Search

Memristive neural networks have extensively been investigated for their capability in handling various artificial intelligence tasks. The training performance of memristive neural networks depends on the pulse scheme applied to the constituent memristors. However, the design of the pulse scheme in most previous studies was approached in an empirical manner or through a trial-and-error method. Here, we choose ferroelectric tunnel junction (FTJ) as a model memristor and demonstrate a physics-based predictive model for the pulse design to achieve high training performance. This predictive model comprises a physical model for FTJ that can adequately describe the polarization switching and memristive switching behaviors of the FTJ and an FTJ-based neural network that uses the long-term potentiation (LTP)/long-term depression (LTD) characteristics of the FTJ for the weight update. Simulation results based on the predictive model demonstrate that the LTP/LTD characteristics with a good trade-off between ON/OFF ratio, nonlinearity, and asymmetry can lead to high training accuracies for the FTJ-based neural network. Moreover, it is revealed that an amplitude-increasing pulse scheme may be the most favorable pulse scheme as it offers the widest ranges of pulse amplitudes and widths for achieving high accuracies. This study may provide useful guidance for the pulse design in the experimental development of high-performance memristive neural networks.

Abstract Reservoir computing (RC) offers efficient temporal information processing with low training cost. All-ferroelectric implementation of RC is appealing because it can fully exploit the merits of ferroelectric memristors (e.g., good controllability); however, this has been undemonstrated due to the challenge of developing ferroelectric memristors with distinctly different switching characteristics specific to the reservoir and readout network. Here, we experimentally demonstrate an all-ferroelectric RC system whose reservoir and readout network are implemented with volatile and nonvolatile ferroelectric diodes (FDs), respectively. The volatile and nonvolatile FDs are derived from the same Pt/BiFeO3/SrRuO3 structure via the manipulation of an imprint field (E imp). It is shown that the volatile FD with E imp exhibits short-term memory and nonlinearity while the nonvolatile FD with negligible E imp displays long-term potentiation/depression, fulfilling the functional requirements of the reservoir and readout network, respectively. Hence, the all-ferroelectric RC system is competent for handling various temporal tasks. In particular, it achieves an ultralow normalized root mean square error of 0.017 in the Hénon map time-series prediction. Besides, both the volatile and nonvolatile FDs demonstrate long-term stability in ambient air, high endurance, and low power consumption, promising the all-ferroelectric RC system as a reliable and low-power neuromorphic hardware for temporal information processing.

The poor contact at poly(3-hexylthiophene)/perovskite interface has led to unsatisfactory performance in solar cells. To resolve this issue, the authors introduce a molecular bridge to form a charge transport channel, realizing devices with maximum efficiency value of 22.87% and long-term stability.

Spiking neural network (SNN) consisting of memristor-based artificial neurons and synapses has emerged as a compact and energy-efficient hardware solution for spatiotemporal information processing. However, it is challenging to develop memristive neurons and synapses based on the same material system because the required resistive switching (RS) characteristics are different. Here, it is shown that SrFeOx (SFO), an intriguing material system exhibiting topotactic phase transformation between insulating brownmillerite (BM) SrFeO2.5 phase and conductive perovskite (PV) SrFeO3 phase, can be engineered into both neuronal and synaptic devices. Using a BM-SFO single layer as the RS medium, the Au/BM-SFO/SrRuO3 (SRO) memristor exhibits nonvolatile RS behavior originating from the formation/rupture of PV-SFO filaments in the BM-SFO matrix. By contrast, using a PV-SFO (matrix)/BM-SFO (interfacial layer) bilayer as the RS medium, the Au/PV-SFO/BM-SFO/SRO memristor exhibits volatile RS behavior originating from the interfacial BM-PV phase transformation. Synaptic and neuronal characteristics are further demonstrated in the Au/BM-SFO/SRO and Au/PV-SFO/BM-SFO/SRO memristors, respectively. Using the SFO-based synapses and neurons, fully memristive SNNs are constructed by simulation, which show good performance on unsupervised image recognition. Our study suggests that SFO is a versatile material platform on which both neuronal and synaptic devices can be developed for constructing fully memristive SNNs.

Abstract High‐performance bilayer In2O3/IGZO thin‐film transistors (TFTs) fabricated by pulsed laser deposition are reported. The TFTs exhibit an on/off current ratio of 109, a reversed subthreshold slope (ss) of 0.08 V dec−1, and a high saturation mobility of 47.9 cm2 V−1 s−1. The reliability of the mobility values is critically validated and assessed by four‐probe measurements, the transfer‐length method, and the temperature‐dependence. X‐ray photoelectron spectra are combined with C–V measurements to characterize the interface, and the results show that a two‐dimensional electron gas (2DEG)‐like state accumulates at the In2O3/IGZO interface. However, this state only forms in the subthreshold region and does not cause the high carrier mobility in the region above the threshold. Instead, the enhanced carrier mobility results from the intrinsic high mobility of the In2O3, the smooth surface, and the low‐defect states in the In2O3/IGZO bilayer with a good percolation transport path.

As a high-k material, hafnium oxide (HfO2) has been used in gate dielectrics for decades. Since the discovery of polar phase in Si-doped HfO2 films, chemical doping has been widely demonstrated as an effective approach to stabilize the ferroelectric phase in HfO2 based thin films. However, the extra capping layer deposition, post-growth annealing and wake-up effect are usually required to arouse the ferroelectricity in HfO2 based thin films, resulting in the increase of complexity for sample synthesis and the impediment of device application. In this study, the ferroelectricity is observed in non-capped dopant-free HfO2 thin films prepared by pulsed laser deposition (PLD) without post-growth annealing. By adjusting the deposited temperature, oxygen pressure and thickness, the maximum polarization up to 14.7 μC/cm2 was obtained in 7.4 nm-thick film. The fraction of orthorhombic phase, concentrations of defects and size effects are considered as possible mechanisms for the influences of ferroelectric properties. This study indicates that PLD is an effective technique to fabricate high-quality ferroelectric HfO2 thin films in the absence of chemical doping, capping layer deposition and post-growth annealing, which may boost the process of nonvolatile memory device application.

Robust, fast, and low-power hardware platforms are desirable for the implementation of real-time machine vision. Here the authors develop a computing-in-sensor network using ferroelectric photo sensors with remanent-polarization-controlled photo responsivities.

Ferroelectric tunnel junction (FTJ) has attracted considerable attention for its potential applications in nonvolatile memory and neuromorphic computing. However, the experimental exploration of FTJs with high ON/OFF ratios is a challenging task due to the vast search space comprising of ferroelectric and electrode materials, fabrication methods and conditions and so on. Here, machine learning (ML) is demonstrated to be an effective tool to guide the experimental search of FTJs with high ON/OFF ratios. A dataset consisting of 152 FTJ samples with nine features and one target attribute (i.e., ON/OFF ratio) is established for ML modeling. Among various ML models, the gradient boosting classification model achieves the highest prediction accuracy. Combining the feature importance analysis based on this model with the association rule mining, it is extracted that the utilizations of {graphene/graphite (Gra) (top), LaNiO3 (LNO) (bottom)} and {Gra (top), Ca[Formula: see text]Ce[Formula: see text]MnO3 (CCMO) (bottom)} electrode pairs are likely to result in high ON/OFF ratios in FTJs. Moreover, two previously unexplored FTJs: Gra/BaTiO3 (BTO)/LNO and Gra/BTO/CCMO, are predicted to achieve ON/OFF ratios higher than 1000. Guided by the ML predictions, the Gra/BTO/LNO and Gra/BTO/CCMO FTJs are experimentally fabricated, which unsurprisingly exhibit [Formula: see text]1000 ON/OFF ratios ([Formula: see text]8540 and [Formula: see text]7890, respectively). This study demonstrates a new paradigm of developing high-performance FTJs by using ML.

Memristor crossbar is extensively investigated as an energy‐efficient accelerator for neural network (NN) computations. However, hardware implementation of NNs using realistic memristors is challenging due to the ubiquity of faults (mainly classified into hard and soft faults) in memristors. Herein, a hardware‐friendly, low‐power multifault‐tolerant training (MFTT) scheme capable of addressing both hard and soft faults simultaneously for memristive NNs is proposed. The MFTT scheme consists of multifault detection, targeted weight pruning, and in situ training with the Manhattan update rule. Specifically, multifault detection is first conducted to detect both hard and large soft faults. The detected faulty weights are subsequently pruned to prevent them from disturbing the NN. The sparsified NN after pruning is in situ trained so that the hard and large soft faults can be effectively tolerated using the sparsity and self‐adaptivity of NNs. In addition, the remaining small soft faults can be well tolerated by the Manhattan update rule. Experimentally, MFTT demonstrates the lowest accuracy losses among several representative fault‐tolerant schemes not only in the hard fault‐only (when the hard fault ratio exceeds 10%) and soft fault‐only (when the soft faults are large) cases, but also in the case where both types of faults coexist.

The aspects of low processing temperature and easy running in oxygen atmosphere contribute to the potential of pulsed laser deposition (PLD) in developing a-IGZO TFTs for flexible applications. However, the realization of low-temperature and high-performance devices with determined strategies requires further exploration. In this work, the effect of oxygen pressure and post-annealing processes and their mechanisms on the performance evolution of a-IGZO TFTs by PLD were systematically studied. A room-temperature a-IGZO TFT with no hysteresis and excellent performances, including a μ of 17.19 cm2/V·s, an Ion/Ioff of 1.7 × 106, and a SS of 403.23 mV/decade, was prepared at the oxygen pressure of 0.5 Pa. Moreover, an O2 annealing atmosphere was confirmed effective for high-quality a-IGZO films deposited at high oxygen pressure (10 Pa), which demonstrates the critical effect of oxygen vacancies, rather than weak bonds, on the device’s performance.

Poly(A) tails are present on almost all eukaryotic mRNAs, and play critical roles in mRNA stability, nuclear export, and translation efficiency. The biosynthesis and shortening of a poly(A) tail are regulated by large multiprotein complexes. However, the molecular mechanisms of these protein machineries still remain unclear. Recent studies regarding the structural and biochemical characteristics of those protein complexes have shed light on the potential mechanisms of polyadenylation and deadenylation. This review summarizes the recent structural studies on pre-mRNA 3′-end processing complexes that initiate the polyadenylation and discusses the similarities and differences between yeast and human machineries. Specifically, we highlight recent biochemical efforts in the reconstitution of the active human canonical pre-mRNA 3′-end processing systems, as well as the roles of RBBP6/Mpe1 in activating the entire machinery. We also describe how poly(A) tails are removed by the PAN2-PAN3 and CCR4-NOT deadenylation complexes and discuss the emerging role of the cytoplasmic poly(A)-binding protein (PABPC) in promoting deadenylation. Together, these recent discoveries show that the dynamic features of these machineries play important roles in regulating polyadenylation and deadenylation.

Even though electrochromism has been around for more than 50 years, it still has several issues. Multi-layered films, high manufacturing costs, and a short lifetime are present in existing electrochromic devices. We demonstrate a unique high-performance device with a basic structure and no solid electrochromic sheets in this work. In this device, the electrolyte layer is also avoided. The device uses an electrochromic solution prepared from a mixture of ammonium metatungstate and iron (II) chloride solution as a functional layer with reversible redox properties. The tungstate ions on the electrode surface are reduced when the device is colored, and the Fe2+ on the electrode surface is oxidized on another electrode surface. The generated Fe3+ in the mixed functional layer oxidizes the previously reduced tungstate ions as the device fades. We determined the ΔT (transmittance modulation) and response time among ammonium metatungstate ratios, iron (II) chloride ratios, and driven current density using DOE (design of experiment) trials. Using 0.175 mol/L ammonium metatungstate and 0.30 mol/L iron (II) chloride, a device with outstanding ΔT (more than 57% at 700 nm), a short response time (less than 10 s), and high coloring efficiency (160.04 cm2/C at 700 nm) is demonstrated.

Abstract We performed a systematic study of the influence of environmental conditions on the electrical performance characteristics of solution-processed 2,7-dioctyl [1] benzothieno[3,2-b][1]-benzothiophene (C8-BTBT) thin-film transistors (TFTs). Four environmental exposure conditions were considered: high vacuum (HV), O2, N2, and air. The devices exposed to O2 and N2 for 2 h performed in a manner similar to that of the device kept in HV. However, the device exposed to air for 2 h exhibited significantly better electrical properties than its counterparts. The average and highest carrier mobility of the 70 air-exposed C8-BTBT TFTs were 4.82 and 8.07 cm2V-1s-1, respectively. This can be compared to 2.76 cm2V-1s-1 and 4.70 cm2V-1s-1, respectively, for the 70 devices kept in HV. Furthermore, device air stability was investigated. The electrical performance of C8-BTBT TFTs degrades after long periods of air exposure. Our work improves knowledge of charge transport behavior and mechanisms in C8-BTBT OTFTs. It also provides ideas that may help to improve device electrical performance further.

Abstract In this work, a low leakage current ZrO2 was fabricated for sputter indium gallium zinc oxide (IGZO) thin-film transistor using direct inkjet-printing technology. Spin-coated and direct inkjet-printed ZrO2 were prepared to investigate the film formation process and electrical performance for different process. Homogeneous ZrO2 films were observed through the high-resolution TEM images. The chemical structure of ZrO2 films were investigated by XPS measurements. The inkjet-printed ZrO2 layer upon IGZO showed a superior performance on mobility and off state current, but a large V th shift under positive bias stress. As a result, the TFT device based on inkjet-printed ZrO2 exhibited a saturation mobility of 12.4 cm2/Vs, an Ion/Ioff ratio of 106, a turn on voltage of 0 V and a 1.4-V V th shift after 1-h PBS strain. Higher density films with less oxygen vacancy were responsible for low off state current for the printed ZrO2 device. The mechanism of deteriorated performance on PBS test can be ascribed to the In-rich region formed at the back channel which easily absorbs H2O and oxygen. The absorbed H2O and oxygen capture electrons under positive bias stress, serving as acceptors in TFT device. This work demonstrates the film formation process of direct inkjet-printed and spin-coated oxide films and reveals the potential of direct inkjet-printed oxide dielectric in high-performance oxide TFT device.

Solution-processed oxide dielectrics are widely studied as alternatives to SiO2, SiNx in thin film transistors for high capacitance and low energy consuming. However, it's still a challenge to achieve high quality of solution-processed oxide dielectrics TFT by low-temperature post-treatment. Here, an effective strategy is proposed to evaluate low temperature solution-processed ZrO2 TFT in terms of uniformity, film density and electrical performance by employing TG/DSC, drop analyzer, XRR and a novel μ-PCD measurement. Particularly, μ-PCD measurement provides valuable information on homogeneity and defect level. The optimized device with ZrNO3 as dielectric precursor and ethyl alcohol as solvent through spin coating process annealed at 200°C shows a saturation mobility of 8.6 cm2V-1S-1, Ion/Ioff ratio about 1.1×106, subthreshold swing about 334 mV/decade with a low leakage current density of 5×10-5 A/cm2 at 10V. This article offers a convenient way for precursor optimization towards low leakage current and high homogeneity solution-processed oxide dielectrics for TFT devices.

Summary: Ferroelectric synapses using polarization switching (a purely electronic switching process) to induce analog conductance change have attracted considerable interest. Here, we propose ferroelectric photovoltaic (FePV) synapses that use polarization-controlled photocurrent as the readout and thus have no limitations on the forms and thicknesses of the constituent ferroelectric and electrode materials. This not only makes FePV synapses easy to fabricate but also reduces the depolarization effect and hence enhances the polarization controllability. As a proof-of-concept implementation, a Pt/Pb(Zr0.2Ti0.8)O3/LaNiO3 FePV synapse is facilely grown on a silicon substrate, which demonstrates continuous photovoltaic response modulation with good controllability (small nonlinearity and write noise) enabled by gradual polarization switching. Using photovoltaic response as synaptic weight, this device exhibits versatile synaptic functions including long-term potentiation/depression and spike-timing-dependent plasticity. A simulated FePV synapse-based neural network achieves high accuracies (>93%) for image recognition. This study paves a new way toward highly controllable and silicon-compatible synapses for neuromorphic computing.

Abstract Write‐once‐read‐many (WORM) memory behavior is often observed in polymer electret memory (PEM) devices, greatly limiting their overall performance. This paper systematically investigates the device physics of PEM devices with poly(α‐methylstyrene) as a charge trapping layer and pentacene as a semiconductor channel. The combined experiments on transistors, capacitances, and optical spectroscopy reveal that both the WORM memory behavior after negative and positive pulses and the gradual formation of memory after the continuous scanning are the results of the deficiency in minority (electrons) transport and trapping. Corresponding quantitative models are established and well explain the two‐stage, gradual trapping processes to form memory. By reducing the structural disorder and lateral channel length, ambipolar, bistable memory and much faster formation of memory window is obtained based on the same PEM device. The insights into device physics of PEM devices are expected to facilitate the design of organic, nonvolatile memory devices with high programming and erasing efficiencies.

Abstract In this work, high-k composite TiAlO film has been investigated as charge-trapping material for nonvolatile memory applications. The annealing formed Al2O3-TiAlO-SiO2 dielectric stack demonstrates significant memory effects and excellent reliability properties. The memory device exhibits a large memory window of ~2.6 V under ±8 V sweeping voltage, and it shows only ~14% charge loss after more than 10 years’ retention, indicating excellent charge retention properties. The electronic structures of the Al2O3-TiAlO-SiO2 have been studied by X-ray photoelectron spectroscopy measurements, and it reveals that the quantum well and the defect traps in TiAlO film can provide a >1.8 eV deep barrier for charge confinement in the TiAlO layer. The mixing between Al2O3 and TiO2 can increase the defects related to the under-coordinated Ti3+ atoms, thereby enhancing the charge-trapping efficiency of the device. Our work implies that high-k TiAlO composite film is promising for applications in future nonvolatile charge-trapping memories.

In photovoltaics window electrodes must display uniform current transport, as well as high light transmission from the substrate. Here, Han et al.show that quasi-fractal metallic networks provide a practical realization of an electrode structure with an optimal surface coverage and a uniform current density.

Aluminum-doped ZnO (AZO) has huge prospects in the field of conductive electrodes, due to its low price, high transparency, and pro-environment. However, enhancing the conductivity of AZO and realizing ohmic contact between the semiconductor and AZO source/drain (S/D) electrodes without thermal annealing remains a challenge. Here, an approach called pulsed laser deposition (PLD) is reported to improve the comprehensive quality of AZO films due to the high energy of the laser and non-existence of the ion damage. The 80-nm-thick AZO S/D electrodes show remarkable optical properties (transparency: 90.43%, optical band gap: 3.42 eV), good electrical properties (resistivity: 16 × 10−4 Ω·cm, hall mobility: 3.47 cm2/V·s, carrier concentration: 9.77 × 1020 cm−3), and superior surface roughness (Rq = 1.15 nm with scanning area of 5 × 5 μm2). More significantly, their corresponding thin film transistor (TFT) with low contact resistance (RSD = 0.3 MΩ) exhibits excellent performance with a saturation mobility (µsat) of 8.59 cm2/V·s, an Ion/Ioff ratio of 4.13 × 106, a subthreshold swing (SS) of 0.435 V/decade, as well as good stability under PBS/NBS. Furthermore, the average transparency of the unpatterned multi-films composing this transparent TFT can reach 78.5%. The fabrication of this TFT can be suitably transferred to transparent arrays or flexible substrates, which is in line with the trend of display development.

In this paper, a high-k metal-oxide film (ZrO2) was successfully prepared by a solution-phase method, and whose physical properties were measured by X-ray diffraction (XRD), X-ray reflectivity (XRR) and atomic force microscopy (AFM). Furthermore, indium–gallium–zinc oxide thin-film transistors (IGZO-TFTs) with high-k ZrO2 dielectric layers were demonstrated, and the electrical performance and bias stability were investigated in detail. By spin-coating 0.3 M precursor six times, a dense ZrO2 film, with smoother surface and fewer defects, was fabricated. The TFT devices with optimal ZrO2 dielectric exhibit a saturation mobility up to 12.7 cm2 V−1 s−1, and an on/off ratio as high as 7.6 × 105. The offset of the threshold voltage was less than 0.6 V under positive and negative bias stress for 3600 s.

In this work, we fabricated a high-mobility amorphous indium-gallium-zinc-oxide (a-IGZO) thin-film transistor (TFT) based on alumina oxide (Al 2 O 3 ) passivation layer (PVL) and copper (Cu) source/drain electrodes (S/D). The mechanism of the high mobility for a-IGZO TFT was proposed and experimentally demonstrated. The conductivity of the channel layer was significantly improved due to the formation of metallic In nanoparticles on the back channel during Al 2 O 3 PVL sputtering. In addition, Ar atmosphere annealing induced the Schottky contact formation between the Cu S/D and the channel layer caused by Cu diffusion. In conjunction with high conductivity channel and Schottky contact, the a-IGZO TFT based on Cu S/D and Al 2 O 3 PVL exhibited remarkable mobility of 33.5–220.1 cm 2 /Vs when channel length varies from 60 to 560 μ m. This work presents a feasible way to implement high mobility and Cu electrodes in a-IGZO TFT, simultaneously.

In this paper, an alumina-zirconia (Al2O3-ZrO2) nanolaminate dielectric was fabricated by spin-coating and the performance was investigated. It was found that the properties of the dielectric can be adjusted by changing the content of Al2O3/ZrO2 in nanolaminates: when the content of Al2O3 was higher than 50%, the properties of nanolaminates, such as the optical energy gap, dielectric strength (Vds), capacitance density, and relative permittivity were relatively stable, while the change of these properties became larger when the content of Al2O3 was less than 50%. With the content of ZrO2 varying from 50% to 100%, the variation of these properties was up to 0.482 eV, 2.12 MV/cm, 135.35 nF/cm2, and 11.64, respectively. Furthermore, it was demonstrated that the dielectric strength of nanolaminates were influenced significantly by the number (n) of bilayers. Every increment of one Al2O3-ZrO2 bilayer will enhance the dielectric strength by around 0.39 MV/cm (Vds ≈ 0.86 + 0.39n). This could be contributed to the amorphous alumina which interrupted the grain boundaries of zirconia.

The characteristics of Cu alloy (0.3 wt. % Cr, 0.2 wt. % Zr) thin film deposited by direct current (DC) magnetron sputtering deposition were investigated. The conductivity and adhesion of the Cu-0.3%Cr-0.2%Zr films were optimized by increasing the sputter power to 150 W and reducing the sputter pressure to 2 mTorr. With an annealing process (at 300 °C for 1 h in argon ambient atmosphere), the resistivity of the alloy film decreased from 4.80 to 2.96 μΩ·cm, and the adhesion classification increased from 2B to 4B on glass substrate. X-ray photoelectron spectroscopy (XPS) analysis showed that Cr aggregated toward the surface of the film and formed a self-protection layer in the annealing process. Transmission electron microscopy (TEM) indicated the aggregation and migration of Cr in the annealing process. A further X-ray diffraction (XRD) analysis showed that Cu2O appeared when the annealing temperature reached above 350 °C, which accounts for the increase of the resistivity. Based on Al2O3 and SiO2 substrate surfaces, the Cu-0.3%Cr-0.2%Zr film also showed high conductivity and adhesion, which has a potential in the application of Cu gate electrodes for thin film transistor (TFT).

Solution-processed high-k dielectric TFTs attract much attention since they cost relatively little and have a simple fabrication process. However, it is still a challenge to reduce the leakage of the current density of solution-processed dielectric TFTs. Here, a simple solution method is presented towards enhanced performance of ZrO2 films by intentionally increasing the concentration of precursor. The ZrO2 films not only exhibit a low leakage current density of 10−6 A/cm2 at 10 V and a breakdown field of 2.5 MV/cm, but also demonstrate a saturation mobility of 12.6 cm2·V−1·s−1 and a Ion/Ioff ratio of 106 in DC pulse sputtering IGZO-TFTs based on these films. Moreover, the underlying mechanism of influence of precursor concentration on film formation is presented. Higher concentration precursor results in a thicker film within same coating times with reduced ZrO2/IGZO interface defects and roughness. It shows the importance of thickness, roughness, and annealing temperature in solution-processed dielectric oxide TFT and provides an approach to precisely control solution-processed oxide films thickness.