We evaluate the sensitivity of neuromorphic inference accelerators based on silicon-oxide-nitride-oxide-silicon (SONOS) charge trap memory arrays to total ionizing dose (TID) effects. Data retention ...statistics were collected for 16 Mbit of 40-nm SONOS digital memory exposed to ionizing radiation from a Co-60 source, showing good retention of the bits up to the maximum dose of 500 krad(Si). Using this data, we formulate a rate-equation-based model for the TID response of trapped charge carriers in the ONO stack and predict the effect of TID on intermediate device states between "program" and "erase." This model is then used to simulate arrays of low-power, analog SONOS devices that store 8-bit neural network weights and support in situ matrix-vector multiplication. We evaluate the accuracy of the irradiated SONOS-based inference accelerator on two image recognition tasks-CIFAR-10 and the challenging ImageNet data set-using state-of-the-art convolutional neural networks, such as ResNet-50. We find that across the data sets and neural networks evaluated, the accelerator tolerates a maximum TID between 10 and 100 krad(Si), with deeper networks being more susceptible to accuracy losses due to TID.
A distributed impedance "field cage" structure is proposed and evaluated for electric field control in GaN-based, lateral high electron mobility transistors operating as kilovolt-range power devices. ...In this structure, a resistive voltage divider is used to control the electric field throughout the active region. The structure complements earlier proposals utilizing floating field plates that did not employ resistively connected elements. Transient results, not previously reported for field plate schemes using either floating or resistively connected field plates, are presented for ramps of dV ds /dt = 100 V/ns. For both dc and transient results, the voltage between the gate and drain is laterally distributed, ensuring that the electric field profile between the gate and drain remains belowthe critical breakdown field as the source-to-drain voltage is increased. Our scheme indicates promise for achieving the breakdown voltage scalability to a few kilovolts.
Biological neural networks continue to inspire new developments in algorithms and microelectronic hardware to solve challenging data processing and classification problems. Here, we survey the ...history of neural-inspired and neuromorphic computing in order to examine the complex and intertwined trajectories of the mathematical theory and hardware developed in this field. Early research focused on adapting existing hardware to emulate the pattern recognition capabilities of living organisms. Contributions from psychologists, mathematicians, engineers, neuroscientists, and other professions were crucial to maturing the field from narrowly-tailored demonstrations to more generalizable systems capable of addressing difficult problem classes such as object detection and speech recognition. Algorithms that leverage fundamental principles found in neuroscience such as hierarchical structure, temporal integration, and robustness to error have been developed, and some of these approaches are achieving world-leading performance on particular data classification tasks. In addition, novel microelectronic hardware is being developed to perform logic and to serve as memory in neuromorphic computing systems with optimized system integration and improved energy efficiency. Key to such advancements was the incorporation of new discoveries in neuroscience research, the transition away from strict structural replication and towards the functional replication of neural systems, and the use of mathematical theory frameworks to guide algorithm and hardware developments.
In article number 2003984, Yiyang Li, A. Alec Talin, and co‐workers design a deterministic nonvolatile resistive memory cell without nanosized filaments. By using the statistical ensemble behavior of ...all point defects within the 3D bulk for information storage, they solve the challenge of stochastic switching that has plagued filament‐based memristors. This provides a compelling artificial synapse for energy‐efficient, neuromorphic computing.
Although understanding filament formation in oxide‐based memristive devices by theory has emerged, there are still fundamental unanswered questions. Importantly, for practical application of thin ...films the material in its amorphous state is to be considered, but mostly lacking so far, and details on sub‐stoichiometry are also scarce. To gain insight into the optical and electronic properties of sub‐stoichiometric amorphous tantalum oxide (TaOx), the electron energy loss spectrum (EELS) of model systems is characterized theoretically and electron transport characteristics are analyzed in detail. Calculated blue‐shifts by increasing sub‐stoichiometry explained the measurements, potentially suggesting estimation of oxygen vacancy concentrations through EEL spectra. Electron transport results based on TaOx material models validated by EELS measurements show that oxygen vacancy filamentary paths are initiated at low bias upon increasing sub‐stoichiometry yet noting an interplay with the local amorphous structure. Contact resistances at interfaces of the TaOx switching layer and a tantalum scavenging layer or titanium nitride electrode are quantified, indicating the possibility for either oxygen vacancy‐ or metal cluster‐based conduction mechanisms at the interface. The computational work, combined with experimental characterization for validation, provides a basis for investigating effects of sub‐stoichiometry on filament formation in TaOx thin film memristive devices.
To explain filament formation in sub‐stoichiometric tantalum oxide (TaOx) thin film memristive devices, theoretical characterization of electron energy loss spectra (EELS) is validated by measurements. Predicted EELS blue‐shifts when increasing sub‐stoichiometry estimate the oxygen vacancy concentration. Electron transport calculations for amorphous TaOx and metal–TaOx–metal stacks of varying sub‐stoichiometry elucidate oxygen vacancy assisted transmission paths of the conduction filaments.
The domain-wall (DW)-magnetic tunnel junction (MTJ) device implements universal Boolean logic in a manner that is naturally compact and cascadable. However, an evaluation of the energy efficiency of ...this emerging technology for standard logic applications is still lacking. In this article, we use a previously developed compact model to construct and benchmark a 32-bit adder entirely from DW-MTJ devices that communicates with DW-MTJ registers. The results of this large-scale design and simulation indicate that while the energy cost of systems driven by spin-transfer torque (STT) DW motion is significantly higher than previously predicted, the same concept using spin-orbit torque (SOT) switching benefits from an improvement in the energy per operation by multiple orders of magnitude, attaining competitive energy values relative to a comparable CMOS subprocessor component. This result clarifies the path toward practical implementations of an all-magnetic processor system.
Abstract
Inspired by biological neuromorphic computing, artificial neural networks based on crossbar arrays of bilayer tantalum oxide memristors have shown to be promising alternatives to ...conventional complementary metal‐oxide‐semiconductor (CMOS) architectures. In order to understand the driving mechanism in these oxide systems, tantalum oxide films are resistively switched by conductive atomic force microscopy (C‐AFM), and subsequently imaged by kelvin probe force microscopy (KPFM) and spatially resolved time‐of‐flight secondary ion mass spectrometry (ToF‐SIMS). These workflows enable induction and analysis of the resistive switching mechanism as well as control over the resistively switched region of the film. In this work it is shown that the resistive switching mechanism is driven by both current and electric field effects. Reversible oxygen motion is enabled by applying low (<1 V) electric fields, while high electric fields generate irreversible breakdown of the material (>1 V). Fully understanding oxygen motion and electrical effects in bilayer oxide memristor systems is a fundamental step toward the adoption of memristors as a neuromorphic computing technology.