Harnessing a WO-based flexible transparent memristor synapse with a hafnium oxide layer for neuromorphic computing

Area of Science:

• Materials Science
• Neuroscience
• Electrical Engineering

Background:

• Transparent memristor-based neuromorphic synapses are crucial for high-speed information processing.
• Synaptic linearity and potentiation/depression cycles are key challenges for memristor applications.

Purpose of the Study:

• To improve switching uniformity and synaptic properties of transparent memristors.
• To investigate the effect of a HfO2 interfacial layer on memristor performance.
• To develop a memristor suitable for flexible electronic devices.

Main Methods:

• Fabrication of a transparent memristor with a WO3/HfO2/ITO structure.
• Characterization of device properties including switching uniformity, linearity, and cycling stability.
• Evaluation of synaptic behavior using spike-timing-dependent plasticity (STDP).
• Analysis of material composition and interface properties using X-ray photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM).

Main Results:

• Optimized HfO2 thickness improved switching properties and synaptic performance.
• The memristor exhibited excellent P/D linearity, 494 epochs of cycling stability, and image recognition capabilities under 3 mm bending.
• Demonstrated reversible short-term and long-term learning behaviors via STDP.
• Achieved 92.2% transparency, confirmed by UV-Vis spectroscopy.

Conclusions:

• The HfO2 interfacial layer effectively enhances memristor uniformity and synaptic properties.
• The optimized transparent memristor is suitable for flexible neuromorphic applications.
• The study provides insights into the switching mechanism and material composition for device optimization.