Influence of defects on polarization distribution in ferroelectrics: a phase-field study
Open Access
- Author:
- Cheng, Xiaoxing
- Graduate Program:
- Materials Science and Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 06, 2020
- Committee Members:
- Long-Qing Chen, Dissertation Advisor/Co-Advisor
Long-Qing Chen, Committee Chair/Co-Chair
Venkatraman Gopalan, Committee Member
Michael T Lanagan, Outside Member
Roman Engel-Herbert, Committee Member
John C Mauro, Program Head/Chair - Keywords:
- Defect
Ferroelectrics
Phase-field simulation
Dislocation
Flexoelectricity
Defect engineering
BiFeO3
SrTiO3 - Abstract:
- Ferroelectric is a large group of functional materials that has outstanding ferroelectric, piezoelectric, dielectric properties, and a wide range of applications, such as capacitors, actuators, transducers, random access memories, waveguides, etc. One of the defining features of the ferroelectric materials is the appearance of a complex switchable spontaneous polarization domain structure, which determines the effective properties of the studied material that leads to usage in different applications. In recent years, due to the increasing demand for novel nanoelectronic devices, scientists have been working on nanoscale domain engineering, which means understanding and manipulating the polarization domain structures at the nanoscale, to search for potential answers to the next generation semiconductors from a materials science perspective. In this dissertation, we study how defect engineering, particularly, dislocations in SrTiO3 and non-stoichiometric charged defects in BiFeO3, may contribute to the control of both local and long-range polarization distribution at the nanoscale. There is a long-standing interest in creating and stabilizing ferroelectric polarity in non-polar crystals such as SrTiO3. Recently, measurable electrical polarization as large as ~28 uC/cm^2 has been reported at the dislocation cores in SrTiO3 bicrystals. The origin of this polarity was attributed to the flexoelectric effect, i.e., induced polarization due to strain gradients. In this work, we systematically study the role of flexoelectricity on inducing polarization around three types of dislocation cores in SrTiO3, (100) edge dislocation, (110) partial edge dislocation, and (010) screw dislocation. We demonstrate that, in the two edge dislocation cases, flexoelectricity has a significant influence on both the polarization's magnitude and distribution. It increases the average polarization value and drives the polarization into a symmetric distribution. For (100) edge dislocation, the ferroelectric phase is mainly tetragonal and can exist even without flexoelectricity, relying purely on the electrostrictive effect. For (110) partial edge dislocation, electrostriction alone is not sufficient to stabilize the ferroelectric phases, and flexoelectricity is essential for the presence of the orthorhombic polarization in this case. Moreover, through tuning of the three flexoelectric coefficients, we recognize the shear flexoelectric coefficient V1212 to have the largest effect on the stable polarization pattern and magnitude for both types of edge dislocations. In contrast, in the (010) screw dislocation cases, we learn that neither electrostriction nor flexoelectricity will be able to stabilize any polar state in SrTiO3. Our findings provide an in-depth understanding of the flexoelectric effects on the induced polarization around three dislocation cores, which may potentially bring new insights into the defect engineering of ferroic materials using dislocations. A discussion about the role of dislocation core's electric effect on polarization distribution and the comparison with the elastic contribution is also crucial for the comprehensive understanding and prediction of polarization patterns near the dislocation cores in SrTiO3. We explore the influence of defect charges on local polarization in room temperature SrTiO3 of three types of dislocations, (100) edge dislocation, (110) partial edge dislocation, and (010) screw dislocation. We find that for edge dislocations, defect charges have a shorter interaction range compares to the flexoelectric effect. The charge induced polarization has a highly anisotropic distribution that is directly related to the local stress state of the system. Defect charges, in the edge dislocation cases, lead to larger polarization value at the dislocation core comparing to the flexoelectric and electrostrictive effect, while the defect's elastic effects have a broader impact region and larger magnitude than the electric ones. Similar polarization distribution can be observed in experimental characterization of regions around (100) edge dislocations in SrTiO3. In the screw dislocation case, the defect charges induce an almost isotropic polarization distribution around the dislocation core. At the same time, flexoelectricity has no influence on the polarization due to the contrary contribution of the non-zero shear stress to the flexoelectric field. Overall, the pure electric effect of the defect charges leads to a nearly isotropic distribution of local polarization within 1 nm around the dislocation core for all three types of dislocations. The flexoelectric effect has a much larger impact on polarization in the two edge dislocation cases than the screw dislocation case. The electrostrictive effect only affects the polarization distribution in the two edge dislocation cases since the location of the total free energy minima is shifted by the local normal stresses while remaining almost unchanged with the presence of the shear stress components. These results provide a comprehensive understanding of how the elastic and electric effects of dislocations in ferroic materials help to stabilize the local polarization around the dislocation cores. Another type of defect engineering system that we investigate to control the ferroelectric domain structures is the non-stoichiometric charged defect in BiFeO3. The differences between this type of defect and the dislocation are, first, it has no lattice mismatch with the matrix (zero eigenstrain); second, it is a two-dimensional defect. We study how the planar non-stoichiometric charged defect configurations, including defect width, interval, and location, may determine the thermodynamically most stable domain structure inside the thin film. We perform high-throughput simulations varying the defect width, interval, and height within a 200 nm BiFeO3 thin film. The trends for every energy term with respect to the defect configurations are explained and analyzed. The stability of the 71 domain strips above the charged defects is explained through the competition between elastic and domain wall energy. We obtain an empirical formula that relates the defect width, position, and thin film thickness with the final domain pattern, which can be used as a predictive tool for the occurrence of the 71 domain strips above the defects in similar BiFeO3 thin film systems. Our conclusion is that there exists a minimal defect width that favors 71 domain above the charged defect over a single domain state. The threshold value is determined by the thin film thickness and the defect configuration. This result provides a novel route to precisely control the 71 domain pattern formation in BiFeO3. The planar non-stoichiometric charged defect in BiFeO3 of a smaller dimension (less than 10 nm) may also affect the equilibrium domain structures in a ferroelectric thin film. We perform high-throughput simulations varying the defect width, location, shape, charge state, electric boundary condition, and initial domain structure within a 100 nm BiFeO3 thin film. We identify the factors that have significant influences on the polarization distribution and several configurations that can stabilize the 109 domain wall, thus applicable as a novel nanoscale domain engineering method. Under the short circuit boundary condition, for a negatively charged defect, we found that the defect thickness (or shape) determines whether we can get a local hedgehog state around the defect or not. Varying the defect thickness also leads to a preference for either 180 or 109 domain pattern below the defect. On the other hand, defect width and defect location have limited influence on the final domain pattern. For neutrally charged defects, the domain pattern is relatively insensitive to defect location and defect thickness. It favors a 109 domain above the defect, except when the defect width is too small, and a single domain state is preferred over the 109 domain pattern for the final polarization state. In the open circuit cases, we observe smaller domains, in other words, a higher density of domain walls, and more interaction between the charged defect and the 109 domain wall, comparing to their short circuit counterparts. We discover that the initial domain structure, whether it is random noise or a single domain, will affect the equilibrium polarization, indicating the possibility of experimental tricks such as small miscut angle or introduction of built-in potential may also contribute to the control of the as-grown domain pattern around the defect. These results illustrate how we can utilize the defects as a novel method to control the occurrence 109 domain wall in future nanoscale domain engineering applications. This work presents in-depth understandings of the local and long-range polarization pattern formation within two defect systems, which provides a solid basis for future experimental design and validation. We discuss the flexoelectric, electrostrictive, and defect charges contribution to local polarization around edge and screw dislocations in SrTiO3. Further, we perform a series of high-throughput simulations to explore the influences of non-stoichiometric charged defect configurations on long-range 71 and 109 domain pattern stability in defect engineered BiFeO3. This work clearly demonstrates the possibility and capability of precise domain pattern control through defect engineering, which could be a viable route to the design and fabrication of more complicated ferroelectric nano-devices.