Process Machine Learning of Iron-Based Superconducting Polycrystalline Bulks

122 phase iron-based superconductors show high upper critical field (<i>H</i>_c2 &gt; 50 T) with small electromagnetic anisotropy (<i>γ</i> ~ 1-2)¹⁾ and large critical grain boundary angle (<i>θ</i>_c ~9deg)²⁾, and therefore is a promising material for applications in polycrystalline form. Foreseeing high field magnet applications, Weiss et al. have reported demonstration of trapped field of 1 T for K-doped BaFe₂As₂ (Ba122) polycrystalline bulks synthesized by hot isostatic pressing³⁾. In this study, we synthesized K-doped Ba122 bulks by Spark Plasma Sintering (SPS). In a highly pure Ar glove box, mechanically alloyed precursor powder was prepared by high-energy ball-milling of elemental metals with the molar ratios of Ba:K:Fe:As = 0.6:0.4:2:2^{4), 5)}. The precursor powder was filled into a graphite mold and sintered using SPS apparatus. Two approaches to optimizing the process conditions were considered: optimization by researchers’ experience and intuition, based on the knowledge of nanostructural analysis and multi-scale observations^{6), 7)}, and optimization by process machine learning⁸⁾. Bayesian optimization was performed to find the best input parameters which maximize the target output property, critical current density (<i>J</i>_c), on the experimentally available range of processing conditions. High <i>J</i>_c value exceeding 10⁵ A/cm², which is among the highest in Ba122 bulk samples, was developed by both experiments guided by researchers and such an adaptive experimental optimization process. In order to demonstrate the rapped field properties, relatively large size, disk-shaped bulk samples with a diameter of 30 mm were synthesized using the optimized processing parameters. The obtained bulk samples were dense with relative density of ~90%. Detailed trapped field characteristics will be reported in the presentation.<br/><b>Acknowledgement</b><br/>This work was supported by JST CREST (JPMJCR18J4), JSPS KAKENHI (JP21H01615), and Nanotechnology Platform (A-18-TU-0037, A-21-KU-1001) of the MEXT, Japan.<br/><b>References</b>:<br/>1) H. Hosono et al., Materials Today 21, 278 (2018).<br/>2) T. Katase et al., Nat. Commun. 2, 409 (2011); K. Iida et al., Supercond. Sci. Technol. 33, 043001 (2020).<br/>3) J. D. Weiss et al., Supercond. Sci. Technol. 28, 112001 (2015).<br/>4) S. Tokuta and A. Yamamoto, APL Materials 7, 111107 (2019).<br/>5) S. Tokuta, Y. Shimada, and A. Yamamoto, Supercond. Sci. Technol. 33, 094010 (2020).<br/>6) Y. Shimada et al., Supercond. Sci. Technol. 32, 084003 (2019).<br/>7) F. Kametani et al., Appl. Phys. Express 13, 113002 (2020).<br/>8) S. Tokuta et al., ASC2020, Wk2MPo3B-05 (2020); A. Yamamoto et al., ASC2020, Wk1MOr3B-02 (2020).