| Abstract |
Spontaneously grown gallium nitride (GaN) nanowires (NWs) by the use of plasma assisted molecular beam epitaxy (PAMBE) are a subject of active research due to their high crystalline quality and promising device applications. The properties of GaN NWs depend strongly on the initial surface on which they are grown. In this work we investigate the effect of different initial surfaces on the morphological properties of GaN NWs, when silicon Si (111) is used as a substrate. We employ several experimental techniques in order to determine the morphological, structural and optoelectronic properties of GaN NWs, including reflection high energy electron diffraction (RHEED), field-emission scanning electron microscopy (FE-SEM), X-ray diffraction (XRD), photoluminescence (PL) spectroscopy and high-resolution transmission electron microscopy (HRTEM). In a first series of experiments, we explore the effect of substrate temperature (Tsub) on the nucleation, growth and properties of GaN NWs. The experiments follow (a) a one-step GaN growth procedure, with constant Tsub throughout GaN deposition, or (b) a two-step GaN growth procedure, using a lower Tsub for the initial stage than the later one. We achieve control of the NW diameter between 20-40 nm. A significant suppression of GaN NW nucleation at high temperatures (790-800°C) was observed for the one-step growth process. We are able to surpass this constraint using a two-step growth process. The GaN nucleation at the initial low Tsub stage allows for main GaN NW growth at much higher Tsub during the second growth stage.
Through a second series of experiments, we investigate how the formation of silicon nitride (SixNy) affects GaN NW growth. SixNy forms unintentionally when GaN is grown under N-rich conditions on a bare Si (111) surface. A better control of SixNy formation can be achieved by intentional nitridation of the surface prior to GaN deposition. We perform the first systematic comparison of GaN NW properties grown on unintentionally and intentionally nitridated Si (111) surfaces. The intentional nitridation resulted in a uniform 1.5 nm amorphous SixNy interlayer at the GaN/Si interface, while an irregular and non-uninform interface, with partial presence of amorphous SixNy, appeared for direct growth of GaN on Si. The homogeneity of the interfacial structure enhanced the degree of crystallographic alignment of the GaN NWs, concerning both tilt and twist. It also decreased the dispersion of NW heights that is otherwise triggered by different nucleation times on structurally different sites of the substrate. The average height of the GaN NWs was similar for both cases but their average diameter was increased from 25 nm to 40 nm on the uniform amorphous SixNy interlayer, possibly an effect of weak epitaxial constraints. Reduced overall intensity and increased defect-related emission at 3.417 eV characterized the 20 K photoluminescence spectra for direct GaN growth on Si. A third series of experiments explores for the first time the effects of ultrathin AlN prelayers, with nominal thicknesses between 0 and 1.5 nm, on the spontaneous growth of GaN NWs on Si (111) substrates. The increase of AlN thickness gradually limits nitridation of the substrate surface and accelerates the nucleation of 3D GaN islands. The formation of amorphous SixNy by Si nitridation is completely avoided for 1.5 nm of AlN that fully covers the Si surface. The dependence of the height, diameter and density of GaN NWs on the AlN thickness was also determined. Surprisingly, infinitely small changes in the AlN prelayer nominal thickness (0.1 nm) can cause significant differences in the GaN NW nucleation procedure, and the final NW characteristics (i.e. differences in height of 300 nm). The 1.5 nm AlN provided the optimum condition for GaN NW nucleation and growth; the NWs exhibited a large homogeneous height with almost no parasitic GaN formation between them. High resolution transmission electron microscopy showed the full relaxation of misfit strain of AlN on Si (111) and of GaN NWs on AlN. Analysis of in-situ RHEED monitoring of the AlN nucleation revealed the immediate relaxation of the AlN prelayer (before GaN deposition). Formation of β-Si3N4 before AlN nucleation was also observed. In a fourth series of experiments, we study the epitaxial relation of the GaN NWs with the Si (111) crystal for off-axis substrates. We focus on the (0001)GaN//(111)Si epitaxial relation for GaN NWs grown on Si (111) substrates with different miscut angles, and for GaN NWs grown on nominally identical Si (111) off-axis substrates but different interfacial structures. Interestingly, the GaN [0001]//Si[111] epitaxial relation is dominant, even in the case where a thick 11.5 nm, amorphous SixNy is present at the interface.
Finally, a method to selectively grow straight, vertical GaN nanowires by PAMBE at sites specified by a silicon oxide mask, which is thermally grown on silicon (111) substrates and patterned by electron-beam lithography and reactive-ion etching is also analyzed. The investigated method requires only one single molecular PAMBE growth process, i.e., the SiO2 mask is formed on silicon instead of on a previously grown GaN or AlN buffer layer. The study of various mask patterns, with the combination of numerical simulations, allowed us to evaluate how the geometrical characteristics (window diameter and spacing) of the mask affect the distribution of the nanowires, their morphology, and alignment for given Ga adatom diffusion length. Capabilities and limitations of this method of selective-area growth of nanowires have been identified. A window diameter less than 50 nm and a window spacing larger than 500 nm could provide selective nucleation of single nanowires in nearly all mask windows.
|
Search
