Neurons and Synapses with Ferroelectrically Modulated Metal-Semiconductor Schottky Diodes and Method

Simple SummaryContent extracted from patent full text and abstract with AI.

This invention describes an artificial synapse component for use in neural networks, using a metal-semiconductor Schottky diode structure that is modulated by ferroelectric materials. The setup mimics characteristics of biological synapses, such as memory and adaptability, by employing ferroelectric layers to change electrical properties in response to applied gate voltage pulses. The design allows for efficient, scalable, and environmentally friendly production of synaptic devices compatible with existing semiconductor technologies.

Use CasesContent extracted from patent full text and abstract with AI.

• Hardware acceleration for artificial intelligence (AI) and neuromorphic computing systems
• Integration into edge devices for low-power smart sensing and adaptive response
• Brain-inspired computing hardware for robotics and autonomous systems
• Large-scale memory and information storage arrays for pattern recognition tasks
• Crossbar array architectures for massively parallel neural network operations
• Development of learning-capable, reconfigurable logic gates (e.g., AND/NAND gates)
• On-chip learning and inference for IoT (Internet of Things) and smart devices

BenefitsContent extracted from patent full text and abstract with AI.

• Mimics key functions of biological synapses including tunable synaptic weights and long-term potentiation/depression (memory effects)
• Reduces manufacturing complexity compared to conventional CMOS-based artificial synapses
• Enables low power consumption and high-speed operation
• Compatible with various environmentally friendly and CMOS-compatible materials
• Scalable for massively parallel integration, enabling high-density neural network chips
• Offers multi-level conductance states for enhanced information storage and processing capability
• Can be fabricated at relatively low temperatures, expanding material choice and process flexibility

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

The invention relates to a synaptic component for a neural network having a layer consisting of a semiconductor (101, 201), and a source electrode (102, 202) which is connected to the semi-conducting layer (101, 201), and a drain electrode (103, 203) which is connected to the semi-conducting layer (101, 201), wherein: the source electrode (102, 202) is spatially separated from the drain electrode (103, 203); the source electrode (102, 202) and the semi-conducting layer (101, 201) form a Schottky diode; and the source electrode (102, 202) is separated from a first gate electrode (105a, 205a) by ferroelectric material (104a, 104b, 204). The invention also relates to a method for the operation of a synaptic component according to the invention, by the first Schottky diode (101, 102; 201, 202) being switched into an inverse position and an electrical voltage (Vg1) being applied in a pulse-shaped manner to the first gate electrode (105a, 205a).