基于机器学习的原子识别以及非线性漂移校正
黄子扬，刘 曦，王怀远，黄瑞龙，郑 赫*，赵培丽，贾双凤，王建波*
（1. 武汉大学物理科学与技术学院，电子显微镜中心，人工微结构教育部重点实验室和高等研究院，
湖北 武汉 430072； 2. 武汉大学科研公共服务条件平台，湖北 武汉 430072）
摘 要 球差校正扫描透射电子显微镜（scanning transmission electron microscope，STEM）是一种重要的微观结构表征手段。然而，由于电子束和样品漂移等问题，极大影响了STEM图像的质量和后续分析。针对上述问题，本文引入机器学习，改进了原子识别的方法，并在此基础上进行了元素分类；另外，针对单张STEM图像，在原子识别的基础上，提出了快速非线性漂移校正的方法，解决了以往漂移校正方法依赖较多数据的问题，此方法适用于辐照敏感材料的漂移校正，显著提高了STEM图像的解析效率。
关键词 非线性漂移校正；原子识别；机器学习；透射电子显微学
中图分类号：TG115. 21+ 5. 3; O766+. 1; O799; O722+. 4; O739; TN16
文献标识码: A doi:10.3969/j.issn.1000- 6281.2024.04.005
Atomic identification and nonlinear drift correction based on machine learning
HUANG Ziyang1, LIU Xi1, WANG Huaiyuan1, HUANG Ruilong1, ZHENG He1*, ZHAO Peili1, JIA Shuangfeng1, WANG Jianbo1, 2*
（1. School of Physics and Technology, Center for Electron Microscopy, MOE Key Laboratory of Artificial Micro- and Nano-structures, and Institute for Advanced Studies, Wuhan University, Wuhan Hubei 430072； 2. Core Facility of Wuhan University, Wuhan Hubei 430072 , China）
Abstract Spherical aberration-corrected scanning transmission electron microscopy (STEM) is a crucial tool for characterizing microscale structures. However, issues such as electron beam and sample drift can significantly affect the quality of STEM images and subsequent analysis. To address these challenges, this paper introduced a machine learning approach to improve atomic identification, followed by elemental classification. Additionally, a rapid nonlinear drift correction method for a single STEM image was proposed, building upon atomic identification. This method overcomed the previous data-dependency issue in drift correction and was applicable for drift correction in radiation-sensitive materials. It significantly enhanced the resolution efficiency of STEM images.
keywords nonlinear drift correction; atomic identification; machine learning; transmission electron microscopy
[1] 董全林, 蒋越凌, 王玖玖, 等. 简述透射电子显微镜发展历程 [J]. 电子显微学报, 2022, 41(6): 685-688.
[2] HETHERINGTON C. Aberration correction for TEM [J]. Materials Today, 2004, 7(12): 50-55.
[3] 姚骏恩. 电子显微镜的现状与展望 [J]. 电子显微学报, 1998，17(6): 81-90.
[4] 姚骏恩. 我国电子显微镜的研制与生产(英文) [J]. 电子显微学报, 2002，21(3): 345-358.
[5] YU Q，LEGROS M，MINOR A M. In situ TEM nanomechanics [J]. MRS Bulletin, 2015, 40(1): 62-70.
[6] CHEN L J，WU W W. In situ TEM investigation of dynamical changes of nanostructures [J]. Materials Science and Engineering R - Reports, 2010, 70(3): 303-319.
[7] STRECKER A, BäDER U, KELSCH M, et al. Progress in the preparation of cross-sectional TEM specimens by ion-beam thinning [J]. International Journal of Materials Research, 2003, 94(3): 290-297.
[8] SOUSA A A，LEAPMAN R D. Development and application of STEM for the biological sciences [J]. Ultramicroscopy, 2012, 123: 38-49.
[9] SCHORB M, HABERBOSCH I, HAGEN W J H, et al. Software tools for automated transmission electron microscopy [J]. Nature Methods, 2019, 16(6): 471-477.
[10] QI M, HUANG Z, ZHENG H, et al. Layer-dependent NO2-sensing performance in MoS2 for room-temperature monitoring [J]. ACS Applied Nano Materials, 2023, 6(11): 9290-9297.
[11] LI P, JIANG R, ZHAO L, et al. Cation defect mediated phase transition in potassium tungsten bronze [J]. Inorganic Chemistry, 2021, 60(23): 18199-18204.
[12] 马龙辉, 孟爽, 蒋仁辉, 等. α-Fe2O3的原子尺度界面结构与演变 [J]. 电子显微学报, 2021, 40(5): 529-536.
[13] 胡捷, 王佳恒, 李凯旋, 等. 超薄SnO2/Ga2O3/SnO2异质结构的原子尺度表征与热稳定性研究 [J]. 电子显微学报, 2022, 41(4): 399-406.
[14] 蒋仁辉, 李佩, 林泽圣, 等. 无序超结构Na5W14O44的制备与原子尺度表征 [J]. 电子显微学报, 2022, 41(4): 413-427.
[15] 蒋仁辉, 李佩, 孟琪, 等. 金属氧化物半导体的界面结构表征与调控 [J]. 中国材料进展, 2023, 42(3): 198-204.
[16] MENG Q, ZHUANG Y, JIANG R, et al. Atomistic observation of desodiation-induced phase transition in sodium tungsten bronze [J]. The Journal of Physical Chemistry Letters, 2021, 12(12): 3114-3119.
[17] CHEN Y, LIU S, TONG P, et al. A small-dataset-trained deep learning framework for identifying atoms on transmission electron microscopy images [J]. Scientific Reports, 2023, 13(1): 2631.
[18] WEI J, FENG B, ISHIKAWA R, et al. Direct imaging of atomistic grain boundary migration [J]. Nature Materials, 2021, 20(7): 951-955.
[19] VON HARRACH H S. Instrumental factors in high-resolution FEG STEM [J]. Ultramicroscopy, 1995, 58(1): 1-5.
[20] MULLER A，GRAZUL J. Optimizing the environment for sub-0.2 nm scanning transmission electron microscopy [J]. Journal of Electron Microscopy, 2001, 50(3): 219-226.
[21] SANG X, LUPINI A R, UNOCIC R R, et al. Dynamic scan control in STEM: Spiral scans [J]. Advanced Structural and Chemical Imaging, 2016, 2(1): 6.
[22] LI X, DYCK O, KALININ S V, et al. Compressed sensing of scanning transmission electron microscopy (STEM) with nonrectangular scans [J]. Microscopy and Microanalysis, 2018, 24(6): 623-633.
[23] JONES L, YANG H, PENNYCOOK T J, et al. Smart align—A new tool for robust non-rigid registration of scanning microscope data [J]. Advanced Structural and Chemical Imaging, 2015, 1(1): 8.
[24] OPHUS C，CISTON J，NELSON C T. Correcting nonlinear drift distortion of scanning probe and scanning transmission electron microscopies from image pairs with orthogonal scan directions [J]. Ultramicroscopy, 2016, 162: 1-9.
[25] BARCENA-GONZALEZ G, GUERRERO-LEBRERO M P, GUERRERO E, et al. CDrift: An algorithm to correct linear drift from a single high-resolution STEM image [J]. Microscopy Microanalysis, 2020, 26(5): 913-920.
[26] WANG Y, SUYOLCU Y E, SALZBERGER U, et al. Correcting the linear and nonlinear distortions for atomically resolved STEM spectrum and diffraction imaging [J]. Microscopy, 2018, 67: i114-i122.
[27] JONES L，NELLIST P D. Identifying and correcting scan noise and drift in the scanning transmission electron microscope [J]. Microscopy and Microanalysis, 2013, 19(4): 1050-1060.
[28] NING S, FUJITA T, NIE A, et al. Scanning distortion correction in STEM images [J]. Ultramicroscopy, 2018, 184(Pt A): 274-283.
[29] BRAIDY N, LE BOUAR Y, LAZAR S, et al. Correcting scanning instabilities from images of periodic structures [J]. Ultramicroscopy, 2012, 118: 67-76.
[30] ISAKOZAWA S, TOMONAGA S, HASHIMOTO T, et al. High-precision image-drift-correction method for EM images with a low signal-to-noise ratio [J]. Microscopy, 2014, 63(4): 301-312.
[31] DHOUIBI S, NEILY S, YOUSSEF S, et al. A semi-coherent interface with nanoledges intersecting the free surface of an elastic half-space [J]. Philosophical Magazine, 2017, 97(4): 213-229.
[32] SANG X，LEBEAU J M. Revolving scanning transmission electron microscopy: Correcting sample drift distortion without prior knowledge [J]. Ultramicroscopy, 2014, 138: 28-35.
[33] DYCUS J H, HARRIS J S, SANG X, et al. Accurate nanoscale crystallography in real-space using scanning transmission electron microscopy [J]. Microscopy and Microanalysis, 2015, 21(4): 946-952.
[34] RECNIK A，MOBUS G，STURM S. IMAGE-WARP: A real-space restoration method for high-resolution STEM images using quantitative HRTEM analysis [J]. Ultramicroscopy, 2005, 103(4): 285-301.
[35] ZUO J M, SHAH A B, KIM H, et al. Lattice and strain analysis of atomic resolution Z-contrast images based on template matching [J]. Ultramicroscopy, 2014, 136: 50-60.
[36] YANKOVICH A B, BERKELS B, DAHMEN W, et al. Picometre-precision analysis of scanning transmission electron microscopy images of platinum nanocatalysts [J]. Nature Communications, 2014, 5(1): 4155.
[37] WANG Y, SALZBERGER U, SIGLE W, et al. Oxygen octahedra picker: A software tool to extract quantitative information from STEM images [J]. Ultramicroscopy, 2016, 168: 46-52.
[38] KALININ S V, OPHUS C, VOYLES P M, et al. Machine learning in scanning transmission electron microscopy [J]. Nature Reviews Methods Primers, 2022, 2(1): 11.
[39] LEE C H, KHAN A, LUO D, et al. Deep learning enabled strain mapping of single-atom defects in two-dimensional transition metal dichalcogenides with sub-picometer precision [J]. Nano Letters, 2020, 20(5): 3369-3377.
[40] RONNEBERGER O，FISCHER P，BROX T. U-Net: Convolutional networks for biomedical image segmentation. in medical image computing and computer-assisted intervention[M]. Springer, 2015: 234-241.
[41] LIN R, ZHANG R, WANG C, et al. TEMImageNet training library and AtomSegNet deep-learning models for high-precision atom segmentation, localization, denoising, and deblurring of atomic-resolution images [J]. Scientific Reports, 2021, 11(1): 5386.
[42] 乔祎, 李建伟, 刘刚, 等. 基于聚焦离子束技术制备金刚石/M(M=Cu,Al,AlN)双相复合材料HRTEM样品 [J]. 电子显微学报, 2016, 35(1): 53-57.
[43] 王雪丽, 张伟, 贾志宏, 等. 聚焦离子束技术制备与样品表面平行的TEM样品 [J]. 电子显微学报, 2013, 32(5): 420-425.
[44] GAO S, LIU Y, ZHAO D, et al. Effect of different Co contents on the microstructure and tensile strength of Mg-Co-Y alloys [J]. Materials Science and Engineering: A, 2019, 750: 91-97.
[45] JIANG R, LI P, GUAN X, et al. Na+ migration mediated phase transitions induced by electric field in the framework structured tungsten bronze [J]. The Journal of Physical Chemistry Letters, 2023, 14(13): 3152-3159.
