Reservoir Element Spin Conduction Layer Stabilization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Neuromorphic elements using spin torque oscillators (STO) face challenges due to manufacturing errors causing resonance frequency discrepancies, leading to insufficient interaction between STO elements and potential failure from long-term high-frequency current application, resulting in instability and increased power consumption.
Innovation Solution
A reservoir element with a spin conduction layer and ferromagnetic layers arranged in a specific configuration, including via wirings with opposite magnetization orientation, tunnel barrier layers, and a reference potential terminal, to stabilize operation and enhance power efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If STO elements are used to build neuromorphic elements, then the correct answer rate can be improved, but manufacturing errors cause resonance frequency discrepancies that prevent sufficient interaction between elements
Solution Approach 1:
The patent replaces the mechanical resonance-based STO interaction system with a spin current-based interaction system. Instead of relying on precise resonance frequency alignment of STO elements, the invention uses spin conduction layers to transmit spin currents between ferromagnetic layers, enabling interaction without mechanical resonance constraints. This substitution resolves the manufacturing precision issue while maintaining the correct answer rate.
Solution Approach 2:
The patent changes the interaction mechanism parameter from resonance frequency coupling to spin current coupling. By altering the fundamental interaction parameter, the system no longer depends on precise frequency alignment, thereby overcoming manufacturing errors in STO elements while preserving computational accuracy.
2Ease of operation
If high frequency current is applied to STO elements for operation, then the neuromorphic element can function, but long-term application causes element failure
Solution Approach 1:
The patent replaces the high-frequency current-driven STO operation with a spin current-driven operation. The spin conduction layer transmits spin currents at lower frequencies, eliminating the need for high-frequency current application while maintaining operational functionality. This substitution improves reliability by avoiding the degradation caused by prolonged high-frequency current exposure.
3Measurement precision
If learning is performed at each level in hierarchical elements, then the correct answer rate increases, but the number of chips increases imposing burden on circuit design and power consumption
Solution Approach 1:
The patent extracts the learning function from the reservoir element layer and concentrates it solely in the output part. The reservoir element performs only signal compression and transformation, while the output part handles all learning operations. This extraction simplifies the reservoir element's circuit design while maintaining the correct answer rate through efficient output-side learning.
Solution Approach 2:
The patent shifts the learning operation from the temporal dimension (distributed across multiple time steps at each level) to a concentrated operation at the output dimension. By moving learning to the output part, the system reduces circuit design complexity while preserving learning effectiveness through a different operational dimension.
4Measurement precision
If more chips are added to hierarchical elements for learning, then the correct answer rate improves, but power consumption increases significantly
Solution Approach 1:
The patent extracts the power-intensive learning operations from multiple distributed chips and concentrates them in a single output part. The reservoir element chips perform only low-power signal transformation, while the output part handles learning. This extraction maintains correct answer rates while dramatically reducing overall power consumption by eliminating redundant learning computations across multiple chips.
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 stable operation and reduced power consumption in neuromorphic elements by compressing signals and learning only at the output part, improving correct answer rates and circuit design simplicity.
Implementation Method 1
a spin conduction layer containing a non-magnetic conductor; a plurality of ferromagnetic layers positioned in a first direction with respect to the spin conduction layer
Implementation Method 2
a plurality of ferromagnetic layers positioned in a first direction with respect to the spin conduction layer and spaced apart from each other in a plan view from the first direction
Data Source
AI summary
A reservoir element of the first aspect of the present disclosure includes: a spin conduction layer containing a non-magnetic conductor; ferromagnetic layers positioned in a first direction with respect to the spin conduction layer and spaced apart from each other in a plan view from the first direction; and via wirings electrically connected to spin conduction layer on a surface opposite to a surface with the ferromagnetic layers.