Tuning analog resistive switching and plasticity in bilayer transition metal oxide based memristive synapses

Li, Jingxian; Duan, Qingxi; Zhang, Teng; Sun, Xinhao; Cai, Yimao; Li, Lidong; Yang, Yuchao; Huang, Ru

doi:10.1039/c7ra07522g

Cited by 28 publications

(24 citation statements)

References 53 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, these devices emulate biomimetic synaptic functions, such as spike‐timing‐dependent plasticity (STDP), paired‐pulse facilitation (PPF), and so on, and STDP is a characteristic of a more biofaithful devices for spiking neural network . According to the interactions between presynaptic and postsynaptic neurons, synaptic effects can also be strengthened (long‐term potentiation) or weakened (long‐term depression) . Up to date, many memristors just reported the single side‐only modulation .…”

Section: Introductionmentioning

confidence: 99%

Graphene Oxide Quantum Dots Based Memristors with Progressive Conduction Tuning for Artificial Synaptic Learning

Yan

Zhang

Chen

et al. 2018

Adv Funct Materials

256

160

View full text Add to dashboard Cite

Memristors as electronic artificial synapses have attracted increasing attention in neuromorphic computing. Emulation of both "learning" and "forgetting" processes requires a bidirectional progressive adjustment of memristor conductance, which is a challenge for cutting-edge artificial intelligence. In this work, a memristor device with a structure of Ag/Zr 0.5 Hf 0.5 O 2 :graphene oxide quantum dots/Ag is presented with the feature of bidirectional progressive conductance tuning. The conductance of proposed memristor is adjusted through voltage pulse number, amplitude, and width. A series of voltage pulses with an amplitude of 0.6 V and a width of 30 ns is enough to modulate conductance. The impacts of pulses with different parameters on conductance modulation are investigated, and the potential relationship between pulse amplitude and energy is revealed. Furthermore, it is proved that the pulse with low energy can realize the almost linear conductance regulation, which is beneficial to improve the accuracy of pattern recognition. The bidirectional progressive conduction modulation mimics various plastic synapses, such as spike-timing-dependent plasticity and paired-pulse facilitation. This progressive conduction tuning mechanism might be attributed to the coexistence of tunneling effect and extrinsic electrochemical metallization effect. This work provides one way for memristor to attain attractive features such as bidirectional tuning, low-power consumption, and fast speed switching that is in urgent demand for further evolution of neuromorphic chips.

show abstract

Section: Introductionmentioning

confidence: 99%

Graphene Oxide Quantum Dots Based Memristors with Progressive Conduction Tuning for Artificial Synaptic Learning

Yan

Zhang

Chen

et al. 2018

Adv Funct Materials

256

160

View full text Add to dashboard Cite

show abstract

“…In particular, high yield and low variability are typically required to maximize the computing accuracy of offline training memristive systems, [179,231] and the nonlinearity and asymmetry between increasing/decreasing responses limit the maximum possible accuracy in online training networks. At the device level, the major obstacle is immature memristive device technology, which has a serious effect on the yield, stability, and uniformity of device arrays.…”

Section: Wwwadvelectronicmatdementioning

confidence: 99%

Memristive Synapses and Neurons for Bioinspired Computing

Yang

Huang

Guo

2019

Adv Elect Materials

188

127

View full text Add to dashboard Cite

and constant data movement are needed while working. In the human brain, huge and complex neural networks composed of gigantic amounts of neurons massively interconnected by an even larger number of synapses are in charge of computing and memory. Unlike a digital computer, information storage and processing happen at the same time in the brain. [3] Artificial intelligence (AI) that could rival biological neural networks is being continuously pursued by human being. There are two approaches to implement AI: one is the software simulation of neural networks with the aid of digital computers, another is developing hardware-integrated circuits of neural networks. In fact, AI based on the software simulation is evolving very fast thanks to the advances in the development of algorithm in recent years, and it even outperforms the human brain in specific complex tasks, such as playing the game of Go. [4] However, software-based AI suffers from critical issues of enormous power consumption and massive data throughput. Hardware-based AI systems are more efficient in terms of speed and power consumption; however, complementary metal oxide semiconductor (CMOS) transistors are inherently different from the basic building blocks of biological neural networks, neurons, and synapses, in terms of operation dynamic and behavior. A circuit consisting of at least ten transistors are required to realize the function of one biological synapse, which is impractical for the implementation of large networks, not to mention the human brain (roughly 10 11 neurons interconnected by ≈10 15 synapses). [5,6] Fortunately, the emergence of memristive devices with innate dynamics resembling biological synapses and neurons provides a feasible and simple way to build hardware-based bio-inspired computing systems. [7,8] For instance, only one compact memristive device with a metalinsulator-metal (MIM) sandwich structure is sufficient to reproduce some functions of a biological synapse. Moreover, the vector-matrix multiplication, which is believed to be the most data-intensive work in neural networks, can be naturally performed in a cross-bar array of memristive devices on the physical level based on Ohm and Kirchhoff laws. [9,10] These features endow the memristive devices ideally suited to realize highly efficient bioinspired computing systems in hardware.To realize highly efficient neuromorphic computing that is comparable to biological counterparts, bioinspired computing systems, consisting of biorealistic artificial synapses and neurons, are developed with memristive devices with native dynamics resembling biological synapses and neurons. Tremendous materials and devices have been successfully used to emulate diverse functions of synapses, as well as neurons, in the last decade. Herein, approaches to realize certain synaptic or neuronal functions are introduced with state-of-art experimental demonstrations. First, the dynamics and working principles of biological synapses and neurons are briefly presented to provide guidance for developing biore...

show abstract

“…f) Schematic of the conductance change mechanism in oxide with and without intermediate phases, leading to different number of weight states. Reproduced with permission . Copyright 2017, The Royal Society of Chemistry.…”

Section: Memristive Synapsesmentioning

confidence: 99%

“…In particular, a high weight precision is usually indispensable in order to achieve online training . Recently, the weight precision in filamentary oxide memristors was found to be related to the phase diagram of the switching oxides used . It is well understood that the facilitation or depression of filamentary synapses usually originates from geometric changes of the conducting filament, including its length or width.…”

Section: Memristive Synapsesmentioning

confidence: 99%

Memristive Devices and Networks for Brain‐Inspired Computing

Zhang

et al. 2019

Physica Rapid Research Ltrs

Self Cite

View full text Add to dashboard Cite

As the era of big data approaches, conventional digital computers face increasing difficulties in performance and power efficiency due to their von Neumann architecture. As a result, there is recently a tremendous upsurge of investigations on brain-inspired neuromorphic hardware with high parallelism and improved efficiency. Memristors are considered as promising building blocks for the realization of artificial synapses and neurons and can therefore be utilized to construct hardware neural networks. Here, a review is provided on existing approaches for the implementation of artificial synapses and neurons based on memristive devices; and the respective advantages and disadvantages of these approaches are evaluated. This is followed by a discussion of hardware accelerators and neuromorphic computing systems that exploit the parallel, in-memory and analog characteristics of memristive crossbar arrays as well as the intrinsic dynamics of memristors. Finally, the outstanding challenges are addressed that have not yet been resolved in the present studies, and future advances are discussed that might be needed for building intelligent and energy efficient neuromorphic systems.

show abstract

Tuning analog resistive switching and plasticity in bilayer transition metal oxide based memristive synapses

Abstract: The existence of rich suboxide phases is favorable for increasing the number of weight states in transition metal oxide synapses.

Cited by 28 publications

References 53 publications

Graphene Oxide Quantum Dots Based Memristors with Progressive Conduction Tuning for Artificial Synaptic Learning

Graphene Oxide Quantum Dots Based Memristors with Progressive Conduction Tuning for Artificial Synaptic Learning

Memristive Synapses and Neurons for Bioinspired Computing

Memristive Devices and Networks for Brain‐Inspired Computing

Contact Info

Product

Resources

About