Machine learning the magnetocaloric effect in manganites from lattice parameters

Efficient solid-state refrigeration techniques have drawn increasing attention due to their potential for improving energy efficiency of refrigeration temperature control systems without using harmful gas as in conventional gas compression techniques. Research on magnetocaloric lanthanum manganites with a large maximum magnetic entropy change near room temperature shows promising results for further developments of magnetic refrigeration devices. By incorporating chemical substitutions, oxygen content modifications, and various synthesis methods, these manganites experience lattice distortions from perovskite cubic structures to pseudocubic, orthorhombic, and rhombohedral structures. Further changes in lattice parameters can also be achieved by the introduction of strain due to lattice mismatches. Empirical results and previous models through thermodynamics and first principles show that changes in lattice parameters correlate with those in MMCE, but correlations are merely general tendencies and obviously not universal. In this work, the Gaussian process regression model is developed as a machine learning tool to find statistical correlations between the MMCE and lattice parameters among lanthanum manganites. More than 100 lattices, cubic, pseudocubic, orthorhombic, and rhombohedral, with the MMCE ranging from 0.65 to 8.00Jkg(-1)K(-1) under a field change of 5 T are explored for this purpose. The modeling approach demonstrates a high degree of accuracy and stability, contributing to efficient and low-cost estimations of the magnetocaloric effect. Furthermore, the machine learning algorithm predicts close MMCE results on epitaxial films with strained lattices against experimental results, which can provide guidance on thin film structure design and help understandings of magnetic phase transformations and magnetocaloric effects.