A Facile Route for Producing Single-Crystalline Epitaxial Perovskite

Sep 24, 2013 - Nano Lett. , 2014, 14 (1), pp 44–49 ... Atomic Layer Deposition of p-Type Epitaxial Thin Films of Undoped and N-Doped Anatase TiO2. K...
18 downloads 0 Views 2MB Size
Letter pubs.acs.org/NanoLett

A Facile Route for Producing Single-Crystalline Epitaxial Perovskite Oxide Thin Films Andrew R. Akbashev, Guannan Chen, and Jonathan E. Spanier* Department of Materials Science and Engineering, Drexel University, Philadelphia, Pennsylvania 19104, United States S Supporting Information *

ABSTRACT: We report how a low vacuum pressure process followed by a few-minute annealing enables epitaxial stabilization, producing high-quality, phase-pure, single-crystalline epitaxial, and misfit dislocation-free BiFeO3(001) thin films on SrTiO3(001) at ∼450 °C less than current routes. These results unambiguously challenge the widely held notion that atomic layer deposition (ALD) is not appropriate for attaining high-quality chemically complex oxide films on perovskite substrates in single-crystalline epitaxial form, demonstrating applicability as an inexpensive, facile, and highly scalable route. KEYWORDS: atomic layer deposition, BiFeO3, perovskite, ferroelectric, multiferroic, epitaxy

T

temperature windows, high vapor pressure, and the presence of distinct catalytic processes on differently terminated surfaces during deposition. In principle, there are no limitations to the single-crystalline epitaxial phase formation via ALD, and ALD can be considered as a potential route for obtaining singlecrystalline epitaxial films of perovskite oxides on perovskite substrates equal in quality to those obtained using hightemperature chemical or physical vapor deposition approaches. Growth of a technologically important multifunctional material, BiFeO3 via ALD is chosen to demonstrate the feasibility of this concept. BiFeO3 is one of the most technologically promising multiferroics.16 It is also attractive among ferroelectric materials for solar energy conversion due to its large photovoltage and visible photon energy bandgap (∼2.7 eV). While high-quality heteroepitaxial BiFeO3 thin films have been obtained through high-vacuum physical vapor deposition4,17 and molecular beam epitaxy (MBE),18 and metal−organic chemical vapor deposition (MOCVD),19,20 an inexpensive low-vacuum surface reaction rate-limited deposition route for obtaining this technologically promising perovskite oxide (or any perovskite oxide) in single-crystalline high-quality heteroepitaxial form has been elusive. BiFeO3 is considered difficult to grow in thin-film

he design of modern electronic and spintronic devices requires materials with unique properties such as high tunneling magnetoresistance and spin filtering efficiency, high-κ dielectricity, strong magnetic anisotropy, multiferroicity, and superconductivity.1−3 These properties in turn require highvacuum and high-temperature thin-film deposition methods that can produce these materials with exceptional crystalline quality.4−6 However, realization of industrially scalable device technologies relies on atomically controlled thin film preparation routes in which heteroepitaxial thin films of comparably high quality can be attained, but at considerably lower temperature and pressure, and cost. Atomic layer deposition (ALD) has become one of the most versatile and widely implemented techniques for obtaining atomically smooth conformal thin films of inorganic and organic materials in an energy-efficient and environmentally friendly manner.7−10 ALD is advantageous in achieving conformal coating of high-aspect-ratio structures and also allows for a very easy atomic control of the thickness over large areas without high vacuum. While ALD of technologically important binary oxides is prevalent,7−10 including high-κ gate insulators, there are far fewer reports of ALD of complex perovskite oxides, which are polycrystalline and/or possess misfit dislocations if epitaxial.11−15 This scarcity is owing to difficulties particularly in the growth optimization and an appropriate choice of the precursors with similar ALD © XXXX American Chemical Society

Received: August 10, 2013

A

dx.doi.org/10.1021/nl4030038 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 1. Schematic illustration of ALD process for obtaining heteroepitaxial BiFeO3: Bi precursor molecules (Bi(mmp)3) delivered by a vapor pulse adsorb on the surface of the Bi−Fe−O amorphous layer producing a Bi−O layer; molecules are then oxidized by ozone (O3). Following this, the sample is exposed to a vapor pulse of Fe(C5H5)2 producing an Fe−O layer, subsequently followed by oxidation using O3. The process of alternating pulses of selected number and duration is optimized to produce the desired postanneal stoichiometry and structural quality in the resulting film.

Figure 2. (a) X-ray reflectivity of typical as-deposited Bi−Fe−O and individual constituent oxide thin films deposited on SiO2/Si, confirming roughness on the atomic scale; (b) 2θ/ω scans of a typical BiFeO3 thin film grown on SrTiO3 obtained by annealed in air at 700 °C; inset: rocking curves (ω scans) of the (001) reflection of the BiFeO3 thin film and the SrTiO3 substrate show a high epitaxial quality of prepared perovskite film with no evidence of secondary phases; (c) BiFeO3 crystallization in the form of 30−50 nm-thick epitaxial thin films after the annealing of amorphous Bi−Fe−O samples at different temperatures.

weight loss after 170 °C attributed to the decomposition of Bi(mmp)3.23 We chose 135−145 °C as an evaporation temperature of Bi(mmp)3 , which provided sufficiently reproducible precursor pulses. We also selected Fe(C5H5)2 (ferrocene) since it has an excellent thermal stability, small molecule size, and can easily be oxidized by ozone. We find the best evaporation temperature for the ferrocene source to be 90 °C. In our preliminary ALD experiments, use of ferrocene led to a slow but very stable growth of iron oxide for substrate temperatures in the range of 150−300 °C when ozone (3−5 min, confirming that even a short thermal annealing (∼3 min) is sufficient to enable crystallization in ALD-grown epitaxial thin films. Since our ALD process involves a sequential deposition of bismuth and iron oxide layers, the resulting amorphous film is already well intermixed and a short exposure to high annealing temperature leads to the crystallization of epitaxial BiFeO3.

each precursor within each subcycle was minimized while maintaining the correct stoichiometry in the final Bi−Fe−O films. As expected, Bi(mmp)3 showed a higher growth rate than Fe(C5H5)2, thus the ratio between the number of pulses was NFe:NBi ≈ 3−5. Each Bi−Fe−O thin film deposition run was carried out on both SiO2/Si and SrTiO3(001) simultaneously. Usually in ALD-limited growth the thickness of a simple oxide film depends nonlinearly on the number of cycles when the total number of cycles is small (