Development of an Automated Reciprocal-Space Navigator in a JEOL FEMTUS Platform
Development of an Automated Reciprocal-Space Navigator in a JEOL FEMTUS Platform
复制标题
在 JEOL FEMTUS 平台中开发自动倒易空间导航器
DOI：
10.1093/micmic/ozad067.359
复制
发表时间：
2023
影响因子：
2.8
通讯作者：
Watanabe, Masashi
中科院分区：
文献类型：
--
作者：
Huang, Surui;Chen, Brian;Bharati, Aparna;Harmer, Martin P;Watanabe, Masashi
电子显微镜操作自动化可以使许多用户受益。新手可能不确定如何操作基本控件，但可以从为他们执行特定任务的自动程序中受益。专家用户可以使用处理日常活动的程序来提高工作效率，例如预扫描样本或校准仪器。无论经验水平如何，每个用户都可以从避免损坏仪器的安全程序中受益，并节省收集数据的时间。在扫描透射电子显微镜 (S/TEM) 中表征晶体材料时，一种常见的操作是在倒易空间中导航。薄样品必须倾斜到一定的方向，不仅用于原子分辨率成像和分析，而且用于位错和包括晶界在内的各种界面的表征。迄今为止，人们已经多次尝试开发用于 S/TEM 中晶体样品的自动载物台控制器，包括 TEAM 载物台 [1]。然而，商业 S/TEM 仪器中尚未提供用于在晶体取向之间导航的自动载物台控制器。因此，我们在理海大学推出了一种使用最新 JEOL FEMTUS 平台在倒易空间中导航的系统，该平台适用于 JEOL JEM-ARM200CF 像差校正 S/TEM。 JEOL FEMTUS 最近被开发用于图像和光谱数据采集以及包括系统校准在内的各种仪器操作。在 FEMTUS 平台中，可以通过基于 JEOL PyJEM Python 的库 [2] 访问用于各种仪器控制的应用程序编程接口 (API) 功能。因此，FEMTUS 可以用作基于 Python 的脚本编写的集成开发环境 (IDE)。该工具可以通过 FEMTUS-PyJEM IDE 为 JEOL S/TEM 仪器开发自动化、灵活的仪器控制和数据收集流程。为了在倒易空间中导航，必须 (1) 识别薄样品相对于入射电子束方向的当前晶体取向，以及 (2) 估计最近的区域轴。这两个步骤都可以通过检查会聚束电子衍射 (CBED) 图案中出现的菊池线来执行。为了实现（1）和（2）的自动化，需要自动检测菊池线，我们在本研究中通过以下步骤实现了这一点。在菊池线检测之前，通过应用掩模去除 CBED 图案扭曲的外围部分来提取衍射图案的中心部分（图 1A）。菊池线是通过基于霍夫变换的线检测 [3] 以极坐标表示 (ρ, θ) 提取的，它为图像中的线提供了比更容易识别的线方程 y= mx+ b 中的标准 y 截距和斜率更通用的数学表示。这里，ρ是直线与原点之间的垂直距离，θ是直线的方向。因此，平行线将具有相同的 θ 但不同的 ρ 值。线提取后，执行非极大值抑制，将所有检测合并到同一对象中，以选择最佳线。为了找到菊池线对，即菊池带，将通过先前基于霍夫变换的线检测步骤获得的θ值分配到72个角度范围以构建（ρ，θ）对的字典。对于字典中的每个（ρ，θ）对，选择ρ值差异最大的两条相邻平行线来表示菊池带（图1B）。结晶区轴是通过中心线的交点来估计的，这些中心线是通过平均各个…的 ρ 值确定的。
Automating the operations of an electron microscope can benefit many users. Novices, who may be uncertain how to manipulate basic controls, could benefit from automatic procedures that perform specific tasks for them. Expert users could become more productive using procedures that take care of routine activities, such as pre-scanning specimens or aligning the instrument. Regardless of experience level, every user benefits from safety routines that avoid damaging the instrument, and by saving time that would have been spent collecting data.One common operation, when characterizing crystalline materials in scanning transmission electron microscopy (S/TEM), is navigation in reciprocal space. A thin specimen must be tilted to a certain orientation not only for atomic resolution imaging and analysis but also for characterization of dislocations and various interfaces including grain boundaries. To date, there have been several attempts to develop an automated stage controller for crystalline specimens in S/TEM, including the TEAM stage [1]. However, an automated stage controller for navigating between crystalline orientations is not yet available in a commercial S/TEM instrument. For this reason, we present a system for navigating in reciprocal space using the latest JEOL FEMTUS platform for a JEOL JEM-ARM200CF aberration-corrected S/TEM at Lehigh University. JEOL FEMTUS has recently been developed for image and spectrometry data acquisition and for various instrument operations including system alignments. In the FEMTUS platform, application programming interface (API) functions for various instrument controls can be accessed via the JEOL PyJEM Python-based library [2]. FEMTUS can thus be used as an integrated development environment (IDE) for python-based scripting. This tool makes it possible to develop automated, flexible instrument control and data collection processes for JEOL S/TEM instruments through the FEMTUS-PyJEM IDE. To navigate in the reciprocal space, it is essential (1) to identify a current crystallographic orientation of a thin specimen against the incident electron beam direction and (2) to estimate the closest zone axis. Both the steps can be performed by examining Kikuchi lines appearing in convergent beam electron diffraction (CBED) patterns. To automate (1) and (2), Kikuchi lines need to be detected automatically, which we achieved through the following steps in this study. Prior to Kikuchi line detection, the central part of the diffraction pattern was extracted by applying a mask to remove distorted peripheral parts of CBED patterns (Fig. 1A). Kikuchi lines were extracted by Hough transform-based line detection [3] in polar representation (ρ, θ), which provide a more general mathematical representation for lines in the image than the standard y-intercept and slope in the more recognizable line equation, y= mx+ b. Here, ρ is the perpendicular distance between the line and the origin, and θ is the orientation of the line. Thus, parallel lines will have the same θ but different ρ values. After line extraction, non-maximum suppression, which merges all detections into the same object, was then performed to select the optimal line. To find a Kikuchi line pair, ie, a Kikuchi band, θ values obtained through the previous Hough transform-based line detection step were distributed into 72 angular ranges to construct a dictionary of the (ρ, θ) pair. For each (ρ, θ) pair in the dictionary, two adjacent parallel lines with the largest difference in the ρ value were selected to represent a Kikuchi band (Fig. 1B). The crystalline zone axis was estimated by intersections of center lines determined by averaging the ρ values of individual …