Method and device for optimizing vertical super-structure grating coupler and medium

Abstract

The invention relates to a method, equipment and medium for optimizing a vertical super-structure grating coupler, and the method comprises the steps: initializing the dielectric constant distribution in a two-dimensional design region, and stretching the dielectric constant distribution into the etching layer height; performing continuous topological optimization based on the initialized dielectric constant distribution; in the continuous topological optimization process, smoothing the dielectric constant distribution according to a preset filtering radius; when the change of the target function is within a predefined threshold range, introducing random disturbance; converting the dielectric constant after continuous topological optimization into two discrete values; according to the discrete dielectric constant distribution obtained through optimization, the geometric structure of the vertical super-structure grating coupler is designed; and carrying out performance simulation on the designed vertical super-structure grating coupler, and evaluating the coupling efficiency and bandwidth parameters of the vertical super-structure grating coupler. The invention provides an excellent input / output interface solution for the realization of a large-scale ultra-compact super-structure photon integrated circuit.

Description

Technical Field

[0001] The present invention relates to the field of meta-photonic devices and vertical grating couplers, and in particular to a method, device and medium for optimizing vertical meta-grating couplers. Background Art

[0002] As an important optical input and output interface, grating coupler has become a widely used photonic device in the fields of laser radar, optical communication, sensors, etc. due to its advantages such as flexible layout, simple optical path design, high integration and broadband characteristics. Among them, the vertical grating coupler reduces the strict requirements on the position of the light source, making it easier to form a large-scale ultra-compact photonic integrated circuit, thereby realizing an ultra-miniaturized system. This feature of the vertical grating coupler is particularly important for the large-scale transceiver antenna array of the photonic integrated laser radar. However, the low coupling efficiency of the traditional vertical grating coupler has always been a difficult problem to solve. This is mainly because the coupling distance between the light source and the grating part is short after the vertical incidence, and the incident light is very easy to leak from the bottom of the grating, especially when the coupling effect of the designed grating structure is weak, only a small amount of light is coupled to the grating structure, and then routed to the output waveguide, resulting in very low coupling efficiency. In addition, in order to achieve low optical transmission loss, it is necessary to ensure that the light is transmitted in single mode in the waveguide. The width of the silicon optical route waveguide in the photonic integrated circuit is generally 0.5μm. Therefore, it is generally necessary to connect a conical spot converter (except for the focusing grating coupler) between the vertical grating coupler and the optical routing waveguide. Only when the conical spot converter is long enough can the two achieve low-loss adiabatic conversion. For example, the most commonly used 10μm grating coupler needs to be efficiently converted to a 0.5μm wide single-mode waveguide through a conical spot converter with a length of several hundred μm. This is undoubtedly a huge obstacle to the high integration of grating couplers. Since the curvature of the focusing grating coupler is difficult to maintain stable during processing, it has a great impact on its performance, so it is relatively limited in practical applications.

[0003] In recent years, the concept of metaphotonics and the application of inverse design methods have given photonic device design more design freedom, and various unprecedented metaphotonic devices with ultra-compact size and high performance have emerged. Among them, the metagrating coupler breaks the traditional periodic structure and realizes a complex spatial structure within sub-wavelength accuracy, making ultra-compact and high-performance grating couplers possible. Commonly used inverse design methods include DBS, genetic algorithm, shape topology optimization, deep learning, adjoint method, etc. Among them, the DBS algorithm is the simplest and can find the global optimal solution, but the calculation is time-consuming and the optimization efficiency is extremely low. Shape topology optimization only optimizes the edge of the structure, and the design freedom is limited. Deep learning requires a lot of data collection before optimization, and the optimization effect is general. The adjoint rule is an efficient optimization algorithm. For devices with single-port input and single-port output, each iteration only requires one forward simulation and one adjoint simulation to obtain the global optimized structure. Therefore, in recent years, many scholars have applied it to the optimization research of metavertical grating couplers. However, it is well known that this method is sensitive to the initial structure and is prone to local optimal convergence, which has certain limitations in device optimization. Summary of the invention

[0004] The present invention provides a method, device and medium for optimizing a vertical metagrating coupler, the purpose of which is to solve the problems of low coupling efficiency and large footprint of traditional vertical grating couplers and to improve the integration and performance of photonic integrated circuits.

[0005] To achieve the above object, the first aspect of the present invention provides a method for optimizing a vertical metagrating coupler, comprising the following steps:

[0006] Initializing the dielectric constant distribution in the two-dimensional design area so that the dielectric constant is randomly distributed between silicon and silicon dioxide or between silicon and air, and stretching it to the height of the etched layer;

[0007] Based on the initialized dielectric constant distribution, performing continuous topology optimization;

[0008] In the continuous topology optimization process, the dielectric constant distribution is smoothed according to the preset filter radius;

[0009] When the change of the objective function is within a predefined threshold range, a random perturbation is introduced to adjust the dielectric constant distribution within the design area;

[0010] Converting the dielectric constant after continuous topology optimization into two discrete values, the conversion comprising applying a binarization algorithm to limit the dielectric constant distribution to two discrete values ​​of silicon and silicon dioxide or silicon and air;

[0011] According to the optimized discrete dielectric constant distribution, the geometric structure of the vertical metagrating coupler is designed;

[0012] The performance of the designed vertical metagrating coupler is simulated to evaluate its coupling efficiency and bandwidth parameters.

[0013] Furthermore, the continuous topology optimization includes:

[0014] Perform forward simulation to calculate the electromagnetic field distribution within the design area;

[0015] Based on the forward simulation results, an objective function is established to evaluate the mode overlap integral between the output waveguide fundamental mode and the incident Gaussian light source;

[0016] performing adjoint simulations to determine the sensitivity of the objective function with respect to the dielectric constant distribution within the design region;

[0017] Using the sensitivity information, adjusting the dielectric constant distribution within the design region to optimize the objective function;

[0018] The above steps are repeated until the objective function converges or the predetermined number of iterations is reached.

[0019] Furthermore, after the forward simulation and the adjoint simulation are completed, the high-frequency noise part in the design area is smoothed according to a preset filtering radius, and the smoothing method includes:

[0020] According to the filter radius, the dielectric constant value of the high-frequency noise part is modified so that the dielectric constant within the filter radius is the same value;

[0021] After filtering is complete, update the dielectric constant distribution in the design area and proceed to the next steps.

[0022] Furthermore, when the change of the objective function is within a predefined threshold range, the method of introducing random perturbations to adjust the dielectric constant distribution within the design area includes:

[0023] Calculate the change in quality factor between consecutive iterations;

[0024] comparing the quality factor change to a predefined threshold range;

[0025] When the quality factor change is within the threshold range, randomly perturbing the dielectric constant distribution within the design area based on a random perturbation method;

[0026] adjusting the dielectric constant distribution within the design area according to the random perturbation;

[0027] After introducing the random perturbations, the forward and adjoint simulations are continued.

[0028] Furthermore, the random disturbance includes:

[0029] Randomly select a number of dielectric constant cells;

[0030] The dielectric constant value of the selected cell is randomly increased or decreased, and the dielectric constant value after random increase or decrease is limited to the dielectric constant range between silicon and silicon dioxide or between silicon and air;

[0031] Repeat the above steps multiple times to achieve random perturbation of the overall dielectric constant distribution;

[0032] After the random perturbation is completed, the dielectric constant distribution within the design region is updated for use in subsequent optimization steps.

[0033] Further, the method of converting the dielectric constant after continuous topology optimization into two discrete values ​​includes:

[0034] Setting two target discrete values ​​of dielectric constant, corresponding to the dielectric constants of silicon and silicon dioxide or silicon and air, respectively;

[0035] For each dielectric constant cell in the optimization region, its dielectric constant is assigned to one of the target discrete values ​​based on how close its current value is to the two target discrete values;

[0036] A binarization algorithm is applied to transform the continuous dielectric constant distribution in the design area into a distribution containing only the above two target discrete values.

[0037] Furthermore, according to the optimized discrete dielectric constant distribution, the method for designing the geometric structure of the vertical metagrating coupler includes:

[0038] extracting an optimized dielectric constant distribution pattern;

[0039] Based on the extracted dielectric constant distribution pattern, the geometric structure of the vertical metagrating coupler is defined.

[0040] Furthermore, the criterion for the convergence of the objective function is:

[0041] When the objective function value is equal to a preset threshold, the objective function is considered to have converged, where the threshold is 1, indicating that the optimization process is stopped when the objective function value is equal to this value;

[0042] When the rate of change of the objective function is within a predetermined threshold range, the objective function is considered to have converged, where the predetermined threshold is [0,10 -8 ], indicating that when the rate of change of the objective function is within this interval, the optimization process is stopped.

[0043] Furthermore, in order to prevent abnormal situations such as infinite execution of optimization, the number of optimization iterations is limited. When the number of optimization iterations reaches the preset value n, the optimization is forced to stop regardless of whether it converges or not. n is set to 500, which means that the optimization process is stopped after 500 generations of optimization iterations.

[0044] To achieve the above objectives, the first aspect of the present invention provides an electronic device, comprising a processor and a memory, wherein the processor is used to implement the steps of the method for optimizing a vertical metagrating coupler when executing a computer program stored in the memory.

[0045] To achieve the above objectives, the first aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for optimizing a vertical metagrating coupler are executed.

[0046] Beneficial effects of the present invention:

[0047] Compared with the prior art, the present invention provides a method, device and medium for optimizing vertical meta-grating couplers, which effectively alleviates the problem of local optimal convergence in traditional adjoint optimization methods by introducing a "random perturbation" strategy and a "continuous topology optimization" method. In the process of continuous topology optimization, when the change of the objective function is within the preset threshold range, the introduction of random perturbations can dynamically adjust the dielectric constant distribution in the design area, avoiding premature convergence to the local optimal solution. Compared with traditional optimization techniques, such as the adjoint method and the DBS algorithm, this method can find the global optimal solution more flexibly and efficiently. In addition, the present invention adjusts the geometric shape of the grating through smoothing, and uses a binarization algorithm to convert the optimized dielectric constant distribution into two discrete values, so that the design result has higher processing feasibility and practical application value, solving the contradiction between high integration and processability in traditional methods. Finally, the optimized grating coupler has higher coupling efficiency and bandwidth performance, meeting the needs of modern photonic integrated circuits for ultra-compact and efficient input and output interfaces. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments.

[0049] Figure 1 It is a flow chart of a method for optimizing a vertical metagrating coupler disclosed in an embodiment of the present invention.

[0050] Figure 2 It is a design flow chart for VMGC disclosed in an embodiment of the present invention.

[0051] Figure 3 It is an optimized structural diagram of a small-sized single-layer VMGC disclosed in an embodiment of the present invention.

[0052] Figure 4 This is the optimized structure diagram using the Lumopt tool.

[0053] Figure 5 It is a schematic diagram of a single-polarization VMGC structure disclosed in an embodiment of the present invention.

[0054] Figure 6 This is a performance diagram of a single-polarization VMGC device disclosed in an embodiment of the present invention.

[0055] Figure 7 It is a schematic diagram of the structure of a polarization beam splitting VMGC disclosed in an embodiment of the present invention.

[0056] Figure 8 It is a performance diagram of a polarization beam splitting VMGC device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0058] According to an embodiment of the present invention, it should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the following method, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0059] like Figure 1 , Figure 2 As shown, the present invention provides a method for optimizing a vertical metagrating coupler, comprising the following steps:

[0060] Step S100, initializing the dielectric constant distribution in the two-dimensional design area, so that the dielectric constant is randomly distributed between silicon and silicon dioxide or between silicon and air, and stretching it to the etching layer height;

[0061] The design region refers to the optimization area of ​​the grating coupler, that is, the specific spatial range in the photonic device where performance needs to be optimized. This region is two-dimensional and represents the cross-section of the entire grating structure, including the distribution of different materials, the optical path, and the photon coupling region. The size and shape of the design region determine the function and performance of the grating coupler. By initializing, optimizing, and adjusting the dielectric constant within this design region, efficient light coupling and minimized light transmission loss can be achieved.

[0062] Initializing the dielectric constant distribution in the design area means assigning an initial dielectric constant value to each pixel unit in the design area at the beginning of the optimization process. The dielectric constant is the response characteristic of a material in an electric field and directly affects the propagation characteristics of light. At this stage, the value of the dielectric constant is randomly distributed, usually between silicon (Si) and silicon dioxide (SiO2), or between silicon and air. This randomly distributed dielectric constant provides a starting point for the subsequent optimization process. The optimization algorithm will adjust the dielectric constant of each pixel unit on this basis to optimize the light coupling efficiency and other performance indicators. Through this initialization method, the design area can provide more design freedom for subsequent topology optimization, thereby helping to achieve a more efficient vertical grating coupler structure.

[0063] Step S200, performing continuous topology optimization based on the initialized dielectric constant distribution;

[0064] Continuous topology optimization allows the material or dielectric constant in the design area to change continuously in space. Specifically for this step, the continuous topology optimization process includes multiple steps: First, forward simulation is performed to calculate the electromagnetic field distribution in the design area by simulating the incidence and propagation of the light source, and the objective function value is calculated based on this to evaluate the optical coupling effect of the current design. Next, based on the adjoint simulation, the adjoint electric field of the design area is obtained, and the sensitivity of the objective function relative to the dielectric constant distribution in the design area is calculated, and then the optimal gradient descent direction (i.e., the negative gradient direction) is determined, and information on how to adjust the dielectric constant distribution in the design area is obtained. This adjustment information will drive the optimization process, gradually adjusting the distribution of the dielectric constant to improve the coupling efficiency and bring the design closer to the final goal.

[0065] In short, continuous topology optimization adjusts the distribution of materials in the design area through repeated iterations, and adjusts the dielectric constant distribution in the design area with the help of objective function and sensitivity analysis to achieve the best optical coupling effect.

[0066] Step S300: During the continuous topology optimization process, the dielectric constant distribution is smoothed according to a preset filtering radius;

[0067] It can be understood that after the forward simulation and the accompanying simulation are completed, the dielectric constant distribution is smoothed according to the preset filter radius to filter out the sharp areas; the purpose of this stage is to avoid discontinuous mutations or high-frequency noise (such as checkerboard effects) in the structure, thereby ensuring that the optimized structure is physically feasible.

[0068] Step S400: when the change of the objective function is within a predefined threshold range, introducing random disturbance to adjust the dielectric constant distribution in the design area;

[0069] During the optimization process, the change rate of the objective function may slow down, indicating that the design has approached a relatively stable state, but this does not necessarily mean that the global optimal solution has been reached. At this time, by introducing random perturbations, the optimization process can be further promoted, the dielectric constant distribution in the design area can be adjusted, and the local optimal limit can be broken through to find a more ideal solution.

[0070] Specifically, first, the optimization process calculates the rate of change of the objective function (also known as the quality factor, FOM) between consecutive iterations. The rate of change of the objective function reflects the progress of the design optimization. If the rate of change of the objective function is within the preset threshold range, it means that the current optimization process has stagnated. In order to break this stagnation, the system will decide whether to introduce random perturbations based on the trend of the objective function changes. In this process, the size and direction of the random perturbations are set based on the amount of change in the previous iteration results.

[0071] When the objective function changes less, it means that the optimization direction tends to be stable and may stagnate near a local optimal solution. To avoid this situation, the dielectric constant of each pixel unit in the design area will be randomly adjusted slightly. These minor adjustments are usually made by randomly changing the value of the dielectric constant to ensure that they are still within the physically feasible range, that is, the dielectric constant remains between silicon (Si) and air (or silicon dioxide). Such perturbations can break the bottleneck of optimization and allow the optimization algorithm to move forward, thereby further improving the design performance on a global scale.

[0072] In addition, the introduction of random perturbations does not mean the randomness of the optimization process, but is carried out through precise control and calculation. The amplitude of the perturbations and the timing of their introduction are precisely set to ensure that they can effectively promote the optimization process without causing disordered changes in the design. For example, the amplitude of the random perturbation can be set to one percent of the dielectric constant, which ensures the effectiveness of the perturbation without destroying the optimized structure.

[0073] Preferably, the random perturbation process is divided into the following steps:

[0074] During the optimization process, the design area is divided into many small cells, each of which has a dielectric constant value that represents the optical properties of the material in that area. To introduce perturbations, some cells whose dielectric constants will be adjusted are first randomly selected.

[0075] Once the cells that need to be adjusted are selected, their dielectric constant values ​​are randomly increased or decreased by a small amount. The increase or decrease is random, with the aim of breaking the limitations of the existing design and allowing the optimization process to explore a wider range of design spaces. The increased or decreased dielectric constant values ​​will be restricted according to a preset range to ensure that the adjusted values ​​still meet physical requirements. Specifically, these adjusted dielectric constant values ​​must be within the range between silicon (Si) and silicon dioxide (SiO2), or between silicon (Si) and air.

[0076] In order to ensure that the optimization process fully explores the design space, the perturbation is not a one-time operation, but is repeated multiple times. Each perturbation is performed in a different cell to adjust the dielectric constant multiple times, thereby achieving a more comprehensive and uniform perturbation in the design area. Through multiple perturbations, the design can further break through the local optimal solution and find a possible better solution.

[0077] When the perturbation is completed, the dielectric constant distribution in the design area will change, which means that the distribution obtained in the original optimization process is adjusted to a new state. This new dielectric constant distribution will be used as input for subsequent optimization processes, and forward simulation, adjoint simulation and other operations will be continued to further optimize the design of the optocoupler.

[0078] Step S500, converting the dielectric constant after continuous topology optimization into two discrete values, wherein the conversion includes applying a binarization algorithm to limit the dielectric constant distribution to two discrete values ​​of silicon and silicon dioxide or silicon and air;

[0079] In the continuous topology optimization process, the dielectric constant in the design region is a continuous value, indicating the transition region of different materials. However, in actual manufacturing, photonic devices can usually only use a few different materials (such as silicon, silicon dioxide, air, etc.), and the dielectric constants of these materials are discrete. Therefore, the purpose of binarization is to convert these continuous dielectric constant values ​​into discrete material states, such as a binary distribution between silicon and air or silicon and silicon dioxide.

[0080] Before binarization, two target discrete values ​​need to be assigned to the dielectric constant. These two discrete values ​​correspond to the dielectric constants of silicon and silicon dioxide (or silicon and air), respectively. By setting these two target values, each pixel unit in the design area will be forced to be classified as one of the materials. Specifically, the numerical value of the dielectric constant will be compared, and if it is close to the dielectric constant of silicon (such as a higher dielectric constant), the pixel unit is assigned to silicon; if it is close to the dielectric constant of air or silicon dioxide (such as a lower dielectric constant), the pixel unit is assigned to air or silicon dioxide.

[0081] During the binarization process, a simple threshold is set to determine whether the dielectric constant of each pixel unit is close to silicon or air / silicon dioxide. For example, when the dielectric constant value of a pixel unit is greater than or equal to the average of the dielectric constants of silicon and air, the dielectric constant of the pixel unit will be set to the value of silicon. Otherwise, it will be set to the value of air or silicon dioxide. This processing method ensures that the dielectric constant in the design area ultimately has only two discrete values, thus meeting the requirements of actual materials.

[0082] After binarization, the dielectric constant of each pixel unit in the design area will be assigned a certain value, so that the entire design area becomes a discrete distribution composed of two materials (such as silicon and air / silicon dioxide). The binary dielectric constant distribution will directly correspond to the actual grating structure, and its physical properties are easier to connect with the manufacturing process.

[0083] Step S600, designing the geometric structure of the vertical metagrating coupler according to the discrete dielectric constant distribution obtained by optimization;

[0084] In the aforementioned optimization step, a discrete dielectric constant distribution consisting of two materials (such as silicon and air or silicon dioxide) has been obtained. These distributions are obtained after optimization and represent the dielectric constant value of each pixel unit. In step S600, it is first necessary to extract the design pattern from this discrete dielectric constant distribution. These patterns include information such as the specific shape of the grating, its optical properties and its distribution in space, and constitute the basic geometric structure of the grating coupler.

[0085] Based on the extracted dielectric constant distribution pattern, the geometric structure of the entire grating coupler will be defined next. The geometric structure not only includes the basic shape of the grating (such as the width, spacing, thickness, etc. of the grating stripes), but also involves the design of the upper and lower layers of the grating, the angle of incidence of the light source, and the layout of the waveguide. This process is to map the optimized theoretical design to the actual physical structure to ensure that the design meets the requirements of the manufacturing process while achieving optical coupling.

[0086] Step S700: Perform performance simulation on the designed vertical metagrating coupler to evaluate its coupling efficiency and bandwidth parameters.

[0087] The performance simulation is performed using electromagnetic simulation tools that simulate the physical phenomena of light propagation, reflection, refraction, etc. in the grating coupler. Through this simulation, the coupling efficiency, wavelength dependence, bandwidth and other parameters of the grating coupler can be evaluated. During the simulation, some specific operating conditions are input, such as the wavelength of the incident light source, the polarization state, the incident angle, etc., in order to fully evaluate the performance of the design under different operating conditions.

[0088] During the simulation, the coupling efficiency can be obtained by calculating the mode overlap integral of the output waveguide port and the incident light source. The higher the coupling efficiency, the more incident light can be effectively transmitted to the output waveguide.

[0089] Through simulation, we can get the performance of the grating coupler at different wavelengths, and then evaluate its bandwidth. The bandwidth determines the signal frequency range that the grating coupler can support. The larger the bandwidth, the better the performance of the coupler at multiple wavelengths.

[0090] In addition to coupling efficiency and bandwidth, performance simulation can also evaluate other important properties of grating couplers, such as polarization-dependent loss, optical transmission loss, beam quality, etc. For example, in multimode waveguide or polarization beam splitting applications, grating couplers need to ensure low polarization-dependent loss and be able to efficiently transmit optical signals. If these performance indicators do not meet the requirements, it may be necessary to further optimize the design, improve material selection or adjust the geometric structure.

[0091] Once the performance simulation is completed, the design team needs to compare the simulation results with the initial design goals. If the performance indicators such as coupling efficiency and bandwidth meet the predetermined requirements, the design is successful and can enter the manufacturing stage. If some indicators do not meet the target, it may be necessary to return to the design stage for further adjustment and optimization.

[0092] It can be understood that the method of this embodiment includes three optimization processes, namely initialization, continuous topology optimization, and binarization.

[0093] The initialization process includes the initialization settings of the device structure size, optimization parameters and optimization objective function (FOM). First, based on the physical basis, the basic structure of the vertical meta-grating coupler (VMGC) based on the 220nmSOI platform is constructed, including a 2μm thick SiO2 buried oxide layer, a 220nm thick partially etched Si lower grating, a fully etched Si upper grating, a SiO2 or air isolation layer between the upper and lower gratings, a 0.5μm wide 220nm thick Si block as the grating coupler output waveguide, and a Gaussian light source located directly above the VMGC. The optimization parameters include the dielectric constant / effective refractive index distribution of the 220nm Si layer optimization area, the partial etching thickness of the lower grating Si layer, the fully etched thickness of the upper grating Si layer, the minimum pixel unit size and the Gaussian light source mode field radius. The optimization area is set to the grating two-dimensional cross section. The optimization objective function depends on the device function. For example, the FOM of the single-polarization VMGC is set to the mode overlap integral of the TE0 fundamental mode at the output waveguide end and the incident Gaussian light source.

[0094] The continuous topology optimization process includes four steps: a forward simulation, a companion simulation, filtering, and random perturbation evaluation and implementation. The forward simulation is to obtain an output beam at the output port when the incident Gaussian light source is placed directly above the grating coupler. The companion simulation is to receive a coupled beam directly above the grating coupler when the TE0 waveguide light source is placed at the output waveguide port. By comparing the electric field distribution of the optimized area in the two simulations, the sensitivity information of each pixel unit in the optimized area can be obtained, thereby obtaining the dielectric constant distribution in the gradient descent direction. The filtering step is to smooth the high-frequency noise (such as spikes, checkerboards) in the design area according to the preset filtering radius to prevent the appearance of non-physical tiny structures, control the minimum feature size, and avoid local oscillations in the gradient to improve the stability of the optimization. The random perturbation evaluation and implementation step is based on the rate of change ΔFOM of the previous FOM values. j To determine, if ΔFOM j If the value is within the preset threshold range (θ, β), a random perturbation Δeps is applied to the entire optimization region. i , that is, the dielectric constant eps of each pixel unit i+1 =eps i +Δeps i When the number of optimization iterations is equal to the preset value n or the FOM approaches the expected value α or the rate of change of the FOM value ΔFOM j When the expected setting range [0,γ] is met, the continuous optimization process ends.

[0095] The binarization process is to binarize the values ​​between the dielectric constants of Si and air in the optimization area after the optimization process is completed. This method adopts the mean binarization method, that is, the dielectric constant eps of each pixel unit is i ≥(eps Si +eps air )*0.5, let eps i =eps Si , on the contrary, let eps i =eps air , where the subscript i represents the i-th pixel unit. So the dielectric constant is eps air The pixel unit is air, and the dielectric constant is eps Si The pixel unit is silicon, and an optimized grating structure can be obtained.

[0096] Furthermore, based on the existing Si processing technology and test conditions, the partial etching thickness of the lower grating Si layer is set to 110nm, the full etching thickness of the upper grating Si layer is set to 110nm, the minimum pixel unit size is set to 40nm, and the isolation layer thickness is 0, to ensure the processability and testability of the device.

[0097] The initial dielectric constant distribution of the grating optimization area adopts a random value between the dielectric constants of air and Si, namely, eps i ∈[1,12.0873].

[0098] The optimization objective function FOM of the single polarization VMGC is shown in formula (1), and the set FOM expected value α=1, that is, the coupling efficiency is 100%. The set ΔFOM j Expected value γ = 1 × 10 -8 Therefore, when FOM=α or 0≤ΔFOM j When ≤γ, the continuous optimization process ends. Where j represents the jth iteration:

[0099]

[0100] Where FOM represents the optimization objective function, λ(i) is the wavelength, and E λ(i) is the electric field strength related to wavelength, is the conjugate complex number of the electric field strength, and dS is the integration of small area elements in the design area.

[0101] In the random disturbance evaluation and implementation phase, the evaluation threshold range (θ, β) and the random disturbance Δeps i It is based on the optimization results and multiple tests. β is the ΔFOM obtained from previous optimizations. j When β is too large, introducing random perturbations too early will interfere with the optimization process; when β is too small, introducing random perturbations too late may result in the optimization ending with a local optimal solution. i The value will also affect the optimization process, Δeps i When the value range is too large, the optimized structure will fail; Δeps i If the value is too small, the perturbation effect cannot be exerted, and the effect is similar to no perturbation. In this method, one percent of the initial dielectric constant distribution is used as the global random perturbation.

[0102] Secondly, the optimization efficiency and optimization effect in this embodiment are greatly improved. Table 1 takes a small-size single-layer VMGC of 3.6μm*3.6μm as an example to compare the optimization efficiency and optimization effect of the method designed by the present invention and the similar reverse design tool lumopt using the adjoint method in Lumerical. By comparison, it can be seen that the optimization efficiency of the present invention is about 3 times that of lumopt, and the optimization effect is greatly improved in terms of both device performance and processability.

[0103] Table 1 Comparison of optimization efficiency and effect between the method of the present invention and the lumopt tool

[0104] Method of the present invention Lumopt Optimize structure See Figure 3 See Figure 4 Iterations 189 488 Time-consuming 15.75 hours 46.6 hours Coupling efficiency 53.32％ 48.18％ 3dB bandwidth 71nm 35nm

[0105] This embodiment adopts the above design method to design a single polarization VMGC, the structure of which is as follows: Figure 5 The single-polarization VMGC can couple a Gaussian beam in free space into a 0.5 μm wide single-mode waveguide, with a coupling efficiency of more than 95% and a 3dB bandwidth of 92 nm. The device performance curve is shown in Figure 6 shown.

[0106] This embodiment adopts the above design method and also designs a polarization beam splitting VMGC, the structure of which is as follows: Figure 7 The single-polarization VMGC can couple Gaussian beams with different polarization angles in free space into two 0.5 μm wide single-mode waveguides respectively, with the total coupling efficiency maintained at about 80% and the polarization-dependent loss lower than 0.16 dB. The device performance curve is shown in Figure 8 shown.

[0107] The embodiment of the present application provides an ultra-compact and high-efficiency vertical meta-grating coupler and design method for a meta-photonic integrated circuit, which effectively solves the problems of low coupling efficiency and large footprint of the vertical grating coupler, has high optimization efficiency and good optimization effect. The present invention provides an excellent input and output interface solution for the realization of large-scale ultra-compact meta-photonic integrated circuits, and is expected to have broad application prospects in many fields such as photon integrated laser radar, optical communication, and optical imaging.

[0108] According to another aspect of this embodiment, an electronic device is further provided, including a processor and a memory, wherein the processor is configured to implement the steps of the method when executing a computer program stored in the memory.

[0109] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0110] In the several embodiments provided in this embodiment, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0111] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0112] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0113] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for optimizing a vertical metagrating coupler, characterized in that: The steps include: Initializing the dielectric constant distribution in the two-dimensional design area so that the dielectric constant is randomly distributed between silicon and silicon dioxide or between silicon and air, and stretching it to the height of the etched layer; Based on the initialized dielectric constant distribution, performing continuous topology optimization; In the continuous topology optimization process, the dielectric constant distribution is smoothed according to the preset filter radius; When the change of the objective function is within a predefined threshold range, a random perturbation is introduced to adjust the dielectric constant distribution within the design area; Converting the dielectric constant after continuous topology optimization into two discrete values, the conversion comprising applying a binarization algorithm to limit the dielectric constant distribution to two discrete values ​​of silicon and silicon dioxide or silicon and air; According to the optimized discrete dielectric constant distribution, the geometric structure of the vertical metagrating coupler is designed; The performance of the designed vertical metagrating coupler is simulated to evaluate its coupling efficiency and bandwidth parameters.

2. The method for optimizing a vertical metagrating coupler according to claim 1, wherein: The continuous topology optimization comprises: Perform forward simulation to calculate the electromagnetic field distribution within the design area; Based on the forward simulation results, an objective function is established to evaluate the mode overlap integral between the output waveguide fundamental mode and the incident Gaussian light source; performing adjoint simulations to determine the sensitivity of the objective function with respect to the dielectric constant distribution within the design region; Using the sensitivity information, adjusting the dielectric constant distribution within the design region to optimize the objective function; The above steps are repeated until the objective function converges or the predetermined number of iterations is reached.

3. The method for optimizing a vertical metagrating coupler according to claim 2, wherein: After the forward simulation and the adjoint simulation are completed, the high-frequency noise part in the design area is smoothed according to the preset filtering radius. The smoothing method includes: According to the filter radius, the dielectric constant value of the high-frequency noise part is modified so that the dielectric constant within the filter radius is the same value; After filtering is complete, update the dielectric constant distribution in the design area and proceed to the next steps.

4. The method for optimizing a vertical metagrating coupler according to claim 2, wherein: When the change of the objective function is within a predefined threshold range, the method of introducing random perturbations to adjust the dielectric constant distribution within the design area includes: Calculate the change in quality factor between consecutive iterations; comparing the quality factor change to a predefined threshold range; When the quality factor change is within the threshold range, randomly perturbing the dielectric constant distribution within the design area based on a random perturbation method; adjusting the dielectric constant distribution within the design area according to the random perturbation; After introducing the random perturbations, the forward and adjoint simulations are continued.

5. The method for optimizing a vertical metagrating coupler according to claim 4, characterized in that: The random disturbance includes: Randomly select a number of dielectric constant cells; The dielectric constant value of the selected cell is randomly increased or decreased, and the dielectric constant value after random increase or decrease is limited to the dielectric constant range between silicon and silicon dioxide or between silicon and air; Repeat the above steps multiple times to achieve random perturbation of the overall dielectric constant distribution; After the random perturbation is completed, the dielectric constant distribution within the design region is updated for use in subsequent optimization steps.

6. The method for optimizing a vertical metagrating coupler according to claim 1, wherein: Methods for converting the dielectric constant after continuous topology optimization into two discrete values ​​include: setting two target discrete values ​​of dielectric constant, corresponding to the dielectric constants of silicon and silicon dioxide or silicon and air, respectively; For each dielectric constant cell in the optimization region, its dielectric constant is assigned to one of the target discrete values ​​based on how close its current value is to the two target discrete values; A binarization algorithm is applied to transform the continuous dielectric constant distribution in the design area into a distribution containing only the above two target discrete values.

7. The method for optimizing a vertical metagrating coupler according to claim 1, wherein: According to the optimized discrete dielectric constant distribution, the method for designing the geometric structure of the vertical metagrating coupler includes: extracting an optimized dielectric constant distribution pattern; Based on the extracted dielectric constant distribution pattern, the geometric structure of the vertical metagrating coupler is defined.

8. The method for optimizing a vertical metagrating coupler according to claim 2, wherein: The criterion for the convergence of the objective function is: When the objective function value is equal to a preset threshold, the objective function is considered to have converged, where the threshold is 1, indicating that the optimization process is stopped when the objective function value is equal to this value; When the rate of change of the objective function is within a predetermined threshold range, the objective function is considered to have converged, where the predetermined threshold range is [0, 10 -8 ], indicating that when the rate of change of the objective function is within this interval, the optimization process is stopped; The conditions for forcing optimization to stop are: When the number of optimization iterations reaches the preset value n, the optimization is forced to stop regardless of whether it converges or not.

9. An electronic device, characterized in that: It comprises a processor and a memory, wherein the processor is used to implement the steps of the method for optimizing a vertical metagrating coupler as claimed in any one of claims 1 to 8 when executing a computer program stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for optimizing a vertical metagrating coupler according to any one of claims 1 to 8 are executed.

Images