Advances in Multi-Principal Elements Alloys X: Structures and Modeling: Modeling and Machine Learning
Sponsored by: TMS Functional Materials Division, TMS Structural Materials Division, TMS: Alloy Phases Committee, TMS: Mechanical Behavior of Materials Committee
Program Organizers: Peter Liaw, University of Tennessee; Michael Gao, National Energy Technology Laboratory; E-Wen Huang, National Chiao Tung University; Jennifer Carter, Case Western Reserve University; Srivatsan Tirumalai; Xie Xie, FCA US LLC; Gongyao Wang, Alcoa Technical Center
Thursday 8:30 AM
March 3, 2022
Room: 251B
Location: Anaheim Convention Center
Session Chair: Ganesh Balasubramanian, Lehigh University
8:30 AM
Expediting the Materials Discovery Process of MPEAs through Efficient Coupling of High-throughput Density Functional Theory, Molecular Dynamics and Machine Learning Techniques: Jacob Startt1; Mitchell Wood1; Sean Donegan2; Remi Dingreville1; 1Sandia National Labs; 2Air Force Research Lab
Refractory high entropy alloys (RHEA), multi-principal element alloys (RMPEA), or concentrated, complex alloys (RCCA) are a promising new class of material offering excellent mechanical strength coupled with high-temperature stability. These alloys present a large compositional space, however, making discovery and optimization difficult. Machine learning (ML) techniques offer a path for efficiently navigating this complex space. In this work we employ two levels of ML to (i) build a highly-accurate computational dataset for the MoNbTa-class alloy and (ii) to rapidly explore the compositional landscape for optimized sets of desired material properties. We first employ high-fidelity density functional theory (DFT) simulations to train a SNAP (spectral neighbor analysis potential) interatomic molecular dynamics (MD) potential. This SNAP potential is then used to generate a large thermodynamic and mechanical property database through which ML algorithms search for optimal alloy compositions. This work demonstrates a viable path towards expediting the materials discovery process for MPEAs.
8:50 AM
Machine Learning Enabled Defect Energies Prediction in Concentrated Alloys: Anus Manzoor1; Gaurav Arora1; Dilpuneet Aidhy1; 1University of Wyoming
Concentrated alloys, including high entropy alloys, consist of multiple principal elements randomly distributed on a crystal lattice that causes large variations in defect energies in a given alloy composition. Statistically capturing the variation requires performing large number of calculations which is computationally highly expensive. The challenge is compounded due to the exponentially large number of compositions that are possible in these alloys. Using machine learning model, we predict the defect energies in complex alloys using the database of the constituent’s binary alloys. Specifically, we demonstrate prediction of vacancy formation and vacancy migration energies in ternary, quaternary, and quinary in Ni-based alloys just by using the database of the binary alloys with high level of accuracy. A Major benefit of this approach is that for every new composition discovered, the defect energies can be computed using only the existing binary alloy database thereby completely bypassing the computationally expensive calculations.
9:10 AM
Exploration of Refractory Complex Concentrated Alloys through the Use of High-throughput Calculations and Experimentation: Austin Hernandez1; Sharmila Karumuri1; Sona Avetian1; Zachary McClure1; Logan Ware1; Alejandro Strachan1; Ilias Bilionis1; Kenneth Sandhage1; Michael Titus1; 1Purdue University
Due to several attractive properties of refractory complex concentrated alloys (RCCAs), such as high temperature ductility, strength, and phase stability, they are considered to be excellent candidate materials to replace Ni-based superalloys in high temperature, oxidizing environments. Unfortunately, RCCAs typically fail catastrophically in oxidizing environments at intermediate and high temperatures. Out of millions of possible high strength and oxidation resistant RCCA compositions, we evaluate several promising compositions exploring a design space of 9 refractory elements (Ti, V, Cr, Zr, Nb, Mo, Hf, Ta, W) and aluminum. To systematically interrogate the oxidation and hardness, the compositions are first evaluated using CALPHAD methods generated from down-selected calculations along a pareto front. In addition, machine learning coupled with an active learning loop is utilized. To facilitate high-throughput experimentation, diffusion couples of Al-containing alloys with Al-free compositions have been explored, and the hardness and oxidation behavior along the concentration gradients have been evaluated.
9:30 AM
Dislocation Motions in Refractory Multi-principal Alloys and Effects of Chemical Order and Disorder: Xinyi Wang1; Francesco Maresca2; Penghui Cao1; 1University of California, Irvine; 2University of Groningen
Body-centered cubic (bcc) refractory multi-principal alloys (r-MPAs) have been the subject of extensive research over the last few years because of their high strength and weak temperature dependence. The origin of extraordinary mechanical properties can be traced back to dislocation motion in concentrated solid solutions. Here, we couple a set of atomistic simulation techniques to investigate the atomistic mechanisms and the associated energy barriers governing screw dislocation motion in bcc r-MPAs. The detailed analysis shows that the screw dislocation reveals a hierarchical and multilevel structure potential energy landscape (PEL) with a collection of small basins nested in large metabasin. This striking feature pertaining to r-MPAs exerts a trapping force and back stress on saddle point activations, retarding dislocation movement. By introducing chemical short-range order, the energy barriers for individual mechanisms all increase but to different extents, leading to the dominant mechanism drift from kink-glide to kink-pair nucleation.
9:50 AM Break
10:10 AM
Factors Affecting Stacking Fault Energy in Concentrated Alloys Using Density Functional Theory and Machine Learning: Gaurav Arora1; Anus Manzoor1; Dilpuneet Aidhy1; 1University Of Wyoming
Recent experimental work has shown that the addition of certain elements can lower the stacking fault energy (SFE) of certain high entropy alloys thereby breaking the strength vs ductility tradeoff. To design alloys with desired SFEs, understanding the underlying mechanisms controlling SFE is critical. In this work, using density functional theory (DFT), we isolate the effect of atomic radii, planar charge density, and nearest neighbor environment on the SFE for 3d, 4d, and 5d doped Ni and Cu alloys. Particularly, we find that not only the radius of the dopant, but the planar charge density plays a significant role in defining the SFE in a particular matrix. Furthermore, we illustrate that using machine learning model, SFE for complex alloys can be predicted with high accuracy by using the database of the constituent’s binary alloys.
10:30 AM
Machine Learning Guided Descriptor Selection for Predicting Corrosion Resistance in Multi-principal Element Alloys (MPEAs): Ankit Roy1; M. F. N. Taufique2; Hrishabh Khakurel3; Ram Devanathan2; Duane Johnson4; Ganesh Balasubramanian1; 1Lehigh University; 2Pacific Northwest National Lab; 3The University of Texas at Arlington; 4Ames Laboratory
More than $ 270 billion is spent on combatting corrosion annually in the USA alone. This work uses machine learning for the development of highly corrosion resistant alloys. The focus is on a new class of alloys called Multi-Principal-Element Alloys (MPEAs) as a potential solution. MPEAs are composed of multiple elements (4 or more) with arbitrary proportions. Some MPEAs exhibit excellent mechanical properties at high temperatures but their design-search space is half a trillion combinations. To overcome this challenge, we employ machine-learning tools to develop a model that predicts the corrosion resistance of any given MPEA, supported by existing (but limited) corrosion data. Such a model reveals important features that determine the corrosion resistance of a given alloy and serves as a tool for swiftly screening a vast number of MPEAs and selecting the best corrosion-resistant systems.