Using Multivariate Analysis of Scanning-Rochigram Data to Reveal Material Functionality
Using Multivariate Analysis of Scanning-Rochigram Data to Reveal Material Functionality
复制标题
使用扫描 Rochigram 数据的多变量分析揭示材料功能
DOI：
10.1017/s1431927616002312
复制
发表时间：
2016
影响因子：
2.8
通讯作者：
A. Lupini
中科院分区：
文献类型：
--
作者：
S. Jesse;M. Chi;A. Borisevich;A. Belianinov;Sergei V. Kalinin;E. Endeve;R. Archibald;C. Symons;A. Lupini
扫描透射电子显微镜（STEM）是研究材料及其多种性质（包括结构、电子、磁性、铁性等）的强大平台。局部地和在其基本长度尺度上。在其当前的标准实施方式中，STEM成像受到仪器连接的严重限制，其中信息丰富的电子衍射图案（Ronchigram）在每个射束位置处被减少到单个值（例如，通过相对大的检测器上的积分强度），导致关于材料性质的重要但微妙方面的信息的损失和失真。先前的理论和实验工作[1，2，3，4]表明，在扫描期间在每个空间位置处完全采集Ronchigram可以实现超分辨率、相衬成像、内部场成像和3D样本重建。数据采集和存储已经发展到可以快速捕获高分辨率4D重叠关联数据集的程度，并且使用大规模计算设施使得这些多GB数据集的处理和挖掘能够在将统计上显著的变化（信号）与噪声分离的同时，消除数据的最显著方面。在当前的工作中，我们已经利用DE-12相机（Direct Electron，LP，San Diego，CA），其配备有安装在以300 ° C操作的像差校正FEI Titan上的4096 X 3072像素直接检测装置（DDD®）传感器[5]。
The scanning transmission electron microscopy (STEM) is a powerful platform for studying materials and their many varied properties (including structural, electronic, magnetic, ferroic etc.) locally and at their fundamental length-scales. In its current standard implementation, STEM imaging is severely restricted by the instrumental linkage in which the information rich electron diffraction pattern (Ronchigram) is reduced to a single value (e.g. through the integrated intensity over a relatively large detector) at each beam location resulting in loss and distortion of information about important but subtle aspects of material properties. Previous theoretical and experimental work [1, 2, 3,4] suggests that full acquisition of the Ronchigram at each spatial location during a scan can enable super resolution, phase-contrast imaging, imaging of internal fields, and 3D sample reconstruction. Data acquisition and storage has evolved to a point now that it is possible to capture high resolution 4D ptychography data sets rapidly, and the use of large-scale compute facilities enables the processing and mining of these multi-GB data sets to distil the most salient aspects of the data while separating the statistically significant variations (signal) from noise. In the current work we have utilized the DE-12 camera (Direct Electron, LP, San Diego, CA), equipped with a 4096 x 3072 pixels Direct Detection Device (DDD®) sensor [5] installed on an aberration corrected FEI Titan operating at 300