Grain Boundary Embrittlement Genome

该数据集是由麻省理工学院的研究人员创建的，包含了FCC和BCB二元合金的Σ5[001](210)晶界 segregation 和 embrittlement 的信息。数据集涵盖了15种基体金属系统和75种溶质元素，总计超过1000种组合。数据集的创建是为了服务合金设计和开发，特别是为了克服缺乏一致数据集的挑战。数据集中的晶界能量计算与DFT数据集相匹配，为合金设计提供了一个设计工具。This dataset was created by researchers from the Massachusetts Institute of Technology (MIT), containing information on grain boundary segregation and embrittlement of FCC and BCC binary alloys with Σ5[001](210) grain boundaries. It covers 15 matrix metal systems and 75 solute elements, totaling over 1000 unique combinations. This dataset was developed to support alloy design and development, specifically to address the challenge of the lack of consistent datasets. The grain boundary energy calculations included in this dataset align with Density Functional Theory (DFT) datasets, serving as a reliable design tool for alloy development.

搜集汇总

数据集介绍

构建方式

Grain Boundary Embrittlement Genome 数据集通过使用通用原子间势能，计算了面心立方（FCC）和体心立方（BCC）二元合金的Σ5[001](210)晶界的晶界偏析和脆化基因组。该数据集涵盖了15种基体金属系统（Ag、Al、Au、Cr、Cu、Fe（包括BCC和FCC）、Mo、Nb、Ni、Pd、Pt、Rh、Ta、V和W）与每种金属相对应的75种溶质元素的组合，共计约1,000种二元合金。数据集的构建采用了机器学习和人工智能的最新进展，利用开放材料数据集和EquiformerV2（eqV2）模型进行计算，从而在保证高精度的同时，大大减少了计算成本。

特点

Grain Boundary Embrittlement Genome 数据集的特点在于其规模庞大、数据一致性高，为合金设计提供了有力的工具。该数据集不仅提供了晶界偏析的能量数据，还包含了晶界脆化的能量数据，为合金的晶界脆化预测提供了重要的参考。此外，数据集还考虑了溶质元素在晶界处的偏析和脆化行为，以及溶质元素在合金中的溶解度限制，为合金的晶界脆化预测提供了更全面的信息。

使用方法

Grain Boundary Embrittlement Genome 数据集的使用方法包括但不限于以下几个方面：首先，可以通过该数据集来评估不同合金的晶界脆化潜能，为合金设计提供参考；其次，可以用于研究晶界偏析和脆化行为，深入理解晶界脆化的机理；最后，该数据集还可以用于开发新的合金设计框架，加速合金的研发过程。

背景与挑战

背景概述

晶界化学对金属和合金的性质起着至关重要的作用，然而，用于合金设计和开发的一致性数据集却相对缺乏。随着人工智能和机器学习在材料科学领域的兴起，开放的材料模型和数据集可以克服这些挑战。本研究使用一种通用的原子间势能，计算了FCC和BCC二元合金的Σ5[001](210)晶界的晶界偏析和脆化基因组。这里计算的晶界数据库作为高角度晶界脆化设计工具，涵盖了15种基本金属系统（Ag、Al、Au、Cr、Cu、Fe（BCC和FCC）、Mo、Nb、Ni、Pd、Pt、Rh、Ta、V和W）和每种金属的75种溶质元素。

当前挑战

晶界偏析与晶界脆化密切相关，然而许多合金元素并不脆化，或者根本不偏析。一些溶剂（宿主）和晶界偏析体的组合会导致晶界强化，或者提供其他有益的性质，如热稳定性、机械性能的提高。因此，成功的合金设计需要精细理解晶界偏析和脆化。近年来，在理解偏析问题方面取得了显著进展，展示了控制多晶环境中GBs中所有原子位点的偏析的热力学量的大图集。然而，脆化问题的脆化部分仍然是一个领域，许多合金还没有可用于计算合金设计框架的大量一致数据。最近试图汇总已发表的数据集，说明了在具有各种方法生成数据之间进行交叉比较的挑战。此外，评估GB脆化潜力的方法基于GB薄片方法，这通常需要大量的计算资源。因此，用于计算合金设计框架的GB偏析和脆化数据有限。本研究利用人工智能和机器学习的最新进展，提供了一种大约1,000种二元合金的大规模、自洽的GB脆化潜力分析。我们利用开放材料数据集和EquiformerV2（eqV2）模型，以比密度泛函理论（DFT）低得多的成本实现了与DFT的非常准确的结果，并在Matbench Discovery上进行了基准测试。具体来说，我们应用了预训练有小型模型（31M参数）的Open Materials数据集，并使用MPtrj和Alexandria数据集进行了微调（eqV2_31M_omat_mp_salex），以计算偏析倾向和脆化潜力。我们考虑了15种基本金属：Ag、Al、Au、Cr、Cu、Fe（BCC和FCC）、Mo、Nb、Ni、Pd、Pt、Rh、Ta、V和W，每种金属与75种溶质元素结合。我们使用GB_code构建了一个Σ5[001](210) GB薄片模型，FCC金属使用2×2×2重复超胞，BCC金属使用3×2×2重复超胞。GB的弛豫使用Fast Inertial Relaxation Engine（FIRE）最小化算法进行，力容忍度为0.01 eV/Å。本研究中的分析使用来自参考文献44-55的Python软件包进行。我们提供了晶格常数、超胞大小、系统大小及其相应的GB能量的值，并与文献中报道的GB能量进行了比较。我们还展示了BCC Fe和FCC Al晶界的示例结构，其中使用了所有4个位点进行偏析计算。这里计算的GB能量与DFT数据集合理匹配，见表I。

常用场景

经典使用场景

Grain boundary chemistry is pivotal in determining the properties of metals and alloys, yet the availability of consistent datasets for alloy design and development has been limited. The Grain Boundary Embrittlement Genome dataset addresses this gap by providing a comprehensive resource for understanding grain boundary segregation and embrittlement in a variety of alloys. This dataset is particularly useful for researchers and engineers in the field of materials science, who can leverage it to design alloys with improved mechanical properties, such as strength and ductility, by avoiding or promoting specific grain boundary segregations. By analyzing the segregation tendency and embrittlement potency of various elements at grain boundaries, this dataset enables the prediction and mitigation of embrittlement, leading to the development of more reliable and durable materials.

衍生相关工作

The Grain Boundary Embrittlement Genome dataset has inspired a range of related research and development activities. It has been used as a foundation for further studies on grain boundary segregation and embrittlement, leading to the development of new models and methodologies. For instance, researchers have used the dataset to validate and refine existing models, as well as to develop new machine learning models for predicting grain boundary properties. The dataset has also been utilized in the design and optimization of materials for specific applications, such as high-strength steels and lightweight alloys. Additionally, the dataset has been employed in educational settings to teach students about the importance of grain boundary chemistry in materials science and engineering. Overall, the Grain Boundary Embrittlement Genome dataset has had a significant impact on the field of materials science, leading to advancements in both fundamental understanding and practical applications.

数据集最近研究

最新研究方向

Grain boundary chemistry holds a pivotal role in determining the properties of metals and alloys, with grain boundary segregation being a critical factor influencing embrittlement. The Grain Boundary Embrittlement Genome dataset, developed by Tuchindaa et al., leverages advanced computational methods, including machine learning and atomistic simulations, to systematically investigate the segregation and embrittlement tendencies of various binary alloys. This dataset offers a comprehensive understanding of how different solute elements interact with grain boundaries in 15 base metals, providing insights into the design of alloys with improved mechanical properties. By employing a universal interatomic potential, the dataset facilitates the computation of grain boundary segregation and embrittlement at a fraction of the computational cost of density functional theory (DFT), thereby enabling high-throughput studies. This research is particularly significant as it bridges the gap in the availability of consistent and large-scale datasets for alloy design, which is crucial for advancing the field of materials science and engineering. The dataset is poised to impact the development of high-strength steels, aluminum alloys, refractory materials, and nanocrystalline alloys, contributing to the enhancement of their thermal stability and mechanical performance.

相关研究论文

• A Grain Boundary Embrittlement Genome for Substitutional Cubic Alloys

  麻省理工学院 · 2025年