TiO₂ based nanostructured memristor for RRAM and neuromorphic applications: a simulation approach

Area of Science:

• Materials Science
• Computational Physics
• Electrical Engineering

Background:

• Memristor devices are crucial for next-generation electronics, including Resistive Random-Access Memory (RRAM) and neuromorphic computing.
• Accurate simulation models are essential for understanding and optimizing memristor performance.
• Existing window functions have limitations in capturing the full range of memristor behavior, particularly nonlinear characteristics.

Purpose of the Study:

• To simulate nanostructured memristor devices using piecewise linear and nonlinear window functions.
• To evaluate the suitability of different window functions for RRAM and neuromorphic applications.
• To propose and validate a new nonlinear window function for improved memristor emulation.

Main Methods:

• Utilized a linear drift model for memristor simulation.
• Employed piecewise linear and nonlinear window functions as the mathematical and scripting basis.
• Developed and tested a novel nonlinear window function with scalable properties.

Main Results:

• Piecewise linear window functions effectively simulate memristor characteristics for RRAM applications.
• Standard nonlinear window functions exhibit limited nonlinear behavior at low control parameter magnitudes.
• The proposed nonlinear window function demonstrates scalable behavior up to f(x)=1 and exhibits nonlinearity at higher control parameter magnitudes.
• Simulation results show smooth nonlinear transitions between low and high resistance states using the proposed function.

Conclusions:

• The proposed nonlinear window function provides a more accurate and versatile model for memristor simulation.
• This new function is particularly well-suited for emulating the nonlinear dynamics required in neuromorphic applications.
• The findings pave the way for more sophisticated memristor-based device design and application development.