Publication Date


Date of Final Oral Examination (Defense)


Type of Culminating Activity


Degree Title

Doctor of Philosophy in Materials Science and Engineering


Materials Science and Engineering

Major Advisor

Rick Ubic, Ph.D.


Lan Li, Ph.D.


Dmitri Tenne, Ph.D.


Predictive models for composition-structure-property relationships are essential to realizing the full potential of electroceramic materials; however, the electroceramics industry has largely failed to invest in predictive models in favor of simple rules of thumb or expensive, time-consuming trial-and-error methods. Empirically derived predictive models have the potential to significantly improve and guide future research in a more cost-effective and timely manner. It may even be possible to predict some intrinsic properties (e.g., polarization) on the order of a unit cell using only the charge and size of each chemical component. Scientists and researchers may ultimately be able to use these types of models to produce compositional recipes from desired property or structural data or conversely predict properties and structures from compositional data. Moreover, empirical models allow for the investigation and discovery of trends in properties and structures that would be practically unobservable via either expensive experimental or computationally demanding first-principles methods. The main focus of this work will be the use of empirical data to derive predictive models for ceramics with the perovskite crystal structure and thereby glean crystallochemical insights which explain them.

Chapter one lays the foundation for this work. A general overview of perovskites is provided and the seemingly endless degree of applications and properties that perovskite exhibit are explored including ferroelectricity, piezoelectricity, pyroelectricity, superconductivity, ionic conductivity, etc. It should be noted that the development of general correlative models which encompass all of these various properties and structural effects would be well beyond the scope of this work. Instead, this work concentrates on developing models for the effects of layered A-site ordering, rock salt B-site ordering, trigonality, and oxygen vacancies on the perovskite structure. Additionally, previous empirical modeling efforts are introduced to lay the groundwork for the development of the correlative models within this work.

Chapter two examines the effect of layered A-site ordering on the perovskite structure. 15 compositions in the [(NayLi1-y)(1-3x)/2La(1+x)/2)TiO3 (NLLT) system (y = 0.25, 0.5, and 0.75; x = 0, 0.0533, 0.1, 0.1733, and 0.225) were produced. Long-range ordering for x ≥ 0.1 was observed via X-ray diffraction, while short-range ordering was shown to exist even for x < 0.1 via electron diffraction. Overall, the degree of ordering was observed to decrease as x increased. The A-site ordering parameter (η) and the resultant expansion (ΔrA) were each modeled for every y series of compositions in the NLLT system. Two general models were developed based on these system specific models, which allow for the prediction of η and ΔrA in layered A-site ordered perovskites with any arbitrary degree of ordering using only published ionic radii data.

Chapter three investigates the effects of 1:2 B-site ordering in the BaMg1/3Ta2/3O3 (BMT) perovskite system. BMT is an ideal compound for the study B-site ordering because the degree of B-site ordering increases with increasing annealing time. Thus, BMT powders were synthesized and annealed incrementally from 0 to 40 hours at 1500°C. The resulting structures were increasingly ordered on the B-site, which also caused a trigonal distortion within the structure. An empirical model was developed which describes the degree of ordering in terms of the B-site shrinkage or the annealing time, which allows for the prediction of the ordering parameter (η) in BMT using only experimental input. This work has major implications in that this modelling technique may be expanded to include other B-site ordered complex perovskites, which could allow for the prediction of the degree of ordering and resultant volume shrinkage in B-site ordered perovskites in general using only published ionic radii data or experimentally-derived pseudocubic lattice constants.

Chapter four analyzes the effects of 1:1 rock salt B-site ordering in general. Four compositions were synthesized in the AZn0.5Ti0.5O3 system (A = Nd, Sm, Nd0.5La0.5, Nd0.5Gd0.5). Long-range 1:1 rock salt cation ordering on the B-site was shown to exist for all four compositions via X-ray diffraction. Additional data was mined from literature for another 38 rock salt B-site ordered perovskites. This data was sorted according to the Bsite species and empirical models were developed which describe the B-site shrinkage (ΔrB) for each of these systems. From these system specific models, a general model was developed for rock salt B-site ordering, which allows for the prediction of ΔrB in rock salt B-site ordered perovskites using only published ionic radii data.

Chapter five focuses on the development of a general correlative model for perovskite trigonality. A data mining approach was employed to gather published structural data for trigonally distorted perovskites. These perovskites were sorted into specific systems, and empirical models were developed for the degree of trigonality for each of these systems, which can be expressed a function of the ratio of the pseudocubic lattice constant (αpc) to the B-X bond length (rBX). From these system specific models, general models were developed for trigonal perovskites with either R3c or R3c symmetry which describe the degree of trigonality as a function of the modified tolerance factor. Additionally, a general model was developed which allows for the prediction of the intrinsic polarization in R3c trigonal perovskites.

Chapter six describes the effects of oxygen vacancies in perovskite ceramics. Eight compositions in the CaTi1-xFexO3-x/2 system (x = 0.05, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, and 0.45), and eleven compositions in the SrTi1-xFexO3-x/2 system (x = 0, 0.025, 0.05, 0.10, 0.15, 0.20, 0.25, 0.30, 0.35, 0.40, and 0.45) were synthesized. A data mining approach was used to supplement this data with more oxygen vacant perovskite systems. Empirical models were developed for each of these oxygen-deficient systems which describe the effective oxygen vacancy size (rV) and the bond deformation (DB). In addition, a general model was developed for the modified tolerance factor, which allows for the prediction of the tolerance factor in oxygen vacant perovskites. In turn, this allows for the accurate prediction of the effective anion size (rX) and pseudocubic lattice constant (αpc) in oxygen vacant perovskites.