Ceramics dielectric capacitors are widely used in technologies due to their fast charge-discharge rate, i.e., high power-density. Their energy-storage density can be significantly improved if antiferroelectric (AFE) ceramics can be used as the dielectric layer. During charging, AFE ceramics abruptly transform to the ferroelectric (FE) state (at a critical field EF), and during discharging revert to the AFE state (at discharge field EA). The difference between these two electric field values (∆E = E_F – E_A) is a measure of the hysteresis in the system. This electric hysteresis makes AFE ceramics mediocre in terms of their energy efficiency (<80% of the stored energy can be discharged), and poor cyclability (<1x10⁶ cycles before failure). The physical origin of the ∆E comes from the crystallographic lattice mismatches between the AFE and FE crystalline regions—they lack “lattice compatibility”. These mismatches cause interfacial defects and lattice distortions, hindering motion of the AFE-FE interface during the phase transition.

Here we present a new methodology to eliminate the drawbacks in AFE dielectrics by eliminating their cause — the electric-field-induced phase-transition hysteresis. We first apply the geometric nonlinear theory of martensitic phase transition to AFE ceramics. In nickel-titanium-based alloys, we previously showed that by tuning the lattice parameters to create highly compatible austenite/martensite interfaces, the cycle life of the alloys can be increased from 20 thousand to over 10 million. We then conduct high-throughput density-functional theory (DFT) calculations to predict the interfacial lattice strain, the polarization in the FE phase, and the stability of the AFE state in a wide range of PbZrO₃-based compositions. In parallel, we perform combinatorial synthesis of thousands of ceramics adjusting simultaneously 4 or 5 variables in compositions, and conduct high-throughput screening of their critical fields (E_F and E_A), maximum polarization, electric hysteresis, energy density and energy efficiency. The large amount of data generated from DFT calculations and high-throughput synthesis/screening are fed to machine-learning (ML) models to identify correlations and predict best compositions. In the final task, predicted compositions are fabricated in bulk form and AFE ceramic disk capacitors are evaluated for their cyclability. Currently, guided by DFT calculations and ML models, high-throughput synthesis has identified ~10 compositions with an energy density of 3.2 J/cm³ and an efficiency of 93% at applied fields below 200 kV/cm in the (Pb,Ba,Sr,La)(Zr,Sn,Ti)O3 system. We have demonstrated that by further modification with Bi(Zn_2/3Nb_1/3)O₃ and Bi(Mg_2/3Nb_1/3)O₃, the energy efficiency can be improved to 96%. Other doping schemes effective on improving efficiency are also explored and compared. Our goal is to produce AFE ceramic capacitors with >3.0 J/cm³ energy density, 98% charge-discharge energy efficiency, and up to 10⁸ cycles charge-discharge lifetime.