[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-121277-en":3,"doc-seo-121277-105":30,"detail-sidebar-cat-0-en-105":95},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},121277,1374391975076,"Riley","https://ap-avatar.wpscdn.com/avatar/14000253ca4ec9f6853?x-image-process=image/resize,m_fixed,w_180,h_180&k=1783305029341752051",8,"Research & Report","Machine learning potentials and investigation of interstitials in metals - Dissertation","Machine learning interatomic potentials are introduced as a computationally efficient alternative to first-principles density-functional-theory simulations for studying materials at the atomic scale. The dissertation constructs a Fe-H machine learning potential for molecular dynamics, trained on DFT molecular-dynamics trajectories using neural networks and atomic energy decomposition to improve training data. Tests verify accurate statistical and dynamic behavior for iron–hydrogen systems, showing hydrogen embrittlement effects that depend on grain-boundary type. An additional Ti-O interstitials potential in ACE format reproduces energies and forces and enables analysis of oxygen segregation at grain boundaries using correlated structural descriptors and ML prediction of segregation energies.","UC Berkeley  \nUC Berkeley Electronic Theses and Dissertations  \nTitle  \nMachine learning potentials and investigation of interstitials in metals  \nPermalink  \n[https://escholarship.org/uc/item/4w54d48p](https://escholarship.org/uc/item/4w54d48p)  \nAuthor  \nZhang, Buyu  \nPublication Date  \n2024  \nPeer reviewed|Thesis/dissertation  \n[eScholarship.org](eScholarship.org) Powered by the California Digital Library  \nUniversity of California  \nMachine Learning Potentials and Investigation of Interstitials in Metals  \nby  \nBuyu Zhang  \nA dissertation submitted in partial satisfaction of the requirements for the degree of  \nDoctor of Philosophy  \nin  \nEngineering – Materials Science and Engineering  \nin the  \nGraduate Division  \nof the  \nUniversity of California, Berkeley  \nCommittee in charge:  \nProfessor Mark Asta, Chair  \nProfessor Daryl Chrzan  \nProfessor Zeyu Zheng  \nFall 2024  \nMachine Learning Potentials and Investigation of Interstitials in Metals  \nCopyright 2024  \nby  \nBuyu Zhang  \n1  \nAbstract  \nMachine Learning Potentials and Investigation of Interstitials in Metals  \nby  \nBuyu Zhang  \nDoctor of Philosophy in Engineering – Materials Science and Engineering University of California, Berkeley  \nProfessor Mark Asta, Chair  \nComputer simulations are an important tool for studying the structure and properties of materials at the atomic scale.. While such simulations can be performed from first-principles density-functional-theory (DFT) calculations directly, the computational cost of such methods scales rapidly with the size of the system. Thus, simulations based on classical interatomic potentials are often used to simulate systems with complex structures (e.g., including defects) with thousands to millions of atoms. The accuracy of such simulations depends sensitively on the classical potential used, and there are limitations to the reliability of commonly used potential models based on simplified mathematical forms. In recent years, machine learning interatomic potentials (MLIPs) have emerged in many areas of material science simulations. Such MLIPs take the pre-calculated DFT data, including total energies and atomic forces, and constructs a force field in an automatic fashion using modern machinelearning approaches. Used within its trained regime, MLIPs have been shown to predict forces and energies with near DFT-level accuracy.  \nIn this dissertation, a MLIP of Fe-H is constructed to study hydrogen embrittlement via molecular dynamics (MD) simulations. The MLIP is obtained by learning the results of the DFT molecular-dynamics (MD) trajectories for various configurations, using a neural network (NN) model. To enhance the amount of data used for the training, we make use of an atomic decomposition of the DFT total energies. Various tests are carried out to ensure that the MLIP describes well the statistic and dynamic properties of iron + hydrogen systems. Using this MLIP, MD simulations of model iron samples with small pre-crack tip at different grain boundaries are carried out with different loadings. We found that a high concentration of H ahead the crack tips will enhance the propagation of the crack, while the degree of enhancement depends on the type of the grain boundary. We also observed the formation of microvoids ahead of the crack, that appear to contribute to its propagation.  \nAdditionally, a MLIP for Ti-O interstitials is developed using the atomic cluster expansion (ACE) format. This ACE potential shows good agreement with reference DFT calculations  \n2  \nfor energies and forces, as well as properties including twin formation energy, interstitial formation energies and segregation energies at grain boundaries (GBs) . We explore the distribution of oxygen segregation energies in a number of GBs in hexagonal-close-packed α-Ti, and analyze some physcially informed features like Voronoi indices and atomic volume that are highly correlated with the segregation energy. Further, we train a machine learning model to pr","cbCaikxN1ZgGWOjR","https://ap.wps.com/l/cbCaikxN1ZgGWOjR","pdf",21874941,1,92,"English","en",105,"# Introduction\n## Machine learning potentials\n## Grain boundary structures and properties\n## Grain boundary segregation\n## Dissertation Outline\n# Theory and Simulation Methods\n## Density functional theory\n## Molecular Dynamics\n## Machine Learning\n## ML potential models\n# Machine learning force field for Fe-H and investigation on role of hydrogen on the crack propagation\n## Introduction\n## Methods\n## Results\n## MD simulations\n## Conclusion\n# Oxygen Grain-Boundary Segregation in HCP Ti-Computational Investigations Using an ACE Potential\n## Introduction\n## Methods\n## Potential Training and Property Predictions\n## Oxygen grain-boundary segregation energies in HCP Ti\n## Summary\n# Summary and future work\n## Summary of results\n## Future work","[{\"question\":\"Why are machine learning interatomic potentials useful compared with DFT-based simulations?\",\"answer\":\"They reduce computational cost while maintaining near DFT-level accuracy within the trained regime by learning force fields directly from pre-calculated DFT energies and forces.\"},{\"question\":\"How is the Fe-H machine learning potential constructed and trained?\",\"answer\":\"A neural-network MLIP is trained on DFT molecular-dynamics trajectories for different configurations, enhanced by atomic decomposition of DFT total energies.\"},{\"question\":\"What do the simulations reveal about hydrogen embrittlement in iron?\",\"answer\":\"High hydrogen concentration ahead of crack tips increases crack propagation, with the enhancement depending on the grain boundary type and accompanied by microvoid formation contributing to propagation.\"},{\"question\":\"How is oxygen segregation in HCP Ti investigated in the dissertation?\",\"answer\":\"An ACE-format Ti-O interstitial potential is used to match DFT for energies and forces, then oxygen segregation energies are analyzed across grain boundaries using correlated descriptors and a ML model trained to predict segregation from unrelaxed structures.\"}]","Machine learning potentials and investigation of interstitials in metals - Dissertation | PDF",1785734859,232,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":90,"head_meta":92,"extra_data":94,"updated_unix":28},"machine-learning-potentials-and-investigation-of-interstitials-in-metals-dissertation","",{"@graph":36,"@context":89},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/machine-learning-potentials-and-investigation-of-interstitials-in-metals-dissertation/121277/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-03",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81,85],{"name":72,"@type":73,"acceptedAnswer":74},"Why are machine learning interatomic potentials useful compared with DFT-based simulations?","Question",{"text":75,"@type":76},"They reduce computational cost while maintaining near DFT-level accuracy within the trained regime by learning force fields directly from pre-calculated DFT energies and forces.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is the Fe-H machine learning potential constructed and trained?",{"text":80,"@type":76},"A neural-network MLIP is trained on DFT molecular-dynamics trajectories for different configurations, enhanced by atomic decomposition of DFT total energies.",{"name":82,"@type":73,"acceptedAnswer":83},"What do the simulations reveal about hydrogen embrittlement in iron?",{"text":84,"@type":76},"High hydrogen concentration ahead of crack tips increases crack propagation, with the enhancement depending on the grain boundary type and accompanied by microvoid formation contributing to propagation.",{"name":86,"@type":73,"acceptedAnswer":87},"How is oxygen segregation in HCP Ti investigated in the dissertation?",{"text":88,"@type":76},"An ACE-format Ti-O interstitial potential is used to match DFT for energies and forces, then oxygen segregation energies are analyzed across grain boundaries using correlated descriptors and a ML model trained to predict segregation from unrelaxed structures.","https://schema.org",{"og:url":52,"og:type":91,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":93,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":96},[97,101,105,109,114,119,124,127,132,135,139],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":106,"show_sort_weight":107,"slug":108},"Exam",70,"exam",{"id":110,"doc_module":4,"doc_module_name":46,"category_name":111,"show_sort_weight":112,"slug":113},5,"Comic",60,"comic",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},6,"Technology",50,"technology",{"id":120,"doc_module":4,"doc_module_name":46,"category_name":121,"show_sort_weight":122,"slug":123},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":125,"slug":126},30,"research-report",{"id":128,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":130,"slug":131},9,"Religion & Spirituality",20,"religion-spirituality",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":130,"slug":134},"World Cup","world-cup",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":136,"slug":138},10,"Lifestyle","lifestyle",{"id":140,"doc_module":4,"doc_module_name":46,"category_name":141,"show_sort_weight":110,"slug":142},19,"General","general"]