Modelling and quantitative interpretation of electron energy-loss spectra using novel density functional theory methods

本文提出的研究旨在通过开发新的计算机建模方法和使用这些方法研究现实世界的材料问题，进一步提高我们使用电子能量损失谱来解决材料科学中的实际问题的能力。我们在两种精心挑选的材料类型中确定了用其他技术很难研究的关键问题；碳纳米管的掺杂及其对核燃料包层合金抗氧化性能的影响。我们对宏观材料特性与原子结构和键合的关系的理解，以及我们如何通过操纵这些特性来控制这些特性，都是在非常短的长度尺度上表征材料的技术发展的结果。一种特别有效的表征方法是测量电子通过薄样品时所损失的能量，即电子能量损失光谱（EELS）。电子损失的能量高度依赖于存在的元素，从而可以确定材料的组成。此外，光谱还包含了原子之间如何化学键合的信息。化学键的性质强烈影响光谱中的精细尺度细节，但用定量的方式解释这些细节并不简单。本提案旨在开发和测试使用计算机建模来预测试验材料系统光谱的方法，以允许从实际材料中观察到的光谱特征进行定量解释和解释。为了计算光谱中的特征，首先必须计算材料自身的电子是如何参与成键的——这是一个固有的量子力学问题。目前最常用的方法是基于密度泛函理论（DFT），它在精度和计算效率之间提供了理想的平衡。即便如此，为了合理的计算时间，可以包含在模型中的原子数量仍然是有限的。这种情况可以通过DFT的有效实现得到改善，特别是使用所谓的伪势。相对而言，伪势方法很少用于模拟EELS光谱，因为其他（所谓的全势或全电子）方法提供了一种更简单的方法，尽管速度较慢。我们提出通过实现新的计算EELS谱的方法来增强伪势方法，这样只有键的初始计算，而不是随后产生的EELS谱的计算，是非常耗时的步骤。该项目的一个关键目标是增加可以建模的系统的大小，以解决实际的材料问题。新开发的方法将用于两种材料的原理验证分析，其中关键特征的表征已被证明是非常有问题的。第一个涉及发展对碳纳米管中氮和硼杂质原子如何控制其性质的理解。这些材料在广泛的新型传感和计算应用中具有潜在的应用前景。第二项应用旨在通过研究锆合金包层氧化的关键机制来提高核燃料棒的寿命。最后，我们希望测试这样一个假设，即在样品后放置一个透镜来重新聚焦失去能量的电子，可以直接成像成键的对称性。这种光学结构被称为能量过滤扫描共聚焦电子显微镜（EFSCEM）。要做到这一点，我们需要计算电子是如何在材料中散射的，就像我们需要计算光谱一样。上述开发的方法将在计算中非常有价值，以检验这一假设，以确定这是否是一个可行的实验，以分配未来的实验资源。

The research proposed here aims to further our ability to use electron energy-loss spectra to solve real problems in Materials Science by developing new computer modelling methods and by using these methods to study real-world materials problems. We have identified in 2 carefully selected materials types key problems proving extremely difficult to study with other techniques; the doping of carbon nanotubes and what determines the oxidation resistance of nuclear fuel cladding alloys. Much of our understanding of how macroscopic materials properties relate to atomic structure and bonding, and how we can control properties by manipulating these, is a result of the development of techniques to characterise materials on very short length scales. A particularly powerful characterisation method is to measure the energy lost by the electrons as they pass through a thin sample, so called electron energy-loss spectroscopy (EELS). The energy lost by the electrons is highly dependent on the elements present, allowing the composition of the material to be determined. Furthermore, the spectra also contain information on how the atoms are chemically bonded to each other. The nature of the bonding strongly affects the fine-scale detail in the spectra, but interpreting these details in a quantitative way is not straightforward. This proposal aims to develop and test methods of using computer modelling to predict spectra for trial materials systems to allow the features observed in spectra from real materials to be quantitatively explained and interpreted.To calculate the features in the spectra, it is necessary first to calculate how the material's own electrons are involved in bonding - an inherently quantum mechanical problem. The most common methods for doing this are currently based on density function theory (DFT), which provides an ideal balance between accuracy and computational efficiency. Even so, the number of atoms that can be included in a model for a reasonable computation time is still limited. The situation can be improved using efficient implementations of DFT, in particular using so-called pseudopotentials. Relatively little use has been made of pseudopotential methods to model EELS spectra because other (so called full potential or all-electron) methods provide a simpler, albeit slower approach. We propose to enhance the pseudopotential approach by implementing new ways of computing the EELS spectrum so that the only the initial calculation of the bonding, and not the subsequent computation of the resulting EELS spectrum, is the significant time consuming step.A key aim of the project is to increase the size of the system that can be modelled to allow real materials problems to be solved. The newly developed methods will then be used in proof-of-principle analysis of two materials where characterisation of key features has proved to be extremely problematical. The first involves developing an understanding of how the addition of nitrogen and boron impurity atoms to carbon nanotubes controls their properties. These materials have potential applications in a wide range of novel sensing and computing applications. The second application aims to improve the lifetime of nuclear fuel rods by studying the critical mechanisms of oxidation in zirconium alloy cladding. Finally, we wish to test the hypothesis that placing a lens after the sample to refocus the electrons that have lost energy may allow the symmetry of the bonding to be directly imaged. The optical configuration to do this has been called the energy-filtered scanning confocal electron microscope (EFSCEM). To do this, we need to calculate how the electrons are scattered in the material much like the calculations we need to compute the spectra. The methods developed as described above will be very valuable in calculations to test this hypothesis to decide whether this is a viable experiment to which to allocate future experimental resources.