You are right, this is a recurring problem in every junction simulation. I guess there are a few different ways to try to answer this question:
1) is experimental knowledge of the exposed surface facets available? That might be a good starting point.
2) when cleaved, a bulk crystal will usually try to expose the facet with lowest surface energy. This is often the (111) facet for fcc metals, the (100) facet for bcc, and (0001) for hcp. However, junctions in devices are usually created by depositing the two materials one at a time on an "inactive" substrate. The growth direction of the first junction material may therefore depend on the exposed surface of the substrate, and the second junction surface on the first one.... In the end, this is a complicated problem.
3) In most cases, one cannot know for sure the correct facet to use. But there may be general agreement that certain facets are more interesting/realistic than others, e.g. (100) for silicon.