DMREF: Real Time Control of Grain Growth in Metals

该项目正在开发一种新的方法，通过集成模拟、实时测量、控制算法以及模拟和实验数据集，自适应地控制热加工过程中材料微观结构的演变。具体实现重点是多晶金属热加工过程中晶粒组织的空间控制。利用蒙特卡罗模拟和相场模拟预测了微观结构演化的轨迹，并利用扫描电子显微镜方法实时测量了局部微观结构、晶体学和温度，对这些轨迹进行了实验跟踪。当测量结果显示与计划轨迹有明显偏差时，通过控制算法模拟并实现新的轨迹。该实验实现包括一个多区电阻加热阵列，该阵列能够通过宏观多晶金属样品的原位扫描电子显微镜对受控温度分布进行编程。利用二次电子成像和电子背散射衍射（EBSD）来监测晶粒结构和晶体学的演变，同时利用EBSD模式的漫射散射量化方法来测量相同电子束下的局部温度分布。由此产生的实时数据流与模拟预测进行比较，当偏差出现时，调整样品中的温度分布以引导微观结构演变回到所需状态。实现这一目标的控制算法是基于线性化相场模型开发的。当测量和模拟之间的偏差过大时，需要在中间状态和最终状态之间使用全蒙特卡罗模拟进行重新优化。随着这些方法的发展，随着相关实验和模拟数据集库的发展，期望在不过度使用这种重新优化的情况下实现反馈控制。虽然这些方法固有地测量表面晶粒参数的二维分布，但集成聚焦离子束层析成像用于将表面晶粒微观结构与体内的三维分布联系起来，并与二维？模拟的三维相关性。该项目侧重于开发一种主要的新材料加工控制能力：在热加工过程中监测、评估和控制系统的局部材料结构，以获得最佳性能的能力。这是通过集成实验、模拟和数据管理的新方法实现的。具体项目实施包括多晶金属退火过程中的晶粒生长控制。然而，开发的方法应该扩展到加速广泛的工艺开发周期，其中特定的内部材料结构已知与最佳材料特性和性能相关。该研究项目与一系列广泛的教育和推广活动相结合。高中生、本科生和研究生都参与了整个项目。特别是，通过这项研究培养的多名研究生正在成为实验、模拟和数据科学综合应用以加速材料加工技术的专家。这使他们为未来在新材料制造技术开发方面的领导地位做好准备。

This project is developing a new methodology to control adaptively the evolution of materials microstructure during thermal processing, through integration of simulation, real-time measurement, control algorithms, and simulated and experimental data sets. The specific implementation focuses on the spatial control of grain microstructure during thermal processing of polycrystalline metals. Monte Carlo and phase field simulations are used to predict the trajectory of microstructure evolution, and real-time measurements of local microstructure, crystallography and temperature using scanning electron microscopy methods are used to track these trajectories experimentally. When measurement shows significant deviation from the planned trajectory, a new trajectory is simulated and implemented through control algorithms. The experimental implementation comprises a multi-zone resistive heating array that enables controlled temperature distributions to be programmed across a macroscopic polycrystalline metal sample in-situ to a scanning electron microscope. Secondary electron imaging and electron backscattered diffraction (EBSD) are employed to monitor the evolution of grain structure and crystallography, while a new method, based upon quantification of diffuse scattering in the EBSD patterns, is employed to measure local temperature distributions using the same electron beam. The resulting real time data streams are compared to the predictions from simulation, and as deviations emerge, the temperature distribution across the sample is adjusted to steer micro-structural evolution back towards the desired state. The control algorithms employed to achieve this are developed based on the linearized phase field models. When the deviations between measurement and simulation become excessive, re-optimization using full Monte-Carlo simulation between intermediate and final states is required. As these methods develop, implementation of feedback control without excessive use of such re-optimization is expected as the libraries of correlated experimental and simulated data sets develop. While these methods inherently measure two-dimensional distributions of grain parameters at the surface, integrated focused ion beam tomography is used to correlate the surface grain microstructure to three dimensional distributions within the bulk, and also to compare to two dimensional ? three dimensional correlations from the simulations.This project focuses on developing a major new capability for control of materials processing: the ability to monitor, evaluate and control the local materials structure of a system as it evolves during thermal processing in order to obtain optimal properties. This is enabled through new methodologies for integrating experiment, simulation, and data management. The specific project implementation comprises control of grain growth during annealing of polycrystalline metals. However, the methodologies developed should be extendible to acceleration of a wide range of process development cycles where specific internal materials structures are known to correlate to optimal materials properties and performance. The research project is integrated with a broad set of educational and outreach activities. High school, undergraduate and graduate students are involved throughout the project. In particular, multiple graduate students who are trained through this research are becoming expert in the integrated application of experimental, simulation and data sciences to the acceleration of materials processing technologies. This is preparing them for future leadership in the development of new materials manufacturing technologies.