Nanoscale wedge resistive-switching synaptic device and experimental verification of vector-matrix multiplication for hardware neuromorphic application

Abstract

In this work, nanoscale wedge-structured silicon nitride (SiN x )-based resistive-switching random-access memory with data non-volatility and conductance graduality has been designed, fabricated, and characterized for its application in the hardware neuromorphic system. The process integration with full Si-processing-compatibility for constructing the unique wedge structure by which the electrostatic effects in the synaptic device operations are maximized is demonstrated. The learning behaviors of the fabricated synaptic devices are shown. In the end, vector-matrix multiplication is experimentally verified in the array level for application in more energy-efficient hardware-driven neuromorphic systems.

Core-Shell Dual-Gate Nanowire Charge-Trap Memory for Synaptic Operations for Neuromorphic Applications

“…The brain functions (observation, reorganization, learning, and memorization) are performed by neurons (computing elements) and synapses (memory elements) [1,2]. In the neuromorphic system, an artificial synaptic device plays a key role in linking the artificial neurons and modulating the connection strength (synaptic weight) between neurons [3][4][5][6][7][8][9][10][11][12][13][14][15]. In order to realize brain-like computing, different types of artificial synaptic devices have been proposed for artificial intelligence applications [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22].…”

“…In the neuromorphic system, an artificial synaptic device plays a key role in linking the artificial neurons and modulating the connection strength (synaptic weight) between neurons [3][4][5][6][7][8][9][10][11][12][13][14][15]. In order to realize brain-like computing, different types of artificial synaptic devices have been proposed for artificial intelligence applications [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. The major applications for these artificial synaptic transistors are neuromorphic in-memory computing chip, artificial sensory perception, humanoid robotics, memorize, and recognize massive and unstructured data through parallel and power-efficient ways [3][4][5][6][7][8][9][10][11][12][13][14][15]…”

“…In order to realize brain-like computing, different types of artificial synaptic devices have been proposed for artificial intelligence applications [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. The major applications for these artificial synaptic transistors are neuromorphic in-memory computing chip, artificial sensory perception, humanoid robotics, memorize, and recognize massive and unstructured data through parallel and power-efficient ways [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22]. Charge tapping/de-trapping based artificial synapse are favorable for in-memory computing applications due to their stable analogue conductance state and nonvolatile characteristic [12].…”

Core-Shell Dual-Gate Nanowire Memory as a Synaptic Device for Neuromorphic Application

“…Short-term plasticity is utilized for keeping the information for a short time with discrimination by duration of data (input frequency) while long-term plasticity provides the function of storing data for a long time. The synaptic behaviors can be realized by different types of memory devices [4]- [14].…”

“…Among these, two-terminal devices have the great resemblance with the biological synapse but lack of completeness in realizing the full functions, and in many cases, the Si-processing compatibility is not practically considered [4], [11]. In order to overcome these issues, Si-processing-based synaptic memory devices with a combination of volatile and nonvolatile memories have been demonstrated as a promising candidate owing to their higher in functionality for synaptic operation reliabilities [5], [7]- [10], [15].…”