金属学报, 2024, 60(10): 1379-1387 DOI: 10.11900/0412.1961.2024.00147

综述

叠熵理论:从材料基因到材料性能

廖名情1, 王毅,2, 王义3, 商顺利3, 刘梓葵,3

1 江苏科技大学 材料科学与工程学院 镇江 212100

2 西北工业大学 凝固技术国家重点实验室 西安 710072

3 Department of Materials Science and Engineering, the Pennsylvania State University, University Park, PA, 16802, USA

Zentropy Theory: Bridging Materials Gene to Materials Properties

LIAO Mingqing1, WANG William Yi,2, WANG Yi3, SHANG Shun-Li3, LIU Zi-Kui,3

1 School of Materials Science and Engineering, Jiangsu University of Science and Technology, Zhenjiang 212100, China

2 State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi'an 710072, China

3 Department of Materials Science and Engineering, the Pennsylvania State University, University Park, PA, 16802, USA

通讯作者: 王 毅,wywang@nwpu.edu.cn,主要从事极端条件先进材料的材料基因工程&集成计算材料工程的研究;刘梓葵,zxl15@psu.edu,主要从事材料热力学、动力学、晶体学以及材料基因组的研究

收稿日期: 2024-05-08   修回日期: 2024-07-05  

基金资助: 江苏省自然科学基金项目(BK20230673)
江苏省双创博士人才项目(JSSCBS20221270)

Corresponding authors: WANG William Yi, professor, Tel:(029)88460294, E-mail:wywang@nwpu.edu.cn;LIU Zi-Kui, professor, Tel:(814)8651934, E-mail:zxl15@psu.edu

Received: 2024-05-08   Revised: 2024-07-05  

Fund supported: Natural Science Foundation of Jiangsu Province(BK20230673)
Doctor of Entrepreneurship and Innovation of Jiangsu Province(JSSCBS20221270)

作者简介 About authors

廖名情,男,1992年生,博士

摘要

熵是科学中一个非常重要的概念,从量子到天文,无处不在。基于统计力学,综合量子力学密度泛函理论和热力学,刘梓葵教授团队建立了关于系统总熵的理论:叠熵理论。该理论以Gibbs统计力学中的微观组态作为材料的基因,把密度泛函理论中的基态作为最基本的微观组态,通过基态的内部自由度遍历所有的微观组态。叠熵理论定义系统总熵为每个微观组态熵的加权平均再加上微观组态之间的Gibbs统计熵。本文系统介绍了叠熵理论的基础方程与原理,简单概述了叠熵理论的典型应用,包括磁性转变、铁电转变、热膨胀机制以及临界现象预测,并对叠熵理论从理论发展、软件生态构建、高通量计算以及与人工智能集成等方面进行了展望。

关键词: 叠熵理论; 热力学; 材料基因工程; 热膨胀机制; 铁电转变温度; 有序-无序转变

Abstract

Entropy is an important concept in science and is ubiquitous from quantum to astronomy. By integrating statistical mechanics, quantum mechanics, and thermodynamics, Professor Zi-Kui Liu proposed the zentropy theory, which stacks entropy over configurations. The zentropy theory takes the configurations in Gibbs' statistical mechanics of a given ensemble as the material gene with the ground state as the basic configuration and additional configurations ergodically derived from its internal degrees of freedom. In the zentropy theory, the total entropy of a system is defined as the weighted average of the entropy of each configuration plus the statistical entropy among all configurations. In this paper, the basic equations and principles of the zentropy theory are introduced, and their typical applications, including magnetic and ferroelectric transformations, thermal expansion mechanisms, and critical phenomenon prediction are outlined. Furthermore, a perspective on the development of this theory, software ecosystems, high-throughput computing, and integration with artificial intelligence is provided in this study.

Keywords: zentropy theory; thermodynamics; materials genome engineering; thermal expansion mechanism; ferroelectric transition temperature; order-disorder transition

PDF (1381KB) 元数据 多维度评价 相关文章 导出 EndNote| Ris| Bibtex  收藏本文

本文引用格式

廖名情, 王毅, 王义, 商顺利, 刘梓葵. 叠熵理论:从材料基因到材料性能[J]. 金属学报, 2024, 60(10): 1379-1387 DOI:10.11900/0412.1961.2024.00147

LIAO Mingqing, WANG William Yi, WANG Yi, SHANG Shun-Li, LIU Zi-Kui. Zentropy Theory: Bridging Materials Gene to Materials Properties[J]. Acta Metallurgica Sinica, 2024, 60(10): 1379-1387 DOI:10.11900/0412.1961.2024.00147

新材料和信息学是先进制造业的2大“底盘技术”,是高端装备和重大工程不可或缺的基础[1,2]。而基于传统的“试错法”开发新材料成本高,研发周期长。面对这一难题,美国在2011年发起了材料基因组计划(materials genome initiative,MGI)[3,4],该计划旨在将高通量实验、高通量计算及数字化数据(数据库及信息学)有机整合,缩短新材料研发周期,降低新材料研发风险和成本[4]。2014年美国更是将MGI提升为“国家战略”[5,6]。2018年,基于近年来集成计算材料工程[7~9]的高速发展,美国航空航天局发布了《2040愿景:材料和系统的集成、多尺度建模和仿真路线图》[10],该愿景旨在整合材料多尺度模拟相关的9大关键要素,创建一个信息-物理-社会一体化的数字化材料研究范式。2022年,美国国家科学院发布了MGI发展前10年的评估报告[11],该报告指出材料数据在材料设计中越来越占据中心地位,机器学习(machine learning,ML)及人工智能(artificial intelligence,AI)在未来会彻底改变材料的设计与开发,目前与MGI相关的项目中33%的受资助项目及45%的在研项目都涉及了机器学习。

美国提出MGI后,中国随后也建立了相应的计划:材料基因工程(materials genome engineering,MGE)[6,12,13]。中国在2011年底就对MGI进行了讨论,2012年启动了“材料基因组计划”重大咨询项目,2015年编制了《材料基因工程关键技术和支撑平台重点专项实施方案》,全面开启了材料基因工程研究。在该专项的支持下,我国材料基因工程在以下方面取得了较大的进展:(1) 新材料开发与优化(如:具有优异机械性能的高温块体金属玻璃[14,15]);(2) 关键技术与装备研发(如:高通量工作流计算软件ALKEMIE[16,17],材料基因工程专用数据库平台MGEData);(3) 工程化应用示范(如:建立了重型燃气轮机专用材料数据库,推动了重型燃气轮机的数字化、智能化设计和制造[12]);(4) 创新平台建设(如:北京材料基因工程高精尖创新中心、上海大学材料基因组工程研究院、西北工业大学材料信息学与基因工程研究中心、哈尔滨工业大学(深圳)材料基因工程及大数据研究院、北京航空航天大学集成计算材料工程中心、云南稀贵金属材料基因工程创新平台[18]);(5) 国内外学术交流(如:材料基因工程高层论坛,创建了专门的刊物:Journal of Materials InformaticsMGE Advances)。此外,材料基因工程大大的提高了计算方法和工具与数据库技术在材料学中的地位,由此催生了材料设计的第四范式:数据驱动的材料设计范式[19,20]和科学研究第五范式:人工智能驱动的科学(AI for Science,AI4S)[21]或智能化科研(AI for Research,AI4R)[22]

尽管AI4S基于数据,但并不意味着AI4S是一个纯粹数据驱动的研究范式,事实上,AI4S是一个用人工智能方法为科学研究所遵循的物理规律/原理开发更高效算法或近似模型的过程,这样才能保证其可拓展性[21]。因此,在材料设计中,找到“材料基因”对材料领域中的AI4S具有重要意义。然而,虽然材料基因工程取得了显著的发展,但是关于何为材料基因,目前尚无明确的定义[23,24]。由于材料基因源于CALPHAD方法[25~28],因此Campbell等[29]提出将材料中的“相”作为“材料基因”;王绍青和叶恒强[24]认为晶体材料基因最确切的表现形式应该为该晶体相的晶体学原胞,这与以“相”为“材料基因”有异曲同工之处。但当材料体系的相中存在更小的微观组态(configuration)时,在温度涨落下,不同的微观组态之间会发生相互转化,而其晶体结构不一定发生变化,如磁性材料的不同自旋组态之间的转变。基于上述研究成果,刘梓葵教授[3]认为单个相内的不同构型就应该成为单个相的基本构成模块,即材料基因。通过与以“相”为基因的第一代材料基因组比较可以发现,在新一代材料基因工程的概念中,需要包含“相”的微观组态信息,为此,刘梓葵教授团队[30~32]提出了叠熵理论(zentropy theory),通过考虑多组态熵的贡献,架起从材料基因到材料性质的“桥梁”(即材料基因的表达决定材料本征性质),以完全第一性原理(无经验参数)计算实现多种性能的预测。

1 叠熵理论

熵是统计力学、热力学、物理化学、材料学等自然学科中一个重要而基础的概念。基于该概念发展出了进行材料设计的“熵工程”方法,设计了多种性能优异的材料,如高熵合金[33]、高熵热电材料[34]、高熵碳化物陶瓷[35,36]、高熵铁弹陶瓷[37]、高熵催化剂[38]等。在经典统计力学上,对熵的表达主要有2种,其一为在假设每个微观组态概率均相等的情况下,Boltzmann通过统计力学推导出:体系的熵(S)可以通过体系当前尺度下的组态数目(Ω)来表达,即Boltzmann熵(kBlnΩ,其中kB为Boltzmann常数);随后该表达被推广到每个组态k概率(pk)不相等的一般情况,即Gibbs熵(-kBkpklnpk)。上述对熵的表达仅局限于当前体系中微观组态之间的统计分布,而熵在不同尺度是普遍存在的,如图1所示,当前体系的每个微观组态中又包含许多亚微观组态,因此当前体系的每个微观组态也有其自身的熵。而Boltzmann熵和Gibbs熵仅表征了系统当前尺度不同微观组态之间的统计熵,没有考虑每个微观组态本身所含的熵,因此如何将亚尺度熵的贡献考虑进来,亚尺度的熵对当前尺度性能有何影响等是当前亟待解决的难题。

图1

图1   多尺度熵示意图

Fig.1   Schematic of entropy over multiscale


针对上述的问题,刘梓葵教授团队[30]经过在材料热力学与材料动力学领域长期的思考与研究,于近期受邀在Journal of Phase Equilibria and Diffusion上发表了题为Zentropy theory for positive and negative thermal expansion的文章,在文中正式提出了一种通用的计算材料体系多尺度熵的理论框架:叠熵理论(zentropy = z + entropy,其中字母z源于Max Planck为配分函数创造的德语词Zusstandssumme,意为状态总和,也通常用于表示配分函数,该名字是刘梓葵教授在杜克大学讲座时由Josiah Roberts建议的。Zentropy即意为不同尺度/状态熵的叠加之和,因此中文译名为叠熵)。该理论一经发表,得到了Phys.org、ScienceDaily、SciTechDaily、AZOMaterials、量子认知百家号等著名科技网站的报道,量子认知百家号给出“综熵理论将是华人科学家们在传统的基础物理理论,特别是热力学基础科学领域作出的重要贡献” (该文中,作者将zentropy译为“综熵”)的高度评价。

在叠熵理论中(如图2所示),体系的熵可由2部分求和而得,即当前尺度系统的统计熵以及每个微观组态的熵(Sk)。当前尺度的统计熵可由统计力学表达,即Gibbs熵(-kBkpklnpk),而微观组态的总熵可由每个组态k熵(Sk)求和得到,即kpkSk,因此叠熵理论中熵(Szentropy)可由 式(1)表达。而每个微观组态可以继续划分为更精细的微观组态,依此递归,直到纯量子态(Sk=0)时,此时 式(1)退化成Gibbs熵形式。

图2

图2   叠熵理论示意图

Fig.2   Schematic of zentropy theory (kB—Boltzmann constant; pk and pi —probabilities of configuration k and its sub-configuration i; Sk and Ski —entropies of configuration k and its sub-configuration i)


Szentropy=kpkSk-kBkpklnpk

在叠熵理论中,各个微观组态的配分函数(partition function)[39]是用其自由能(Fk )来计算的,而在Gibbs统计力学里用的是其总能(Ek ),2种框架下各个物理量的对比如表1所示。刘梓葵教授在其近期发表的英文综述中[25,40~42]对叠熵理论进行了详解。

表1   叠熵理论框架与经典统计力学下热力学物理量的比较[41]

Table 1  Comparison of thermodynamic terminology between zentropy theory and classic statistical mechanics[41]

PropertyStatistical mechanicsZentropy framework
EntropyS=-kBkpklnpkS=kpkSk-kBkpklnpk
Free energyF=kpkEk+kBTkpklnpkF=kpkFk+kBTkpklnpk
Partition functionZ=exp(-FkBT)=kexp(-EkkBT)Z=exp(-FkBT)=kexp(-FkkBT)
Probabilitypk=ZkZ=exp(-Ek-FkBT)pk=ZkZ=exp(-Fk-FkBT)

Note:Ek, Fk, and Zk are the total energy, free energy, and partition function of configuration k, respectively; T is the temperature in unit of K

新窗口打开| 下载CSV


2 基于叠熵理论的材料本征性质预测

基于叠熵理论,材料的本征性质是由其材料基因(微观组态)宏观概率表达的结果。对于每个微观组态(包括基态与其他微观组态),第一性原理高通量计算工具DFTTK[43]能够考虑静态能量、声子及热电子贡献从而自动计算系统的熵与自由能,如 式(2)所示:

FkV, T=EkV+FvibkV, T+FelkV, T

式中,V为体积,T为温度;Ek (V)为0 K下微观组态k在不同体积的静态总能,可由第一性原理计算直接得到;FvibkV, T为振动对微观组态k自由能的贡献;FelkV, T为热电子对微观组态k自由能的贡献。

在传统理论中,通常以最稳定的微观组态性质来表示材料的宏观性质,没有考虑其他微观组态的贡献,尤其是非振动因素造成的微观组态,如磁性材料的自旋组态,铁电材料的畴组态,高熵材料的不同原子排列组态等。因此,在实际应用中,可以通过构建非振动微观组态,并对单一的微观组态采用第一性原理方法(不需要任何经验参数)计算得到其能量、体积等各项性质,进而综合叠熵理论,从而实现完全通过从头算(ab initio)方法对材料中的多种现象或性能给出新解释,如临界现象[44]、热膨胀机理[30,32,45]、铁磁-顺磁转变[46]以及铁电-顺电转变[47]等。

2.1 材料热膨胀机制新解

对于负膨胀材料,文献中提出了多种不同的负膨胀机制[48],如振动机制(包括刚性单元耦合振动以及桥位原子的横向振动等)、铁磁相变、铁电相变等,不同的材料需要用不同的机制解释,部分材料中还存在多种机制共存,如Cu2P2O7中振动机制与铁磁转变机制共存[49],为负膨胀材料机制的解释增加了困难。而叠熵理论能很好地统一不同的负膨胀机制:从叠熵理论角度来看,宏观材料的体积源于各个微观组态体积的加权平均,而每个微观组态的能量决定各个微观组态的概率,即决定其权重,如 式(3)所示,当基态体积大于其他微观组态体积的情况占多数时,即 式(3)中的右边第二项为负时,材料就可能表现出负膨胀现象。

V=kpkVk=Vg+kpkVk-Vg

式中,V k 为微观组态k的体积,Vg为基态的体积。

此处的微观组态包括振动造成的,如桥位原子横向振动、刚性单元耦合振动等,也包括非振动因素造成的,如磁性转变、铁电转变等。由振动因素造成的微观组态可直接由第一性原理方法结合声子谱计算得到,如采用DFTTK[43]软件直接计算预测了pentadiamond的负膨胀现象[50],而非振动因素造成的微观组态必须考虑其相应自由度,如磁性转变的自旋组态等。以磁性材料Ce (正膨胀)[51]和Fe3Pt (负膨胀)[52]为例,在磁性材料中,其熵或能量除了需要考虑振动、热电子等贡献外,还需要考虑其不同自旋组态的贡献,如图3[30]所示。在Ce中,考虑无磁态(nonmagnetic,NM)、铁磁态(ferromagnetic,FM)以及反铁磁态(antiferromagnetic,AFM) 3种不同的自旋组态,NM为基态,其平衡体积小于FM以及AFM的体积,因此在热输入的条件下,FM和AFM微观组态的概率增加,而NM微观组态的概率降低,因此其表现为正常的膨胀效应;而在9个Fe原子的Fe3Pt超胞中,考虑Fe原子的不同自旋组态,共有29 = 512种不同的自旋组态,根据对称性,其中共有37种独立的自旋组态,图3[30]中展示了独立的自旋组态的体积-能量曲线。可以看出,其基态(FM)的平衡体积均大于其他自旋组态的平衡体积,因此在热输入条件下,基态的概率减小而其他组态的概率增加,材料表现出负热膨胀效应。除Ce与Fe3Pt外,该理论也能很好地解释Fe[53]和Ni[54]温度-压力相图中的异常。此外,刘梓葵教授团队[55]提出了一个基于压力-温度相图预测负热膨胀效应(NTE)的简单判据,即在压力-温度相图中存在具有负斜率的两相平衡边界时,材料表现出负膨胀效应。

图3

图3   不同微观组态0 K下的体积-能量曲线[30]

Fig.3   Volume-energy curves at 0 K of different configurations[30] (FM, AFM, and NM mean ferromagnetic, antiferromagnetic, and nonmagnetic, respectively; V and Etot mean atomic volume and total energy at 0 K, respectively)

(a) Ce (b) Fe3Pt


2.2 临界现象预测

系统的临界点描述了体系的稳定性极限,在热力学上定义摩尔量对其共轭的热力学势,如STV与压强(P)等的偏导发散,如 式(4)所示。与此同时,体积对温度的偏导也发散,如 式(5)所示,正膨胀材料其临界点处该值为+∞,而负膨胀材料其临界点处该值为-∞。

ST=V-P=+
VT=±

材料临界行为的预测需要精确地表达熵或体积与温度的关系,因此,材料在临界点附近的异常性能通常很难直接通过第一性原理方法计算单一组态得到,在本团队前期工作中[56],通过引入磁性经验项,很好地预测了Ce的γα转变及其临界现象。而采用叠熵理论,只需要考虑NM、FM以及AFM 3种简单的不同组态熵及不同组态间的统计熵贡献,不需要引入任何经验项,即能完全从第一性原理方法很好地复现Ce的临界行为;此外,考虑512种不同的自旋组态(37种独立的自旋组态)就能很好地复现Fe3Pt的临界行为[51,44],如图4[45]所示。

图4

图4   不同压强下的温度-体积相图[45]

Fig.4   Temperature-volume phase diagram with isobaric volumes at various pressures[45] (T means temperature; the volume (V) is normalized to their respective equilibrium volume (VN) at atmospheric pressure and room temperature; CPTE and NTE mean colossal positive thermal expansion and negative thermal expansion, respectively; Exp. means experimental results)

(a) Ce (b) Fe3Pt


2.3 铁电-顺电转变临界温度预测

在以往的铁电-顺电转变的临界温度预测中,通常需要通过实验或计算数据来拟合模型参数,如含时Ginzburg-Landau理论中Landau自由能表达式中的系数需要通过拟合自由能曲线得到[57]。但在叠熵理论中,铁电材料中铁电畴壁(畴壁两侧极化方向不同)按一定概率存在,其概率由其畴壁能相对铁电基态的能量高低决定,低温时铁电基态占多数,材料表现出铁电性,而当温度升高,铁电基态概率降低,当降低到不占多数时,材料表现出顺电性。以铁电材料PbTiO3[47]为例,该材料铁电基态为沿<001>方向极化的四方相,有90°畴壁(90DW)和180°畴壁(180DW) 2种铁电畴壁。通过叠熵理论,考虑了铁电基态(ferroelectric ground state,FEG)以及90DW和180DW的2个非基态的微观组态(其多重度比值为1∶4∶1),通过第一性原理计算得到90DW和180DW的畴壁能[58],此时体系的组态熵(Sconf)如 式(6)所示:

Sconf=-kB(pFEGlnpFEG+4p90DWlnp90DW+
p180DWlnp180DW)

式中,pFEGp90DWp180DW分别为FEG、90DW和180DW的概率。

结合表1中的公式,可计算得到各个组态的概率分布,如图5[47]所示。从图中可以看出,随着温度的升高,铁电基态的概率越来越小,而非基态(尤其是90DW)的概率越来越大,当非基态组态占多数时,即铁电态占比小于50%,可作为铁电-顺电转变的判据,由此得到PbTiO3的铁电-顺电转变温度为776 K (实验结果763 K),实现了完全通过第一性原理无任何经验或者拟合参数的铁电-顺电转变临界温度的预测。此外,考虑到不同微观组态均以一定概率表现,随着温度的升高,PbTiO3中90DW的概率逐渐升高,若铁电态PbTiO3沿晶体c轴极化,则随着温度的增大,将出现大量朝向晶体a轴和b轴方向的极化(与原铁电极化方向形成90DW),在宏观上表现为PbTiO3发生从四方相到立方相的转变。通过该理论能很好地解释X射线衍射(XRD)与X射线吸收精细结构(XAFS)方法(2种方法在时间上的分辨率不同,XRD可以认为是一段时间平均的结果,而XAFS可认为是瞬时的结果)测定PbTiO3宏观和微观晶格参数的矛盾[59]

图5

图5   PbTiO3各个组态概率随温度的变化[47]

Fig.5   Probability of configurations as a function of temperature in PbTiO3[47] (p—probability, DW—domain wall, FEG—ground state of ferroelectric without DW, TC—critical temperature, the domain wall energies for 90DW and 180DW are 35 and 132 mJ/m2, respectively, which is taken from Ref.[58])


2.4 无序程度定量描述与有序-无序转变预测

叠熵理论还能对材料体系的无序程度(fDoD)进行定量描述,即通过其实际组态熵(Sconf)与理想状态下(完全无序)组态熵(Sideal)的比值来表达体系的无序程度(fDoDS)[46]。对于给定的体系,完全无序状态下的组态熵是一个常数,可用该体系的微观组态数目ξ表达,如 式(7)所示:

fDoDS=SconfSideal=-pklnpklnξ

原则上来说,在0 K下体系为完全有序状态,即fDoDS = 0,当用 式(7)计算时,某些情况在0 K下其fDoD ≠ 0,此时可采用 式(8)进行归一化处理:

fDoD=fDoD-fDoD0 K1-fDoD0 K

基于叠熵理论可以得到每个微观组态的概率分布,进而自然可以反映整个材料的无序程度。因此,除了用熵来表征材料的无序程度外,还可以用其他实验上可测定的参数来表征材料的无序程度,从而和实验进行比较,如平均磁矩(fDoDm),第一配位壳层的平均配位数(fDoDcn)等,具体见 式(9)和(10)。

fDoDm=1-mTm0=1-pkmkp0kmk
fDoDcn=1-δcnTδcn0=1-pkδcnkp0kδcnk

式中,< >表示基于pk 的系综平均,m表示磁矩,下标0和T表示0 K和有限温度下的参数,上标k表示为微观组态k的性质,δcn为材料的不同自旋或原子配位数差的绝对值,可通过同步辐射测定其对分布函数来表征。

在定义了材料无序程度的定量度量后,其拐点,即二阶导数为0 (2fDoDT2=0)的位置可认为是无序-有序(如磁性转变、调幅分解等)转变位置。以Fe3Pt为例,在考虑了512种不同的自旋微观组态后,其无序程度随温度变化关系如图6[46]所示。从图中可以看出,在利用准谐近似(quasi harmonic approximation,QHA,考虑了其声子以及热电子贡献)得到自由能后,Fe3Pt中的Curie温度为400 K,与实验结果的425 K基本一致,而若仅以经典统计力学中的平衡结构能量(ΔE0)或静态总能(Ek (V))作为参数,得到的Curie温度(分别为45和650 K)误差较大。同时,从预测的无序程度来看,完全无序(fDoD = 1)在实际材料中很难达到,Fe3Pt在1300 K时,其无序度仅达到0.75左右。此外,叠熵理论还被成功应用于预测钙钛矿型镍酸盐中的反铁磁与顺磁之间的转变,如YNiO3[60] (根据叠熵理论预测的Neel温度为144 K,实验测定值为145 K)和SmNiO3[61] (叠熵理论预测的Neel温度为266 K,实验测定为225 K)。

图6

图6   通过熵预测的Fe3Pt无序程度(fDoDS)随温度的变化[46]

Fig.6   Predicted degree of disorder (fDoDS) as a function of temperature in Fe3Pt via entropy[46]E0 and Ek (V) are equilibrium energy and static total energy at 0 K, respectively. QHA means quasiharmonic approach, in which the free energy with contributions from vibrations and thermal electrons is evaluated)


3 总结与展望

熵是热力学上一个重要的概念,其普遍存在于从量子尺度到天文尺度的各个尺度中,以往熵的理论忽视了子系统熵对当前研究体系熵的贡献,而叠熵理论通过递归公式将各个尺度熵的贡献考虑进来,实现了熵的跨尺度“堆叠”。在考虑子系统熵的贡献后,明晰了经典统计力学中的物理量,如自由能,配分函数等中的微观组态的总能需要用微观组态的自由能来替代。单个相是材料构成的基本模块,从叠熵理论中可以看出,单个相中的不同微观组态是单个相的基本构成模块,这些微观组态被认为是材料的基因[3],微观组态的能量控制着该微观组态的概率,而不同微观组态的概率高低控制该“基因”的表达。这些微观组态在不同的材料中有不同的表现形式,如磁性材料中的不同自旋组态,铁电材料中的不同畴态,高熵材料中不同的原子排列等。基于上述观点,可以完全采用第一性原理方法(无需任何经验参数)对材料的多种行为进行预测与解释:通过考虑各个微观组态自由能进而表达出体系准确的自由能,能够很好地描述材料的临界行为;材料的负膨胀行为源于体积小于基态的非基态组态占多数;铁电-顺电相变源于铁电微观组态的概率低于50%,进而宏观上不能表现出该微观组态特征;有序-无序相变源于无序度的变化率由正变为负。未来围绕着叠熵理论可以重点开展如下研究。

(1) 发展可预测更多材料性能的叠熵理论。叠熵理论是在统计力学底层的创新,因此利用叠熵理论除了能对上述几个“老问题”给出新的理解外,还有望应用于其他领域[25],如超导[62]、熔点预测[63]、高熵材料性能的解释与预测等。

(2) 开发叠熵理论自动化高通量计算工作流及相关软件生态。叠熵理论应用过程较为复杂,如独立微观组态及其多重度分析、组态概率计算、所有独立微观组态的第一性原理计算等,因此需要开发叠熵理论自动化高通量计算工作流及相关的软件,加速叠熵理论的应用。目前刘梓葵教授主导成立了旨在通过组织专题讨论会推动材料计算工具与材料数据库发展的非盈利性材料基因组基金(Materials Genome Foundation,https://materialsgenomefoundation.org/),目前已经支持了一系列材料基因组计划相关软件生态,如第一性原理高通量计算工具DFTTK[43]、材料热力学数据库构建及优化软件ESPEI[64]、开源计算相图软件PyCalphad[65]、量化相图数据不确定性的工具PDUQ[66]、面向材料科学家的机器学习软件ASCENDS[67]、基于晶体结构预测材料形成能的机器学习模型SIPFENN/pySIPFENN[68],关于叠熵理论的软件也正在开发中。

(3) 构建基于材料基因(微观组态)的“数据海”[69,70]。以微观组态为基础,结合开发的软件生态,对基于微观组态的材料基因数据进行自动高通量计算、存储以及管理等,形成基于微观组态的材料“数据海”,如钙钛矿铁电畴“数据海”、高熵材料“数据海”等。

(4) 集成AI4S和叠熵理论,构建基于材料基因的AI4Materials模型。叠熵理论需要各态历经,即需要考虑所有的微观组态,因此微观组态的数目随着体系的增大(包括体系原子数目的增多或原子种类的增多)急剧增长,而AI4S的主要目标正是有效应对难解的组合爆炸问题[22],因此有机集成AI4S和叠熵理论,以AI4S加速基于微观组态的计算,以叠熵理论赋予AI4S物理意义,提高AI4S精度与可解释性,最终形成基于材料基因的AI4Materials模型。

参考文献

Gan Y.

Research on the innovative development of new materials science and technology in China

[J]. Engineering, 2024, 32: 10

[本文引用: 1]

Xie M, Gan Y, Wang H.

Research on new material power strategy by 2035

[J]. Strategic Study CAE, 2020, 22(5): 1

[本文引用: 1]

谢 曼, 干 勇, 王 慧.

面向2035的新材料强国战略研究

[J]. 中国工程科学, 2020, 22(5): 1

[本文引用: 1]

Liu Z K.

Perspective on materials genome

[J]. Chin. Sci. Bull., 2013, 58: 3618

[本文引用: 3]

刘梓葵.

关于材料基因组的基本观点及展望

[J]. 科学通报, 2013, 58: 3618

[本文引用: 3]

National Science and Technology Council.

Materials genome initiative for global competitiveness

[EB/OL]. (2011-06-24).

[本文引用: 2]

National Science and Technology Council.

Materials genome initiative strategic plan

[EB/OL]. (2014-12-04).

[本文引用: 1]

Su Y J, Fu H D, Bai Y, et al.

Progress in materials genome engineering in China

[J]. Acta. Metall. Sin., 2020, 56: 1313

DOI      [本文引用: 2]

Materials genome engineering (MGE) is a frontier technology in the field of material science and engineering, which is well capable to revolutionize the research and development (R&D) mode of new materials, greatly improve the R&D efficiency, shorten the R&D time, and reduce the cost. This paper reviews the progress of MGE in China from the aspects of the fundamental theory and methods, key technology and equipment, the R&D of new materials and related engineering application, talents training, formation and promotion of new concept of material genetic engineering. The paper also looks forward to the future development of MGE in China.

宿彦京, 付华栋, 白 洋 .

中国材料基因工程研究进展

[J]. 金属学报, 2020, 56: 1313

DOI      [本文引用: 2]

材料基因工程是材料领域的颠覆性前沿技术,将对材料研发模式产生革命性的变革,全面加速材料从设计到工程化应用的进程,大幅度提升新材料的研发效率,缩短研发周期,降低研发成本,促进工程化应用。本文从基础理论与方法、关键技术与装备、新材料研发与工程化应用、人才培养以及材料基因工程新理念的形成和推广等方面,综述了中国材料基因工程的研究进展,并提出了未来发展方向建议。

National Research Council. Integrated Computational Materials Engineering: A Transformational Discipline for Improved Competitiveness and National Security [M]. Washington, D.C.: The National Academies Press, 2008: 9

[本文引用: 1]

Wang W Y, Yin J L, Chai Z X, et al.

Big data-assisted digital twins for the smart design and manufacturing of advanced materials: From atoms to products

[J]. J. Mater. Inf., 2022, 2: 1

Liu Z K, Chen L Q, Raghavan P, et al.

An integrated framework for multi-scale materials simulation and design

[J]. J. Comput. Aided Mater. Des., 2004, 11: 183

[本文引用: 1]

Liu X, Furrer D, Kosters J, et al.

Vision 2040 : A roadmap for integrated, multiscale modeling and simulation of materials and systems

[EB/OL]. (2018-03-22).

[本文引用: 1]

National Academies of Sciences, Engineering, and Medicine. NSF Efforts to Achieve the Nation's Vision for the Materials Genome Initiative: Designing Materials to Revolutionize and Engineer Our Future (DMREF) [M]. Washington, D.C.: The National Academies Press, 2023: 57

[本文引用: 1]

Xie J X.

Prospects of materials genome engineering frontiers

[J]. Mater. Genome Eng. Adv., 2023, 1: e17

[本文引用: 2]

Xie J X, Su Y J, Zhang D W, et al.

A vision of materials genome engineering in China

[J]. Engineering, 2022, 10: 10

[本文引用: 1]

Li M X, Zhao S F, Lu Z, et al.

High-temperature bulk metallic glasses developed by combinatorial methods

[J]. Nature, 2019, 569: 99

[本文引用: 1]

Lu Z C, Zhang Y B, Li W Y, et al.

Materials genome strategy for metallic glasses

[J]. J. Mater. Sci. Technol., 2023, 166: 173

DOI      [本文引用: 1]

Metallic glasses (MGs) have attracted extensive attention in the past decades due to their unique chem-ical, physical and mechanical properties promising for a wide range of engineering applications. A thor-ough understanding of their structure-property relationships is the key to the development of novel MGs with desirable performance. New strategies, as proposed by Materials Genome Initiative (MGI), construct a new paradigm for high-throughput materials discovery and design, and are being increas-ingly implemented in the search of new MGs. While a few reports have summarized the application of high-throughput and/or machine learning techniques, a comprehensive assessment of materials genome strategies for developing MGs is still missing. Herein, this paper aims to present a timely overview of key advances in this fascinating subject, as well as current challenges and future opportunities. A holistic approach is used to cover the related topics, including high-throughput preparation and characterization of MGs, and data-driven machine learning strategies for accelerating the development of novel MGs. Fi-nally, future research directions and perspectives for MGI-assisted design of MGs are also proposed and surmised.

Wang G J, Peng L Y, Li K Q, et al.

ALKEMIE: An intelligent computational platform for accelerating materials discovery and design

[J]. Comput. Mater. Sci., 2021, 186: 110064

[本文引用: 1]

Wang G J, Li K Q, Peng L Y, et al.

High-throughput automatic integrated material calculations and data management intelligent platform and the application in novel alloys

[J]. Acta. Metall. Sin., 2022, 58: 75

DOI      [本文引用: 1]

The development of novel materials has experienced three paradigms: purely empirical, theoretical models, and computational materials science. Currently, the huge amount of data generated by experiments and simulations has facilitated a shift in materials science to a data-driven fourth paradigm. Therefore, the development of high-throughput automatic integrated computations and data mining algorithms based on material databases and artificial intelligence algorithms is critical for accelerating the design of novel materials. This paper presents an open-source distributed computational platform called Artificial Learning and Knowledge Enhanced Materials Informatics Engineering 2.0 (ALKEMIE2.0) based on the AMDIV (automation-modular-database-intelligence-visualization) design concepts. The ALKEMIE2.0 platform includes five core components of automation, modular, materials database, artificial intelligence, and visualization, which are suitable for the computational design of novel materials. The overall characteristics of ALKEMIE2.0 are divided into five pillars. ALKEMIE-Core integrates multiscale calculations and simulation software using the ALKEMIE-Plugin application programming interface. Its high-throughput calculation workflows that support 104 magnitude concurrencies are implemented by integrating the automatic frameworks of model constructions, calculation workflows, and data analyses. Furthermore, the platform is based on the ALKEMIE-Server, which can easily and automatically open daemon services and realize information interactions in distributed supercomputers. With its strong portability and scalability, ALKEMIE has been deployed in the National Supercomputing Tianjin Center. In addition, the multitype materials database called the ALKEMIE-Data Vault contains structure, task, workflow, and material property databases, which combined with the power of supercomputing, enables the rapid application of artificial intelligence algorithms in the design of new materials. In particular, the many user-friendly interfaces, which were elaborately designed using the ALKEMIE-GUI and are suitable for scientists with broad backgrounds, make structural building, work flowcharts, data analysis, and machine learning models more transparent and maneuverable. Finally, the main features of ALKEMIE2.0 are demonstrated using two examples of multiplatform deployment and high-throughput screening of binary aluminum alloys.

王冠杰, 李开旗, 彭力宇 .

高通量自动流程集成计算与数据管理智能平台及其在合金设计中的应用

[J]. 金属学报, 2022, 58: 75

DOI      [本文引用: 1]

材料研发模式经历了经验主导的第一范式、理论模型主导的第二范式和计算模拟主导的第三范式,如今正处于数据驱动的第四范式。为加速新材料的设计与研发,发展基于材料数据库和人工智能算法的高通量自动集成计算和数据挖掘算法变得至关重要。本文介绍了作者团队自主开发的分布式高通量自动流程集成计算和数据管理智能平台ALKEMIE2.0 (Artificial Learning and Knowledge Enhanced Materials Informatics Engineering 2.0),该平台基于AMDIV设计理念,包含了自动化、模块化、数据库、人工智能和可视化流程等5个适用于数据驱动的材料研发模式核心要素。概括来说,ALKEMIE2.0以模块化的方式集成了多个不同尺度的计算模拟软件;其高通量自动纠错流程可实现从建模、运行到数据分析,全程自动无人工干预;支持单用户不低于10<sup>4</sup>量级的并发高通量自动计算模拟。进一步而言,ALKEMIE2.0具有强大的可移植性和可扩展性,目前已部署在国家超算天津中心,基于多类型材料数据库结合超算强大的计算能力使得人工智能算法在新材料设计与研发中得以快速的应用和实践。更重要的是,ALKEMIE2.0设计了用户友好的可视化操作界面,使得结构建模、工作流计算逻辑、数据分析和机器学习模型具有更高的透明性和更强的可操作性,且适用于对材料计算模拟掌握程度从初级到专业的所有材料研究人员。最后,通过多平台部署和高通量筛选二元铝合金2个算例详细展示了ALKEMIE2.0的主要特色及功能。

Chong X Y, Yu W, Liang Y X, et al.

Understanding oxidation resistance of Pt-based alloys through computations of Ellingham diagrams with experimental verifications

[J]. J. Mater. Inf., 2023, 3: 21

[本文引用: 1]

Agrawal A, Choudhary A.

Perspective: Materials informatics and big data: Realization of the “fourth paradigm” of science in materials science

[J]. APL Mater., 2016, 4: 053208

[本文引用: 1]

Liu Y L, Niu C, Wang Z, et al.

Machine learning in materials genome initiative: A review

[J]. J. Mater. Sci. Technol., 2020, 57: 113

DOI      [本文引用: 1]

Discovering new materials with excellent performance is a hot issue in the materials genome initiative. Traditional experiments and calculations often waste large amounts of time and money and are also limited by various conditions. Therefore, it is imperative to develop a new method to accelerate the discovery and design of new materials. In recent years, material discovery and design methods using machine learning have attracted much attention from material experts and have made some progress. This review first outlines available materials database and material data analytics tools and then elaborates on the machine learning algorithms used in materials science. Next, the field of application of machine learning in materials science is summarized, focusing on the aspects of structure determination, performance prediction, fingerprint prediction, and new material discovery. Finally, the review points out the problems of data and machine learning in materials science and points to future research. Using machine learning algorithms, the authors hope to achieve amazing results in material discovery and design.

E W N.

AI helps to establish a new paradigm for scientific research

[J]. Bull. Chin. Acad. Sci., 2024, 39(1): 10

[本文引用: 2]

鄂维南.

AI助力打造科学研究新范式

[J]. 中国科学院院刊, 2024, 39(1): 10

[本文引用: 2]

Li G J.

AI4R: The fifth scientific research paradigm

[J]. Bull. Chin. Acad. Sci., 2024, 39: 1

[本文引用: 2]

李国杰.

智能化科研(AI4R):第五科研范式

[J]. 中国科学院院刊, 2024, 39: 1

[本文引用: 2]

Ågren J.

The materials genome and CALPHAD

[J]. Chin. Sci. Bull., 2013, 58: 3633

[本文引用: 1]

Ågren J.

材料基因组与相图计算

[J]. 科学通报, 2013, 58: 3633

[本文引用: 1]

Wang S Q, Ye H Q.

First-principles calculation of crystalline materials genome

[J]. Chin. Sci. Bull., 2013, 58: 3623

[本文引用: 2]

王绍青, 叶恒强.

晶体材料基因组问题第一原理计算研究

[J]. 科学通报, 2013, 58: 3623

[本文引用: 2]

Liu Z K.

Computational thermodynamics and its applications

[J]. Acta Mater., 2020, 200: 745

[本文引用: 3]

Liu Z K.

First-principles calculations and CALPHAD modeling of thermodynamics

[J]. J. Phase Equilib. Diffus., 2009, 30: 517

Olson G B, Liu Z K.

Genomic materials design: CALculation of PHAse Dynamics

[J]. Calphad, 2023, 82: 102590

Liu Z K.

Thermodynamics and its prediction and CALPHAD modeling: Review, state of the art, and perspectives

[J]. Calphad, 2023, 82: 102580

[本文引用: 1]

Campbell C E, Kattner U R, Liu Z K.

File and data repositories for next generation CALPHAD

[J]. Scr. Mater., 2014, 70: 7

[本文引用: 1]

Liu Z K, Wang Y, Shang S L.

Zentropy theory for positive and negative thermal expansion

[J]. J. Phase Equilib. Diffus., 2022, 43: 598

[本文引用: 7]

Liu Z K.

Building materials genome from ground‐state configuration to engineering advance

[J]. Mater. Genome Eng. Adv., 2023, 1: e15

Liu Z K, Hew N L E, Shang S L.

Zentropy theory for accurate prediction of free energy, volume, and thermal expansion without fitting parameters

[J]. Microstructures, 2024, 4: 2024009

[本文引用: 2]

Mooraj S, Chen W.

A review on high-throughput development of high-entropy alloys by combinatorial methods

[J]. J. Mater. Inf., 2023, 3: 4

[本文引用: 1]

Jiang B B, Yu Y, Cui J, et al.

High-entropy-stabilized chalcogenides with high thermoelectric performance

[J]. Science, 2021, 371: 830

DOI      PMID      [本文引用: 1]

Thermoelectric technology generates electricity from waste heat, but one bottleneck for wider use is the performance of thermoelectric materials. Manipulating the configurational entropy of a material by introducing different atomic species can tune phase composition and extend the performance optimization space. We enhanced the figure of merit () value to 1.8 at 900 kelvin in an n-type PbSe-based high-entropy material formed by entropy-driven structural stabilization. The largely distorted lattices in this high-entropy system caused unusual shear strains, which provided strong phonon scattering to largely lower lattice thermal conductivity. The thermoelectric conversion efficiency was 12.3% at temperature difference Δ = 507 kelvin, for the fabricated segmented module based on this n-type high-entropy material. Our demonstration provides a paradigm to improve thermoelectric performance for high-entropy thermoelectric materials through entropy engineering.Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works.

Liu Y, Lu Y H, Wang W Y, et al.

Effects of solutes on thermodynamic properties of (TMZrU)C (TM = Ta, Y) medium-entropy carbides: A first-principles study

[J]. J. Mater. Inf., 2023, 3: 17

[本文引用: 1]

Liao M Q, Gong H S, Qu N, et al.

CALPHAD aided mechanical properties screening in full composition space of NbC-TiC-VC-ZrC ultra-high temperature ceramics

[J]. Int. J. Refract. Met. Hard Mater., 2023, 113: 106191

[本文引用: 1]

Wang J, Chong X Y, Lv L, et al.

High-entropy ferroelastic (10RE0.1)TaO4 ceramics with oxygen vacancies and improved thermophysical properties

[J]. J. Mater. Sci. Technol., 2023, 157: 98

[本文引用: 1]

Chen L X, Chen Z W, Yao X, et al.

High-entropy alloy catalysts: High-throughput and machine learning-driven design

[J]. J. Mater. Inf., 2022, 2: 19

[本文引用: 1]

Ceder G.

A derivation of the Ising model for the computation of phase diagrams

[J]. Comput. Mater. Sci., 1993, 1: 144

[本文引用: 1]

Liu Z K.

Theory of cross phenomena and their coefficients beyond Onsager theorem

[J]. Mater. Res. Lett., 2022, 10: 393

[本文引用: 1]

Liu Z K.

Quantitative predictive theories through integrating quantum, statistical, equilibrium, and nonequilibrium thermodynamics

[J]. J. Phys.: Condens. Matter., 2024, 36: 343003

[本文引用: 2]

Liu Z K.

On Gibbs Equilibrium and hillert nonequilibrium thermodynamics

[DB/OL]. arXiv: 2402. 14231, 2024

[本文引用: 1]

Wang Y, Liao M Q, Bocklund B J, et al.

DFTTK: Density functional theory toolKit for high-throughput lattice dynamics calculations

[J]. Calphad, 2021, 75: 102355

[本文引用: 3]

Liu Z K, Li B, Lin H.

Multiscale entropy and its implications to critical phenomena, emergent behaviors, and information

[J]. J. Phase Equilib. Diffus., 2019, 40: 508

[本文引用: 2]

Liu Z K, Wang Y, Shang S L.

Thermal expansion anomaly regulated by entropy

[J]. Sci. Rep., 2014, 4: 7043

[本文引用: 4]

Shang S L, Wang Y, Liu Z K.

Quantifying the degree of disorder and associated phenomena in materials through zentropy: Illustrated with Invar Fe3Pt

[J]. Scr. Mater., 2023, 225: 115164

[本文引用: 5]

Liu Z K, Shang S L, Du J L, et al.

Parameter-free prediction of phase transition in PbTiO3 through combination of quantum mechanics and statistical mechanics

[J]. Scr. Mater., 2023, 232: 115480

[本文引用: 5]

Liang E J, Sun Q, Yuan H L, et al.

Negative thermal expansion: Mechanisms and materials

[J]. Front. Phys., 2021, 16: 53302

[本文引用: 1]

Zhou C, Tang Z Y, Kong X Q, et al.

High-performance zero thermal expansion in Al metal matrix composites

[J]. Acta Mater., 2024, 275: 120076

[本文引用: 1]

Liao M Q, Wang Y, Wang F J, et al.

Unexpected low thermal expansion coefficients of pentadiamond

[J]. Phys. Chem. Chem. Phys., 2022, 24: 23561

[本文引用: 1]

Wang Y, Hector Jr L G, Zhang H, et al.

A thermodynamic framework for a system with itinerant-electron magnetism

[J]. J. Phys.: Condens. Matter., 2009, 21: 326003

[本文引用: 2]

Wang Y, Shang S L, Zhang H, et al.

Thermodynamic fluctuations in magnetic states: Fe3Pt as a prototype

[J]. Philos. Mag. Lett., 2010, 90: 851

[本文引用: 1]

Shang S L, Wang Y, Liu Z K.

Thermodynamic fluctuations between magnetic states from first-principles phonon calculations: The case of bcc Fe

[J]. Phys. Rev., 2010, 82B: 014425

[本文引用: 1]

Shang S L, Saal J E, Mei Z G, et al.

Magnetic thermodynamics of fcc Ni from first-principles partition function approach

[J]. J. Appl. Phys., 2010, 108: 123514

[本文引用: 1]

Liu Z K, Wang Y, Shang S L.

Origin of negative thermal expansion phenomenon in solids

[J]. Scr. Mater., 2011, 65: 664

[本文引用: 1]

Wang Y, Hector Jr L G, Zhang H, et al.

Thermodynamics of the Ce γ-α transition: Density-functional study

[J]. Phys. Rev, 2008, 78B: 104113

[本文引用: 1]

Li Y L, Hu S Y, Liu Z K, et al.

Effect of substrate constraint on the stability and evolution of ferroelectric domain structures in thin films

[J]. Acta Mater., 2002, 50: 395

[本文引用: 1]

Meyer B, Vanderbilt D.

Ab initio study of ferroelectric domain walls in PbTiO3

[J]. Phys. Rev., 2002, 65B: 104111

[本文引用: 2]

Fang H Z, Wang Y, Shang S L, et al.

Nature of ferroelectric-paraelectric phase transition and origin of negative thermal expansion in PbTiO3

[J]. Phys. Rev., 2015, 91B: 024104

[本文引用: 1]

Du J L, Malyi O I, Shang S L, et al.

Density functional thermodynamic description of spin, phonon and displacement degrees of freedom in antiferromagnetic-to-paramagnetic phase transition in YNiO3

[J]. Mater. Today Phys., 2022, 27: 100805

[本文引用: 1]

Du J L, Zhang Z L, Shang S L, et al.

Underpinnings behind the magnetic order-to-disorder transition and property anomaly of disproportionated insulating samarium nickelate

[J]. Acta Mater., 2024, 268: 119783

[本文引用: 1]

Liu Z K, Shang S L.

Revealing symmetry-broken superconducting configurations by density functional theory

[DB/OL]. arXiv: 2404. 00719, 2024

[本文引用: 1]

Hong Q J, Liu Z K.

A generalized approach for rapid entropy calculation of liquids and solids

[DB/OL]. arXiv: 2403. 19872, 2024

[本文引用: 1]

Bocklund B, Otis R, Egorov A, et al.

ESPEI for efficient thermodynamic database development, modification, and uncertainty quantification: Application to Cu-Mg

[J]. MRS Commun., 2019, 9: 618

DOI      [本文引用: 1]

The software package ESPEI has been developed for efficient evaluation of thermodynamic model parameters within the CALPHAD method. ESPEI uses a linear fitting strategy to parameterize Gibbs energy functions of single phases based on their thermochemical data and refines the model parameters using phase equilibrium data through Bayesian parameter estimation within a Markov Chain Monte Carlo machine learning approach. In this paper, the methodologies employed in ESPEI are discussed in detail and demonstrated for the Cu-Mg system down to 0 K using unary descriptions based on segmented regression. The model parameter uncertainties are quantified and propagated to the Gibbs energy functions.

Otis R, Liu Z K.

pycalphad: CALPHAD-based computational thermodynamics in python

[J]. J. Open Res. Softw., 2017, 5: 1

[本文引用: 1]

Paulson N H, Bocklund B J, Otis R A, et al.

Quantified uncertainty in thermodynamic modeling for materials design

[J]. Acta Mater., 2019, 174: 9

DOI      [本文引用: 1]

Phase fractions, compositions and energies of the stable phases as a function of macroscopic composition, temperature, and pressure (X-T-P) are the principle correlations needed for the design of new materials and improvement of existing materials. They are the outcomes of thermodynamic modeling based on the CALculation of PHAse Diagrams (CALPHAD) approach. The accuracy of CALPHAD predictions vary widely in X-T-P space due to experimental error, model inadequacy and unequal data coverage. In response, researchers have developed frameworks to quantify the uncertainty of thermodynamic property model parameters and propagate it to phase diagram predictions. In most previous studies, uncertainty was represented as intervals on phase boundaries (with respect to composition or temperature) and was unable to represent the uncertainty in invariant reactions or in the stability of phase regions. In this work, we propose a suite of tools that leverages samples from the multivariate model parameter distribution to represent uncertainty in forms that surpass previous limitations and are well suited to materials design. These representations include the distribution of phase diagrams and their features, as well as the dependence of phase stability and the distributions of phase fraction, composition, activity and Gibbs energy on X-T-P location - irrespective of the total number of components. Most critically, the new methodology allows the material designer to interrogate a certain composition and temperature domain and get in return the probability of different phases to be stable, which can positively impact materials design. (C) 2019 Acta Materialia Inc. Published by Elsevier Ltd.

Peng J, Lee S, Williams A, et al.

Advanced data science toolkit for non-data scientists—A user guide

[J]. Calphad, 2020, 68: 101733

[本文引用: 1]

Krajewski A M, Siegel J W, Xu J, et al.

Extensible structure-informed prediction of formation energy with improved accuracy and usability employing neural networks

[J]. Comput. Mater. Sci., 2022, 208: 111254

[本文引用: 1]

Liu Z K.

Ocean of data: Integrating first-principles calculations and CALPHAD modeling with machine learning

[J]. J. Phase Equilib. Diffus., 2018, 39: 635

[本文引用: 1]

Liu Z K.

View and comments on the data ecosystem: “Ocean of data”

[J]. Engineering, 2020, 6: 604

[本文引用: 1]

/