High-performance one-dimensional halide perovskite crossbar memristors and synapses for neuromorphic computing

Area of Science:

• Materials Science
• Neuroscience
• Computer Engineering

Background:

• Current memristive devices for neuromorphic computing lack clear design pathways and reliable performance.
• Existing halide perovskite approaches are limited in scalability and analog programming capabilities.

Purpose of the Study:

• To systematically design and evaluate pyridinium-templated 1D halide perovskites for neuromorphic applications.
• To compare the memristive performance of (propyl)pyridinium lead iodide and (benzyl)pyridinium lead iodide.

Main Methods:

• Fabrication and characterization of 1D halide perovskite memristors in a 16x16 crossbar array.
• Evaluation of resistive switching performance, including on/off ratio, retention, and endurance.
• Development of a universal approach to map the analog programming window.

Main Results:

• (Propyl)pyridinium lead iodide exhibited superior resistive switching performance due to enhanced electronic isolation of 1D chains.
• Achieved high on/off ratio (>10^5), long-term retention (10^5 s), and high endurance (2000 cycles) on a large flexible crossbar.
• Demonstrated accurate handwritten digit recognition using spike-timing-dependent plasticity.

Conclusions:

• Pyridinium-templated 1D halide perovskites offer a robust pathway for neuromorphic computing applications.
• The developed materials and methods facilitate the design of scalable and high-performance artificial neural networks.
• This work provides a critical step towards realizing efficient and accurate brain-inspired computing systems.