🔬 optics

Silicon Photonic Beam Steerer Based on Metalens Focal Plane Array

This paper presents a monolithically integrated silicon photonic beam steerer that combines a metalens focal plane array with thermo-optic prisms to achieve continuous, blind-spot-free steering over a 62° field of view, thereby overcoming the size limitations of external lenses and the resolution constraints of discrete switching for LiDAR and free-space optical communications.

Original authors: Chung-Yu Hsu, Ping-Yen Hsieh, Hsun-Sung Chiu, Li-Jun Tung, Chieh-Chih Yu, Ko-Chi Chen, Yu-Heng Hong, Hao-Chung Kuo, Chi-Wai Chow, You-Chia Chang

Published 2026-03-30

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Chung-Yu Hsu, Ping-Yen Hsieh, Hsun-Sung Chiu, Li-Jun Tung, Chieh-Chih Yu, Ko-Chi Chen, Yu-Heng Hong, Hao-Chung Kuo, Chi-Wai Chow, You-Chia Chang

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to shine a flashlight on a specific spot on a wall, but you need to do it incredibly fast, without any moving parts, and you want the beam to be able to point anywhere on that wall smoothly, not just jump from one spot to another.

This is the challenge scientists face with LiDAR (the "eyes" of self-driving cars) and free-space optical communication (sending data through the air like invisible lasers). Traditional methods use big, clunky mechanical mirrors that spin or tilt. They are slow, break easily, and are too big for tiny chips.

This paper introduces a new, tiny "smart flashlight" built entirely on a silicon chip. Here is how it works, broken down into simple concepts:

1. The Problem with Old "Smart Flashlights"

Previous attempts to make these chips had two main flaws:

The "Jumping" Problem: Imagine a row of 15 tiny flashlights. If you turn them on one by one, the beam jumps from spot to spot. If you need to track a moving car, the beam might "miss" the car while it jumps to the next spot. These are called blind spots.
The "External Lens" Problem: To make the light travel straight and far, you usually need to glue a big, heavy glass lens onto the chip. This makes the whole system bulky and expensive to assemble.

2. The New Solution: A "Magic Lens" and a "Heat Steering Wheel"

The researchers built a device that solves both problems using two clever tricks:

Trick A: The "Magic Lens" (The Metalens)

Instead of gluing a big glass lens on top, they printed a microscopic lens directly onto the chip.

Analogy: Think of it like drawing a tiny, perfect magnifying glass directly onto a piece of paper using a super-fine pen. This "metalens" is made of thousands of microscopic slots (smaller than a hair).
What it does: It takes the light coming out of the chip and instantly straightens it out, sending it flying in a straight line. Because it's built right into the chip, the whole device is flat, tiny, and doesn't need any extra parts.

Trick B: The "Heat Steering Wheel" (Thermo-Optic Prisms)

This is the secret sauce that fixes the "jumping" problem.

Analogy: Imagine you are driving a car on a straight road. To turn slightly left or right, you don't need to jump to a new lane; you just gently turn the steering wheel.
How it works: The chip has tiny heaters (like miniature hair dryers) placed on the path of the light. When they heat up a specific triangular area of the silicon, the silicon gets slightly "thicker" for the light (its refractive index changes). This acts like a prism, bending the light beam smoothly.
The Result: Instead of the beam jumping from spot to spot, the heat allows the beam to slide smoothly between the spots. This eliminates the "blind spots" and lets the beam track moving objects perfectly.

3. Why This is a Big Deal

Triple the Resolution: Because they can slide the beam smoothly between the 15 main spots, they can actually hit 45 distinct points instead of just 15. It's like having a 15-key piano that can play 45 different notes.
No Blind Spots: The beam never stops moving; it flows continuously. This is crucial for self-driving cars that need to see a pedestrian stepping out from behind a car right now, not 10 milliseconds later.
Super Clean Beam: The light stays focused and doesn't scatter (which creates "noise" or "ghost images"). The beam is so clean that the unwanted side-light is 19 times weaker than the main beam.
Fast and Efficient: It can switch directions in about 10 microseconds (faster than a blink of an eye) and uses very little power.

The Bottom Line

The team has created a solid-state, chip-sized laser pointer that can sweep a wide area (62 degrees) smoothly and accurately. By printing the lens directly onto the chip and using heat to steer the beam, they have made a device that is smaller, more reliable, and smarter than anything currently used in self-driving cars or high-speed air-to-air internet.

It's a major step toward making the "eyes" of our future robots and cars as small and cheap as a microchip, but as powerful as a giant radar system.

1. Problem Statement

The paper addresses critical limitations in current solid-state beam-steering technologies used for LiDAR and Free-Space Optical (FSO) communications:

Bulkiness: Conventional Focal Plane Arrays (FPAs) require external collimation lenses, hindering chip-scale integration and increasing packaging costs.
Discrete Steering & Blind Spots: Standard FPAs rely on discrete switching between emitter elements, creating "blind spots" (gaps) in the Field of View (FOV). This makes them unsuitable for applications requiring continuous tracking, such as FSO link establishment.
Resolution vs. Complexity: The number of resolvable points in both Optical Phased Arrays (OPAs) and FPAs is typically limited to the number of array elements ( $N$ ). Increasing resolution requires massive arrays, which complicates control circuitry and increases chip area.
Calibration Issues: OPAs require complex, time-consuming phase calibration for thousands of elements, whereas FPAs are robust but lack continuous control.

2. Methodology

The authors propose a monolithically integrated silicon photonic beam steerer that combines three key components to overcome these limitations:

On-Chip Metalens FPA:
- Instead of an external lens, a metasurface metalens is fabricated directly on the silicon chip.
- The metalens consists of subwavelength slots (500 nm period, 140 nm width) designed to provide an aberration-free F-theta phase profile.
- It collimates light from 15 discrete waveguide channels (part of a 1×16 optical switch network) into specific propagation angles within the silicon slab.
Thermo-Optic (TO) Prisms for Fine-Tuning:
- To eliminate blind spots and enable continuous steering, the device integrates TO prisms.
- These are formed by tungsten microheaters arranged in triangular patterns over the silicon slab waveguide.
- By applying current, the heaters locally increase the refractive index of the silicon (via the thermo-optic effect), creating a dynamic prism that deflects the beam.
- Two independent TO prisms allow fine-tuning in both positive and negative directions relative to the coarse steering angle.
Shallow-Etched Grating Emitter:
- A grating emitter diffracts the guided light from the slab into free space.
- It enables wavelength-dependent steering in the altitude ( $\theta$ ) direction, while the FPA and TO prisms control the azimuthal ( $\phi$ ) direction.

Device Fabrication:
The device was fabricated on a Silicon-on-Insulator (SOI) wafer (220 nm top silicon) using electron-beam lithography and plasma etching. It includes a 1.3-µm SiO2 cladding and Ti/W microheaters for thermal actuation.

3. Key Contributions

Monolithic Integration: Successfully integrated the collimation lens (metalens) directly onto the chip, eliminating the need for bulky external optics and complex alignment packaging.
Continuous Steering without Blind Spots: Combined discrete coarse steering (via FPA switching) with continuous fine-tuning (via TO prisms) to achieve seamless beam coverage.
Enhanced Resolution: Demonstrated a method to triple the number of resolvable points ( $3N$ ) using only $N$ waveguide channels, significantly improving resolution without increasing the physical array size.
Robust Calibration: Retained the inherent calibration simplicity of FPAs (power monitoring only) while adding continuous control, avoiding the complex phase calibration required by OPAs.

4. Experimental Results

Field of View (FOV): Achieved continuous beam steering over an azimuthal FOV of 62°.
Resolution Enhancement: Using only 15 waveguide channels, the system generated 45 resolvable points (a 3-fold increase compared to the channel count).
Beam Quality:
- Maintained high beam quality with an average Sidelobe Suppression Ratio (SLSR) of 19 dB (significantly better than the theoretical 13 dB limit of non-apodized OPAs).
- Average divergence angle in the azimuthal direction was 1.48°.
Fine-Tuning Capability:
- The TO prisms enabled beam displacement up to 168% of the beam's Full Width at Half Maximum (FWHM) for specific channels, effectively filling the gaps between discrete FPA angles.
Speed and Power:
- Switching Speed: Rise/fall times were measured at 17.0 µs / 8.8 µs for coarse steering and 15.1 µs / 22.3 µs for fine-tuning.
- Power Consumption: The total electrical power consumption was 1.46 W (160 mW for the switch network, up to 1.3 W for the TO prism).
- Insertion Loss: Measured at 16.5 dB.
Demonstration: The system successfully traced a trajectory forming the letter "H" in the far field, proving programmable active beam steering.

5. Significance

This work presents a major advancement in solid-state beam steering for next-generation LiDAR and FSO systems. By solving the "blind spot" problem of FPAs and the "calibration complexity" of OPAs, the proposed architecture offers a compact, robust, and high-resolution solution.

Scalability: The architecture is scalable; increasing the channel count ( $N$ ) and metalens focal length ( $f$ ) proportionally can further increase the number of resolvable points.
Application Impact: The ability to continuously track and align beams is critical for dynamic FSO links (e.g., between UAVs) and high-fidelity 3D mapping for autonomous vehicles.
Integration: The monolithic integration of the metalens paves the way for fully integrated, chip-scale optical systems that are easier to manufacture and package.