6-22. Cross-scale design of precious metal superalloys based on thermodynamic database: high-throughput first-principles calculations combined with CALPHAD modeling

XiaoYu Chong¹, Jing Feng¹, Shun-Li Shang², Yi Wang², Zi-Kui Liu²

1.Key laboratory of material genetic engineering, Kunming University of Science and Technology,

2.Kunming, Yunnan, China 3.Materials Genome Inc., State College, PA, 16803, USA

Abstract: The three major elements of material genome/genetic engineering include high-throughput computing, high-throughput experiments, database and machine learning. The key technical characteristics of high-throughput computing differ from traditional computing methods as following: realize parallel and/or automatic calculation for 10²~10⁴ tasks; The integrated computing (ICME) method focuses on breaking through the bottleneck of material design/computing cross-scales. In order to realize the leap from microscopic simulation to mesoscopic phase diagram and phase field simulation to macroscopic finite element simulation, the key point is the automatic transmission and correlation modeling of parameters at different levels/scales. The CALPHAD modeling and database are the basis of ICME. The previous CALPHAD modeling methods are all from experiments, but it takes too long and the cost is high. We develop a standard workflow to construct the multiphase thermodynamic database based on the high-throughput first-principles quasi-harmonic calculations. Through the independently developed first-principles calculations software DFTTK, the 10⁴ thermodynamic, mechanical and diffusion properties for single phase are realized at the finite temperature by parallel calculations. The obtained parameters are directly imported into another independently developed thermodynamic modeling program ESPEI, which can automatically optimize the parameters of higher-order thermodynamic database at the same time, which overcomes the traditional "snowball" effect of the previous thermodynamic modeling and reduce the threshold. Our team optimize the thermodynamic modeling and high temperature performance of Pt-Ir-Rh-Ni-Zr-Hf-Si-Cr system by this way. Combined with the properties from first-principles calculations for single-phase, we build the mapping relationship between the composition and performance under different temperature. All the work is helpful to optimize the composition design, processing technology, high temperature performance and achieve long service under 1500-1600 ℃.

基于热力学数据库的稀贵金属高温合金跨尺度设计：高通量第一性原理计算耦合相图热力学模型

种晓宇¹，冯晶¹，Shun-Li Shang²，Yi Wang²，Zi-Kui Liu²

1.昆明理工大学材料基因工程校重点实验室，昆明，云南，中国

2.Materials Genome Inc., State College, PA, 16803, USA

摘要：材料基因组/工程的三大要素包含高通量计算、高通量实验和数据库及机器学习，高通量计算不同于传统计算方法关键技术特点主要为：实现并发式计算、自动流程计算，实现 10²~10⁴量级任务； 集成计算 (ICME)方法，重点突破跨层次/跨尺度材料设计/计算瓶颈。而要实现从微观原子模拟到介观相图、相场模拟再到宏观有限元工程模拟的跨跃，其关键核心是不同层次/不同尺度计算参数的自动传递和关联建模。其中相图热力学模型及数据库是集成计算材料工程（ICME）的基础。以前的相图建模和优化方法都是通过实验测定，但是周期长、成本高。我们发展了基于第一性原理高通量准简谐计算来建立多元多相热力学数据库的方法和标准化工作流。通过自主开发的第一性原理计算软件DFTTK，实现10⁴量级晶体结构的有限温度下热力学性质、力学性质和扩散性质的并发式计算；获取的参数直接传递到另一个自主研发的多元相图热力学建模程序ESPEI中，该程序能实现多元高阶体系多个热力学参数的同时自动优化，克服了传统热力学建模方法的“雪球”效应，尽量减少人的工作量，降低热力学模型优化的门槛。我们团队以此方法对Pt-Ir-Rh-Ni-Zr-Hf-Si-Cr多元合金体系进行热力学建模和高温性能优化，以热力学数据库为基础，结合第一性原理获取单相的性能，构建了该合金不同温度下成分和性能的全映射关系，以此进行成分设计、优化加工工艺、提高高温性能，实现该类合金在1500-1600℃的长时稳定服役。