Hybrid computational modeling of thermomagnetic material systems

Computational material science;Rare-earth material;Monte-Carlo simulation;Computational physics;High performance computing;GPGPU;DFT

Kim, Sookyung Kyung ; Garmestini, Hamid Materials Science and Engineering Jang, Seung Soon Fujimoto, Richard Deo, Chaitanya S. Benedict, Lorin Khaleel, Mohammad A. ; Garmestini, Hamid

University:Georgia Institute of Technology

Department:Materials Science and Engineering

关键词: Computational material science;    Rare-earth material;    Monte-Carlo simulation;    Computational physics;    High performance computing;    GPGPU;    DFT;   

Others  :  https://smartech.gatech.edu/bitstream/1853/58213/1/KIM-DISSERTATION-2017.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

Current knowledge of computational material modeling for engineering demands accurate prediction of thermo-magnetic properties of material. Different of computational modeling approaches should be considered and selected in a way that fit the best to the specific material systems. In general, thermo-magnetic properties of material should benefit from ab-initio Density Functional Theory (DFT) calculation in some degree. However, DFT alone has the limitation in fully modeling finite temperature properties in that the concept of statistical physics such as magnetic excitation and magnetic spin interaction are not considered in DFT. The nature of atomic scale simulation also made it difficult to extend to meso-scale simulation. Therefore, a promising route toward this goal is a combination of DFT with concepts of statistical physics, which was shown to yield accurate predictions for a wide range of magnetic and nonmagnetic materials. There are two aims for this work.Firstly, a review and comparison of various computational modeling techniques currently available for predicting thermo-magnetic properties of materials is presented. Specifically, different approaches for those computational modeling methods are presented for different material systems. Secondly, new computational modeling frameworks based on currently available methodologies is developed and proposed for particular material systems for engineering task purposes. (1) For rare-earth replacement permanent magnets, the new program combining DFT based Korringa–Kohn–Rostoker (KKR) calculation and Heisenberg Monte Carlo has been developed and applied to (Fe1-xCox)2B. (2) For stainless steel, the new quantum-mechanically driven computational material discovery framework is proposed. (3) For meso-scale simulation of strong ferromagnetic material, GPU based parallel computing technique has been applied for Ising Monte-Carlo simulation and applied to Fe. The results from the proposed modeling routines show that we can achieve our exact aim to understand better the theoretical origin of thermo-magnetic properties of different material systems.