Insight into the atomic structure and misfit relaxation mechanisms in micrometer-thick SrMoO₃ oxide electrodes grown on scandate substrates

Abstract text (incl. figure legends and references)

Introduction: Understanding the microstructure and related defect structures and their relationship with the physical properties is fundamental for the realization and optimization of oxide electronics-based devices. For instance, in the case of varactors or so-called metal-insulator-metal (MIM) tunable ferroelectric microwave capacitors, the tunable permittivity of the insulator layer and/or the thickness of the bottom oxide electrode exceeding the skin depth is crucial for its performance. To investigate the relationship between the thickness and other structural properties of these oxide thin films, i.e., alteration of stoichiometry and lattice constants, we performed Scanning Transmission Electron Microscopy along with X-ray diffraction (XRD) and Photoemission spectroscopy (PES).

Objectives: We characterized a set of heterostructure multilayer devices (scandates/SrTiO3/SrMoO3) by STEM. For the analysis of the HAADF-STEM images we implemented an automated-quantitative STEM method that allows to analyze a set of real-space images to (i) extract atomic position information and (ii) to identify local structural defects. The goal was to provide insight into the underlying strain accommodation mechanism in micrometer-thick SrMoO3 (SMO) oxide electrodes grown on various scandate substrates.

Materials and methods: The preparation procedure of the oxide-based multilayer devices can be found in P. Salg et al. [1]. The TEM lamellae were obtained using a JEOL JIB 4600F. For STEM, images with 5x5 frames along [001] and [100] directions of the epitaxial SMO layer were obtained using a JEOL ARM200F. Local in-plane/out-of-plane lattice constants were extracted along [001] growth direction by implementing customized codes, based on the python packages Atomap and Hyperspy. Picometer level precision was achieved. First principal calculations for SMO using density function theory (DFT) with the CASTEP package.

Results:

Both, XRD and STEM imaging (Fig.1a) confirmed the epitaxial growth (in-plane locked) of the SMO layer on top of all scandate substrates used. By performing automated-quantitative STEM analysis, atomic position information was obtained, as well as statistical out-of-plane/in-plane lattice constants (Fig.1b). The variation of this statistical lattice constants implies another strain accommodation mechanism, i.e., off stoichiometry, in heterostructure material in addition to interfacial defects. The DFT calculations for SMO layer confirmed the relationship between the variation of lattice constants and the off stoichiometry.

Conclusions: A set of heterostructure multilayer devices were characterized by the combination of PES, XRD and STEM. The use of the automated-quantitative image analysis showed a reliable way to measure the local lattice constants with picometer-scale accuracy. Together with the help of DFT calculations, the relationship between the interfacial defects, the lattice constant, the off stoichiometry, and the accumulative strain was studied.

References:

[1] P. Salg, D. Walk and L. Zeinar et al., APL Mater., vol. 7 (2019), p. 051107. doi: 10.1063/1.5094855.

The authors acknowledge funding from the ERC "Horizon 2020" Program under Grant No. 805359-FOXON and Grant No. 957521-STARE.

Fig 1(a). HAADF-STEM image of heterostructure interface scandate/SrTiO3/SrMoO3. (b) Atom position information were extracted using customized codes based on the Atomap python package.