Zhi-Wei Zheng

Ferroelectric hafnium zirconium oxide (HZO) thin films have emerged as one of the most promising candidates for nonvolatile memory and logic applications. The ferroelectricity of HZO film stems from the non-centrosymmetric orthorhombic phase, which is metastable and can be stabilized via the introduction of oxygen vacancies (Vo). To modulate Vo behaviors, a variety of processing strategies, including doping engineering, thermal annealing, and interfacial engineering, have been demonstrated to be effective in promoting phase stabilization and performance enhancement of HZO films. Nevertheless, despite the effectiveness of these approaches in enhancing ferroelectricity, the underlying mechanisms, especially the role of Vo, remain unclear. In this study, we employed capping-layer-based interfacial engineering via the deposition of Al ₂ O ₃ capping layers with varying thicknesses on HZO films to construct Al ₂ O ₃ /HZO capacitor devices. First-principles calculations based on density functional theory were carried out to calculate the interfacial energies, Vo formation energies, and Vo migration barriers, with the aim of investigating the influences of Vo on phase stabilization and ferroelectric polarization. The calculation results reveal that both the Al ₂ O ₃ thickness and the Vo concentration requires precise tailoring for optimizing the ferroelectric performance. These theoretical findings are in good agreement with our experimental observations. This work provides an alternative solution to facilitate the advancement of high-performance HZO-based ferroelectric devices.

This study develops a physics-based modeling framework for back-end-of-line (BEOL)-compatible oxide-semiconductor-channel field-effect transistors (FeFETs), integrating nucleation-limited switching theory and technology computer-aided design technology. Based on this framework, the interdomain screening effect, which significantly modulates domain electric fields and subsequently impacts polarization switching dynamics, is systematically investigated as a focus herein. The core of this model lies in the asymmetric characterization of ferroelectric domains (denoted as PFE+/PFE−), which originates from the weak erase effect. This framework successfully reproduces the ID−VG hysteresis loops, achieving excellent consistency with experimental data including the abrupt current jump and asymmetric subthreshold swings, thereby validating the model's physical accuracy. Key findings reveal that the depolarization field not only acts as the intrinsic driver of polarization instability in individual ferroelectric domains but also modulates the switching behaviors of neighboring domains via the interdomain screening effect, ultimately constraining device reliability. This highlights the interdomain screening effect as a pivotal link between microscopic domain polarization and macroscopic device performance. Furthermore, the study uncovers the polarization reversal mechanism that underlies the memory window enhancement induced by channel length scaling. These results provide critical physical insights and actionable design guidance for optimizing the performance and reliability of BEOL-compatible FeFET devices.

Lanthanum-doped hafnium zirconium oxide (La:HZO) has demonstrated high potential to solve the endurance challenge of hafnia-based ferroelectric memory. To date, La:HZO has been experimentally reported to exhibit ferroelectricity (FE) with lower La concentration (<1 mol. %) but anti-ferroelectricity (AFE) with higher La concentration that exceeds 1 mol. %. Despite the improved endurance with the introduction of La dopants, there still lacks an in-depth understanding to explain the mechanisms behind FE to AFE transition. In this work, we theoretically studied the origin of FE and AFE characteristics in La:HZO thin films through first-principles calculations. The polarization switching process was also analyzed based on the Landau–Ginzburg–Devonshire theory. The calculation results reveal that the dominant phase in La:HZO changes from non-FE monoclinic to FE orthorhombic to AFE tetragonal phase with increasing the concentration of La dopants, which results in FE-AFE transition. During the process, the presence of oxygen vacancies also facilitates the formation of the tetragonal phase, which highlights the significant role of oxygen vacancies in phase stabilization. Our findings explain the physical origin of the narrow doping window in La:HZO and provide a theoretical basis for optimizing HZO-based ferroelectric devices by regulating the concentration of dopants and oxygen vacancies within the film.

In this study, a photothermal pain sensing neural platform for fire scenarios was implemented based on the integration of a multigate Li-ion electrolyte-gated transistor device with a thermoelectric generator and a photoresistor. Peripheral sensitization, desensitization, pain threshold, and other basic nociceptive behaviors of nociceptive afferents were successfully emulated. This photothermal pain sensing platform realized the judgment of the distance of the fire from the incident light of the photothermal source. The secondary injury process and the composite injury sensing process were also mimicked. This work extends the applications of Li-ion synaptic transistors in artificial nervous systems and provides new ideas for the development of practical and feasible pain sensing systems.

Neuromorphic computing that enables parallel computation and adaptive learning process has the potential to overcome the traditional von Neumann bottleneck. To build a neuromorphic system, synapse is deemed as one of the most essential elements. In this study, adopting a metal-ferroelectric-metal-insulate-semiconductor (MFMIS) structure, we demonstrated a hafnia-based oxide-semiconductor-channel ferroelectric field-effect transistor (OS-FeFET) as a synapse with multilevel characteristics, and investigated the impact of area ratio (AR) between the MOS capacitor (A _MOS) and the ferroelectric capacitor (A _FE) on the memory and synaptic performances through the simulation. The present results show that not only the memory window but also the linearity of synaptic weight update and dynamic range could be significantly improved as the AR increases. Furthermore, an artificial neural network (ANN) based on these OS-FeFET synapses was constructed, which could achieve 85.3% accuracy in recognizing the handwritten digits. This study may provide insights for future OS-FeFET synapse design for neuromorphic application.

Although a nanosheet ferroelectric field-effect transistor (FeNSFET) with the gate-all-around structure has proved to be a promising solution to enhance memory density, device performances could be greatly affected by the corner dead zone of ferroelectric polarization. In this study, a novel FeNSFET with a surrounded gate (S-FeNSFET), which is different from that with the conventional all-around gate (A-FeNSFET), was proposed to alleviate the dead zone issue. Through Technology Computer Aided Design simulation, in the case of two nanosheets (NSs), the memory window (MW) of the S-FeNSFET is ∼85% larger than that of the A-FeNSFET. By analyzing the ferroelectric polarization distribution, it is confirmed that the corner dead zone could be effectively mitigated to improve ferroelectric polarization in the S-FeNSFET, thereby enhancing memory performances. Further comparison between these two devices with different NS numbers reveals that the MW and subthreshold swing (SS) of the A-FeNSFET almost do not change with the NS number, whereas the MW of the S-FeNSFET increases as the NS number increases, together with a slightly degraded SS. Based on this S-FeNSFET configuration, a complementary reconfigurable logic-in-memory circuit was further designed and validated. The dynamic reconfiguration among AND, OR, and XOR logic functions was successfully achieved by using N-type and P-type S-FeNSFETs in the case of three NSs. This work provides important guidance for advancing the development of FeNSFET technology in the next-generation memory and logic applications.