首 页
滚动信息 更多 >>
本刊2022年SCI影响因子9.7 (2023年6月发布) (2023-10-23)
本刊2021年SCI影响因子12.256 (2022-07-07)
npj Computational Materials 2019年影响因子达到9... (2020-07-04)
npj Computational Materials获得第一个SCI影响因... (2018-09-07)
英文刊《npj Computational Materials(计算材料学... (2017-05-15)
快捷服务
最新文章 研究综述
过刊浏览 作者须知
期刊编辑 审稿须知
相关链接
· 在线投稿
会议信息
友情链接
  中国科学院上海硅酸盐研究所
  无机材料学报
  OQMD数据库
近期文章
Generalization of the mixed-space cluster expansion method for arbitrary lattices
发布时间:2024-02-06

Generalization of the mixed-space cluster expansion method for arbitrary lattices 

Kang Wang, Du Cheng & Bi-Cheng Zhou

npj Computational Materials 9: 75 (2023).

编辑概述:混合空间团簇展开:跨越实空间和倒空间的新视界

在现代材料科学中,原子在母晶格上的占位分布至关重要,它不仅决定了合金热力学性质,还影响着材料由无序到有序的转变过程,进而关系到材料的相稳定性及其它性质。团簇展开(CE)方法结合密度泛函理论(DFT)能够精确地描述这些现象,尽管理论上需要无限多的团簇,实际中简单团簇已足以实现高精度的能量参数化。针对长程应变相互作用的精确建模,Laks等人提出了混合空间团簇展开(MSCE)方法,这一方法通过兼顾实空间和倒空间,使得对复杂合金体系中短程和长程相互作用的描述变得可能。美国弗吉尼亚大学材料科学与工程系的Kang Wang博士和Bi-Cheng Zhou博士领导的团队,提出了一种混合空间团簇展开(MSCE)方法。该方法通过构建兼容的倒空间相互作用,并与一个晶体对称性不可知算法结合,用于计算组分应变能量,从而将MSCE方法推广到了具有多子晶格的系统。

MSCE方法之所以比实空间CE方法具有更高的精度,归因于三个关键方面:(1)首先,通过k空间的形式化处理,MSCE方法明确地包含了由尺寸不匹配产生的组分应变能(CSE)的长程极限。(2)其次,长程相互作用的衰减考虑了中程结构,这在训练集中占多数。(3)最后,实空间有效团簇相互作用(ECIs)的正则化实现了包括更大数量的团簇,从而提升了拟合能力。利用这一方法,作者在假定的六方密堆(HCP)系统和镁-锌(Mg-Zn)合金中进行了验证。在Mg-Zn合金体系中,Zn原子更倾向于沿着[0001]方向的锌棒和局域的C14MgZn2结构来排列。值得关注的是,C14MgZn2棒与实验中的Mg-Zn体系中沿[0001]方向成长的峰龄沉淀相β1棒具有一致的取向关系。在目前的MSCEMC模拟中,包含了所有在DFT中完全弛豫HCP排序,使得MSCE预测的能量可以全面反映HCP晶格位置的弛豫过程。与此相比,当只考虑中等程度的弛豫结构(位移量d < 0.1)时,实空间CE将排除许多弛豫过度的结构。这样的计算体现了非常接近HCP晶格位点的排序能量,并揭示了类似吉尼尔-普雷斯顿区(Guinier-Preston zoneGP)的相干结构。然而,这些潜在的GP区域在目前的MSCEMC模拟中并未观察到。尽管对镁锌合金中GP区的实验证据已有报道,但这些证据被认为并不充分。实空间CEMSCE的计算对比为此提供了可能的解释:当锌原子开始在镁基局域区域聚集时,由于镁和锌原子之间的晶格不匹配,会涉及晶格畸变。在畸变幅度较小,原子位置接近理想HCP晶格位点的情况下,可以发现类似GP区的锌原子局部排列。随着局部锌浓度的增加和晶格畸变的加剧,富锌区域更倾向于转变为更稳定的结构,如C14C15MgZn2。考虑到温度升高使得晶格变得更加柔软,GP区在老化温度较低且整体锌浓度较低的样品中更可能存在。这或许可以解释为何即便利用现代显微技术,Mg-Zn合金中GP区的确切证据依旧难以捕捉。

Editorial Summary: Mixed-Space Cluster Expansion: A New Perspective Across Real and Reciprocal Space

In modern materials science, the occupancy distribution of atoms on the parent lattice is crucial, which not only determines the thermodynamic properties of alloys, but also influences the transition process from disorder to order, which in turn relates to the phase stability and other properties of materials. The cluster expansion (CE) method combined with density functional theory (DFT) is able to accurately describe these phenomena, and although an infinite number of clusters is theoretically required, in practice, simple clusters are sufficient for high-precision energy parameterization. For the accurate modeling of long-range strain interactions, Laks et al. proposed the mixed-space cluster expansion (MSCE) method, which makes it possible to characterize both short- and long-range interactions in complex alloy systems by taking into account both real and inverse space. A team lead by Dr. Kang Wang and Dr. Bi-Cheng Zhou from Department of Materials Science and Engineering, University of Virginia, USA, proposed a mixed-space cluster expansion (MSCE) method, which is generalized to systems with multiple sublattices by formulating compatible reciprocal space interactions and combined with a crystal-symmetry-agnostic algorithm for the calculation of constituent strain energy. The MSCE method owes the high accuracy, compared with r-space CE, to three aspects. (1) The long-ranged limit of the constituent strain energy (CSE) due to size-mismatch are explicitly incorporated using the k-space formalism. (2) The attenuation of the long-ranged interactions accommodates for medium-ranged structures, which is the case for the majority of the structures in the training set. (3) The regularization of r-space effective cluster interactions (ECIs) allow us to include much larger number of clusters, which enhances the fitting capability. This generalized approach is demonstrated in a hypothetical HCP system and Mg-Zn alloys. In Mg-Zn system, the preferred orderings of Zn atoms are identified with Zn-rods and local C14 MgZn2 arranged along [0001]. Noteworthy, C14 MgZn2 rods here agree with the peak-age precipitates β1’ rods along [0001]α with the orientation relationship in Mg-Zn system. In the current MSCE and MC, all the fully relaxed HCP orderings in DFT are included in the training set, which enables the energy predicted by MSCE to be fully incorporate the relaxations from HCP lattice sites. In the calculations with only moderately relaxed structures (with d < 0.1), many overly relaxed structures are excluded in r-space CE. Such calculation reflects the energies of orderings very close to HCP lattice sites and coherent orderings resembling GP zones were revealed. However, such potential GP zones were not found in the current MSCE and MC simulations. In Mg-Zn system, experimental evidences of GP zones were reported, which, however, is deemed insufficient. Comparison of the calculations using r-space CE and MSCE offers a possible explanation. When Zn atoms start to aggregate in local regions of Mg matrix, lattice distortion will be involved due to the lattice mismatch between Mg and Zn atoms. When such distortions are small and the atoms sit close to the ideal HCP lattice sites, local arrangements of Zn atoms resembling GP zones can be found. When the local Zn concentration gets high and local lattice distortion becomes severe, the local Zn-rich regions prefer to transform to more stable structures, such as C14 or C15 MgZn2. Considering the lattice becomes less rigid when the temperature is elevated, the GP zones are more likely to be found at low aging temperatures in the samples with low overall Zn concentrations. This may explain why concrete evidence of GP zones in Mg-Zn is elusive even with modern microscopy.

 
【打印本页】【关闭本页】
版权所有 © 中国科学院上海硅酸盐研究所  沪ICP备05005480号-1    沪公网安备 31010502006565号
地址:上海市长宁区定西路1295号 邮政编码:200050