Dielectric ceramic capacitors are essential core components for electronics, smart grids and new energy vehicles, prized for their high power density. As electronic devices move toward miniaturization and intelligence, the demand for lead-free dielectric ceramics with ultrahigh recoverable energy storage density ( W rec ) and high efficiency ( η ) is becoming increasingly urgent. Relaxor ferroelectrics (RFE) suffer from relatively large remnant polarization, while superparaelectric relaxor ferroelectrics (SPE-RFE) see a sharp drop in maximum polarization. Neither can achieve an ideal polarization difference ( Δ P), which is the key to high energy storage performance, making it hard to develop ceramics that meet industrial application requirements. The high-entropy strategy has been proven effective in optimizing the energy storage performance of BT-based lead-free ceramics, yet the vast compositional space of high-entropy systems makes traditional trial-and-error development methods extremely inefficient and costly, severely slowing down the discovery of high-performance materials.
Recently, a research team led by Xiwei Qi from Shijiazhuang Tiedao University and Xiaoyan Zhang from Northeastern University achieved a critical breakthrough in this field. The team developed a machine learning (ML) accelerated design strategy, successfully screening out the optimal BT-based high-entropy ceramic composition (Ba 0.24 Sr 0.24 Bi 0.26 Na 0.26 )(Ti 0.85 Zr 0.15 )O 3 , and revealed that its ultrahigh energy storage performance stems from its location in the crossover region between RFE and SPE-RFE, where nanodomains and polar nanoclusters coexist synergistically. Experimental verification showed this ceramic achieves an ultrahigh W rec of 10.8 J·cm -3 and a high η of 86% under 600 kV·cm -1 , and also exhibits excellent temperature and frequency stability as well as pulsed charge-discharge performance.
The team published their work in Journal of Advanced Ceramics on March 3, 2026.
“This unique crossover region structure is the core reason for the ceramic’s excellent performance,” said Xiwei Qi, corresponding author of the study from Shijiazhuang Tiedao University. “It allows the material to retain the high maximum polarization characteristic of RFE and the low remnant polarization characteristic of SPE-RFE at the same time, resulting in an ultrahigh polarization difference of about 51.02 μC·cm -2 , which lays the structural foundation for ultrahigh energy storage density and high efficiency.”
“Traditional methods relying on equimolar ratios and trial-and-error can hardly cover the huge compositional space of high-entropy ceramics,” said Xiaoyan Zhang, another corresponding author and researcher at the Key Laboratory of Dielectric and Electrolyte Functional Material Hebei Province. The team built a random forest regression model based on 71 BT-based bulk ceramic samples, and combined it with the expected improvement acquisition function to intelligently screen 660,000 candidate compositions. This method takes into account both model prediction values and uncertainty, achieving a balance between exploration and exploitation of new compositions, and drastically reducing experimental burden and research cycle compared with conventional methods.
This research verifies the great potential of machine learning in accelerating the design of complex functional ceramics, breaking the limitations of traditional methods in the compositional screening of high-entropy energy storage ceramics. The designed BT-based high-entropy ceramic not only has ultrahigh energy storage performance, but also features excellent environmental stability and pulsed charge-discharge performance, laying a solid material foundation for the practical application of lead-free dielectric capacitors in high-end fields such as pulsed power equipment and new energy vehicles. Haowen Liu, Zhiyuan Ma and Kailong Ma from Northeastern University also made important contributions to the model construction, experimental verification and result analysis of the study.
About Author
Xiwei Qi is a professor at the School of Materials Science and Engineering, Northeastern University, China. He received his PhD degree in Materials Science from Tsinghua University in 2004. His main research interests focus on multiferroic materials, high-entropy ceramics, and oxide glasses.
Funding
This work was supported by the National Natural Science Foundation of China under Grant (No. U23A20605), (No. 52572011).
About Journal of Advanced Ceramics
Journal of Advanced Ceramics (JAC) is an international academic journal that presents the state-of-the-art results of theoretical and experimental studies on the processing, structure, and properties of advanced ceramics and ceramic-based composites. JAC is Fully Open Access, monthly published by Tsinghua University Press, and exclusively available via SciOpen . JAC’s 2024 IF is 16.6, ranking in Top 1 (1/34, Q1) among all journals in “Materials Science, Ceramics” category, and its 2024 CiteScore is 25.9 (5/130) in Scopus database. ResearchGate homepage: https://www.researchgate.net/journal/Journal-of-Advanced-Ceramics-2227-8508
Journal of Advanced Ceramics
Machine learning-driven BaTiO3-based high-entropy ceramics with ultrahigh energy storage density from crossover region
3-Mar-2026