Variability in HfO₂-based memristors described with a new bidimensional statistical technique.

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Science

Background:

• Resistive switching (RS) memories are crucial for advanced computing applications.
• Understanding cycle-to-cycle variability is essential for reliable device performance.
• Current analysis methods often lack the granularity to fully capture complex variability.

Purpose of the Study:

• To introduce a novel statistical analysis for assessing cycle-to-cycle variability in resistive memories.
• To enhance the understanding of parameter distributions and their variation.
• To provide a more robust method for evaluating memory technology suitability.

Main Methods:

• Employing two-dimensional (2D) distributions to analyze set/reset voltages and currents.
• Utilizing a 2D coefficient of variation (CV) for comprehensive data assessment.
• Applying the methodology to resistive switching data from hafnium oxide-based technologies.

Main Results:

• The 2D methodology provides a more thorough analysis compared to traditional 1D methods.
• Demonstrated effectiveness in analyzing variability in hafnium oxide resistive memory devices.
• The 2D CV offers a compact yet powerful metric for variability assessment.

Conclusions:

• The proposed 2D statistical analysis significantly improves the assessment of resistive memory variability.
• This method facilitates a more accurate evaluation of technology readiness for non-volatile memories, neuromorphic computing, and random number generation.
• Enhanced understanding of variability is key to advancing next-generation electronic devices.