Ferroelastic domain dynamics

: a multiscale study through in situ microscopy

Abstract

Ferroelastic components are crucial elements in multiferroics, mediating and influencing key properties in ferroelectric and ferromagnetic materials. Thus, ferroelastics play a pivotal role in the development of active-adaptable devices in neuromorphic computing as well as a host of standalone applications in sensors and switches. In this context and considering the current drive for device miniaturisation towards increasing computational output, an understanding of the fundamental physics towards the microscale become imperative.Historically, limitations in imaging techniques have hindered the observation of subtle yet crucial dynamics towards the microscale. However, with the emergence of new in situ microscopy capabilities, observations of physical mechanisms such as domain switching are now accessible.This thesis investigates the behaviours of ferroelastic domains and domain walls, both statically and dynamically, extending from bulk to freestanding samples at the microscale, using LaAlO3 as a standard ferroelastic system.Initial investigations on the bulk examined the effect of varying the aspect ratio (length/width), in an optical set up, using heat cycles (beyond LaAlO3’s critical temperature) as the driving force. This revealed divergent behaviours during the ramp down of said heat cycles. Pixel-by-pixel analysis, based on the collective propagation of domains known as avalanches, identified that the observable differences in behaviour was characteristic of an exponential mixing, a previously unreported phenomenon for ferroelastics.Subsequently, the sample size was reduced to explore scaling effects on the microscale, where the influence of the chemical environment was investigated; a dynamical treatment not previously explored in ferroelastics. Several environments were tested, using an in situ Transmission Electron Microscopy (TEM) heating and gas, consisting of ultra-high vacuum (UHV), N2, O2/Ar, H2/N2/Ar. Dynamic analysis revealed the active suppression on domain patterning under the H2/N2/Ar environment. While the UHV, N2 and O2/Ar were comparable in terms of domain behaviours and to that of the bulk, the UHV was selected due to the experimental success rate.Building upon these findings, the thickness effect of specially designed freestanding samples was tested in the UHV environment via in situ heating TEM. Experimental results correlated to the behavioural regimes known as the monopolar-crossover-dipolar regimes based on the interactions between kinks in the domain walls. Strain analysis from Scanning Procession Electron Diffraction (SPED) data was used correspondingly with Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) simulations to confirm the transition between these regimes and the behaviours therein. Additionally, the foundation for automated measurements, using machine learning, was established to aid the measurement of domains’ characteristic such as the domain wall density and curvature.

Date of Award	Jul 2024
Original language	English
Awarding Institution	Queen's University Belfast
Supervisor	Miryam Arredondo (Supervisor) & Marty Gregg (Supervisor)