Jihang Lee

Tuning Resistive Switching Characteristics of Tantalum Oxide Memristors through Si Doping

An oxide memristor device changes its internal state according to the history of the applied voltage and current. The principle of resistive switching (RS) is based on ion transport (e.g., oxygen vacancy redistribution). To date, devices with bi-, triple-, or even quadruple-layered structures have been studied to achieve the desired switching behavior through device structure optimization. In contrast, the device performance can also be tuned through fundamental atomic-level design of the switching materials, which can directly affect the dynamic transport of ions and lead to optimized switching characteristics. Here, we show that doping tantalum oxide memristors with silicon atoms can facilitate oxygen vacancy formation and transport in the switching layer with adjustable ion hopping distance and drift velocity. The devices show larger dynamic ranges with easier access to the intermediate states while maintaining the extremely high cycling endurance (>10(10) set and reset) and are well-suited for neuromorphic computing applications. As an example, we demonstrate different flavors of spike-timing-dependent plasticity in this memristor system. We further provide a characterization methodology to quantitatively estimate the effective hopping distance of the oxygen vacancies. The experimental results are confirmed through detailed ab initio calculations which reveal the roles of dopants and provide design methodology for further optimization of the RS behavior.

Oxide Resistive Memory with Functionalized Graphene as Built-in Selector Element

DOI: 10.1002/adma.201400270 transfer was repeated several times (typically ∼3) in order to further improve the conductance of the electrodes, therefore forming a multilayer graphene (MLG) stack. An optical graph of the glass substrate with the MLG fi lm on top is displayed in Figure 1 b, showing high transparency. Subsequently, the MLG fi lm was patterned into bottom electrodes (BEs) with widths of 1–5 μm by photolithography and O 2 plasma etching (Figure 1 a2). Au/Ti metal contacts (B1 and B2) were then deposited onto the two ends of the MLG BE afterwards to minimize the contact resistance to the external probes in electrical measurements (Figure 1 a3). Following BE defi nition, a bilayer oxide structure consisting of an oxygen rich Ta 2 O 5– x layer (∼5 nm) and an oxygen defi cient TaO y layer (∼40 nm) serving as the switching medium of the RRAM devices was successively deposited by radio-frequency (RF) sputtering and reactive sputtering (400 °C, O 2 /Ar = 3%), respectively, without breaking the vacuum (Figure 1 a4). Similar to steps 1–3 in Figure 1 a, multiple monolayer graphene transfer, O 2 plasma etching and Au/Ti metal contacts deposition were conducted again to fabricate the top electrodes (TEs) that complete the crossbar memory structure (Figure 1 a5–7). Finally, a pad opening step was performed through a timed reactive ion etching (RIE) process to remove the Ta 2 O 5– x /TaO y bilayer on top of the bottom contact Au/Ti pads (Figure 1 a8). The sizes of the RRAM devices in this work range from 1–5 μm × 1–5 μm, and during measurements the voltage was applied on the TE with the BE grounded. Figure 1 c shows an optical graph of the as-fabricated chip. Compared with Figure 1 b, the transparency decrease is likely due to the relatively high concentration of metallic components in the TaO y base layer of the switching medium. Since resistive switching in Ta 2 O 5– x /TaO y bilayer devices is driven by internal oxygen vacancy redistribution, the change in deposition sequence should in principle only lead to a reversal of the switching polarity, as has been verifi ed by studies on devices with inert metal electrodes. [ 25 ] However, in the case of devices with MLG electrodes, the stacking sequence of the bilayer was found to be critical in determining the device behavior, as shown in Figure 1 d,e. When the Ta 2 O 5– x layer was deposited fi rst on the graphene BE followed by TaO y deposition, the MLG/TaO y /Ta 2 O 5– x /MLG (top to bottom) devices exhibited conventional bipolar resistive switching characteristics similar to the results obtained with metal electrodes, [ 25,26 ] as shown in Figure 1 d. Such switching behavior can be well understood in the picture of V O exchange between the Ta 2 O 5– x and the V O -rich TaO y layers, [ 8,26,27 ] and the switching polarity with positive SET and negative RESET voltages is a natural result of the device confi guration since the V O reservoir (the oxygen defi cient TaO y layer) sits on top in this case. The linear on-state behavior is also consistent with Ohmic conduction for the V O -based conducting fi laments in Ta 2 O 5– x /TaO y RRAMs. [ 8,26,27 ] Driven by the continuing demand for improved computing capability, the semiconductor industry has been constantly looking for a fast, reliable, scalable yet nonvolatile memory technology. Nanoscale resistive switching devices (RRAMs) have recently generated signifi cant interest as a potential building block for novel data storage and computing applications [ 1–7 ] and have shown exciting performance metrics including endurance of >10 12 , [ 8 ] programming time of <1 ns, [ 9 ] scalability better than 10 nm [ 10 ] and retention of >10 years. [ 8 ] In addition, since RRAM devices do not rely on conventional Si-based transistors, they can potentially be fabricated in three-dimensional (3D) multistack fashion or on diverse substrates for fl exible and/or transparent electronics applications. [ 11,12 ] This is particularly suitable for oxide-based valency-change type devices, where the resistance change is caused by the redistribution of oxygen vacancies inside the switching layer so potentially any inert metal can be used as the electrode material. [ 13 ] To this end, instead of using conventional metal electrodes RRAM devices can be integrated with novel electrode materials such as graphene to take advantage of the excellent electrical, mechanical and optical properties of graphene. [ 14,15 ] Furthermore, graphene can be readily functionalized or modifi ed to exhibit a number of interesting electrical characteristics, [ 16–18 ] and resistive switching effects in some graphene based materials, for example graphene oxide, have also been reported. [ 19–21 ] In this study, we show that RRAM devices can be successfully integrated with graphene electrodes in a compact device structure. More importantly, functionalized graphene electrodes exhibit an intrinsic threshold switching behavior analogous to insulator-metal transition (IMT), and provide a built-in selector element for the RRAM devices that is very desirable for mitigating the sneak path problem in crossbar memory arrays. [ 22,23 ]

Iodine Vacancy Redistribution in Organic–Inorganic Halide Perovskite Films and Resistive Switching Effects

Organic-inorganic halide perovskite (OHP) materials, for example, CH3 NH3 PbI3 (MAPbI3 ), have attracted significant interest for applications such as solar cells, photodectors, light-emitting diodes, and lasers. Previous studies have shown that charged defects can migrate in perovskites under an electric field and/or light illumination, potentially preventing these devices from practical applications. Understanding and control of the defect generation and movement will not only lead to more stable devices but also new device concepts. Here, it is shown that the formation/annihilation of iodine vacancies (VI s) in MAPbI3 films, driven by electric fields and light illumination, can induce pronounced resistive switching effects. Due to a low diffusion energy barrier (≈0.17 eV), the VI s can readily drift under an electric field, and spontaneously diffuse with a concentration gradient. It is shown that the VI diffusion process can be suppressed by controlling the affinity of the contact electrode material to I- ions, or by light illumination. An electrical-write and optical-erase memory element is further demonstrated by coupling ion migration with electric fields and light illumination. These results provide guidance toward improved stability and performance of perovskite-based optoelectronic systems, and can lead to the development of solid-state devices that couple ionics, electronics, and optics.

Experimental Demonstration of Feature Extraction and Dimensionality Reduction Using Memristor Networks

Memristors have been considered as a leading candidate for a number of critical applications ranging from nonvolatile memory to non-Von Neumann computing systems. Feature extraction, which aims to transform input data from a high-dimensional space to a space with fewer dimensions, is an important technique widely used in machine learning and pattern recognition applications. Here, we experimentally demonstrate that memristor arrays can be used to perform principal component analysis, one of the most commonly used feature extraction techniques, through online, unsupervised learning. Using Sangers rule, that is, the generalized Hebbian algorithm, the principal components were obtained as the memristor conductances in the network after training. The network was then used to analyze sensory data from a standard breast cancer screening database with high classification success rate (97.1%).

Retention failure analysis of metal-oxide based resistive memory

Resistive switching devices (RRAMs) have been proposed a promising candidate for future memory and neuromorphic applications. Central to the successful application of these emerging devices is the understanding of the resistance switching and failure mechanism, and identification of key physical parameters that will enable continued device optimization. In this study, we report detailed retention analysis of a TaOx based RRAM at high temperatures and the development of a microscopic oxygen diffusion model that fully explains the experimental results and can be used to guide future device developments. The device conductance in low resistance state (LRS) was constantly monitored at several elevated temperatures (above 300 °C), and an initial gradual conductivity drift followed by a sudden conductance drop were observed during retention failure. These observations were explained by a microscopic model based on oxygen vacancy diffusion, which quantitatively explains both the initial gradual conductance drift and the sudden conductance drop. Additionally, a non-monotonic conductance change, with an initial conductance increase followed by the gradual conductance decay over time, was observed experimentally and explained within the same model framework. Specifically, our analysis shows that important microscopic physical parameters such as the activation energy for oxygen vacancy migration can be directly calculated from the failure time versus temperature relationship. Results from the analytical model were further supported by detailed numerical multi-physics simulation, which confirms the filamentary nature of the conduction path in LRS and the importance of oxygen vacancy diffusion in device reliability. Finally, these high-temperature stability measurements also reveal the existence of multiple filaments in the same device.

Electronic properties of tantalum pentoxide polymorphs from first-principles calculations

Tantalum pentoxide (Ta2O5) is extensively studied for its attractive properties in dielectric films, anti-reflection coatings, and resistive switching memory. Although various crystalline structures of tantalum pentoxide have been reported, its structural, electronic, and optical properties still remain a subject of research. We investigate the electronic and optical properties of crystalline and amorphous Ta2O5 structures using first-principles calculations based on density functional theory and the GW method. The calculated band gaps of the crystalline structures are too small to explain the experimental measurements, but the amorphous structure exhibits a strong exciton binding energy and an optical band gap (∼4 eV) in agreement with experiment. We determine the atomic orbitals that constitute the conduction band for each polymorph and analyze the dependence of the band gap on the atomic geometry. Our results establish the connection between the underlying structure and the electronic and optical prope...

On‐Demand Reconfiguration of Nanomaterials: When Electronics Meets Ionics

Rapid advances in the semiconductor industry, driven largely by device scaling, are now approaching fundamental physical limits and face severe power, performance, and cost constraints. Multifunctional materials and devices may lead to a paradigm shift toward new, intelligent, and efficient computing systems, and are being extensively studied. Herein examines how, by controlling the internal ion distribution in a solid-state film, a materials chemical composition and physical properties can be reversibly reconfigured using an applied electric field, at room temperature and after device fabrication. Reconfigurability is observed in a wide range of materials, including commonly used dielectric films, and has led to the development of new device concepts such as resistive random-access memory. Physical reconfigurability further allows memory and logic operations to be merged in the same device for efficient in-memory computing and neuromorphic computing systems. By directly changing the chemical composition of the material, coupled electrical, optical, and magnetic effects can also be obtained. A survey of recent fundamental material and device studies that reveal the dynamic ionic processes is included, along with discussions on systematic modeling efforts, device and material challenges, and future research directions.

Resonant coherent Bragg rod analysis of strained epitaxial heterostructures

The resonant response of the complex x-ray scattering factor has been used in conjunction with the coherent Bragg rod analysis phase-retrieval algorithm to determine the composition and strain profiles of ultrathin layers of GaAs grown on InGaAs buffers. The buffer layers are nominally latticed matched with the InP substrate and the subsequent GaAs growth is compared at two different temperatures: 480 and 520°C. We show that electron density maps extracted from Bragg rod scans measured close to the Ga and As K-edges can be used to deconvolute roughness and intermixing. It is found that indium incorporation and roughening lead to a significant reduction of the strain in this system.

Focused-ion-beam-directed nucleation of InAs quantum dots

GaAs buffer layers were patterned with Ga+ ions via a focused ion beam and then overgrown with InAs. Atomic force microscopy reveals a strong influence of the ion dose upon subsequent formation of InAs quantum dots. Uniformly dosed areas show an apparent reduction in the critical thickness for quantum dot formation and the area density of the dots increases with increasing ion dose, which is related to ion beam induced roughening of the surface.

K-means Data Clustering with Memristor Networks

Memristor-based neuromorphic networks have been actively studied as a promising candidate to overcome the von-Neumann bottleneck in future computing applications. Several recent studies have demonstrated memristor networks capability to perform supervised as well as unsupervised learning, where features inherent in the input are identified and analyzed by comparing with features stored in the memristor network. However, even though in some cases the stored feature vectors can be normalized so that the winning neurons can be directly found by the (input) vector-(stored) vector dot-products, in many other cases, normalization of the feature vectors is not trivial or practically feasible, and calculation of the actual Euclidean distance between the input vector and the stored vector is required. Here we report experimental implementation of memristor crossbar hardware systems that can allow direct comparison of the Euclidean distances without normalizing the weights. The experimental system enables unsupervised K-means clustering algorithm through online learning, and produces high classification accuracy (93.3%) for the standard IRIS data set. The approaches and devices can be used in other unsupervised learning systems, and significantly broaden the range of problems a memristor-based network can solve.