2O3(0001) with a stoichiometric Al/O atoms at the topmost layer. If you optimize the structure of Al-terminate Al2O3(0001), the Al atoms at topmost layer may move down and such displacement can suppress the surface dipole to stabilize the Al-terminated Al2O3(0001). It may also lead to the suppression of surface states. This is already discussed in a PRB paper:Bing Huang, Qiang Xu, and Su-Huai Wei , Theoretical study of corundum as an ideal gate dielectric material for graphene transistors,
Phys. Rev. B 84, 155406 – Published 7 October 2011,
http://journals.aps.org/prb/abstract/10.1103/PhysRevB.84.155406In the paper of PRB84,155406(2011), the authors considered three different substrates: Al-terminated Al2O(0001) surface after reconstruction, O-terminated Al2O3(0001) surface, and fully hydroxylated Al2O3(0001) surface. You can follow them to conduct your study on the system of monolayer MoS2/Al2O3(0001). Before putting monolayer MoS2 on Al2O3 substrate, you need to perform the geometry optimization for the substrates of Al2O3(0001) surface. When the monolayer MoS2 is put on Al2O3 substrate, you have to choose proper surface area to make the lattice be matched as much as possible. Finally, you need to do the geometry optimization for the whole system (i.e., the interface) of monolayer MoS2/Al2O3(0001).
The increase of vacuum space cannot modify the surface states.