Session: 15-01-01: ASME International Undergraduate Research and Design Exposition

Paper Number: 144641

144641 - Investigating Creep Behavior and Short-Range Ordering in High-Entropy Alloy: A Data Driven Molecular Dynamics Study 

Research Summary: The analysis of creep behavior in Al_xCoCrFeNi_y alloy through molecular dynamics (MD) and deep learning is crucial for its widespread engineering use. Renowned for its high-temperature strength, the alloy finds applications in aerospace, engine parts, turbine blades etc. This study examines the impact of temperature, pressure, and composition variations on creep properties, comparing HEA polycrystalline with short-range ordered equilibrated polycrystalline using MD. Through short-range ordering (SRO) this study investigates how nanostructures affect the strength and ductility of HEAs. Deep learning aids in predicting optimal operating conditions and alloying element ratios for improved creep performance. By optimizing model performance with hyperparameter tuning, this research enhances understanding of high-temperature deformation and creep behavior in HEAs, ensuring their reliable application. 

Background: High-entropy alloys (HEAs), with equal atomic quantities of multiple elements, typically form simple solid-solution structures like FCC/BCC due to strong mixing entropy effects. Combining FCC/BCC structures provides a balance between strength and ductility. AlCrCoFeNi HEA, for example, exhibits multicomponent Al-lean FCC and Al-rich BCC phases with distinct elemental preferences. To further explore this, we focus on the Al_xCoCrFeNi_y alloy system, where x and y represent the ratios of Al and Ni content, respectively. Previous research on the AlCoCrFeNi_2.1 alloy highlighted microstructural evolution in creep deformation but overlooked short-range ordering (SRO). Recent studies emphasize the role of SRO in enhancing strength and ductility simultaneously in HEAs.

AlCrCoFeNi is utilized in the manufacturing of structural components for aircraft and automobiles due to its high strength and resistance to fatigue and corrosion. MD simulations aid in understanding the creep behavior, crucial for optimizing its composition and processing, enhancing creep resistance. Here SRO aids in designing components for improved mechanical properties. Hence, AlCrCoFeNi is a suitable material for conducting molecular dynamics simulation for creep analysis.

Our study integrates molecular dynamics (MD) and deep learning to analyze SRO-induced creep behavior, aiming to predict lifetimes and improve performance for engineering applications.

Methodology: Phase 1: Molecular Dynamics (MD) study of Al_xCoCrFeNi_y High-Entropy Alloy: Large-scale Atomic/Molecular Massively Parallel Simulator (LAMMPS) will be used to run all MD simulations. The simulation data will be visualized using the Open Visualization Tool Pro (OVITO Pro).  

1. Interatomic Potential for Al_xCoCrFeNi_y:The interatomic potential required for this study was developed by Farkas et al. The potential utilizes the embedded atom method (EAM).

2. Preparation of Al_xCoCrFeNi_y sample with Short-Range Ordering (SRO): The sample was generated by the voronoi tessellation method within Atomsk. A single crystalline Al structure with a lattice constant of a = 4.046 Å and size 20a × 20a × 20a containing 32,000 atoms was created. Then Al atoms were replaced with Co, Cr, Fe, and Ni based on ratios, yielding a single-crystalline FCC structure. Finally, a polycrystalline Al_xCoCrFeNi_y sample (125 Å × 125 Å × 125 Å, 10 randomly oriented grains, 1,17,929 atoms) was generated. The system was equilibrated using the Conjugate Gradient (CG) method until the force reached 1x10^-6 eV/atom and the total energy fluctuation of the system was under 1x10^-8 eV. Isothermal-isobaric (NPT) ensemble relaxation followed at 300K and 1250 bars. Monte Carlo (MC) simulations facilitated atom swapping and NPT relaxation to observe Short-Range Ordering (SRO) at 500K.

3. Application of Constant Height Temperature and Stress for the Study of Creep: After equilibration, stress and temperature will be varied for both polycrystalline and SRO-equilibrated structures. We will vary the temperature in the 0.3T_m - 0.5T_m range where T_m is the melting temperature of the alloy. The stress is varied in the 0.5 S_y - 0.7S_y range with S_y being the yield strength of the material. We will evaluate the material under lean Al and rich Ni conditions, known to enhance overall creep performance of the alloy. The testing conditions are:

Temperature - 500K, 700K, 900K, 1000K

Stress - 1GPa, 1.25 GPa, 1.5GPa, 1.75 GPa

SRO compositions - Al_0.1CoCrFeNi, Al_0.12CoCrFeNi, Al_0.15CoCrFeNi, AlCoCrFeNi_2.8

4, Visualization and analysis of the MD Open Visualization Tool Pro (OVITO Pro): We will generate Strain vs Time plot using Python script with MD creep simulation data file. The steady-state creep rate (SSCR) will also be calculated. Using the dump files, we will analyze mean square displacement for each atom in OVITO PRO, obtaining the Diffusion Coefficient. The dislocation analysis will be done with the OVITO PRO software. Moreover, we intend to employ it for the generation of simulated XRD patterns. All this data will be combined in a CSV file as input for the deep learning analysis. 

Phase 2: Lifetime prediction using Deep Learning: The model inputs include temperature, stress, and composition, while Long Short-Term Memory (LSTM) networks aid in time series predictions. We will predict the operating conditions for Al_xCoCrFeNi_y and the optimal ratio of alloying elements for better creep performance. 

Results and Conclusion: The study conducted a creep test on Al_xCoCrFeNi_y to validate its setup, which was conducted on a continuum scale and a nanoscale. The results showed that the Al_0.15CoCrFeNi alloy had a higher creep resistant property than the Al_0.60CoCrFeNi alloy. As the Al concentration increased, the creep curve became steeper and the steady state creep region became shorter. Monte Carlo (MC) simulations were used to simulate the polycrystalline structure at 500K, and the phase concentration was visualized in OVITO. Tensile tests were conducted on each specimen for 25,00,000 timesteps, revealing that lower Al concentration resulted in higher yield strength and Young's modulus. This suggests that a lower Al concentration may lead to better creep resistance. In a nutshell, this study focuses on the creep performance of a high entropy alloy varying temperature, stress, and composition. Short-range ordering is used to minimize potential energy and equilibrate the polycrystalline structure. The study will also investigate creep performance using deep learning and varying temperatures and stresses, aiming to improve creep performance and application of the alloy in various applications.

Presenting Author: Afia Zaman Bangladesh University of Engineering and Technology

Presenting Author Biography: Afia Zaman is a dedicated researcher and an undergraduate student, working towards a BSc. in Mechanical Engineering at Bangladesh University of Engineering and Technology. She is characterized by her calm demeanor, composed nature, and exceptional problem-solving skills. Her work primarily revolves around robot and machine learning based projects, where she applied her keen analytical abilities and innovative approaches to problem-solving. Outside of her academic pursuits, she is an active participant in various club activities on campus. Whether it's organizing events, leading workshops, or participating in community outreach programs, she eagerly contributes her time and energy to enriching the university experience for her peers. In her free time, she enjoys painting to express herself creatively.

Authors:

Munirul Alam Bangladesh University of Engineering and Technology
Afia Zaman Bangladesh University of Engineering and Technology
Anik Shabab Soudha Bangladesh University of Engineering and Technology