Spintronic Neuron MTJ Structure for Leaky-Integrate-Reset

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing neuron devices struggle with simulating the leaky characteristic of neurons, requiring large hard magnets that disrupt integration and limited mutual inhibition, failing to achieve global inhibition and 'winner takes all' functions.

Innovation Solution

An all-electrically-controlled spintronic neuron device with a specific layered structure, including a bottom antiferromagnetic pinning layer, synthetic antiferromagnetic layer, potential barrier layer, ferromagnetic reference layer, and antiferromagnetic pinning layers, utilizing electric field control for magnetic domain wall movement to simulate leaky and integrate functions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If magnetic tunnel junction is used to switch magnetization direction through spin transfer torque or spin orbit torque, then non-volatility and high read/write speed are achieved, but the ability to simulate full neuron functionality including leaky and integrate processes is limited

Engineering Contradiction:
Improveread/write speedVSAvoidneuron functionality simulation
Core Design Contradiction:
SpeedVSAdaptability or versatility

Solution Approach 1:

The magnetic tunnel junction structure is segmented into multiple functional regions: a threshold region formed by the potential barrier layer and a pinning region formed by the antiferromagnetic pinning layer. This segmentation allows different regions to perform different functions - the threshold region enables threshold switching behavior while the pinning region provides stable magnetization anchoring, together simulating complete neuron leaky-integrate-reset functionality

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different regions of the magnetic tunnel junction are given different local properties: the potential barrier layer creates a threshold region with specific magnetization switching characteristics, while the antiferromagnetic pinning layer creates a pinning region with stable magnetization. This local quality differentiation enables the device to simulate both threshold activation and leaky integration behaviors simultaneously

Inventive Principle:
Principle #3Local quality

2Reliability

If focus is placed on magnetic random access memories based on spin transfer torque and spin orbit torque, then high density and high endurance are achieved, but research on neuromorphic devices remains limited

Engineering Contradiction:
ImproveenduranceVSAvoidneuromorphic device functionality
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The magnetic tunnel junction device is designed to perform multiple functions: it maintains the high endurance characteristics of spintronic memory while simultaneously simulating neuron functions including leaky integration, threshold activation, and reset processes. The device can operate both as a stable memory element and as a dynamic neuromorphic computing unit, achieving universal functionality

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Device complexity

If existing neuron devices use stray fields for inhibition, then simple structure is maintained, but global inhibition capability is insufficient

Engineering Contradiction:
Improvestructure simplicityVSAvoidglobal inhibition capability
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent introduces a circuit with negative differential resistance characteristic as an intermediary element that converts local magnetic domain wall movements into global electrical signals. This intermediary enables the neuron device to achieve global inhibition capability while maintaining relatively simple magnetic layer structure, as the complex inhibition logic is implemented through the electrical circuit rather than magnetic structure

Inventive Principle:
Principle #24Intermediary (Mediator)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The device achieves reliable leaky-integrate-reset characteristics with minimal disruption to integrated circuits, enabling global inhibition and 'winner takes all' functionality through antiferromagnetic coupling and spin orbit torque.

Implementation Method 1

a synthetic antiferromagnetic layer formed on the bottom antiferromagnetic pinning layer, wherein the synthetic antiferromagnetic layer includes a bottom ferromagnetic layer, a synthetic antiferromagnetic coupling layer and a ferromagnetic free layer

Methodology Applied
Scientific EffectAntiferromagnetic coupling: Magnetism

Implementation Method 2

The magnetic tunnel junction switches a magnetization direction of a free layer through the Spin Transfer Torque (STT) generated by a polarized current of a ferromagnetic reference layer

Methodology Applied
Scientific EffectSpin transfer torque:

Implementation Method 3

or the Spin Orbit Torque (SOT) of a vertical injected spin current generated by Spin Hall Effect of a heavy metal

Methodology Applied
Scientific EffectSpin Hall effect: Hall Effect

Implementation Method 4

a relative magnetization direction of the ferromagnetic layer on two sides of the tunneling layer is switched, and a change of a resistance value may be characterized according to the tunneling magnetoresistance effect

Methodology Applied
Scientific EffectTunneling magnetoresistance effect: Magnetoresistance

Data Source

PatentUS12453292B2All-electrically-controlled spintronic neuron device, neuron circuit and neural network
Publication Date: 2025.10.21 INST OF MICROELECTRONICS CHINESE ACAD OF SCI LTD
  • US12453292B2 patent drawing
  • US12453292B2 patent drawing
  • US12453292B2 patent drawing

AI summary

Provided is an all-electrically-controlled spintronic neuron device, a neuron circuit and a neural network. The neuron device includes: a bottom antiferromagnetic pinning layer; a synthetic antiferromagnetic layer formed on the bottom antiferromagnetic pinning layer; a potential barrier layer formed on the ferromagnetic free layer, wherein a region of the ferromagnetic free layer directly opposite to the potential barrier layer forms a threshold region; a ferromagnetic reference layer formed on the potential barrier layer; wherein the potential barrier layer, the ferromagnetic reference layer and the ferromagnetic free layer form a magnetic tunnel junction; a first antiferromagnetic pinning layer and a second antiferromagnetic pinning layer formed on an exposed region of the ferromagnetic free layer except the region directly opposite the potential barrier layer, and located on two sides of the potential barrier layer; and a first electrode formed on the ferromagnetic reference layer.