Development of X-ray Ptychography at the Diamond Light Source for the study of Domain Walls in Cu3Au.

相干X射线衍射(CXD)是研究纳米晶体内部三维结构的一种强有力的新方法。它利用了最新一代同步辐射(SR)源(如钻石)的壮观相干特性。当光束的相干长度内有足够的通量时，整个晶体就可以被相干地照射。则衍射图是对结构的傅里叶变换的幅度的完整测量。如果该衍射被充分地过采样，则丢失的相位信息被编码在衍射图案的更精细的特征中，其中它可以通过使用迭代相移算法来恢复。然后，傅里叶变换可以直接逆成样本的图像。在单一入射角的情况下，产生2D衍射图；然而，如果然后摇动样品，可以建立一系列衍射图来创建完整的3D数据集。最近，我们展示了可以获得完全复杂的3D图像，并根据样品内的应变场对其进行有意义的解释。晶格形变(应变)到动量传递上的投影是样品内部每一点相的量化值。Cu3Au是一种有趣的材料，用于纳米成像研究，它是经历有序-无序相变的金属合金的模型系统。这种无序是由于Au在合金的面心立方晶格中占据不同的晶格位置而形成的反相畴(APD)。它的X射线研究有很长的历史，可以追溯到沃伦的工作，然后被曾傑瑞·科恩、西蒙·莫斯、赫尔穆特·多施等大型研究小组广泛研究。薄膜特别令人感兴趣，因为还有薄膜微区的微结构问题。因此，在像我们这样的给定样本中，在整体和不同类别的上层建筑反射中应该有不同的和不同的结构。不同的布拉格反射对不同的序参数分量很敏感，通过相干衍射的反转会给出不同的APDS图像：基波反射将成像薄膜的颗粒结构，而超结构峰将成像反相区。已知在300℃左右的相变附近，磁区的平均尺寸变化非常关键，但形状的变化在很大程度上是未知的。CXD依赖于寻找缺失相的可靠算法。到目前为止，我们的经验是，来自致密、孤立和稳定物体的高质量衍射数据通常可以使用已知的方法进行反演。高质量不仅意味着具有良好统计数据，而且不受背景污染，通常以类似取向的杂散颗粒的形式存在。到目前为止，CXD在成像连续物体(如雪崩光电二极管阵列)以取代孤立的纳米物体方面并不成功。这被认为是因为定义样品上的光束的照明所需的孔径的不确定性，以及因为典型的连续物体具有高度的内部对称性，通常具有不同取向的分离物体的多个副本。这里特别有趣的是，APD的大小和形状看起来很相似，因此提出了一个挑战，我们相信可以通过使用新的体层摄影术方法来克服。

Coherent X-ray Diffraction (CXD) is a powerful new method for investigating the internal structure of nanocrystals in three dimensions (3D). It utilizes the spectacular coherence properties of the latest generation of synchrotron radiation (SR) sources, such as Diamond. When there is sufficient flux within the beam's coherence length, an entire crystal can be illuminated coherently. Then the diffraction pattern is a complete measurement of the amplitude of the Fourier transform of the structure. If this diffraction is sufficiently oversampled, the missing phase information is encoded in the finer features of the diffraction pattern from where it can be recovered by use of iterative phasing algorithms. Then the Fourier transform can be inverted directly to an image of the sample. At a single incidence angle of the beam on the sample, a 2D diffraction pattern results; however, if the sample is then rocked, a series of diffraction patterns can be built up to create a full 3D data set Recently we showed that a fully complex 3D image can be obtained and meaningfully interpreted in terms of strain fields within the sample. The projection of the lattice deformation (strain) onto the momentum transfer is the quantitative value of the phase at every point inside the sample.Cu3Au is an interesting material to use for these investigations on nanoscale imaging, being a model system of a metallic alloy that undergoes an order-disorder phase transition. The disorder arises because of the formation of antiphase domains (APDs), with the Au occupying different lattice sites in the FCC crystal lattice of the alloy. It has a long history of investigation with X-rays, dating back to the work of Warren, then extensively studied by the large research groups of Jerry Cohen, Simon Moss, Helmut Dosch and others. Thin films are especially interesting because there are questions of microstructure of the thin-film domains as well. In a given sample, such as ours, distinct and separate structure is therefore expected in the bulk and different classes of superstructure reflections. Different Bragg reflections are sensitive to different order-parameter components and will give different images of the APDs by inversion of the coherent diffraction: the fundamental reflections will image the film's grain structure, while the superstructure peaks will image the antiphase domains. The average sizes of the domains are know to vary in a critical way close to the phase transition around 300C, but the variation of the shapes is largely unexplored.CXD depends on a reliable algorithm for finding the missing phases. Our experience to date is that high quality diffraction data from compact, isolated and stable objects can usually be inverted using known methods. High quality implies not only data with good statistics, but also free from contamination by background, usually in the form of stray grains of similar orientation. CXD has to date been unsuccessful at imaging continuous objects, such as APD arrays, in the place of isolated nano-sized objects. This is believed to be because of uncertainties with the aperture needed to define the illumination of the beam on the sample and because there is a high degree of internal symmetry in a typical continuous object, often with multiple copies of separated objects with differing orientations. Of specific interest here, APDs tend to look similar is size and shape, and so present a challenge, which we believe can be overcome by use of the new ptychography method.