EELS and STEM studies of perovskite oxide heterostructures

Abstract

Perovskite oxide heterostructures have attracted much research interest due to a strong coupling between the local crystal structure and the functional properties. Of special interest are the interfaces of thin film systems, where novel phases can occur from this phenomena. This gives these materials a large range of functional properties, making them good candidates for use in novel device concepts like spintronics. To tailor-make devices from perovskite oxide heterostructures we need to understand their detailed structure at subnanometer scales. Transmission Electron Microscopy is an ideal tool for this due to the atomic resolution and wealth of experimental signals. In particular Scanning Transmission Electron Microscopy (STEM) and Electron Energy Loss Spectroscopy (EELS) are powerful techniques to investigate the crystal and electronic structure in these materials.

EEL spectra contain a wealth of information used to examine the electronic structure of a material at the atomic level. However, EELS data is often difficult to analyse quantitatively due to its complicated spectral features. In this project, a method to quantitatively and automatically analyse EEL spectra has been developed. The method is based on the model based approach, and has been implemented in the free and open source program HyperSpy. Utilizing this method, detailed information about oxidation state and the presence of oxygen vacancies has been extracted.

From high quality atomic resolution STEM images the position of individual atomic columns is resolvable. This technique can therefore be used to investigate small structural changes in a material. Most challenging is the analysis of light elements, such as the oxygen column. As displacement of oxygen atoms is a common type of structural distortion in perovskite oxide heterostructures, there is a need for good methods to find the position and shape of oxygen columns. In this project, the program Atomap was developed to be able to automatically, accurately and robustly find the position and shape of all cation and oxygen columns. In this work, STEM, EELS, Atomap and HyperSpy have been used to study the atomic and electronic structure of three different material systems. These have been La0.7Sr0.3MnO3 (LSMO) grown on (001)-oriented SrTiO3 (STO), LSMO grown on (111)-oriented STO and a bilayer system La0.7Sr0.3MnO3/LaFeO3 (LSMO/LFO) grown on (111)-oriented STO.

In the LSMO/STO-(001) system, analysis of STEM and EELS data revealed the presence of oxygen vacancies and a reduction in the Mn oxidation state at the interface. In co-junction with STEM-imaging, a different structural phase was observed on the LSMO side of the interface, compared to the bulk of the thin film. This phase was identified as a Brownmillerite structure with unordered oxygen vacancies, which became ordered over time.

Using STEM-EELS, chemical intermixing was studied in the heterostructures grown on (111)-oriented STO. Cation intermixing was observed on the STO interfaces. The terminating layer of the STO substrate was Sr-deficient, leading to a thin LaSrTiMnO3 layer at the interface.

When a LSMO/LFO bilayer was grown on (111)-oriented STO, structural changes were seen in both LFO and LSMO. Atomap was used to analyse the atomic structure. The position of all the atomic columns including oxygen was found, this revealed structural coupling between the LSMO and LFO.

An important consideration is potential sample damage from the electron probe. Therefore, beam damage on the perovskite oxide materials was studied by exposing the TEM sample to varying electron doses. Beam damage effects were observed in both EELS and STEM data.

Importantly, it was found that in LSMO, sample damage can occur without it being visible in the STEM images. The damaged area also extended into the region surrounding the exposed area.

This shows careful examination of potential beam damage is vital. To facilitate this, a standard method for assessing a TEM sample for beam damage susceptibility was suggested.

Sammendrag:

Grenseflatestudier med elektronmikroskop

Nanomaterialer kan ha unike elektriske og magnetiske egenskaper, spesielt i grenseflatene mellom forskjellige materialer der lagene tilpasser seg hverandre ved å flytte på atomer eller elektroner.

For eksempel, kan en slik grenseflate være superledende eller magnetisk selv om de separate materialene ikke er det. Dette kan utnyttes i framtidens elektronikk. Målet for mitt prosjekt er å studere hva som skjer med atomene og elektronene i disse grenseflatene.

Et elektronmikroskop er et ideelt instrument for dette. For det første, kan man se atomene i materialet og grenseflatene. Jeg har brukt et slikt mikroskop til å ta bilder av atomene. Deretter utviklet jeg et program som automatisk finner atomene i bildene. Med dette programmet finner jeg posisjonen til alle atomene i bildene med 1 pikometers presisjon, og jeg kan se om atomene har flyttet på seg sammenlignet med de langt borte fra grenseflaten. Programmet jeg har utviklet er derfor nyttig for å studere årsaken til at nanomaterialene har unike egenskaper.

For det andre, snapper atomene i prøven til seg energi fra elektronstrålen. Ved å måle hvor mye energi hvert elektron i elektronstrålen har mistet, kan jeg finne ut hvilket grunnstoff hvert atom i bildene er og hvilken elektrisk ladning det har. Da kan jeg vite om atomene har byttet elektroner.

Jeg har funnet ut at når et tynt lag La0.7Sr0.3MnO3 legges på en sideflate av en kubisk SrTiO3-krystall, vil noen av oksygen-atomene i grenseflaten forsvinne, og Mangan får flere elektroner. Da blir grenseflaten ikke-magnetisk, men oksygenet får stor bevegelighet. Hvis La0.7Sr0.3MnO3 i stedet legges på en skråskjært flate, bytter SrTiO3 og La0.7Sr0.3MnO3 noen atomer slik at materialene blandes sammen i den tynne grenseflaten. Til slutt, hvis man legger et lag med LaFeO3 mellom disse to vil i tillegg til sammenblanding i grenseflaten, oksygen-atomene i LaFeO3 forskyves.