skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Structure classification and melting temperature prediction in octet AB solids via machine learning

Abstract

Machine learning methods are being increasingly used in condensed matter physics and materials science to classify crystals structures and predict material properties. However, the reliability of these methods for a given problem, especially when large data sets are unavailable, has not been well studied. By addressing the tasks of classifying crystal structure and predicting melting temperatures of the octet subset of AB solids, we performed such a study and found potential problems with using machine learning methods on relatively small data sets. At the same time, however, we can reaffirm the potential power of such methods for these tasks. In particular, we uncovered an important new material feature, the excess Born effective charge, that significantly increased the accuracy of the predictions for the classification problem we defined. This discovery leads us to propose a new scale for the degree of ionicity and covalency in these solids. More specifically, we partitioned the crystal structures of a set of 75 octet solids into those that are ionic and covalent bonded and thus performed a binary classification task. We found that using the standard indices (rσ,rπ), suggested by St. John and Bloch several decades ago, enabled an average success in classification of 92%.more » Using just rσ and the excess Born effective charge ΔZA of the A atom enabled an average success of 97%, but we also found relatively large variations about these averages that were dependent on how certain machine learning methods were used and for which a standard deviation was not a proper measure of the degree of confidence we can place in either average. Instead, we calculated and report with 95 % confidence that the traditional classification pair predicts an accuracy in the interval [ 89%, 95%] and the accuracy of the new pair lies in the interval [96 %, 99%]. For melting temperature predictions, the size of our data set was 46. We estimate the root-mean-squared error of our resulting model to be 11% of the mean melting temperature of the data, but we note that if the accuracy of this predicted error is itself measured, our estimated fitting error itself has a root-mean-square error of 50%. In short, what we illustrate is that classification and regression predictions can vary significantly, depending on the details of how machine learning methods are applied to small data sets. This variation makes it important, if not essential, to average the predictions and compute confidence intervals about these averages to report results meaningfully. However, when properly used, these statistical methods can advance our understanding and improve predictions of material properties even for small data sets.« less

Authors:
ORCiD logo [1]; ORCiD logo [1]; ORCiD logo [1]
  1. Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Publication Date:
Research Org.:
Los Alamos National Lab. (LANL), Los Alamos, NM (United States)
Sponsoring Org.:
USDOE Laboratory Directed Research and Development (LDRD) Program
OSTI Identifier:
1469521
Alternate Identifier(s):
OSTI ID: 1184701
Report Number(s):
LA-UR-14-28547
Journal ID: ISSN 1098-0121; PRBMDO
Grant/Contract Number:  
AC52-06NA25396
Resource Type:
Accepted Manuscript
Journal Name:
Physical Review. B, Condensed Matter and Materials Physics
Additional Journal Information:
Journal Volume: 91; Journal Issue: 21; Journal ID: ISSN 1098-0121
Publisher:
American Physical Society (APS)
Country of Publication:
United States
Language:
English
Subject:
36 MATERIALS SCIENCE; 97 MATHEMATICS AND COMPUTING; Information Science; Material Science

Citation Formats

Pilania, Ghanshyam, Gubernatis, James E., and Lookman, Turab. Structure classification and melting temperature prediction in octet AB solids via machine learning. United States: N. p., 2015. Web. https://doi.org/10.1103/PhysRevB.91.214302.
Pilania, Ghanshyam, Gubernatis, James E., & Lookman, Turab. Structure classification and melting temperature prediction in octet AB solids via machine learning. United States. https://doi.org/10.1103/PhysRevB.91.214302
Pilania, Ghanshyam, Gubernatis, James E., and Lookman, Turab. Mon . "Structure classification and melting temperature prediction in octet AB solids via machine learning". United States. https://doi.org/10.1103/PhysRevB.91.214302. https://www.osti.gov/servlets/purl/1469521.
@article{osti_1469521,
title = {Structure classification and melting temperature prediction in octet AB solids via machine learning},
author = {Pilania, Ghanshyam and Gubernatis, James E. and Lookman, Turab},
abstractNote = {Machine learning methods are being increasingly used in condensed matter physics and materials science to classify crystals structures and predict material properties. However, the reliability of these methods for a given problem, especially when large data sets are unavailable, has not been well studied. By addressing the tasks of classifying crystal structure and predicting melting temperatures of the octet subset of AB solids, we performed such a study and found potential problems with using machine learning methods on relatively small data sets. At the same time, however, we can reaffirm the potential power of such methods for these tasks. In particular, we uncovered an important new material feature, the excess Born effective charge, that significantly increased the accuracy of the predictions for the classification problem we defined. This discovery leads us to propose a new scale for the degree of ionicity and covalency in these solids. More specifically, we partitioned the crystal structures of a set of 75 octet solids into those that are ionic and covalent bonded and thus performed a binary classification task. We found that using the standard indices (rσ,rπ), suggested by St. John and Bloch several decades ago, enabled an average success in classification of 92%. Using just rσ and the excess Born effective charge ΔZA of the A atom enabled an average success of 97%, but we also found relatively large variations about these averages that were dependent on how certain machine learning methods were used and for which a standard deviation was not a proper measure of the degree of confidence we can place in either average. Instead, we calculated and report with 95 % confidence that the traditional classification pair predicts an accuracy in the interval [ 89%, 95%] and the accuracy of the new pair lies in the interval [96 %, 99%]. For melting temperature predictions, the size of our data set was 46. We estimate the root-mean-squared error of our resulting model to be 11% of the mean melting temperature of the data, but we note that if the accuracy of this predicted error is itself measured, our estimated fitting error itself has a root-mean-square error of 50%. In short, what we illustrate is that classification and regression predictions can vary significantly, depending on the details of how machine learning methods are applied to small data sets. This variation makes it important, if not essential, to average the predictions and compute confidence intervals about these averages to report results meaningfully. However, when properly used, these statistical methods can advance our understanding and improve predictions of material properties even for small data sets.},
doi = {10.1103/PhysRevB.91.214302},
journal = {Physical Review. B, Condensed Matter and Materials Physics},
number = 21,
volume = 91,
place = {United States},
year = {2015},
month = {6}
}

Journal Article:

Citation Metrics:
Cited by: 11 works
Citation information provided by
Web of Science

Save / Share:

Works referenced in this record:

Pauli-Force Model Potential for Solids
journal, March 1973


Electron localization: Band-by-band decomposition and application to oxides
journal, December 2002


Quantum-Defect Electronegativity Scale for Nontransition Elements
journal, October 1974


Dielectric Classification of Crystal Structures, Ionization Potentials, and Band Structures
journal, April 1969


Non-nominal value of the dynamical effective charge in alkaline-earth oxides
journal, June 1997


Debye temperature and melting point of II-VI and III-V semiconductors
journal, July 2010

  • Kumar, V.; Jha, V.; Shrivastava, A. K.
  • Crystal Research and Technology, Vol. 45, Issue 9
  • DOI: 10.1002/crat.201000268

Projector augmented-wave method
journal, December 1994


A chemical scale for crystal-structure maps
journal, July 1984


Phonons and related crystal properties from density-functional perturbation theory
journal, July 2001

  • Baroni, Stefano; de Gironcoli, Stefano; Dal Corso, Andrea
  • Reviews of Modern Physics, Vol. 73, Issue 2
  • DOI: 10.1103/RevModPhys.73.515

Bond-Orbital Model and the Properties of Tetrahedrally Coordinated Solids
journal, November 1973


Data mining for materials: Computational experiments with A B compounds
journal, March 2012


Thousands of samples are needed to generate a robust gene list for predicting outcome in cancer
journal, April 2006

  • Ein-Dor, L.; Zuk, O.; Domany, E.
  • Proceedings of the National Academy of Sciences, Vol. 103, Issue 15
  • DOI: 10.1073/pnas.0601231103

Special points for Brillouin-zone integrations
journal, June 1976

  • Monkhorst, Hendrik J.; Pack, James D.
  • Physical Review B, Vol. 13, Issue 12, p. 5188-5192
  • DOI: 10.1103/PhysRevB.13.5188

Small Sample Issues for Microarray-Based Classification
journal, January 2001

  • Dougherty, Edward R.
  • Comparative and Functional Genomics, Vol. 2, Issue 1
  • DOI: 10.1002/cfg.62

Semiconductor effective charges from tight-binding theory
journal, June 1996


On the crystal chemistry of normal valence compounds
journal, December 1959


Role of tight-binding parameters and scaling laws on effective charges in semiconductors
journal, February 2000


Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set
journal, October 1996


Performance of Error Estimators for Classification
journal, March 2010


The Illusion of Distribution-Free Small-Sample Classification in Genomics
journal, August 2011

  • R. Dougherty, Edward; Zollanvari, Amin; M. Braga-Neto, Ulisses
  • Current Genomics, Vol. 12, Issue 5
  • DOI: 10.2174/138920211796429763

Melting point trends in intermetallic alloys
journal, January 1987


Ionicity of the Chemical Bond in Crystals
journal, July 1970


Electric polarization as a bulk quantity and its relation to surface charge
journal, August 1993


Localized Effective Charges in Diatomic Crystals
journal, August 1971


Scaling Theory of Melting Temperatures of Covalent Crystals
journal, September 1972


Big Data of Materials Science: Critical Role of the Descriptor
journal, March 2015


The infrared effective charge in IV-VI compounds. I. A simple one-dimensional model
journal, November 1979


The melting points of intermetallic compounds
journal, March 1986


Ground State of the Electron Gas by a Stochastic Method
journal, August 1980


Electronic Structure
book, January 2004


Search for regularities in the melting points of AB compounds
journal, June 1993


Theory of polarization of crystalline solids
journal, January 1993


Polarization and dynamical charge of ZnO within different one-particle schemes
journal, December 1995


Structural Stability of 495 Binary Compounds
journal, March 1980


Dynamical atomic charges: The case of AB O 3 compounds
journal, September 1998


Macroscopic polarization in crystalline dielectrics: the geometric phase approach
journal, July 1994


Semiconductor effective charges and dielectric constants in the tight-binding approach
journal, November 1997


Piezoelectricity
journal, February 1972


Electronic structure of AlFeN films exhibiting crystallographic orientation change from c- to a-axis with Fe concentrations and annealing effect
journal, February 2020


    Works referencing / citing this record:

    Machine learning properties of binary wurtzite superlattices
    journal, January 2018


    Predicting defect behavior in B2 intermetallics by merging ab initio modeling and machine learning
    journal, December 2016


    Machine learning in materials informatics: recent applications and prospects
    journal, December 2017

    • Ramprasad, Rampi; Batra, Rohit; Pilania, Ghanshyam
    • npj Computational Materials, Vol. 3, Issue 1
    • DOI: 10.1038/s41524-017-0056-5

    Identifying Pb-free perovskites for solar cells by machine learning
    journal, March 2019


    Recent advances and applications of machine learning in solid-state materials science
    journal, August 2019

    • Schmidt, Jonathan; Marques, Mário R. G.; Botti, Silvana
    • npj Computational Materials, Vol. 5, Issue 1
    • DOI: 10.1038/s41524-019-0221-0

    Machine learning bandgaps of double perovskites
    journal, January 2016

    • Pilania, G.; Mannodi-Kanakkithodi, A.; Uberuaga, B. P.
    • Scientific Reports, Vol. 6, Issue 1
    • DOI: 10.1038/srep19375

    Accelerated search for BaTiO 3 -based piezoelectrics with vertical morphotropic phase boundary using Bayesian learning
    journal, November 2016

    • Xue, Dezhen; Balachandran, Prasanna V.; Yuan, Ruihao
    • Proceedings of the National Academy of Sciences, Vol. 113, Issue 47
    • DOI: 10.1073/pnas.1607412113

    Uncovering structure-property relationships of materials by subgroup discovery
    journal, January 2017

    • Goldsmith, Bryan R.; Boley, Mario; Vreeken, Jilles
    • New Journal of Physics, Vol. 19, Issue 1
    • DOI: 10.1088/1367-2630/aa57c2

    Discovering phase transitions with unsupervised learning
    journal, November 2016


    Predictions of new AB O 3 perovskite compounds by combining machine learning and density functional theory
    journal, April 2018


    Machine learning in materials design and discovery: Examples from the present and suggestions for the future
    journal, December 2018


    Classification of AB O 3 perovskite solids: a machine learning study
    journal, September 2015

    • Pilania, G.; Balachandran, P. V.; Gubernatis, J. E.
    • Acta Crystallographica Section B Structural Science, Crystal Engineering and Materials, Vol. 71, Issue 5
    • DOI: 10.1107/s2052520615013979

    Finding New Perovskite Halides via Machine Learning
    journal, April 2016

    • Pilania, Ghanshyam; Balachandran, Prasanna V.; Kim, Chiho
    • Frontiers in Materials, Vol. 3
    • DOI: 10.3389/fmats.2016.00019