Ferroelectric Devices Research Articles

Enhancement of Interfacial Polarization in BaTiO3 Thin Films via Oxygen Inhomogeneity

AbstractEnhancement of interfacial ferroelectricity is crucial for the development of nanoscale ferroelectric devices, such high density and nonvolatile memory. Although the epitaxial strain can lead to improved properties, the limited choices of substrates and strain relaxation have hindered the further tuning of ferroelectricity by strain engineering. To overcome this limitation, the interaction between polarization and defects is regarded as a potential route to enhance the interfacial polarization. Here, a universal method to achieve a polarization‐enhanced interface by tuning the distribution of oxygen in lead‐free BaTiO3 films is demonstrated. The interfacial phase processes largely increased c‐axis lattice constant, tetragonality, and about 100% increasement of Ti atom displacement near interface within thickness of 8 unit cells, suggesting enhancement of spontaneous polarization up to 70 μC cm‐2. The gradient distribution of oxygen along c‐axis is also revealed. In addition, the first‐principle Monte Carlo simulation of the tetragonality and polarization support the experimental results. It is suggested that the inhomogeneity of oxygen and interfacial charges could induce an electric field and further enhance the interfacial polarization. This work provides a new route of utilizing the oxygen distribution to engineer the ferroelectrics and paves the way for the development of nanoscale ferroelectric devices.

(Invited) HfZrO-Based Ferroelectric Devices for Lower Power AI and Memory Applications

Electron devices using ferroelectric materials have recently stirred a strong interest as a technology booster for future integrated systems, since the discovery of HfO2-based ferroelectrics [1], which are CMOS-friendly and scalable. At present, a variety of devices using HfO2-based ferroelectrics such as FeRAM, FeFETs and FTJ have been extensively studied for low power memories and computing-in-memory applications. This enthusiasm has been enhanced by the increasing demands and strong concerns of AI calculations, which needs coexistence of both logic and memory functions. Thus, we are currently investigating the possibilities of FeRAM and FeFETs using Hf0.5Zr0.5O2 (HZO) films from the viewpoints of ultralow power memory and AI applications. For FeRAM, we have examined the possibility of low voltage operation and resulting thinning HZO films under low thermal budget needed for formation in BEOL for embedded non-volatile memory applications under advanced technology nodes [2]. However, we have found that there is a trade-off between crystallization temperature and HZO film thickness less than 6 nm. We have demonstrated that MFM capacitors with HZO less than 5 nm can realize crystallization temperature lower than 500ºC, excellent ferroelectricity (2Pr > 25 mC/cm2), low operating voltage (0.7-1.2 V), and high read/write endurance by performing sufficient amounts of wake-up operations to the thin HZO films. For FeFETs, we have quantitatively studied the behaviors of free carriers and the trapping around MFIS interfaces in order to well understand the FeFET memory operation and the reliability, on which the MFIS interfaces have a crucial influence [3-6]. We have clarified that a large amount of electron trapping near the HZO/interfacial layer interfaces can help polarization of HZO while hole trapping is less significant, inducing asymmetric P-V loops in the FeFET memory operation, by employing quasi-static split C-V and Hall measurements to Si FeFETs. This electron trapping properties can significantly affect the memory window, the read-out operation and the FeFET reliability. On the other hand, the complicated dynamics of the FeFET operation including polarization switching and carrier trapping can provide a possibility to utilize this device as AI calculation hardware. We have proposed a new AI calculation scheme by reservoir computing using FeFETs for neuromorphic applications [7, 8]. Here, reservoir computing, which is one of recurrent neural networks [9, 10], is known as promising for AI calculations with low-power consumption, real-time learning and high potential to time-series data processing. Also, reservoir computing by any physical systems holding short-term memory and non-linear response functions can lead to further improvement in energy efficiency of AI calculations [11]. FeFETs with the memory effects due to polarization and the non-linearity due to the threshold operation and the complicated interactions between domains and between domains/channels are eligible as physical reservoirs. We have experimentally demonstrated memory and parity check task operations of time-series input data and the ability to classify them by taking time responses of the drain current for gate voltage input as the virtual nodes. As a result, reservoir computing systems based on Si FeFETs are expected to provide CMOS-platform-friendly hardware for neuromorphic AI calculations with low power consumption, low cost and fast learning time.This work was supported by JST CREST (JPMJCR20C3), JSPS KAKENHI (19K15021), and Nanotechnology Platform project by MEXT (JPMXP09A20UT0056), Japan.

Real-Time Monitoring of Nanoscale Polarization Switching

Researchers have visualized the nanoscale jumps in a ferroelectric’s polarization that are thought to play a key role in how well some ferroelectric devices function.

Probing Ultrafast Dynamics of Ferroelectrics by Time-Resolved Pump-Probe Spectroscopy.

Ferroelectric materials have been a key research topic owing to their wide variety of modern electronic and photonic applications. For the quick exploration of higher operating speed, smaller size, and superior efficiencies of novel ferroelectric devices, the ultrafast dynamics of ferroelectrics that directly reflect their respond time and lifetimes have drawn considerable attention. Driven by time‐resolved pump‐probe spectroscopy that allows for probing, controlling, and modulating dynamic processes of ferroelectrics in real‐time, much research efforts have been made to understand and exploit the ultrafast dynamics of ferroelectric. Herein, the current state of ultrafast dynamic features of ferroelectrics tracked by time‐resolved pump‐probe spectroscopy is reviewed, which includes ferroelectrics order parameters of polarization, lattice, spin, electronic excitation, and their coupling. Several potential perspectives and possible further applications combining ultrafast pump‐probe spectroscopy and ferroelectrics are also presented. This review offers a clear guidance of ultrafast dynamics of ferroelectric orders, which may promote the rapid development of next‐generation devices.

Wake-up and fatigue mechanisms in ferroelectric Hf0.5Zr0.5O2 films with symmetric RuO2 electrodes

The mechanisms leading to wake-up and fatigue in ferroelectric hafnium zirconium oxide thin film devices with symmetric RuO2 electrodes are investigated via polarization, relative permittivity, dielectric nonlinearity, pyroelectric coefficient, and microfocus x-ray diffraction (XRD) measurements. The devices are observed to wake-up for up to 103 bipolar pulsed field cycles, after which fatigue occurs with polarization approaching zero following 108 cycles. Wake-up is accompanied by a decrease in both high-field permittivity and hysteresis loop pinching and an increase in the pyroelectric coefficient, indicating that the wake-up process involves a combination of transformations from the tetragonal to the orthorhombic phase and domain depinning from defect redistribution. Fatigue is observed to coincide with an increase in irreversible domain wall motion and a decrease in pyroelectric coefficient. Finite pyroelectric coefficients are measured on fully fatigued devices, indicating that domain pinning is a strong contributor to fatigue and that fatigued devices contain domain structures that are unable to switch under the fields applied for measurement. Microfocus XRD patterns measured on each device reveal that the phase constitution is qualitatively unaffected by field cycling and resultant polarization fatigue. These data indicate that the wake-up process has contributions from both phase transformations and domain depinning, whereas the fatigue process is driven primarily by domain pinning, and the near-zero measured switchable polarization is actually a poled device with immobile domains. These observations provide insight into the physical changes occurring during field cycling of HfO2-based ferroelectrics while examining a possible oxide electrode material for silicon CMOS device implementation.

(Invited) HfZrO-Based Ferroelectric Devices for Lower Power AI and Memory Applications

Si-friendly HfO2-based ferroelectric devices have been strongly recognized as a novel technology booster for future integrated memory and logic systems. In this paper, we address our recent activities on TiN/Hf0.5Zr0.5O2(HZO)/TiN MFM capacitors, HZO/Si FeFETs for memory applications and a newly-proposed reservoir computing using HZO/Si FeFETs and MFM capacitors for AI applications. We have demonstrated that MFM capacitors with HZO less than 5 nm can realize low crystallization temperature, excellent ferroelectricity, low operating voltage and high read/write endurance by performing sufficient wake-up operations to the thin HZO films. For the FeFET memory, we have found the importance of interfacial layers (ILs) between HZO and Si on the memory window. It has been revealed that the IL thickness is sensitive to the process temperature and that electron trapping around HZO/ILs has significant impacts on the memory operation. Finally, we have proposed and experimentally demonstrated reservoir computing using FeFETs for neuromorphic applications.

Engineering crystal phase of Nylon-11 films for ferroelectric device and piezoelectric sensor

Among non-fluorinated ferroelectric polymers, odd number Nylons have recently attracted attention because of their solution-processibility and decent ferroelectric characteristics. Here, we demonstrate a ferroelectric and piezoelectric Nylon-11 δ` phase crystal film manufactured via a humidity modulated casting and subsequent melt-quenching process. The low relative humidity (RH) below 20% during the spin casting process effectively mitigates the formation of hygroscopic holes resulting in the formation of pin-hole free Nylon-11 thin films. In addition, the subsequent melt-quenching gives rise to the evolution of the ferroelectric δ` phase crystals, which is carefully confirmed by grazing incidence X-ray diffraction and infrared spectroscopy measurements. The resulting ferroelectric thin films show decent remnant polarization of ~4.6 μC/cm2 at a low operating voltage of ±30 V, suggesting the formation of high-quality Nylon-11 δ` phases thin film. Furthermore, the centrifugal casting at low RH facilitates the fabrication of freestanding and transparent 20-μm thick ferroelectric δ` phase film after the subsequent melt-quenching process. The resulting Nylon-11 film provides excellent piezoelectric performance upon external vertical pressure, which enables a proof-of-concept demonstration as a Morse code detector.

Intrinsic synaptic plasticity of ferroelectric field effect transistors for online learning

Nanoelectronic devices emulating neuro-synaptic functionalities through their intrinsic physics at low operating energies are imperative toward the realization of brain-like neuromorphic computers. In this work, we leverage the non-linear voltage dependent partial polarization switching of a ferroelectric field effect transistor to mimic plasticity characteristics of biological synapses. We provide experimental measurements of the synaptic characteristics for a 28 nm high-k metal gate technology based device and develop an experimentally calibrated device model for large-scale system performance prediction. Decoupled read-write paths, ultra-low programming energies, and the possibility of arranging such devices in a cross-point architecture demonstrate the synaptic efficacy of the device. Our hardware-algorithm co-design analysis reveals that the intrinsic plasticity of the ferroelectric devices has potential to enable unsupervised local learning in edge devices with limited training data.

Helical Nanostructures of Ferroelectric Liquid Crystals as Fast Phase Retarders for Spectral Information Extraction Devices: A Comparison with the Nematic Liquid Crystal Phase Retarders.

Extraction of spectral information using liquid crystal (LC) retarders has recently become a topic of great interest because of its importance for creating hyper- and multispectral images in a compact and inexpensive way. However, this method of hyperspectral imaging requires thick LC-layer retarders (50 µm–100 µm and above) to obtain spectral modulation signals for reliable signal reconstruction. This makes the device extremely slow in the case of nematic LCs (NLCs), since the response time of NLCs increases proportionally to the square of the LC-layer thickness, which excludes fast dynamic processes monitoring. In this paper, we explore two approaches for solving the speed problem: the first is based on the use of faster nanospiral ferroelectric liquid crystals as an alternative to NLCs, and the second is based on using a passive multiband filter and focuses on multispectral extraction rather than hyperspectral. A detailed comparative study of nematic and ferroelectric devices is presented. The study is carried out using a 9-spectral bands passive spectral filter, covering the visible and near-infrared ranges. We propose the concept of multispectral rather than hyperspectral extraction, where a small number of wavelengths are sufficient for specific applications.

Ferroelectric HfO2-based synaptic devices: recent trends and prospects

Neuro-inspired deep learning algorithms have shown promising futures in artificial intelligence. Despite the remarkable progress in software-based neural networks, the traditional von-Neumann hardware architecture has suffered from limited energy efficiency while facing unprecedented large amounts of data. To meet the performance requirements of neuro-inspired computing, large-scale vector-matrix multiplication is preferred to be performed in situ, namely compute-in-memory. Non-volatile memory devices with different materials have been proposed for weight storage as synaptic devices. Among them, HfO2-based ferroelectric devices have attracted great attention because of their low energy consumption, good complementary-metal-oxide-semiconductor (CMOS) compatibility and multi-bit per cell potential. In this review, recent trends and prospects of the ferroelectric synaptic devices are surveyed. First, we present the three-terminal synaptic devices based on the ferroelectric field effect transistor (FeFET), and discuss the switching physics of the intermediate states, the back-end-of-line integration and the 3D NAND architecture design. Then, we introduce a hybrid precision synapse concept that leverages the volatile charges on the gate capacitor of the FeFET and the non-volatile polarization on the gate dielectric of the FeFET. Lastly, we review two-terminal synaptic devices using the ferroelectric tunnel junction (FTJ) and ferroelectric capacitor (FeCAP). The design margins of the crossbar array with FTJ and FeCAP analyzed.

Interlayer engineering for enhanced ferroelectric tunnel junction operations in HfO x -based metal-ferroelectric-insulator-semiconductor stack

Ferroelectric tunnel junction (FTJ) has been considered as a promising candidate for next-generation memory devices due to its non-destructive and low power operations. In this article, we demonstrate the interlayer (IL) engineering in the FTJs to boost device performances. Through the analysis on the material and electrical characteristics of the fabricated FTJs with engineered IL stacks, it is clearly found that the insertion of an Al2O3 layer between the SiO2 insulator and the pure-HfO x FE improves the read disturbance (2V c = 2.2 V increased), the endurance characteristics (tenfold improvement), and the cell-to-cell TER variation simultaneously without the degradation of the ferroelectricity (less than 5%) and the polarization switching speeds through grain size modulation. Based on these investigations, the guidelines of IL engineering for low power ferroelectric devices were provided to obtain stable and fast memory operations.

Fully epitaxial ferroelectric ScGaN grown on GaN by molecular beam epitaxy

We report on the ferroelectric properties of single-phase wurtzite ScGaN grown on GaN by plasma-assisted molecular beam epitaxy. Distinct ferroelectric switching behavior was confirmed by detailed electrical characterization. Coercive fields in the range of 2.0–3.0 MV/cm and large, retainable remnant polarization in the range of 60–120 μC/cm2 are unambiguously demonstrated for ScGaN epilayers with Sc contents of 0.31–0.41. Taking advantage of the widely tunable energy bandgap of III-nitride semiconductors, the demonstration of ferroelectricity in ScGaN, together with the recently reported ferroelectric ScAlN, will enable a broad range of emerging applications with combined functionality in ferroelectric, electronic, optoelectronic, photovoltaic, and/or photonic devices and systems.

Pyroelectric dependence of atomic layer-deposited Hf0.5Zr0.5O2 on film thickness and annealing temperature

Ferroelectric Hf0.5Zr0.5O2 is a prime candidate material for integrated HfO2-based ferroelectric devices due to its simple composition, low crystallization temperature, and significant remanent polarization. It is particularly promising for integrated pyroelectric devices used in infrared sensing and energy harvesting, although the appearance of nonferroelectric tetragonal and monoclinic phases should be avoided to achieve high-performance pyroelectric sensors. Both nonferroelectric phases are strongly influenced by the Hf0.5Zr0.5O2 film thickness and annealing temperature. The sensitivity of the pyroelectric coefficient on film thickness is investigated with atomic layer-deposited Hf0.5Zr0.5O2 films within a 10–30 nm thickness range. The films are capped with TiN and undergo post-metallization anneals at 450 °C and 600 °C. An optimum pyroelectric coefficient of −56 μC K−1 m−2 is found in the 15 nm thick Hf0.5Zr0.5O2. The pyroelectric coefficient is found to be sensitive to thickness-dependent depolarization effects and monoclinic phase growth. Ferroelectric, dielectric, and pyroelectric properties are improved with a lower annealing temperature, demonstrating the back-end-of-line compatibility of Hf0.5Zr0.5O2 pyroelectric devices.

Controllable band offset in monolayer MoSe2 driven by surface termination and ferroelectric field of BiFeO3(0001) substrate

Integration of 2D van der Waals crystals into ferroelectric materials is a reliable way for the technological leap toward next-generation (opto)electronic applications. In present work, first-principles density functional theory calculations were performed to describe nonvolatile carrier modulation and ferroelectric field effect caused by interface coupling of MoSe2 monolayer with ferroelectric BiFeO3(0001) substrate. The Fermi level of MoSe2 monolayer on BiFeO3(0001) surface showed strong dependence on surface termination and polarization direction. Furthermore, polarization switching made MoSe2/BiFeO3(0001) heterointerface evolve into energetically metastable state and remarkable ferroelectric field effect on band offset in MoSe2 monolayer was observed due to distinct interface charge transfer. Therefore, the results of this work open up new prospects for surface chemistry and domain structure engineering of MoSe2 homojunctions for MoSe2/BiFeO3(0001)-based ferroelectric field effect devices.

Synthesis, characterization, and visible light photocatalytic activity of solution-processed free-standing 2D Bi2O2Se nanosheets

Owing to their unique structural and electronic properties such as layered structure with tuneable bandgap and high electron mobility, 2D materials have emerged as promising candidates for photocatalysis. Recently, bismuth oxyselenide (Bi2O2Se), a member of bismuth oxychalcogenide’s family has shown great potential in high-speed field-effect transistors, infrared photodetectors, ferroelectric devices, and electrochemical sensors. However, the potential of Bi2O2Se in photocatalysis has not yet been explored. In the current work, Bi2O2Se nanosheets with an average size of ∼170 nm and a lattice strain of 0.01 were synthesized at room temperature using a facile solution-processed method and the as-synthesized material was investigated with various characterization techniques such as x-ray diffraction, FE-SEM, UV–vis spectroscopy. The bandgap for the indirect transition in Bi2O2Se nanosheets was estimated to be 1.19 eV. Further, the visible-light-driven photocatalytic degradation of methylene blue (MB) dye using Bi2O2Se as a photocatalyst is presented. The photocatalytic experiments demonstrate the promising photocatalytic ability of Bi2O2Se as it leads to 25.06% degradation of MB within 80 min of light illumination. The effect of active species trapping agents (carrier and radical scavengers) on photocatalytic activity is also presented and discussed.

Structural, electronic, and polarization properties of YN and LaN

ScN has attracted great attention for its electronic properties and its ability to enhance polarization of AlN; however, its sister compounds, YN and LaN, remain much less studied. Here, we use first-principles calculations to evaluate YN and LaN in their cubic and hexagonal phases. Rocksalt YN and LaN are semiconductors, although we show that LaN differs from ScN and YN in having a direct band gap, which we attribute to its weaker $p\text{\ensuremath{-}}p$ coupling. Both have low electron effective masses. In addition to their rocksalt structures, we evaluate the layered hexagonal and wurtzite phases of YN and LaN. For YN, the wurtzite phase cannot be stabilized, and hexagonal YN is higher in energy than rocksalt YN. In contrast, for LaN, the wurtzite phase is favored, and it is comparable in energy to rocksalt LaN. Wurtzite LaN has a polarization of $0.608\phantom{\rule{0.28em}{0ex}}\mathrm{C}/{\mathrm{m}}^{2}$ (referenced to the centrosymmetric layered hexagonal phase), and a high piezoelectric coefficient ${e}_{33}=1.78\phantom{\rule{0.16em}{0ex}}\mathrm{C}/{\mathrm{m}}^{2}$. Interestingly, we find that the polarization of wurtzite LaN may be reversible; we find a relatively small switching barrier of 0.06 eV per formula unit, offering the potential for its use as a ferroelectric. Since wurtzite LaN is closely lattice matched to InP, we investigate a heterostructure between (0001) wurtzite LaN and (111) zinc-blende InP, and find the polarization discontinuity would yield a bound charge of $1.3\ifmmode\times\else\texttimes\fi{}{10}^{14}\phantom{\rule{0.16em}{0ex}}e/{\mathrm{cm}}^{2}$, offering the potential for novel electronic applications such as tunnel junctions. Our results compare and contrast ScN, YN, and LaN, and highlight the potential of these materials for adoption in electronic and ferroelectric devices.

Preisach modeling of imprint on hafnium zirconium oxide ferroelectric capacitors

Imprint, the preferential orientation of the polarization of a ferroelectric device subjected to elevated temperatures, is a primary reliability concern afflicting data retention in ferroelectric RAM. In this paper, we demonstrate Preisach-based hysteresis modeling, which can be used to predict imprint behavior in ferroelectric thin films. A method was developed for capturing imprint in the context of a Preisach model and a numerical approach for evaluating the Preisach distribution was expanded upon. Interpolation and curve fitting were used to make predictions of the Preisach distributions of imprinted ferroelectric hafnium zirconium oxide devices after short-duration bakes at 23–260 °C and long-term bakes at 85 and 125 °C. In the case of long-term bakes, imprint-induced coercive shifts were modeled as shifts in the derivative of the top and bottom hysteretic polarization curves. The shift in the curves is modeled by fitting experimental data to a commonly used empirically logarithmic relationship reported in the literature. Simulations give remanent polarizations and coercive fields within <5.0 μC/cm2 and 0.1 V, respectively, of the raw data average.

Ferroelectric phase transitions induced by a strain gradient

In perovskite oxide ferroelectrics, gradients of lattice strain are known to induce nanoscale topological structures, leading to novel or enhanced functionality. Here, we experimentally detect and theoretically analyze thickness distribution of structural properties in epitaxial ${\mathrm{Pb}}_{0.5}{\mathrm{Sr}}_{0.5}\mathrm{Ti}{\mathrm{O}}_{3}$ films grown on (001) $\mathrm{SrTi}{\mathrm{O}}_{3}$ substrates. We show that the relaxation of substrate-imposed stress produces a strain gradient, which leads to the formation of distinct ferroelectric phases as a function of distance from the film-substrate interface within the same film. Charge carriers trapped at phase boundaries stabilize the induced phases and manifest themselves under electric field. Crosstalk between the phases, where polarization may rotate in one phase and invert in the other one, opens perspectives for advanced ferroelectric thin film devices.

Mexican-hat potential energy surface in two-dimensional III2-VI3 materials and the importance of entropy barrier in ultrafast reversible ferroelectric phase change

First-principles calculations reveal a Mexican-hat potential energy surface (PES) for two-dimensional (2D) In2Se3. This unique PES leads to a pseudo-centrosymmetric paraelectric β phase that resolves the current controversy between theory and experiment. We further show that while the α-to-β (ferroelectric-to-paraelectric) phase transition is fast and coherent, assisted by an in-plane shear phonon mode, a random distribution of the atoms in the trough of the PES acts as an entropy barrier against the reverse β-to-α transition. This will be the origin of the speed limitation of current In2Se3 ferroelectric devices. However, if one orders the β phase (due to the formation of in-plane ferroelectric domains), the reverse transition can take place within tens of picoseconds in the presence of a perpendicular electric field. Finally, the Mexican-hat PES is a general feature for the entire family of 2D III2-VI3 materials. Our finding offers a critical physical picture in controlling the ultrafast reversible phase transition in 2D In2Se3 and other III2-VI3 materials, which will benefit their practical industrial development for advanced ferroelectric devices.

N-polar ScAlN and HEMTs grown by molecular beam epitaxy

Molecular beam epitaxy of single-phase wurtzite N-polar ScxAl1−xN (x ∼ 0.11–0.38) has been demonstrated on sapphire substrates by locking its lattice-polarity to the underlying N-polar GaN buffer. Coherent growth of lattice-matched Sc0.18Al0.82N on GaN has been confirmed by x-ray diffraction reciprocal space mapping of the asymmetric (105) plane, whereas lattice-mismatched, fully relaxed Sc0.11Al0.89N and Sc0.30Al0.70N epilayers exhibit a crack-free surface. The on-axis N-polar crystallographic orientation is unambiguously confirmed by wet chemical etching. High electron mobility transistors using N-polar Sc0.18Al0.82N as a barrier have been grown on sapphire substrates, which present the existence of well confined two-dimensional electron gas. A Hall mobility of ∼564 cm2/V s is measured for a 15-nm-thick Sc0.18Al0.82N barrier sample with a sheet electron concentration of 4.1 × 1013 cm−2, and the corresponding sheet resistance is as low as 271 Ω/sq. The polarity-controlled epitaxy of ScxAl1−xN provides promising opportunities for applications in high-frequency and high-power electronic and ferroelectric devices.