Accelerating the discovery of novel magnetic materials using machine learning–guided adaptive feedback

Abstract

Magnetic materials are essential for energy generation and information devices, and they play an important role in advanced technologies and green energy economies. Currently, the most widely used magnets contain rare earth (RE) elements. An outstanding challenge of notable scientific interest is the discovery and synthesis of novel magnetic materials without RE elements that meet the performance and cost goals for advanced electromagnetic devices. Here, we report our discovery and synthesis of an RE-free magnetic compound, Fe 3 CoB 2 , through an efficient feedback framework by integrating machine learning (ML), an adaptive genetic algorithm, first-principles calculations, and experimental synthesis. Magnetic measurements show that Fe 3 CoB 2 exhibits a high magnetic anisotropy ( K 1 = 1.2 MJ/m 3 ) and saturation magnetic polarization ( J s = 1.39 T), which is suitable for RE-free permanent-magnet applications. Our ML-guided approach presents a promising paradigm for efficient materials design and discovery and can also be applied to the search for other functional materials.

Accelerating materials discovery using integrated deep machine learning approaches

“…One significant advance of the integrated deep machine learning approach presented in this paper over that used in ref. 25 and 36 is that interatomic potential trained by artificial neural network (ANN) has been incorporated into our ML framework. The accuracy of the ANN-ML interatomic potentials for complex materials was also demonstrated.…”

“…Because of the incorporation of the ANN-ML interatomic potential, only a few tens or a few hundred structures (instead of several thousand structures with the approach in ref. 25 and 36) need to be final checked by first-principles calculations. Thus, the pace of the novel compound discovery can be sped up 100–1000 times without losing the accuracy of first-principles predictions.…”

Data-driven deep generative design of stable spintronic materials

“…Xia et al predicted the Fe 3 CoB 2 magnetic compound employing the machine learning (ML)-guided adaptive feedback method with DFT, and also they synthesized it using a conventional arc-melting process. 21 ML models were also developed by Long et al 22 for intermetallic compounds to classify the AFM and FM materials and predict their Curie temperatures. Lu et al developed an adaptive ML framework to search the chemical space with over 2 × 10 5 candidates to realize new 2-dimensional magnetic compositions.…”

Using a computationally driven screening to enhance magnetocaloric effect of metal monoborides

“…Many relevant magnetic properties can also be calculated with DFT, such as the atomic magnetic moments, the saturation magnetization, exchange couplings or the spin-resolved density of states. Currently, it is common practice to use DFT-based methods, sometimes in conjunction with machine learning algorithms, in the search for new magnetic materials using large data sets of materials [22][23][24]. Moreover, several theoretical approaches have been applied as high-throughput computational screening schemes in more focused contexts, such as the discovery of new two-dimensional magnetic materials [25] the search of better permanent magnets [26][27][28], high-spin metal organic frameworks [29] or bipolar magnetic semiconductors [30].…”