[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-120347-en":3,"doc-seo-120347-105":30,"detail-sidebar-cat-0-en-105":91},{"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},120347,1099514067438,"River Wang","https://ap-avatar.wpscdn.com/avatar/100002539ee87300030?x-image-process=image/resize,m_fixed,w_180,h_180&k=1780474512215547542",8,"Research & Report","Accelerating High-Throughput Phonon Calculations via Machine Learning Universal Potentials","Phonons are central to predicting material properties, yet conventional phonon calculations remain computationally expensive and restrict high-throughput applications. This study accelerates harmonic phonon calculations using machine learning universal potentials (MLIPs) with an efficient training dataset generation strategy. Instead of single-atom displacements across many supercells, it uses randomly perturbed all-atom structures. A MACE-based MLIP is trained on DFT data covering 2,738 materials and 77 elements, totaling 15,670 supercell structures, and is validated on a held-out set with low errors, stability classification, and thermodynamic agreement.","Portland State University  \nPDXScholar  \n\n| Mechanical and Materials Engineering Faculty Publications and Presentations | Mechanical and Materials Engineering |\n| --- | --- |\n| 4-1-2025\u003Cbr>Accelerating High-Throughput Phonon Calculations via Machine Learning Universal Potentials\u003Cbr>Huiju Lee\u003Cbr>Portland State University\u003Cbr>Vinay I. Hegde\u003Cbr>Northwestern University\u003Cbr>Chris Wolverton\u003Cbr>Northwestern University\u003Cbr>Yi Xia\u003Cbr>Portland State University, [yxia@pdx.edu](yxia@pdx.edu)\u003Cbr>Follow this and additional works at: [https://pdxscholar.library.pdx.edu/mengin_fac](https://pdxscholar.library.pdx.edu/mengin_fac)\u003Cbr> Part of the Mechanical Engineering Commons\u003Cbr>Let us know how access to this document benefits you. |  |\n\nCitation Details  \nPublished as: Lee, H., Hegde, V. I., Wolverton, C., & Xia, Y. (2025) . Accelerating high-throughput phonon calculations via machine learning universal potentials. Materials Today Physics, 53, 101688.  \nThis Pre-Print is brought to you for free and open access. It has been accepted for inclusion in Mechanical and Materials Engineering Faculty Publications and Presentations by an authorized administrator of PDXScholar. Please contact us if we can make this document more accessible: [pdxscholar@pdx.edu](pdxscholar@pdx.edu).  \nAccelerating High-Throughput Phonon Calculations via Machine Learning Universal  \nPotentials  \nHuiju Lee, 1 Vinay I. Hegde,2 Chris Wolverton,2 and Yi Xia 1, ∗  \n1 Department of Mechanical and Materials Engineering, Portland State University, Portland, OR 97201, USA  \n2 Department of Materials Science and Engineering, Northwestern University, Evanston, IL 60208, USA (Dated: April 2, 2025)  \nPhonons play a critical role in determining various material properties, but conventional methods for phonon calculations are computationally intensive, limiting their broad applicability. In this study, we present an approach to accelerate high-throughput harmonic phonon calculations using machine learning universal potentials (MLIPs) combined with an efficient training dataset generation strategy. Instead of computing phonon properties from a large number of supercells with small atomic displacements of a single atom, our approach uses a smaller subset of supercell structures with all atoms randomly perturbed by displacements ranging from 0 .01 to 0 .05 ˚A, significantly reducing computational costs. We train a state-of-the-art MLIP based on multi-atomic cluster expansion (MACE), on a comprehensive dataset of 2,738 materials with 77 elements, totaling 15,670 supercell structures, computed using high-fidelity density functional theory (DFT) calculations. The trained model is validated against phonon calculations for a held-out subset of 384 materials, achieving a mean absolute error (MAE) of 0.18 THz for vibrational frequencies from full phonon dispersions, 2.19 meV/atom for Helmholtz vibrational free energies at 300K, as well as a classification accuracy of 86 .2% for dynamical stability of materials. A thermodynamic analysis of polymorphic stability in  \n126 systems demonstrates good agreement with DFT results at 300 K and 1000 K. In addition, the diverse and extensive high-quality DFT dataset curated in this study serves as a valuable resource  \nfor researchers to train and improve other machine learning interatomic potential models.  \nI. INTRODUCTION  \nPhonons, as quasiparticles representing the collective vibrational modes of atoms within a crystalline material, are ubiquitous and play a crucial role in determining various material properties, including thermal conductivity, mechanical behavior, electrical conductivity, and superconductivity [1–3] . Additionally, they are vital in assessing dynamic and thermodynamic stability, as well as phase transitions of crystalline materials. This is particularly significant in materials discovery, where accurately predicting stability is essential for identifying new materials with desired properties [4–6] .  \nOne popular approach for first-princ","cbCaikAsNcHAK6Sb","https://ap.wps.com/l/cbCaikAsNcHAK6Sb","pdf",3504742,1,20,"English","en",105,"# Introduction\n## Conventional finite-displacement phonon calculations\n## Computational limits for high-throughput screening\n## Machine learning approaches for phonons\n## Direct prediction of phonon properties using ML models","[{\"question\":\"Why are conventional phonon calculations computationally intensive?\",\"answer\":\"Finite-displacement phonon methods require many DFT supercell calculations to capture interactions across different ranges and reach converged results, making them costly for large or low-symmetry systems.\"},{\"question\":\"What training-data strategy is used to accelerate high-throughput harmonic phonon calculations?\",\"answer\":\"The approach generates a smaller subset of supercell structures where all atoms are randomly perturbed with displacements between 0.01 and 0.05 Å, reducing the number of required computations.\"},{\"question\":\"How is the ML potential validated and what performance is reported?\",\"answer\":\"The trained MACE-based MLIP is validated on a held-out subset of materials, achieving mean absolute errors for vibrational frequencies and Helmholtz free energies at 300 K, plus an accuracy for dynamical stability classification, with thermodynamic agreement across temperatures.\"}]","Accelerating High-Throughput Phonon Calculations via Machine Learning Universal Potentials | PDF",1785729604,50,{"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":86,"head_meta":88,"extra_data":90,"updated_unix":28},"accelerating-high-throughput-phonon-calculations-via-machine-learning-universal-potentials","",{"@graph":36,"@context":85},[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/accelerating-high-throughput-phonon-calculations-via-machine-learning-universal-potentials/120347/",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],{"name":72,"@type":73,"acceptedAnswer":74},"Why are conventional phonon calculations computationally intensive?","Question",{"text":75,"@type":76},"Finite-displacement phonon methods require many DFT supercell calculations to capture interactions across different ranges and reach converged results, making them costly for large or low-symmetry systems.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What training-data strategy is used to accelerate high-throughput harmonic phonon calculations?",{"text":80,"@type":76},"The approach generates a smaller subset of supercell structures where all atoms are randomly perturbed with displacements between 0.01 and 0.05 Å, reducing the number of required computations.",{"name":82,"@type":73,"acceptedAnswer":83},"How is the ML potential validated and what performance is reported?",{"text":84,"@type":76},"The trained MACE-based MLIP is validated on a held-out subset of materials, achieving mean absolute errors for vibrational frequencies and Helmholtz free energies at 300 K, plus an accuracy for dynamical stability classification, with thermodynamic agreement across temperatures.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,114,119,122,126,129,133],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":29,"slug":113},6,"Technology","technology",{"id":115,"doc_module":4,"doc_module_name":46,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":21,"slug":125},9,"Religion & Spirituality","religion-spirituality",{"id":21,"doc_module":4,"doc_module_name":46,"category_name":127,"show_sort_weight":21,"slug":128},"World Cup","world-cup",{"id":130,"doc_module":4,"doc_module_name":46,"category_name":131,"show_sort_weight":130,"slug":132},10,"Lifestyle","lifestyle",{"id":134,"doc_module":4,"doc_module_name":46,"category_name":135,"show_sort_weight":106,"slug":136},19,"General","general"]