Making BaZrS\n 3\n Chalcogenide Perovskite Thin Films by Molecular Beam Epitaxy

Abstract

We demonstrate the making of BaZrS3 thin films by molecular beam epitaxy (MBE). BaZrS3 forms in the orthorhombic distorted-perovskite structure with corner-sharing ZrS6 octahedra. The single-step MBE process results in films smooth on the atomic scale, with near-perfect BaZrS3 stoichiometry and an atomically-sharp interface with the LaAlO3 substrate. The films grow epitaxially via two, competing growth modes: buffered epitaxy, with a self-assembled interface layer that relieves the epitaxial strain, and direct epitaxy, with rotated-cube-on-cube growth that accommodates the large lattice constant mismatch between the oxide and the sulfide perovskites. This work sets the stage for developing chalcogenide perovskites as a family of semiconductor alloys with properties that can be tuned with strain and composition in high-quality epitaxial thin films, as has been long-established for other systems including Si-Ge, III-Vs, and II-Vs. The methods demonstrated here also represent a revival of gas-source chalcogenide MBE. Introduction Sulfides and selenides in the perovskite and related crystal structures – chalcogenide perovskites, for brevity – may be the next family of high-performing semiconductors. Chalcogenide perovskites share key physical properties with oxide perovskites, including recordhigh dielectric polarizability, while also featuring band gap (E ) in the visible and near-infrared (VIS-NIR). Chalcogenide perovskites are also distinguished by their good thermal and chemical stability, and non-toxic and abundant elemental components. Chalcogenide perovskites have been demonstrated to have slow non-radiative excited-state charge recombination, direct band gap (in most cases), and strong above-band-gap optical absorption. 9] Theoretical predictions, preliminary experiments, and chemical intuition suggest that chemical alloying may produce materials with continuously-tunable direct band gap spanning from E = 0.5 to 2.3 eV. These and other, related results suggest that the chalcogenide perovskite semiconductor alloy system may be useful for optoelectronic and energy-conversion technologies, particularly for solid-state lighting and solar energy conversion. Most experimental studies to-date on chalcogenide perovskites have focused on bulk materials (e.g. powders) and microscopic single crystals. Thin-film synthesis is the next, outstanding challenge. High-quality thin films are needed to enable fundamental studies of excited-state charge transport and applied studies of device performance. Thin-film synthesis may also be essential for studies of chemical doping and alloying. The good thermal stability of chalcogenide perovskites comes hand-in-hand with high materials processing temperature. The formation of compounds including low-vapor pressure refractory metals and high-vapor pressure chalcogens poses a particular challenge, common to chalcogenide perovskites and many layered and two-dimensional materials. Zr and Hf are sluggish to form crystalline compounds, requiring very high temperature for synthesis, but at very high temperatures Sand Se-containing precursors are extremely volatile, leading to chalcogen loss from the growing material, and highly-corrosive conditions for the experimental equipment. Published report of chalcogenide perovskite thin-film synthesis have appeared recently. All reports to-date are of two-step processes that separate the processes of sulfide formation (all are sulfides to-date) and film synthesis, and all have resulted in smallgrained and randomly-oriented thin films. High-quality, epitaxial, single-crystal film synthesis remains an essential goal to enable the potential of chalcogenide perovskites, as the history of other semiconductor and complex oxide materials systems teaches us. Epitaxial film growth may also be able to stabilize high-selenium-content alloys in the perovskite structure, since pure selenides form instead in non-perovskite, needle-like structures. Here we demonstrate the making of BaZrS3 thin films by molecular beam epitaxy (MBE). BaZrS3 forms in the orthorhombic distorted-perovskite structure with corner-sharing ZrS6 octahedra (space group Pnma, no. 62), has a direct band gap energy (E ) in the range E = 1.8 − 1.9 eV, and is the most-studied chalcogenide perovskite. The single-step MBE process results in films smooth on the atomic scale, evidenced by reflection high-energy electron diffraction (RHEED) measurements during growth, and by atomic-force microscopy (AFM) and scanning electron microscopy (SEM). The films are mirror-smooth and are brightly-colored even at 20 nm thick, indicating strong optical interaction. Epitaxial growth on LaAlO3 is confirmed by X-ray diffraction (XRD) and scanning transmission electron microscopy (STEM). Films grow via two, competing epitaxial growth modes: (M1) buffered epitaxy, with a self-assembled interface layer that relieves the epitaxial strain, and (M2) direct epitaxy, with rotated-cube-on-cube growth that accommodates the large lattice constant mismatch between the oxide and the sulfide. Results We deposit films on (001)PC-oriented LaAlO3 single-crystal substrates (MTI Corp.). LaAlO3 transforms from a rhombohedral (space group R3c) to a cubic crystal structure at elevated temperature (approximately at 500 °C). In this work, Miller indices marked with substrate “PC” (as above) indicate pseudo-cubic indexing; otherwise, we index LaAlO3 in its rhombohedral structure, with room-temperature lattice constants a = b = 5.370 Å, c = 13.138 Å. For a material ABX3 with corner-sharing BX6 octahedra, the pseudo-cubic lattice constant aPC is an average B-XB distance; for LaAlO3, aPC = 3.8114 Å. Orthorhombic BaZrS3 has lattice constants a = 7.056 Å, b = 9.962 Å, c = 6.996 Å, and aPC = 4.975 Å. [12] The growth modes M1 and M2 are competing mechanisms to accommodate this large lattice constant mismatch. We prepare the substrates by outgassing in the MBE chamber at 900 °C in flow of H2S gas. We deposit films from elemental Ba and Zr, and H2S gas. The chamber pressure during film growth varies between 5×10 to 9×10 torr and is controlled by the H2S gas flow. For the results shown here, the H2S flow rate (QH2S) during outgassing and film growth is between 0.6 and 0.8 sccm. The Ba and Zr rates are 0.021 Å/s and 0.0075 Å/s (approximately 1:1 stoichiometry), measured at the substrate position by a quartz crystal monitor (QCM). The substrate temperature during growth is held at 900 °C (measured at the thermocouple), and the film growth rate is 0.04 Å/s, confirmed by X-ray reflectivity (XRR). The H2S gas flow is maintained during cooldown after growth. More details on the deposition methods are presented below (Methods). Reflection high-energy electron diffraction (RHEED) data acquired during growth shows evidence of atomically-smooth, crystalline, epitaxial films. In Fig. 1a we present the RHEED pattern measured on the outgassed substrate along the [100]PC azimuth at 900 °C, before film growth. Within a short period of time after film growth starts, the substrate RHEED pattern disappears and a new pattern corresponding to the BaZrS3 film appears, as shown in Fig. 1b-c. This pattern remains consistent throughout the whole film growth process. Quantitative analysis of the RHEED data (Fig. 1b) shows that the in-plane d-spacing of the film measured along the [100]PC substrate azimuth is 4.98 Å, corresponding (020) planes of BaZrS3 (d-spacing of 4.991 Å for the relaxed structure). Measured along the [110]PC substrate azimuth (Fig. 1c), the film dspacing is 3.52 Å, corresponding to the (121), (200), and (002) planes (d-spacings of 3.525, 3.530, and 3.513 Å, respectively for the relaxed structure). This analysis revealed that the film is fullyrelaxed with the in-plane d-spacings matching their counterparts from the BaZrS3 reference pattern. This is consistent with the self-assembled buffered epitaxial growth mode (M1), described below. In Fig. 1d we show a photograph of a typical film, which is a deep orange-reddish color, appropriate for a material with a direct band gap in the range 1.8 – 1.9 eV and strong light absorption. In Fig. 1e we present atomic force microscopy (AFM) data measured on a film after growth, showing that it is atomically-smooth with roughness of 3.8 Å. In Fig. 1f we present the result of the XRR measurement on the film. The presence of well-defined Kiessig interference fringes in the XRR curve indicates that the film surface is smooth and the film-substrate interface is well defined; by modeling the XRR data we find the film thickness to be 23.52 ± 0.21 nm. Figure 1: Growth of smooth and epitaxial BaZrS3 thin films on LaAlO3 by MBE. (a) RHEED data measured on a LaAlO3 substrate at 900 °C before the start of film growth. The indices and light grey lines mark reflections measured along the substrate [100]PC azimuth. (b, c) RHEED data measured during film growth with the sample oriented along the [100]PC and [110]PC substrate azimuths, respectively. The RHEED data in (a-c) are scaled and centered similarly, so that the spacing of the different reflections can be compared directly. The substrate (light grey) and film (orange) reflections are labeled using pseudo-cubic indexing. (c) Photograph of a typical sample; film is 23.7 nm thick and is uniform across the 1 cm substrate. (d) AFM data showing a smooth surface interrupted by depressions; the image root-mean-square roughness is 3.92 Å. (f) XRR result showing Kiessig fringes indicating a smooth surface. In Fig. 2 we present x-ray diffraction (XRD) data. The out-of-plane scan (Fig. 2a) shows the BaZrS3 (H0K) family appearing with the LaAlO3 (00L) family of reflections. The rocking curve of the (202) film peak has a full-width of 0.409°. Two competing epitaxial growth modes contribute to this rocking curve width, as discussed below. In Fig. 2b we present azimuthal (φ) scans that identify the in-plane epitaxial relationship between film and substrate. The film (200), (121) and (002)