Orientation disparity in GaN/graphene/$m$-sapphire: control-based re-examination of thru-hole epitaxy
The crystallographic orientation of films grown on 2D-masked substrates is often used to infer the pathway among remote, van der Waals, and thru-hole (pinhole-seeded) epitaxy. However, attribution of a specific growth mechanism based on orientation can be ambiguous unless mask continuity and substrate pre-treatment are evaluated within a single process window. We compare GaN grown under identical conditions on four m-plane sapphire templates: (i) bare, (ii) "graphene-grown" (high-temperature Ar/H2 with CH4 on), (iii) "anneal-only" (high-temperature Ar/H2 with CH4 off), and (iv) graphene oxide spin-coated and reduced on pristine sapphire. GaN selects (103) on graphene-grown and anneal-only m-plane sapphire, selects (100) on bare m-plane sapphire, and is predominantly (100) with a minority (103) on graphene oxide spin-coated and reduced/pristine m-plane sapphire. High-resolution TEM shows that, on partly graphene-covered samples, nucleation occurs on exposed sapphire (thru-hole), not on graphene, providing mechanism evidence independent of orientation. Within this window, the substrate surface state set by high-temperature Ar/H2 pre-treatment (rather than mask continuity) primarily governs orientation, while open-area effects can play a secondary role. Thus, preferred orientation alone may not determine the growth mechanism; mask continuity and substrate pre-treatment must be explicitly controlled when using orientation as evidence for mechanism assignment.