Method for improving silicon wafer dicing topography

Abstract

This invention provides a method for improving the morphology of silicon wafer dicing, relating to the field of semiconductor material processing technology. The method includes: establishing a spatial distribution model of the wire bow along the axial direction of the guide wheel; determining a coating thickness distribution function along the axial direction based on this model, such that the coating thickness decreases from the middle region to both ends of the guide wheel; preparing the guide wheel according to the coating thickness distribution function; ensuring that the coating hardness and groove depth of the guide wheel exhibit a gradient distribution along the axial direction matching the coating thickness; installing the prepared guide wheel on a multi-wire dicing device, arranging dicing lines to form a wire mesh to perform dicing operations on the silicon wafer; during the dicing process, using marker layers pre-embedded at different radial depths within the coating to detect the cumulative wear of the coating in real time, determining the wear state of the guide wheel; and adjusting the dicing parameters based on the wear state of the guide wheel. This solution can improve the morphology quality of silicon wafer dicing by stably controlling the wire bow within a target range throughout the guide wheel's lifespan.

Description

Technical Field

[0001] This invention relates to the field of semiconductor material processing technology, and in particular to a method for improving the morphology of silicon wafer cutting. Background Technology

[0002] As the size of semiconductor silicon wafers continues to increase, from 4 inches to 12 inches, the quality requirements for wire cutting are becoming increasingly stringent. Currently, common silicon wafer defects can be categorized into five types: thickness variation (TTV), bow, wire saw mark, wire bow, and residual stress.

[0003] In multi-wire dicing, wire bowing is the core factor affecting the aforementioned defects. There are two main reasons for abnormal wire bowing. First, during the silicon ingot cutting, the wire mesh bends within the span between adjacent guide rollers, resulting in a non-uniform distribution of wire bowing along the guide roller axis—larger in the middle and smaller at both ends—directly causing uneven thickness in the width direction of the silicon wafer. Second, as the number of cuts increases, the guide roller coating gradually wears down, and the outer diameter decreases. With a fixed axial spacing, the equivalent span increases, causing the wire bowing to gradually rise, and the silicon wafer cutting quality continuously declines with the extended use of the guide rollers.

[0004] In existing technologies, the problem of wire bowing is addressed by either increasing the diameter of the guide rollers, adjusting the cutting tension, or replacing the guide rollers periodically. However, none of these methods fundamentally solve the problem of wire bowing, which in turn leads to poor morphological performance of the silicon wafers. Summary of the Invention

[0005] In view of this, and to address the above shortcomings, it is necessary to propose a method for improving the morphology of silicon wafer cutting, in order to improve the morphology quality of silicon wafer cutting by stably controlling the bow within the target range throughout the life cycle of the guide roller.

[0006] This invention provides a method for improving the morphology of silicon wafer dicing, comprising:

[0007] Establish a spatial distribution model of the bow along the axial direction of the guide wheel, and determine the coating thickness distribution function of the guide wheel coating thickness along the axial direction based on the spatial distribution model, so that the coating thickness decreases from the middle region of the guide wheel to both ends.

[0008] The guide wheel is prepared according to the coating thickness distribution function; wherein the coating hardness and guide wheel groove depth of the guide wheel are both distributed in a gradient along the axial direction that matches the coating thickness;

[0009] The prepared guide wheel is installed on the multi-wire cutting equipment, and the cutting lines are arranged to form a wire mesh to perform the cutting operation on the silicon wafer.

[0010] During the cutting process, the cumulative wear of the coating is sensed in real time by using marking layers embedded at different radial depths inside the coating, thereby determining the wear state of the guide wheel;

[0011] The cutting parameters are adjusted according to the wear condition of the guide wheel.

[0012] Preferably, the spatial distribution model is:

[0013]

[0014] in, The spatial distribution of the bow length. These are the position coordinates along the guide wheel axis. The effective working width of the guide wheel, This represents the maximum theoretical line bow at the center of the guide wheel.

[0015] Preferably, determining the coating thickness distribution function along the axial direction of the guide wheel coating thickness based on the spatial distribution model includes:

[0016] The coating thickness distribution function is determined based on the following formula:

[0017]

[0018]

[0019] in, The contour function is used for compensation, and K is the compensation coefficient. Let be the coating thickness distribution function. The reference coating thickness.

[0020] Preferably, the step of preparing the guide wheel according to the coating thickness distribution function includes:

[0021] According to the coating thickness distribution function A coated guide wheel is prepared using a mold forming method. The coating has three hardness levels: a high-hardness coating in the middle area of ​​the guide wheel, a low-hardness coating in the two end areas, and a medium-hardness coating in the transition area between the middle and end areas. The hardness is linked to the thickness, so that the areas with greater thickness have higher hardness and the areas with less thickness have lower hardness.

[0022] A guide groove is created using a diamond roller; wherein the depth of the guide groove along the axial direction is positively correlated with the coating thickness.

[0023] Preferably, during the coating preparation process, a marking layer is pre-embedded in the coating; the marking layer is at least one layer, the marking layer material has a color difference from the coating body material, and different marking layers are embedded at different radial depths, each corresponding to a different cumulative wear threshold.

[0024] Preferably, the relationship between hardness and thickness is as follows:

[0025]

[0026] in, The hardness is at the axial position x. and These are the hardness values ​​of the high-hardness coating and the low-hardness coating, respectively. This is the linkage coefficient. The minimum thickness of the regions at both ends of the guide wheel. This represents the maximum thickness of the middle region of the guide wheel.

[0027] Preferably, the marking layer consists of two carbon black marking layers, located at 1 / 3 and 2 / 3 of the total coating thickness, respectively.

[0028] Preferably, the step of determining the wear state of the guide wheel by real-time sensing of the cumulative wear of the coating through marker layers pre-embedded at different radial depths within the coating includes:

[0029] When the coating surface has a uniform color and no carbon black marker layer is detected, the current wear condition is determined to be mild wear.

[0030] When the first carbon black marker layer is detected to be exposed, the current wear condition is determined to be moderate wear.

[0031] When the second carbon black marker layer is detected to be exposed, the current wear condition is determined to be severe wear.

[0032] Preferably, adjusting the cutting parameters according to the wear state of the guide wheel includes: adjusting the cutting tension and / or feed speed according to the determined wear state, and measuring the actual bow value after parameter adjustment for closed-loop verification.

[0033] Preferably, the closed-loop verification of measuring the actual bow value after parameter adjustment includes: when the deviation between the measured bow value and the target value exceeds a preset threshold, automatically increasing or decreasing the preset tension amplitude for fine-tuning correction until the bow returns to the target range.

[0034] As can be seen from the above technical solution, the method for improving the morphology of silicon wafer cutting provided by the present invention first considers establishing a spatial distribution model of the wire bow along the axial direction of the guide wheel, and determining the coating thickness distribution function of the guide wheel coating thickness along the axial direction based on the spatial distribution model, so that the coating thickness decreases from the middle region of the guide wheel to both ends. Then, the guide wheel is prepared according to the coating thickness distribution function, and the coating hardness and guide wheel groove of the guide wheel are both distributed in a gradient along the axial direction that matches the coating thickness. Further, the prepared guide wheel is installed on a multi-wire cutting machine, and cutting lines are arranged to form a wire mesh to perform the cutting operation on the silicon wafer. At the same time, during the cutting process, the cumulative wear of the coating is sensed in real time by a marking layer pre-embedded at different radial depths inside the coating, the wear state of the guide wheel is determined, and the cutting parameters are adjusted according to the wear state. It can be seen that this solution, by considering the design of a coating on the surface of the guide wheel, can shorten the equivalent span and thus reduce the overall wire bow, and this method does not require modification of the main structure of the wire cutting machine. Furthermore, considering that in multi-wire dicing, the wire bows exhibit a spatial distribution that is larger in the middle and smaller at both ends, if a traditional guide with a uniform outer diameter is used, the dicing wire in the middle region will experience the greatest bending deformation, resulting in a thinner silicon wafer in the middle position and an increased TTV (Total TV). Therefore, this application constructs a spatial distribution of the wire bows and then designs a non-uniform coating thickness and hardness based on this spatial distribution. This effectively flattens the non-uniform wire bows from the source, making them more consistent along the width direction of the silicon wafer, thereby improving the wafer dicing morphology. Furthermore, this solution also embeds a marking layer in the coating to achieve online direct detection of the guide wheel wear condition. This allows for accurate identification of the current wear state without stopping the machine, avoiding reduced production efficiency due to downtime for wear measurement. Attached Figure Description

[0035] Figure 1 This is a flowchart of a method for improving the morphology of silicon wafer cutting, provided as an embodiment of the present invention. Detailed Implementation

[0036] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] like Figure 1 As shown, the present invention provides a method for improving the morphology of silicon wafer cutting, the method comprising the following steps:

[0038] Step 101: Establish a spatial distribution model of the bow along the axial direction of the guide wheel, and determine the coating thickness distribution function of the guide wheel coating thickness along the axial direction based on the spatial distribution model, so that the coating thickness decreases from the middle region of the guide wheel to both ends.

[0039] Step 102: Prepare the guide wheel according to the coating thickness distribution function; wherein the coating hardness and guide wheel groove depth of the guide wheel are both distributed in a gradient along the axial direction that matches the coating thickness;

[0040] Step 103: Install the prepared guide wheel onto the multi-wire cutting equipment, arrange the cutting lines to form a wire mesh, and perform the cutting operation on the silicon wafer;

[0041] Step 104: During the cutting process, the cumulative wear of the coating is sensed in real time by using the marking layers embedded in the coating at different radial depths, and the wear state of the guide wheel is determined.

[0042] Step 105: Adjust the cutting parameters according to the wear condition of the guide wheel.

[0043] In this embodiment, by designing a coating on the guide wheel surface, the equivalent span can be shortened, thereby reducing the overall bow size. Moreover, this method does not require modification to the main structure of the wire EDM machine. Furthermore, considering that in multi-wire EDM, the bow size is spatially distributed with a larger center and smaller ends, if a traditional guide wheel with a uniform outer diameter is used, the cutting wire in the middle region will experience the greatest bending deformation, resulting in a thinner wafer thickness in the middle and an increased TTV (Total Transmission Value). Therefore, this application constructs a spatial distribution of the bow size and then designs a non-uniform coating thickness and hardness based on this distribution, flattening the non-uniform bow size from the source and making the bow size more consistent along the wafer width direction, thereby improving the wafer cutting morphology. Furthermore, this solution also embeds a marking layer in the coating to achieve online direct detection of the guide wheel wear condition. This allows for accurate knowledge of the current wear condition without stopping the machine, avoiding reduced production efficiency due to downtime for wear measurement.

[0044] The following provides a more detailed explanation of each of the above steps.

[0045] For step 101, a spatial distribution model of the bow along the axial direction of the guide wheel is established, and the coating thickness distribution function along the axial direction of the guide wheel coating thickness is determined based on the spatial distribution model, so that the coating thickness decreases from the middle region of the guide wheel to both ends.

[0046] In multi-wire cutting, the wire bows exhibit an inherent non-uniform distribution along the guide wheel axis, being larger in the middle and smaller at both ends. To eliminate this non-uniformity at its source, it is first necessary to understand the specific distribution pattern of the wire bows. This step quantifies the spatial distribution of the wire bows by establishing a parabolic distribution model, thereby deriving the required outer diameter compensation at each position of the guide wheel. More compensation is applied where the wire bow is large, and less compensation is applied where the wire bow is small, providing a design basis for subsequent guide wheel fabrication. Specifically, this step first obtains the parameters of the multi-wire cutting equipment, such as the distance between adjacent guide wheel shafts L, the effective working width of the guide wheel W, the outer diameter of the metal core D1, the initial set tension F0, and the minimum safe distance d between the guide wheel surface and the crystal rod surface. min Then, considering that the distribution of the pantograph along the guide wheel axis follows a parabolic law, the spatial distribution model of the pantograph is constructed with the center of the guide wheel as the origin as follows:

[0047]

[0048] in, The spatial distribution of the bow length. These are the position coordinates along the guide wheel axis. The effective working width of the guide wheel, The maximum theoretical line camber at the center of the guide wheel can be determined through trial and error measurement.

[0049] Furthermore, considering compensation for the aforementioned nonlinear arc distribution, making the outer diameter increment of the middle region of the guide wheel greater than that at both ends, the axial distribution function of the coating thickness is then constructed as follows:

[0050]

[0051]

[0052] in, The contour function is used for compensation, and K is the compensation coefficient. Let be the coating thickness distribution function. The reference coating thickness is determined by safety clearance conditions. For example... Where h is the guide wheel groove depth, d min The minimum safe distance between the guide wheel surface and the crystal rod surface can be 5mm.

[0053] In this embodiment, the cutting line is considered as a slender beam with hinged ends within the span of adjacent guide wheels. It bends under the pressure load of the crystal ingot, with the largest deflection at the center of the span, decreasing towards both ends. This application establishes a spatial distribution model of the bow wires and accurately calculates the reverse compensation profile of the outer diameter of the guide wheel coating, making the middle region of the guide wheel thicker and providing additional support for the cutting line in this region. This flattens the non-uniform bow wires at the source, thereby making the bow wires more uniform in the width direction of the silicon wafer and improving the cutting morphology of the silicon wafer.

[0054] For step 102, a guide wheel is prepared according to the coating thickness distribution function; wherein the coating hardness and guide wheel groove depth of the guide wheel are both distributed in a gradient along the axial direction that matches the coating thickness.

[0055] Considering the two practical problems that guide rollers face in actual cutting: first, the middle area experiences the greatest stress and wears the fastest, and the compensation profile degrades over time; second, the middle area has the largest line swing amplitude, making it prone to line mark defects. Therefore, this step considers simultaneously introducing coating depth gradients and hardness gradients to synchronize the wear rates of each area, and introducing groove depth gradients to enhance the linear constraint force in the middle area, thereby improving the coating's performance and lifespan.

[0056] In this step, we consider preparing a polyurethane coating on the outer periphery of the metal shaft, while ensuring that the coating has gradient characteristics in terms of coating thickness and coating hardness.

[0057] When preparing the axial gradient of the coating thickness, consider the coating thickness distribution function. The coated guide wheel is prepared using a mold forming method. Specifically, liquid polyurethane prepolymer is mixed with a curing agent, vacuum degassed, and then injected into a mold whose inner cavity contour is precisely machined according to a T(x) function. The mold is cured at 120℃ for 8 hours to bring the outer contour of the coated blank close to the target contour (leaving a 1mm fine grinding allowance). After curing, it is finely ground using a CNC cylindrical grinder according to a T(x) function, achieving a surface roughness Ra≤0.8μm and an outer diameter tolerance of ±0.05mm. Finally, a diamond roller is used for grooving, with a groove depth tolerance of ±0.01mm.

[0058] To create an axial hardness gradient for the coating, three hardness levels are designed: a high-hardness coating in the middle region of the guide wheel, a low-hardness coating at both ends, and a medium-hardness coating in the transition region between the middle and end regions. Hardness is linked to thickness, ensuring higher hardness at thicker areas and lower hardness at thinner areas. Specifically, a multi-component zone casting method can be used to prepare three polyurethane prepolymers with different hardness segments. The high-hardness coating is used in the middle region (target hardness 98HA), the medium-hardness coating in the transition region, and the low-hardness coating at both ends (target hardness 88HA). Through the axial zone pouring head of the multi-component casting equipment, the corresponding components are simultaneously injected at different axial positions of the mold. The components diffuse into each other at the interfaces, forming a continuous hardness gradient, and are cured in a single step.

[0059] The relationship between hardness and thickness can be described as follows:

[0060]

[0061] in, The hardness is at the axial position x. and These are the hardness values ​​of the high-hardness coating and the low-hardness coating, respectively. This is the linkage coefficient. The minimum thickness of the regions at both ends of the guide wheel. This represents the maximum thickness in the middle region of the guide wheel. This ensures that areas with greater thickness have higher hardness, while areas with less thickness have lower hardness.

[0062] In one embodiment, an axial gradient in the guide wheel groove depth can also be considered. During grooving, the feed depth of the diamond roller can be controlled by a CNC program, ensuring that the groove depth along the axial direction is positively correlated with the coating thickness. In this way, the thickness gradient can provide spatial pre-compensation, the hardness gradient can achieve wear matching, and the groove depth gradient can enhance the lateral constraint force of the cutting line in the middle region. These three gradients work together to achieve optimal configuration of the guide wheel's spatial dimensional performance, making the wear rate of each axial position of the guide wheel tend to be consistent, avoiding the uneven wear of traditional guide wheels where pits are first worn into the middle.

[0063] In one embodiment, a wear marking layer can also be embedded in the coating. Specifically, during the coating preparation process, two annular marking layers can be embedded at radial depths of 1 / 3 and 2 / 3 of the total coating thickness, respectively. The marking layer material is polyurethane with 1.0 wt% carbon black added, providing a color contrast with the light-colored base polyurethane, and has a thickness of 0.8 mm. The embedding method involves pausing the casting process when the predetermined thickness is reached, applying the marking layer material to the circumferential surface of the coating, and continuing to cast the upper layer material after it has partially cured.

[0064] For step 103, the prepared guide wheel is installed on the multi-wire cutting equipment, the cutting lines are arranged to form a wire mesh, and the silicon wafer is cut.

[0065] In this step, the prepared guide wheel is installed on a multi-wire cutting machine, the cutting parameters of the multi-wire cutting machine are set, and the silicon wafer is cut.

[0066] For step 104, during the cutting process, the cumulative wear of the coating is sensed in real time by using the marking layers embedded in the coating at different radial depths, and the wear state of the guide wheel is determined.

[0067] Considering that with increasing cutting cycles, the guide wheel coating wears continuously, the outer diameter gradually decreases, and the equivalent span increases, the bow will slowly recover, meaning the initial spatial compensation effect is eroded over time. If this change is not detected, subsequent targeted intervention is impossible. Therefore, this step considers using a pre-embedded marking layer to determine the degree of wear, thereby providing trigger signals and input parameters for dynamic compensation. Specifically, step 104 can be implemented as follows:

[0068] When the coating surface has a uniform color and no carbon black marker layer is detected, the current wear condition is determined to be mild wear.

[0069] When the first carbon black marker layer is detected to be exposed, the current wear condition is determined to be moderate wear.

[0070] When the second carbon black marker layer is detected to be exposed, the current wear condition is determined to be severe wear.

[0071] In this embodiment, a CCD image sensor can be installed near the guide wheel to acquire images of the guide wheel surface in real time and analyze the surface color characteristics. When the coating surface color is uniform and no carbon black marker layer is detected, it is determined to be light wear, with the cumulative wear amount less than the embedding depth of the first marker layer. This corresponds to the early stage of the guide wheel's lifespan, approximately the first 1 / 3 of its lifespan. When the first marker layer is exposed, such as when ring-shaped black stripes appear on the surface, it is determined to be moderate wear, with the cumulative wear amount between the embedding depth of the first and second marker layers. This corresponds to the middle stage of the guide wheel's lifespan, approximately the middle 1 / 3 of its lifespan. When the second marker layer is exposed, it is determined to be heavy wear, with the cumulative wear amount between the embedding depth of the second marker layer and the maximum allowable wear amount. This corresponds to the late stage of the guide wheel's lifespan, approximately the last 1 / 3 of its lifespan. When both marker layers are completely exposed, it is determined to be at the end of its lifespan, at which point the guide wheel needs to be replaced. In this way, by setting marker layers, the wear status judgment can be improved from the traditional indirect estimation to direct detection, without the need for machine downtime for real-time online sensing, and without affecting production efficiency. Moreover, the cost of preparing the marking layer is extremely low, and it does not significantly increase the cost of the guide wheel.

[0072] For step 105, the cutting parameters are adjusted according to the wear condition of the guide wheel.

[0073] In this step, after understanding the wear condition, corresponding measures are needed to counteract the bow rise. Increasing the cutting wire tension can straighten the mesh and compensate for the bow, but increased tension will increase the risk of wire breakage. Therefore, it is recommended to simultaneously reduce the feed rate to reduce the cutting load.

[0074] For example, in one embodiment, for mild wear, an initial tension is used. and initial feed rate No compensation is provided. For moderate wear, tension compensation can be... Feed rate compensation can be For severe wear, tension compensation can provide... Feed rate compensation can be , and Both are coefficients, representing tension per unit length and time, respectively. and These represent the embedment depths of the first and second marker layers, respectively. This is the maximum permissible wear level. In practice, a state-switching approach can be used. For example, when the wear state is detected to switch from light wear to moderate wear, after the current cutting batch is completed and before the next batch begins, the tension and speed are switched to the tension compensation and feed speed compensation corresponding to moderate wear. The same operation is performed when switching from moderate wear to heavy wear to avoid disturbances caused by adjustments during the cutting process. In this way, process parameters are proactively adjusted based on the sensed wear state to compensate for the bow recovery caused by guide roller wear. Throughout the entire lifecycle of the guide roller, from its initial use to its scrapping, the bow is stably maintained at a constant target level, ensuring that each batch of silicon wafers has a consistent and excellent morphology, eliminating quality drift from good in the early stages to poor in the later stages.

[0075] The modules or units in the device of this invention can be merged, divided, and deleted according to actual needs. The above-disclosed embodiments are merely preferred embodiments of the present invention and should not be construed as limiting the scope of the invention. Those skilled in the art will understand that implementing all or part of the processes of the above embodiments and making equivalent changes according to the claims of this invention still fall within the scope of the invention.

Claims

1. A method for improving the morphology of silicon wafer cutting, characterized in that, include: Establish a spatial distribution model of the bow along the axial direction of the guide wheel, and determine the coating thickness distribution function of the guide wheel coating thickness along the axial direction based on the spatial distribution model, so that the coating thickness decreases from the middle region of the guide wheel to both ends. The guide wheel is prepared according to the coating thickness distribution function; wherein the coating hardness and guide wheel groove depth of the guide wheel are both distributed in a gradient along the axial direction that matches the coating thickness; The prepared guide wheel is installed on the multi-wire cutting equipment, and the cutting lines are arranged to form a wire mesh to perform the cutting operation on the silicon wafer. During the cutting process, the cumulative wear of the coating is sensed in real time by using marking layers embedded at different radial depths inside the coating, thereby determining the wear state of the guide wheel; The cutting parameters are adjusted according to the wear condition of the guide wheel.

2. The method for improving the morphology of silicon wafer cutting according to claim 1, characterized in that, The spatial distribution model is as follows: ； in, The spatial distribution of the bow length. These are the position coordinates along the guide wheel axis. The effective working width of the guide wheel, This represents the maximum theoretical line bow at the center of the guide wheel.

3. The method for improving the morphology of silicon wafer cutting according to claim 2, characterized in that, The step of determining the coating thickness distribution function along the axial direction of the guide wheel coating thickness based on the spatial distribution model includes: The coating thickness distribution function is determined based on the following formula: ； ； in, The contour function is used for compensation, and K is the compensation coefficient. Let be the coating thickness distribution function. The reference coating thickness.

4. The method for improving the morphology of silicon wafer cutting according to claim 3, characterized in that, The process of preparing the guide wheel according to the coating thickness distribution function includes: According to the coating thickness distribution function A coated guide wheel is prepared using a mold forming method. The coating has three hardness levels: a high-hardness coating in the middle area of ​​the guide wheel, a low-hardness coating in the two end areas, and a medium-hardness coating in the transition area between the middle and end areas. The hardness is linked to the thickness, so that the areas with greater thickness have higher hardness and the areas with less thickness have lower hardness. A guide groove is created using a diamond roller; wherein the depth of the guide groove along the axial direction is positively correlated with the coating thickness.

5. The method for improving the morphology of silicon wafer cutting according to claim 4, characterized in that, During the coating preparation process, a marking layer is pre-embedded in the coating; the marking layer is at least one layer, and there is a color difference between the marking layer material and the coating body material. Different marking layers are buried at different radial depths, each corresponding to a different cumulative wear threshold.

6. The method for improving the morphology of silicon wafer cutting according to claim 5, characterized in that, The relationship between hardness and thickness is as follows: ； in, The hardness is at the axial position x. and These are the hardness values ​​of the high-hardness coating and the low-hardness coating, respectively. This is the linkage coefficient. The minimum thickness of the regions at both ends of the guide wheel. This represents the maximum thickness of the middle region of the guide wheel.

7. The method for improving the morphology of silicon wafer cutting according to claim 5, characterized in that, The marking layer consists of two carbon black marking layers, located at 1 / 3 and 2 / 3 of the total coating thickness, respectively.

8. The method for improving the morphology of silicon wafer cutting according to claim 7, characterized in that, The method of determining the wear state of the guide wheel by real-time sensing of the cumulative wear of the coating through marker layers pre-embedded at different radial depths inside the coating includes: When the coating surface has a uniform color and no carbon black marker layer is detected, the current wear condition is determined to be mild wear. When the first carbon black marker layer is detected to be exposed, the current wear condition is determined to be moderate wear. When the second carbon black marker layer is detected to be exposed, the current wear condition is determined to be severe wear.

9. The method for improving the morphology of silicon wafer cutting according to claim 1, characterized in that, The step of adjusting the cutting parameters according to the wear state of the guide wheel includes: adjusting the cutting tension and / or feed speed according to the determined wear state, and measuring the actual bow value after parameter adjustment for closed-loop verification.

10. The method for improving the morphology of silicon wafer dicing according to claim 9, characterized in that, The closed-loop verification of measuring the actual bow value after parameter adjustment includes: when the deviation between the measured bow value and the target value exceeds a preset threshold, automatically increasing or decreasing the preset tension amplitude for fine-tuning correction until the bow returns to the target range.

Images