Controlling Phase and Functionality of Ultra-Thin Chalcogenide Films Through Heteroepitaxy

Low-dimensional crystal structures and polymorphism feature prominently in the science and applications of chalcogenide semiconductors. These can present processing challenges, as it can be difficult to prepare uniform thin films over large areas. However, with great challenges come great opportunities for controlling phase and functionality through epitaxy. A diversity of crystal phases and optoelectronic properties can be accessed through slight changes in thermodynamic boundary conditions. Opportunities abound for fundamental studies. Opportunities also abound for applications, but only if the processing outcomes can be well controlled. I will present several examples wherein we control the phase and properties of ultra-thin chalcogenide films through molecular beam epitaxy, focusing on heteroepitaxial growth on oxide substrates.<br/><br/>The layered material SnSe has pronounced in-plane structural and optical anisotropy, which may be useful for integrated photonics and ferroelectric field-effect devices, but only if the anisotropy can be controlled. By choosing to grow SnSe on oxide substrates with square or rectangular symmetry, we control the distribution of ferroelastic domains, and the optical properties. This is a relatively straightforward example that can be understood by considering the mismatch between substrate and film in-plane lattice constants.<br/><br/>A more complicated example is the growth of the perovskite BaZr(S,Se)₃ on LaAlO₃. In this case, Nature deviates widely from what we expect based on straightforward considerations of lattice mismatch. Instead, we observe a process of surface passivation and the self-assembly of a nm-thick buffer layer that supports epitaxial growth of fully-relaxed films with cube-on-cube orientation, despite a lattice mismatch exceeding 30%. This ultra-thin, self-assembled chalcogenide layer has a rock-salt-like structure, and fully accommodates the heteroepitaxial misfit strain without misfit dislocations. This illustrates how complex, multicomponent interface chemistry can produce surprises – that, in this case, are enabling for growth and characterization of strain-free and epitaxial chalcogenide perovskite thin films.<br/><br/>As a third example, I will present control of polymorphism and domain epitaxy in BaZrSe₃ thin films. We have previously shown that high-Se-content BaZr(S,Se)₃ perovskite alloys can be made by direct co-deposition of Ba, Zr, S, and Se, or by post-growth selenization of BaZrS₃. However, BaZrSe₃ is thermodynamically most stable in a hexagonal, face-sharing structure (<i>i.e.</i>, not a perovskite). The orthorhombic perovskite phase (orth-BaZrSe₃) features a direct band gap near 1.4 eV and may be of interest for photovoltaics. The hexagonal phase (h-BaZrS₃) is theoretically predicted to have a band gap in the mid-infrared (MIR), is expected to have exceptionally strong birefringence, and may be of interest for infrared photonics. We find that sufficiently-aggressive selenization of BaZrS₃ converts the film from the perovskite to the hexagonal phase. orth-BaZrSe₃ and h-BaZrS₃ both crystallize on oxide substrates, but the choice of substrate affects the relative stability. On hexagonal substrates, the perovskite-to-hexagonal transformation occurs more readily. On square planar substrates, the perovskite phase is more stable; when h-BaZrSe₃ does form, it does so with the optic axis in-plane, with domain degeneracy that renders it challenging to measure the film birefringence. However, on substrates with rectangular symmetry, h-BaZrSe₃ films form with the optic axis in a single orientation across the substrate. As result, we are able to demonstrate wave-plate functionality for infrared light. This illustrates how relatively subtle changes in thermodynamic boundary conditions can produce functionally diverse outcomes, spanning from a candidate solar cell absorber material, to a highly-birefringent optical coating for infrared photonics.