Micro-led trapezoidal mesa manufacturing method and system based on half-tone mask

Abstract

This invention discloses a method and system for fabricating MicroLED trapezoidal mesa based on halftone masks. The method first determines the photoresist and etching parameters and calibrates the process correction factor. Then, it establishes a full-link quantitative mapping model, including a first sub-model of mask transmittance to exposure dose distribution, a second sub-model of exposure dose to photoresist residual thickness profile and equivalent sidewall angle, and a third sub-model of multi-layer etching transfer angle mapping. Next, based on the target GaN trapezoidal mesa angle, the design parameters of the halftone mask transition region are calculated inversely, and the width and transmittance of each sub-region are determined using an equal thickness gradient allocation rule. Finally, halftone mask exposure and development are sequentially used to form the photoresist ramp profile, hard mask layer etching, and GaN layer etching to form the GaN trapezoidal mesa at the target angle. This invention achieves forward design from mask parameters to the final mesa angle, with a wide process window and high uniformity of the entire wafer angle.

Description

Technical Field

[0001] This invention relates to the field of MicroLED microdisplay device manufacturing technology, and in particular to a method and system for fabricating MicroLED trapezoidal mesa based on halftone masks. Background Technology

[0002] MicroLED (micro light-emitting diode) is a new generation of micro-display technology with major advantages such as high brightness and high contrast, low power consumption, long life, small size and thin design, wide color gamut coverage and scalability. It is widely used in TV, AR / VR, automotive display and other application scenarios.

[0003] In the MicroLED manufacturing process, to achieve convergent light emission and improve light extraction efficiency in GaN (gallium nitride)-based light-emitting devices, the GaN mesa needs to be fabricated into a trapezoidal structure with a specific tilt angle. Currently, the GaN trapezoidal process mainly adopts a hard mask process, which involves sequentially depositing ITO (indium tin oxide) and SiO2 on the GaN surface, first etching SiO2 to form the upper step (ITO serves as an etching stop layer), and then etching GaN to form the lower step. To support this multi-step etching process, the photoresist needs to be shaped into a three-dimensional form with trapezoidal steps after development. Precise control of the trapezoidal mesa angle and overall uniformity are the core challenges of this process.

[0004] Currently, two methods are known in the industry for achieving trapezoidal morphology in photoresist. The first is the defocus exposure method, which creates a sloping morphology at the edge of the photoresist by intentionally deviating the exposure focus from the optimal position. The second is halftone mask technology, which has mature applications in the TFT-LCD field and can generate multiple levels of photoresist thickness in a single exposure by utilizing the differences in transmittance in different areas of the mask.

[0005] However, these existing technologies have the following drawbacks when applied to the fabrication of GaN trapezoidal mesa in MicroLEDs: (1) The defocusing exposure method is affected by wafer warping, bending, and total thickness deviation. The actual defocusing amount in different areas of the wafer is inconsistent, resulting in poor uniformity of photoresist morphology after exposure and development. Some areas have a serious risk of defocusing, which may lead to complete distortion of the pattern. Moreover, the relationship between the defocusing amount and the step angle is nonlinear. Small defocusing deviations can lead to large fluctuations in the step angle, resulting in an extremely narrow process window.

[0006] (2) When halftone mask technology is directly applied to MicroLED, there is a huge gap in the precision requirements. Specifically, the GaN trapezoidal mesa of MicroLED requires precise control of a specific mesa angle of 60° to 75° within the extremely small pixel scale of several micrometers to tens of micrometers. The angle accuracy requirement is within ±2°, which exceeds the morphology control precision of TFT-LCD process.

[0007] (3) The angle of the GaN trapezoidal mesa is not determined solely by the photoresist morphology, but is affected by the cascading influence of the multi-layer pattern transfer chain of the photoresist morphology, SiO2 hard mask, ITO cut-off layer and GaN body. The existing technology lacks a systematic design method to accurately correlate the halftone mask design parameters with the final GaN mesa angle, which means that engineers can only rely on repeated trial and error to adjust the halftone parameters, resulting in low development efficiency and poor process repeatability.

[0008] (4) The GaN layer thickness in MicroLED is usually 3-5μm. The consumption of photoresist and the lateral etching of SiO2 mask during dry etching will change the angular fidelity of pattern transfer. If a quantitative mapping model from mask parameters to the final mesa angle is not established, the forward design of the target angle cannot be achieved.

[0009] In summary, existing technologies suffer from the inability to establish a precise quantitative mapping relationship between halftone mask design parameters and the final GaN trapezoidal mesa angle, resulting in the inability to achieve forward design of the target angle, a narrow process window, and poor uniformity across the entire wafer. Summary of the Invention

[0010] This invention provides a method and system for fabricating MicroLED trapezoidal mesa based on halftone masks. It can establish a quantitative mapping relationship across the entire chain from the design parameters of the halftone mask transition area to the final GaN trapezoidal mesa angle, enabling forward design of the target mesa angle. This solves the technical problems in the prior art, such as the inability to accurately correlate mask parameters with the final mesa angle, the reliance on trial and error iteration, the narrow process window, and the poor uniformity of the angle across the entire wafer.

[0011] In a first aspect, the present invention provides a method for fabricating a trapezoidal mesa of MicroLEDs based on a halftone mask, the method comprising the following steps: Step S1: Determine process constants and calibrate process correction factors. The process constants include photoresist parameters and etching parameters. The photoresist parameters include initial coating thickness h0 and threshold dose D. th Clearance dose D cl The etching parameters include the photoresist contrast ratio γ and the base exposure dose E0, and the etching parameters include the SiO2 etching selectivity ratio S1 to the photoresist, the GaN etching selectivity ratio S2 to SiO2, the SiO2 etching process correction factor K1, and the GaN etching process correction factor K2. Step S2: Based on the process constants, establish a full-link quantitative mapping model from the halftone mask transition region design parameters to the final GaN trapezoidal mesa angle. This full-link quantitative mapping model includes a cascaded first sub-model of mask transmittance to exposure dose distribution, a second sub-model of exposure dose to photoresist residual thickness profile and equivalent sidewall angle, and a third sub-model of angle mapping for multi-layer etching transfer. The comprehensive formula for the full-link quantitative mapping model is: W eff The effective horizontal distance of the transition zone is equal to the sum of the widths of all sub-regions within the transition zone; Step S3: Based on the target GaN trapezoidal mesa angle θ target Based on the aforementioned end-to-end quantitative mapping model, the design parameters of the halftone mask transition region are calculated in reverse, including the inverse calculation of the photoresist target sidewall angle θ. PR,target Calculate the effective transition zone width W eff,target And calculate the width w of each sub-region of the transition zone according to the grade number N and the transmittance allocation rule. i and transmittance T i ; Step S4: Fabricate a halftone mask according to the design parameters determined in step S3. The graphic area corresponding to each MicroLED pixel on the halftone mask is divided into a central fully opaque area, an edge transition area, and a peripheral fully transparent area. The transition area is divided into N sub-regions, and the transmittance of each sub-region is achieved through a sub-resolution grating pattern. Step S5: Exposure and development using a halftone mask to form a photoresist slope profile. Specifically, positive photoresist is coated onto a GaN epitaxial substrate with sequentially deposited ITO and SiO2 hard mask layers to an initial coating thickness h0. Using the halftone mask fabricated in step S4, the photoresist is exposed and developed at the optimal focal plane position with a base exposure dose E0, forming a profile with an equivalent sidewall angle θ. PR The photoresist slope profile; Step S6: Sequentially etch the hard mask layer and the GaN layer to form a GaN trapezoidal mesa. Use a photoresist with a slope profile as a mask to etch the SiO2 hard mask layer, and use the ITO layer as the etching stop layer to form a structure with a sidewall angle θ. SiO2 A SiO2 hard mask slope is then used; subsequently, the GaN layer is etched using the SiO2 hard mask as a mask to form a target angle θ. GaN GaN trapezoidal platform.

[0012] The core of the MicroLED trapezoidal mesa fabrication method based on halftone masks provided by this invention lies in the first-ever establishment of a full-link quantitative mapping model from the design parameters of the halftone mask transition region to the final GaN mesa angle. This model unifies the parameter relationships of the three process stages—mask transmittance distribution, photoresist morphology transformation, and multilayer etching transfer—into an explicit mathematical formula. This overcomes the limitation of existing technologies where halftone masks are only used as a qualitative morphology control method, making it a precision process method with quantitative forward design capabilities. Compared with defocus exposure, this method uses the spatial transmittance distribution of the mask rather than focal plane offset to control the photoresist morphology. Therefore, it has natural robustness to wafer warpage, bending, and thickness deviation, a significantly wider process window, and significantly higher uniformity of the entire wafer angle.

[0013] Preferably, in step S2, the first sub-model is: the effective exposure dose D received by the photoresist at the i-th sub-region of the transition region. i Equal to the transmittance T of this sub-region i The product of the base exposure dose E0, i.e. ; The second sub-model is a three-segment partitioned photoresist residual thickness model: when D i <D th At that time, the residual thickness h r (D i )=h0; when D th ≤D i ≤D cl hour, When D i >D cl At that time, h r (D i =0; Equivalent sidewall angle θ of photoresist PR satisfy ,in The summation range is from 1 to N; The third sub-model is a two-stage etching angle transfer model: the relationship between the SiO2 hard mask sidewall angle and the photoresist sidewall angle is as follows: The relationship between the GaN mesa angle and the SiO2 sidewall angle is as follows: ; The photoresist contrast γ and threshold dose D th Clearance dose D cl Satisfies strict mathematical consistency relation: .

[0014] Preferably, the process correction factors K1 and K2 in step S1 are calibrated in the following manner: A calibration mask is prepared, wherein a set of transition regions with a total physical width W is provided on the calibration mask. HTA series of gradient halftone test patterns were generated, with each group of test patterns using the same transmittance grading method. The complete photolithography, SiO2 etching, and GaN etching process was performed on a standard GaN / ITO / SiO2 laminated substrate using the aforementioned calibration mask. The actual GaN mesa angle θ corresponding to each group of test patterns was measured using cross-sectional SEM. GaN,meas ; The calibration methods include two types: integrated calibration mode and independent calibration mode. The comprehensive calibration model is as follows: a comprehensive correction factor K = K1 × K2 is fitted using the least squares method, with the fitting objective being to sum over all test groups j. Take the minimum value, where W eff (j) represents the sum of the widths of each sub-region of the transition area in the j-th test pattern, i.e., the total physical width W of the transition area in this group of test patterns. HT (j); The independent calibration mode is as follows: after the SiO2 etching step is completed and before the GaN etching step begins, cross-sectional SEM measurements are performed on each set of test patterns on the calibration mask to obtain the actual sidewall angle θ of the SiO2 hard mask. SiO2,meas K1 was independently fitted using the least squares method; then, the measured angle after GaN etching and the calibrated K1 were used as the basis. Independently fit K2; The value of K1 ranges from 0.85 to 1.15, and the value of K2 ranges from 0.80 to 1.10.

[0015] Preferably, the reverse calculation in step S3 includes the following sub-steps: S3.1 Calculate the target sidewall angle of the photoresist based on the inverse calculation form of the end-to-end synthesis formula: ; S3.2 Calculate the required effective transition area width directly from the photoresist sidewall angle definition: ; S3.3. Determine the transition level N (N is greater than or equal to 3), and use the equal thickness gradient allocation rule to allocate parameters to the sub-regions, ensuring that each sub-region bears an equal amount of photoresist thickness variation h0 / N, and that the width of each sub-region is equal. The transmittance of each sub-region is determined by the following formula: ; S3.4. Verify the self-consistency of the design parameters and confirm their accuracy. The summation range is from 1 to N.

[0016] Preferably, the equal thickness gradient distribution rule enables the photoresist to form a stepped profile with optimal linearity in the transition region, resulting in a uniform and equidistant decrease in residual photoresist thickness among the sub-regions; the transmittance range of each sub-region calculated by this rule falls within the effective transmittance range D corresponding to the photoresist working range. th / E0 to D cl Within / E0, the 0% transmittance of the central fully shaded area and the 100% transmittance of the outer fully transparent area belong to different levels.

[0017] Preferably, in step S4, the light transmittance of the central fully opaque area is 0%, corresponding to the area where the photoresist retains its full thickness, and is used to define the top of the GaN trapezoidal mesa; the light transmittance of the outer fully transparent area is 100%, corresponding to the area where the photoresist is completely removed, and is used to define the bottom of the GaN trapezoidal mesa; the edge transition area is divided into N sub-regions along the direction from the central fully opaque area to the outer fully transparent area, with a light transmittance distribution that increases from the inside to the outside; Within each sub-region, a period p is passed through a point smaller than the resolution limit of the exposure system. i Arranging light-blocking lines and light-transmitting gaps to achieve a sub-resolution grating pattern, with a grating period p i satisfy Where λ is the exposure wavelength and NA is the numerical aperture; the transmittance T of the i-th sub-region i The duty cycle of the light-blocking lines determines the relationship. , where d i The width of the light-blocking line.

[0018] Preferably, the GaN epitaxial layer has a thickness of 3 to 5 micrometers, and the target GaN trapezoidal mesa angle θ target The value ranges from 60 degrees to 75 degrees; When the base exposure dose E0 fluctuates within ±5% of the nominal value, the angle deviation of the fabricated GaN mesa is less than ±1.5 degrees; when the focal plane shifts by ±0.3 micrometers, the angle deviation of the fabricated GaN mesa is less than ±0.5 degrees.

[0019] Secondly, the present invention provides a MicroLED trapezoidal mesa system based on a halftone mask for performing the steps involved in establishing a full-link quantitative mapping model, calculating inverse parameters, and calibrating process correction factors in the method described in the first aspect. The system includes: The process parameter acquisition module is used to acquire and store photoresist parameters and etching parameters, including initial coating thickness h0, contrast γ, and threshold dose D. th Clearance dose D clThe etching parameters, including the SiO2 etching selectivity ratio to the photoresist (S1), the GaN etching selectivity ratio to SiO2 (S2), the SiO2 etching process correction factor (K1), and the GaN etching process correction factor (K2), are set to the base exposure dose E0. The module automatically performs consistency checks when storing parameters to ensure that γ matches D. th D cl satisfy Mathematical relationships; The end-to-end mapping calculation module is used to establish an end-to-end quantitative mapping model based on the photoresist parameters and the etching parameters. The end-to-end quantitative mapping model is established through a comprehensive formula. A quantitative relationship is established between the design parameters of the transition zone of the halftone mask and the angle of the GaN trapezoidal mesa. The reverse engineering module receives the target GaN trapezoidal mesa angle, calculates the target sidewall angle and effective transition region width of the photoresist based on the full-link quantitative mapping model, and calculates the width and transmittance of each sub-region of the transition region according to the number of levels and transmittance allocation rules including equal thickness gradient allocation rules, and then executes... The self-consistency verification is performed, and the summation range is from 1 to N; The correction factor calibration module is used to obtain and update the process correction factors K1 and K2 by fitting the measured GaN mesa angle data based on the calibration mask using the least squares method. The module supports a comprehensive calibration mode of fitting K=K1×K2 and an independent calibration mode of fitting K1 and K2 respectively by combining intermediate measurement data after SiO2 etching. After calibration, the module automatically updates the K1 and K2 values ​​stored in the process parameter acquisition module.

[0020] The beneficial effects of the present invention are as follows: Compared with the prior art, the technical means of the present invention include: establishing a two-level etching angle transfer model with process correction factors and a photoresist equivalent sidewall angle model based on the total physical width of the transition region; proposing an equal thickness gradient allocation rule to obtain the photoresist slope profile with optimal linearity; and a systematic process correction factor calibration method based on calibrating the mask and least squares fitting. The contribution of the present invention lies in the first establishment of a full-link quantitative mapping model from the halftone mask transition region design parameters to the final GaN trapezoidal mesa angle, realizing the process chain of mask transmittance-exposure dose distribution-photoresist residual thickness profile-SiO2 hard mask angle transfer-GaN mesa angle transfer, achieving forward design of the target angle rather than trial-and-error iteration. Attached Figure Description

[0021] Figure 1 This is a schematic flowchart of the method for fabricating a MicroLED trapezoidal mesa based on a halftone mask provided by the present invention.

[0022] Figure 2This is a schematic diagram of the halftone mask planar structure provided by the present invention, with the full light-blocking area, the transition area, each sub-region, and the full light-transmitting area marked.

[0023] Figure 3 This is a partially enlarged view of the sub-resolution grating structure in the transition region provided by the present invention, with the grating period p marked. i and the width d of the light-blocking line i .

[0024] Figure 4 This is a schematic diagram of the full-link quantitative mapping model framework provided by the present invention, showing the input-output relationship of the three cascaded sub-models.

[0025] Figure 5 The diagram shows the cross-sectional profile of the photoresist after exposure and development, and the multilayer transfer cross-section after SiO2 etching and GaN etching, provided by the present invention.

[0026] Figure 6 This is a schematic diagram showing the comparison of halftone transition zone parameters under different target angles provided by the present invention.

[0027] Figure 7 A schematic diagram showing the comparison curves of the process window sensitivity of the halftone scheme and the out-of-focus exposure scheme provided by the present invention.

[0028] Figure 8 The structural block diagram of the MicroLED trapezoidal mesa system based on halftone mask provided by the present invention. Detailed Implementation

[0029] The following description and accompanying drawings fully illustrate specific embodiments of this application to enable those skilled in the art to practice them. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. The scope of embodiments of this application includes the entire scope of the technical solutions described herein, as well as all available equivalents of said technical solutions. In this document, each embodiment may be referred to individually or collectively with the term "invention," which is merely for convenience and, if more than one invention is disclosed, is not intended to automatically limit the scope of the application to any single invention or inventive concept. Relational terms such as "first" and "second" are used herein only to distinguish one entity or operation from another, without requiring or implying any actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed. The various embodiments in this document are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the structures, products, etc., disclosed in the embodiments, since they correspond to the disclosed parts, the descriptions are relatively simple; relevant details can be found in the method section.

[0030] Terminology Explanation Before describing the technical solution in detail, the key terms used in this application will be explained first: MicroLED: Miniature light-emitting diodes, with pixel sizes typically in the range of 1-100μm, are core components of next-generation microdisplay technology, possessing advantages such as high brightness and high contrast, low power consumption, long lifespan, small and thin design, wide color gamut coverage, and scalability. In this application, MicroLED is the target device type, and the trapezoidal angle of its GaN mesa directly affects light extraction efficiency and display performance.

[0031] GaN: Gallium nitride, a III-V group wide bandgap semiconductor material, is the core light-emitting layer material of MicroLEDs. In this application, the GaN epitaxial layer is the target layer that needs to be etched into a trapezoidal mesa, with a typical thickness of 3-5 μm. The specific tilt angle of the GaN trapezoidal mesa (typically 60°-75°) helps to form convergent light rays during the light emission process, thereby improving light extraction efficiency.

[0032] Halftone: Halftone refers to the control of exposure of different areas of photoresist by varying the transmittance of different regions on a photomask. In this application, the transition area of ​​the halftone photomask forms a three-dimensional slope profile of the photoresist in a single exposure through a spatial transmittance gradient, which is the core photolithography tool of this invention.

[0033] ITO: Indium Tin Oxide, a transparent conductive thin film. In the process route of this application, ITO is deposited on the GaN surface, and SiO2 is deposited on the ITO surface. ITO acts as an etching stop layer during the SiO2 etching process to prevent SiO2 from over-etching and damaging the GaN layer.

[0034] ICP: Inductively Coupled Plasma, a commonly used dry etching technique. In this application, ICP etching is used for pattern transfer between the SiO2 hard mask and the GaN layer. Different etching gas systems (CF4 / CHF3 for SiO2, Cl2 / BCl3 for GaN) are used to achieve selective etching of different material layers.

[0035] Subresolution grating: A structure of light-blocking lines and light-transmitting gaps arranged on a photomask with a period less than the resolution limit of the exposure system. In this application, the exposure system uses an i-line lithography machine (exposure wavelength λ = 365 nm), with a typical numerical aperture NA = 0.50, corresponding to a resolution limit of... Grating period p i The wavelength must be less than 365nm (e.g., 180nm). At this wavelength, the exposure system cannot distinguish individual lines and gaps in the grating. Instead, it treats the area as a semi-transparent region with uniform transmittance, which is determined by the duty cycle of the light-blocking lines.

[0036] Photoresist contrast γ: A parameter describing the steepness of the thickness change of photoresist under varying exposure levels, defined as: ; Where D cl For clearance dose, D th γ is the threshold dose, and lg is the common logarithm base 10. The larger the γ value, the narrower the range of exposure dose required to completely remove the photoresist. It is important to note that the value of γ is strictly determined by D. th and D cl The decision requires that all three satisfy the mathematical consistency defined by the above formula.

[0037] Etching selectivity: The ratio of the etching rates of two materials under the same etching conditions. In this application, The etching selectivity ratio of SiO2 to photoresist. These two selection ratios, representing the etching selectivity of GaN to SiO2, determine the angle transformation relationship during the pattern transfer process.

[0038] Process correction factors (K1, K2): Calibration coefficients used to correct the deviation between the theoretical etching angle transfer model and the actual process results. K1 reflects the influence of ion bombardment angle distribution and sidewall passivation effect in the SiO2 etching step, while K2 reflects the influence of the ratio of chemical etching to physical sputtering in the GaN etching step. The value range of K1 is usually 0.85-1.15, and the value range of K2 is usually 0.80-1.10. Both need to be obtained through experimental calibration.

[0039] TTV: Total Thickness Variation, an indicator describing the uniformity of wafer thickness. TTV, along with wafer warp and bending, constitutes a major factor affecting the uniformity of defocusing exposure schemes.

[0040] Detailed technical solution This invention provides a method for fabricating a trapezoidal mesa surface for MicroLEDs based on a halftone mask. For example... Figure 1 As shown, Figure 1 This is a schematic flowchart illustrating the method for fabricating a MicroLED trapezoidal mesa based on a halftone mask provided by the present invention. The method provided by the present invention can be executed by the process system of a semiconductor manufacturing platform in conjunction with manufacturing equipment. The calculation steps involving the establishment of a full-link quantitative mapping model, inverse parameter calculation, and process correction factor calibration can be implemented in software by a device equipped with a processor and memory. The steps involving photolithography and etching are executed by the corresponding semiconductor manufacturing equipment.

[0041] The method includes the following steps: Step S1: Determine process constants and calibrate process correction factors. The purpose of this step is to obtain all the basic process parameter data required for the end-to-end quantitative mapping model, including photoresist parameters and etching parameters. This step corresponds to the prerequisite for reverse engineering the design parameters in the first aspect of the technical solution, providing input data for all subsequent calculation steps.

[0042] (1) Regarding the determination of photoresist parameters, measurements were taken using standard process wafers. Specifically, a positive photoresist (e.g., i-line photoresist AZ4620 or equivalent) was spin-coated onto a standard test substrate, and the initial coating thickness h0 was measured and recorded. Then, a series of exposure tests were performed on the photoresist with different exposure doses. After development, the residual thickness at each dose was measured using a profilometer or ellipsometry, and the photoresist characteristic curve was plotted to extract the threshold dose D. th (Below this value, the photoresist does not dissolve) and removal dose D cl(Photoresist exceeding this value is completely removed), and the photoresist contrast γ is calculated according to the following formula: ; Three parameters D th D cl The consistency relationship between γ and γ must satisfy the definition of the above formula. Simultaneously, determine the base exposure dose E0, which is the nominal exposure dose received in the fully transparent area; E0 should be greater than D. cl This ensures that the photoresist in the fully transparent area is completely removed.

[0043] (2) Regarding the determination of etching parameters, the etching rate R of SiO2 in the CF4 / CHF3 gas system ICP etching was first measured under standard process conditions. SiO2 Etching rate R of photoresist under the same conditions PR Calculate the etching selectivity of SiO2 for photoresist: ; Then, the etching rate R of GaN was measured under ICP etching conditions in a Cl2 / BCl3 gas system. GaN The etching rate R of SiO2 SiO2 'Calculate the etching selectivity of GaN for SiO2:' '; (3) Regarding the calibration of process correction factors K1 and K2, a calibration mask is prepared, on which a set of halftone test patterns with a series of gradient transition widths are set, for example W HT The transmittance grading methods for each test pattern are 3, 4, 5, 6, 7, and 8 μm, respectively, with the same method applied to each group. Using this calibration mask, a complete photolithography (coating, exposure, development), SiO2 etching, and GaN etching process was completed on a standard GaN / ITO / SiO2 laminated substrate. The actual GaN mesa angle θ corresponding to each test pattern was measured using cross-sectional SEM. GaN,meas The measured angle is compared with the predicted angle θ of the theoretical model under the assumptions K1=K2=1. GaN,calc By comparison, a comprehensive correction factor K = K1 × K2 is fitted using the least squares method. The fitting objective is to find K that minimizes the following objective function: ; To calibrate K1 and K2 separately, the specific procedure is as follows: After the SiO2 etching step is completed and before the GaN etching step begins, perform cross-sectional SEM measurements on each set of test patterns on the calibration mask to obtain the actual sidewall angle θ of the SiO2 hard mask. SiO2,meas The measured SiO2 angle was compared with the theoretical prediction. A comparison was performed, and K1 was independently fitted using the least squares method. Then, the measured angle θ after GaN etching was used. GaN,meas And the already calibrated K1, according to K2 is fitted independently. After calibration, K1 and K2 can be used repeatedly as intrinsic constants of the process platform. Recalibration is required when the etching equipment or process gas formulation changes.

[0044] Step S2: Establish a full-link quantitative mapping model The purpose of this step is to construct a quantitative mapping model of the entire process chain, from the halftone mask transition region design parameters to the final GaN mesa angle, based on the process constants obtained in step S1. This step corresponds to the establishment of the quantitative mapping model of the entire process chain in the technical solution and is the core component of this invention. Figure 4 As shown, Figure 4 This is a schematic diagram of the full-link quantitative mapping model framework provided by the present invention.

[0045] The end-to-end quantitative mapping model consists of three cascaded sub-models.

[0046] (1) The first sub-model is a mapping from mask transmittance to exposure dose distribution. When using positive photoresist, the effective exposure dose D received by the photoresist in the i-th sub-region of the transition region is... i The transmittance T of this sub-region i The product of the base exposure dose E0 is directly determined: ; This relationship establishes a direct correspondence between mask geometry design parameters and photoresist physical receptivity, due to the transmittance T of each sub-region. i The effective exposure dose for each sub-region is completely predictable and controllable, as it is determined by the duty cycle of the sub-resolution grating and E0 is a controllable parameter of the exposure equipment.

[0047] (2) The second sub-model is the mapping from exposure dose to the residual thickness profile of the photoresist and the equivalent sidewall angle. The residual thickness h of the positive photoresist after development... r The relationship between the exposure dose D and the photoresist contrast γ is determined by the photoresist contrast γ, and is modeled using a three-segment partitioning model: 1) When D i <D th Time: h r (D i =h0 (If the exposure dose is below the threshold, the photoresist is completely unaffected). 2) When D th ≤D i ≤D cl hour: (The photoresist is in the working range, and the residual thickness decreases linearly with the logarithm of the exposure dose.) 3) When Di >D cl Time: h r (D i =0 (exposure dose is higher than removal dose, photoresist is completely removed).

[0048] Based on this, the equivalent sidewall angle θ of the slope profile formed in the transition region after photoresist development is... PR satisfy: ; Wherein the effective horizontal distance W eff Defined as the sum of the widths of all sub-regions in the transition zone: ; The physical meaning of this definition is: W eff The equivalent sidewall angle θ represents the total horizontal distance traversed by the photoresist as it transitions from full thickness h0 to zero thickness. PR Let W be the angle between the line connecting the top (h0) and bottom (0) of the slope and the horizontal direction. For the case of equal thickness gradient distribution, the stepped profile formed by N equal-width sub-regions has the best approximation accuracy for the ideal linear slope, where W... eff =W HT (Total physical width of the transition zone), the equivalent sidewall angle is consistent with the average slope of the actual stepped profile.

[0049] (3) The third sub-model is the angle mapping of multi-layer etching transfer. The slope profile of the photoresist needs to undergo two stages of etching transfer to finally form the GaN trapezoidal mesa. In the SiO2 etching step, the photoresist slope is transferred to the SiO2 hard mask, and the sidewall angle θ of the SiO2 hard mask is... SiO2 Angle θ with photoresist sidewall PR The relationship is: ; K1 is a process correction factor for the SiO2 etching step, reflecting the influence of ion bombardment angle distribution and sidewall passivation effect during etching. Its value typically ranges from 0.85 to 1.15. K1 > 1 indicates that the actual sidewall angle is steeper than the ideal calculated value (sidewall passivation enhances anisotropy), while K1 < 1 indicates that the actual sidewall angle is gentler than the ideal value (lateral etching weakens anisotropy). In the GaN etching step, the slope edge of the SiO2 hard mask shifts towards GaN, and the GaN mesa angle θ... GaN Angle θ with SiO2 sidewall SiO2 The relationship is: ; K2 is a process correction factor for the GaN etching step, reflecting the influence of the ratio of chemical etching to physical sputtering on the GaN sidewalls in the etching gas system such as Cl2 / BCl3. Its value typically ranges from 0.80 to 1.10. In GaN ICP etching, physical sputtering, i.e., ion bombardment, provides directional anisotropic etching, while chemical etching (the chemical reaction between reactive gases and the material) tends to be isotropic. The ratio of these two processes determines the value of K2.

[0050] Combining the above three sub-models, we obtain the comprehensive formula for the end-to-end quantitative mapping model: ; Will After substituting the values, this formula can be used to calculate the final GaN mesa angle θ. GaN The parameters that can be designed for a halftone mask (width w of each sub-region) are expressed as follows: i and transmittance T i ) and explicit functions of measurable process parameters (h0, E0, S1, S2, K1, K2). h0 / W eff The tangent of the equivalent sidewall angle of the photoresist represents the steepness of the photoresist slope itself; S2 / S1 represents the amplification effect of the two-stage etching selection comparison angle. Since S2 is usually much larger than S1, the overall two-stage etching transfer has an angle amplification effect (i.e., the GaN mesa angle is much larger than the photoresist sidewall angle); K1×K2 is a comprehensive process correction factor that corrects the influence of non-ideal effects in the two-step etching on the angle transfer.

[0051] Step S3: Calculate the design parameters of the halftone mask transition region in reverse based on the angle of the target GaN trapezoidal mesa. The purpose of this step is to use the full-link quantitative mapping model established in step S2 to determine the target GaN trapezoidal mesa angle θ. target The design parameters of the halftone mask transition region are calculated in reverse. This step corresponds to the technical solution regarding the reverse design process and equal gradient allocation rules, realizing the forward design from the target angle to the mask parameters.

[0052] (1) Inverse calculation of the target sidewall angle of the photoresist. Based on the inverse calculation form of the end-to-end synthesis formula: ; This formula reverses the angular amplification effect of the two-stage etching process. Since S2 / S1 is large and K1×K2 is close to 1, θ PR,target Typically much smaller than θ target That is, the slope angle of the photoresist is much smaller than the final GaN mesa angle.

[0053] (2) Determine the effective transition region width. This is directly obtained from the definition of the photoresist sidewall angle: This width is the sum of the physical widths of all sub-regions in the transition zone.

[0054] (3) Sub-region parameter allocation. After determining the number of transition zone levels N (N≥3), the sub-region parameters are allocated using the equal thickness gradient allocation rule. The equal thickness gradient allocation rule ensures that each sub-region bears an equal amount of photoresist thickness variation h0 / N. Under the equal gradient condition, the transmittance of each sub-region is determined according to the following formula: ; This formula is derived from the working interval of the photoresist residual thickness model, requiring that the residual thickness corresponding to the i-th sub-region is exactly h0×(1-i / N). Substituting this into the residual thickness formula, the corresponding exposure dose D is solved. i Then through This formula is obtained by converting it to light transmittance.

[0055] It should be noted that the transmittance range calculated by this formula is T1 (slightly higher than D when i=1). th / E0) to T N (When i=N, T) N =D cl These values ​​( / E0) fall within the effective transmittance range corresponding to the working area of ​​the photoresist, and belong to different levels of concepts from the 0% transmittance of the fully shaded area and the 100% transmittance of the fully transparent area in the overall structure of the photomask.

[0056] The fully shaded area (0%) ensures that the exposure dose received by the photoresist is well below the threshold D. th This maintains the full thickness of the photoresist; the full light-transmitting area (100%) ensures that the photoresist receives an exposure dose higher than the removal dose D. cl This allows for the complete removal of the photoresist; the transmittance of each sub-region in the transition area is precisely controlled within D. th / E0 to D cl Within the / E0 range, the residual thickness of the photoresist decreases gradually according to the design gradient within the working range.

[0057] The width of each sub-region is equal under the condition of equal gradient: ; This is because, under a uniform thickness gradient distribution, each of the N sub-regions bears the same h0 / N thickness variation, and the N equal-width sub-regions are evenly distributed within the transition zone, with the sum of their widths exactly equal to the effective transition zone width W. eff,target .

[0058] Self-consistency can be verified: This rule enables the photoresist to form a stepped profile with optimal linearity in the transition region. That is, the residual thickness of the photoresist decreases uniformly and equally between each sub-region without local steep or gentle segments, which is beneficial for obtaining uniform and flat GaN mesa sidewalls after subsequent etching and transfer.

[0059] Step S4: Fabricate a halftone mask according to the design parameters. The purpose of this step is to fabricate a halftone mask based on the transition region design parameters determined in step S3. This step corresponds to the technical features regarding the mask structure and sub-resolution grating implementation method in the technical solution. For example... Figure 2 As shown, Figure 2 This is a schematic diagram of the planar structure of the halftone mask provided by the present invention.

[0060] The halftone mask divides the patterned area corresponding to each MicroLED pixel into three functional regions. The central, fully opaque region has 0% transmittance, corresponding to the area where the photoresist is fully retained, and is used to define the top of the GaN trapezoidal mesa. The edge transition region has a transmittance distribution that increases from the inside out, forming a photoresist slope; its transmittance gradient directly determines the final GaN mesa angle. The outer, fully transparent region has 100% transmittance, corresponding to the area where the photoresist is completely removed, and is used to define the bottom of the GaN trapezoidal mesa.

[0061] The transition zone is divided into N sub-regions (N≥3) along the direction from the central fully shaded area to the outer fully transparent area. The i-th sub-region has a width w calculated in step S3. i and transmittance T i The transmittance of each sub-region is achieved through sub-resolution grating patterns. For example... Figure 3 As shown, Figure 3 This is a partially enlarged view of the sub-resolution grating structure in the transition region provided by the present invention. Within each sub-region, light-blocking lines and light-transmitting gaps are arranged at a period less than the resolution limit of the exposure system. For an i-line exposure machine (exposure wavelength λ = 365 nm, numerical aperture NA = 0.50), the resolution limit is: ; grating period p i The transmittance (T) must be less than 365nm (e.g., 180nm). i Determined by the duty cycle of the light-blocking lines: ; Where d i This refers to the width of the light-blocking lines. When the grating period is less than the resolution limit, the exposure system cannot distinguish individual lines and gaps in the grating. Instead, it treats the area as a semi-transparent region with uniform transmittance. By adjusting the duty cycle of the light-blocking lines and the light-transmitting gaps, different transmittance values ​​can be achieved in different sub-regions on the same mask.

[0062] In addition, as an alternative, the transmittance of the transition region can also be achieved through a continuous gradient, that is, by continuously changing the duty cycle of the grayscale grating to achieve a linear gradient of transmittance without separating sub-regions. This approach can further improve the smoothness of the photoresist ramp, but the mask manufacturing complexity is higher. Another alternative is to use a semi-permeable film material (such as MoSi thin film) to achieve halftone, depositing MoSi thin films of different thicknesses in the transition region to adjust the transmittance of each sub-region. This approach has higher transmittance control precision, but the mask requires additional MoSi deposition and patterning steps. The mapping model and design method of the above two alternatives are the same as the main approach.

[0063] Step S5: Use a halftone mask for exposure and development to form the photoresist slope profile. The purpose of this step is to use the halftone mask prepared in step S4 to form a photoresist pattern with a specific slope profile on the substrate to be processed. This step corresponds to the technical feature in the technical solution regarding exposure and development to form the slope profile.

[0064] The substrate to be processed is a GaN epitaxial substrate, on which an ITO layer and a SiO2 hard mask layer have been deposited sequentially. Positive photoresist is then coated on the SiO2 surface to the target initial thickness h0.

[0065] The halftone mask prepared in step S4 is installed in the exposure machine, and the photoresist is exposed at the optimal focal plane position with a base exposure dose E0. During exposure, the photoresist below the central fully shaded area of ​​the mask does not receive any exposure dose and remains intact; the photoresist below the outer fully transparent area receives the full E0 dose (E0>D). cl The photoresist beneath each sub-region of the transition region is completely removed; the photoresist below each sub-region of the transition region receives... The modulation dose is within the photoresist's working range. After exposure, development is performed, and each sub-region forms a residual thickness corresponding to its respective exposure dose. For example... Figure 5 As shown in the cross-section of the photoresist, a stepped slope profile from the initial thickness h0 to zero thickness is finally formed at the corresponding position in the transition region, with an equivalent sidewall angle θ. PR This aligns with the design goals outlined in step S3.

[0066] Because the halftone scheme exposes at the optimal focal plane, the depth-of-focus margin is a standard value (typically ±0.5μm or more). Even if there is a height deviation of ±0.3μm in a local area of ​​the wafer, the change in photoresist morphology is negligible, and the GaN mesa angle deviation is <±0.5°. Simultaneously, when the base exposure dose E0 fluctuates by ±5% around the nominal value, since the transmittance ratio of each sub-region is fixed (determined by the duty cycle of the mask grating), the overall fluctuation of the exposure dose proportionally affects all sub-regions. Mathematically, E0×(1±0.05) ensures that all D...i A proportional change, in the logarithmic coordinates of the residual thickness formula, is equivalent to changing all lg(D)... i Overall translation: ; Since the logarithmic span of the photoresist working range is lg(D) cl )-lg(D th (Typical value approximately 0.56), a translation of 0.021 accounts for only about 3.8% of the working area, resulting in an overall translation of the residual thickness of each sub-region, but the thickness difference between adjacent sub-regions remains almost unchanged. Since the slope angle is determined by the thickness gradient (the ratio of the thickness difference between adjacent sub-regions to the width of the sub-region) rather than the absolute thickness value, the angle deviation is <±1.5°.

[0067] In contrast, in the defocus exposure scheme, the same ±5% dose fluctuation combined with wafer distortion will result in an angle deviation of more than ±5°, because the angle of the defocus scheme directly depends on the absolute exposure dose and the focal plane position, rather than the spatial transmittance ratio distribution.

[0068] Step S6: Perform hard mask layer etching and GaN layer etching sequentially to form GaN trapezoidal mesa. The purpose of this step is to use a photoresist with a sloping profile as a mask to ultimately form a GaN trapezoidal mesa with the target angle through a two-stage etching process. This step corresponds to the technical features of hard mask layer etching and GaN layer etching in the technical solution. Figure 5 As shown, Figure 5 The diagram shows the cross-sectional profile of the photoresist after exposure and development, and the multilayer transfer cross-section after SiO2 etching and GaN etching, provided by the present invention.

[0069] (1) ICP etching of the SiO2 hard mask layer. ICP etching was performed using a CF4 / CHF3 gas system, with a photoresist having a ramp profile as the mask to etch the SiO2 layer. During the etching process, the stepped ramp profile of the photoresist gradually shifts towards the SiO2 layer as the photoresist itself is consumed. That is, the thinner sub-regions of photoresist are exhausted first, thus exposing the underlying SiO2, while the thicker sub-regions are exposed later. This time difference results in different etching times for different positions of the SiO2 layer, thus forming a specific sidewall angle θ. SiO2 A continuous SiO2 hard mask slope is used, with an ITO layer acting as an etching stop layer to prevent over-etching. The relationship between the SiO2 hard mask sidewall angle and the photoresist sidewall angle is as follows: ; K1 corrects for the effects of ion bombardment angle distribution and sidewall passivation during the etching process.

[0070] (2) ICP etching of the GaN layer. ICP etching was performed using a Cl2 / BCl3 gas system, with the slope edge of the SiO2 hard mask serving as the mask for etching the GaN layer. During the etching process, the slope profile of the SiO2 hard mask gradually shifted towards the GaN layer as the SiO2 was slowly consumed, forming a shape with the target angle θ. GaN The GaN trapezoidal mesa. The relationship between the GaN mesa angle and the SiO2 sidewall angle is: ; Because the etching selectivity ratio of GaN to SiO2 in S2 is much greater than 1, the GaN mesa angle is significantly amplified, and the final GaN trapezoidal mesa angle is different from the target angle θ set in step S3. target Consistent.

[0071] After etching is completed, the actual angle of the GaN trapezoidal mesa can be measured and verified by cross-sectional SEM to confirm that the deviation from the target angle is within the design allowable range (usually ±2°). Example 1

[0072] The technical solution of the present invention will be described in detail below through a specific embodiment. The objective of this embodiment is to fabricate a GaN trapezoidal structure with a trapezoidal mesa angle of 65° on a MicroLED substrate.

[0073] This embodiment is based on the following substrate and film structure: GaN epitaxial layer thickness 4.0 μm, SiO2 hard mask thickness 500 nm, ITO thickness 150 nm.

[0074] (1) Corresponding to step S1: Select positive photoresist AZ4620, spin-coat it on a standard test substrate and measure to determine the initial coating thickness h0=2.5μm. Obtain the photoresist characteristic curve through a series of exposure dose tests, and extract the threshold dose D. th =50mJ / cm 2 Clearance dose D cl =180mJ / cm 2 Calculate the photoresist contrast: γ=1 / (lg(180)-lg(50))=1 / (2.2553-1.6990)=1 / 0.5563=1.80; Set the base exposure dose E0 = 200 mJ / cm 2 Greater than D cl To ensure complete removal of the photoresist in the fully transparent area. For SiO2 etching (CF4 / CHF3 system ICP), the measured SiO2 etching rate was 220 nm / min, and the photoresist etching rate was 120 nm / min. Calculations were made as follows: S1 = 220 / 120 = 1.83; For GaN etching (Cl2 / BCl3 system ICP), the measured etching rate of GaN is 350 nm / min, and the etching rate of SiO2 is 35 nm / min. Calculate: S2 = 350 / 35 = 10.0; Using a calibration mask (W) HT (Six groups with transmittance grades of 3, 4, 5, 6, 7, and 8 μm) were processed on a standard GaN / ITO / SiO2 laminated substrate. After completing the entire process, the actual GaN mesa angles for each group were measured using cross-sectional SEM. Following the SiO2 etching step, the SiO2 sidewall angles for each group were measured, and K1 = 0.95 was obtained through independent fitting using the least squares method. Then, K2 = 0.88 was obtained by independently fitting the measured angles after GaN etching and the calibrated K1. Overall angle magnification factor: (S2 / S1)×K1×K2=(10.0 / 1.83)×0.95×0.88=5.464×0.836=4.57; That is, the tangent of the photoresist sidewall corner is amplified by about 4.57 times after two stages of etching.

[0075] (2) Corresponding to step S2: Substitute all the above process constants into the three cascaded sub-models to establish a full-link quantitative mapping model. First sub-model: D i =T i ×200mJ / cm 2 The second sub-model adopts a three-segment partitioning model with parameter D. th =50mJ / cm 2 D cl =180mJ / cm 2 h0 = 2.5 μm, lg(D cl )-lg(D th =0.5563. Equivalent sidewall angle of photoresist: tan(θ PR )=2.5 / W eff ,in ; The parameters of the third sub-model are S1=1.83, S2=10.0, K1=0.95, and K2=0.88. The comprehensive formula is: tan(θ GaN )=(2.5 / W eff )×(10.0 / 1.83)×0.95×0.88=(2.5 / W eff )×4.57; (3) Corresponding step S3: Target GaN trapezoidal mesa angle θ target =65°. First, calculate the sidewall angle of the photoresist target: tan(θ PR,target )=tan(65°)×1.83 / (10.0×0.95×0.88)=2.1445×1.83 / 8.36=3.924 / 8.36=0.470; θ PR,target =arctan(0.470)=25.2°; Then calculate the effective transition zone width: W eff,target =2.5 / 0.470=5.32μm; A uniform gradient distribution with N=4 is adopted, with each sub-region bearing a thickness variation of h0 / N=2.5 / 4=0.625μm. The transmittance of each sub-region is calculated as follows: T1 = (50 / 200) × 10^{(1 / 4) × lg(180 / 50)} = 0.25 × 10^{0.25 × 0.5563} = 0.25 × 10^{0.1391} = 0.25 × 1.377 = 0.344 (i.e., 34.4%). T2 = (50 / 200) × 10^{(2 / 4) × 0.5563} = 0.25 × 10^{0.2781} = 0.25 × 1.896 = 0.474 (i.e., 47.4%). T3 = (50 / 200) × 10^{(3 / 4) × 0.5563} = 0.25 × 10^{0.4172} = 0.25 × 2.612 = 0.653 (i.e., 65.3%). T4 = (50 / 200) × 10^{(4 / 4) × 0.5563} = 0.25 × 10^{0.5563} = 0.25 × 3.600 = 0.900 (i.e. 90.0%).

[0076] The width of each sub-region is: w i =5.32 / 4=1.33μm; Self-consistency verification: W eff =4×1.33=5.32μm=W eff,target .

[0077] The transmittance of each sub-region was cross-validated using residual thickness: 1) For T1 = 34.4%: D1 = 0.344 × 200 = 68.8 mJ / cm 2 ; h r=2.5×[1-(lg(68.8)-lg(50)) / 0.5563]=2.5×[1-(1.8376-1.6990) / 0.5563] =2.5×[1-0.1386 / 0.5563]=2.5×[1-0.2492]=2.5×0.7508=1.877μm≈1.875μm.

[0078] 2) For T2 = 47.4%: D2 = 0.474 × 200 = 94.8 mJ / cm 2 ; h r =2.5×[1-(lg(94.8)-lg(50)) / 0.5563]=2.5×[1-0.2778 / 0.5563]=2.5×[1-0.4995]=2.5×0.5005=1.251μm≈1.25μm.

[0079] 3) For T3 = 65.3%: D3 = 0.653 × 200 = 130.6 mJ / cm 2 ; h r =2.5×[1-(lg(130.6)-lg(50)) / 0.5563]=2.5×[1-0.4169 / 0.5563]=2.5×[1-0.7496]=2.5×0.2504=0.626μm≈0.625μm.

[0080] 4) For T4 = 90.0%: D4 = 0.900 × 200 = 180 mJ / cm 2 =D cl ; h r =0μm.

[0081] The corresponding residual photoresist thickness sequence is: h0 = 2.50 μm at the inner boundary, h r (T1) = 1.875 μm, h r (T2) = 1.25 μm, h r (T3) = 0.625 μm, h r (T4)=0μm, and the thickness change at each level is exactly 0.625μm=h0 / 4, which verifies the correctness of the equal gradient distribution.

[0082] (4) Corresponding step S4 (fabrication of halftone mask): Fabricate the halftone mask according to the above design parameters. For each MicroLED pixel, the graphic area is configured with a central fully opaque area (0%), an edge transition area consisting of four sub-regions (each sub-region has a width of 1.33 μm and transmittances of 34.4%, 47.4%, 65.3%, and 90.0%, respectively), and a peripheral fully transparent area (100%). The transmittance of each sub-region is achieved through a sub-resolution grating pattern. The corresponding grating structure is fabricated on the mask substrate using a chromium film standard patterning process. The exposure system uses an i-line lithography machine (λ=365nm, NA=0.50), with a grating period of 180nm, which is less than the resolution limit of 365nm, satisfying the sub-resolution condition. The widths of the opaque lines corresponding to each sub-region are as follows: d1 = (1 - 0.344) × 180 = 118.1 nm; d2 = (1 - 0.474) × 180 = 94.7 nm; d3 = (1 - 0.653) × 180 = 62.5 nm; d4 = (1 - 0.900) × 180 = 18.0 nm.

[0083] (5) Corresponding to step S5: Coat the SiO2 surface of the GaN / ITO / SiO2 substrate with AZ4620 photoresist to a thickness of 2.5 μm. Install the fabricated halftone mask into the i-line exposure machine, and at the optimal focal plane position, apply E0 = 200 mJ / cm. 2 Exposure is performed. After exposure, development is performed. After development, the photoresist at the corresponding position of the transition area forms a contour that descends from 2.50 μm to 0 μm in a step, and the residual thicknesses of the four sub-regions are 1.875, 1.25, 0.625 and 0 μm, respectively.

[0084] Equivalent sidewall angle: θ PR =arctan(2.50 / 5.32)=arctan(0.470)=25.2°; This aligns with the design goals outlined in step S3.

[0085] (6) Corresponding step S6: 1) Using a photoresist slope as a mask, the SiO2 layer was etched using ICP with a CF4 / CHF3 system. The SiO2 etching rate was 220 nm / min, the photoresist etching rate was 120 nm / min, and the ITO layer was used as the etching stop layer. The resulting SiO2 hard mask sidewall corners are as follows: θ SiO2=arctan(tan(25.2°)×(1 / 1.83)×0.95)=arctan(0.470×0.519)=arctan(0.244)≈13.7° 2) Using a SiO2 hard mask as the substrate, the GaN layer was etched using an ICP etching system based on a Cl2 / BCl3 matrix. The GaN etching rate was 350 nm / min, and the SiO2 etching rate was 35 nm / min. The final GaN trapezoidal mesa angle is: θ GaN =arctan(tan(13.7°)×10.0×0.88)=arctan(0.244×8.80)=arctan(2.147)≈65.0° like Figure 6 As shown, the final result of this embodiment is that the measured angle of the GaN trapezoidal mesa is approximately 65.0°, which matches the target angle of 65°, verifying the correctness and effectiveness of the full-link quantitative mapping model and the four-step reverse design process. Example 2

[0086] The applicability of the present invention's technical solution under different target angles is demonstrated below through a second embodiment. The objective of this embodiment is to fabricate a GaN trapezoidal structure with a mesa angle of 70° on a MicroLED substrate. This embodiment uses the same substrate and film structure (4.0 μm GaN epitaxial layer, 500 nm SiO2 hard mask, 150 nm ITO) and the same process platform as Embodiment 1.

[0087] (1) Corresponding to step S1 (determining process constants and calibrating process correction factors): This embodiment uses the same process platform as Example 1, therefore all process constants follow the calibration results of Example 1: h0=2.5μm, γ=1.80, D th =50mJ / cm 2 D cl =180mJ / cm 2 E0 = 200 mJ / cm 2 S1=1.83, S2=10.0, K1=0.95, K2=0.88. This demonstrates the engineering practicality of the method of the present invention, which allows for the reuse of process constants after a single calibration. That is, when switching between different target angles on the same process platform, there is no need to recalibrate the process parameters; only the new target angle needs to be input in step S3.

[0088] (2) Corresponding step S2 (establishing the full-link quantitative mapping model): The full-link quantitative mapping model is exactly the same as in Example 1. First sub-model: D i =T i ×200mJ / cm 2The second sub-model uses the same three-segment partitioning model parameters (D). th =50, D cl =180, h0=2.5, lg(D cl )-lg(D th =0.5563). The synthesis formula for the third sub-model: tan(θ GaN )=(2.5 / W eff )×4.57 The model structure and parameters remain unchanged; the only difference is the target angle input in step S3.

[0089] (3) Corresponding step S3 (reverse calculation of mask transition zone design parameters): Target GaN trapezoidal mesa angle θ target =70°. First, calculate the sidewall angle of the photoresist target: tan(θ PR,target )=tan(70°)×1.83 / (10.0×0.95×0.88)=2.7475×1.83 / 8.36=5.028 / 8.36=0.601 θ PR,target =arctan(0.601)=31.0° Then calculate the effective transition zone width: W eff,target =2.5 / 0.601=4.16μm Compared with Example 1, the target angle increased from 65° to 70°, the photoresist target sidewall angle increased from 25.2° to 31.0°, and the effective transition region width decreased from 5.32μm to 4.16μm. This reflects the core control mechanism of the full-link model: a larger GaN mesa angle requires a steeper photoresist slope, which is achieved by narrowing the transition region width while keeping the photoresist parameters unchanged.

[0090] A uniform gradient distribution with N=4 was used. Since the photoresist type remained unchanged (D... th D cl (Same as E0), the transmittance of each sub-region is the same as in Example 1: T1=34.4%, T2=47.4%, T3=65.3%, T4=90.0%. However, the width of each sub-region is narrowed to: w i =4.16 / 4 = 1.04 μm (1.33 μm in Example 1) Self-consistency verification: W eff =4×1.04=4.16μm=W eff,target The corresponding residual photoresist thickness is also: h0 = 2.50 μm at the inner boundary, h r(T1) = 1.875 μm, h r (T2) = 1.25 μm, h r (T3) = 0.625 μm, h r (T4) = 0 μm. The thickness change per level remains 0.625 μm, but the slope is steeper due to the narrower width of the sub-region.

[0091] (4) Corresponding to step S4 (fabrication of halftone mask): Fabricate the halftone mask according to the above design parameters. Compared with Example 1, the only difference is that the width of each sub-region of the transition area is narrowed from 1.33μm to 1.04μm, while the transmittance gradation remains unchanged. The graphic area corresponding to each MicroLED pixel is also set with a central fully opaque area (0%), an edge transition area composed of 4 sub-regions (each sub-region is 1.04μm wide, with transmittances of 34.4%, 47.4%, 65.3%, and 90.0%, respectively), and an outer fully transparent area (100%). The transmittance of each sub-region is achieved by a sub-resolution grating pattern, with a grating period of 180nm, and the width of the opaque lines in each sub-region is the same as in Example 1.

[0092] Corresponding to step S5 (exposure and development): AZ4620 photoresist is coated onto the SiO2 surface of the GaN / ITO / SiO2 substrate to a thickness of 2.5 μm. The halftone mask prepared in step S4 is used at the optimal focal plane position with E0 = 200 mJ / cm². 2 Exposure is performed. After development, the photoresist at the corresponding location in the transition region forms a contour that descends from 2.50 μm to 0 μm in a stepped manner, with residual thicknesses of 1.875, 1.25, 0.625, and 0 μm in the four sub-regions, respectively. Due to the narrower transition region and steeper slope, the equivalent sidewall angle is: θ PR =arctan(2.50 / 4.16)=arctan(0.601)=31.0° This aligns with the design goals outlined in step S3.

[0093] (5) Corresponding to step S6 (multi-layer etching to form GaN trapezoidal mesa): First, using the photoresist slope as a mask, the SiO2 layer is etched using the CF4 / CHF3 system ICP, with ITO as the etching stop layer. The sidewall corners of the formed SiO2 hard mask: θ SiO2 =arctan(tan(31.0°)×(1 / 1.83)×0.95)=arctan(0.601×0.519)=arctan(0.312)≈17.3° Then, using a SiO2 hard mask as a substrate, the GaN layer was etched using ICP with a Cl2 / BCl3 system. The final GaN trapezoidal mesa angle is: θGaN =arctan(tan(17.3°)×10.0×0.88)=arctan(0.312×8.80)=arctan(2.746)≈70.0° The final result of this embodiment is that the measured angle of the GaN trapezoidal mesa is approximately 70.0°, which perfectly matches the target angle of 70°. In contrast to Embodiment 1, both embodiments, under the same process platform and process constants, successfully adjusted the GaN mesa angle from 65° to 70° simply by adjusting the width of the sub-region of the halftone mask transition area (from 1.33 μm to 1.04 μm). This fully verifies the applicability, computational consistency, and reproducibility of the end-to-end quantitative mapping model under different target angles.

[0094] like Figure 6 As shown, the calculation results (W) of the 60° target angle in the technical principle analysis are presented. eff =6.60μm, N=5, width of each sub-region 1.32μm), the three sets of data together confirm the model's ability to continuously adjust the target angle within the commonly used angle range of 60°-75°.

[0095] like Figure 7 As shown, compared with the defocusing exposure scheme, this method has an angle deviation of <±1.5° (the defocusing scheme exceeds ±5°) under the condition of exposure dose fluctuation of ±5%, and an angle deviation of <±0.5° under the condition of focal plane offset of ±0.3μm, showing a significantly wider process window.

[0096] like Figure 8 As shown, the present invention also provides a MicroLED trapezoidal mesa system based on a halftone mask, the system comprising: The process parameter acquisition module is used to acquire and store photoresist parameters and etching parameters. The photoresist parameters include initial coating thickness h0, contrast γ, and threshold dose D. th Clearance dose D cl The etching parameters include the SiO2 etching selectivity ratio to the photoresist (S1), the GaN etching selectivity ratio to SiO2 (S2), the SiO2 etching process correction factor (K1), and the GaN etching process correction factor (K2). This module also automatically performs a consistency check when storing parameters to ensure that γ matches D. th D cl satisfy The mathematical relationship. This module corresponds to the process constant determination function in step S1 of the aforementioned method embodiment.

[0097] The end-to-end mapping calculation module is used to establish an end-to-end quantitative mapping model based on the photoresist parameters and the etching parameters. The end-to-end quantitative mapping model uses a comprehensive formula... This module characterizes the quantitative relationship between the design parameters of the halftone mask transition region and the angle of the GaN trapezoidal mesa. It contains the computational logic of three cascaded sub-models, corresponding to the end-to-end mapping model establishment function in step S2 of the aforementioned method embodiment.

[0098] The reverse design module receives the target GaN trapezoidal mesa angle, calculates the photoresist target sidewall angle and effective transition region width based on the full-link quantitative mapping model, and calculates the width and transmittance of each sub-region of the transition region according to the number of grades and transmittance allocation rules (including equal thickness gradient allocation rules). This module also performs self-consistency verification when outputting design parameters, confirming W... eff =Σ_{i=1}^{N}w i With W eff,target Consistent. This module corresponds to the reverse design calculation function in step S3 of the aforementioned method embodiment.

[0099] The correction factor calibration module is used to obtain and update the process correction factors K1 and K2 based on the measured GaN mesa angle data of the calibration mask using the least squares method. This module supports both a comprehensive calibration mode (fitting K = K1 × K2) and an independent calibration mode (fitting K1 and K2 separately using intermediate measurement data after SiO2 etching), and automatically updates the K1 and K2 values ​​stored in the process parameter acquisition module after calibration. This module corresponds to the calculation function of the process correction factor calibration in step S1 of the aforementioned method embodiments. The specific functions and implementation methods of the above modules correspond to the corresponding steps in the aforementioned method embodiments and will not be repeated here.

Claims

1. A method for fabricating a trapezoidal mesa for MicroLEDs based on a halftone mask, characterized in that, The method includes the following steps: Step S1: Determine process constants and calibrate process correction factors. The process constants include photoresist parameters and etching parameters. The photoresist parameters include initial coating thickness h0 and threshold dose D. th Clearance dose D cl The etching parameters include the photoresist contrast ratio γ and the base exposure dose E0, and the etching parameters include the SiO2 etching selectivity ratio S1 to the photoresist, the GaN etching selectivity ratio S2 to SiO2, the SiO2 etching process correction factor K1, and the GaN etching process correction factor K2. Step S2: Based on the process constants, establish a full-link quantitative mapping model from the halftone mask transition region design parameters to the final GaN trapezoidal mesa angle. This full-link quantitative mapping model includes a cascaded first sub-model of mask transmittance to exposure dose distribution, a second sub-model of exposure dose to photoresist residual thickness profile and equivalent sidewall angle, and a third sub-model of angle mapping for multi-layer etching transfer. The comprehensive formula for the full-link quantitative mapping model is: W eff The effective horizontal distance of the transition zone is equal to the sum of the widths of all sub-regions within the transition zone; Step S3: Based on the target GaN trapezoidal mesa angle θ target Based on the aforementioned end-to-end quantitative mapping model, the design parameters of the halftone mask transition region are calculated in reverse, including the inverse calculation of the photoresist target sidewall angle θ. PR,target Calculate the effective transition zone width W eff,target And calculate the width w of each sub-region of the transition zone according to the grade number N and the transmittance allocation rule. i and transmittance T i ; Step S4: Fabricate a halftone mask according to the design parameters determined in step S3. The graphic area corresponding to each MicroLED pixel on the halftone mask is divided into a central fully opaque area, an edge transition area, and a peripheral fully transparent area. The transition area is divided into N sub-regions, and the transmittance of each sub-region is achieved through a sub-resolution grating pattern. Step S5: Exposure and development using a halftone mask to form a photoresist slope profile. Specifically, positive photoresist is coated onto a GaN epitaxial substrate with sequentially deposited ITO and SiO2 hard mask layers to an initial coating thickness h0. Using the halftone mask fabricated in step S4, the photoresist is exposed and developed at the optimal focal plane position with a base exposure dose E0, forming a profile with an equivalent sidewall angle θ. PR The photoresist slope profile; Step S6: Sequentially etch the hard mask layer and the GaN layer to form a GaN trapezoidal mesa. Use a photoresist with a slope profile as a mask to etch the SiO2 hard mask layer, and use the ITO layer as the etching stop layer to form a structure with a sidewall angle θ. SiO2 A SiO2 hard mask slope is then used; subsequently, the GaN layer is etched using the SiO2 hard mask as a mask to form a target angle θ. GaN GaN trapezoidal platform.

2. The method for fabricating a MicroLED trapezoidal mesa based on a halftone mask according to claim 1, characterized in that, In step S2, the first sub-model is: the effective exposure dose D received by the photoresist at the i-th sub-region of the transition region. i Equal to the transmittance T of this sub-region i The product of the base exposure dose E0, i.e. ; The second sub-model is a three-segment partitioned photoresist residual thickness model: when D i <D th At that time, the residual thickness h r (D i )=h0; when D th ≤D i ≤D cl hour, When D i >D cl At that time, h r (D i =0; Equivalent sidewall angle θ of photoresist PR satisfy ,in The summation range is from 1 to N; The third sub-model is a two-stage etching angle transfer model: the relationship between the SiO2 hard mask sidewall angle and the photoresist sidewall angle is as follows: The relationship between the GaN mesa angle and the SiO2 sidewall angle is as follows: ; The photoresist contrast γ and threshold dose D th Clearance dose D cl Satisfies strict mathematical consistency relation: .

3. The method for fabricating a MicroLED trapezoidal mesa based on a halftone mask according to claim 1, characterized in that, The process correction factors K1 and K2 mentioned in step S1 are calibrated in the following way: A calibration mask is prepared, wherein a set of transition regions with a total physical width W is provided on the calibration mask. HT A series of gradient halftone test patterns were generated, with each group of test patterns using the same transmittance grading method. The complete photolithography, SiO2 etching, and GaN etching process was performed on a standard GaN / ITO / SiO2 laminated substrate using the aforementioned calibration mask. The actual GaN mesa angle θ corresponding to each group of test patterns was measured using cross-sectional SEM. GaN,meas ; The calibration methods include two types: integrated calibration mode and independent calibration mode. The comprehensive calibration model is as follows: a comprehensive correction factor K = K1 × K2 is fitted using the least squares method, with the fitting objective being to sum over all test groups j. Take the minimum value, where W eff (j) represents the sum of the widths of each sub-region of the transition area in the j-th test pattern, i.e., the total physical width W of the transition area in this group of test patterns. HT (j); The independent calibration mode is as follows: after the SiO2 etching step is completed and before the GaN etching step begins, cross-sectional SEM measurements are performed on each set of test patterns on the calibration mask to obtain the actual sidewall angle θ of the SiO2 hard mask. SiO2,meas K1 was independently fitted using the least squares method; then, the measured angle after GaN etching and the calibrated K1 were used as the basis. Independently fit K2; The value of K1 ranges from 0.85 to 1.15, and the value of K2 ranges from 0.80 to 1.

10.

4. The method for fabricating a MicroLED trapezoidal mesa based on a halftone mask according to claim 1, characterized in that, The reverse calculation described in step S3 includes the following sub-steps: S3.1 Calculate the target sidewall angle of the photoresist based on the inverse calculation form of the end-to-end synthesis formula: ; S3.2 Calculate the required effective transition area width directly from the photoresist sidewall angle definition: ; S3.

3. Determine the transition level N (N is greater than or equal to 3), and use the equal thickness gradient allocation rule to allocate parameters to the sub-regions, ensuring that each sub-region bears an equal amount of photoresist thickness variation h0 / N, and that the width of each sub-region is equal. The transmittance of each sub-region is determined by the following formula: ; S3.

4. Verify the self-consistency of the design parameters and confirm their accuracy. The summation range is from 1 to N.

5. The method for fabricating a MicroLED trapezoidal mesa based on a halftone mask according to claim 4, characterized in that, The uniform thickness gradient distribution rule enables the photoresist to form a stepped profile with optimal linearity in the transition region, ensuring that the residual thickness of the photoresist decreases uniformly and at equal intervals between each sub-region. The transmittance range of each sub-region calculated by this rule falls within the effective transmittance range D corresponding to the working range of the photoresist. th / E0 to D cl Within / E0, the 0% transmittance of the central fully shaded area and the 100% transmittance of the outer fully transparent area belong to different levels.

6. The method for fabricating a MicroLED trapezoidal mesa based on a halftone mask according to claim 1, characterized in that, In step S4, the light transmittance of the central fully opaque area is 0%, corresponding to the area where the photoresist retains its full thickness, and is used to define the top of the GaN trapezoidal mesa; the light transmittance of the outer fully transparent area is 100%, corresponding to the area where the photoresist is completely removed, and is used to define the bottom of the GaN trapezoidal mesa; the edge transition area is divided into N sub-regions along the direction from the central fully opaque area to the outer fully transparent area, with a light transmittance distribution that increases from the inside to the outside; Within each sub-region, a period p is passed through a point smaller than the resolution limit of the exposure system. i Arranging light-blocking lines and light-transmitting gaps to achieve a sub-resolution grating pattern, with a grating period p i satisfy Where λ is the exposure wavelength and NA is the numerical aperture; the transmittance T of the i-th sub-region i The duty cycle of the light-blocking lines determines the relationship. , where d i The width of the light-blocking line.

7. The method for fabricating a MicroLED trapezoidal mesa based on a halftone mask according to claim 1, characterized in that, The GaN epitaxial layer has a thickness of 3 to 5 micrometers, and the target GaN trapezoidal mesa angle θ target The value ranges from 60 degrees to 75 degrees; When the base exposure dose E0 fluctuates within ±5% of the nominal value, the angle deviation of the fabricated GaN mesa is less than ±1.5 degrees; when the focal plane shifts by ±0.3 micrometers, the angle deviation of the fabricated GaN mesa is less than ±0.5 degrees.

8. A MicroLED trapezoidal mesa system based on halftone masks, characterized in that, The system is used to perform the steps involved in establishing a full-link quantitative mapping model, calculating inverse parameters, and calibrating process correction factors in the method of any one of claims 1 to 7. The system includes: The process parameter acquisition module is used to acquire and store photoresist parameters and etching parameters, including initial coating thickness h0, contrast γ, and threshold dose D. th Clearance dose D cl The etching parameters, including the SiO2 etching selectivity ratio to the photoresist (S1), the GaN etching selectivity ratio to SiO2 (S2), the SiO2 etching process correction factor (K1), and the GaN etching process correction factor (K2), are set to the base exposure dose E0. The module automatically performs consistency checks when storing parameters to ensure that γ matches D. th D cl satisfy Mathematical relationships; The end-to-end mapping calculation module is used to establish an end-to-end quantitative mapping model based on the photoresist parameters and the etching parameters. The end-to-end quantitative mapping model is established through a comprehensive formula. A quantitative relationship is established between the design parameters of the transition zone of the halftone mask and the angle of the GaN trapezoidal mesa. The reverse engineering module receives the target GaN trapezoidal mesa angle, calculates the target sidewall angle and effective transition region width of the photoresist based on the full-link quantitative mapping model, and calculates the width and transmittance of each sub-region of the transition region according to the number of levels and transmittance allocation rules including equal thickness gradient allocation rules, and then executes... The self-consistency verification is performed, and the summation range is from 1 to N; The correction factor calibration module is used to obtain and update the process correction factors K1 and K2 by fitting the measured GaN mesa angle data based on the calibration mask using the least squares method. The module supports a comprehensive calibration mode of fitting K=K1×K2 and an independent calibration mode of fitting K1 and K2 respectively by combining intermediate measurement data after SiO2 etching. After calibration, the module automatically updates the K1 and K2 values ​​stored in the process parameter acquisition module.

Images