Versatile Titanium Carbide MXene Thin-Film Memristors with Adaptive Learning Behavior

Area of Science:

• Materials Science
• Nanotechnology
• Computer Engineering

Background:

• Growing demand for neuromorphic computing hardware to address the von Neumann bottleneck.
• Need for advanced materials beyond traditional silicon-based architectures.
• Exploration of two-dimensional materials for next-generation computing.

Purpose of the Study:

• To prepare and characterize two-dimensional titanium carbide (Ti₃C₂T) MXene for memristor applications.
• To fabricate and evaluate the performance of a Ti₃C₂T-based memristor.
• To demonstrate the potential of this memristor for edge computation and artificial learning.

Main Methods:

• Synthesis of Ti₃C₂T MXene using conventional HF etching.
• Thin-film deposition via immiscible liquid-liquid interfacial growth.
• Fabrication of memristors by sandwiching Ti₃C₂T between electrodes.
• Characterization using Raman scattering, crystallinity measurements, current-voltage (I-V) analysis, and conductive AFM.

Main Results:

• Successful preparation and characterization of Ti₃C₂T MXene.
• Memristor exhibits nonvolatile resistive switching with a switching speed of 30 ns and an On/Off ratio of ~10³.
• Device shows high endurance (500 cycles), retention (>10⁴ s), and low energy consumption (fJ per reading).
• Filamentary conduction mechanism confirmed via I-V characteristics and conductive AFM.
• Demonstration of edge computation, logic gate operations, and classical conditioning.

Conclusions:

• Ti₃C₂T MXene is a promising material for high-performance memristors.
• The fabricated memristor offers excellent switching characteristics, endurance, and energy efficiency.
• This versatile device shows significant potential for advanced neuromorphic computing and AI applications.