Dynamic and Static Switching in ITO/SnOx/ITO and Its Synaptic Application.

Area of Science:

• Materials Science
• Neuroscience
• Computer Science

Background:

• Artificial neural networks (ANNs) increasingly utilize diverse neural network (NN) architectures like CNNs and RNNs.
• Resistive random-access memory (RRAM) is a promising candidate for memory devices in these networks.
• Developing efficient memory devices is crucial for advancing brain-inspired computing.

Purpose of the Study:

• To investigate the potential of a fabricated ITO/SnOX/TaN device as a synaptic element.
• To analyze the current-voltage (I-V) characteristics and data retention properties of the RRAM device.
• To confirm the device's ability to emulate synaptic plasticity, including paired-pulse facilitation (PPF).

Main Methods:

• Fabrication of an ITO/SnOX/TaN RRAM device.
• Measurement of current-voltage (I-V) characteristics, distinguishing between dynamic and static behaviors.
• Data retention tests at room temperature for 10^3 seconds.
• X-ray photoelectron spectroscopy (XPS) analysis to understand conduction mechanisms.
• Application of rectangular voltage pulses to assess synaptic plasticity (PPF, STP to LTP transition).

Main Results:

• The device exhibited two distinct I-V curve types: dynamic and static, post-forming process.
• Demonstrated data retention capabilities at room temperature.
• XPS analysis and linear fitting provided insights into the conduction mechanisms responsible for data preservation.
• The device successfully emulated synaptic plasticity, transitioning from short-term potentiation (STP) to long-term potentiation (LTP) via PPF.

Conclusions:

• The ITO/SnOX/TaN RRAM device shows potential for implementing synaptic functions in artificial neural networks.
• The observed dynamic and static I-V characteristics, along with plasticity, are key features for neuromorphic computing.
• This research contributes to the development of advanced memory devices for next-generation AI.