Overview
A novel wavelength-addressable diffractive optical network has been introduced, converting illumination wavelength into a high-dimensional control parameter for programmable 2D beam steering. This passive architecture employs cascaded spatially optimized diffractive layers, co-designed through deep learning, to rapidly map specific wavelengths to pre-defined output angles. This approach aims to eliminate the reliance on mechanical scanning or electronic phase control for beam steering.
Research Context
Conventional single-layer dispersive optical elements are inherently limited to 1D linear mapping due to their physical constraints. The presented framework addresses this limitation by leveraging complex wavefront transformations. It positions illumination wavelength as an intrinsic addressing key for achieving arbitrary 2D beam steering. This contrasts with traditional gratings that restrict wavelength routing to linear trajectories.
Approach
The core of the proposed system is a passive architecture composed of cascaded diffractive layers. These layers were jointly designed using deep learning techniques to achieve spatial optimization for their specific function. The deep learning process facilitated the creation of a network capable of nonlocal wavefront transformations, which are critical for enabling arbitrary wavelength-to-angle mappings across a 2D field of view.
The research involved both numerical demonstrations and experimental validations:
- Numerical Demonstration: The system was numerically tested for wavelength-controlled beam steering across 625 wavelength channels, spanning the 400-750 nm range. This simulation aimed to achieve a 25 x 25 array of independently addressable beam positions.
- Experimental Validation: The proposed framework underwent experimental validation in two distinct spectral regimes: terahertz and visible. For terahertz frequencies, 3D fabricated passive diffractive layers were utilized to demonstrate wavelength-multiplexed beam steering. In the visible spectrum, phase-only spatial light modulators were employed for similar validation.
Findings
The numerical demonstration indicated the capability for wavelength-controlled beam steering, realizing a 25 x 25 array of independently addressable beam positions. This was achieved with subwavelength positioning accuracy and high channel fidelity across 625 wavelength channels ranging from 400 nm to 750 nm. The experimental validations, conducted in both the terahertz and visible spectral regimes, further confirmed the framework's ability to perform wavelength-multiplexed beam steering using the specified passive diffractive layers and spatial light modulators, respectively.
A key finding is that the diffractive network executes nonlocal wavefront transformations. This capability differentiates it from conventional gratings by enabling arbitrary wavelength-to-angle mappings across a 2D field of view, thereby overcoming the 1D linear mapping restriction.
Why This Matters
This wavelength-addressable diffractive architecture offers a compact and scalable paradigm for high-speed programmable beam steering. Potential applications for this technology include optical communications, routing, imaging, sensing, and emerging photonic information-processing systems.