Tailoring Synaptic Properties of the Band-Engineered Charge Trap Memory for a Flexible Edge Neuromorphic Processor

Area of Science:

• Materials Science
• Neuroscience
• Electrical Engineering

Background:

• Biological signals are crucial for medicine, security, and interface technologies, demanding high-performance memory systems.
• Existing inorganic synaptic devices offer linearity but lack flexibility, while organic devices are flexible but unreliable.
• Neuromorphic synaptic devices are essential for processing complex biological signals.

Purpose of the Study:

• To overcome the limitations of current synaptic devices by developing a flexible, high-performance neuromorphic device.
• To create a hybrid organic-inorganic structure that combines flexibility with electrical robustness.

Main Methods:

• Developed a band-engineered charge trap memory (BE-CTM) device using initiated chemical vapor deposition (iCVD).
• Fabricated uniform hybrid organic-inorganic dielectric layers with nanometer-scale thickness (≤10 nm) and tunable composition.
• Conducted electrical and reliability tests using biologically relevant voltage pulses and performed system-level simulations.

Main Results:

• The BE-CTM device demonstrated excellent synaptic weight linearity and operational stability under flexible conditions.
• System-level simulations showed high recognition accuracy: 93.4% for handwritten data and 95.8% for noisy images.
• ECG classification using a multilayer perceptron (MLP) model maintained high accuracy with minimal conductance updates.

Conclusions:

• The developed BE-CTM device successfully integrates mechanical flexibility with electrical robustness for neuromorphic applications.
• This hybrid device shows significant potential for next-generation neuromorphic and biometric information processing.
• The iCVD method enables precise control over dielectric layer properties for tailored device performance.