A SPICE Model of Phase Change Memory for Neuromorphic Circuits

Chen, Xuhui; Hu, Huifang; Huang, Xiaoqing; Cai, Weiran; Liu, Ming; Lam, C.; Lin, Xinnan; Zhang, Lining; Chan, Mansun

doi:10.1109/access.2020.2995907

Cited by 11 publications

(5 citation statements)

References 31 publications

(40 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…If not, the skyrmion will move in the backward direction, thus showing the leaky behavior. It can be clearly seen from figure 7 that the leaky behavior in case of linearly tapered nanotrack is much slower as compared to the one in exponentially tapered nanotrack which can be justified from equations (10) and (11). Whenever the skyrmion reaches at the detection point i.e.…”

Section: Resultsmentioning

confidence: 83%

See 1 more Smart Citation

Antiferromagnetic skyrmion based shape-configured leaky-integrate-fire neuron device

Bindal

Raj

Kaushik

2022

J. Phys. D: Appl. Phys.

View full text Add to dashboard Cite

Spintronic devices based on antiferromagnetic skyrmion (AFM) motion on the nanotracks have gained significant interest as a key component of neuromorphic data processing systems. AFM skyrmions are favorable over the ferromagnetic skyrmions as they follow the straight trajectories and prevent its annihilation at the nanotrack edges. In this paper, the AFM skyrmion-based neuron device that exhibits the leaky-integrate-fire (LIF) functionality is proposed for the first time. It exploits the current-driven skyrmion dynamics on the shape-configured nanotracks that are linearly decreasing and exponentially decaying. The device structure creates the regions from lower to higher energy states for the AFM skyrmions during its motion from the wider to narrower region. This causes the repulsion force from the nanotrack edges to act on the AFM skyrmion thereby, drifting it in the backward direction in order to minimize the system energy. This provides the leaking functionality to the neuron device without any external stimuli and additional hardware cost. The average velocities during the integration and leaky processes are in the order of 103 and 102 m/sec, respectively for the linearly and exponentially tapered nanotracks. Moreover, the energy of the skyrmion is in the order 10-20 J. Hence, the suggested device opens up the path for the development of high-speed and energy-efficient devices in antiferromagnetic spintronics for neuromorphic computing.

show abstract

Section: Resultsmentioning

confidence: 83%

“…In the literature, leaky-integrate-fire (LIF) neuron has been extensively adopted [7,8]. Few single-device artificial neurons based on phase change devices, domain walls, and ferromagnetic (FM) skyrmions have recently been developed to imitate the LIF neuronal functionality [9][10][11][12][13].…”

Section: Introductionmentioning

confidence: 99%

Antiferromagnetic skyrmion based shape-configured leaky-integrate-fire neuron device

Bindal

Raj

Kaushik

2022

J. Phys. D: Appl. Phys.

View full text Add to dashboard Cite

show abstract

“…PCM as a neuromorphic semiconductor memory is also often used in neural networks to simulate the role of synapses [27]. So the PCM compact models with SC approach and DE approach are compared in the neural network circuit with 49*2 synapses for simulation [28], as shown in Fig. 9(a).…”

Section: Simulation Results and Discussionmentioning

confidence: 99%

A Robust and Efficient Compact Model for Phase-Change Memory Circuit Simulations

Chen

Ding

Huang

et al. 2021

IEEE Trans. Electron Devices

Self Cite

View full text Add to dashboard Cite

show abstract

“…A statistical model in an evolutionary manner for PCM conductance in the SET process has been developed [67] . The model is intended to capture the statistical characteristics of PCM crystallization.…”

Section: Modeling For Variations Of Pcm and Circuitsmentioning

confidence: 99%

A review of compact modeling for phase change memory

Ding¹,

Peng²,

Li³

et al. 2022

J. Semicond.

Self Cite

View full text Add to dashboard Cite

Phase change memory (PCM) attracts wide attention for the memory-centric computing and neuromorphic computing. For circuit and system designs, PCM compact models are mandatory and their status are reviewed in this work. Macro models and physics-based models have been proposed in different stages of the PCM technology developments. Compact modeling of PCM is indeed more complex than the transistor modeling due to their multi-physics nature including electrical, thermal and phase transition dynamics as well as their interactions. Realizations of the PCM operations including threshold switching, set and reset programming in these models are diverse, which also differs from the perspective of circuit simulations. For the purpose of efficient and reliable designs of the PCM technology, open issues and challenges of the compact modeling are also discussed.

show abstract

A SPICE Model of Phase Change Memory for Neuromorphic Circuits

Cited by 11 publications

References 31 publications

Antiferromagnetic skyrmion based shape-configured leaky-integrate-fire neuron device

Antiferromagnetic skyrmion based shape-configured leaky-integrate-fire neuron device

A Robust and Efficient Compact Model for Phase-Change Memory Circuit Simulations

A review of compact modeling for phase change memory

Contact Info

Product

Resources

About