Neuron Synapse Bifunctional Element for CMOS Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neuromorphic semiconductor devices face challenges in integrating neuron and synapse functions while maintaining compatibility with existing CMOS devices and achieving low power consumption.
Innovation Solution
A neuron synapse bifunctional element is designed with a gate electrode, an oxide dielectric charge supply layer, an oxide semiconductor channel layer, and source and drain electrodes, where electrons or holes trapped in a charge trapping layer between the channel and charge supply layers change the current flow based on the signal applied to the gate electrode, enabling both neuron and synapse functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If separate CMOS circuits are used to implement neuron and synapse functions, then functional capability is achieved, but device complexity and integration difficulty increase
Solution Approach 1:
The patent combines neuron and synapse functions into a single device structure. The neuron synapse bifunctional element integrates the neuron's spike generation capability with the synapse's weight adjustment functionality, eliminating the need for separate circuit implementations and reducing overall system complexity.
Solution Approach 2:
The bifunctional element is designed to perform multiple functions simultaneously - it can operate as a neuron element, a synapse element, or both together. This multi-functionality is achieved through a unified structure that responds to electrical stimulation in ways that enable both neuronal spiking and synaptic weight modulation.
2Adaptability or versatility
If complex CMOS circuits are used to implement neuron functions with Leaky-Integrate-Fire characteristics, then neuron functionality is achieved, but manufacturing complexity increases
Solution Approach 1:
The patent replaces complex CMOS circuit implementations with a physics-based device structure. Instead of using multiple transistors and capacitors to simulate neuronal behavior, the invention uses a single device with specific material layers (oxide semiconductor channel layer, charge trapping layer, dielectric layers) that naturally exhibit neuron-like spiking and synapse-like plasticity through physical mechanisms.
3Adaptability or versatility
If RRAM or memristor devices are used for synapse elements, then synapse functionality is achieved, but compatibility with existing CMOS devices becomes problematic
Solution Approach 1:
The device employs a composite structure with multiple material layers including oxide semiconductor, charge trapping materials, and dielectric layers. This composite approach enables the device to exhibit both neuronal and synaptic functions while being compatible with standard CMOS fabrication processes, as each layer can be deposited using conventional semiconductor manufacturing techniques.
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 bifunctional element allows for high circuit integration by performing both neuron and synapse functions with the same element type, maintaining CMOS compatibility, and enabling functionality beyond electrical stimulation, including light stimulation.
Implementation Method 1
as electrons or holes are trapped in a charge trapping layer between the channel layer and the charge supply layer, a magnitude and pattern of a current flow flowing in the channel layer change depending on a change in a signal applied to the gate electrode
Implementation Method 2
a channel layer provided on the charge supply layer and formed of an oxide semiconductor, and a source electrode and a drain electrode provided on both sides of the channel layer
Implementation Method 3
a charge supply layer provided on the gate electrode and formed of an oxide dielectric
Implementation Method 4
the learning direction determining dielectric may be configured to control an intensity of an electric field distributed to the charge supply layer based on a thickness, a material type, and a density of a thin film
Data Source
AI summary
Disclosed is a neuron synapse bifunctional element, which includes a gate electrode, a charge supply layer provided on the gate electrode and formed of an oxide dielectric, a channel layer provided on the charge supply layer and formed of an oxide semiconductor, and a source electrode and a drain electrode provided on both sides of the channel layer, respectively, and as electrons or holes are trapped in a charge trapping layer between the channel layer and the charge supply layer, a magnitude and pattern of a current flow flowing in the channel layer change depending on a change in a signal applied to the gate electrode.