MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials

MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials Lei Wang, Yan Chen, Xueru Wang, Xiaobin Deng, Yilun Liu, Xi Chen, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7463697/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Inverse design of solid-state materials with desired properties remains a central challenge in materials science, requiring exploration of vast chemical spaces containing potentially 10 100 possible structures. Current generative approaches face limitations in computational efficiency, multi-property targeting precision and mechanistic interpretability. Here, we introduce MatterGPT, an autoregressive Transformer-decoder architecture that leverages SLICES (Simplified Line-Input Crystal-Encoding System) representation to generate novel crystals through conditional next-token prediction. Trained on 306,533 crystal structures, MatterGPT achieves > 99% structural validity, > 99% structural uniqueness and > 50% novelty rates while targeting both specific lattice-insensitive and lattice-sensitive properties. Critically, MatterGPT enables direct multi-property generation without post-generation filtering. Interpretability analysis reveals clear property-guided generation mechanisms and systematic chemical space exploration. The comprehensive open-source release, including MatterGPT Hub integration platform, establishes sequence-based autoregressive generation as a computationally efficient and interpretable paradigm for inverse crystal design, accelerating materials discovery across energy storage, electronics, and functional applications. Physical sciences/Materials science/Theory and computation/Computational methods Physical sciences/Chemistry/Theoretical chemistry/Structure prediction Autoregressive Generation SLICES Solid-State Materials On-demand Generation Inverse Design Full Text Additional Declarations There is NO Competing Interest. Supplementary Files MatterGPTSINC.docx Supporting Information for MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials MLChecklist1.11.pdf Machine Learning Checklist Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7463697","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":510253904,"identity":"3fd7039c-63cd-4361-a543-b43e0b162778","order_by":0,"name":"Lei Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2ElEQVRIiWNgGAWjYBACPgbmw4//VNgwMLAzMDAzGIDEEvBrYWNgSzPgOZMGUk+0Fh4DCd62w1AtDMRokUgwMJBsOy/Pz8xjurmgwI6Bnz3HgJCWhAcG524bzmzmMbs9wyCZQbLnDQEt0gkHDBLKbicYHAZq4TEAeucGIVukExskDrCdg2mpZ7AnrAXolIa2AzAthxkMJAhpkX/GZsxwJhnoF7YyoF+O80iceVaAVws/z/nPjxkq7OT52Zu33S74Uy3H3568Aa8WDMBDmvJRMApGwSgYBVgBAKwTPtj90oIoAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-1919-9107","institution":"Nanjing University","correspondingAuthor":true,"prefix":"","firstName":"Lei","middleName":"","lastName":"Wang","suffix":""},{"id":510253905,"identity":"9f23a947-19d6-42d0-bdfb-411af9b3b552","order_by":1,"name":"Yan Chen","email":"","orcid":"https://orcid.org/0000-0003-1723-0112","institution":"Xi'an Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Yan","middleName":"","lastName":"Chen","suffix":""},{"id":510253906,"identity":"d1b6b04d-9fc4-4019-abd0-1b6a7de5aa7f","order_by":2,"name":"Xueru Wang","email":"","orcid":"","institution":"Lingnan University","correspondingAuthor":false,"prefix":"","firstName":"Xueru","middleName":"","lastName":"Wang","suffix":""},{"id":510253907,"identity":"f6eaef2b-1f9b-493d-9edf-0543a2486b8d","order_by":3,"name":"Xiaobin Deng","email":"","orcid":"","institution":"Lingnan University","correspondingAuthor":false,"prefix":"","firstName":"Xiaobin","middleName":"","lastName":"Deng","suffix":""},{"id":510253908,"identity":"67a0112d-3b3c-4a08-a56b-6d6783308e87","order_by":4,"name":"Yilun Liu","email":"","orcid":"https://orcid.org/0000-0002-3070-096X","institution":"Xi'an Jiaotong University","correspondingAuthor":false,"prefix":"","firstName":"Yilun","middleName":"","lastName":"Liu","suffix":""},{"id":510253909,"identity":"694f0bfc-c5c1-4bd5-b26d-30497e0ec027","order_by":5,"name":"Xi Chen","email":"","orcid":"https://orcid.org/0000-0002-8911-4172","institution":"School of Interdisciplinary Studies, Lingnan University","correspondingAuthor":false,"prefix":"","firstName":"Xi","middleName":"","lastName":"Chen","suffix":""},{"id":510253910,"identity":"d2b6d702-653e-4afc-ab74-1e4928c70f51","order_by":6,"name":"Jun Zhang","email":"","orcid":"","institution":"Shenzhen Bay Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Jun","middleName":"","lastName":"Zhang","suffix":""},{"id":510253911,"identity":"1b3961fd-31d2-405c-9e44-7d120d0c17ae","order_by":7,"name":"Yunwei Zhang","email":"","orcid":"","institution":"School of Physics, Sun Yat-Sen University, Guangzhou, Guangdong 510275","correspondingAuthor":false,"prefix":"","firstName":"Yunwei","middleName":"","lastName":"Zhang","suffix":""},{"id":510253912,"identity":"d2d250c6-2d97-48f8-852b-47057a92a9a7","order_by":8,"name":"Hang Xiao","email":"","orcid":"","institution":"Lingnan University","correspondingAuthor":false,"prefix":"","firstName":"Hang","middleName":"","lastName":"Xiao","suffix":""}],"badges":[],"createdAt":"2025-08-26 14:00:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7463697/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7463697/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106401547,"identity":"e9910580-2c95-495b-bb4f-33427e9ce66c","added_by":"auto","created_at":"2026-04-08 09:06:42","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1469718,"visible":true,"origin":"","legend":"","description":"","filename":"MatterGPTNC.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7463697/v1_covered_b62d5ba5-3ad0-4631-be5f-2568dac48c51.pdf"},{"id":90645594,"identity":"bc814833-0a26-4ba1-8fb4-b29abe884891","added_by":"auto","created_at":"2025-09-05 07:42:10","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":8426157,"visible":true,"origin":"","legend":"Supporting Information for MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials","description":"","filename":"MatterGPTSINC.docx","url":"https://assets-eu.researchsquare.com/files/rs-7463697/v1/cd25f73527dbd3060256f3f1.docx"},{"id":90645592,"identity":"d44b1a0f-0690-47ff-aac1-0b28e7cf7ad2","added_by":"auto","created_at":"2025-09-05 07:42:10","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":365298,"visible":true,"origin":"","legend":"Machine Learning Checklist","description":"","filename":"MLChecklist1.11.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7463697/v1/b230b96f431636fa932a5c37.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Autoregressive Generation, SLICES, Solid-State Materials, On-demand Generation, Inverse Design","lastPublishedDoi":"10.21203/rs.3.rs-7463697/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7463697/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eInverse design of solid-state materials with desired properties remains a central challenge in materials science, requiring exploration of vast chemical spaces containing potentially 10\u003csup\u003e100\u003c/sup\u003e possible structures. Current generative approaches face limitations in computational efficiency, multi-property targeting precision and mechanistic interpretability. Here, we introduce MatterGPT, an autoregressive Transformer-decoder architecture that leverages SLICES (Simplified Line-Input Crystal-Encoding System) representation to generate novel crystals through conditional next-token prediction. Trained on 306,533 crystal structures, MatterGPT achieves\u0026thinsp;\u0026gt;\u0026thinsp;99% structural validity, \u0026gt;\u0026thinsp;99% structural uniqueness and \u0026gt;\u0026thinsp;50% novelty rates while targeting both specific lattice-insensitive and lattice-sensitive properties. Critically, MatterGPT enables direct multi-property generation without post-generation filtering. Interpretability analysis reveals clear property-guided generation mechanisms and systematic chemical space exploration. The comprehensive open-source release, including MatterGPT Hub integration platform, establishes sequence-based autoregressive generation as a computationally efficient and interpretable paradigm for inverse crystal design, accelerating materials discovery across energy storage, electronics, and functional applications.\u003c/p\u003e","manuscriptTitle":"MatterGPT: A Generative Transformer for Multi-Property Inverse Design of Solid-State Materials","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-05 07:42:05","doi":"10.21203/rs.3.rs-7463697/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0a06ea81-ae1b-4279-930e-9f2ecc9fd636","owner":[],"postedDate":"September 5th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":54198596,"name":"Physical sciences/Materials science/Theory and computation/Computational methods"},{"id":54198597,"name":"Physical sciences/Chemistry/Theoretical chemistry/Structure prediction"}],"tags":[],"updatedAt":"2026-04-01T12:20:59+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-05 07:42:05","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7463697","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7463697","identity":"rs-7463697","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}