Assessment of functional performance in self-rectifying passive crossbar arrays utilizing sneak path current

Area of Science:

• Materials Science and Engineering
• Electrical Engineering
• Computer Science

Background:

• Self-rectifying memristive devices offer potential for low-power in-memory computing.
• Characterizing passive crossbar arrays with these devices is difficult due to parasitic sneak path currents.
• Existing methods like read margin assessments are limited by the voltage divider effect.

Purpose of the Study:

• To introduce a novel performance metric, Sneak-path Current (SC), for evaluating passive crossbar arrays.
• To address the challenge of characterizing memristive devices in crossbar architectures.
• To provide a quantitative measure of sneak path current interaction with read margin.

Main Methods:

• Proposed a new metric, $$ SC, utilizing sneak path currents for functional behavior assessment.
• Employed negative rectification factors ($$ and $$) to analyze dynamic rectification across bias and resistance states.
• Evaluated the metric using self-rectifying BiFeO$$ memristive cells.

Main Results:

• The $$ SC metric quantitatively evaluates the interplay between sneak path currents and read margin.
• Demonstrated efficacy in characterizing passive crossbar arrays, addressing a key research gap.
• Achieved $$ SC < 2.19E-2 with a read margin > 0 in BiFeO$$ crossbar arrays.

Conclusions:

• The $$ SC metric offers a superior alternative to conventional metrics for assessing memristive crossbar arrays.
• Successfully validated the functional performance of passive crossbar arrays using self-rectifying memristive devices.
• This work paves the way for more reliable and efficient in-memory computing architectures.