Multi-Dimensional Electron Microscope

在过去十年左右的时间里，电子显微镜发生了一些革命，这是材料科学的核心技术，也是固态化学和凝聚态物理学的一部分。这场革命是基于硬件的发展，改进了电子光学，单色，相机灵敏度和光谱仪效率，以及软件，代码现在能够以强大而快速的方式处理大量数据集，以提取关键信息。这个关于“多维电子显微镜”（MDEM）的提议将许多这些发展汇集到一个单一的仪器中，该仪器致力于在二维和三维的原子和纳米尺度上分析材料。现代显微镜的灵活性和功能还在于能够同时使用多个检测器，相机和光谱仪，以便可以从单个电子束位置获取多个信号-这被称为“多模态”显微镜，当与MDEM方法相结合时，可以对结构和成分，晶体学和物理化学行为进行详细的研究。MDEM基于扫描电子显微镜，其可以从低电压（例如60 kV）到高电压（例如300 kV）操作，前者用于研究具有低原子序数和/或低维度的样品，例如石墨烯，其中撞击损伤可能是主要的，后者用于辐射分解可能是非常有害的有机晶体。MDEM旨在研究之前被认为对光束过于敏感而无法用传统方法检查的样本。通过使用最新一代的直接电子探测器，具有非常灵敏和线性的响应，我们能够在几毫秒内记录有机晶体的衍射图案，然后晶体在光束下降解。我们将应用这种方法来研究药物晶体中的纳米级缺陷结构，位错，堆垛层错和孪晶的发展，以及重要的是不同有机晶体之间的界面。值得注意的是，人们对加工的“半结晶”聚合物的微观结构知之甚少，特别是脂肪族聚合物，如聚乙烯和相关烷烃。通过使用扫描电子衍射方法，我们将使用MDEM来揭示迄今未见的聚合物纳米结构。电子断层扫描，或3D成像，现在可以扩展到大量的纳米级材料，并可以与衍射，X射线和能量损失光谱相结合，提供材料结构，成分和晶体学的完整3D图片。该方法几乎是普遍适用的，并且该方法所实现的材料科学范围是巨大的。MDEM的多模态多维方面意味着我们能够获取大量信息，并将开发新的软件算法，以稳健，高效和有意义的方式处理数据。这些算法使用了机器学习和压缩传感领域的最新思想，在压缩传感领域，先验信息被构建到图像、断层图像或光谱的任何重建或解释中。有许多材料系统和设备将受益于MDEM方法，除了已经提到的那些，我们还提供了一些例子：钙钛矿太阳能电池、氮化物半导体、工程合金如纳米结构钢和Ni基超合金、低维二硫属化物、磁性skyrmionic材料、非均相催化剂、MOF和金属玻璃。

In the past decade or so, there has been something of a revolution in electron microscopy, a technique central to much of materials science, and parts of solid state chemistry and condensed matter physics. That revolution has been based around developments both in hardware, improving electron optics, monochromation, camera sensitivity and spectrometer efficiency, and in software, with code now able to process vast data sets in a robust and speedy fashion to extract the key information. This proposal, for a 'multi-dimensional electron microscope', or MDEM, brings together many of these development into a single instrument that is dedicated to analysing materials at the atomic- and nano-scales in two and three dimensions. The flexibility and power of modern microscopy resides also in the ability to use multiple detectors, cameras and spectrometers simultaneously so that multiple signals can be acquired from a single electron beam position - this is known as 'multi-modal' microscopy and when combined with the MDEM approach leads to a remarkable detailed investigation of structure and composition, crystallography and physico-chemical behaviour. The MDEM is based around a scanning electron microscope that can operate from low voltages (e.g. 60kV) to high voltages (e.g. 300kV), the former being used for the study of samples with low atomic number and/or of low dimension, such as graphene, where knock-on damage may be predominant, the latter for organic crystals where radiolysis can be hugely detrimental. The MDEM is designed to investigate samples that have previously been considered too beam-sensitive to examine with conventional methods. By using the latest generation of direct electron detectors, with remarkably sensitive and linear response, we are able to record diffraction patterns from organic crystals in just a few milliseconds, before the crystal degrades under the beam. We will apply this method to study the nanoscale defect structure in pharmaceutical crystals, the development of dislocations, stacking faults and twins and importantly the interfaces between dissimilar organic crystals. Remarkably little is known about the microstructure of processed 'semi-crystalline' polymers, especially aliphatic polymers such as polyethylene and related alkanes. By using scanning electron diffraction methods we will use the MDEM to reveal hitherto unseen polymer nano-structure.Electron tomography, or 3D imaging, can now be extended to a huge range of nanoscale materials and can be combined with diffraction, x-ray and energy loss spectroscopy to provide a full 3D picture of the materials' structure, composition and crystallography. The method is almost universally applicable and the range of materials science enabled by this method is huge. The multi-modal multi-dimensional aspect of the MDEM means we are able to acquire vast amounts of information and new software algorithms will be developed to process the data in a robust, efficient and meaningful fashion. These algorithms use the latest ideas in machine learning and in compressed sensing, where prior information is built into any reconstruction or interpretation of the image, tomogram or spectrum.There are numerous material systems and devices that will benefit from the MDEM approach and, in addition to those already mentioned, we present a few more examples: perovskite solar cells, nitride semiconductors, engineering alloys such as nano-structured steels and Ni-base superalloys, low-dimensional dichalcogenides , magnetic skyrmionic materials, heterogeneous catalysts, MOFs and metallic glasses.