From density functional theory to machine learning predictive models for electrical properties of spinel oxides
2024

Predicting Electrical Properties of Spinel Oxides Using Machine Learning

Sample size: 190 publication Evidence: moderate

Author Information

Author(s): Elbaz Yuval, Caspary Toroker Maytal

Primary Institution: Technion – Israel Institute of Technology

Hypothesis

Can machine learning models accurately predict the electronic conductivity and band gaps of spinel oxides based on their compositions?

Conclusion

The study successfully developed machine learning models that can predict the electronic conductivity and band gaps of spinel oxides, showing that higher nickel content leads to increased conductivity.

Supporting Evidence

  • Models were trained on a dataset of 190 different ternary spinel oxides.
  • High nickel content in spinel oxides was associated with increased conductivity.
  • Machine learning algorithms showed good predictive performance with mean absolute errors around 30-37 μA.

Takeaway

Scientists created a computer program that can guess how well certain materials conduct electricity just by looking at their ingredients.

Methodology

The study used density functional theory (DFT) calculations to create a dataset of spinel oxides, followed by machine learning algorithms to predict conductivity and band gaps based on compositions.

Potential Biases

Potential biases may arise from the choice of DFT parameters and the limited diversity of the dataset.

Limitations

The dataset was relatively small, and inconsistencies in target values may affect model accuracy.

Digital Object Identifier (DOI)

10.1038/s41598-024-62788-4

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication