Resistive switching (RS) is an interesting property shown by some materials systems that, especially during the last decade, has gained a lot of interest for the fabrication of electronic devices, with electronic nonvolatile memories being those that have received the most attention. The presence and quality of the RS phenomenon in a materials system can be studied using different prototype cells, performing different experiments, displaying different figures of merit, and developing different computational analyses. Therefore, the real usefulness and impact of the findings presented in each study for the RS technology will be also different. This manuscript describes the most recommendable methodologies for the fabrication, characterization, and simulation of RS devices, as well as the proper methods to display the data obtained. The idea is to help the scientific community to evaluate the real usefulness and impact of an RS study for the development of RS technology.
We present an investigation of structural changes in silicon-rich silicon oxide metal-insulatormetal resistive RAM devices. The observed unipolar switching, which is intrinsic to the bulk oxide material and does not involve movement of metal ions, correlates with changes in the structure of the oxide. We use atomic force microscopy, conductive atomic force microscopy, x-ray photoelectron spectroscopy, and secondary ion mass spectroscopy to examine the structural changes occurring as a result of switching. We confirm that protrusions formed at the surface of samples during switching are bubbles, which are likely to be related to the outdiffusion of oxygen. This supports existing models for valence-change based resistive switching in oxides. In addition, we describe parallel linear and nonlinear conduction pathways and suggest that the conductance quantum, G 0 , is a natural boundary between the high and low resistance states of our devices. V C 2015 AIP Publishing LLC. [http://dx
electroluminescent devices, [ 2 ] electrochromic windows, [ 3 ] and transparent amorphous conductors [ 4 ] remains poorly understood. Abrupt changes of resistance in response to electrical stress are hallmarks of correlated electron and ion dynamics and an obvious manifestation of structural dynamics; such phenomena have been reported in a variety of oxides with a range of stoichiometries in different applications. Materials studied include indium tin oxide, zinc, vanadium, nickel, and titanium oxides, and silicon oxide, all of which have a variable degree of substoichiometry that affects the dynamics of resistance change. Here, we employ a suite of structural characterization techniques, including bias-induced resistance changes, to probe dynamic structural changes in amorphous oxides. Abrupt resistance changes are not necessarily central to device functionality, and in some cases, such as electroluminescence, are detrimental. Nevertheless, they demonstrate an extreme response to electrical stress. Silicon oxide, which we study here, is a technologically important oxide representative of the broader class of metastable amorphous oxides. It is historically one of the most studied materials, and the remarkable dynamics that we uncover are thus all the more surprising.Existing models for defect generation and electrical breakdown in oxides are often restricted to crystalline and stoichiometric materials; amorphous oxides present a formidable modeling challenge. Nevertheless, recent work on resistance switching highlights local structural and chemical changes driven by sub-breakdown electrical stress. Research into resistance changes in silicon oxide dates back to the 1960s and 1970s, when irreversible electrical breakdown was widely studied. [5][6][7][8] More recently, there have been reports of intrinsic reversible (soft) breakdown of silicon oxide, [9][10][11] usually ascribed to the formation of chains of oxygen vacancies [ 12 ] produced by fi elddriven movement of oxygen ions. The reversibility of these changes is of the greatest interest, as it probes the dynamics of oxides under controlled stress and provides a model for the initial stages of irreversible dielectric breakdown. In terms of applications, nonvolatile resistive random access memory (RRAM) [ 1 ] or analogue neuromorphic devices [ 13,14 ] are important technological areas that exploit reversible dynamic changes in oxide local structure. However, they are by no means the only applications relying on electrically stressed amorphous oxides. [2][3][4] The observation of quantized conductance in electrically stressed silicon oxide suggests further applications in quantum technology [ 15 ] while, in other fi elds, studies of electroluminescence from silicon-rich silicon oxide demonstrate that its optical properties depend critically on the sequence of applied voltage stress; over-stressing produces permanent Functional oxides are fundamental to modern microelectronics as high quality insulators, transparent conductors, electroluminescent and electrochro...
Resistive Random Access Memory (RRAM) is a promising technology for power efficient hardware in applications of artificial intelligence (AI) and machine learning (ML) implemented in non-von Neumann architectures. However, there is an unanswered question if the device non-idealities preclude the use of RRAM devices in this potentially disruptive technology. Here we investigate the question for the case of inference. Using experimental results from silicon oxide (SiO x ) RRAM devices, that we use as proxies for physical weights, we demonstrate that acceptable accuracies in classification of handwritten digits (MNIST data set) can be achieved using non-ideal devices. We find that, for this test, the ratio of the high- and low-resistance device states is a crucial determinant of classification accuracy, with ~96.8% accuracy achievable for ratios >3, compared to ~97.3% accuracy achieved with ideal weights. Further, we investigate the effects of a finite number of discrete resistance states, sub-100% device yield, devices stuck at one of the resistance states, current/voltage non-linearities, programming non-linearities and device-to-device variability. Detailed analysis of the effects of the non-idealities will better inform the need for the optimization of particular device properties.
Conductive atomic force microscopy was used to etch through SiOx resistance switching devices to produce three-dimensional renderings of conductive filaments.
Artificial neural networks are notoriously power- and time-consuming when implemented on conventional von Neumann computing systems. Consequently, recent years have seen an emergence of research in machine learning hardware that strives to bring memory and computing closer together. A popular approach is to realise artificial neural networks in hardware by implementing their synaptic weights using memristive devices. However, various device- and system-level non-idealities usually prevent these physical implementations from achieving high inference accuracy. We suggest applying a well-known concept in computer science—committee machines—in the context of memristor-based neural networks. Using simulations and experimental data from three different types of memristive devices, we show that committee machines employing ensemble averaging can successfully increase inference accuracy in physically implemented neural networks that suffer from faulty devices, device-to-device variability, random telegraph noise and line resistance. Importantly, we demonstrate that the accuracy can be improved even without increasing the total number of memristors.
We studied intrinsic resistance switching behaviour in sputter-deposited amorphous silicon suboxide (a-SiOx) films with varying degrees of roughness at the oxide-electrode interface. By combining electrical probing measurements, atomic force microscopy (AFM), and scanning transmission electron microscopy (STEM), we observe that devices with rougher oxide-electrode interfaces exhibit lower electroforming voltages and more reliable switching behaviour. We show that rougher interfaces are consistent with enhanced columnar microstructure in the oxide layer. Our results suggest that columnar microstructure in the oxide will be a key factor to consider for the optimization of future SiOx-based resistance random access memory.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.