The most of my time at the university I spent on performing calculation ( and experiments ) on ZnO wires, so hopefull
my suggestion can bring you further.
In general ZnO is semi-conductor, and hence ZnO wires are semi-conductors. The carrier confinement is strong in ZnO,
but the confinement of excitations are stronger, and hence the band gap remains more of less constant depending on size.
However calculation on ZnO shows that they are metal, as you calculation do. I would like to draw your attention to the two bands
located around (at Gamma) in the area between 0.0 - 0.3 eV. These two bands are surface bands, and is effect of the termination of
the crystal. You can identify these as being surfaces in three ways. First simply due to the very flat nature of the bands, they hardly changes
through the brillouin zone. Alternative you can calculate the bloch states for them, and you will see that they are located at the surface of the wire.
Finally you can dope the surface with Hydrogen, in order to artifical remove these surface states.
So in a sense your calculations, are very good, since they shows us what we except (or at least what I excepted ).
The question is then, how does this calculation relate to experiments, and my experience tells me quite good.
Four years ago, I was working on building a blue-ray laser from ZnO wires, and I found that my wires were really good semi-conductors,
and the light emitted correspond really well with the band gap. However as I gained more and more experience in creating cleaner and cleaner
wires, my light emission became weaker and weaker, and my wires began to act like metals rather than semi-conductors, and the reason
was that I needed the surface to be polluted in order to preserve the semi-conductor properties of the wires, as the pollution destroyed
the surface states.
Therefore when trying to model this behavior I choose a strategy that went along these lines:
1) If the wire is small in cross section, I could choose to add hydrogen termination, as it would make the results agree with my measurements,
and removing them I could make the calculation agree with my most purified ZnO wires.
2) If the wire is larger in cross section, then I left out the hydrogen termination, as the surface states becomes a minor effect as the system size increases.
3) If the wire were very large in cross section, then the results are the same as they are for ZnO (bulk), and hence they were not soo fun to calculate on
On a note I should say, that I used that I have checked the correct of the results of ATK for ZnO Wires against Abinit, and they agree very good, however for these things
ATK is way faster