Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning

Silicon nitride (Si3N4) is an extensively used material in the automotive, aerospace, and semiconductor industries. However, its widespread use is in contrast to the scarce availability of reliable interatomic potentials that can be employed to study various aspects of this material on an atomistic...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:The Journal of chemical physics 2023-05, Vol.158 (19)
Hauptverfasser: Milardovich, Diego, Wilhelmer, Christoph, Waldhoer, Dominic, Cvitkovich, Lukas, Sivaraman, Ganesh, Grasser, Tibor
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue 19
container_start_page
container_title The Journal of chemical physics
container_volume 158
creator Milardovich, Diego
Wilhelmer, Christoph
Waldhoer, Dominic
Cvitkovich, Lukas
Sivaraman, Ganesh
Grasser, Tibor
description Silicon nitride (Si3N4) is an extensively used material in the automotive, aerospace, and semiconductor industries. However, its widespread use is in contrast to the scarce availability of reliable interatomic potentials that can be employed to study various aspects of this material on an atomistic scale, particularly its amorphous phase. In this work, we developed a machine learning interatomic potential, using an efficient active learning technique, combined with the Gaussian approximation potential (GAP) method. Our strategy is based on using an inexpensive empirical potential to generate an initial dataset of atomic configurations, for which energies and forces were recalculated with density functional theory (DFT); thereafter, a GAP was trained on these data and an iterative re-training algorithm was used to improve it by learning on-the-fly. When compared to DFT, our potential yielded a mean absolute error of 8 meV/atom in energy calculations for a variety of liquid and amorphous structures and a speed-up of molecular dynamics simulations by 3–4 orders of magnitude, while achieving a first-rate agreement with experimental results. Our potential is publicly available in an open-access repository.
doi_str_mv 10.1063/5.0146753
format Article
fullrecord <record><control><sourceid>proquest_scita</sourceid><recordid>TN_cdi_scitation_primary_10_1063_5_0146753</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2813728606</sourcerecordid><originalsourceid>FETCH-LOGICAL-c445t-1b623a5892d1135fed0f8ea2b4b854e27539b6c59ef265e76aba81103bfcb8533</originalsourceid><addsrcrecordid>eNp90V1LHDEUBuAgim63vfAPlMHeqDA235O5LGJbYW0v2r0OmcyZmmU2WZOs4L83ulsXBL0KBx5ezsmL0DHBFwRL9lVcYMJlI9gemhCs2rqRLd5HE4wpqVuJ5RH6kNICY0wayg_REWuI4mWYoPmNsbfOQzWCid75f5XzGaLJYelstQoZfHZmrIYQq-RGZ4OvvcvR9VCd_nHsFz-ruofK2OzudyEf0cFgxgSftu8Uzb9f_b38Wc9-_7i-_DarLeci16STlBmhWtoTwsQAPR4UGNrxTgkOtFzUdtKKFgYqBTTSdEYRglk32CIYm6KTTW5I2elkXQZ7W1b0YLOmHHNevmeKTjdoFcPdGlLWS5csjKPxENZJU0WYUqptSKFfXtFFWEdfTnhWDVXyOfBso2wMKUUY9Cq6pYkPmmD9VIgWeltIsZ-3ietuCf2L_N9AAecb8LS9yS74d9PexPch7qBe9QN7BFz8nzY</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2813728606</pqid></control><display><type>article</type><title>Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning</title><source>AIP Journals Complete</source><source>Alma/SFX Local Collection</source><creator>Milardovich, Diego ; Wilhelmer, Christoph ; Waldhoer, Dominic ; Cvitkovich, Lukas ; Sivaraman, Ganesh ; Grasser, Tibor</creator><creatorcontrib>Milardovich, Diego ; Wilhelmer, Christoph ; Waldhoer, Dominic ; Cvitkovich, Lukas ; Sivaraman, Ganesh ; Grasser, Tibor ; Argonne National Laboratory (ANL), Argonne, IL (United States)</creatorcontrib><description>Silicon nitride (Si3N4) is an extensively used material in the automotive, aerospace, and semiconductor industries. However, its widespread use is in contrast to the scarce availability of reliable interatomic potentials that can be employed to study various aspects of this material on an atomistic scale, particularly its amorphous phase. In this work, we developed a machine learning interatomic potential, using an efficient active learning technique, combined with the Gaussian approximation potential (GAP) method. Our strategy is based on using an inexpensive empirical potential to generate an initial dataset of atomic configurations, for which energies and forces were recalculated with density functional theory (DFT); thereafter, a GAP was trained on these data and an iterative re-training algorithm was used to improve it by learning on-the-fly. When compared to DFT, our potential yielded a mean absolute error of 8 meV/atom in energy calculations for a variety of liquid and amorphous structures and a speed-up of molecular dynamics simulations by 3–4 orders of magnitude, while achieving a first-rate agreement with experimental results. Our potential is publicly available in an open-access repository.</description><identifier>ISSN: 0021-9606</identifier><identifier>EISSN: 1089-7690</identifier><identifier>DOI: 10.1063/5.0146753</identifier><identifier>PMID: 37184017</identifier><identifier>CODEN: JCPSA6</identifier><language>eng</language><publisher>United States: American Institute of Physics</publisher><subject>Active learning ; Aerospace industry ; Algorithms ; Amorphous materials ; Availability ; Density functional theory ; INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY ; Machine learning ; Molecular dynamics ; Silicon nitride</subject><ispartof>The Journal of chemical physics, 2023-05, Vol.158 (19)</ispartof><rights>Author(s)</rights><rights>2023 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c445t-1b623a5892d1135fed0f8ea2b4b854e27539b6c59ef265e76aba81103bfcb8533</citedby><cites>FETCH-LOGICAL-c445t-1b623a5892d1135fed0f8ea2b4b854e27539b6c59ef265e76aba81103bfcb8533</cites><orcidid>0000-0001-6536-2238 ; 0000-0003-2453-1693 ; 0000-0002-9996-2589 ; 0000-0003-2453-507X ; 0000-0002-8631-5681 ; 0000-0001-9056-9855 ; 0000000286315681 ; 0000000190569855 ; 0000000165362238 ; 0000000324531693 ; 000000032453507X ; 0000000299962589</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://pubs.aip.org/jcp/article-lookup/doi/10.1063/5.0146753$$EHTML$$P50$$Gscitation$$Hfree_for_read</linktohtml><link.rule.ids>230,314,776,780,790,881,4498,27901,27902,76126</link.rule.ids><backlink>$$Uhttps://www.ncbi.nlm.nih.gov/pubmed/37184017$$D View this record in MEDLINE/PubMed$$Hfree_for_read</backlink><backlink>$$Uhttps://www.osti.gov/servlets/purl/2404406$$D View this record in Osti.gov$$Hfree_for_read</backlink></links><search><creatorcontrib>Milardovich, Diego</creatorcontrib><creatorcontrib>Wilhelmer, Christoph</creatorcontrib><creatorcontrib>Waldhoer, Dominic</creatorcontrib><creatorcontrib>Cvitkovich, Lukas</creatorcontrib><creatorcontrib>Sivaraman, Ganesh</creatorcontrib><creatorcontrib>Grasser, Tibor</creatorcontrib><creatorcontrib>Argonne National Laboratory (ANL), Argonne, IL (United States)</creatorcontrib><title>Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning</title><title>The Journal of chemical physics</title><addtitle>J Chem Phys</addtitle><description>Silicon nitride (Si3N4) is an extensively used material in the automotive, aerospace, and semiconductor industries. However, its widespread use is in contrast to the scarce availability of reliable interatomic potentials that can be employed to study various aspects of this material on an atomistic scale, particularly its amorphous phase. In this work, we developed a machine learning interatomic potential, using an efficient active learning technique, combined with the Gaussian approximation potential (GAP) method. Our strategy is based on using an inexpensive empirical potential to generate an initial dataset of atomic configurations, for which energies and forces were recalculated with density functional theory (DFT); thereafter, a GAP was trained on these data and an iterative re-training algorithm was used to improve it by learning on-the-fly. When compared to DFT, our potential yielded a mean absolute error of 8 meV/atom in energy calculations for a variety of liquid and amorphous structures and a speed-up of molecular dynamics simulations by 3–4 orders of magnitude, while achieving a first-rate agreement with experimental results. Our potential is publicly available in an open-access repository.</description><subject>Active learning</subject><subject>Aerospace industry</subject><subject>Algorithms</subject><subject>Amorphous materials</subject><subject>Availability</subject><subject>Density functional theory</subject><subject>INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY</subject><subject>Machine learning</subject><subject>Molecular dynamics</subject><subject>Silicon nitride</subject><issn>0021-9606</issn><issn>1089-7690</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><recordid>eNp90V1LHDEUBuAgim63vfAPlMHeqDA235O5LGJbYW0v2r0OmcyZmmU2WZOs4L83ulsXBL0KBx5ezsmL0DHBFwRL9lVcYMJlI9gemhCs2rqRLd5HE4wpqVuJ5RH6kNICY0wayg_REWuI4mWYoPmNsbfOQzWCid75f5XzGaLJYelstQoZfHZmrIYQq-RGZ4OvvcvR9VCd_nHsFz-ruofK2OzudyEf0cFgxgSftu8Uzb9f_b38Wc9-_7i-_DarLeci16STlBmhWtoTwsQAPR4UGNrxTgkOtFzUdtKKFgYqBTTSdEYRglk32CIYm6KTTW5I2elkXQZ7W1b0YLOmHHNevmeKTjdoFcPdGlLWS5csjKPxENZJU0WYUqptSKFfXtFFWEdfTnhWDVXyOfBso2wMKUUY9Cq6pYkPmmD9VIgWeltIsZ-3ietuCf2L_N9AAecb8LS9yS74d9PexPch7qBe9QN7BFz8nzY</recordid><startdate>20230521</startdate><enddate>20230521</enddate><creator>Milardovich, Diego</creator><creator>Wilhelmer, Christoph</creator><creator>Waldhoer, Dominic</creator><creator>Cvitkovich, Lukas</creator><creator>Sivaraman, Ganesh</creator><creator>Grasser, Tibor</creator><general>American Institute of Physics</general><general>American Institute of Physics (AIP)</general><scope>AJDQP</scope><scope>NPM</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>8FD</scope><scope>H8D</scope><scope>L7M</scope><scope>7X8</scope><scope>OIOZB</scope><scope>OTOTI</scope><orcidid>https://orcid.org/0000-0001-6536-2238</orcidid><orcidid>https://orcid.org/0000-0003-2453-1693</orcidid><orcidid>https://orcid.org/0000-0002-9996-2589</orcidid><orcidid>https://orcid.org/0000-0003-2453-507X</orcidid><orcidid>https://orcid.org/0000-0002-8631-5681</orcidid><orcidid>https://orcid.org/0000-0001-9056-9855</orcidid><orcidid>https://orcid.org/0000000286315681</orcidid><orcidid>https://orcid.org/0000000190569855</orcidid><orcidid>https://orcid.org/0000000165362238</orcidid><orcidid>https://orcid.org/0000000324531693</orcidid><orcidid>https://orcid.org/000000032453507X</orcidid><orcidid>https://orcid.org/0000000299962589</orcidid></search><sort><creationdate>20230521</creationdate><title>Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning</title><author>Milardovich, Diego ; Wilhelmer, Christoph ; Waldhoer, Dominic ; Cvitkovich, Lukas ; Sivaraman, Ganesh ; Grasser, Tibor</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c445t-1b623a5892d1135fed0f8ea2b4b854e27539b6c59ef265e76aba81103bfcb8533</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Active learning</topic><topic>Aerospace industry</topic><topic>Algorithms</topic><topic>Amorphous materials</topic><topic>Availability</topic><topic>Density functional theory</topic><topic>INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY</topic><topic>Machine learning</topic><topic>Molecular dynamics</topic><topic>Silicon nitride</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Milardovich, Diego</creatorcontrib><creatorcontrib>Wilhelmer, Christoph</creatorcontrib><creatorcontrib>Waldhoer, Dominic</creatorcontrib><creatorcontrib>Cvitkovich, Lukas</creatorcontrib><creatorcontrib>Sivaraman, Ganesh</creatorcontrib><creatorcontrib>Grasser, Tibor</creatorcontrib><creatorcontrib>Argonne National Laboratory (ANL), Argonne, IL (United States)</creatorcontrib><collection>AIP Open Access Journals</collection><collection>PubMed</collection><collection>CrossRef</collection><collection>Technology Research Database</collection><collection>Aerospace Database</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>MEDLINE - Academic</collection><collection>OSTI.GOV - Hybrid</collection><collection>OSTI.GOV</collection><jtitle>The Journal of chemical physics</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Milardovich, Diego</au><au>Wilhelmer, Christoph</au><au>Waldhoer, Dominic</au><au>Cvitkovich, Lukas</au><au>Sivaraman, Ganesh</au><au>Grasser, Tibor</au><aucorp>Argonne National Laboratory (ANL), Argonne, IL (United States)</aucorp><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning</atitle><jtitle>The Journal of chemical physics</jtitle><addtitle>J Chem Phys</addtitle><date>2023-05-21</date><risdate>2023</risdate><volume>158</volume><issue>19</issue><issn>0021-9606</issn><eissn>1089-7690</eissn><coden>JCPSA6</coden><abstract>Silicon nitride (Si3N4) is an extensively used material in the automotive, aerospace, and semiconductor industries. However, its widespread use is in contrast to the scarce availability of reliable interatomic potentials that can be employed to study various aspects of this material on an atomistic scale, particularly its amorphous phase. In this work, we developed a machine learning interatomic potential, using an efficient active learning technique, combined with the Gaussian approximation potential (GAP) method. Our strategy is based on using an inexpensive empirical potential to generate an initial dataset of atomic configurations, for which energies and forces were recalculated with density functional theory (DFT); thereafter, a GAP was trained on these data and an iterative re-training algorithm was used to improve it by learning on-the-fly. When compared to DFT, our potential yielded a mean absolute error of 8 meV/atom in energy calculations for a variety of liquid and amorphous structures and a speed-up of molecular dynamics simulations by 3–4 orders of magnitude, while achieving a first-rate agreement with experimental results. Our potential is publicly available in an open-access repository.</abstract><cop>United States</cop><pub>American Institute of Physics</pub><pmid>37184017</pmid><doi>10.1063/5.0146753</doi><tpages>8</tpages><orcidid>https://orcid.org/0000-0001-6536-2238</orcidid><orcidid>https://orcid.org/0000-0003-2453-1693</orcidid><orcidid>https://orcid.org/0000-0002-9996-2589</orcidid><orcidid>https://orcid.org/0000-0003-2453-507X</orcidid><orcidid>https://orcid.org/0000-0002-8631-5681</orcidid><orcidid>https://orcid.org/0000-0001-9056-9855</orcidid><orcidid>https://orcid.org/0000000286315681</orcidid><orcidid>https://orcid.org/0000000190569855</orcidid><orcidid>https://orcid.org/0000000165362238</orcidid><orcidid>https://orcid.org/0000000324531693</orcidid><orcidid>https://orcid.org/000000032453507X</orcidid><orcidid>https://orcid.org/0000000299962589</orcidid><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 0021-9606
ispartof The Journal of chemical physics, 2023-05, Vol.158 (19)
issn 0021-9606
1089-7690
language eng
recordid cdi_scitation_primary_10_1063_5_0146753
source AIP Journals Complete; Alma/SFX Local Collection
subjects Active learning
Aerospace industry
Algorithms
Amorphous materials
Availability
Density functional theory
INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CHEMISTRY
Machine learning
Molecular dynamics
Silicon nitride
title Machine learning interatomic potential for silicon-nitride (Si3N4) by active learning
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T19%3A51%3A11IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_scita&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Machine%20learning%20interatomic%20potential%20for%20silicon-nitride%20(Si3N4)%20by%20active%20learning&rft.jtitle=The%20Journal%20of%20chemical%20physics&rft.au=Milardovich,%20Diego&rft.aucorp=Argonne%20National%20Laboratory%20(ANL),%20Argonne,%20IL%20(United%20States)&rft.date=2023-05-21&rft.volume=158&rft.issue=19&rft.issn=0021-9606&rft.eissn=1089-7690&rft.coden=JCPSA6&rft_id=info:doi/10.1063/5.0146753&rft_dat=%3Cproquest_scita%3E2813728606%3C/proquest_scita%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2813728606&rft_id=info:pmid/37184017&rfr_iscdi=true