A critical examination of compound stability predictions from machine-learned formation energies

被引:0
|
作者
Bartel, Christopher J. [1 ]
Trewartha, Amalie [1 ]
Wang, Qi [2 ]
Dunn, Alexander [1 ,2 ]
Jain, Anubhav [2 ]
Ceder, Gerbrand [1 ,3 ]
机构
[1] Univ Calif Berkeley, Dept Mat Sci & Engn, Berkeley, CA 94720 USA
[2] Lawrence Berkeley Natl Lab, Energy Technol Area, Berkeley, CA 94720 USA
[3] Lawrence Berkeley Natl Lab, Div Mat Sci, Berkeley, CA 94720 USA
关键词
51;
D O I
10.1038/s41524-020-00362-y
中图分类号
O64 [物理化学(理论化学)、化学物理学];
学科分类号
070304 ; 081704 ;
摘要
Machine learning has emerged as a novel tool for the efficient prediction of material properties, and claims have been made that machine-learned models for the formation energy of compounds can approach the accuracy of Density Functional Theory (DFT). The models tested in this work include five recently published compositional models, a baseline model using stoichiometry alone, and a structural model. By testing seven machine learning models for formation energy on stability predictions using the Materials Project database of DFT calculations for 85,014 unique chemical compositions, we show that while formation energies can indeed be predicted well, all compositional models perform poorly on predicting the stability of compounds, making them considerably less useful than DFT for the discovery and design of new solids. Most critically, in sparse chemical spaces where few stoichiometries have stable compounds, only the structural model is capable of efficiently detecting which materials are stable. The nonincremental improvement of structural models compared with compositional models is noteworthy and encourages the use of structural models for materials discovery, with the constraint that for any new composition, the ground-state structure is not known a priori. This work demonstrates that accurate predictions of formation energy do not imply accurate predictions of stability, emphasizing the importance of assessing model performance on stability predictions, for which we provide a set of publicly available tests.
引用
收藏
页数:11
相关论文
共 50 条
  • [11] Stability and Equilibrium Structures of Unknown Ternary Metal Oxides Explored by Machine-Learned Potentials
    Hwang, Seungwoo
    Jung, Jisu
    Hong, Changho
    Jeong, Wonseok
    Kang, Sungwoo
    Han, Seungwu
    [J]. JOURNAL OF THE AMERICAN CHEMICAL SOCIETY, 2023, 145 (35) : 19378 - 19386
  • [12] Phase Transitions in Inorganic Halide Perovskites from Machine-Learned Potentials
    Fransson, Erik
    Wiktor, Julia
    Erhart, Paul
    [J]. JOURNAL OF PHYSICAL CHEMISTRY C, 2023, 127 (28): : 13773 - 13781
  • [13] Machine-learned 3D Building Vectorization from Satellite Imagery
    Wang, Yi
    Zorzi, Stefano
    Bittner, Ksenia
    [J]. 2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION WORKSHOPS, CVPRW 2021, 2021, : 1072 - 1081
  • [14] Machine-learned tuning of artificial Kitaev chains from tunneling spectroscopy measurements
    Benestad, Jacob
    Tsintzis, Athanasios
    Souto, Ruben Seoane
    Leijnse, Martin
    van Nieuwenburg, Evert
    Danon, Jeroen
    [J]. PHYSICAL REVIEW B, 2024, 110 (07)
  • [15] Towards Automatically Generating Security Analyses from Machine-Learned Library Models
    Kober, Maria
    Arzt, Steven
    [J]. COMPUTER SECURITY - ESORICS 2021, PT II, 2021, 12973 : 752 - 758
  • [16] Machine-Learned Decision Trees for Predicting Gold Nanorod Sizes from Spectra
    Shiratori, Katsuya
    Bishop, Logan D. C.
    Ostovar, Behnaz
    Baiyasi, Rashad
    Cai, Yi-Yu
    Rossky, Peter J.
    Landes, Christy F.
    Link, Stephan
    [J]. JOURNAL OF PHYSICAL CHEMISTRY C, 2021, 125 (35): : 19353 - 19361
  • [17] Insights from symmetry: Improving machine-learned models for grain boundary segregation
    Borges, Y.
    Huber, L.
    Zapolsky, H.
    Patte, R.
    Demange, G.
    [J]. COMPUTATIONAL MATERIALS SCIENCE, 2024, 232
  • [18] Machine-Learned Electronically Excited States with the MolOrbImage Generated from the Molecular Ground State
    Chen, Ziyong
    Yam, Vivian Wing-Wah
    [J]. JOURNAL OF PHYSICAL CHEMISTRY LETTERS, 2023, 14 (07): : 1955 - 1961
  • [19] Machine-Learned Electronically Excited States with the MolOrbImage Generated from the Molecular Ground State
    Chen, Ziyong
    Yam, Vivian Wing-Wah
    [J]. JOURNAL OF PHYSICAL CHEMISTRY LETTERS, 2023, : 1955 - 1961
  • [20] Machine-Learned Potentials by Active Learning from Organic Crystal Structure Prediction Landscapes
    Butler, Patrick W. V.
    Hafizi, Roohollah
    Day, Graeme M.
    [J]. JOURNAL OF PHYSICAL CHEMISTRY A, 2024, 128 (05): : 945 - 957