Mixed metal halide perovskites (AB(XₓY₍₃₋ₓ₎)₃) have attracted significant attention due to their outstanding optoelectronic properties and potential for low-cost device fabrication. However, their practical implementation is limited by ion migration processes that strongly affect stability and performance under operation. One demonstration of this instability is light-induced halide segregation, where initially homogeneous mixed-halide compositions dynamically separate into iodide-rich and bromide-rich domains.

In this work, we demonstrate that photoinduced ionic redistribution, typically viewed as a degradation pathway, can instead be exploited as a functional mechanism for information storage. By tracking the time-dependent photoluminescence (PL) response under femtosecond pulsed excitation, we show that excitation conditions directly govern the kinetics and extent of halide migration. This enables controlled formation of distinct segregation states, each associated with a characteristic PL emission wavelength.

We systematically investigate MAPb(Br₀.₈I₀.₂)₃ and Cs₀.₀₇(FA₀.₈₃MA₀.₁₇)₀.₉₃Pb(Br₀.₅I₀.₅)₃ across a wide excitation parameter space, varying repetition rates from 10 kHz to 20 MHz and average intensities between 9 W/cm² and 1800 W/cm². By doing so, we can access reproducible, input-dependent ionic configurations corresponding to well-defined halide distributions. We demonstrate that this tunable ionic response enables parallel writing of at least nine independent, diffraction-limited, memory channels in a single thin film, with each channel supporting over 100 distinguishable states. This behavior reflects a multi-level encoding scheme far beyond conventional binary systems, corresponding to an effective 7-bit storage capability per diffraction-limited memory channel.

Importantly, the segregated states exhibit long retention times after excitation is removed, indicating metastable ionic configurations that can be read out with temporal delay. Furthermore, we show that thermal activation rapidly restores the homogeneous phase, providing a reliable pathway to erase stored information and reset the system.

Our findings highlight the potential of perovskites for high-density, rewritable optical memory and neuromorphic computing, utilizing their intrinsic dynamic responses for next-generation photonic technologies.