Reconfigurable Multistate Optical Memory in Mixed Halide Perovskites.

Area of Science:

• Materials Science
• Photonics
• Neuromorphic Computing

Background:

• The von Neumann bottleneck hinders data-intensive computing, including large language model training.
• Optical memory in materials capable of computation offers a potential solution to this bottleneck.

Purpose of the Study:

• To report a reconfigurable multistate optical memory based on mixed halide perovskites (MHPs).
• To explore the use of MHPs for storing and processing optical information, addressing challenges in data-intensive computing.

Main Methods:

• Utilizing the dynamic ionic behavior of mixed halide perovskites under femtosecond pulsed excitation.
• Encoding memory states via photoluminescence (PL) emission wavelength, controlled by excitation power and repetition rate.
• Demonstrating rapid erasure of memory states using thermal pulses.

Main Results:

• Achieved rewritable, multistate optical memory where PL emission wavelength encodes excitation parameters.
• Observed long-lived retention of memory states with rapid thermal erasure.
• Demonstrated analog-like synaptic behavior and spatially resolved writing for high-density memory potential.

Conclusions:

• Mixed halide perovskites function as active photonic materials for optical information storage and processing.
• This MHP-based approach offers a scalable platform for low-power, high-bandwidth memory and neuromorphic computing.
• The findings present a promising direction for overcoming the von Neumann bottleneck in future computing architectures.