TEM Characterization of LaFeO3Thin Films on SrTiO3 (111) Substrates

Abstract

In this thesis transmission electron microscopy (TEM) was used to study the atomic structure of LaFeO3 (LFO) thin films grown on (111) oriented SrTiO3 (STO) substrates by pulsed laser deposition. To prepare TEM samples both tripod polishing and focused ion beam (FIB) techniques were used, and samples of cross sectional (CS) and plan view (PV) geometries were studied. The films studied had a thickness of approximately 20 nm, except for one which measured about 55 nm. Tripod polishing yielded fair PV samples, but poor CS samples as the tripod is suspected of altering the film structure of these samples. FIB was used to prepare CS samples by lift-out, and to improve tripod PV samples. It is thought that even better samples than the ones prepared in the present work is achievable.

By electron diffraction, dark field (DF) imaging, high resolution TEM (HRTEM), and scanning TEM (STEM) the LFO thin film is found to be orthorhombic, with one of the shorter unit cell vectors lying in the STO (111) plane and the other two axes pointing out of this plane. Because of the six-fold symmetry of the substrate surface, the film exhibits three distinguishable domains, labelled A, B, and C. These domains are equally likely to nucleate on the substrate and are found in equal shares. They have no regular shape or boundary, and range in sizes from some tens to some thousands square nanometres. Substrate miscut induced step edges are not found to affect the domains.Internal domain boundaries (IDBs) are found in A, B, and C domains alike, and are credited to the long orthorhombic axis of two adjacent domains being antiparallel.

The film is found to couple to the substrate in certain areas, causing an out-of-plane elongation of the substrate unit cell. In these disturbed areas the cubic symmetry is lost and the STO unit cell is thought to be rhombohedral, orthorhombic, or monoclinic. This effect is credited to increased strain in triple points between substrate and domain boundaries. Such coupling may prove to be a way to engineer functional properties of oxide multilayers.