CoChat

Data‐Driven Materials Science: Status, Challenges, and Perspectives

Open access

The DOI matches its Crossref record; 201 of the 201 references it cites were found in a registry or an index.

Abstract

Data‐driven science is heralded as a new paradigm in materials science. In this field, data is the new resource, and knowledge is extracted from materials datasets that are too big or complex for traditional human reasoning—typically with the intent to discover new or improved materials or materials phenomena. Multiple factors, including the open science movement, national funding, and progress in information technology, have fueled its development. Such related tools as materials databases, machine learning, and high‐throughput methods are now established as parts of the materials research toolset. However, there are a variety of challenges that impede progress in data‐driven materials science: data veracity, integration of experimental and computational data, data longevity, standardization, and the gap between industrial interests and academic efforts. In this perspective article, the historical development and current state of data‐driven materials science, building from the early evolution of open science to the rapid expansion of materials data infrastructures are discussed. Key successes and challenges so far are also reviewed, providing a perspective on the future development of the field.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1002/advs.201900808

Cites

201 of 201 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 Materials discovery schematic. In the traditional approach, new materials are discovered by experimentation, theory, or computation (also referred to as 1st, 2nd, and 3rd paradigms and symbolized by the three icons at the top of the left panel). In the 4th paradigm of data‐driven materials science, available data is gathered in data infrastructures, and machine learning approaches discover new materials.
Figure 1 Materials discovery schematic. In the traditional approach, new materials are discovered by experimentation, theory, or computation (also referred to as 1st, 2nd, and 3rd paradigms and symbolized by the three icons at the top of the left panel). In the 4th paradigm of data‐driven materials science, available data is gathered in data infrastructures, and machine learning approaches discover new materials.CC BY · PubMed Central
Figure 2 Timeline and geographic distribution of materials data infrastructures and companies. The colour of the dots represents the time of establishment. The map shows that historically more centers have emerged in the U.S. and Europe, with Asia catching up over time. In addition, the U.S. has a higher renewal rate than Europe, as can be seen in the larger number of ligher colored dots. CSD: Cambridge Structural Database, ICSD: Inorganic Crystal Structure Database, ESP: Electronic Structure Project, AFLOW: Automatic‐Flow for Materials Discovery, AIST: National Institute of Advanced Industrial Science and Technology Databases, COD: Crystallography Open Database, MatDL: Materials Digital Library, CMR: Computational Materials Repository, NREL CID: NREL Center for Inverse Design, CEPDB: The Clean Energy Project Database, MGI: Materials Genome Initiative, CMD: Computational Materials Network, OQMD: Open Quantum Materials Database, NOMAD: Novel Materials Discovery Laboratory, MaX: Materials Design at the Exascale, MICCOM: Midwest Integrated Center for Computational Materials, MPDS: Materials Platform for Data Science, CMI2: Center for Materials Research by Information Integration, HTEM: High Throughput Experimental Materials Database, JARVIS: Joint Automated Repository for Various Integrated Simulations, OMDB: Organic Materials Database, QCArchive: The Quantum Chemistry Archive.
Figure 2 Timeline and geographic distribution of materials data infrastructures and companies. The colour of the dots represents the time of establishment. The map shows that historically more centers have emerged in the U.S. and Europe, with Asia catching up over time. In addition, the U.S. has a higher renewal rate than Europe, as can be seen in the larger number of ligher colored dots. CSD: Cambridge Structural Database, ICSD: Inorganic Crystal Structure Database, ESP: Electronic Structure Project, AFLOW: Automatic‐Flow for Materials Discovery, AIST: National Institute of Advanced Industrial Science and Technology Databases, COD: Crystallography Open Database, MatDL: Materials Digital Library, CMR: Computational Materials Repository, NREL CID: NREL Center for Inverse Design, CEPDB: The Clean Energy Project Database, MGI: Materials Genome Initiative, CMD: Computational Materials Network, OQMD: Open Quantum Materials Database, NOMAD: Novel Materials Discovery Laboratory, MaX: Materials Design at the Exascale, MICCOM: Midwest Integrated Center for Computational Materials, MPDS: Materials Platform for Data Science, CMI2: Center for Materials Research by Information Integration, HTEM: High Throughput Experimental Materials Database, JARVIS: Joint Automated Repository for Various Integrated Simulations, OMDB: Organic Materials Database, QCArchive: The Quantum Chemistry Archive.CC BY · PubMed Central
Figure 3 Number of materials informatics projects and infrastructures as function of time (see Figure 2 and Table 1 for details on individual projects and infrastructures). We divide the time axis into three periods that reflect the evolution of the data infrastructures (see text for details).
Figure 3 Number of materials informatics projects and infrastructures as function of time (see Figure 2 and Table 1 for details on individual projects and infrastructures). We divide the time axis into three periods that reflect the evolution of the data infrastructures (see text for details).CC BY · PubMed Central
Figure 4 Challenges faced by materials data infrastructures (on the left) on the way to increase the adoption by stakeholders from academia, industry, governments and the public (depicted on the right).
Figure 4 Challenges faced by materials data infrastructures (on the left) on the way to increase the adoption by stakeholders from academia, industry, governments and the public (depicted on the right).CC BY · PubMed Central
Figure 5 Example of an ontological hierarchy for the structural characterization of materials: a materials tree of life. Adapted under the terms of the Creative Commons Attribution 4.0 International License.145 Copyright 2018, the Authors, Published by Springer Nature.
Figure 5 Example of an ontological hierarchy for the structural characterization of materials: a materials tree of life. Adapted under the terms of the Creative Commons Attribution 4.0 International License.145 Copyright 2018, the Authors, Published by Springer Nature.CC BY · PubMed Central
Figure 6 A computational workflow used in creating a dataset of elastic tensors with the FireWorks workflow manager. The indigo boxes correspond to inputs or results, lighter blue boxes correspond to actions, and green diamonds correspond to decisions. Adapted with permission.188 Copyright 2019, Wiley.
Figure 6 A computational workflow used in creating a dataset of elastic tensors with the FireWorks workflow manager. The indigo boxes correspond to inputs or results, lighter blue boxes correspond to actions, and green diamonds correspond to decisions. Adapted with permission.188 Copyright 2019, Wiley.CC BY · PubMed Central
Figure 7 Key steps in building a machine learning model. The white arrows indicate the flow of data, green arrows indicate actions that can be identified and performed after analysis to improve the performance of the model.
Figure 7 Key steps in building a machine learning model. The white arrows indicate the flow of data, green arrows indicate actions that can be identified and performed after analysis to improve the performance of the model.CC BY · PubMed Central
Figure 8 The machine learning domain in terms of data volume and the complexity of the physical process, with selected examples placed in this domain. The complexity of a physical process here means the complex, nonlinear structures present in the data. Two opposing learning scenarios, a hard and an easy one, are illustrated in the lower panel. In these two cases, the underlying physical process is represented by a colored contour map, and the sampling of this process is represented by black crosses.223, 224, 225, 226, 227, 228
Figure 8 The machine learning domain in terms of data volume and the complexity of the physical process, with selected examples placed in this domain. The complexity of a physical process here means the complex, nonlinear structures present in the data. Two opposing learning scenarios, a hard and an easy one, are illustrated in the lower panel. In these two cases, the underlying physical process is represented by a colored contour map, and the sampling of this process is represented by black crosses.223, 224, 225, 226, 227, 228CC BY · PubMed Central
Figure 9 a) Schematic of an ecosystem in data‐driven materials science with materials data platforms at the center. In this ecosystem, different stakeholders from universities, the public, industry, and government facilitate the development of a technology. b–d) Possible relationships between data platforms and industry, discussed in the text.
Figure 9 a) Schematic of an ecosystem in data‐driven materials science with materials data platforms at the center. In this ecosystem, different stakeholders from universities, the public, industry, and government facilitate the development of a technology. b–d) Possible relationships between data platforms and industry, discussed in the text.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 201 references

Research works printed but not linked: 57

  1. Eagar T. W., 1995, Technol. Rev., 98, 42
  2. Boren M., 2012, McKinsey on Chemicals
  3. Hey T., 2009, The Fourth Paradigm: Data‐Intensive Scientific Discovery
  4. Brynjolfsson E., 2014, Second Machine Age. Work, Progress, and Prosperity in a Time of Brilliant Technologies
  5. Mueller T., 2016, Machine Learning in Materials Science, 186
  6. 2017, Building a Materials Data Infrastructure: Opening New Pathways to Discovery and Innovation in Science and Engineering
  7. Walsh K., 2016, Open Innovation, Open Science, Open to the World—A Vision for Europe
  8. Regazzi J. J., 2015, Scholarly Communications: A History from Content as King to Content as Kingmaker, 10.5040/9798881815738 · doi:10.5040/9798881815738
  9. Esanu J. M., 2003, The Role of Scientific and Technical Data and Information in The Public Domain: Proc. of A Symp
  10. PLOS https://www.plos.org(accessed: August 2019).
  11. Scientific Data https://www.nature.com/sdata/(accessed: August 2019).
  12. Sharing Research Data for Journal Authors https://www.elsevier.com/authors/author‐services/research‐data(accessed: August 2019).
  13. 10.1215/9780822389002 · DOI · Paper page
  14. Pedregosa F., 2011, J. Mach. Learn. Res., 12, 2825
  15. Noordik J., 2004, Cheminformatics Developments: History, Reviews and Current Research
  16. Weisgerber D. W., 1997, J. Assoc. Inf. Sci. Technol., 48, 349
  17. Wierling C., 2008, Nucleic Acids Res., 37, D623
  18. 10.1063/1.4812323 · DOI · Paper page
  19. Blokhin E., 2018, Handbook of Materials Modeling: Methods: Theory and Modeling, 1
  20. SpringerMaterials https://materials.springer.com(accessed: August 2019).
  21. Springer Nature Recommended Repositories https://www.springernature.com/de/authors/research‐data‐policy/repositories/12327124(accessed: August 2019).
  22. Kluyver T., 2016, Positioning and Power in Academic Publishing: Players, Agents and Agendas, 87
  23. 10.1016/j.commatsci.2012.10.028 · DOI · Paper page
  24. Smith B., 2003, Blackwell Guide to the Philosophy of Computing and Information, 155
  25. Vet P. E., 1994, Proc. of 11th European Conference on Artificial Intelligence (ECAI'94), 187
  26. European Materials Modelling Council Report on Workshop on Interoperability in Materials Modelling 2017.
  27. Cheung K., 2008, AAAI Spring Symposium: Semantic Scientific Knowledge Integration, 9
  28. Copestake A., 2006, Proc. of the UK e‐Science Programme All Hands Meeting 2006 (AHM2006)
  29. 10.1016/j.commatsci.2011.02.023 · DOI · Paper page
  30. Mitchell T., 1997, Machine Learning
  31. Russell S., 2010, Artificial Intelligence: A Modern Approach
  32. 10.1103/PhysRevLett.120.145301 · DOI · Paper page
  33. Jolliffe I. T., 2002, Principal Component Analysis
  34. Maaten L., 2008, J. Mach. Learn. Res., 9, 2579
  35. Samanta A., 2018, Nat. Commun., 9
  36. Liu R., 2015, Sci. Rep., 5, 1
  37. 10.1038/ncomms15679 · DOI · Paper page
  38. Montavon G., 2012, Advances in Neural Information Processing Systems, 440
  39. H.Huo M.Rupp e‐prints arXiv:1704.06439 2017. · arXiv
  40. Pronobis W., 2018, J. Chem. Theory Comput.
  41. L.Himanen M. O. J.Jäger E. V.Morooka F.Federici Canova Y. S.Ranawat D. Z.Gao P.Rinke A. S.Foster e‐prints arXiv:1904.08875 2019. · arXiv
  42. Goodfellow I., 2016, Deep Learning
  43. Rasmussen C., 2006, Gaussian Processes for Machine Learning
  44. Chen T., 2016, Proc. of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785
  45. Y.Jia E.Shelhamer J.Donahue S.Karayev J.Long R.Girshick S.Guadarrama T.Darrell inProc. of the 22Nd ACM Int. Conf. on Multimedia(Eds:K. A.Hua Y.Rui R.Steinmetz A.Hanjalic A.Natsev W.Zhu) MM '14. ACM New York NY USA2014.
  46. Theano Development Team e‐prints abs/1605.02688 2016. · arXiv
  47. Gilmer J., 2017, Proc. of the 34th Int. Conf. on Machine Learning, 1263
  48. LeCun Y., 1998, Convolutional Networks for Images, Speech, and Time Series, 255
  49. R.Chard Z.Li K.Chard L.Ward Y.Babuji A.Woodard S.Tuecke B.Blaiszik M. J.Franklin I.Foster e‐prints arXiv:1811.11213 2018. · arXiv
  50. Xue D., 2016, Proc. Natl. Acad. Sci. USA, 113, 47
  51. Meredig B., 2014, Phys. Rev. B, 094104, 1
  52. M. J.Kusner B.Paige J. M.Hernández‐Lobato inProc. of the 34th Int. Conf. on Machine Learning(Eds:D.Precup Y. W.Teh) PMLR Sydney Australia 2017 pp.1945–1954.
  53. P.Schwaller T.Laino T.Gaudin P.Bolgar C.Bekas A. A.Lee e‐prints arXiv:1811.02633 2018. · arXiv
  54. 10.1021/acs.jpcc.6b10908 · DOI · Paper page
  55. Bartók A. P., 2018, Phys. Rev. X, 8, 041048
  56. J. S.Smith B. T.Nebgen R.Zubatyuk N.Lubbers C.Devereux K.Barros S.Tretiak O.Isayev A.Roitberg ChemRxiv Preprint 2018.
  57. Gawer A., 2002, Platform Leadership How Intel, Microsoft, and Cisco Drive Industry Innovation

Cited by

Items 401–500 of 821

Metadata

Authors
Himanen, L., Geurts, A., Foster, A. S., & Rinke, P.
Affiliations
  • Aalto University (FI)
  • Netherlands Organisation for Applied Scientific Research (NL)
  • Kanazawa University (JP)
  • Technical University of Munich (DE)
Corresponding author
Rinke, Patrick
Venue
Advanced Science
Publisher
Wiley
Published
2019-09
Volume
6
Issue
21
Pages
1900808-1900808
ISSN
2198-3844 (ISSN-L)
OpenAlex
W2972219719
Cited by
808
References
193
Citations by year
2026: 109 · 2025: 135 · 2024: 148 · 2023: 138 · 2022: 122 · 2021: 105 · 2020: 46 · 2019: 5
FWCI
3.53
Type
journal article
Language
en
Open access
Locations
  • Advanced Science (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • KNAW Research Portal (The Royal Netherlands Academy of Arts and Sciences) (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • Data Archiving and Networked Services (DANS) (submittedVersion)
  • Aaltodoc (Aalto University) (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • mediaTUM – the media and publications repository of the Technical University Munich (Technical University Munich) (submittedVersion)
  • KNAW research portal (Royal Academy of Art and Sciences (KNAW)) (submittedVersion)
  • arXiv (Cornell University)
Funders
  • Jenny ja Antti Wihurin Rahasto
  • Academy of Finland (284621)
  • Ministry of Education, Culture, Sports, Science and Technology
Topics
  • Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
  • Physical Sciences › Computer Science › Information Systems › Research Data Management Practices
  • Social Sciences › Decision Sciences › Information Systems and Management › Scientific Computing and Data Management
Keywords
data-driven materials science, open science, materials database, machine learning for materials, high-throughput materials synthesis, data veracity, data standardization, data longevity, materials discovery

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.