Modeling ferroelectric LiNb_(1-x)Ta_xO_3 solid solutions and their defect structure

本论文是“极性氧化物中周期性低维缺陷结构”研究组的一部分，该研究组致力于用LiNb_1-xTa_xO_3（LNT）模型系统研究缺陷结构、电子和离子输运以及机电性质之间的关系，本项目的目标是在整个组成范围内对LNT固溶体进行理论描述。宏观材料性质是在微观结构的基础上从第一性原理计算的。因此，我们联合收割机结合了长期的经验，在原子计算领域的大量专业知识的铁电体建模。LiNbO_3是目前研究最多的铁电体之一，但我们对LNT固溶体的了解却相当有限。特别是，它们很少被原子论研究。这种缺乏的主要原因是无序合金的微观建模和其性质的定量测定是理论材料科学的主要挑战。一方面，在周期性超原胞方法中对无序系统的正确描述并不简单。另一方面，由于所涉及的超晶胞的大小，激发态性质的计算（例如光学答案）是非常苛刻的。为了解决这些问题，我们开发和扩展计算算法的从头建模的固体溶液和它们的缺陷结构，并将它们应用到LNT。我们的原子模型提供了两个至关重要的优势。第一个优点是可以模拟人工或实验无法访问的结构，以解开复杂的物理过程，同时采取行动，以确定材料的性能。因此，我们可以单独处理对物理性质的不同贡献，这些贡献同时取决于许多参数。第二个优点是，由于方法广泛，我们可以获得大量可直接测量的数量，从而可以将本提案与每个实验项目直接联系起来。随着结构和热力学性质的测定（例如，晶格参数、激活能、扩散系数、过量折射率）以及线性和非线性光学特征（例如光学的，拉曼和二次谐波光谱）我们的目标是预测LNT基态的组成依赖性质，为深入理解所研究的模型系统奠定了基础及其结构和电子激发。这使我们能够理解结构，化学计量和掺杂对宏观尺度的影响，从而允许解释实验结果。这反过来又使我们有条件提出具有定制机电或电光特性的固溶体的最佳组合物。

The proposal is part of the research group 'Periodic low-dimensional defect structures in polar oxides', which is dedicated to the correlation of defect structure, electron and ion transport and electromechanical properties using the model system lithium niobate-lithium tantalate (LiNb_(1-x)Ta_xO_3, LNT).The goal of the present project is the theoretical description of LNT solid solutions over the whole composition range. Macroscopic materials properties are calculated from first principles on the basis of the microscopic structure. Thereby, we combine a long-standing experience in the modelling of ferroelectrics with substantial expertise in the field of atomistic calculations. While LiNbO_3 is one of the most investigated ferroelectrics, our knowledge of LNT solid solutions is rather limited. In particular, they have been rarely investigated by atomistic theories. The main reason for this lack is that the microscopic modelling of disordered alloys and the quantitative determination of their properties are major challenges in the theoretical materials science. On the one hand, the correct description of disordered systems within the periodic supercell approach is not straightforward. On the other hand, the calculation of excited state properties such as the optical answer is very demanding due to the size of the involved supercells. In order to tackle these problems, we develop and extend computational algorithms for the ab initio modelling of solid solutions and their defect structures, and apply them to LNT.Our atomistic models offer two crucial advantages. The first advantage is the possibility to model artificial or experimentally not accessible structures to disentangle the complex physical processes that simultaneously act to determine the materials properties. Thus, we can singularly address the distinct contributions to physical properties, which simultaneously depend on many parameters. The second advantage is that, due to the broad methodological spectrum, we have access to a multitude of directly measurable quantities, which allow for the direct connection of the present proposal to each experimental project. With the determination of structural and thermodynamic properties (e.g. lattice parameters, activation energies, diffusion coefficients, excess enthalpies) as well as linear and non-linear optical signatures (e.g. optical, Raman- and second-harmonic-spectra) we lay the basis for the thorough understanding of the investigated model system.The aim of our proposal is the prediction of the composition dependent properties of the LNT ground state and its structural and electronic excitations. This enables us to understand the influence of structure, stoichiometry and doping on the macroscopic scale, and allows thereby the interpretation of the experimental results. This, in turn, puts us in the condition to propose optimal compositions of solid solutions with tailored electro-mechanical or electro-optical properties.