Phase-Change Neuromorphic Synapse Tunable Time Delay

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current neuromorphic networks face limitations in achieving compact, high-performance, low-power hardware implementations with tunable time-delay elements, as existing solutions like mercury delay lines require replacing the entire device for different time delays, which is inefficient.

Innovation Solution

A synapse for neuromorphic networks is developed using a phase-change material time-delay element with actuators at both ends, allowing for tunable time delays by altering the phase change material's properties through electrical or thermal stimuli, enabling adjustable signal propagation times.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If mercury delay line is used for time delay, then time delay function is achieved, but device complexity and power consumption increase

Engineering Contradiction:
Improvetime delay functionVSAvoiddevice complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical mercury delay line system with an electrical/photonic integrated circuit implementation. The time delay function is achieved through photodetectors, transimpedance amplifiers, and capacitors on an integrated circuit substrate, eliminating the need for liquid mercury and complex mechanical tuning mechanisms.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent creates an electrical/photonic copy of the mercury delay line's time delay function using integrated circuit components. Instead of physically delaying signals through sound propagation in mercury, the invention uses electrical signal processing through amplifiers and capacitive elements to achieve equivalent time delay functionality.

Inventive Principle:
Principle #26Copying

2Adaptability or versatility

If mercury delay line length is changed to tune time delay, then time delay adjustment is achieved, but device complexity and power consumption increase

Engineering Contradiction:
Improvetime delay tuningVSAvoiddevice complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements dynamic time delay adjustment through electronic control of capacitor values or amplifier gain settings on the integrated circuit. This allows continuous tuning of the time delay parameter without physical reconfiguration, enabling adaptive adjustment through electrical signals rather than mechanical changes.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The patent achieves time delay tuning by changing electrical parameters such as capacitance values or amplification factors within the integrated circuit. This allows the time delay characteristic to be adjusted by modifying circuit parameters through voltage or current control, rather than changing physical dimensions of the delay medium.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If high-density hardware neuromorphic networks are constructed, then recognition performance is improved, but power consumption increases

Engineering Contradiction:
Improverecognition performanceVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent replaces power-intensive mechanical or optical hardware components with energy-efficient integrated circuit implementations. The use of photodetectors and transimpedance amplifiers on an integrated substrate reduces power consumption compared to traditional hardware neuromorphic approaches while maintaining high-density connectivity for improved recognition performance.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

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

This solution allows for compact, high-density neuromorphic networks with tunable time delays, enhancing learning and recognition capabilities by adjusting signal pathways between nodes, reducing energy consumption and recognition time.

Implementation Method 1

The time-delay portion is formed from a phase change material wherein a change in the material of the time-delay portion alters a propagation time of a signal transmitted from the first actuator to the second actuator

Methodology Applied
Scientific EffectPhase change: Phase Change

Data Source

PatentUS9547819B1Phase-change material time-delay element for neuromorphic networks
Publication Date: 2017.01.17 INTERNATIONAL BUSINESS MACHINE CORPORATION
  • US9547819B1 patent drawing
  • US9547819B1 patent drawing
  • US9547819B1 patent drawing

AI summary

A synapse for a neuromorphic network is provided. The synapse includes a time-delay portion having a first end and a second end, a first actuator located at the first end and in operational contact with the time-delay portion, and a second actuator located at the second end and in operational contact with the time-delay portion. The time-delay portion is formed from a phase change material wherein a change in the material of the time-delay portion alters a propagation time of a signal transmitted from the first actuator to the second actuator.