Yuan Yu, Demers Hendrix, Brodusch Nicolas, Wang Xianglong, Gauvin Raynald
Department of Mining and Materials Engineering, McGill University, 3610 Rue University, Montreal, Québec, Canada, H3A 0C5.
Centre d'excellence en électrification des transports et stockage d'énergie, Hydro- Québec, 1806 Boulevard Lionel-Boulet, Varennes, Québec, Canada, J3X 1S1.
Ultramicroscopy. 2020 Dec;219:113117. doi: 10.1016/j.ultramic.2020.113117. Epub 2020 Sep 20.
Current quantitative X-ray microanalysis methods are only available for homogeneous materials. This paper presents a newly developed inverse modeling algorithm to determine both the structure and composition of two-dimensional (2D) heterogeneous materials from a series of X-ray intensity measurements under different beam energies and beam positions. It utilizes an iterative process of forward modeling to determine the optimal specimen to minimize the relative differences between the simulated and experimental characteristic X-ray intensities. The Monte Carlo method is used for the forward modeling to predict the X-ray radiation for a given specimen and experimental setup. Several examples of applications are presented for different types of samples with one-dimensional (1D) and 2D structures, in which the simulated X-ray intensities from phantom samples are used as input. Most of the results obtained from our algorithm agree well with the phantom samples. Some discrepancies are found for the voxels located at deeper depths of the 2D samples. And the discrepancies may be attributed to errors from the Monte Carlo simulations and from the variation of the X-ray range with beam energy. As a proof-of-concept work, this paper confirms the feasibility of our inverse modeling algorithm applied to 2D heterogeneous materials.
当前的定量X射线微分析方法仅适用于均质材料。本文提出了一种新开发的反演建模算法，用于根据在不同束能量和束位置下的一系列X射线强度测量结果，确定二维（2D）非均质材料的结构和组成。它利用正向建模的迭代过程来确定最佳样本，以最小化模拟和实验特征X射线强度之间的相对差异。蒙特卡罗方法用于正向建模，以预测给定样本和实验设置下的X射线辐射。针对具有一维（1D）和二维结构的不同类型样本给出了几个应用示例，其中将虚拟样本的模拟X射线强度用作输入。我们算法得到的大多数结果与虚拟样本吻合良好。在二维样本较深深度处的体素中发现了一些差异。这些差异可能归因于蒙特卡罗模拟的误差以及X射线范围随束能量的变化。作为概念验证工作，本文证实了我们的反演建模算法应用于二维非均质材料的可行性。
