# US 2026/0148054 A1 · Three-dimensional heterogeneous integrated optical computing chip

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0148054 A1
- Document ID: US20260148054A1
- Status: Published application
- Kind: A1
- Application number: 19/186,506
- Filed: 2025-04-22
- Published: 2026-05-28
- Earliest claimed: 2024-11-28
- Assignees: Huazhong University of Science and Technology
- Applicants: Huazhong University of Science and Technology
- Inventors: Jian WANG; Yuanjian Wan
- Web page: https://patentgrep.com/patent/US20260148054A1

## CPC

- G06N 3/067: using optical means
- G06N 3/0675: {using electro-optical, acousto-optical or opto-electronic means}
- G06E 3/001: {Analogue devices in which mathematical operations are carried out with the aid of optical or electro-optical elements}
- G06E 3/005: {using electro-optical or opto-electronic means}
- G06N 3/044: Recurrent networks, e.g. Hopfield networks
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/047: Probabilistic or stochastic networks
- G06N 3/048: Activation functions
- G06N 3/063: using electronic means
- G06N 3/065: Analogue means
- G06N 3/08: Learning methods
- G06N 3/082: modifying the architecture, e.g. adding, deleting or silencing nodes or connections
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/09: Supervised learning
- G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06E: OPTICAL COMPUTING DEVICES (digital storage using optical elements G11C13/04)

## Abstract

Provided is a three-dimensional (3D) heterogeneous integrated optical computing chip, which belongs to the field of optical computing. The computing chip is formed by three parts: a spatial diffraction optical neural network chiplet, a planar matrix reconfigurable optical neural network component, and a planar switchable optical switch module. Through a 3D heterogeneous integration method, one or more spatial diffraction optical neural network chiplets, the planar matrix reconfigurable optical neural network component, and the planar switchable optical switch module are combined together to achieve efficient multi-task optical computing processing. The disclosure utilizes the combination of the spatial diffraction optical neural network and the planar matrix reconfigurable optical neural network, breaking through the conventional optical computing architecture, which has broad application prospects in the field of optical computing and fills the gap in the relevant technical field.

## Claims

### Claim 1

A three-dimensional (3D) heterogeneous integrated optical computing chip, comprising a spatial diffraction optical neural network chiplet, a planar matrix reconfigurable optical neural network component, and a planar switchable optical switch module, the spatial diffraction optical neural network chiplet, the planar matrix reconfigurable optical neural network component, and the planar switchable optical switch module are integrated through a 3D heterogeneous integration method, the spatial diffraction optical neural network chiplet comprises no less than 1 layer of multi-phase plane with propagation layers between the multi-phase plane, the planar matrix reconfigurable optical neural network component comprises a linear matrix network module and a detector array connected in sequence,
the spatial diffraction optical neural network chiplet is configured to perform feature extraction and dimensionality reduction on a task to be processed so as to obtain a feature output result, the feature output result is input to the planar switchable optical switch module for state configuration through a coupling manner, and the planar matrix reconfigurable optical neural network component is configured to perform operations on the feature output result after state configuration.

### Claim 2 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein pixel points of the multi-phase plane serve as neurons, with approximately full connections established between layers through optical diffraction, thereby realizing linear operations.

### Claim 3 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein the 3D heterogeneous integration method comprises transfer printing, direct bonding, or femtosecond laser 3D direct writing.

### Claim 4 (depends on 2)

The 3D heterogeneous integrated optical computing chip according to claim 2, wherein the multi-phase plane is a metasurface unit or a diffraction optical unit.

### Claim 5 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein the linear matrix network module comprises a Mach-Zehnder interferometer, a microring array, or a multimode interference waveguide.

### Claim 6 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein the planar switchable optical switch module comprises a coupler array and an optical switch, the coupler array is configured to receive feature optical signals from different spatial diffraction optical neural network chiplets and convert spatial light modes to waveguide modes, and the optical switch is configured to implement a working state configuration of the planar switchable optical switch module and selectively output feature output results from the different spatial diffraction optical neural network chiplets to the planar matrix reconfigurable optical neural network component.

### Claim 7 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein a processing material of the spatial diffraction optical neural network chiplet comprises silicon, silicon nitride, metal, silicon dioxide, and polymer, and a processing material of the planar matrix reconfigurable optical neural network component and the planar switchable optical switch module comprises silicon, germanium, silicon dioxide, indium phosphide, gallium arsenide, thin film lithium niobate, polymer, phase change material, or a mixture of the above materials.

### Claim 8 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein the spatial diffraction optical neural network chiplet further comprises a first nonlinear activation function layer, the first nonlinear activation function layer is between layers of the multi-phase plane, the first nonlinear activation function layer is configured to implement nonlinear operation, a material of the nonlinear activation function layer is graphene, phase change material, or reverse saturable absorber materials, the planar matrix reconfigurable optical neural network component further comprises a second nonlinear activation function layer, the second nonlinear activation function layer is between the matrix network module and the detector array, and a material of the second nonlinear activation function layer is a germanium-silicon material or an SOA material.

### Claim 9 (depends on 3)

The 3D heterogeneous integrated optical computing chip according to claim 3, wherein the 3D heterogeneous integration method is through multi-chiplet integration, directly bonding the spatial diffraction optical neural network chiplet onto a planar chip formed by the planar matrix reconfigurable optical neural network component and the planar switchable optical switch module; alternatively, the different spatial diffraction optical neural network chiplets, the planar matrix reconfigurable optical neural network component, and the planar switchable optical switch module are bonded to preset areas of an optical interposer, and connection among the spatial diffraction optical neural network chiplets, the planar matrix reconfigurable optical neural network component, and the planar switchable optical switch module is realized through the optical interposer.

### Claim 10 (depends on 1)

The 3D heterogeneous integrated optical computing chip according to claim 1, wherein the planar switchable optical switch module dynamically switches routing connections between different spatial diffraction optical neural network chiplets and the planar matrix reconfigurable optical neural network component according to task types; simultaneously, a matrix weight configuration of the planar matrix reconfigurable optical neural network component is dynamically adjusted to adapt to computational requirements of different tasks; further, comprising combining feedback control and intelligent algorithms to adaptively optimize and adjust a routing state of the planar switchable optical switch module and matrix parameters of the planar matrix reconfigurable optical neural network component, task-driven self-learning and self-configuring multi-task configuration are realized.
