A Self-Powered Polarization-Sensitive Neuromorphic Vision Device Enabled by Laser-Induced Symmetry Engineering

The trend toward ultimate edge computing systems requires a paradigm shift from integrating individual components to intrinsic multifunctional devices. However, a primary challenge lies in engineering a single device capable of harvesting energy, sensing complex environmental information, and performing on-device computation without overcomplicated device configuration. Herein, we address the challenge by developing a single-step, facile, ultrafast laser-induced symmetry engineering (LISE) process to fabricate a self-powered, polarization-sensitive neuromorphic vision device in a single MoTe₂-based architecture. By engineering the localized phase transition, we achieve simultaneous symmetry engineering of both the energy band and crystal structures. This dual asymmetry allows for self-powered operation via a built-in photovoltaic effect and polarization sensitivity from the engineered crystal anisotropy. Leveraging the photovoltaic volatile memory, an engineered FeFET operating as a physical reservoir achieves fully self-powered and all-optical reservoir computing for underwater imaging. Computation can be actively modulated by the polarization state of incident light and preconditioned by gate voltage, revealing a powerful hardware-level method for tuning computation. The proposed LISE approach demonstrates the ultrafast laser as a powerful tool for the local manipulation of material-symmetry-related properties and facilitates the creation of high-performance multifunctional neuromorphic systems.