Electron Microscopy Characterization of III-Nitride Nanowires grown on Graphene

Abstract

GaN nanowires grown on graphene glass using molecular beam epitaxy have been studied using scanning electron microscopy and different transmission electron microscopy techniques. The nanowires have an AlN nucleation particle at the bottom. The nanowires were found to be quite evenly distributed on the substrate, and of approximately the same length (ca. 400 nm). Their cross-sectional shape varied from thin (less than 40 nm) perfect hexagonal, to thicker (40 - 70 nm) distorted hexagonal, to coalesced into larger clusters. It is assumed that the distribution of AlN nucleation particles contribute to the even distribution of nanowires. The AlN particles are dome-shaped with a typical height of 10-15 nm. Furthermore, very little lateral GaN growth around the base of the AlN nucleation particles was found, indicating that the size and shape of the AlN particle control the size and shape of the nanowire. Although the AlN nucleation particles show promising properties as nucleation particles for GaN nanowire growth on graphene, and an ability to control the shape of the nanowires, the presence of a material with a relatively high band gap pose challenges for the use of these graphene/AlN/GaN nanowire heterostructures in applications. The studied nanowires can be categorized, and types occurring most often were studied in further detail. The nanowires of type I (diameter larger than 40 nm) and II (diameter less than 40 nm) are considered to be closest to the targeted structure, with a single nanowire growing from a single nucleation particle. Furthermore, they are defect-free above the nucleation particle. The type III nanowire is a single nanowire grown from multiple nucleation particles, creating a foot at the bottom of the nanowire. Type III has approximately the same length as type I and II, but the diameter is in generally larger, and varies more (74 +- 32 nm). The type III nanowires are also defect-free above the foot. Compositional analysis was performed using energy-dispersive spectroscopy, which confirmed the presence of AlN at the nucleation particle, and GaN in the rest of the nanowire. Quanti fication using the Cliff-Lorimer method was attempted, but the use of calculated k-factors from black-box software and absorption lead to less accurate compositional quantification. In the future, the zeta-method should be investigated for quantification, as it could aid in solving some of the shortcomings that the Cliff-Lorimer method suffer from. For example, the zeta- method enable thickness independent element maps and absorption correction, which could possibly quantify the composition where the AlN nucleation particle and the GaN nanowire overlap. Images and diffraction patterns were obtained at two different tilts rotated 30 degrees relative to each other for all the studied nanowires. It is shown that this was useful for obtaining additional structural and morphological information, which together with the EDS analysis gave a more accurate impression of the studied nanowires, than single 2D projections usually obtained in the TEM would do. For example, these tilt series could determine that the wurtzite GaN nanowire facets are [-1100], and not [-2110] as commonly observed in wurtzite GaAs nanowires.