Optimization design method of metasurface optical coupler
By combining gradient descent algorithm with adjoint optimization, the design of metasurface optical couplers was optimized, solving the design challenges of complex geometries in silicon photonic devices, improving coupling efficiency and reducing losses, and achieving efficient optical field management.
Patent Information
- Application Number
- CN202411038122.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2026-02-03
AI Technical Summary
Existing technologies struggle to quickly and effectively optimize the complex geometries in silicon photonic devices, resulting in high crosstalk and scattering losses at waveguide intersections, which negatively impact system performance.
By combining gradient descent algorithm with adjoint optimization, a metasurface optical coupler is designed and optimized. The gradient update of dielectric constant and Gaussian filter are used to achieve fast design and efficient coupling.
This enables rapid design of metasurface couplers, reducing losses and improving coupling efficiency and signal-to-noise ratio.
Smart Images
Figure CN121454768A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of optimization methods, and specifically relates to an optimization design method for metasurface optical couplers. Background Technology
[0002] Silicon photonics offers the ability to manage light via subwavelength silicon waveguides on a chip, enabling extremely tight integration of photonic components and traditional CMOS electronics. Functions that previously required many separate components can now be implemented on a single chip, reducing cost, power consumption, and size. However, many challenges remain, one of which is effectively managing the optical field at the micro- and nano-scale. While waveguides exhibit extremely low losses and enable good light transmission throughout the chip, other functions (such as splitters, waveguide crossings, and multimode interferometers) are affected by evanescent fields present outside the waveguides and incomplete total internal reflection at silicon-oxide interfaces. These contribute to scattering losses, which are highly detrimental to the overall system performance.
[0003] Therefore, a great deal of research has been conducted in recent years on topology optimization of photonic devices. This has significantly reduced losses in Y-splitters, crosstalk at waveguide intersections, and boundary losses. Most of these optimizations are based on genetic optimization, particle swarm optimization, or hybrid methods tailored to specific problems. Heuristic optimization relies on finite parameterization of the solution space and subsequent random testing on a large number of different parameter sets. Due to the high computational cost of solving Maxwell's equations, these optimization methods are suitable for relatively simple geometries.
[0004] For more complex geometries and functions, the above methods will not be able to solve the optimization problem quickly. Therefore, it is necessary to have a more efficient way to implement topology optimization and improve coupling efficiency. Summary of the Invention
[0005] This application provides a metasurface optical coupler and its optimized design method to at least solve the above-mentioned technical problems existing in the prior art.
[0006] This application provides an optimization design method for a metasurface optical coupler, which couples obliquely incident single-mode / multimode laser light into an optical fiber / on-chip waveguide after modulation by a silicon photonic device. The method is as follows: Importing an initial dielectric constant distribution, determining system parameters and optimization objectives, and filtering and gradient binarization of the initial structure; calculating the value of the current system evaluation function, calculating the forward and adjoint fields, and solving for the gradient; performing gradient optimization, updating the dielectric constant distribution based on the current gradient, and filtering and gradient binarization of the updated structure; continuing to solve for the evaluation function of the new structure, and updating the weight coefficients in the evaluation function based on the current and previous values; calculating the forward and adjoint fields, solving for the gradient, and repeating the above process until the evaluation function meets the design requirements.
[0007] In one implementation, the optimization objective of the evaluation function includes two directions: maximum value optimization and minimum value optimization.
[0008] In one possible implementation, the evaluation function is constructed based on the following information:
[0009] The parameter β to be optimized can be related to the dielectric constant of the dielectric material by establishing a partial derivative relationship, which can be used to determine the optimization direction.
[0010] In one possible implementation, the evaluation function is expressed as:
[0011]
[0012] Where F represents the evaluation function, and βi represents the i-th parameter to be optimized.
[0013] In one possible implementation, the parameter β to be optimized is either the coupling efficiency or the signal-to-noise ratio.
[0014] In one possible implementation, when multiple β values exist, the derivative of the evaluation function with respect to the dielectric constant is expressed as:
[0015]
[0016] Where ε represents the dielectric constant of the material, σ i This represents the weighting coefficient.
[0017] In one possible implementation, the dielectric constant is updated as follows:
[0018]
[0019] Among them, G <r>This represents Gaussian smoothing filtering, where r represents the filter radius, and B...<n,iter> Let ε represent the gradient binarization function. old (x,y) represents the dielectric constant distribution in the previous cycle, ε new (x, y represent the updated dielectric constants.)
[0020] In one embodiment, the incident light field is left-handed circularly polarized, right-handed circularly polarized, linearly polarized, or naturally polarized, and the wavelength of the incident light is between 940 nm and 1550 nm. The angle at which the incident light is incident on the metasurface optical coupler is adjustable from 0° to 70°.
[0021] In one embodiment, the device material of the metasurface optical coupler is monocrystalline silicon or polycrystalline silicon, with a refractive index ranging from 3.1 to 3.5 and a thickness between 300 nm and 1 μm.
[0022] In one embodiment, the structural region is a periodically or non-periodically arranged porous or columnar structure with a minimum structural size >50nm.
[0023] Compared with the prior art, this application has the following advantages:
[0024] 1. The metasurface coupler optimized by the optimization design method of this application can achieve rapid design and reduce the design time required;
[0025] 2. The metasurface coupler optimized by the optimization design method of this application can improve coupling efficiency and reduce loss.
[0026] 3. The metasurface coupler optimized by the optimization design method of this application can improve the signal-to-noise ratio. Attached Figure Description
[0027] Figure 1 This is a schematic diagram illustrating the working principle of the coupler in the embodiments of this application;
[0028] Figure 2 This is a flowchart of the optimized design method in the embodiments of this application;
[0029] Figure 3 This is a schematic diagram of the initial structure dielectric constant in the embodiments of this application;
[0030] Figure 4 This is a schematic diagram of the optimized dielectric constant in the embodiments of this application;
[0031] Figure 5 This is a schematic diagram of the evaluation function optimization process in an embodiment of this application. Detailed Implementation
[0032] The present invention will now be described in further detail with reference to the accompanying drawings.
[0033] In the description of this application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.
[0034] This application discloses an optimized design method for metasurface optical couplers. The device designed using this method couples obliquely incident single-mode or multi-mode laser light, modulated by silicon photonics devices, into an optical fiber or on-chip waveguide. The incident light angle at the metasurface coupler is adjustable from 0° to 70°. The wavelength of the incident light can cover 940nm–1550nm. The device material of the metasurface optical coupler includes, but is not limited to, single-crystal silicon or polycrystalline silicon, with a refractive index range of 3.1–3.5. The thickness is between 300nm and 1µm. The structural regions can be periodically or non-periodically arranged aperture-like or columnar structures, with a minimum structural size >50nm.
[0035] A typical arrangement of any metasurface coupler is as follows:
[0036] like Figure 1 As shown, the metasurface optical coupler includes a substrate 12, on which a structural region 13 and an optical waveguide 14 are provided. The substrate 12 serves as a support and protection element. The structural region 13 and the optical waveguide 14 are on the same plane. The structural region 13 couples the incident light field 11 to the optical fiber or on-chip optical waveguide 14 with high efficiency. The incident light field 11 can be left-handed circularly polarized, right-handed circularly polarized, linearly polarized, or naturally polarized. The wavelength of the incident light is between 940-1550 nm, and it is incident on the silicon photonic device at an angle θ, which is adjustable between 0° and 70°.
[0037] The optimization design method for metasurface optical couplers is mainly achieved by combining gradient descent and adjoint optimization. Gradient descent is responsible for determining the optimization direction. Adjoint optimization is responsible for accurately calculating the gradient of the current evaluation function with respect to the dielectric constant, reducing computational cost and improving accuracy. Furthermore, a Gaussian filter is needed to ensure the fabricationability of the structure, and the gradient binarization function makes the final structure binary, which helps reduce manufacturing difficulty.
[0038] Specific optimization design methods are as follows: Figure 2 As shown.
[0039] The gradient optimization process first requires determining the evaluation function for the device performance, expressed as:
[0040]
[0041] Where F represents the evaluation function, and β represents the i-th parameter to be optimized. F, as the optimization objective, has two directions: maximum optimization and minimum optimization. Typical β parameters include coupling efficiency and signal-to-noise ratio. All β values can be correlated with the dielectric constant of the dielectric material through partial derivatives to determine the optimization direction.
[0042] The goal of adjoint optimization is to solve problems quickly. Where ε represents the dielectric constant of the material. When multiple β values exist, the derivative of the evaluation function with respect to the dielectric constant is expressed as:
[0043]
[0044] Where σ i This represents the weighting coefficient; terms with higher priority for optimization have higher weights, and vice versa. For example, when the signal-to-noise ratio is close to the target value, but the coupling efficiency deviates significantly from it, the weighting coefficient for coupling efficiency is increased. Common methods for solving the partial derivatives here include genetic algorithms and particle swarm optimization. The larger the structure region and the more pixels, the higher the cost of full-wave simulation. An adjoint method can obtain the complete forward propagating electric field E of the structure region through two full-wave simulations. f (x,y,z) and accompanying field E a (x, y, z). The optimization term and its derivative with respect to the permittivity can be obtained. The update method for the dielectric constant is expressed as follows:
[0045]
[0046] G here <r>This indicates Gaussian smoothing filtering, where r represents the filter radius. This value determines the minimum size of the aperture or columnar structure of the device. B<n,iter> This represents the gradient binarization function, where the value of n gradually increases with the iteration number `iter`. It forces the binarization of the dielectric constant of the structural region, where the minimum value is the background dielectric constant and the maximum value is the material dielectric constant. ε old (x,y) represents the dielectric constant distribution in the previous cycle, ε new (x,y) represents the updated dielectric constant.
[0047] Taking a specific design requirement as an example, the design process is as follows:
[0048] First, based on design requirements, aSi with a refractive index of 3.45 was selected as the optimized dielectric material, with a thickness of 800 nm. Linearly polarized light with a wavelength of 1550 nm was chosen as the light source. The incident angle θ was 45 degrees. The light was coupled to an on-chip optical waveguide, and the coupling efficiency was used as the evaluation function, expressed as:
[0049]
[0050] Where β represents the current coupling efficiency, β i This represents the coupling efficiency of the objective. According to the formula, when the coupling efficiency approaches 1, gradient optimization proceeds along the direction of gradient ascent.
[0051] First, let's give the following... Figure 3 The random initial structure shown has a continuously distributed dielectric constant. The evaluation value for the current structure is denoted as F1. Electromagnetic field simulation is performed using FDTD. The electromagnetic field distribution E in the positive structure region is solved separately. f (x,y,z) and accompanying field E a (x,y,z) and solve for the gradient value:
[0052]
[0053] The structural region is updated using this gradient value, and the specific update method can be expressed as follows:
[0054]
[0055] G here <r>This indicates Gaussian smoothing filtering, where r represents the filter radius. This value determines the minimum size of the aperture or columnar structure of the device. Considering fabrication capabilities, the value of r is chosen to be 20nm. B<n,iter> This represents the gradient binarization function, where the value of n gradually increases with the iteration number `iter`. It forces the binarization of the dielectric constant of the structural region, where the minimum value is the background dielectric constant and the maximum value is the material dielectric constant. ε old (x,y) represents the dielectric constant distribution in the previous cycle, ε new (x,y) represents the updated dielectric constant.
[0056] Then, perform another forward full-wave simulation using the current dielectric constant, and re-evaluate the current coupling efficiency as F2. Repeat the above process until the structural region is completely binarized as shown. Figure 4 The distribution is shown. The trend of the evaluation function value changes throughout the optimization process is as follows. Figure 5 As shown in the figure, this is an example of maximum value optimization. After optimization, the result will be as follows: Figure 1 The diagram shows the working operation of the coupler device.
[0057] The above are merely specific embodiments of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.< / r> < / r> < / r>
Claims
1. An optimized design method for a metasurface optical coupler, characterized in that, The obliquely incident single-mode / multimode laser is modulated by silicon photonic devices and coupled into an optical fiber / on-chip waveguide; The method is as follows: Import the initial distribution of the initial dielectric constant, determine the system parameters and optimization objectives, and perform filtering and gradient binarization on the initial structure; Calculate the value of the current system evaluation function, perform calculations of the positive and adjoint fields, and solve for the gradient; Gradient optimization is performed, the distribution of dielectric constant is updated based on the current gradient, and the updated structure is filtered and the gradient is binarized. Continue solving the evaluation function for the new structure, and update the weight coefficients in the evaluation function based on the current value and the previous value; Perform calculations for the positive and adjoint fields, solve for the gradient, and repeat the above process until the evaluation function meets the design requirements.
2. The optimized design method for metasurface optical couplers according to claim 1, characterized in that: The optimization objectives of the evaluation function include two directions: maximization and minimization.
3. The optimized design method for metasurface optical couplers according to claim 2, characterized in that, The evaluation function is constructed based on the following information: The parameter β to be optimized can be related to the dielectric constant of the dielectric material by establishing a partial derivative relationship, which can be used to determine the optimization direction.
4. The optimized design method for the metasurface optical coupler according to claim 3, characterized in that, The evaluation function is expressed as follows: Where F represents the evaluation function, and βi represents the i-th parameter to be optimized.
5. The optimized design method for the metasurface optical coupler according to claim 3, characterized in that: The parameter β to be optimized is either the coupling efficiency or the signal-to-noise ratio.
6. The optimized design method for a metasurface optical coupler according to claim 3, characterized in that, When multiple β values exist, the derivative of the evaluation function with respect to the dielectric constant is expressed as: Where ε represents the dielectric constant of the material, σ i This represents the weighting coefficient.
7. The optimized design method for a metasurface optical coupler according to claim 6, characterized in that, The update method for the dielectric constant is expressed as follows: Among them, G <r> This represents Gaussian smoothing filtering, where r represents the filter radius, and B...<n,iter> Let ε represent the gradient binarization function, εold(x,y) represent the dielectric constant distribution in the previous iteration, and εnew(x,y) represent the updated dielectric constant.< / r> 8. The optimized design method for metasurface optical couplers according to claim 1, characterized in that: The incident light field is left-handed circularly polarized, right-handed circularly polarized, linearly polarized, or naturally polarized. The wavelength of the incident light is between 940nm and 1550nm. The angle at which the incident light is incident on the metasurface optical coupler is adjustable from 0° to 70°.
9. The optimized design method for a metasurface optical coupler according to claim 1, characterized in that: The device material of the metasurface optical coupler is monocrystalline silicon or polycrystalline silicon, with a refractive index range of 3.1-3.5 and a thickness between 300nm and 1um.
10. The optimized design method for a metasurface optical coupler according to claim 1, characterized in that: The structural regions are periodically or non-periodically arranged porous and columnar structures, with a minimum structural size >50nm.