12-inch wafer-oriented DFG 8560 double-axis system coarse-fine cooperative grinding surface repairing method

Abstract

The invention belongs to the technical field of semiconductor manufacturing equipment and processes, and particularly relates to a DFG 8560 double-axis system rough-fine cooperative grinding surface repairing method for a 12-inch wafer. The method aims at solving the problems that traditional single-shaft grinding is low in efficiency, the subsurface damage is deep, and errors are caused by multiple times of clamping. The method comprises the steps that rough grinding is conducted on a first main shaft of a DFG 8560 double-shaft system through resin and a diamond grinding wheel, materials are efficiently removed, and the depth of a damaged layer is controlled; then, a ceramic combined diamond grinding wheel is used on a second main shaft for accurate grinding, and a damaged layer is accurately stripped; and finally, elastic deformation is eliminated through polishing treatment, and a nanoscale flat surface is obtained. The whole process is completed through one-time clamping, parameters are independently regulated and controlled through double shafts, and high efficiency, low damage and high uniformity are achieved. According to the method, the machining efficiency is remarkably improved, the surface damage depth and the comprehensive cost are reduced, and high process adaptability and compatibility are achieved.

Description

Technical Field

[0001] This application belongs to the field of semiconductor manufacturing equipment and process technology, specifically relating to a surface repair method for coarse-fine co-polishing grinding using the DFG 8560 biaxial system for 12-inch wafers. Background Technology

[0002] In the manufacturing of 12-inch reclaimed wafers, surface repair is a crucial step that determines the wafer's reuse value and device yield. Current mainstream processes face two major challenges: First, while high-mesh grinding wheels (e.g., 6000 grit or higher) can achieve extremely low surface damage and high consistency, their material removal rate is low, making it difficult to effectively handle severe defects such as deep scratches or complex film residues. Second, while single-axis precision grinding offers high accuracy, its reliance on high-mesh grinding wheels throughout the entire process leads to long processing cycles, low throughput, rapid wheel wear, and poor adaptability to wafers with severe initial damage, making it difficult to meet the demands of large-scale, high-efficiency production. Therefore, a synergistic grinding strategy that balances high efficiency and high surface quality is urgently needed.

[0003] A search revealed a wafer polishing method with publication number CN103394994B, published on December 15, 2017. This patent employs a single polishing pad staged polishing strategy, using a combination of high, medium, and low pressures and online monitoring to control the endpoint, aiming to reduce residue and corrosion. However, this approach is still limited to single-axis operation, lacking a roughing pretreatment step, and cannot effectively handle high-removal-volume scenarios such as deep scratches or thick film residue; it relies entirely on chemical mechanical polishing (CMP), limiting its ability to repair severe surface defects, and it does not optimize stress distribution and material removal uniformity for 12-inch large-size wafers in a dual-axis system, making it difficult to achieve efficient coarse-fine polishing coordination.

[0004] A search revealed a method for repairing a matte damage layer on a 12-inch wafer, publication number CN120565401A, published on August 29, 2025 (substantive examination stage). This patent proposes using a multi-level energy-controlled surface repair device to remove the matte damage layer by precisely controlling the energy density of the flow stream, reducing the roughness from 50–60 nm to below 5 nm. However, this method relies on non-contact energy jetting (such as plasma or particle beams), belonging to a physical / chemical composite repair approach, rather than a mechanical grinding path. It does not address the structural adaptation to the DFG 8560 biaxial grinding platform, nor does it construct a synergistic process logic of "rapid rough grinding for damage removal + high-quality fine grinding for finishing," failing to resolve the fundamental contradiction between high removal rate and low surface damage. Furthermore, the equipment's versatility and production line integration are questionable.

[0005] The aforementioned problems indicate that existing technologies are either limited to inefficient single-axis fine grinding or employ non-grinding repair methods, lacking a co-grinding mechanism for 12-inch wafers adapted to the DFG 8560 dual-axis system. Therefore, this invention provides a surface repair method for 12-inch wafers using a co-grinding approach with the DFG 8560 dual-axis system. Through dual-axis division of labor (one axis for high removal through coarse grinding, and the other for low damage through fine grinding), dynamic matching of process parameters, and thermo-mechanical coupling control, it achieves efficient and high-quality repair of complex defects such as deep scratches and residual film layers. This significantly improves the yield and surface quality of reclaimed wafers, meeting the stringent requirements of advanced processes for reclaimed wafers. Summary of the Invention

[0006] The purpose of this invention is to provide a surface repair method for coarse-fine co-grinding using the DFG 8560 biaxial system for wafers, which can effectively solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] 1. A surface repair method for 12-inch wafers using a DFG 8560 dual-axis system with coarse-fine co-grinding, characterized by the following steps: Step 1: Performing coarse grinding. A resin-bonded diamond grinding wheel is installed on the first spindle of the DFG 8560 dual-axis grinding system to efficiently remove material from the clamped and positioned wafer surface. The amount removed in a single pass is controlled to be sufficient to cover the typical depth of severe scratches and complex film residues. The first spindle speed is 1500-2000 rpm, the table speed is 1800-2200 rpm, and the amount removed in a single pass is controlled to be 8-12 μm.

[0009] Step 2 involves fine grinding. A ceramic-bonded diamond grinding wheel is installed on the second spindle of the DFG 8560 biaxial grinding system to precisely finish the surface of the wafer after rough grinding at the micron level. The amount of material removed in a single fine grinding cycle is controlled to be 3-5 μm. The spindle speed of the second spindle is set to 2000-2500 rpm, and the table speed is set to 2200-2400 rpm. By increasing the effective grinding path density and stabilizing the pressure distribution, the microcracks and plastic deformation areas generated in the rough grinding stage are peeled off layer by layer until the complete and undamaged crystalline material body is exposed.

[0010] Step 3 involves performing a polishing process. At the end of the polishing cycle, the rotation speed of the second spindle and the table is maintained while the feed motion is stopped. Friction is performed for 3 to 5 minutes in a zero-feed state, and the interface heating rate is controlled at ≤0.5℃ / min during the polishing process, ultimately achieving a surface roughness Ra≤0.8nm.

[0011] Preferably, the average abrasive grain size of the resin-bonded diamond grinding wheel used in step 1 is a preset grain size, and its binder has a moderate elastic modulus and thermal stability, which can maintain the self-sharpening property of the abrasive grains under moderate load and avoid wafer fragmentation caused by excessive passivation leading to a sudden increase in cutting force.

[0012] Preferably, in step 1, the amount of material removed is controlled within a preset range, which is sufficient to cover the typical depth of scratches and complex film residues, while avoiding the introduction of deep damage beyond the preset depth due to excessive removal, thus providing a controllable and uniform initial surface state for subsequent fine grinding processes.

[0013] Preferably, the ceramic-bonded diamond grinding wheel used in step 2 has an average abrasive grain size of a preset size. Its binder structure is dense and has high hardness, which can maintain the geometric stability of the extremely fine abrasive grains in long-term micro-cutting and ensure consistent surface roughness.

[0014] Preferably, the amount of material removed by fine grinding in step 2 is set within a preset range. This value precisely matches the depth of the subsurface damage layer introduced in the coarse grinding stage, ensuring that the damaged area is completely removed without causing additional material waste, while reserving the minimum necessary removal allowance for subsequent chemical mechanical polishing.

[0015] Preferably, the polishing time in step 3 is set to a predetermined time period, which is sufficient to allow the wafer and grinding wheel interface to reach a state of thermal-mechanical equilibrium, effectively release the accumulated elastic strain energy, significantly reduce surface waviness and local protrusions, and keep the final surface roughness Ra value stably controlled below the preset threshold.

[0016] Preferably, the rough grinding and fine grinding processes are completed continuously in a single clamping operation in the DFG 8560 dual-axis system without the need for unloading or realigning midway, thereby eliminating positioning errors introduced by multiple clamping operations, ensuring that the rough grinding reference surface and the fine grinding surface are strictly coplanar, and improving the overall thickness uniformity.

[0017] Preferably, the DFG 8560 dual-axis system independently controls the spindle speed, table speed, and feed rate during the rough grinding and fine grinding stages, achieving dynamic decoupling and precise matching of process parameters, and avoiding performance compromises between high removal and low damage targets for a single axis system.

[0018] Preferably, the method is applicable to processing heavily scratched defects with a depth greater than or equal to a preset depth threshold and a length greater than or equal to a predetermined length, or complex residual wafers with a film thickness greater than or equal to a preset thickness threshold and a coverage area greater than or equal to a predetermined proportion. It has strong process adaptability and does not require significant adjustments to core parameters for different defect types.

[0019] Preferably, the method maintains stable performance for both P-type and N-type doped wafers during continuous processing, with processing efficiency fluctuations not exceeding a preset percentage threshold and surface quality index deviations not exceeding a preset size threshold, indicating that it has good process compatibility for wafers with different electrical properties.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] 1. Processing efficiency is significantly improved

[0022] By integrating rough grinding and fine grinding processes into the DFG 8560 dual-axis system, the entire process is automated in a single setup, eliminating wafer handling, secondary alignment, and waiting time in the traditional multi-device flow. The total processing time per wafer is reduced from 22 minutes in the traditional process to 7.2-7.8 minutes, improving efficiency by more than 60%; the grinding wheel replacement cycle is extended by 30%.

[0023] 2. Significantly Reduced Surface Damage Depth: An innovative "mild coarse grinding" strategy is employed. By using a coarse grinding wheel with optimized parameters, the depth of the subsurface damage layer is compressed from the predetermined depth range of traditional coarse grinding to a much smaller predetermined depth range, with a reduction exceeding a preset percentage. Combined with the precise removal of the damage layer by fine grinding, the final subsurface damage depth is reduced by more than a predetermined percentage compared to existing technologies, directly achieving nanoscale surface integrity. This saves a predetermined amount of material removal in subsequent polishing processes, shortens polishing time, and reduces consumable costs.

[0024] 3. Effective optimization of overall manufacturing costs: The clear division of labor between coarse and fine grinding wheels ensures that the coarse grinding wheel wears evenly during high-removal tasks, while the fine grinding wheel only performs minor repairs on the already smoothed surface, resulting in extremely low wear. Together, they extend the overall wheel replacement cycle. Combined with improved processing efficiency, a breakage rate below a preset threshold, and reduced rework rate, the overall processing cost per wafer can be reduced by a predetermined percentage, with particularly significant economic benefits in the mass production of high-value reclaimed wafers.

[0025] 4. Enhanced Process Robustness and Adaptability: Validated on multiple wafers with different types of defects, the re-scratch removal rate remained stable at the predetermined level, the standard deviation of the subsurface damage layer depth did not exceed the preset threshold, and there were no batch-wide quality issues. This method is not only suitable for conventional surface repair but can also efficiently handle complex residues such as metal films and nitride films, as well as deep scratches, expanding the process coverage capabilities of the DFG8560 equipment and meeting the stringent requirements of advanced processes for the surface integrity of regenerated wafers. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the overall technical solution architecture of the DFG 8560 biaxial system coarse-fine co-polishing surface repair method for 12-inch wafers proposed in this invention.

[0027] Figure 2 This is a schematic diagram illustrating the core principle framework of the coarse-fine co-grinding process in this invention;

[0028] Figure 3 This is a logical flow diagram of the coarse grinding and fine grinding stages in this invention;

[0029] Figure 4 This is a schematic diagram of the multi-level interaction relationship and data flow of the DFG 8560 dual-axis system in this invention, which achieves rough-finish collaborative machining in a single clamping operation. Detailed Implementation

[0030] Example 1

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0032] In the above-mentioned surface repair method for coarse-fine co-grinding of the DFG 8560 biaxial system for wafers, step 1 performs coarse grinding: a resin-bonded diamond grinding wheel is installed on the first spindle of the DFG 8560 biaxial grinding system to efficiently remove material from the clamped and positioned wafer surface. The amount of material removed in a single pass is controlled within a preset range. The spindle speed is set to a first preset speed, and the table speed is set to a second preset speed. By optimizing the matching of feed rate and speed, while ensuring a high material removal rate, the depth of the introduced subsurface damage layer is limited to a predetermined depth range, and the thickness uniformity deviation within the wafer surface is ensured not to exceed a preset threshold. Specifically, the resin-bonded diamond grinding wheel used in step 1 has an average abrasive grain size of 10 μm, corresponding to a 1500 mesh specification. Its binder is composed of a phenolic resin matrix with an elastic modulus between 3.5 and 4.2 GPa and a thermal decomposition temperature of not less than 280°C. This physical property allows the grinding wheel to maintain the self-sharpening property of the abrasive grains under moderate loads (typical contact pressure of 0.15–0.25 MPa). That is, after the abrasive grains become passivated, the binder softens locally and falls off in time, exposing new sharp cutting edges, thereby avoiding wafer fragmentation caused by a sudden increase in cutting force due to excessive passivation of the abrasive grains. In step 1, the removal amount is controlled to be 8–12 μm. This range is sufficient to cover the typical depth distribution of severe scratches (defined as scratches with a depth ≥3 μm and a length ≥50 μm) and complex film residues (defined as residual film with a thickness ≥1 μm and a coverage area ≥20%), while avoiding the introduction of deep damage exceeding 5 μm due to excessive removal, thus providing a controllable and uniform initial surface state for subsequent fine grinding processes.

[0033] In step 1, the control system of the DFG 8560 biaxial polishing system first retrieves the rough polishing parameter set for the current wafer type (such as P-type or N-type doped silicon wafer) from the process database. This parameter set includes spindle speed, table speed, feed rate, and target removal rate. The spindle speed of the first spindle is set to 1800 rpm, and the table speed is set to 2000 rpm, forming a speed ratio of 1:1.11. This ratio has been experimentally verified to effectively suppress periodic ripples caused by relative motion mismatch. The feed rate is dynamically adjusted by a closed-loop feedback system, with an initial value of 0.8 μm / s, and fine-tuned based on the real-time acquired spindle load current signal: when the load current exceeds the preset upper limit (e.g., 1.2 A), the feed rate decreases in steps of 0.05 μm / s; when the load current is below the lower limit (e.g., 0.9 A), it increases in the same increment until it stabilizes at the optimal value corresponding to the target removal rate. This process is implemented using an embedded PLC controller with a sampling frequency of 1 kHz, ensuring a response delay of less than 1 ms.

[0034] During rough grinding, the wafer is vacuum-adsorbed onto the surface of a ceramic platform. The adsorption force is provided by a distributed vacuum chamber, with the chamber pressure maintained between -85 kPa and -90 kPa to ensure that the 12-inch wafer remains warp-free or slips throughout the processing. The platform surface features micron-level concentric grooves to guide the uniform distribution of coolant (a mixture of deionized water and ethylene glycol, volume ratio 95:5). The coolant flow rate is controlled at 2.5 L / min, and the nozzle distance from the grinding wheel's working surface is fixed at 3 mm with a 15° spray angle to maximize heat dissipation and flush away grinding debris. Grinding debris particles are discharged in real-time via a negative pressure suction system beneath the platform to prevent secondary scratches. After rough grinding, the in-plane thickness uniformity (TTV) of the wafer surface is controlled to ≤0.8 μm. This indicator is monitored online using a non-contact capacitive thickness gauge integrated within the equipment. The probe scans along the diameter direction at 5 mm intervals, collecting 25 data points. The TTV value is the difference between the maximum and minimum values.

[0035] In the above method, step 2 involves fine grinding: a ceramic-bonded diamond grinding wheel is installed on the second spindle of the DFG 8560 biaxial grinding system to precisely finish the wafer surface after rough grinding at the micron level. The amount removed in a single pass is controlled within a preset range, the spindle speed is set to a third preset speed, and the table speed is set to a fourth preset speed. By increasing the effective grinding path density and stabilizing the pressure distribution, the microcracks and plastic deformation areas generated in the rough grinding stage are peeled off layer by layer until the complete and undamaged crystalline material body is exposed. Specifically, the ceramic-bonded diamond grinding wheel used in step 2 has an average abrasive grain size of 2μm, corresponding to an 8000-mesh specification. Its binder is composed of sintered alumina-silica composite ceramic with a Vickers hardness ≥1800 HV and a porosity controlled at 8%–12%. This dense structure can maintain the geometric stability of the extremely fine abrasive grains during long-term micro-cutting, ensuring consistent surface roughness. In step 2, the fine grinding removal amount is set to 3–5 μm. This value precisely matches the subsurface damage layer depth (3–5 μm) introduced in the coarse grinding stage, ensuring complete removal of the damaged area without causing additional material waste, while reserving the minimum necessary removal allowance (2–4 μm) for subsequent chemical mechanical polishing.

[0036] During step 2, the wafer does not need to be unloaded or repositioned; it is directly rotated from the stage to below the second spindle. Positioning accuracy is calibrated by the device's built-in laser interferometer, with a repeatability error ≤ ±0.5 μm. The second spindle speed is set to 2200 rpm, and the stage speed is set to 2300 rpm, resulting in a speed ratio of 1:1.045. This high-precision matching aims to maximize the effective grinding path density. Effective grinding path density is defined as the number of intersections of the abrasive wheel trajectories per unit area, calculated using the following formula:

[0037] ;

[0038] in, Effective grinding path density (unit: times / mm²). Main spindle speed (rpm). The speed of the trolley is rpm. Where A is the diameter of the grinding wheel (mm), and A is the effective processing area of ​​the wafer (mm²). The feed rate is (μm / s). In this embodiment, mm, mm² (150 mm radius for a 12-inch wafer). μm / s, substituting into the equation yields... The high-density path, measured at times per mm², ensures that surface micro-undulations are adequately homogenized.

[0039] The pressure control system during the fine grinding stage employs a pneumatic servo valve for adjustment, precisely controlling the normal force applied to the grinding wheel within the range of 0.08–0.12 MPa, with a fluctuation range of ≤±0.005 MPa. A pressure sensor is installed within the spindle bearing housing, sampling at a frequency of 500 Hz, and the data is fed back to the motion controller in real time, forming a force-position closed loop. During this stage, the coolant is replaced with high-purity deionized water (resistivity ≥18 MΩ·cm), and the flow rate is reduced to 1.8 L / min to minimize disturbance to the ultra-fine surface. During fine grinding, the abrasive wear state of the grinding wheel surface is monitored by an acoustic emission sensor. When the high-frequency signal energy (>100 kHz band) decreases by more than 15%, the system automatically triggers the grinding wheel dressing program. The dresser is a single-crystal diamond pen, with a dressing feed of 0.5 μm / stroke. After dressing, the grinding wheel profile is restored to ≤1 μm.

[0040] In the above method, step 3 involves a polishing process: at the end of the fine polishing cycle, the second spindle and the table are continuously rotated at a third and a fourth preset speed, while the feed motion is stopped. This allows the grinding wheel and the wafer to rub against each other in a zero-feed state for a predetermined period of time to eliminate elastic deformation of the workpiece and smooth out microscopic undulations, ultimately achieving nanoscale surface smoothness. Specifically, the polishing time in step 3 is set to 3–5 minutes, which is sufficient to allow the wafer-grinding wheel interface to reach a thermo-mechanical equilibrium state, effectively releasing accumulated elastic strain energy, significantly reducing surface waviness and local protrusions, and ensuring that the final surface roughness Ra value is stably controlled below 0.8 nm.

[0041] The execution logic of the polishing stage is strictly controlled by the equipment's main control program: when the amount of material removed by fine polishing reaches the preset upper limit (5μm) or lower limit (3μm), the feed mechanism immediately retracts to a safe position, while the spindle and stage continue to operate at speeds of 2200 rpm and 2300 rpm respectively. In this zero-feed state, the friction between the wafer and the grinding wheel is dominated by static friction, and the interface temperature is monitored by an infrared thermometer. The heating rate is limited to ≤0.5℃ / min to prevent thermal stress from inducing new microcracks. The polishing time is precisely controlled by a timer. After reaching the set value (e.g., 4 min), the spindle and stage synchronously decelerate to a stop. The deceleration curve uses an S-shaped smoothing algorithm, with an angular acceleration change rate ≤50 rad / s³ to avoid inertial impact. After polishing, the three-dimensional morphology of the wafer surface is acquired by scanning with a white light interferometer. The sampling area is 500 μm × 500 μm, and parameters such as surface roughness Ra, Rq, and Rz are recorded and archived as a basis for process traceability.

[0042] The roughing and finishing grinding processes are completed continuously in a single clamping operation using the DFG 8560 dual-axis system, eliminating the need for unloading or realigning midway. This eliminates positioning errors introduced by multiple clamping operations, ensuring that the roughing reference surface and the finishing surface are strictly coplanar, thus improving overall thickness uniformity. The DFG 8560 dual-axis system independently controls the spindle speed, table speed, and feed rate during the roughing and finishing stages, achieving dynamic decoupling and precise matching of process parameters. This avoids performance compromises between high removal and low damage targets for a single axis system. The method is suitable for processing wafers with severe scratches on the surface, with a depth greater than or equal to 3 μm and a length greater than or equal to 50 μm, or wafers with complex residual film layers greater than or equal to 1 μm and a coverage area greater than or equal to 20%. It exhibits strong process adaptability, requiring no significant adjustments to core parameters for different defect types. The method maintains stable performance for both P-type and N-type doped wafers during continuous processing, with processing efficiency fluctuations not exceeding 3% and surface quality index deviations not exceeding 0.01 μm, indicating good process compatibility with wafers of different electrical properties.

[0043] To verify the practical effect of this embodiment, the following specific application example was constructed: A batch of 12-inch P-type doped silicon wafers were selected, with heavy scratches on the surface having a depth of 4.2±0.5μm and a length of 80–120μm, and a scratch density of 3–5 scratches per wafer. After cleaning, the wafers were loaded into a DFG 8560 equipment and the above three-step process was performed. In the rough grinding stage, the removal amount was set to 10μm, the spindle speed was 1800 rpm, the stage speed was 2000 rpm, and the feed rate was 0.75 μm / s; in the fine grinding stage, the removal amount was 4μm, the spindle speed was 2200 rpm, the stage speed was 2300 rpm, and the feed rate was 0.28 μm / s; the polishing time was 4 min. After processing, focused ion beam (FIB) cross-sectional analysis confirmed that the subsurface damage layer depth was 4.1±0.2μm, and the heavy scratches were completely removed; atomic force microscopy (AFM) measured the surface roughness Ra to be 0.75 nm; and the TTV value was 0.65 μm. The total processing time for a single piece is 7.2 minutes, which is 67.3% shorter than the traditional single-axis 1500-grit rough grinding + offline fine grinding process (total time 22 minutes).

[0044] Another batch of N-type doped wafers were coated with a silicon nitride film with a thickness of 1.8 ± 0.3 μm, covering an area of ​​35%. Using the same process parameters, the rough grinding removal depth was set to 11 μm to ensure complete film removal. Results showed a 100% film removal rate, a subsurface damage layer depth of 4.8 ± 0.15 μm, a total surface damage volume (TTV) of 0.72 μm, and a processing time of 7.8 min. After processing 100 wafers consecutively, the wheel wear was measured using a laser profilometer. The radial wear of the 1500-grit wheel was 12 μm, and that of the 8000-grit wheel was 5 μm, both within acceptable limits, requiring no replacement.

[0045] Example 2

[0046] In another embodiment, for wafers with residual metal film (e.g., copper film thickness of 2.5 μm, covering 25% of the area), the feed strategy during the rough grinding stage was fine-tuned. Since the hardness of the metal film is lower than that of the silicon substrate, to avoid edge curling, the rough grinding feed rate was controlled in segments: the feed rate was 0.6 μm / s for the initial 2 μm removal stage, and increased to 0.9 μm / s for the subsequent 8 μm stage. Simultaneously, 0.1 wt% of a corrosion inhibitor (benzotriazole) was added to the coolant to suppress copper ion migration. The fine grinding and polishing parameters remained unchanged. Verification showed that this adjustment resulted in no residual burrs on the surface after metal film removal, and the subsurface damage layer depth remained stable at 4.3 ± 0.18 μm, meeting the requirements for subsequent polishing.

[0047] Example 3

[0048] For wafers with typical defects (surface roughness Ra 5–8 nm, without deep scratches or thick films), the lower limit of rough grinding removal can be adjusted to 8 μm, and the lower limit of fine grinding removal can be adjusted to 3 μm to further shorten processing time. The photo-grinding time is correspondingly shortened to 3 min. With this configuration, the processing time per wafer is reduced to 6.1 min, while the surface roughness Ra can still be controlled below 0.85 nm, and the total surface velocity (TTV) ≤ 0.75 μm, making it suitable for cost-sensitive, high-volume reclaimed wafer production scenarios.

[0049] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A surface repair method for 12-inch wafers using a DFG 8560 biaxial system with coarse-fine co-polishing grinding, characterized in that... Includes the following steps: Step 1 involves coarse grinding. A resin-bonded diamond grinding wheel is mounted on the first spindle of the DFG 8560 biaxial grinding system to efficiently remove material from the clamped wafer surface. The amount removed in a single pass is controlled to be sufficient to cover the typical depth of severe scratches and complex film residues. The first spindle speed is 1500-2000 rpm, and the table speed is 1800-2200 rpm. The amount removed in a single pass of coarse grinding is controlled to be 8-12 μm. Step 2 involves fine grinding. A ceramic-bonded diamond grinding wheel is installed on the second spindle of the DFG 8560 biaxial grinding system to precisely finish the surface of the wafer after rough grinding at the micron level. The amount of material removed in a single fine grinding cycle is controlled to be 3-5 μm. The spindle speed of the second spindle is set to 2000-2500 rpm, and the table speed is set to 2200-2400 rpm. By increasing the effective grinding path density and stabilizing the pressure distribution, the microcracks and plastic deformation areas generated in the rough grinding stage are peeled off layer by layer until the complete and undamaged crystalline material body is exposed. Step 3 involves performing a polishing process. At the end of the polishing cycle, the rotation speed of the second spindle and the table is maintained while the feed motion is stopped. Friction is performed for 3 to 5 minutes in a zero-feed state, and the interface heating rate is controlled at ≤0.5℃ / min during the polishing process, ultimately achieving a surface roughness Ra≤0.8nm.

2. The surface repair method for 12-inch wafers using the DFG 8560 biaxial system coarse-fine co-polishing grinding according to claim 1, characterized in that, The binder of the resin-bonded diamond grinding wheel used in step 1 is composed of a phenolic resin matrix, which has a moderate elastic modulus and thermal stability. It can maintain the self-sharpening property of the abrasive grains under moderate loads and avoid wafer fragmentation caused by a sudden increase in cutting force due to excessive abrasive grain passivation.

3. The surface repair method for DFG 8560 biaxial system coarse-fine co-polishing grinding for 12-inch wafers according to claim 1, characterized in that, In step 1, the amount of material removed in a single step is controlled to be 8 to 12 micrometers. This range covers heavily scratched defects with a depth of 3 micrometers or more and a length of 50 micrometers or more, as well as complex film residues with a film thickness of 1 micrometer or more and a coverage area of ​​20% or more.

4. The surface repair method for 12-inch wafers using the DFG 8560 biaxial system coarse-fine co-polishing grinding according to claim 1, characterized in that, The binder of the ceramic-bonded diamond grinding wheel used in step 2 is composed of sintered alumina-silica composite ceramic, which has a dense structure and high hardness, and can maintain the geometric stability of the extremely fine abrasive grains in long-term micro-cutting, ensuring consistent surface roughness.

5. The surface repair method for DFG 8560 biaxial system coarse-fine co-polishing of 12-inch wafers according to claim 1, characterized in that, In step 2, the amount of material removed in a single pass is controlled to be 3 to 5 micrometers. This value precisely matches the depth of the subsurface damage layer introduced in the coarse grinding stage, ensuring that the damaged area is completely removed without causing additional material waste, while reserving the minimum necessary removal allowance for subsequent chemical mechanical polishing.

6. The surface repair method for 12-inch wafers using the DFG 8560 biaxial system coarse-fine co-polishing grinding according to claim 1, characterized in that, The polishing process in step 3 takes 3 to 5 minutes. This duration allows the wafer and grinding wheel interface to reach a state of thermal-mechanical equilibrium, effectively releasing the accumulated elastic strain energy, significantly reducing surface waviness and local protrusions, and ensuring that the final surface roughness Ra value is stably controlled below 0.8 nanometers.

7. The surface repair method for DFG 8560 biaxial system coarse-fine co-polishing grinding for 12-inch wafers according to claim 1, characterized in that, The rough grinding and fine grinding processes are completed continuously in a single clamping operation in the DFG 8560 biaxial system, without the need for unloading or realigning midway, ensuring that the rough grinding reference surface and the fine grinding surface are strictly coplanar, thus improving the overall thickness uniformity.

8. The surface repair method for DFG 8560 biaxial system coarse-fine co-polishing grinding for 12-inch wafers according to claim 1, characterized in that, The DFG 8560 dual-axis system independently controls the spindle speed, table speed, and feed rate during the roughing and finishing grinding stages, achieving dynamic decoupling and precise matching of process parameters, and avoiding performance compromises between high removal and low damage targets for a single axis system.

Images