Silicon Waveguide Crossing for Low-Loss Photonic Switching

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Silicon photonic switch matrices face limitations due to high optical loss and crosstalk in waveguide crossings, which restrict the size of switch ports and hinder the development of large-scale, low-cost, high-density photonic devices.

Innovation Solution

A silicon waveguide crossing design that uses adiabatic and non-adiabatic transitions to transform single-mode light into multimode light, creating a converging beam that self-images across the gap, achieving low-loss and low-crosstalk by supporting multiple even modes and utilizing CMOS-compatible fabrication processes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional waveguide crossings are used in silicon photonic switch matrices, then the device can be fabricated with standard processes, but high optical loss and crosstalk occur which limit switch port count size

Engineering Contradiction:
Improveoptical loss and crosstalk performanceVSAvoidswitch matrix size capability
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The waveguide crossing is divided into multiple functional sections: an input waveguide section, a first multimode section with expanded width, a second multimode section, and an output waveguide section. Each section has specific dimensions optimized for its function, transforming the light progressively through mode coupling and self-imaging effects to achieve low loss and crosstalk while maintaining scalability for large switch matrices

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The waveguide width is varied across different sections to control optical mode behavior. The first multimode section has a width W1 greater than the input waveguide width W0, and the second multimode section has width W2 greater than W1. These parameter changes enable mode expansion, self-imaging, and efficient light transfer, reducing optical loss and crosstalk to enable larger switch matrices

Inventive Principle:
Principle #35Parameter changes

2Productivity

If waveguide crossings are increased to enable larger switch matrices, then higher port count is achieved, but total insertion loss and optical crosstalk increase which limits further scaling

Engineering Contradiction:
Improveswitch port count sizeVSAvoidtotal insertion loss and optical crosstalk
Core Design Contradiction:
ProductivityVSLoss of energy

Solution Approach 1:

The patent employs self-imaging effects where the optical field distribution in the first multimode section is replicated in the second multimode section. This copying mechanism ensures that the light pattern is preserved through the crossing with minimal distortion and loss, enabling each waveguide crossing to maintain high efficiency even as the overall switch matrix scales to larger port counts

Inventive Principle:
Principle #26Copying

Solution Approach 2:

By segmenting the crossing into optimized sections with controlled mode coupling, the patent achieves low loss per crossing. This modular approach allows multiple crossings to be cascaded in large switch matrices without cumulative loss becoming prohibitive, thereby enabling higher port count sizes while maintaining acceptable total insertion loss and crosstalk levels

Inventive Principle:
Principle #1Segmentation

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The waveguide crossing exhibits very low loss (less than 20 mdB over a large bandwidth) and high efficiency (99.95% light transfer), enabling the creation of large-scale, low-loss, low-crosstalk waveguide crossings suitable for high-density photonic switch matrices.

Implementation Method 1

A silicon waveguide crossing design that uses adiabatic and non-adiabatic transitions to transform single-mode light into multimode light

Methodology Applied
Scientific EffectAdiabatic transition:

Implementation Method 2

The widths of the two waveguides and the length of the multimode waveguide are selected such that particular mode coupling between the waveguides and multimode interference (MMI) effects in the multimode waveguide cause the latter to output a converging beam of light

Methodology Applied
Scientific EffectMultimode interference (MMI):

Data Source

PatentEP3411737B1A waveguide crossing
Publication Date: 2021.07.28 HUAWEI TECH CO LTD
  • EP3411737B1 patent drawingFigure 1
  • EP3411737B1 patent drawingFigure 2
  • EP3411737B1 patent drawingFigure 3

AI summary

A waveguide crossing includes a first waveguide(200) and a second waveguide(300) intersecting the first waveguide(200) such that a gap(G) equal to a width of the second waveguide(300) is formed in the first waveguide(200), the second waveguide(300) having a centerline(C L) defining a plane of symmetry. The first waveguide(200) has a first waveguide section(202) through which a single optical mode propagates, followed by a first non-adiabatic diverging taper(204), followed by a second waveguide section(206) wider than the first waveguide section(202) through which two even-order optical modes propagate, followed by a second non-adiabatic diverging taper(208), followed by a third waveguide section(210) wider than the second waveguide section(206) through which three even-order optical modes propagate. The three even-order modes synthesize to form a quasi-Gaussian beam that self-replicates symmetrically across the gap(G), thereby providing a low-loss waveguide crossing useful for photonic switching.