Search

The reversible transition between the on and off states of threshold switches under a constant pulse generates voltage oscillation, which can be exploited for compact neuron element in neuromorphic systems. Because the transition voltages play an important role in the oscillation behavior, area and thickness scaling analysis of NbOx-based devices is performed to identify the underlying mechanism. The threshold voltage ( $V_{th}$ ) is sensitive to the device area, indicating that the on state of the device is achieved by the local formation of a conductive phase. On the other hand, the area-independent hold voltage ( $V_{\mathrm{ hold}}$ ) becomes smaller due to the higher compliance current and increased temperature ambient, which implies that spontaneous dissolution of the phase is retarded. Through HSPICE simulation, we reveal that the greater difference between $\text{V}_{\mathrm{ hold}}$ and $\text{V}_{\mathrm{ th}}$ enables a high degree of freedom of oscillation frequency modulation.

Resistive random-access memory (RRAM) has been widely studied because of its high information storage density and easy integration, and it has potential to replace traditional memory storage means. However, it is prone to random formations and fracture of conductive filaments, which limits its commercial applications. In this study, magnetron sputtering was used to deposit NbO x films on different substrates modified with different concentrations of Au-nanoparticles (NPs). The as developed Pt-Au min /NbO x /W tip device with the lowest concentration of Au-NPs modified bottom electrode exhibited the best performance; the V set distribution of the device was more concentrated, V reset decreased significantly, the distribution range was narrow, and the resistance distribution of high and low resistance states (HRS and LRS, respectively) was more stable. Modification of the bottom electrode with Au-NPs did not change the ohmic conduction behavior of NbO x RRAM in LRS; however, it changed the resistance switching mechanism from SCLC to P–F effect modified SCLC. The Au-NPs modified bottom electrode acted like a tip, enhancing the local electric field and promoting the formation of conductive filaments near the NPs. This work explains the internal mechanism for improving RRAM properties using a local electric field and provides theoretical guidance for enabling the rapid commercialization of RRAM. • NbO x films were deposited on substrates-modified with Au-nanoparticles (NPs) using magnetron sputtering. • Pt-Au min /NbO x /W tip device with the lowest concentration of Au-NPs exhibited the best performance. • The bottom electrode decorated with Au NPs enhanced the local electric field and promoted the formation of conductive filaments. • Modification of the bottom electrode with Au-NPs changed the RS mechanism from SCLC to P–F effect modified SCLC. [ABSTRACT FROM AUTHOR]

The spike-response model (SRM) describes the adaptive behaviors of a biological neuron in response to repeated or prolonged stimulation, so that SRM neurons can avoid information overload and support neural networks for competitive learning. In this work, an artificial SRM neuron with the leaky integrate-and-fire (LIF) functions and the adaptive threshold is firstly implemented by the volatile memristive device of Pt/NbOx/TiN. By modulating the volatile speed of the device, the threshold of the SRM neuron is adjusted to achieve the adaptive behaviors, such as the refractory period and the lateral inhibition. To demonstrate the function of the SRM neuron, a spiking neural network (SNN) is constructed with the SRM neurons and trained by the unsupervised learning rule, which successfully classifies letters with noises, while a similar SNN with LIF neurons fails. This work demonstrates that the SRM neuron not only emulates the adaptive behaviors of a biological neuron, but also enriches the functionality and unleashes the computational power of SNNs.

Valence change memory (VCM) attracts numerous attention in memory applications, due to its high stability and low energy consumption. However, owing to the low on/off ratio of VCM, increasing the difficulty of information identification hinders the development of memory applications. We prepared N-doped NbOx:N films (thickness = approximately 15 nm) by pulsed laser deposition at 200 °C. N-doping significantly improved the on/off ratio, retention time, and stability of the Pt/NbOx:N/Pt devices, thus improving the stability of data storage. The Pt/NbOx:N/Pt devices also achieved lower and centralized switching voltage distribution. The improved performance was mainly attributed to the formation of oxygen vacancy (VO) + 2N clusters, which greatly reduced the ionic conductivity and total energy of the system, thus increasing the on/off ratio and stability. Moreover, because of the presence of Vo + 2N clusters, the conductive filaments grew in more localized directions, which led to a concentrated distribution of SET and RESET voltages. Thus, in situ N-doping is a novel and effective approach to optimize device performances for better information storage and logic circuit applications. [ABSTRACT FROM AUTHOR]

To achieve the highest possible integration storage density in the V-point structure, the working current of the selector in the one-selection one-resistance (1S1R) structure should match with the resistance random access memory (RRAM). In this study, a selector device is designed with a Ti/NbOx/Ti/Pt structure through the magnetron sputtering method and achieves excellent performance of threshold switching under ultra-large compliance current (CC) up to 100 mA. Furthermore, both the switching voltages and the OFF-state resistance of the device demonstrate excellent stability even when CC is increased to a milliampere level, attributed from the existence of metallic NbO in the switching layer. This study provides evidence that a Ti/NbOx/Ti/Pt device has a great potential to drive RRAM in the V-point structure.

Valence change memory (VCM) attracts numerous attention in memory applications, due to its high stability and low energy consumption. However, owing to the low on/off ratio of VCM, increasing the difficulty of information identification hinders the development of memory applications. We prepared N-doped NbOx:N films (thickness = approximately 15 nm) by pulsed laser deposition at 200 °C. N-doping significantly improved the on/off ratio, retention time, and stability of the Pt/NbOx:N/Pt devices, thus improving the stability of data storage. The Pt/NbOx:N/Pt devices also achieved lower and centralized switching voltage distribution. The improved performance was mainly attributed to the formation of oxygen vacancy (VO) + 2N clusters, which greatly reduced the ionic conductivity and total energy of the system, thus increasing the on/off ratio and stability. Moreover, because of the presence of Vo + 2N clusters, the conductive filaments grew in more localized directions, which led to a concentrated distribution of SET and RESET voltages. Thus, in situ N-doping is a novel and effective approach to optimize device performances for better information storage and logic circuit applications.

The research on anti-reflection (AR) optical thin film has long sought to obtain high-performance reflection and transmission properties in photovoltaic and photonic devices. The study of multi-layer AR (M-AR) film with low- and high-refractive-index materials is essential to increase the selective transmittance and reflectance at visible light wavelengths. However, M-AR film exhibits low substrate adhesion and slow deposition rates. We developed a DC pulse sputter system incorporating an inductively coupled plasma (ICP) source of high density to obtain high-quality M-AR film. Six-layer AR optical thin film was simulated using SiOx as a low-refractive-index material and NbOx as a high-refractive-index material. The multi-layer AR film based on SiOx and NbOx (M-SiNb) was fabricated using DC pulse sputtering which incorporated an ICP source. M-SiNb film exhibited better properties than the optical simulation results at 550 nm (transmittance: 99.19%, reflectance: 0.87%). Similarly, the M-SiNb film fabricated using the ICP source had high transmittance and reflectance in the visible light region and excellent adhesion to the substrate notwithstanding the various mechanical tests it was subjected to. Consequently, the development of the DC pulse sputter system included the ICP source, and this study represents important research in the field of optical film.

In this letter, we demonstrate a new concept of threshold selector by combining Ag-based selector with NbOx insulator metal transition (IMT) selector. To overcome intrinsic limits of slow turn-off speed ( $> 1\mu \text{s}$ ) of the Ag-based selector, we adopted NbOx IMT selector with fast turn-off speed. The hybrid device exhibits excellent selector characteristics, such as extremely low off-current (<1 pA) and fast turn-off speed ( $< 1~\mu \text{s}$ ). It shows excellent read-out margin (>75% at >210 word lines) in a fast sequential operation of cross point memory array. [ABSTRACT FROM AUTHOR]

Recent years have witnessed dramatic upsurge of artificial intelligence and neural networks in various application areas. The classification accuracy of neural networks is closely related to the number of training samples, the complexity of the neural network structure, and the number of training epochs. To achieve high performance of the neural network, a large number of training samples are usually required. However, since the number of training samples available can be limited in many real scenes, significant overfitting issue commonly arises and seriously deteriorates the training performance of neural networks. Here, we show a dropout neuronal unit based on NbO x volatile memristor, which can enable dropout function in the neural network constructed. The proposed neural network with such dropout neuron is able to suppress the overfitting problem effectively in the training process, and the role of dropout probability is investigated based on MNIST and FashionMNIST datasets. The results clearly imply the high potential of the dropout neuronal unit in building highly efficient neuromorphic computing systems. [Display omitted] • The overfitting phenomenon is a serious problem of neural networks. • Dropout techniques in software are often used to solve overfitting problems. • A dropout neuronal unit based on NbO x volatile memristor is presented. • Neural networks with such dropout neuron can suppress overfitting efficiently. [ABSTRACT FROM AUTHOR]

Abstract: In this work, ultrathin layers of GdOx, NbOx and GdNbOx, deposited with ACSD have been investigated. Because of the high temperature anneals utilized in the process flow of electronic devices, interactions of the deposited high-k materials and the substrate are analyzed. The deposited layers of GdOx, NbOx and GdNbOx on SiO2 and Al2O3 are annealed in an oxidizing and inert atmosphere and studied by XRD, GATR–FTIR and ellipsometry to assess layer-substrate interactions and crystallization behavior. With temperatures up to 900°C, some minor interactions with the SiO2/Si substrates and no interaction with the Al2O3 substrate were observed. At 1000°C, however, more intense interactions with both substrates in both ambients are observed (severe silicate formation, interlayer regrowth and interaction with Al2O3). HT–XRD on the deposited layers shows that all the layers crystallized well below 900°C. It is concluded that GdNbOx could be a more advantageous high-k material, compared to its monometal counterparts, based on its limited interaction with the substrate at high temperatures and in oxidative ambient. [Copyright &y& Elsevier]

To address the problem of sneak currents in large-scale crossbar arrays without extra access devices like transistors or selectors, memristive devices with self-rectifying characteristics are highly desired. Herein, memristive devices of Pt/Cu-NbO x /TiN with a large rectification ratio (~1.43 × 105 at ±2 V) are realized in Cu-doped NbO x thin films. The large rectification ratio is due to the large Schottky barrier height at the Pt/Cu-doped NbO x interface, which is closely related to the changes of the band structure through Cu doping. More importantly, the devices can facilitate the fabrication of memristive crossbar arrays whose scale can reach maximal 486 × 486 (~230 kbits) with 10% read margin. This work clearly proves the feasibility of the self-rectifying memristive devices for bio-inspired computing. • Self-rectifying memristive devices of Pt/Cu-NbO x /TiN are fabricated. • Pt/Cu-NbO x /TiN devices perform a high rectification ratio of ~1.43 × 105. • The high self-rectifying behavior can be ascribed to the changes of the band structure by the Cu doping in NbO x. • A large crossbar array of 486 × 486 (~230 Kbits) can be fabricated by the self-rectifying memristive devices. [ABSTRACT FROM AUTHOR]

The research on anti-reflection (AR) optical thin film has long sought to obtain high-performance reflection and transmission properties in photovoltaic and photonic devices. The study of multi-layer AR (M-AR) film with low- and high-refractive-index materials is essential to increase the selective transmittance and reflectance at visible light wavelengths. However, M-AR film exhibits low substrate adhesion and slow deposition rates. We developed a DC pulse sputter system incorporating an inductively coupled plasma (ICP) source of high density to obtain high-quality M-AR film. Six-layer AR optical thin film was simulated using SiOx as a low-refractive-index material and NbOx as a high-refractive-index material. The multi-layer AR film based on SiOx and NbOx (M-SiNb) was fabricated using DC pulse sputtering which incorporated an ICP source. M-SiNb film exhibited better properties than the optical simulation results at 550 nm (transmittance: 99.19%, reflectance: 0.87%). Similarly, the M-SiNb film fabricated using the ICP source had high transmittance and reflectance in the visible light region and excellent adhesion to the substrate notwithstanding the various mechanical tests it was subjected to. Consequently, the development of the DC pulse sputter system included the ICP source, and this study represents important research in the field of optical film. [ABSTRACT FROM AUTHOR]

Further enhancing activity and durability of Pt-based catalysts for oxygen reduction is needed for the automotive application of fuel cells. This is however a very challenging task. Herein, we describe a new method to nail down low-oxidation-state NbO x nanoparticles at high temperature without overgrowth using the surface pores on carbon and then deposit Pt head selectively on top of NbO x derived by the redox reduction between themselves. This new structured catalyst, Pt-NbO x C, exhibited mass activities higher than 0.5 A mg−1 after 50,000 cycles between 0.6 and 1.0 V and 5,000 cycles between 1 and 1.5 V, which are 7 and 4 times of the 0.07 and 0.13 A mg−1, respectively, on Pt/C. That high sustainable ORR activity is ascribed to strong metal-support interaction that maximizes Pt utilization and minimizes particle agglomeration, carbon corrosion and metal oxidation. Furthermore, the method embedded nanoparticles into carbon pores will likely be used in developing small particles with uniform dispersion for various applications. Image 1 • Low-oxidation-state metal oxides nanoparticles are confined by pores on carbon at high temperature without overgrowth. • Pt head is selectively deposited and fully covered on metal oxides by the redox reduction between themselves. • The structured Pt-NbO x C catalyst exhibits enhanced ORR activity and durability concurrently. [ABSTRACT FROM AUTHOR]

Amorphous NbO x electron transport layer (ETL) shows great potential for boosting the power conversion efficiency (PCE) of perovskite solar cells (PSCs) at low temperatures (<100 °C). To date, it is still a challenge to simultaneously control the hydrolysis of the NbO x precursor solution and reduce the impurities of NbO x ETLs during low-temperature solution processing under ambient conditions. Herein, for the first time, we report ozone (O 3 ) as a strong ligand to stabilize Nb salt solutions under ambient conditions. The above procedure not only enables the formation of a highly repeatable amorphous NbO x film by suppressing the hydrolysis of the solution but also reduces the OH content in the film, which decreases the defect intensity and improves the conductivity of the NbO x ETL. Thus, the formation of highly repeatable NbO x ETL-based PSCs are obtained; moreover, these PSCs have high PCE values of 19.54 and 16.42% on rigid and flexible substrates, respectively, much higher than the devices based on ETLs from a solution without an O 3 treatment.