Abstract The effects of process history on the particle morphology of uranium ore concentrates (UOC) synthesized from natural uranium ores were investigated. A purified UOC was extracted from a uranium-rich ore by sulfuric acid leaching and purified by two commercial solvent extraction processes: the Dapex process using di-(2-ethylhexyl) phosphoric acid (DEHPA), and the Amex process using Alamine® 336. The UOC from either route was precipitated from its respective strip solution by ammonia and calcined. Powder X-ray diffraction (P-XRD) was used to confirm the crystallography of the UOC; while Rietveld refinement was performed for quantification of U oxide phase composition. Inductively coupled plasma - mass spectrometry (ICP-MS) was used to quantify impurity concentrations of the calcined material. As expected, different impurities were present due to processing via the Dapex and Amex process. To further correlate the impurities to the process history, particle morphology was quantified from scanning electron microscopy (SEM) micrographs using the MAMA segmentation software. In addition, a deep neural network was trained on the SEM images to distinguish between process histories. The results provide a real-world representation of how particle morphology will likely change based on processing at commercial facilities as opposed to high purity material experiments in a laboratory setting.

中文翻译：

---

工艺历史对从矿石中提取的铀精矿表面形貌的影响

摘要 研究了工艺过程对天然铀矿合成铀精矿 (UOC) 颗粒形态的影响。通过硫酸浸出从富含铀的矿石中提取纯化的 UOC，并通过两种商业溶剂提取工艺进行纯化：使用二-（2-乙基己基）磷酸 (DEHPA) 的 Dapex 工艺和使用 Alamine® 336 的 Amex 工艺。来自任一路线的UOC通过氨从其各自的汽提溶液中沉淀并煅烧。粉末 X 射线衍射 (P-XRD) 用于确认 UOC 的晶体学；而 Rietveld 精修用于量化 U 氧化物相组成。电感耦合等离子体质谱法 (ICP-MS) 用于量化煅烧材料的杂质浓度。正如预期的那样，由于通过 Dapex 和 Amex 工艺进行加工，存在不同的杂质。为了进一步将杂质与工艺历史相关联，使用 MAMA 分割软件从扫描电子显微镜 (SEM) 显微照片中量化颗粒形态。此外，在 SEM 图像上训练了一个深度神经网络，以区分过程历史。结果提供了基于商业设施处理而不是实验室环境中的高纯度材料实验的粒子形态可能如何变化的真实世界表现。在 SEM 图像上训练了一个深度神经网络，以区分过程历史。结果提供了基于商业设施处理而不是实验室环境中的高纯度材料实验的粒子形态可能如何变化的真实世界表现。在 SEM 图像上训练了一个深度神经网络，以区分过程历史。结果提供了基于商业设施处理而不是实验室环境中的高纯度材料实验的粒子形态可能如何变化的真实世界表现。