Variable Resistance Element for Neuromorphic Synapse

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current variable resistance elements used in neuromorphic devices exhibit abrupt resistance changes between set and reset operations, which limits their ability to mimic the gradual conductivity changes seen in biological synapses, affecting the accuracy of learning and recognition processes in neuromorphic devices.

Innovation Solution

A variable resistance element with an ion-receiving layer and an ion supply layer, where the resistance changes gradually based on the amount of ions supplied in response to applied voltages, allowing for multiple resistance levels and symmetrical conductivity characteristics, suitable for use as a synapse in neuromorphic devices.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If conventional variable resistance elements are used in neuromorphic devices, then device structure is simple, but resistance changes abruptly between set and reset operations, limiting the ability to mimic biological synapses

Engineering Contradiction:
Improveresistance control precisionVSAvoiddevice structure
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The variable resistance element is segmented into distinct functional layers: an ion supply layer containing mobile ions, an ion-receiving layer that accepts ions, and a gate electrode that controls ion movement. This segmentation allows independent optimization of each layer's properties to achieve gradual resistance changes while maintaining overall device manageability

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different layers are assigned specific local qualities: the ion supply layer provides a reservoir of mobile ions, the ion-receiving layer provides ion acceptance capability with specific stoichiometry, and the gate electrode provides controlled electric field. Each layer's local properties are optimized to contribute to the overall gradual resistance modulation function

Inventive Principle:
Principle #3Local quality

2Measurement precision

If abrupt resistance changes are used in variable resistance elements, then switching speed is fast, but accuracy of learning and recognition processes in neuromorphic devices deteriorates

Engineering Contradiction:
Improvelearning accuracyVSAvoidresistance change speed
Core Design Contradiction:
Measurement precisionVSSpeed

Solution Approach 1:

The resistance change mechanism is made dynamic and continuous through controlled ion migration. By applying different voltages to the gate electrode, the resistance can be continuously adjusted between high and low states, enabling analog-like behavior that mimics biological synapse weight adjustment while maintaining relatively fast response times

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The resistance state is controlled by changing the concentration of mobile ions in the ion-receiving layer through voltage-controlled ion migration. This parameter change approach allows continuous resistance modulation rather than abrupt switching, improving learning accuracy while maintaining acceptable speed through efficient ion transport

Inventive Principle:
Principle #35Parameter changes

3Adaptability or versatility

If gradual resistance changes are implemented through ion supply and ion-receiving layers, then ability to mimic biological synapses is improved, but device complexity increases

Engineering Contradiction:
Improvesynapse mimicking capabilityVSAvoidlayer structure
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The ion supply layer and ion-receiving layer structure serves multiple functions: it enables gradual resistance changes for synapse mimicking, provides non-volatile memory characteristics, and allows for programmable resistance states. This multi-functionality justifies the increased structural complexity by delivering diverse neuromorphic capabilities from a single device architecture

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

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 solution enables neuromorphic devices to achieve accurate and efficient learning and recognition by allowing gradual resistance changes, improving the operating characteristics of neuromorphic devices and enabling them to process information more effectively.

Implementation Method 1

a resistance of the ion-receiving layer varies depending on an amount of ions supplied from the ion supply layer based on a voltage applied to the gate pattern

Methodology Applied
Scientific EffectIon migration: Ion Repulsion/Attraction

Data Source

PatentUS11380844B2Semiconductor device including variable resistance element
Publication Date: 2022.07.05 SK HYNIX INC
  • US11380844B2 patent drawing
  • US11380844B2 patent drawing
  • US11380844B2 patent drawing

AI summary

A semiconductor device including at least one variable resistance device is provided. A variable resistance element includes: an ion supply layer having a top, a bottom and a sidewall connecting the top to the bottom; an ion-receiving layer having an inner sidewall connected to at least a portion of the sidewall of the ion supply layer; a gate pattern connected to an outer sidewall of the ion-receiving layer; and a source pattern connected to one of the top or bottom of the ion supply layer, and a drain pattern connected to the other one or the top or bottom of the ion supply layer. A resistance of the ion supply layer is varies depending on an amount of ions supplied from the ion supply layer to the ion-receiving layer in response to a voltage applied to the gate pattern.