Neuromorphic Synapse Stack Structure for Linear Conductivity Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neuromorphic devices face challenges in achieving symmetric and linear changes in electrical conductivity for learning and recognition operations, with existing synapses exhibiting abrupt changes in resistance states and non-constant rate of change, which affects their learning and recognition accuracy.
Innovation Solution
A synapse design featuring an oxygen-containing layer with a stack structure of reactive metal layers alternately arranged with oxygen diffusion-retarding layers, where the thickness of a dielectric oxide layer changes in response to electrical pulses, allowing for gradual and symmetric changes in conductivity during potentiation and depression operations, and a resistance layer is added to increase the synapse's resistance value.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a simple reactive metal layer structure is used, then the device complexity is reduced, but the linearity and symmetry of conductivity changes deteriorate
Solution Approach 1:
The reactive metal layer is segmented into multiple layers (first reactive metal layer, second reactive metal layer, third reactive metal layer) with different thicknesses and positions. This segmentation allows each layer to contribute differently to the overall conductivity change, enabling precise control over the linearity and symmetry of the potentiation and depression operations while maintaining a manageable structural complexity.
Solution Approach 2:
The patent deliberately introduces asymmetry in the stack structure by placing oxygen diffusion-retarding layers at specific positions (between the first and second reactive metal layers, and between the second and third reactive metal layers) and giving different thicknesses to different reactive metal layers. This asymmetric arrangement compensates for non-linear effects and achieves symmetric conductivity changes during potentiation and depression operations.
2Speed
If the dielectric oxide layer thickness changes rapidly, then the response speed is improved, but the abrupt resistance state changes worsen learning accuracy
Solution Approach 1:
The patent implements dynamic control of the dielectric oxide layer thickness through multi-stage potential operations. The thickness changes progressively through multiple steps rather than abruptly, allowing the synapse to transition smoothly between conductive states. This dynamic, gradual adjustment improves learning accuracy by preventing abrupt resistance changes while maintaining adequate response speed through optimized pulse sequences.
Solution Approach 2:
The patent employs periodic electrical pulses to drive the formation and removal of the dielectric oxide layer in a controlled, step-by-step manner. By applying sequences of pulses with specific amplitudes and durations, the system achieves gradual conductivity changes that improve learning accuracy while maintaining responsive behavior through the periodic nature of the stimulation.
3Productivity
If the rate of conductivity change is increased, then the learning speed is improved, but the symmetry between potentiation and depression operations deteriorates
Solution Approach 1:
The patent utilizes parameter changes in the electrical pulses (amplitude, duration, frequency) to control the rate of dielectric oxide layer formation and removal. By carefully adjusting these parameters, the system achieves symmetric conductivity changes during potentiation and depression operations while maintaining adequate learning speed. The different thicknesses of reactive metal layers also serve as fixed parameters that balance the symmetry of operations.
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 design ensures linearity and symmetry in the electrical conductivity changes, preventing abrupt resistance state changes and improving the learning and recognition accuracy of neuromorphic devices by controlling the rate of change in conductivity.
Implementation Method 1
a dielectric oxide layer is formed or disappears according to a voltage or current applied to the first electrode and the second electrode, the dielectric oxide layer being formed in at least one of the plurality of reactive metal layers at an interface with a corresponding one of the plurality of oxygen diffusion-retarding layers when the at least one of the plurality of reactive metal layers reacts with the oxygen ions
Implementation Method 2
the plurality of oxygen diffusion-retarding layers interfere with a movement of the oxygen ions from the oxygen-containing layer to the plurality of reactive metal layers
Data Source
AI summary
A neuromorphic device includes a synapse. The synapse includes a first electrode, a second electrode spaced apart from the first electrode, an oxygen-containing layer disposed between the first electrode and the second electrode, the oxygen-containing layer including oxygen ions, and a stack structure disposed between the oxygen-containing layer and the second electrode, the stack structure including a plurality of reactive metal layers alternately arranged with a plurality of oxygen diffusion-retarding layers. The plurality of reactive metal layers are capable of reacting with oxygen ions of the oxygen-containing layer. The plurality of oxygen diffusion-retarding layers interfere with a movement of the oxygen ions from the oxygen-containing layer to the plurality of reactive metal layers.