Thumbnail Image
Publication

Development of a grain boundary genome: atomically-informed defect analysis

Tavenner, Jacob P.
Citations
Altmetric:
Advisor
Tucker, Garritt J.
Editor
Date
Date Issued
2022
Date Submitted
Keywords
atomic descriptors
 computational modeling
 grain boundaries
 machine learning
 mechanical behavior
 molecular dynamics
Collections
2022 - Mines Theses & Dissertations
Show all metadata
Files
Thumbnail Image
Tavenner_mines_0052E_12388.pdf
Adobe PDF, 48.45 MB
Tavenner_mines_0052E_316/GB_Failure.mp4
MP4, 28.58 MB
Tavenner_mines_0052E_316/GB_Decohesion.mp4
MP4, 19.91 MB
Tavenner_mines_0052E_316/GB_Compression.mp4
MP4, 26.33 MB
Tavenner_mines_0052E_316/GB_Tension.mp4
MP4, 29.57 MB
Research Projects
Organizational Units
Journal Issue
URI
https://hdl.handle.net/11124/15418
Embargo Expires
2023-09-30
Abstract
In most metallic systems, the behavior of grain boundaries (GBs) is often categorized by general grain size effects. Although distinct GB structures are known to exhibit drastically different properties, the specific features which give rise to such property variations are largely unknown. To address such questions, this thesis investigates the effects of atomic-scale behaviors on bulk GB properties. Due to the challenging nature of investigating such behaviors, simulation techniques such as Molecular Dynamics are employed to provide fully 3-D, atomically accurate data regarding aluminum GB mechanics. A large database of minimum energy, zero pressure bicrystals is created and tested under a variety of tensile and compressive loading conditions. These mechanical simulations are analyzed to extract key features describing the GB response under the various applied constraints. Analysis of the resulting features provides unique insight into the range of expected behavior across a large distribution of GB characters and illustrates that many traditional methods for determining GB properties are severely lacking. To address such challenges and incorporate atomic information into predictions of GB behavior, a new set of atomic descriptors has been developed, the strain functional descriptors (SFDs). This SFD formulation offers a significant improvement compared to existing atomic descriptors through a complete and minimally-spanning nature over higher-order spatial tensors. Leveraging these SFDs for describing local atomic environments enables analysis of the atomistic origins of complex behaviors such as segregation energy. Through the application of the SFDs on the atomic environments which make up GBs, multiple methods for incorporating atomic information to a GB scale are addressed. Using these newly developed GB descriptions, the variety of GB structures are analyzed across the dataset and compared to previously studied GB structures. Furthermore, the similarity of individual GBs to known defects is analyzed. These analysis methods enable the impact of specific atomic structures on the mechanical response of GBs to be identified.
Associated Publications
Rights
Copyright of the original work is retained by the author.
Embedded videos