Theoretical polarization vector distribution, P-E loops (phase-field simulation), and energy storage performance of (1-x)(0.80Bi_0.5Na_0.5TiO₃-0.20NaNbO₃)-xBa(Ni_1/3Nb_2/3)O₃ ceramics.

Dielectric ceramic capacitors with high dielectric constants possess extremely high power density and ultra-fast charge-discharge rates, making them widely used in fields such as new energy vehicles and high-power pulsed power systems.

However, a high dielectric constant often comes at the cost of low breakdown strength and high dielectric loss. This results in lower recoverable energy storage density (W_rec) and low charge-discharge efficiency (h), hindering the development of miniaturized, lightweight, and integrated electronic devices.

To address these challenges, Professor Zhou Di's team from the School of Electronic Science and Engineering, Faculty of Electronic and Information Engineering at Xi'an Jiaotong University (XJTU), proposed a research strategy utilizing entropy-driven local multiphase polarization states to reduce the size of polar nanoregions (PNRs), thereby achieving ultra-high energy storage efficiency.

The team first used phase-field simulations to study the arrangement of microscopic polarization vectors and variations in macroscopic hysteresis loops within the (1-x)(0.80Bi_0.5Na_0.5TiO₃-0.20NaNbO₃)-xBa(Ni_1/3Nb_2/3)O₃ ternary system.

They discovered that as entropy increased, long-range ordered ferroelectric domains were gradually broken down into R-T type polar nanoregions, accompanied by a significant reduction in remnant polarization.

Consequently, in the 1.76R high-entropy composition, they achieved a high energy storage density W_rec of 6.8 J·cm^-3 and a high efficiency η of 95.1 percent. The material demonstrated excellent stability under a 300 kV·cm^-1 electric field across a wide temperature range (20–140 C), wide frequency range (1–100 Hz), and long cycling (1-105), showcasing the immense application potential of this dielectric ceramic capacitor.

Furthermore, using aberration-corrected scanning transmission electron microscopy (STEM), the team observed R-T type PNRs approximately 1 nm in size within the system's cubic matrix. The presence of the cubic matrix suppresses polarization switching and internal stress during the application of an electric field, while acting as a restorative force that promotes polarization recovery to the initial state once the field is removed. This delays polarization saturation, thereby achieving near-zero loss (high η).

Based on the aforementioned polarization structure regulation mechanism, and to further enhance energy storage performance while accelerating material optimization, the team collaborated with Professor Huang Houbing's group from the Beijing Institute of Technology.

They employed machine learning (ML) to synergistically build a coupled search space of material chemical composition and ferroelectric domain structures, developing efficient optimization algorithms to design high-entropy energy storage dielectric ceramics.

Inspired by the Landau-Ginzburg-Devonshire theory, this study proposed an inverse design framework. It combines variational generative models with active learning optimization to accelerate the development of ferroelectric ceramics with enhanced energy storage performance under low electric fields.

By formulating the solution of the Time-Dependent Ginzburg-Landau (TDGL) equations, which govern domain structure evolution, as conditional sampling within a model's latent space, the team achieved synergistic optimization of chemical and polarization configurations.

These research findings were published online in the internationally renowned journals ACS Nano and Nature Communications, titled Entropy-regulated local multiphase polarization states for near-zero energy loss in relaxor ferroelectrics and Active learning in latent spaces enables rapid inverse design of ferroelectric ceramics for energy storage, respectively.