Multivariate data analysis approach to understand magnetic properties of perovskite manganese oxides

Here we apply statistical multivariate data analysis techniques to obtain some insights into the complex structure-property relations in antiferromagnetic (AFM) and ferromagnetic (FM) manganese perovskite systems, AMnO3. The 131 samples included in the present analyses are described by 21 crystal-structure or crystal-chemical (CS/CC) parameters. Principal component analysis (PCA), carried out separately for the AFM and FM compounds, is used to model and evaluate the various relationships among the magnetic properties and the various CS/CC parameters. Moreover, for the AFM compounds, PLS (partial least squares projections to latent structures) analysis is performed so as to predict the magnitude of the Néel temperature on the bases of the CS/CC parameters. Finally, so-called PLS-DA (PLS discriminant analysis) method is employed to find out the most influential/characteristic CS/CC parameters that differentiate the two classes of compounds from each other.

Graphical abstractStatistical multivariate data analysis techniques are applied to detect structure-property relations in antiferromagnetic (AFM) and ferromagnetic (FM) manganese perovskites. For AFM compounds, partial least squares projections to latent structures analysis predict the magnitude of the Néel temperature on the bases of structural parameters only. Moreover, AFM and FM compounds are well separated by means of so-called partial least squares discriminant analysis method.Figure optionsDownload full-size imageDownload as PowerPoint slide