Ultralow Energy Domain Wall Device for Spin-Based Neuromorphic Computing

Kumar, Dheeraj; Chung, Hong Jing; Chan, JianPeng; Jin, Tianli; Lim, Sze Ter; Parkin, S. S. P.; Sbiaa, R.; Piramanayagam, S. N.

doi:10.1021/acsnano.2c09744

Cited by 27 publications

(13 citation statements)

References 100 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As stated in section 2, the length of our simulated device (the ferromagnetic layer in which the domain wall moves) is 1000 nm and the width is 100 nm. Our device dimensions are about one order lower than that of the domain-wall devices experimentally demonstrated thus far [26,29,31,32]. Also, they are of comparable size as the domain-wall devices reported through simulation [31], if they have to store synaptic weights of the same bit resolution.…”

Section: Scaling Issuesmentioning

confidence: 75%

“…So estimating non-linearity and asymmetry factors for Leonard et al's reported data is difficult. Kumar et al have recently reported synaptic behaviour in a similar domain-wall device through magneto-optic Kerr effect (MOKE) images and change in TMR of the MTJ structure due to current-driven domain-wall motion [32]. But they have not reported any LTP-LTD plot from which the non-linearity and asymmetry coefficients can be inferred.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Impact of edge defects on the synaptic characteristic of a ferromagnetic domain-wall device and on on-chip learning

Singh Yadav,

Sadashiva,

Holla

et al. 2023

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

Topological-soliton-based devices, like the ferromagnetic domain-wall device, have been proposed as non-volatile memory (NVM) synapses in electronic crossbar arrays for fast and energy-efficient implementation of on-chip learning of neural networks. High linearity and symmetry in the synaptic weight-update characteristic of the device (long-term potentiation (LTP) and long-term depression (LTD)) are important requirements to obtain high classification/ regression accuracy in such an on-chip learning scheme. However, obtaining such linear and symmetric LTP and LTD characteristics in the ferromagnetic domain-wall device has remained a challenge. Here, we first carry out micromagnetic simulations of the device to show that the incorporation of defects at the edges of the device, with the defects having higher perpendicular magnetic anisotropy (PMA) compared to the rest of the ferromagnetic layer, leads to massive improvement in the linearity and symmetry of the LTP and LTD characteristics of the device. This is because these defects act as pinning centres for the domain wall and prevent it from moving during the delay time between two consecutive programming current pulses, which is not the case when the device doesn't have defects. Next, we carry out system-level simulations of two crossbar arrays with synaptic characteristics of domain-wall synapse devices incorporated in them: one without such defects, and one with such defects. For on-chip learning of both Long Short Term Memory Networks (LSTMs) (using a regression task) and Fully Connected Neural Networks (FCNN) (using a classification task), we show improved performance when the domain-wall synapse devices have defects at the edges. We also estimate the energy consumption in these synaptic devices and project their scaling, with respect to on-chip learning in corresponding crossbar arrays.

show abstract

Section: Scaling Issuesmentioning

confidence: 75%

Section: Related Workmentioning

confidence: 99%

Impact of edge defects on the synaptic characteristic of a ferromagnetic domain-wall device and on on-chip learning

Singh Yadav,

Sadashiva,

Holla

et al. 2023

Neuromorph. Comput. Eng.

View full text Add to dashboard Cite

show abstract

“…Recently, Kumar et al studied the hybrid high pressure‐W/low pressure‐W spin‐Hall layer and demonstrated the DW motion at ultralow current densities of 10 6 A m −2 . [ 51 ] A small assisting OOP magnetic field is required to move the DWs at such low current densities. Such a low‐current density is attributed to the moderate SHA values of 0.15 and ultralow pinning field of 2 mT.…”

Section: Spin–orbit Torquementioning

confidence: 99%

“…[ 50 ] Kumar et al demonstrated the synaptic functionality in meander devices. [ 51 ] In the meander device, two segments of DW devices join at an offset that forms the pinning site. The number of multiple resistances is determined by the number of pinning sites.…”

Section: Spintronic Devices For Neuromorphic Applicationsmentioning

confidence: 99%

Spintronic Heterostructures for Artificial Intelligence: A Materials Perspective

Maddu

Kumar

Bhatti

et al. 2023

Physica Rapid Research Ltrs

View full text Add to dashboard Cite

With the advent of the Big Data era, neuromorphic computing (NC) (also known as brain‐inspired computing) has gained a lot of research interest. Spintronic devices are the emerging candidates for implementing the NC due to their intrinsic nonvolatility, extremely high endurance, low‐power consumption, and complementary metal‐oxide compatibility. Many research groups have proposed various NC architectures based on spintronic devices. Herein, a collective survey of different spintronic‐based approaches is given for NC. The reviewed approaches include the progress of stochastic magnetic tunnel junction (MTJ) devices, spin‐torque nano‐oscillator, spin‐Hall nano‐oscillator, domain walls, and skyrmion devices. In all of these approaches, spin–orbit torque (SOT)‐based magnetization control, which is achieved via spintronics heterostructures, plays a significant role. Various heterostructures of heavy metal and ferromagnetic layers that have been proposed are reviewed for generating SOT. In addition, the phenomena and materials involved in the generation of orbital torque are summarized due to the orbital Hall effect (OHE), which has recently gained researchers’ attention. Finally, an outlook on the opportunities and challenges for spintronic‐based NC hardware is provided, shedding light on its great potential for artificial intelligence (AI) applications.

show abstract

“…Along with charge‐based devices, synaptic devices exploiting their spin degree of freedom have been studied recently. [ 11,12,13,14 ] Devices based on spin‐transfer‐torque magnetic random memory (STT‐MRAM), [ 15,16 ] skyrmion, [ 17–19 ] and spin‐orbit‐torque magnetic random memory (SOT‐MRAM), [ 20,21,22 ] etc have been implemented in synaptic electronics. The spintronics materials are appealing for their fast switching, unlimited endurance and compatibility with the complementary metal oxide semiconductors (CMOS) based technology.…”

Section: Introductionmentioning

confidence: 99%

A Multilevel Magnetic Synapse Based on Voltage‐Tuneable Magnetism by Nitrogen Ion Migration

Monalisha

Pellicer

et al. 2023

Adv Elect Materials

View full text Add to dashboard Cite

Advanced synaptic devices with simultaneous memory and processor capabilities are envisaged as core elements of neuromorphic computing (NC) for low‐power artificial intelligence. So far, most synaptic devices are based on resistive memories, where the device resistance is tuned with applied voltage or current. However, the use of electric current in such resistive devices causes significant power dissipation due to Joule heating. Higher energy efficiency has been reported in materials exhibiting voltage control of magnetism (VCM). In particular, voltage‐driven ion motion to modulate magnetism (magneto‐ionics) is an emerging VCM mechanism that can offer new prospects for low‐power implementation of NC. In the present work, voltage‐driven nitrogen ion motion is exploited in transition metal nitride (CoFeN) thin films (i.e., nitrogen magneto‐ionics) to emulate biological synapses. In the proposed device, distinct multilevel non‐volatile magnetic states for analog computing and multi‐state storage are realized. Moreover, essential synaptic functionalities of the human brain are successfully simulated. The device exhibits an excellent synapse with a remarkable retention time (≈6 months), high switching ratio and large endurance (≈103), for hardware implementation of NC. This research provides new insight into exploiting magneto‐ionic‐based synaptic devices for spin‐based neuromorphic systems.

show abstract

Ultralow Energy Domain Wall Device for Spin-Based Neuromorphic Computing

Cited by 27 publications

References 100 publications

Impact of edge defects on the synaptic characteristic of a ferromagnetic domain-wall device and on on-chip learning

Impact of edge defects on the synaptic characteristic of a ferromagnetic domain-wall device and on on-chip learning

Spintronic Heterostructures for Artificial Intelligence: A Materials Perspective

A Multilevel Magnetic Synapse Based on Voltage‐Tuneable Magnetism by Nitrogen Ion Migration

Contact Info

Product

Resources

About