The interface tool will not work for this - it's for creating a flat interface. I would first use the Nanowire tool (not Wulff, although they are related, sort of) for Si. Then I would make a similar nanowire of SiO2, but after it's created I would cut out the middle to fit the Si part, and merge the configurations by simply dropping the Si nanowire stash item onto the SiO2 when it's active in the Builder.
Not entirely trivial - and probably not very realistic, since the Si-O bonds at the surface will be rather arbitrary.
A simpler choice, to begin with at least, would be to just represent the SiO2 with a dielectric region. Then at least the effects of the dielectric on the confinement of the wavefunction (compared to vacuum) could be included.
If a real model of SiO2 on Si is desired, one may need to perform some simulated growth, using MD. That need not be too difficult since there are fairly good Tersoff potentials for Si-SiO2, and which are included in ATK. Or, one can build a crude model and run an MD simulation on it to let it relax.
