Deep ultraviolet light emitting diode and optimization method thereof

Abstract

The invention discloses a deep ultraviolet light-emitting diode and an optimization method thereof, which are applied to the technical field of deep ultraviolet light-emitting equipment. The method comprises the following steps that a structure optimization device model of the deep ultraviolet light emitting diode is established, a substrate nano array is arranged on the light emitting surface of a transparent substrate of the device, and a p-GaN nano array is arranged on the surface of a p-GaN layer and / or a p-AlGaN layer of the device; parameters of the p-GaN nano array and the substrate nano array are set as to-be-optimized parameters, corresponding to-be-optimized parameter constraint ranges are set according to the to-be-optimized parameters, and a multi-dimensional parameter space is constructed according to the to-be-optimized parameters and the corresponding to-be-optimized parameter constraint ranges; and with maximization of the total light extraction efficiency of the deep ultraviolet light emitting diode model as an optimization target, performing global collaborative optimization in the multi-dimensional parameter space by adopting a global optimization algorithm to obtain the optimal parameter of each to-be-optimized parameter. And the light extraction efficiency of the deep ultraviolet light-emitting diode is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of deep ultraviolet light-emitting device technology, and in particular to a deep ultraviolet light-emitting diode and its optimization method. Background Technology

[0002] Deep ultraviolet light-emitting diodes (DUV-LEDs) are semiconductor solid-state light sources with emission wavelengths shorter than 300 nm. They have broad application prospects in fields such as sterilization and disinfection, biochemical detection, non-line-of-sight secure communication, and water / air purification, and are an ideal alternative to traditional mercury lamp light sources. In recent years, with the advancement of aluminum gallium nitride (AlGaN) epitaxial growth and device technology, the performance of DUV-LEDs has been improved. However, their external quantum efficiency (EQE) is still significantly lower than that of visible light LEDs, which has become the main bottleneck restricting their large-scale commercial application.

[0003] The key factors leading to the low external quantum efficiency of DUV-LEDs are the high defect density and extremely low light extraction efficiency (LEE) within the material. Although the internal quantum efficiency can be optimized by improving crystal quality and quantum well structure, improving the light extraction efficiency remains a significant challenge. Traditional flip-chip DUV-LEDs typically consist of a transparent substrate, an n-AlGaN layer, a multi-quantum-well active region, a p-AlGaN layer, a p-GaN layer, and a metal mirror. This structure uses a flip-chip packaging method, with light emitted through a sapphire substrate. However, due to the large refractive index difference at the semiconductor-air interface and the dominance of transverse magnetic polarization mode in the deep ultraviolet band, the light extraction efficiency of traditional planar structures is usually less than 5% (statistically 2.4%), with the light extraction efficiency of transverse magnetic (TM) polarization mode often less than 1%. The fundamental reason for this phenomenon is that, on the one hand, the huge refractive index difference at the semiconductor-air interface leads to severe total internal reflection; on the other hand, the band structure of high-aluminum AlGaN materials determines that its outgoing light is mainly TM polarized, and the Fresnel transmittance of TM light at the interface is much lower than that of transverse electromagnetic (TE) polarized light, and it is easily confined to propagate in the planar waveguide and absorbed by the material.

[0004] To address the issue of low light extraction efficiency, researchers have developed various technical solutions, such as fabricating nanostructures like photonic crystals on the device surface, using patterned sapphire substrates, or embedding scattering structures like air holes inside the device. These techniques have indeed improved device performance to some extent by enhancing photon scattering and suppressing total internal reflection. However, for conventional planar device structures, the light extraction efficiency is usually still below 10%, indicating that existing light extraction enhancement technologies still have significant room for improvement in industrial applications. More importantly, structural optimization in existing technologies is mostly independent optimization, that is, designing the surface structure or the embedded layer structure separately, ignoring the optical coupling and interference effects between different structures. This approach often only achieves a simple superposition of effects rather than synergistic gain. In addition, the design process often relies on cumbersome parameter scanning methods, which are not only computationally expensive but also difficult to find the true global optimum in the multidimensional parameter space. Summary of the Invention

[0005] This invention provides a deep ultraviolet light-emitting diode and its optimization method, aiming to solve the problems in the background art and improve the light extraction efficiency of deep ultraviolet light-emitting diodes.

[0006] The first aspect of this invention provides an optimization method for deep ultraviolet light-emitting diodes, comprising the following steps: S1. Construct a deep ultraviolet light-emitting diode model, wherein the light-emitting surface of the transparent substrate of the deep ultraviolet light-emitting diode model is provided with a substrate nanoarray, and the surface of the p-GaN layer and / or p-AlGaN layer of the deep ultraviolet light-emitting diode model is provided with a p-GaN nanoarray. S2. Set the parameters of the p-GaN nanoarray and the substrate nanoarray as parameters to be optimized, and set the constraint range of the corresponding parameters to be optimized according to each parameter to be optimized. Construct a multi-dimensional parameter space according to the multiple parameters to be optimized and the constraint range of the corresponding parameters to be optimized. S3. Taking the maximization of the total light extraction efficiency of the deep ultraviolet light-emitting diode model as the optimization objective, a global optimization algorithm is used to perform global collaborative optimization in the multi-dimensional parameter space to obtain the optimal parameters of each parameter to be optimized.

[0007] In some embodiments of the first aspect, in step S2, the parameters to be optimized include the period of the substrate nanoarray, the duty cycle of the substrate nanoarray, the height of the substrate nanoarray, the period of the p-GaN nanoarray, the duty cycle of the p-GaN nanoarray, and the height of the p-GaN nanoarray.

[0008] In some embodiments of the first aspect, the period of the substrate nanoarray is constrained to be in the range of 10 to 1000 nm; The duty cycle of the substrate nanoarray is constrained to a range of 0.1 to 1; The height of the substrate nanoarray is constrained to the range of 10 to 1000 nm; The period of the p-GaN nanoarray is constrained to a range of 10 to 1000 nm; The duty cycle of the p-GaN nanoarray is constrained to a range of 0.1 to 1; The height of the p-GaN nanoarray is constrained to the range of 10 to 100 nm.

[0009] In some embodiments of the first aspect, step S3 specifically includes the following steps: S30. Initialize a population of a preset number of particles, wherein each particle represents a candidate scheme containing all parameters to be optimized in the multidimensional parameter space, and the initial position and velocity of each particle are randomly generated within the constraints. S31. For each of the particles, solve the light field value using a preset theoretical model; S32. Input the light field data into the global optimization objective function to obtain the current optimal fitness function value of the particle; S33. Based on the fitness values ​​mentioned above, determine the optimal individual position of each particle that has been searched so far and the global optimal position of the entire population. S34. Based on the individual best position searched for by each particle so far and the global best position of the entire population, update the velocity and position of the particles using the standard dynamics formula of the particle swarm optimization algorithm. S35. Repeat the above iterative process to determine whether the convergence criterion is met. If yes, output the result; otherwise, repeat the process from S31 to S35.

[0010] In some embodiments of the first aspect, the expression for the global optimization objective function is:

[0011] In the formula, To optimize the objective function globally, To achieve effective light extraction efficiency, As a penalty weighting coefficient, An aspect ratio penalty function based on geometric structure. For height, This refers to the bottom feature size.

[0012] In some embodiments of the first aspect, in the global optimization objective function, The following expression exists:

[0013] In the formula, To achieve effective light extraction efficiency, As a wavelength-dependent weighting factor, The extraction efficiency of the pure TE model obtained from numerical simulation. The extraction efficiency of the pure TM model obtained from numerical simulation.

[0014] In some embodiments of the first aspect, in the global optimization objective function, The following expression exists:

[0015] In the formula, An aspect ratio penalty function based on geometric structure. For height, For the bottom feature size, The limit threshold allowed by existing micro-nano etching processes.

[0016] In some embodiments of the first aspect, the nanostructures of the substrate nanoarray and the p-GaN nanoarray are one or a combination of nanoparabolic cones, nanocones, nanocylinders, and nanopyramids.

[0017] In some embodiments of the first aspect, the substrate nanoarray and the p-GaN nanoarray are arranged in an aperiodic or periodic manner; wherein the periodic arrangement is one of a tetragonal lattice, a hexagonal lattice, or a quasi-lattice.

[0018] A second aspect of the present invention provides a deep ultraviolet light-emitting diode manufactured according to the optimal parameters obtained by the optimization method described in the first aspect.

[0019] As can be seen from the above technical solutions, the present invention has at least the following advantages: (1) Significant synergistic gain: This invention breaks away from the traditional layer-by-layer independent optimization design approach and treats the multilayer nanostructure as a whole system for global optimization. This method makes full use of the optical coupling effect between the surface structure and the embedded structure, overcomes the defect that linear superposition design cannot match the interlayer light field, and achieves a synergistic gain effect of "1+1>2".

[0020] (2) High efficiency in optimization: The intelligent optimization algorithm is used to perform parallel search in the high-dimensional parameter space. Compared with the traditional single-parameter scanning method or control variable method, the design efficiency is greatly improved. It can quickly converge to the global optimal solution with low computational cost, and is particularly suitable for solving the design problem of complex optical systems with strong coupling of multiple parameters.

[0021] (3) Excellent device performance: The results show that at a wavelength of 280 nm, the light extraction efficiency of the device designed by this method can reach 12.98%, which is about 425% higher than that of the traditional flip-chip structure (2.47%). In particular, for the transverse magnetic (TM) polarization mode that is difficult to extract in deep ultraviolet LEDs, its extraction efficiency is increased from 0.29% to 10.70%, achieving an enhancement of about 36 times, effectively breaking through the technical bottleneck of TM light extraction. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the structure of a conventional deep ultraviolet light-emitting diode provided by the present invention and the deep ultraviolet light-emitting diode of this embodiment; Figure 2 A schematic diagram of the steps of the optimization method provided by the present invention; Figure 3 A schematic diagram of the particle dynamic evolution process provided by the present invention; Figure 4 This is a schematic diagram of the LEE comparison results provided by the present invention.

[0024] Figure label: 1. Metal mirror; 2. p-GaN layer; 3. p-AlGaN layer; 4. Multiple quantum well active region; 5. n-AlGaN layer; 6. Transparent substrate; 7. Substrate nanoarray; 8. p-GaN nanoarray. Detailed Implementation

[0025] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0026] Please refer to Figure 1As shown in (a), a traditional flip-chip DUV-LED typically includes a sapphire substrate, an n-AlGaN layer, a multi-quantum-well active region, a p-AlGaN layer, a p-GaN layer, and a metal mirror. This structure is packaged using a flip-chip method, with light emitted through the sapphire substrate. However, due to factors such as the large refractive index difference at the semiconductor-air interface and the dominance of transverse magnetic polarization modes in the deep ultraviolet band, the light extraction efficiency of traditional planar structures is typically 2.4%, severely limiting device performance. This paper addresses the limitations of existing technologies, such as the limited optimization effect of single-layer nanostructures, the inability to utilize interlayer optical coupling effects through independent layer-by-layer optimization of multi-layer structures, and the low efficiency of traditional parameter scanning methods.

[0027] This invention provides a deep ultraviolet light-emitting diode and its optimization method, aiming to solve the problems of limited optimization effect of single-layer nanostructures, inability to utilize interlayer optical coupling effect when optimizing multi-layer structures layer by layer independently, and low efficiency of traditional parameter scanning methods in the prior art. This invention achieves a significant improvement in the overall light extraction efficiency of deep ultraviolet light-emitting diodes, especially for improving the light extraction efficiency of transverse magnetic (TM) polarization mode.

[0028] See Figure 1 and Figure 2 The first aspect of this invention provides an optimization method for deep ultraviolet light-emitting diodes, comprising the following steps: S1. Construct a deep ultraviolet light-emitting diode model, such as Figure 1 As shown in (b), the deep ultraviolet light-emitting diode model includes a metal mirror 1, a p-GaN layer 2, a p-AlGaN layer 3, a multi-quantum well active region 4, an n-AlGaN layer 5, and a transparent substrate 6 stacked from bottom to top; the light-emitting surface of the transparent substrate 6 is provided with a substrate nanoarray 7, and the surface of the p-GaN layer and / or the p-AlGaN layer 3 is provided with a p-GaN nanoarray 8; S2. Set the parameters of p-GaN nanoarray 8 and substrate nanoarray 7 as parameters to be optimized, and set the constraint range of each parameter to be optimized according to each parameter to be optimized. Construct a multi-dimensional parameter space based on multiple parameters to be optimized and the constraint range of each parameter to be optimized. S3. Taking the maximization of the total light extraction efficiency of the deep ultraviolet light-emitting diode model as the optimization objective, a global optimization algorithm is used to perform global collaborative optimization in the multi-dimensional parameter space to obtain the optimal parameters of each parameter to be optimized.

[0029] It is understandable that the effects and purposes of each step S1-S3 are as follows: In step S1, a substrate nanoarray 7 is set on the light-emitting surface to achieve a continuous transition of refractive index from semiconductor to air by utilizing the contour of the substrate nanoarray 7, so as to suppress total internal reflection; a p-GaN nanoarray 8 is set on the surface of the p-GaN layer and / or the surface of the p-AlGaN layer 3 to initially modulate photons in the vicinity of the active region through scattering effect and change the propagation path of photons. In step S2, all the geometric parameters of the above nanostructure are established as a whole set to be optimized. The reason for selecting these two surfaces for joint modeling is that the p-GaN layer and p-AlGaN layer 3 are adjacent to the active region and control the initial emission angle of photons, while the light-emitting surface of the transparent substrate 6 determines whether photons can escape into the air. Combining the two for global optimization can break the limitations of traditional single-sided optimization or step-by-step optimization, and make full use of the interlayer synergistic effect between inner layer scattering and outer layer anti-reflection, thus laying the foundation for maximizing the light extraction efficiency (LEE) in the future. Furthermore, setting a specific search range aims to provide clear physical boundary constraints for the intelligent algorithm, ensuring that the optimization results have practical engineering application value. In step S3, a global optimization algorithm is used to optimize the multiple structural geometric parameters of the bilayer nanostructure as a whole, making full use of the interlayer synergistic effect to maximize the LEE.

[0030] Based on the above description of the embodiments, the advantages of this embodiment are as follows: (1) Significant synergistic gain: This invention breaks away from the traditional layer-by-layer independent optimization design approach and treats the multilayer nanostructure as a whole system for global optimization. This method makes full use of the optical coupling effect between the surface structure and the embedded structure, overcomes the defect that linear superposition design cannot match the interlayer light field, and achieves a synergistic gain effect of "1+1>2".

[0031] (2) High efficiency in optimization: The intelligent optimization algorithm is used to perform parallel search in the high-dimensional parameter space. Compared with the traditional single-parameter scanning method or control variable method, the design efficiency is greatly improved. It can quickly converge to the global optimal solution with low computational cost, and is particularly suitable for solving the design problem of complex optical systems with strong coupling of multiple parameters.

[0032] (3) Excellent device performance: The results show that at a wavelength of 280 nm, the light extraction efficiency of the device designed by this method can reach 12.98%, which is about 425% higher than that of the traditional flip-chip structure (2.47%). In particular, for the transverse magnetic (TM) polarization mode that is difficult to extract in deep ultraviolet LEDs, its extraction efficiency is increased from 0.29% to 10.70%, achieving an enhancement of about 36 times, effectively breaking through the technical bottleneck of TM light extraction.

[0033] In one specific embodiment, a specific manner in which the substrate nanoarray 7 and p-GaN nanoarray 8 in step S1 can be realized is further provided. In step S1, the nanostructures of the substrate nanoarray 7 and p-GaN nanoarray 8 are one or a combination of nanoparabolic cones, nanocones, nanocylinders, and nanopyramids. The p-GaN surface nanostructures may also be nanopore arrays or other scattering structures. The substrate nanoarray 7 and p-GaN nanoarray 8 are arranged in an aperiodic or periodic manner. The periodic arrangement is one of a tetragonal lattice, a hexagonal lattice, or a quasi-lattice.

[0034] It is understandable that the light-emitting surface of the transparent substrate 6 is provided with a surface nanocone array structure, whose conical profile realizes a continuous transition of refractive index from semiconductor to air, effectively suppressing total internal reflection at the interface; at the same time, a nanopillar array structure is prepared on the p-GaN layer, which initially modulates the photons in the vicinity of the active region through the scattering effect, thereby changing the propagation path of the photons.

[0035] In one specific embodiment, a feasible structure for the transparent substrate 6 in step S1 is further provided, wherein the transparent substrate 6 is one of a sapphire substrate and an aluminum nitride substrate.

[0036] In one specific embodiment, a further feasible method for step S2 is provided. In step S2, the parameters to be optimized include the period p1 of the substrate nanoarray 7, the duty cycle d1 of the substrate nanoarray 7, the height h1 of the substrate nanoarray 7, the period p2 of the p-GaN nanoarray 8, the duty cycle d2 of the p-GaN nanoarray 8, and the height h2 of the p-GaN nanoarray 8. The duty cycle of the substrate nanoarray 7 is constrained within a range of 0.1 to 1; the height of the substrate nanoarray 7 is constrained within a range of 10 to 1000 nm; the period of the p-GaN nanoarray 8 is constrained within a range of 10 to 1000 nm; the duty cycle of the p-GaN nanoarray 8 is constrained within a range of 0.1 to 1; and the height of the p-GaN nanoarray 8 is constrained within a range of 10 to 100 nm. The constructed multidimensional parameter space is a six-dimensional parameter space constructed from the above parameters and constraint ranges.

[0037] Understandably, this step sets a specific search range to provide clear physical boundary constraints for the intelligent algorithm, ensuring that the optimization results have practical engineering application value. Setting the lower limit of the parameter to a non-zero value (such as 10 nm or 0.1) instead of 0 is to avoid extreme values ​​where the structure size is zero, resulting in the absence of structure or loss of physical meaning. This ensures that all candidate solutions have actual physical structure and mechanical stability, and meet the limit resolution of existing micro-nano fabrication processes. The upper limit covers the effective working area of ​​subwavelength optics. In particular, strictly controlling the upper limit of the p-GaN nanopillar height to 100 nm is to adapt to the characteristics of p-GaN thin layers and prevent etching penetration that could damage the underlying active region, thereby improving the convergence efficiency of the algorithm while ensuring process feasibility.

[0038] In one specific embodiment, a concrete implementation method for step S3 is further provided, which specifically includes the following steps: S30. Initialize a population of a preset number of particles, where each particle represents a candidate scheme containing all parameters to be optimized in the multidimensional parameter space, and the initial position and velocity of each particle are randomly generated within the constraints. S31. For each particle, solve the light field value using a preset theoretical model; S32. Input the light field data into the global optimization objective function to obtain the optimal fitness function value of the current particle; S33. Based on the fitness values ​​mentioned above, determine the best individual position that each particle has searched so far and the global best position of the entire population. S34. Based on the individual best position searched by each particle so far and the global best position of the entire population, update the velocity and position of the particles using the standard dynamics formula of the particle swarm optimization algorithm. S35. Repeat the above iterative process to determine whether the convergence criterion is met. If yes, output the optimal parameters. If not, repeat the process of S31-S35.

[0039] Understandably, the processes described in steps S30-S35 efficiently search for the optimal solution in the multidimensional parameter space using the particle swarm optimization algorithm. This algorithm significantly improves parameter optimization efficiency due to its swarm intelligence characteristics. Specifically, the randomly distributed particle swarm generated during the initialization phase covers the entire constraint space, effectively avoiding the risk of getting trapped in local optima. Through the collaborative calculation of the theoretical model and the optimization function, a quantitative evaluation of the matching degree between the light field characteristics and the parameters is achieved. The iterative update mechanism gradually approaches the globally optimal parameter combination by dynamically adjusting the particle motion trajectory. This optimization strategy based on swarm intelligence not only improves computational accuracy but also significantly shortens the development cycle of traditional trial-and-error methods.

[0040] In one embodiment, based on the embodiment of step S30 above, a further implementation method of S30 is provided, wherein an initial population containing a certain number of particles is initialized, and each particle represents a candidate design scheme containing all the parameters to be optimized of the double-layer nanostructure [p1, d1, h1, p2, d2, h2] in the six-dimensional parameter space, and its initial position and velocity are randomly generated within the preset constraint range of step 2.

[0041] In one embodiment, based on the above-mentioned step S31, a further implementation method for S31 is provided. The preset theoretical model is a strict electromagnetic field theoretical model (such as the solution kernel of the finite difference time domain method FDTD or the strict coupled wave analysis method RCWA). In specific implementation, each particle will enter the iterative evaluation stage, and for each particle's current combination of structural parameters, the optical field is numerically solved based on the strict electromagnetic field theoretical model.

[0042] In one embodiment, based on the above-described step S32, a further implementation method for S32 is provided, wherein the specific expression of the global optimization objective function is:

[0043] In the formula, To optimize the objective function globally, To achieve effective light extraction efficiency, As a penalty weighting coefficient, An aspect ratio penalty function based on geometric structure. For height, This refers to the bottom feature size; In the global optimization objective function The following expression exists:

[0044] In the formula, To achieve effective light extraction efficiency, As a wavelength-dependent weighting factor, The extraction efficiency of the pure TE model obtained from numerical simulation. The extraction efficiency of the pure TM model obtained from numerical simulation; In the global optimization objective function The following expression exists:

[0045] In the formula, An aspect ratio penalty function based on geometric structure. For height, For the bottom feature size, The limit threshold allowed by existing micro-nano etching processes.

[0046] It is understood that the above embodiments construct a global optimization objective function that incorporates both physical light effects and process constraints. The algorithm aims to maximize the fitness function value of the current combination of structural parameters X; effective light extraction efficiency. The competition between transverse electric (TE) and transverse magnetic (TM) polarization modes in deep ultraviolet LEDs was considered; wavelength-dependent weighting factors were also considered. The introduction of this concept can accurately describe the variation of polarization degree with wavelength in high-aluminum AlGaN materials. The physical definition of ,in and These represent the luminous intensities of TE-polarized light and TM-polarized light, respectively. In this embodiment, they can be expressed using the correlation polarization formula. ,(in (where is the emission wavelength, in nm), derived from... With wavelength The linear functional relationship is as follows By substituting the wavelength, the value of the long-dependent weighting factor is obtained, thereby realizing the adaptive and precise assignment of the TE / TM light extraction weight; The penalty weighting coefficient is used to adjust the guiding intensity of process constraints. For deep ultraviolet LEDs with a wavelength of 280 nm, a balance is made between optimizing optical performance and the feasibility of micro / nano fabrication processes. Set to 0.5; This is a high aspect ratio penalty function based on geometric structure, used to eliminate extreme structures that are difficult to fabricate. It is defined as the penalty function applied when the aspect ratio of the nanostructure is high. (in For height, The bottom feature size is less than or equal to the limit allowed by existing micro / nano etching processes. (For example, when set to 5), the function value is 0, while when Greater than The function value is This forces the algorithm to automatically avoid non-physical regions that are prone to structural collapse or incomplete etching, thereby searching for a globally optimal solution that combines high performance and manufacturability in the multidimensional parameter space.

[0047] In one embodiment, based on the embodiment of step S33 described above, a further implementation of S33 is provided, which determines the optimal individual position searched for each particle so far based on the fitness value obtained in S32. And the global optimal position of the entire population. In practice, step S33 does not change the actual position and velocity of any particle. The rest simply refreshes the current record of each particle and the entire population based on the results of this round of detection, providing a basis for decision-making in the subsequent execution of S34.

[0048] In one embodiment, based on the embodiment of step S34 described above, a further implementation of S34 is provided, which is based on the individual optimal position of S33. And the global optimal position of the entire population. The particle velocities are updated using the standard dynamics formula of the particle swarm optimization algorithm. and location .

[0049] The specific expression for the standard dynamics formula of the particle swarm optimization algorithm is as follows:

[0050]

[0051] Among them, superscript and Representing the current iteration algebra and the next iteration algebra respectively; subscript Representing the first in the particle swarm One particle; and They represent the first The particle in the first generation and first The flight velocity vector of the time period; and They represent the first The particle in the first generation and first The spatial position vector of the time, that is, the corresponding combination of geometric parameters of the nanostructure; This represents the individual's historical best position, i.e., the position of the first individual. The spatial location with the highest fitness that the particle has passed through during the previous search process; This represents the global optimal position of the population, that is, the spatial position with the highest fitness that the entire particle swarm has searched so far. Inertial weights are used to balance global search and local exploit capabilities. and The learning factor (or acceleration constant) represents the step size weights for a particle to learn towards its own historical best position and the group's historical best position, respectively. and For in the interval Uniformly distributed random numbers are used to increase the randomness of the search, and the updated position vector is used. Boundary checks are required to ensure that the solution strictly meets the constraints of the aforementioned six-dimensional geometric parameter space, thus guaranteeing the feasibility of the solution in actual manufacturing processes.

[0052] In one embodiment, based on the embodiment of step S35 above, a further implementation method of S35 is provided, which repeats the process iteration of S31-S35 above until the convergence judgment criterion is met. The convergence criterion is set as follows: when the change of the global optimal fitness value within a certain number of consecutive generations is less than a preset threshold, or when the number of iterations reaches a preset maximum number of generations, the algorithm terminates and the result is output.

[0053] Understandably, please refer to Figure 3 This figure visually illustrates the dynamic evolution of particles in the parameter space. Driven by the aforementioned optimization mechanism, particles gradually gather from an initial random discrete distribution to a high-fitness region, eventually converging rapidly to the global optimum. The entire optimization process requires only about 165 numerical operations, which improves the design and solution efficiency by tens of thousands of times compared to the traditional parameter scanning method, greatly reducing the R&D and trial-and-error costs of high-performance devices.

[0054] The second aspect of the present invention provides a deep ultraviolet light-emitting diode, which is manufactured according to the optimal parameters obtained by the optimization method of the first aspect.

[0055] In this embodiment, the deep ultraviolet light-emitting diode includes a metal mirror 1, a p-GaN layer 2, a p-AlGaN layer 3, a multi-quantum well active region 4, an n-AlGaN layer 5, and a transparent substrate 6 stacked sequentially from bottom to top; a substrate nanoarray 7 is provided on the light-emitting surface of the transparent substrate 6, and a p-GaN nanoarray 8 is provided on the surface of the p-GaN layer and / or the p-AlGaN layer 3; and the parameters of the substrate nanoarray 7 and the p-GaN nanoarray 8 are determined according to the optimal parameters obtained by the optimization method of the first aspect.

[0056] It should be noted that the specific assembly and processing methods of deep ultraviolet light-emitting diodes are all existing technologies, and this article does not impose too many restrictions. Those skilled in the art can manufacture the deep ultraviolet light-emitting diode of this embodiment based on the content given in this application and in combination with existing technologies.

[0057] Experimental Example Based on the above description, this experimental example presents a deep ultraviolet light-emitting diode (DUV). The DUV includes a metal mirror 1, a p-GaN layer 2, a p-AlGaN layer 3, a multi-quantum-well active region 4, an n-AlGaN layer 5, and a sapphire transparent substrate 6 stacked sequentially from bottom to top. The light-emitting surface of the sapphire transparent substrate 6 is provided with a substrate nanoarray 7, which is configured as a parabolic cone array. The surface of the p-GaN layer is provided with a p-GaN nanoarray 8, which is configured as a nanopillar array. The parameters of the substrate nanoarray 7 and the p-GaN nanoarray 8 are determined according to the optimal parameters obtained by the optimization method in the first aspect.

[0058] The optimization method for the parameters of the substrate nanoarray 7 and the p-GaN nanoarray 8 for deep ultraviolet light with an emission wavelength of 280 nm specifically includes the following steps: S1. Construct a deep ultraviolet light-emitting diode model, wherein the deep ultraviolet light-emitting diode model includes a metal mirror 1, a p-GaN layer 2, a p-AlGaN layer 3, a multi-quantum well active region 4, an n-AlGaN layer 5, and a sapphire transparent substrate 6 stacked from bottom to top; the light-emitting surface of the sapphire transparent substrate 6 is provided with a substrate nanoarray 7, which is set as a parabolic cone array; the surface of the p-GaN layer is provided with a p-GaN nanoarray 8, which is set as a nanopillar array; S2. Set the period p1, duty cycle d1, height h1 of substrate nanoarray 7, period p2, duty cycle d2, and height h2 of p-GaN nanoarray 8 as variables to be optimized. The constraint range of the duty cycle of substrate nanoarray 7 is 0.1 to 1; the constraint range of the height of substrate nanoarray 7 is 10 to 1000 nm; the constraint range of the period of p-GaN nanoarray 8 is 10 to 1000 nm; the constraint range of the duty cycle of p-GaN nanoarray 8 is 0.1 to 1; and the constraint range of the height of p-GaN nanoarray 8 is 10 to 100 nm. Construct a six-dimensional parameter space according to the constraint range corresponding to each variable. S3. A particle swarm optimization algorithm is used to perform global collaborative optimization in a six-dimensional parameter space, with 15 particles and a maximum iteration of 40 generations. The overall light extraction efficiency is set as the optimization objective. Specifically, in the global optimization objective function, for this embodiment... Calculation of center wavelength in nm Penalty weight coefficient Set to 0.5. Set to 5; in the standard dynamics formula of the particle swarm optimization algorithm, Set to 0.9. and All parameters were set to 2.0. Finally, after 11 iterations, the algorithm converged, and the parameter combination that maximized the overall LEE was obtained as follows: the period of the sapphire light-emitting parabolic array p1 = 720 nm, the duty cycle d1 = 1.0, and the height h1 = 316 nm; the period of the p-GaN surface nanopillar array p2 = 504 nm, the duty cycle d2 = 0.1, and the height h2 = 69 nm.

[0059] Comparative Example The comparative example provides a conventional flip-chip deep ultraviolet light-emitting diode (DUV-LED), which includes a metal mirror 1, a p-GaN layer 2, a p-AlGaN layer 3, a multi-quantum well active region 4, an n-AlGaN layer 5, and a sapphire transparent substrate 6 stacked from bottom to top.

[0060] The experimental examples and comparative examples were verified experimentally, and the following results were obtained: For comparison, see Figure 1 (a): For planar devices without any nanostructures, the LEE in TE mode is 5.48%, the LEE in TM mode is 0.29%, and the weighted total LEE is 2.47%.

[0061] For experimental examples, see Figure 1 (b): The optimal parameter combination obtained by using this scheme is TE mode LEE of 16.12%, TM mode LEE of 10.70%, and weighted total LEE of 12.98%.

[0062] Please refer to Figure 4 The comparative results show that, compared with the traditional flip-chip structure, the method of this invention increases the LEE from 2.47% to 12.98%, an improvement of approximately 425%. In particular, the LEE of the TM mode is increased from 0.29% to 10.70%, achieving a significant enhancement of approximately 36 times, effectively solving the technical problem of difficult TM polarization mode light extraction in DUV-LEDs.

[0063] This embodiment fully demonstrates that the overall synergistic optimization design method proposed in this invention can discover and utilize the optical coupling effect between bilayer nanostructures, find the global optimal solution, and achieve a synergistic gain effect of "1+1>2", providing an efficient and feasible design method for improving the performance of DUV-LEDs.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

[0065] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented, for example, in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0066] Finally, it should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

Claims

1. An optimization method for deep ultraviolet light-emitting diodes, characterized in that, Includes the following steps: S1. Construct a deep ultraviolet light-emitting diode model, wherein the light-emitting surface of the transparent substrate of the deep ultraviolet light-emitting diode model is provided with a substrate nanoarray, and the surface of the p-GaN layer and / or p-AlGaN layer of the deep ultraviolet light-emitting diode model is provided with a p-GaN nanoarray. S2. Set the parameters of the p-GaN nanoarray and the substrate nanoarray as parameters to be optimized, and set the constraint range of the corresponding parameters to be optimized according to each parameter to be optimized. Construct a multi-dimensional parameter space according to the multiple parameters to be optimized and the constraint range of the corresponding parameters to be optimized. S3. Taking the maximization of the total light extraction efficiency of the deep ultraviolet light-emitting diode model as the optimization objective, a global optimization algorithm is used to perform global collaborative optimization in the multi-dimensional parameter space to obtain the optimal parameters of each parameter to be optimized.

2. The optimization method according to claim 1, characterized in that, In step S2, the parameters to be optimized include the period of the substrate nanoarray, the duty cycle of the substrate nanoarray, the height of the substrate nanoarray, the period of the p-GaN nanoarray, the duty cycle of the p-GaN nanoarray, and the height of the p-GaN nanoarray.

3. The optimization method according to claim 2, characterized in that: The period of the substrate nanoarray is constrained to the range of 10 to 1000 nm; The duty cycle of the substrate nanoarray is constrained to a range of 0.1 to 1; The height of the substrate nanoarray is constrained to the range of 10 to 1000 nm; The period of the p-GaN nanoarray is constrained to a range of 10 to 1000 nm; The duty cycle of the p-GaN nanoarray is constrained to a range of 0.1 to 1; The height of the p-GaN nanoarray is constrained to the range of 10 to 100 nm.

4. The optimization method according to claim 1, characterized in that, Step S3 specifically includes the following steps: S30. Initialize a population of a preset number of particles, wherein each particle represents a candidate scheme containing all parameters to be optimized in the multidimensional parameter space, and the initial position and velocity of each particle are randomly generated within the constraints. S31. For each of the particles, solve the light field value using a preset theoretical model; S32. Input the light field data into the global optimization objective function to obtain the current optimal fitness function value of the particle; S33. Based on the fitness values ​​mentioned above, determine the optimal individual position of each particle that has been searched so far and the global optimal position of the entire population. S34. Based on the individual best position searched for by each particle so far and the global best position of the entire population, update the velocity and position of the particles using the standard dynamics formula of the particle swarm optimization algorithm. S35. Repeat the above iterative process to determine whether the convergence criterion is met. If yes, output the result; otherwise, repeat the process from S31 to S35.

5. The optimization method according to claim 4, characterized in that, The expression for the global optimization objective function is: ； In the formula, To optimize the objective function globally, To achieve effective light extraction efficiency, For the penalty weighting coefficient, An aspect ratio penalty function based on geometric structure. For height, This refers to the bottom feature size.

6. The optimization method according to claim 4, characterized in that, In the global optimization objective function, The following expression exists: ； In the formula, To achieve effective light extraction efficiency, As a wavelength-dependent weighting factor, The extraction efficiency of the pure TE model obtained from numerical simulation. The extraction efficiency of the pure TM model obtained from numerical simulation.

7. The optimization method according to claim 4, characterized in that, In the global optimization objective function, The following expression exists: ； In the formula, An aspect ratio penalty function based on geometric structure. For height, For the bottom feature size, The limit threshold allowed by existing micro-nano etching processes.

8. The optimization method according to claim 1, characterized in that, The nanostructures of the substrate nanoarray and the p-GaN nanoarray are one or a combination of nanoparabolic cones, nanocones, nanocylinders, and nanopyramids.

9. The optimization method according to claim 1, characterized in that, The substrate nanoarray and the p-GaN nanoarray are arranged in an aperiodic or periodic manner; wherein, the periodic arrangement is one of a tetragonal lattice, a hexagonal lattice, or a quasi-lattice.

10. A deep ultraviolet light-emitting diode, characterized in that, It is manufactured using the optimal parameters obtained by the optimization method according to any one of claims 1 to 9.

Images