Scholarlink

<p>Composite materials are promising magnetic materials for achieving the challenges of magnetic cooling, offering a broader range of properties compared to single-phase materials. In this study we investigate the magnetocaloric properties of composite materials with the general formula (La<sub>0.67</sub>Ca<sub>(0.33-x)</sub>SrₓMn<sub>0.98</sub>Ni<sub>0.02</sub>O<sub>3</sub>)<sub>1-y</sub>/(La<sub>0.67</sub>Ca<sub>0.33</sub>MnO<sub>3</sub>)<sub>y</sub> (with y values of 1/3 and 1/2, where the Sr-doped manganite phase is a 50/50 mixture of x = 0.1 and x = 0.2), synthesised using the sol–gel Pechini method. These composites exhibit broad magnetic entropy changes (ΔS) due to the synergistic interaction between their constituent phases, which is an essential characteristic for efficient magnetic refrigeration. To quantitatively assess and optimise their magnetocaloric performance, artificial neural network (ANN) modelling was employed. This offers a novel computational approach for predicting and tuning ΔS in such complex material systems. The model performed exceptionally well, achieving a coefficient of determination (R<sup>2</sup>) ranging from 0.990 to 0.996 around the magnetic transition temperature for the composition with <i>y</i> = 1/3, and from 0.990 to 0.995 for the composition with <i>y</i> = 1/2. This demonstrates strong agreement with experimental data across the 1–5 T magnetic field range. Furthermore, the model produced remarkably low values for the mean squared error (MSE), mean absolute error (MAE), and root mean square error (RMSE). This underscores its precision in identifying optimal parameters with minimal deviation. The predicted ΔS values were found to align closely with experimental measurements, highlighting the potential of AI-driven modelling to accelerate the development of advanced materials for solid-state refrigeration. This study therefore provides a standardised framework for applying machine learning techniques to the design and characterisation of magnetocaloric materials, paving the way for more efficient and environmentally sustainable cooling technologies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning based technique for characterization of magnetocaloric entropy performance: a study on formula (La0.67Ca(0.33-x)SrₓMn0.98Ni0.02O3)1-y/(La0.67Ca0.33MnO3)y manganite composites

  • 期刊论文

  • 发表日期：

    2026年3月23日

查看全文
  • Kawther Laajimi,
  • Radhouane Laajimi,
  • Rafik M’nassri

摘要

Composite materials are promising magnetic materials for achieving the challenges of magnetic cooling, offering a broader range of properties compared to single-phase materials. In this study we investigate the magnetocaloric properties of composite materials with the general formula (La0.67Ca(0.33-x)SrₓMn0.98Ni0.02O3)1-y/(La0.67Ca0.33MnO3)y (with y values of 1/3 and 1/2, where the Sr-doped manganite phase is a 50/50 mixture of x = 0.1 and x = 0.2), synthesised using the sol–gel Pechini method. These composites exhibit broad magnetic entropy changes (ΔS) due to the synergistic interaction between their constituent phases, which is an essential characteristic for efficient magnetic refrigeration. To quantitatively assess and optimise their magnetocaloric performance, artificial neural network (ANN) modelling was employed. This offers a novel computational approach for predicting and tuning ΔS in such complex material systems. The model performed exceptionally well, achieving a coefficient of determination (R2) ranging from 0.990 to 0.996 around the magnetic transition temperature for the composition with y = 1/3, and from 0.990 to 0.995 for the composition with y = 1/2. This demonstrates strong agreement with experimental data across the 1–5 T magnetic field range. Furthermore, the model produced remarkably low values for the mean squared error (MSE), mean absolute error (MAE), and root mean square error (RMSE). This underscores its precision in identifying optimal parameters with minimal deviation. The predicted ΔS values were found to align closely with experimental measurements, highlighting the potential of AI-driven modelling to accelerate the development of advanced materials for solid-state refrigeration. This study therefore provides a standardised framework for applying machine learning techniques to the design and characterisation of magnetocaloric materials, paving the way for more efficient and environmentally sustainable cooling technologies.