A multi-focal energy gradient laser stealth cutting method for hard and brittle materials

Abstract

A method for multifocal energy gradient laser holographic cutting of hard and brittle materials includes the following steps: S1, determining the parameters of the incident beam and the high numerical aperture focusing objective; S2, determining the three-dimensional multifocal beam and parameters of the target hard and brittle material; S3, setting a two-dimensional linear phase distribution on the spatial light modulator plane as the initial phase for subsequent iterative calculations; S4, establishing a forward-backward propagation iterative loop based on phase optimization to calculate the multifocal holographic phase distribution; S5, superimposing a linear blazed grating phase on the multifocal holographic phase distribution, and generating a composite diffraction phase hologram using MATLAB code and inputting it into the spatial light modulator; achieving uniform holographic cutting with consistent actual absorbed energy in each layer. This invention features controllable three-dimensional focal positions, adjustable energy gradients along the depth direction at each focal point, strong deep-layer energy compensation capability, simultaneous aberration correction, and high consistency in cutting at different depths.

Description

Technical Field

[0001] This invention relates to the fields of laser precision machining and optical holography, specifically to a multi-focal energy gradient laser hidden cutting method for hard and brittle materials. Background Technology

[0002] Hard and brittle materials (such as silicon carbide, optical glass, and silicon) are a special class of engineering materials, typically possessing high hardness, high wear resistance, and excellent thermal and chemical stability, making them valuable in aerospace, optical devices, microelectronics, and precision machinery. However, due to their high hardness and brittleness, traditional mechanical diamond cutting easily leads to edge chipping, microcrack propagation, and increased cross-sectional roughness, resulting in low device yield and poor reliability. Laser stealth cutting technology achieves non-contact, chip-free cutting by forming a modified layer inside the material, significantly improving surface quality.

[0003] However, most existing laser stealth cutting processes employ a single-focus, layer-by-layer scanning method. For example, Yang et al. (Micromachines 2022, 13(7), 1011) modified the focal depth layer by layer and adjusted the incident power by 2% per layer to cut a 508μm thick 4H laser. Processing SiC wafers still relies on inefficient multi-layer repeated scanning and struggles to achieve independent and precise control of axial energy distribution. Results show that due to crystal anisotropy, the surface roughness varies by approximately 20% across different crystal orientations, reflecting the limited consistency and process window of this technology. The root of this problem lies in the extremely high refractive index and absorptivity of hard and brittle materials like silicon carbide (SiC). Laser propagation within these materials results in severe refractive distortion and energy loss. This leads to a serious process contradiction: to ensure consistent modification effects at different depths, the actual energy absorbed by each layer must remain constant. However, in actual processing, improper incident energy distribution at deeper focal points causes a sharp decrease in energy due to energy loss along the way, ultimately leading to phenomena such as the inability to cut deeper layers or extremely uneven modified layers. This becomes a major bottleneck restricting high-quality stealth cutting. While some industry solutions propose layered energy gradient cutting (i.e., manually increasing the incident energy of deeper layers during layer-by-layer scanning), this layer-by-layer processing method is not only inefficient but also makes it difficult to precisely match the positions of modified layers between different layers. Traditional three-dimensional multifocal parallel processing technology lacks the ability to independently and precisely control the energy ratio of each focal point along the axial direction, failing to achieve an energy gradient distribution of "high energy distribution in deep layers and low energy distribution in shallow layers." To improve cutting efficiency and achieve fine control over energy distribution in the depth direction, spatial light modulators (especially LCOS-type SLMs) and computational generative holography (CGH) technology have been introduced in recent years. Axial / three-dimensional multifocal light fields are generated through wavefront encoding for parallel processing. Existing technologies include various phase retrieval algorithms for generating multifocals, such as the Gerchberg-Saxton (GS) algorithm and its weighted improved version, the GSW algorithm. However, the traditional GSW algorithm mainly optimizes the uniformity of focal points in a two-dimensional plane, exhibiting poor scalability for the number of focal points in three-dimensional space and failing to achieve independent control over the energy ratio of focal points at different depths. This limits its application in laser stencil cutting processes that require fine control over the energy of each layer. In addition, existing algorithms still need improvement in terms of convergence speed, computational efficiency and light field uniformity. Furthermore, they generally do not consider the impact of spherical aberration introduced by the difference in refractive index during material processing on the focal point quality, resulting in deviations between the actual processing effect and the design target.

[0004] Therefore, there is an urgent need to develop a method that can generate a three-dimensional multifocal beam with an energy gradient in a single operation using a spatial light modulator (SLM). Summary of the Invention

[0005] To overcome the shortcomings of the existing technologies, this invention provides a multi-focal energy gradient laser stencil method for hard and brittle materials. This method proposes an iterative phase generation algorithm (hereinafter referred to as DO-3D-GSW) that seamlessly combines a three-dimensional beam energy ratio adaptive control mechanism with spherical aberration correction caused by the material's refractive index. It uses a spatial light modulator (SLM) to shape a single laser beam into a multi-focal array with energy gradients in one go. It features controllable three-dimensional focal positions, adjustable energy gradients of each focal point along the depth direction, strong deep energy compensation capability, synchronous aberration correction, and high consistency of cutting at different depths.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for multi-focal energy gradient laser concealed cutting of hard and brittle materials includes the following steps; S1. Determine the parameters of the incident beam and the high numerical aperture focusing objective. S2. Determine the three-dimensional multifocal beam of the target hard and brittle material and the parameters of the target hard and brittle material; S3. Based on the parameters determined by S1 and S2, a two-dimensional linear phase distribution with a constant slope is set on the spatial light modulator plane as the initial phase for subsequent iterative calculations; this realizes the initialization phase and aberration compensation calculations. S4. Establish a forward-backward propagation iterative loop based on phase optimization to calculate the phase distribution of the multifocal hologram; The forward propagation process is calculated based on the two-dimensional linear initial phase, the pure spherical aberration corrected phase, and the free space propagation function corresponding to the three-dimensional spatial coordinates of each focus. The backward propagation process updates the weights according to the deviation between the actual light intensity of each focus and the expected energy gradient weights, thereby obtaining a multi-focus energy gradient phase distribution that meets the requirements of hidden cutting of hard and brittle materials. S5. Superimpose a linear blazed grating phase onto the multifocal holographic phase distribution, and use MATLAB code to generate a composite diffraction phase hologram from the superimposed phase function and input it into the spatial light modulator. After the incident beam is modulated by the spatial light modulator, the zero-order diffracted light is filtered out by the post-positioned 4F spatial filter system. After being focused by the focusing objective lens, a multi-focal energy gradient beam with high energy in the deep layer and low energy in the shallow layer and the influence of spherical aberration is generated in one go inside the target hard and brittle material. This compensates for the absorption and refraction loss generated inside the hard and brittle material as the depth increases, and achieves uniform stealth cutting with consistent actual absorption energy in each layer.

[0007] Preferably, in step S1, the incident beam is a laser beam with a Gaussian spot energy spatial distribution; the incident beam is used to perform internal hidden cutting on the target hard and brittle material. The parameters of the incident beam include: the wavelength of the incident beam, the beam waist radius, and the spatial distribution of the beam energy. The parameters of the high numerical aperture focusing objective include numerical aperture (NA).

[0008] The incident beam parameters are determined by the optical penetration characteristics of the target hard and brittle material and the requirements of the hidden cutting process; the focusing objective parameters are determined by the energy density requirements of the internal nonlinear absorption and the need for aberration calculation.

[0009] In S2, the parameters of the three-dimensional multifocal beam for hard and brittle materials include: the number of focal points, the three-dimensional spatial coordinates of each focal point inside the target hard and brittle material, the processing depth corresponding to each focal point, and the expected energy gradient weight of each target focal point set to accurately compensate for absorption or refraction loss in the depth direction inside the target hard and brittle material, wherein the deeper the spatial depth of the focal point, the greater its expected energy weight; the parameters of the target hard and brittle material include: the refractive index of the target processed material; Among them, the three-dimensional multifocal parameters are determined by the physical dimensions of the target material (such as wafer thickness) and the expected continuity of the internal modified surface; the expected energy gradient weight is determined based on compensating for the absorption and refraction attenuation of the laser as it accumulates with depth inside the material. The rule set is that the deeper the spatial depth of the focal point, the greater its expected energy weight, and the target material parameter is the inherent physical property of the selected material at a specific processing wavelength.

[0010] In step S3, based on the numerical aperture of the focusing objective, the processing depth of each focal point, and the refractive index of the processing material, a spherical aberration compensation phase, i.e., a pure spherical aberration correction phase, is calculated to compensate for the spherical aberration caused by the sudden change in refractive index when the laser penetrates the target hard and brittle material. The pure spherical aberration correction phase is used to compensate for the spherical aberration caused by the discontinuity of refractive index and the increase in focusing depth after the laser enters the interior of the target hard and brittle material from the air. Spherical aberration refers to the aberration that light rays of different apertures cannot simultaneously converge at the same ideal focal point inside the target hard and brittle material, resulting in focal spot stretching, energy divergence, and a decrease in the quality of deep processing.

[0011] The expression for the pure spherical aberration correction phase is: in, λ The incident wavelength, NA This is the numerical aperture of the focusing objective lens. ρ For normalized radial coordinates, n 2 represents the refractive index of the target material being processed. dLet be the center depth of the target three-dimensional multifocal array. The coefficient s is determined by inner product calculation and reflects the influence of the material's refractive index on the effective focusing depth. This coefficient can be obtained through numerical integration or approximated by ray tracing: s ≈ n².

[0012] Preferably, in step S3, the two-dimensional linear phase distribution corresponds to the target three-dimensional multifocal beam parameters. The role of the two-dimensional linear phase distribution as the initial phase for iteration is to provide a stable initial solution for subsequent phase optimization based on the known number of focal points, three-dimensional spatial coordinates, processing depth, and desired energy gradient weight, thereby improving the convergence speed and iteration stability of multifocal energy allocation.

[0013] Preferably, in step S3, the two-dimensional linear phase distribution is chosen as the initial phase for iterative calculation mainly because, comparing the three initial phase strategies—all-zero phase, random phase, and linear phase—it can be seen that: the all-zero phase improves uniformity quickly in the early stages, but uniformity decays in the later stages of iteration, resulting in insufficient stability in long-sequence iterations; the random phase has high initial uniformity, but the convergence process fluctuates violently and the numerical oscillation is obvious, which is not conducive to result prediction and reproduction; while the linear phase improves slowly in the first few iterations, but the subsequent convergence speed is fast, requiring only about 10 iterations to achieve a focal uniformity of over 0.85, and the iteration curve is smooth and stable throughout without performance regression. It performs best in terms of convergence speed, iterative stability, and final reconstruction quality, and can balance algorithm efficiency and reconstruction accuracy. It can also prevent the algorithm from getting trapped in local minima when calculating multifocals with large energy gradient differences. Therefore, it is selected as the initial phase for subsequent multifocal light field generation.

[0014] In algorithm initialization, a two-dimensional linear phase distribution is used. = c 1 x + c 2 y It provides a stationary initial solution with a constant slope, i.e., the linear coefficients in the formula. c 1 and c 2. The constant slope is constructed based on the set number of focal points, coordinates and weight parameters to avoid numerical oscillations in subsequent iterations. It is independent of the phase of pure spherical aberration correction, but plays a joint role in the superposition of physical fields. Aberration compensation calculation refers to the calculation of the pure spherical aberration phase matrix in advance using geometric optical path difference to compensate for focal spot distortion caused by abrupt changes in high refractive index. And force it to be embedded into the forward and backward propagation models of each iteration to achieve aberration correction synchronously; The correspondence between the two-dimensional linear phase distribution and the three-dimensional multifocal beam parameters of the target hard and brittle material refers to the fact that, under the premise of knowing the number of focal points, coordinates and energy weights, a constant slope phase is preset on the coordinate system corresponding to each SLM pixel. This provides a stable physical starting point for the subsequent iterative calculation of stacking depth and aberration information, which can ensure that the energy gradient converges rapidly and without oscillation.

[0015] Preferably, the forward-backward propagation iterative loop process described in S4 actually constitutes an improved algorithm relative to the traditional weighted Gerchberg-Saxton (GSW) algorithm, namely the DO-3D-GSW algorithm. The improvements of the algorithm compared to the traditional GSW algorithm are specifically reflected in the following three aspects: First, in terms of the algorithm initialization mechanism, it abandons the traditional GSW algorithm's approach of using all-zero phase or random phase matrix as the starting point, and innovatively uses the two-dimensional linear phase with constant slope as the initial phase. This fundamentally eliminates the defects of traditional algorithms that are prone to getting trapped in local optima when calculating multifoci and numerical oscillations in the later stages of iteration, and greatly improves the convergence speed. Second, in the spatial light field propagation dimension, by independently introducing into the propagation model different depth coordinates (z) of each focal point... m The free space propagation function of ) enables the algorithm to be extended from two-dimensional plane to three-dimensional solid space; Third, in terms of physical environment modeling, the traditional GSW algorithm performs optical field extrapolation in an ideal, uniformly refractive medium space. In contrast, this invention forcibly embeds the pre-calculated pure spherical aberration correction phase matrix in each core calculation stage of forward and backward propagation. This improvement transforms the optimized target environment from an ideal vacuum / air to a real physical processing space with severe refractive index abrupt changes (such as entering high-refractive-index silicon carbide from air), achieving simultaneous controllable multifocal energy, homogenization, and aberration compensation.

[0016] Preferably, the iterative loop process in S4 specifically includes: S41 (Forward Propagation Calculation): The initial phase of the spatial light modulator in the current cycle, the pure spherical aberration correction phase obtained in S3, and the free space propagation function corresponding to the three-dimensional spatial coordinates and processing depth of each focus are superimposed to calculate the actual complex amplitude and actual light intensity at each target focus position. in, For the current number kThe phase of the spatial light modulator in the secondary cycle, through complex exponential superposition, forcibly embeds the three-dimensional propagation bias and refractive index aberration compensation, transforming the optimized boundary conditions into the actual interior of the medium; thus, the actual light intensity at the current focal point can be calculated. ; S42 (Adaptive Weight Update): Compare the actual light intensity of each focus with the expected energy gradient weight of each target focus set in S2 to compensate for the material's deep absorption loss, calculate the deviation value, and use the feedback coefficient to adaptively update the weight factor corresponding to each focus. in, Indicates the first In the nth iteration Weighting factors corresponding to each target focus; For the first m The expected energy weight set for each target focus; For the first The complex amplitude modulus of the light field at the focal point after each iteration reflects the relative intensity currently achieved. The adaptive weighted feedback coefficient has a range of values. This mechanism exponentially amplifies the weight of dark focal points and suppresses bright focal points, thereby achieving precise redistribution of system energy.

[0017] S43 (Backpropagation Update): Multiply the updated weighting factor by the actual complex amplitude of each focus, and propagate back to the spatial light modulator plane to superimpose and calculate the phase angle, thus obtaining the updated phase for the next cycle. in, M The total number of target focuses; arg{ The function is used to extract this. M The complex field argument (principal value interval is [ ) after the inverse evolution of each focus and the superposition of the entire matrix) is [ ] π,π]); S44 (Loop Termination Judgment): Calculate the uniformity of the actual light intensity of all currently generated focal points. If the uniformity reaches the set threshold or the maximum number of iterations is reached, terminate the loop and output the current updated phase as the final multifocal diffraction phase distribution.

[0018] In step S42, the adaptive update formula for the weighting factor is: in, Indicates the first In the nth iteration Weighting factors corresponding to each target focus; For the first mThe expected energy weights set for each target focus (the deeper the depth, the higher the energy weight). The larger the set value); For the first After the nth iteration The complex amplitude modulus of the light field at the focal point of each target reflects the relative intensity currently achieved; The adaptive weighted feedback coefficient has a range of values. The preferred value is 0.4 to 0.6.

[0019] In S44, the uniformity is defined as: in, I max and I min These represent the maximum and minimum actual light intensities of each target focus generated in a single iteration. This index effectively reflects the uniformity of light intensity distribution among the focuses; the closer the uniformity is to 1, the more precise the energy allocation control of the algorithm.

[0020] Preferably, in step S5, the purpose of superimposing the linear blazed grating phase is to deflect the target's three-dimensional multifocal light field to the +1st order diffraction direction. By setting a pinhole aperture on the Fourier spectrum of the 4F spatial filter system, and physically ensuring that the pinhole aperture only allows +1st order diffracted light to pass through while blocking the zeroth order diffracted light at the center of the optical axis, the potential thermal damage of the unmodulated beam to the target material is completely eliminated. Preferably, in step S5, the multifocal energy gradient beam with high energy at deep depths and low energy at shallow depths refers to assigning higher incident energy weights to deeper focal points based on the transmission loss differences at different depths within the target hard and brittle material, in order to compensate for absorption and refraction attenuation during laser propagation within the material. The elimination of spherical aberration refers to pre-compensating for the focal spot distortion caused by deep focusing through the pure spherical aberration correction phase, so that each depth focal point forms a uniformly sized and energy-controllable modified region within the target hard and brittle material, thereby achieving high-quality and uniform stealth cutting.

[0021] Preferably, in S5, the spatial light modulator is a pure phase spatial light modulator 5 (SLM). The 4F spatial filtering system includes a first lens 6, a mechanical pinhole aperture 8, and a second lens 10 arranged sequentially along the optical path. It is used to perform spatial filtering and order selection on the beam modulated by the spatial light modulator, filter out the zero-order diffraction light and retain only the target-order diffraction light to enter the focusing objective lens, so as to avoid the zero-order direct light causing local thermal damage to the target hard and brittle material. The mechanical pinhole aperture 8 is set on the Fourier spectrum plane of the two lenses.

[0022] The laser processing hardware optical path system for realizing the multi-focus energy gradient laser hidden cutting method for hard and brittle materials includes an ultrafast infrared laser 1, a half-wave plate 2, a Glan Taylor prism 3, a beam expander 4, and a pure phase spatial light modulator 5 arranged sequentially along the optical path. The ultrafast infrared laser beam emitted by the ultrafast infrared laser 1 is phase-modulated by the pure phase spatial light modulator 5, and after passing through the first lens 6 and the first reflector 7, it is split into two parts: the unmodulated zero-order central direct light and the +1st order diffracted light carrying SiC spherical aberration compensation and multifocal array information; the two beams enter the 4F spatial filtering system, and the position of the mechanical aperture 8 located on the spectral plane is adjusted to completely block the zero-order direct strong light in physical space, allowing only the +1st order diffracted light to pass through and enter the high numerical aperture focusing objective lens 11 with a numerical aperture of 0.65; A second reflecting mirror 9 and a second lens 10 are arranged between the mechanical pinhole aperture 8 and the high numerical aperture focusing objective lens 11, and the SiC wafer 12 to be processed is placed on the three-dimensional high-precision moving platform 13.

[0023] The beneficial effects of this invention are: This invention directly addresses the problems of deep-penetration and uneven layer cutting in the stealth cutting of high-absorption, high-refractive-index hard and brittle materials (such as silicon carbide), overcoming the shortcomings of traditional layered multi-scan cutting, which suffers from low efficiency and is prone to misalignment of modified layers. To this end, this invention utilizes an improved DO-3D-GSW algorithm in conjunction with a spatial light modulator to generate a three-dimensional multi-focal array with incident energy increasing with depth in a single scan. Through a light blade design that distributes high energy to deep layers and low energy to shallow layers, it precisely counteracts the absorption and refraction losses of the laser accumulated with depth within the material, ultimately achieving absolute consistency in actual absorbed energy at different depths, ensuring extremely high cutting yield and processing efficiency. Simultaneously, to ensure the sharpness of the deep, high-energy focal point, this invention independently introduces a three-dimensional free-space propagation function and forcibly embeds a pure spherical aberration correction phase in the algorithm's bottom-level iterative loop, completely eliminating deep-focus distortion and energy divergence caused by laser penetration of high-refractive-index materials.

[0024] Furthermore, this method uses a two-dimensional linear initial phase to eliminate numerical oscillations in the multi-focus energy redistribution process, achieving extremely high convergence speed. In the post-processing stage, by superimposing blazed gratings and combining them with the soft and hard closed-loop design of the 4F spatial filtering system, the overburning of the wafer shallow layer caused by zero-order strong light is completely eliminated, ultimately fully meeting the stringent industrial requirements for speckle-free, high-quality, uniform, and stealth dicing of semiconductor wafers.

[0025] This invention can precisely control the energy gradient and position of each focal point, has a high convergence speed, and can compensate for the influence of system aberrations, thereby effectively compensating for the absorption loss caused by the material depth, and finally achieving high-quality cutting with consistent depths, so as to improve the process quality of stealth cutting of high-hardness and brittle materials such as silicon carbide. Attached Figure Description

[0026] Picture 1 This is a flowchart of a multi-focus energy gradient shaping method for laser hidden cutting of hard and brittle materials according to the present invention.

[0027] Picture 2 The diagram shows a schematic of the laser processing hardware optical path system upon which the embodiments of the present invention depend.

[0028] Picture 3 The diagram shows the target eight-focus diffraction phase hologram generated by the DO-3D-GSW algorithm iteration of this invention, the simulation results of the position and intensity of the eight-focus light field on the xz section, and the simulation curves of the position and intensity of each focus in the z direction (the diagram shows that the energy of each focus is distributed in a gradient along the z axis that increases with depth).

[0029] Picture 4 This is a comparison of the processing results of single-crystal SiC wafers cut by eight-focus energy gradient in an embodiment of the present invention with the results of traditional equal-energy processing without aberration correction, showing the morphology of the wafer edge and the modified layer.

[0030] Explanation of reference numerals in the attached figures: 1 Ultrafast infrared laser; 2 Half-wave plate; 3 Glan Taylor prism; 4 Beam expander; 5 Pure phase spatial light modulator (SLM); 6 First lens; 7 First reflecting mirror; 8 Mechanical pinhole aperture; 9 Second reflecting mirror; 10 Second lens; 11 High numerical aperture focusing objectives; 12 SiC wafers to be processed; 13 Three-dimensional high-precision mobile platform. Detailed Implementation

[0031] The present invention will now be described in further detail with reference to the accompanying drawings.

[0032] refer to Pictures 1-4This invention provides a multifocal energy gradient laser stealth cutting method for hard and brittle materials. Based on the DO-3D-GSW algorithm proposed in this invention, the multifocal diffraction phase distribution of the target is calculated through iterative optimization. This phase distribution is then input into a spatial light modulator, and finally, after passing through a focusing objective lens, a multifocal energy gradient beam with "high energy in deep layers and low energy in shallow layers" and eliminating spherical aberration can be generated in one step inside the target material, achieving uniform stealth cutting with consistent actual energy absorption in each layer.

[0033] This invention provides a multi-focus energy gradient laser lithography method for hard and brittle materials, comprising: S1. Determine the parameters of the incident beam and the focusing objective lens; The incident beam is used for internal hidden cutting of hard and brittle materials. The parameters of the incident beam include: wavelength, beam waist radius, and spatial distribution of beam energy. The parameters of the focusing objective include numerical aperture (NA). S2. Determine the parameters of the target three-dimensional multifocal beam and the target processing material; The parameters of the target three-dimensional multifocal beam include: the number of focal points, the three-dimensional spatial coordinates of each focal point within the target hard and brittle material, the processing depth corresponding to each focal point, and the expected energy gradient weight of each target focal point set to accurately compensate for absorption / refractive loss in the depth direction within the hard and brittle material, wherein the deeper the focal point, the greater its expected energy weight; the parameters of the target hard and brittle material include: the refractive index of the target processed material; S3. Initialize phase and aberration compensation calculations; Based on the numerical aperture of the focusing objective, the processing depth of each focal point, and the refractive index of the target hard and brittle material, a spherical aberration compensation phase, i.e., a pure spherical aberration correction phase, is calculated to compensate for the abrupt change in refractive index when the laser penetrates the target hard and brittle material. At the same time, based on the parameters determined by S1 and S2, a two-dimensional linear phase distribution with a constant slope is set on the spatial light modulator plane as the initial phase for subsequent iterative calculations. S4. Establish a forward-backward propagation iterative loop based on phase optimization to calculate the multi-focus holographic phase distribution; wherein, the forward propagation process is calculated based on the initial phase of the spatial light modulator, the pure spherical aberration corrected phase, and the free space propagation function corresponding to the three-dimensional spatial coordinates of each focus; the backward propagation process updates the weights according to the deviation between the actual light intensity of each focus and the expected energy gradient weights, thereby obtaining the multi-focus energy gradient phase distribution that meets the requirements of the hidden cutting of hard and brittle materials; S5. Superimpose a linear blazed grating phase onto the multifocal holographic phase distribution, and use MATLAB code to generate a composite diffraction phase hologram from the superimposed phase function and input it into the spatial light modulator. After the incident beam is modulated by the spatial light modulator, the zero-order diffracted light is filtered out by the post-positioned 4F spatial filter system. Finally, after being focused by the focusing objective lens, a multi-focal energy gradient beam with high energy in the deep layer and low energy in the shallow layer and with the influence of spherical aberration is generated in one go inside the target hard and brittle material. This compensates for the absorption and refraction loss generated inside the hard and brittle material as the depth increases, and achieves uniform stealth cutting with consistent actual absorption energy in each layer.

[0034] like Picture 1 The diagram shows a flowchart of a multi-focus energy gradient shaping method for laser hidden cutting of hard and brittle materials provided by an embodiment of the present invention.

[0035] The incident beam is a laser beam with a Gaussian spot energy spatial distribution.

[0036] The expression for the pure spherical aberration correction phase is: in, λ The incident wavelength, NA This is the numerical aperture of the focusing objective lens. ρ For normalized radial coordinates, n 2 represents the refractive index of the target material being processed. d Let be the center depth of the target three-dimensional multifocal array. The coefficient s is determined by inner product calculation and reflects the influence of the material's refractive index on the effective focusing depth. This coefficient can be obtained through numerical integration or approximated by ray tracing: s ≈ n².

[0037] The aforementioned forward-backward propagation iterative loop process actually constitutes an improved algorithm compared to the traditional weighted Gerchberg-Saxton (GSW) algorithm, namely the DO-3D-GSW algorithm. The improvements of this algorithm compared to the traditional ordinary GSW algorithm are specifically reflected in the following three aspects: First, in terms of the algorithm initialization mechanism, it abandons the traditional GSW algorithm's approach of using all-zero phase or random phase matrix as the starting point, and innovatively uses the two-dimensional linear phase with constant slope as the initial phase. This fundamentally eliminates the defects of traditional algorithms that are prone to getting trapped in local optima when calculating multifoci and numerical oscillations in the later stages of iteration, and greatly improves the convergence speed. Second, in the spatial light field propagation dimension, by independently introducing into the propagation model different depth coordinates (z) of each focal point... m The free space propagation function of ) enables the algorithm to be extended from two-dimensional plane to three-dimensional solid space; Third, in terms of physical environment modeling, the traditional GSW algorithm performs optical field extrapolation in an ideal, uniformly refractive medium space. In contrast, this invention forcibly embeds the pre-calculated pure spherical aberration correction phase matrix in each core calculation stage of forward and backward propagation. This improvement transforms the optimized target environment from an ideal vacuum / air to a real physical processing space with severe refractive index abrupt changes (such as entering high-refractive-index silicon carbide from air), achieving simultaneous controllable multifocal energy, homogenization, and aberration compensation.

[0038] Preferably, the iterative loop process in S4 specifically includes: S41 (Forward Propagation Calculation): The initial phase of the spatial light modulator in the current cycle, the pure spherical aberration correction phase obtained in S3, and the free space propagation function corresponding to the three-dimensional spatial coordinates and processing depth of each focus are superimposed to calculate the actual complex amplitude and actual light intensity at each target focus position. S42 (Adaptive Weight Update): Compare the actual light intensity of each focus with the expected energy gradient weight of each target focus set in S2 to compensate for the material's deep absorption loss, calculate the deviation value, and use the feedback coefficient to adaptively update the weight factor corresponding to each focus. S43 (Backpropagation Update): Multiply the updated weighting factor by the complex amplitude of each focus, and propagate back to the spatial light modulator plane to superimpose and calculate the phase angle, thus obtaining the updated phase for the next cycle. S44 (Loop Termination Judgment): Calculate the uniformity of the actual light intensity of all currently generated focal points. If the uniformity reaches the set threshold or the maximum number of iterations is reached, terminate the loop and output the currently updated phase as the final multifocal diffraction phase distribution.

[0039] The purpose of superimposed linear blazed grating phase is to deflect the target three-dimensional multifocal light field to the +1st order diffraction direction; by setting a pinhole aperture on the Fourier spectrum surface of the 4F spatial filter system, and in physical space making the pinhole aperture allow only the +1st order diffraction light to pass through while blocking the zeroth order diffraction light at the center of the optical axis, the potential thermal damage of the unmodulated beam to the target processing material is completely eliminated.

[0040] DO-3D-GSW Algorithm Principle: First, during the initialization phase, the algorithm constructs a discrete two-dimensional coordinate system that corresponds one-to-one with the pixels of the physical space light modulator. x , y Based on this, two core physical bias matrices that run the entire system are pre-calculated. One is a free-space propagation function that breaks the limitations of traditional two-dimensional propagation. This function establishes the path from the spatial light modulator plane to any three-dimensional target focus within the material. x m ,y m , z m Spatial mapping of ) in, λ The wavelength of the laser in a vacuum. f Let be the focal length of the focusing objective. The first term on the right-hand side of the formula is a linear phase term, the gradient of which determines the lateral position of the focal point. x m , y m The second term is a quadratic phase term, the curvature of which is directly related to the axial position of the focus. z m .

[0041] Secondly, regarding the focal stretching phenomenon caused by the light beam penetrating air and entering a high-refractive-index material, a pure spherical aberration correction phase is derived and calculated based on the principle of optical path difference in geometric optics: in, NA This is the numerical aperture of the objective lens; Normalized radial coordinates of the pupil plane ( R max (where the pupil radius is) d The depth of the center of the target multifocal array within the material.

[0042] After completing the pre-calculation, to avoid oscillations caused by getting trapped in local minima during multi-focus optimization, the algorithm assigns the spatial light modulator a linear initial phase with a constant slope. And set the initial weighting factor for all target focuses. w m (0) =1, thus establishing a stable starting point for iterative optimization.

[0043] After entering the core iterative optimization loop, the algorithm alternately executes forward propagation and backward propagation to achieve optical field reconstruction. In the forward propagation phase, the algorithm simulates the physical process of a light beam penetrating a high-refractive-index medium, calculating the first... k During the nth iteration m The actual complex amplitude at each focal point V m (k) : in, For the current number kThe phase of the spatial light modulator in the next cycle. This complex exponential superposition forcibly embeds the "three-dimensional propagation bias" and "refractive index aberration compensation," transforming the optimized boundary conditions into the actual interior of the medium. From this, the actual light intensity at the current focal point can be calculated. .

[0044] Subsequently, the algorithm triggers an adaptive weight update mechanism, comparing the actual light intensity with the expected energy, and updates the weight factors using the following formula: in, Indicates the first In the nth iteration The weight corresponding to each focus; T m The target amplitude of the desired energy is preset for this focus; For the first The complex amplitude modulus of the light field at the focal point after each iteration reflects the relative intensity currently achieved. The adaptive weighted feedback coefficient has a range of values. This mechanism exponentially amplifies the weight of dark focal points and suppresses bright focal points, achieving precise redistribution of system energy.

[0045] The algorithm then performs backpropagation using the updated weighting factors. Treating each adjusted focus as a secondary virtual wave source, the complex amplitude field is backpropagated back to the spatial light modulator plane. The new phase for the next cycle is generated using the following comprehensive phase update formula: in, M The total number of target focuses; arg{ The function is used to extract this. M The complex field argument (principal value interval is [ ) after the inverse evolution of each focus and the superposition of the entire matrix) is [ ] π,π]).

[0046] This "forward detection-adaptive weighting-backward synthesis" cycle will continue indefinitely, with the uniformity of the current multifocal optical field evaluated at the end of each cycle. U : in I max and I min These represent the maximum and minimum values ​​of the actual light intensity at each target focus generated in a single iteration. U When the set convergence threshold is reached, the loop exits, and the converged 3D multifocal aberration-free raw phase is output. Φ MFSA ( x ,y ).

[0047] Finally, considering that physical hardware is highly susceptible to reflecting unmodulated zero-order central intense light and burning up materials, the algorithm superimposes a periodic linear blazed grating phase onto the optimized bare phase. Φ grating ( x , y This forces the reconstructed three-dimensional multifocal light field to deflect entirely to the +1st order diffraction direction in order to achieve physical avoidance.

[0048] Finally, the composite diffraction phase is folded and mapped to convert it into an 8-bit integer grayscale image that can be recognized by the computer terminal of the spatial light modulator. I CGH ( x , y ): The beam is directly input into a spatial light modulator. After passing through a focusing lens, the beam generated by the spatial light modulator creates a three-dimensional multifocal beam with a precise energy gradient (high energy in deeper layers, low energy in shallower layers) and eliminated spherical aberration within the target material. The core advantage of this invention lies in completely solving the pain points of "inability to cut" and "uneven cutting" in deep laser processing of high-absorption, high-refractive-index materials. It not only completely eliminates focal distortion and energy divergence during deep focusing by synchronously embedding three-dimensional spatial mapping and dynamic spherical aberration compensation at the algorithm's underlying layer, but more importantly, it significantly improves the computational convergence speed and eliminates numerical oscillations through optimized linear initial phase and adaptive gradient weighting mechanisms. This ensures extremely high accuracy in the energy gradient allocation of the multifocal array, perfectly offsetting the absorption and refraction losses accumulated along the laser's depth, ultimately achieving absolute consistency in the actual absorbed energy of each modified layer. Furthermore, the ingenious combination of blazed gratings and physical space filtering successfully removes the zero-order interference from the shallow layers of materials, which are prone to burning, greatly improving the quality of one-time uniform stealth laser cutting of hard and brittle semiconductor wafers such as silicon carbide.

[0049] Example: This embodiment demonstrates how the ultrafast laser three-dimensional multi-focus beam shaping method of the present invention can be used to achieve "one-time uniform stealth cutting of high-absorption, high-refractive-index single-crystal SiC wafers with energy gradient".

[0050] The laser processing hardware optical path system relied upon in this embodiment includes: an ultrafast infrared laser 1, a half-wave plate 2, a Glan Taylor prism 3, a beam expander 4, a pure phase spatial light modulator (SLM) 5, a 4F spatial filtering system composed of two plano-convex lenses (with a mechanical pinhole aperture on its Fourier spectrum plane), and a high numerical aperture focusing objective lens 11 arranged sequentially along the optical path. The optical path also includes several mirrors. The SiC wafer 12 to be processed is placed on a three-dimensional high-precision moving platform 13.

[0051] like Picture 2 The diagram shown is a schematic of the laser processing hardware optical path system upon which this embodiment relies.

[0052] This invention presents a one-time shaping method for the energy gradient of ultrafast laser multifoci based on the DO-3D-GSW algorithm, which is implemented according to the following steps: S1: Input physical hardware parameters.

[0053] Setting the center wavelength of ultrafast infrared laser 1 λ =1064 nm, beam waist radius of 2 mm, and Gaussian spot distribution; set the numerical aperture of the high numerical aperture focusing objective 11. NA =0.65, focal length f =4 mm, pupil radius is 2.5 mm; the pixel resolution of the pure phase spatial light modulator 5 (SLM) is 1272×1024, and the pixel size is 12.5 μm. A two-dimensional discrete coordinate system corresponding to the pixels of the pure phase spatial light modulator 5 (SLM) is established in the computer control system. x , y ).

[0054] S2: Set the target parameters for SiC wafer processing.

[0055] The target material to be processed is a 350 μm thick SiC wafer 12, with a refractive index at a wavelength of 1064 nm. n 2≈2.6 (air refractive index) n (1=1.0). To form a large-area, highly uniform parallel modified layer inside the SiC wafer, the goal was to generate a three-dimensional microarray containing eight focal points longitudinally distributed along the z-axis, with the processing depth of the first focal point located below the surface of the SiC wafer. d At a focal depth of 10 μm, the longitudinal distance between two adjacent focal points is set to 15 μm. Since the proportion of energy absorbed and refracted increases significantly when the laser strikes deeper into the SiC wafer, using the same initial incident energy at each focal point will result in the material at deeper depths being unable to be cut, while the material at shallower depths will be excessively ablated under the same processing pulse.

[0056] Therefore, the expected energy weights of these 8 foci are set to increase linearly with depth, constructing a multifoci with an increasing energy gradient distribution. Specifically, the expected energy weights from shallow to deep are set as follows: T 1 = 0.65, T 2 = 0.7, T 3 = 0.75, T 4 = 0.8, T 5 = 0.85, T 6 = 0.9 T 7 = 0.95, T 8 = 1.0. This energy gradient distribution is specifically designed to offset deep transport losses, ensuring that the energy actually deposited at each focal point within the material remains consistent.

[0057] S3: Algorithm initialization and SiC aberration compensation pre-calculation.

[0058] The computer first uses a formula to calculate the pure spherical aberration correction phase matrix used to compensate for the abrupt change in the refractive index of SiC. : At the same time, using the formula The three-dimensional free-space propagation function matrices corresponding to the eight focal points are pre-calculated. Finally, a two-dimensional linear phase distribution with a constant slope is generated on the two-dimensional grid of the spatial light modulator. = c 1 x + c 2 y As the initial phase of the iterative algorithm, random initial values ​​are discarded to prevent computational oscillations.

[0059] S4: Perform iterative optimization of the DO-3D-GSW algorithm for gradient weights.

[0060] Start the forward-backward propagation iterative loop: The current spatial light modulator phase, the aforementioned three-dimensional propagation functions at eight different depths, and the SiC pure spherical aberration correction phase are summed using complex matrices to calculate the actual complex amplitudes at the eight longitudinal focal points. V m (k) And actual light intensity.

[0061] Introducing feedback coefficients ω =0.5, using the formula: The weight factors for eight focal points at different depths are adaptively updated.

[0062] The weighted complex field is backpropagated to the spatial light modulator plane to extract the phase angle and generate a new phase. Φ (k)In this embodiment, after approximately 15 rapid iterations, the computer determines the uniformity of light intensity at eight different depth focal points. U When the value reaches 0.992 (greater than the set threshold of 0.98) and the actual light intensity ratio precisely matches the energy gradient setting of 0.65~1.0, the algorithm automatically exits the loop and outputs the converged multifocal bare phase function. Φ MFSA ( x , y ).

[0063] S5: Zero-level light processing and SiC wafer stealth dicing.

[0064] Computers Φ MFSA ( x , y Based on this, a linear blazed grating phase is superimposed. Φ grating The final target multifocal diffraction phase hologram is generated through modulus extraction and 256-level grayscale mapping. This hologram is then loaded onto the screen of the spatial light modulator in the hardware optical path. The ultrafast infrared laser beam emitted by the ultrafast infrared laser 1 is phase-modulated by the pure phase-type spatial light modulator 5 and divided into two parts: an unmodulated zero-order central direct beam and a +1st order diffracted beam carrying SiC spherical aberration compensation and multifocal array information. The two beams enter the 4F spatial filtering system, adjusting the position of the mechanical aperture stop located on the spectral plane to completely block the zero-order direct intense light in physical space, allowing only the +1st order diffracted light to pass through and enter the focusing objective with a numerical aperture of 0.65. Because the algorithm precisely controls the energy distribution at different depths according to the set target weights, the actual laser energy of these eight focal points exhibits a good linear increasing gradient distribution from shallow to deep (i.e., strictly corresponding to the energy ratio from 0.65 to 1.0). This energy gradient laser blade design, characterized by "high energy at deep layers and low energy at shallow layers," ingeniously compensates for material loss when the laser penetrates a thick silicon carbide wafer. For example... Picture 3 The figure shows the target eight-focal diffraction phase hologram generated by the DO-3D-GSW algorithm iteration, the simulation results of the position and intensity of the eight-focal light field on the xz section, and the simulation curves of the position and intensity of each focal point in the z direction (the figure shows that the energy of each focal point is distributed in a gradient along the z axis that increases with depth).

[0065] With the single translation of the three-dimensional high-precision moving platform, not only can extremely deep continuous thermally modified surfaces be cut out inside the thick single-crystal SiC wafer in one go, but also due to the precise compensation of the energy gradient, the modification effect of each layer from shallow to deep is highly consistent, strongly inducing the internal cracks to extend upward smoothly and straight, which greatly improves the cleaving yield and cutting edge quality of SiC wafers.

[0066] like Picture 4 The image shows a comparison of the results of single-crystal SiC wafer dicing using an eight-focus energy gradient method with those of traditional "equal energy and uncorrected aberration" dicing, focusing on the morphology of the dicing edge and the modified layer. Picture 4 It can be seen that when using the eight-focus energy gradient to cut single-crystal SiC wafers in one go, the energy of each focus is reasonably distributed according to the depth direction, and the spherical aberration in the focusing process inside the material is effectively compensated. The edge of the wafer is straighter and more continuous, the edge chipping and burrs are significantly reduced, and the modified layer is more uniform and has better continuity. In contrast, the traditional "equal energy and uncorrected aberration" processing method is more prone to uneven modified layer, local over-ablation or incomplete cutting due to the obvious deep energy attenuation and focal spot distortion, resulting in poor edge quality of the wafer.

[0067] It should be emphasized that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any way. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. A method for multi-focal energy gradient laser stencil cutting of hard and brittle materials, characterized in that, Includes the following steps; S1. Determine the parameters of the incident beam and the high numerical aperture focusing objective. S2. Determine the three-dimensional multifocal beam of the target hard and brittle material and the parameters of the target hard and brittle material; S3. Based on the parameters determined by S1 and S2, a two-dimensional linear phase distribution with a constant slope is set on the spatial light modulator plane as the initial phase for subsequent iterative calculations, thereby realizing the initialization phase and aberration compensation calculations. S4. Establish a forward-backward propagation iterative loop based on phase optimization to calculate the phase distribution of the multifocal hologram; S5. Superimpose a linear blazed grating phase onto the multifocal holographic phase distribution, and use MATLAB code to generate a composite diffraction phase hologram from the superimposed phase function and input it into the spatial light modulator. After the incident beam is modulated by the spatial light modulator, the zero-order diffracted light is filtered out by the post-positioned 4F spatial filter system. After being focused by the high numerical aperture focusing objective lens, a multi-focal energy gradient beam with high energy in the deep layer and low energy in the shallow layer and with the influence of spherical aberration is generated in one go inside the target hard and brittle material. This compensates for the absorption and refraction loss generated inside the target hard and brittle material with increasing depth, and achieves uniform stealth cutting with consistent actual absorption energy in each layer.

2. The method for multi-focus energy gradient laser stencil cutting of hard and brittle materials according to claim 1, characterized in that, In S1, the incident beam is a laser beam with a Gaussian spot energy spatial distribution; the incident beam is used to perform internal hidden cutting on the target hard and brittle material; The parameters of the incident beam include: the wavelength of the incident beam, the beam waist radius, and the spatial distribution of the beam energy. The parameters of the high numerical aperture focusing objective include numerical aperture (NA). The incident beam parameters are determined by the optical penetration characteristics of the target hard and brittle material and the requirements of the hidden cutting process; the focusing objective parameters are determined by the energy density requirements of the internal nonlinear absorption and the need for aberration calculation.

3. The method for multi-focus energy gradient laser stencil cutting of hard and brittle materials according to claim 2, characterized in that, In S2, the parameters of the three-dimensional multifocal beam include: the number of focal points, the three-dimensional spatial coordinates of each focal point inside the target hard and brittle material, the processing depth corresponding to each focal point, and the expected energy gradient weight of each target focal point set to accurately compensate for the absorption or refraction loss in the depth direction inside the target hard and brittle material, wherein the deeper the spatial depth of the focal point, the greater its expected energy weight; the parameters of the target hard and brittle material include: the refractive index of the target processed material; Among them, the three-dimensional multifocal parameters are determined by the physical dimensions of the target material and the expected continuity of the internal modified surface; the expected energy gradient weight is determined based on compensating for the absorption and refraction attenuation of the laser as it accumulates with depth inside the material. The rule set is that the deeper the spatial depth of the focal point, the greater its expected energy weight, and the target material parameter is the inherent physical property of the selected material at a specific processing wavelength.

4. The multi-focus energy gradient laser concealment method for hard and brittle materials according to claim 3, characterized in that, In step S3, based on the numerical aperture of the focusing objective, the processing depth of each focal point, and the refractive index of the processing material, the spherical aberration compensation phase, i.e., the pure spherical aberration correction phase, is calculated to compensate for the aberration caused by the sudden change in refractive index when the laser penetrates the target hard and brittle material. The expression is as follows: in, λ The incident wavelength, NA This is the numerical aperture of the focusing objective lens. ρ For normalized radial coordinates, n 2 represents the refractive index of the target material being processed. d The center depth of the target three-dimensional multifocal array; the coefficient s is determined by inner product calculation, reflecting the influence of the material refractive index on the effective focusing depth. This coefficient is obtained by numerical integration or by ray tracing and is approximately s ≈ n2. The pure spherical aberration correction phase is used to compensate for the spherical aberration caused by the discontinuity of refractive index and the increase in focusing depth after the laser enters the interior of the target hard and brittle material from the air. Spherical aberration refers to the aberration that light rays of different apertures cannot converge simultaneously to the same ideal focal point inside the target hard and brittle material, resulting in focal spot stretching, energy divergence, and a decrease in the quality of deep processing; The two-dimensional linear phase distribution corresponds to the three-dimensional multifocal beam parameters of the target hard and brittle material.

5. The method for multi-focus energy gradient laser stencil cutting of hard and brittle materials according to claim 4, characterized in that, In algorithm initialization, a two-dimensional linear phase distribution is used. = c 1 x + c 2 y It provides a stationary initial solution with a constant slope, i.e., the linear coefficients in the formula. c 1 and c 2. The constant slope is constructed based on the set number of focal points, coordinates and weight parameters to avoid numerical oscillations in subsequent iterations. It is independent of the phase of pure spherical aberration correction, but plays a joint role in the superposition of physical fields. Aberration compensation calculation refers to the calculation of the pure spherical aberration phase matrix in advance using geometric optical path difference to compensate for focal spot distortion caused by abrupt changes in high refractive index. And force it to be embedded into the forward and backward propagation models of each iteration to achieve aberration correction synchronously; The correspondence between the two-dimensional linear phase distribution and the three-dimensional multifocal beam parameters of the target hard and brittle material refers to the fact that, under the premise of knowing the number of focal points, coordinates and energy weights, a constant slope phase is preset on the coordinate system corresponding to each SLM pixel. This provides a stable physical starting point for the subsequent iterative calculation of stacking depth and aberration information, which can ensure that the energy gradient converges rapidly and without oscillation.

6. The multi-focus energy gradient laser concealment method for hard and brittle materials according to claim 5, characterized in that, The forward-backward propagation iterative loop process described in S4 includes a forward propagation process that calculates the two-dimensional linear initial phase, the pure spherical aberration corrected phase, and the free space propagation function corresponding to the three-dimensional spatial coordinates of each focus. The backward propagation process updates the weights based on the deviation between the actual light intensity of each focus and the expected energy gradient weights, thereby obtaining a multi-focus energy gradient phase distribution that meets the hidden cutting requirements of the target hard and brittle material.

7. The method for multi-focus energy gradient laser stencil cutting of hard and brittle materials according to claim 5, characterized in that, The iterative loop process in S4 specifically includes: S41: The initial phase of the spatial light modulator in the current cycle, the pure spherical aberration correction phase obtained in S3, and the free space propagation function corresponding to the three-dimensional spatial coordinates and processing depth of each focal point are superimposed to calculate the actual complex amplitude and actual light intensity at the focal position of each target hard and brittle material. in, For the current number k The phase of the spatial light modulator in the next cycle is used to determine the actual light intensity at the current focal point. ; S42: Compare the actual light intensity of each focal point with the expected energy gradient weight of each target focal point set in S2 to compensate for the deep absorption loss of the target hard and brittle material, calculate the deviation value, and use the feedback coefficient to adaptively update the weight factor corresponding to each focal point. in, Indicates the first In the nth iteration Weighting factors corresponding to each target focus; For the first m The expected energy weight set for each target focus; For the first The complex amplitude modulus of the light field at the focal point after the next iteration; For adaptive weighted feedback coefficients; S43: Multiply the updated weighting factor by the actual complex amplitude of each focus, and propagate it back to the spatial light modulator plane to superimpose and calculate the phase angle, thus obtaining the updated phase for the next cycle. in, M The total number of target focuses; arg{ The function is used to extract this. M The complex field argument after the inverse evolution of each focal point and the superposition of the entire matrix; S44: Calculate the uniformity of the actual light intensity of all currently generated focal points. If the uniformity reaches the set threshold or the maximum number of iterations, terminate the loop and output the current updated phase as the final multifocal diffraction phase distribution.

8. The method for multi-focus energy gradient laser stencil cutting of hard and brittle materials according to claim 7, characterized in that, In step S42, the adaptive update formula for the weighting factor is: in, Indicates the first In the nth iteration Weighting factors corresponding to each target focus; For the first m The expected energy weight set for each target focus; For the first After the nth iteration The complex amplitude modulus of the light field at the focal point of each target; For adaptive weighted feedback coefficients; In S44, the uniformity is defined as: in, I max and I min These represent the maximum and minimum values ​​of the actual light intensity at the focal point of each target hard and brittle material generated in a single iteration.

9. The multi-focus energy gradient laser concealment method for hard and brittle materials according to claim 8, characterized in that, In S5, the purpose of superimposing the phase of the linear blazed grating is to deflect the three-dimensional multifocal light field of the target hard and brittle material to the +1st order diffraction direction; by setting a mechanical pinhole aperture on the Fourier spectrum surface of the 4F spatial filter system, and in physical space making the pinhole aperture allow only the +1st order diffraction light to pass through while blocking the zeroth order diffraction light at the center of the optical axis, the potential thermal damage of the unmodulated beam to the target hard and brittle material is completely eliminated. The multi-focal energy gradient beam with high energy at deep depths and low energy at shallow depths refers to assigning higher incident energy weights to deeper focal points based on the differences in transmission loss at different depths within the target hard and brittle material, in order to compensate for absorption and refraction attenuation during laser propagation within the material; the elimination of spherical aberration refers to pre-compensating for focal spot distortion caused by deep focusing through pure spherical aberration phase correction, so that each focal point at a depth forms a modified region of uniform size and controllable energy within the target hard and brittle material, thereby achieving high-quality and uniform stealth cutting; The spatial light modulator is a pure phase spatial light modulator (5); The 4F spatial filtering system includes a first lens (6), a mechanical pinhole aperture (8), and a second lens (10) arranged sequentially along the optical path. It is used to perform spatial filtering and order selection on the beam modulated by the spatial light modulator, filter out the zero-order diffraction light and retain only the target order diffraction light to enter the focusing objective lens, so as to avoid the zero-order direct light causing local thermal damage to the target hard and brittle material. The mechanical pinhole aperture (8) is set on the Fourier spectrum plane of the two lenses.

10. A laser processing hardware optical path system for implementing the multi-focus energy gradient laser hidden cutting method for hard and brittle materials as described in any one of claims 1-9, characterized in that, It includes an ultrafast infrared laser (1), a half-wave plate (2), a Glan Taylor prism (3), a beam expander (4), and a pure phase spatial light modulator (5) arranged sequentially along the optical path; The ultrafast infrared laser beam emitted by the ultrafast infrared laser (1) is phase-modulated by the pure phase spatial light modulator (5), and after passing through the first lens (6) and the first reflector (7), it is divided into two parts: the unmodulated zero-order central direct light and the +1 order diffracted light carrying SiC spherical aberration compensation and multifocal array information. The two beams enter the 4F spatial filter system, and the position of the mechanical aperture stop (8) located on the spectral plane is adjusted to completely block the zero-order direct strong light from the physical space, allowing only the +1 order diffracted light to pass through and enter the high numerical aperture focusing objective (11). A second mirror (9) and a second lens (10) are set between the mechanical pinhole aperture (8) and the high numerical aperture focusing objective (11), and the target hard and brittle material is placed on a three-dimensional high-precision moving platform (13); The target hard and brittle material is the SiC wafer to be processed (12).

Images