Balanced cutting process for reducing wafer rejection rate

Abstract

This invention discloses a balanced cutting process to reduce silicon wafer scrap rate, relating to the field of silicon wafer processing. It involves initial parameter setting and loading, dynamic determination of the cutting stage, execution of cutting parameters in stages, feedback after batch cutting, real-time dynamic adjustment during the cutting process, and comprehensive verification and process solidification of cutting results. By dynamically dividing the cutting process into three stages—initial, intermediate, and final—and matching differentiated line running cycle and return-to-line ratio parameter combinations to each stage, the wear rate of the steel wire is indirectly obtained by real-time monitoring of the main motor current. When the wear rate is abnormal, the return-to-line ratio of the current stage is dynamically fine-tuned to achieve dynamic adjustment of the line running cycle and return-to-line ratio parameters. For potential comprehensive quality anomalies such as curvature and thickness after cutting, the parameters of the most directly affected stage are adjusted, and parameters are gradually adjusted to optimize the process.

Description

Technical Field

[0001] This invention relates to the field of silicon wafer processing, and in particular to a balanced cutting process for reducing silicon wafer scrap rate. Background Technology

[0002] Wire cutting technology is currently the mainstream method for processing crystal rods into silicon wafers. Its core is to use a high-speed moving steel wire to carry cutting slurry to grind the crystal rod.

[0003] The wire EDM process is complex and is affected by many parameters, including wire tension, wire speed, feed rate, and mortar properties.

[0004] Currently, the cutting process parameters commonly used in the industry are mostly fixed values. This static parameter setting method has significant drawbacks: Unable to adapt to the dynamic changes in the cutting process: The cutting process can be divided into three stages: initial entry, mid-term stable cutting, and final separation. The stress state, heat dissipation conditions, and fracture risk are completely different in each stage, and fixed parameters cannot reach the optimal in all stages.

[0005] During the cutting process, the steel wire will continuously wear down, and its cutting ability will dynamically decrease. The existing process cannot make online adjustments based on the real-time wear rate of the steel wire or the cutting status. When abnormal wear or abnormalities occur, it cannot respond in time, resulting in batch thickness abnormalities or wire breakage accidents.

[0006] Therefore, we propose a balanced cutting process to reduce the scrap rate of silicon wafers. Summary of the Invention

[0007] The purpose of this invention is to provide a balanced cutting process that reduces the scrap rate of silicon wafers, thereby solving the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a balanced dicing process for reducing silicon wafer scrap rate, comprising the following steps: S1 Initial Parameter Setting and Loading: Detect the initial state of the steel wire and crystal rod, select the basic process scheme according to the crystal rod specifications, set the corresponding wire running cycle and return ratio parameters, and load them as the initial cutting parameters; S2 Cutting Stage Dynamic Determination: After cutting begins, the system calculates cyclically according to the cutting cycle and dynamically determines the current cutting stage based on the real-time cutting depth; S3 Stage-by-Stage Cutting Parameter Execution: Based on the cutting stage determined in step S2, the corresponding preset steel wire parameter group is called to control the cutting process. S4 batch cutting feedback: After a batch cutting is completed, the thickness uniformity of the cut silicon wafers and the wear degree of the steel wires are detected. Based on the steel wire wear difference, the return line ratio parameter of the next batch cutting is adjusted and / or the line running cycle parameter in the middle of the next batch cutting is adjusted based on the overall thickness of the silicon wafers, forming optimized new process parameters for the next batch cutting. S5 cutting process real-time dynamic adjustment: real-time status input and stage judgment. At the beginning of the cutting cycle, the real-time cutting depth is input, the cutting stage is judged in real time, and the wear rate is judged to realize the dynamic fine adjustment of parameters during the cutting process, so as to reduce the abnormal wear of the steel wire. S6 Cutting Results Comprehensive Verification and Process Consolidation: The silicon wafers after the complete batch of cutting are subjected to comprehensive quality testing to verify the stability of the current process parameters.

[0009] Preferably, the cutting stage based on step S2 includes at least the initial cutting stage, the middle cutting stage, and the final cutting stage; The preset steel wire parameter group described in step S3 includes at least a first preset steel wire parameter, a second preset steel wire parameter, and a third preset steel wire parameter, which correspond to the initial cutting stage, the middle cutting stage, and the final cutting stage, respectively, and the wire running cycle and return ratio in each group of parameters are different.

[0010] Preferably, the adjustment of the hysteresis ratio parameter based on the wire wear difference in step S4 specifically includes: Calculate the difference in steel wire wear before and after this batch of cutting; If the difference is less than the set threshold, the current hysteresis ratio parameter remains unchanged; If the difference is greater than the set threshold, the loop ratio in subsequent cuts will be reduced proportionally according to the proportion exceeding the threshold.

[0011] Preferably, the operating cycle parameters based on the overall thickness adjustment line of the silicon wafer mentioned in step S4 specifically include: Inspect the overall thickness of this batch of silicon wafers; If the overall thickness is too thick, shorten the line operation cycle in the middle of the next batch of cutting; If the overall thickness is appropriate, then keep the current line operating cycle parameters unchanged; If the overall thickness is too thin, extend the line running cycle in the middle of the next batch of cutting.

[0012] Preferably, the adjustment is based on real-time dynamic adjustment during the cutting process in step S5: Real-time wear monitoring monitors the changing trend of the main motor current and indirectly converts it into a steel wire wear rate signal. When the system determines that the wear rate has not exceeded the preset safety threshold, it will not interfere with the current parameters; When the wear rate exceeds the threshold, the hysteresis ratio of the current stage is dynamically reduced proportionally based on the trend of the excess. Motion control execution calculates the incoming and outgoing times for the current cycle based on the final determined line running cycle and return line ratio, outputs motion control commands, and drives the actuator to complete the cutting for the current cycle.

[0013] Preferably, dynamic adjustment is based on step S5: If the wear rate of the steel wire exceeds the set wear rate threshold, the return line ratio is adjusted proportionally, and the line operation cycle and return line ratio data are dynamically set. Perform calculations and allocate time, calculate the inbound and return times for this cycle, execute the inbound and return times, complete the cutting, and determine whether the batch of cyclic cutting has ended; If the batch of cyclic cutting has not ended, a new cutting cycle calculation begins. Real-time status is input again, and the trend of main motor current change is monitored in real time. Wear acceleration signal and normal signal are output. Calculations are performed and time is allocated. Based on the final determined cycle and return line ratio, the entry time and return time of this cycle are calculated. Motion control commands are output, and the execution time and return time are executed until the cutting is completed. It is then determined whether the batch of cyclic cutting has ended. If it has not ended, the cycle continues, and the results are verified.

[0014] Preferably, a comprehensive quality inspection is performed on a complete batch of cut silicon wafers based on step S6, and the inspection items include overall thickness, total thickness deviation and curvature; Overall thickness detection and judgment, measuring the thickness at the center point; Total thickness deviation was measured at multiple points, including the center point and edges of the silicon wafer. Curvature, measuring the curvature data of the silicon wafer's centerline; The collected data is categorized, and the silicon wafer thickness parameter is compared with the threshold for judgment. If all test results are within the set threshold range, the current process parameters will be saved as the standard cutting process. If a specific anomaly is detected, a special anomaly diagnosis and parameter matching adjustment process will be initiated, and parameters will be adjusted only for one major anomaly at a time.

[0015] Preferably, when an abnormal curvature is detected, the parameter matching and adjustment process includes: Prioritize adjusting the line running cycle and return line ratio parameters at the end of the cutting process, with the adjustment direction being to reduce the line running cycle and reduce the return line ratio; If the problem is not resolved or the optimization is not significant after adjustment, further adjust the line running cycle and return line ratio parameters during the cutting process.

[0016] Preferably, when an abnormality in the overall thickness of the silicon wafer is detected, and after ruling out equipment problems, it is determined to be a parameter abnormality, the parameter matching and adjustment process includes: Adjust the line running cycle and return line ratio parameters during the cutting process, and execute the next batch of cutting according to the adjusted parameters; The silicon wafer thickness is re-inspected. If it meets the standard, the parameters are saved, the next batch of cutting is executed, and batch thickness is continuously inspected to detect any abnormalities in a timely manner. If the target is not met, determine whether the adjustment will improve the situation. If it does improve the situation, calculate the adjustment range to determine whether the adjustment was too much and caused a pullback, or too little and increased the parameters accordingly. If the problem doesn't improve, readjust the parameters.

[0017] Preferably, when an abnormal total thickness deviation is detected, the parameter matching and adjustment process includes: If the deviation exhibits periodic fluctuations, prioritize adjusting the line running cycle parameters during the cutting phase; If the current running cycle is lower than the preset range, the cycle is extended; if the current running cycle is higher than the preset range, the cycle is shortened. If a geometric tilt deviation occurs, it is determined that a static positional deviation exists; If random fluctuations occur without any pattern, and the source of the interference is random and non-periodic, then mechanical problems should be investigated.

[0018] The technical effects and advantages of this invention are as follows: By dynamically dividing the cutting process into three stages—initial, intermediate, and final—and matching differentiated combinations of line running cycle and return line ratio parameters for each stage, and indirectly obtaining the steel wire wear rate by monitoring the main motor current in real time, the return line ratio of the current stage is dynamically fine-tuned when the wear rate is abnormal, thereby achieving dynamic adjustment of the line running cycle and return line ratio parameters.

[0019] Then, by analyzing the thickness of the silicon wafers after batch cutting and the wear data of the steel wires, the process parameters for the next batch are automatically adjusted. The return line ratio is adjusted according to the wear of the steel wires to control the wear rate, and the intermediate line running cycle is adjusted according to the thickness of the silicon wafers to control the cutting accuracy.

[0020] For comprehensive quality anomalies such as curvature and thickness that may occur after cutting, a specialized diagnostic and matching adjustment process is provided. Prioritize the adjustment of parameters at the most directly affected stage, and gradually adjust parameters to resolve anomalies and optimize the process. When matching abnormal parameters, the principle of single anomaly control should be followed, that is, only one anomaly is adjusted at a time. This avoids the uncontrollable interactive effects caused by adjusting multiple parameters at the same time, improves the optimization process, achieves final process determination, and reduces the overall scrap rate of silicon wafers. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the overall process of the present invention. Figure 2 This is a flowchart illustrating the cutting stage judgment process of the present invention. Figure 3 This is a flowchart illustrating the feedback process for detecting the steel wire wear difference and the thickness of the cut silicon wafer in this invention. Figure 4 This is a flowchart illustrating the line operation cycle and return line ratio of the present invention; Figure 5 This is a flowchart illustrating the overall cutting parameter analysis and process adjustment of this invention. Figure 6 This is a flowchart illustrating the dynamic adjustment process of steel wire wear during the cutting stage of this invention. Figure 7 This is a flowchart for verifying the process parameters of the present invention; Figure 8 This is a flowchart of the abnormal parameter matching process of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] This invention provides, for example Figures 1-8 The illustrated equalization dicing process for reducing silicon wafer scrap rate includes the following steps: S1 Initial Parameter Setting and Loading: Perform initial state detection of steel wire and crystal rod, input the detection data, select the specified basic process scheme according to the crystal rod specifications, set the corresponding line running cycle and return line ratio parameters, and load them as the initial cutting parameters.

[0024] The initial state of the steel wire and crystal rod is detected, and the detection data is input. A specified process is selected, and the corresponding running cycle and return line ratio are set. The process data is loaded as the initial parameters for cutting. At the start of cutting, a cutting cycle is calculated, the real-time status is input, and the current cutting depth is given. It is judged as one of the early, middle and late stages of cutting. The judgment state is also set in the process parameters. For example, in this embodiment, the cutting time is set to 20% for the early stage of cutting, 20%-80% for the middle stage of cutting, and 80% for the late stage of cutting. It can be adjusted according to the size of the crystal rod and the cutting thickness. It is not a unique and constant parameter. S2 Cutting Stage Dynamic Determination: After cutting begins, the system performs cyclic calculations according to a cutting cycle, inputs status data such as the current cutting depth in real time, and dynamically determines the current cutting stage based on preset rules.

[0025] S3 Stage-by-Stage Cutting Parameter Execution: Based on the cutting stage determined in step S2, the corresponding preset steel wire parameter group is called for cutting control.

[0026] Load parameters to start cutting, determine the current cutting stage, and divide the cutting process into three stages according to the cutting time: initial cutting stage, middle cutting stage, and final cutting stage. Cutting is performed according to the first preset steel wire parameters in the initial stage of cutting to improve the penetration efficiency of the cutting fluid into the cutting kerf and lay the foundation for stable cutting.

[0027] During the middle stage of cutting, cutting is carried out according to the second preset steel wire parameters to balance wire wear, silicon wafer edge stress, and ensure that the cutting fluid continuously and stably enters the cutting seam. This stage is crucial to ensuring the quality of the silicon wafer body.

[0028] Cutting is performed at the end of the cutting process according to the third preset steel wire parameters. The core objective is to reduce the risk of wire breakage and to stably complete the final separation of the silicon wafer from the parent material.

[0029] The first, second, and third preset steel wire parameters all include the wire running cycle and return ratio, which improve the wire cutting force in the three stages at the end of the cutting process and reduce high-level wire breakage. By using differentiated parameters in different stages, especially the optimized parameter combination in the end of the cutting process, the probability of high-level wire breakage can be effectively reduced while maintaining the necessary cutting force.

[0030] Specific parameter range examples are as follows: The first preset steel wire parameters correspond to a wire running cycle of 70 to 190 seconds and a return-to-wire ratio of 1.3 to 1.7:1. The second preset steel wire parameters correspond to a wire running cycle of 110 to 230 seconds and a return wire ratio of 1.4 to 2:1. The line running cycle corresponding to the third preset steel wire parameters is 70 to 190 seconds, and the return line ratio is 1.1 to 1.7:1.

[0031] S4 Batch Cutting Feedback: After a batch cutting is completed, result detection and parameter optimization are performed. Inspection and Comparison: The thickness uniformity of the cut silicon wafers and the wear degree of the steel wires are inspected. The yield rate, thickness data, and steel wire wear rate of the silicon wafers are compared with ideal data or historical compliance data.

[0032] Adjustment based on wire wear: Calculate the difference in wire wear before and after this batch of cutting. If the difference is less than the set threshold, the wear is considered normal, and the current loop ratio parameter remains unchanged. If the difference is greater than the threshold, the loop ratio in subsequent cuts will be reduced proportionally according to the proportion exceeding the threshold.

[0033] Adjustment based on silicon wafer thickness: Detect the overall thickness of this batch of silicon wafers.

[0034] If the overall thickness is too large, shorten the running cycle of the intermediate cutting line; If the thickness is appropriate, then maintain it; If the overall thickness is too thin, extend the running cycle of the intermediate cutting line.

[0035] Generate new process parameters: By comprehensively adjusting the return line ratio based on the difference in steel wire wear and the operating cycle based on the silicon wafer thickness, and combining the initially set threshold parameters for each stage, the return line ratio and operating cycle parameters for the initial, middle, and final stages of the next batch of cutting are comprehensively adjusted to form a set of optimized new process parameters.

[0036] Iterative optimization: The next batch of cuts will be executed using the adjusted parameters.

[0037] Once the batch is completed, repeat step S4 above to conduct a new round of testing, comparison, and parameter fine-tuning, thereby achieving continuous iterative optimization of the parameters.

[0038] The threshold for the wear difference of the steel wire can be dynamically set according to the strength and size of each batch of crystal rods and the total number of cuts planned for this batch, so as to improve the targeting of the adjustment.

[0039] The specific operation is as follows: After the previous batch of silicon wafers is cut, the thickness of the cut silicon wafers and the wear of the steel wires are tested. The pass rate of the silicon wafers and the wear rate of the steel wires are compared with the ideal data to determine whether the pass rate is within the allowable threshold. If it is not within the threshold, corresponding adjustments are made. Specifically, when the wear difference of the steel wire is detected, that is, the wear value of the steel wire after the batch of cutting is detected, the difference between the wear value of the steel wire before the batch of cutting is calculated, which is the wear difference of the steel wire in this batch. The system determines whether the current steel wire wear difference is greater than a set threshold. If the steel wire wear difference is less than the threshold, the steel wire wear difference in this batch of cutting is considered acceptable. At this time, the current return line ratio parameter is kept unchanged, and the value is output. When the wear difference of the steel wire exceeds the threshold, the return line ratio is reduced proportionally by adjusting the value as it exceeds the range of the difference, and the value is then output. When the thickness of the cut silicon wafer is detected and determined, if the overall thickness of the cut silicon wafer is too thick, the intermediate operation cycle is shortened and the value is output. When the detected thickness is within a suitable range, keep the current operating cycle parameters unchanged and output the value; When the overall thickness of the cut silicon wafer is found to be too thin, the intermediate operation cycle is extended, and this value is output. By adjusting the return line ratio based on the difference in steel wire wear and the operating cycle based on the thickness of the cut silicon wafer, and combining the initially set threshold, the return line ratio and operating cycle parameters in the early, middle and late stages of the cutting process are adjusted to form new process parameters. The adjusted line running cycle and return line ratio are used for cutting, and the running cycle and return line ratio parameters for the initial, middle and final stages are selected according to the corresponding cutting stage until the cutting is completed. After a batch of cutting is completed, the wire wear difference and the thickness of the cut silicon wafer are detected and compared again. The same comparison is performed as above, and the return line ratio and running cycle are adjusted. The adjusted data is output. Combined with the new process parameters of the previous batch, the return line ratio and running cycle parameters in the early, middle and late stages of the cutting stage are adjusted to form new process parameters. Cutting is carried out according to this process.

[0040] The threshold for the wire wear difference can be adjusted based on the crystal rod strength, crystal rod size, and total number of cuts in the batch. Furthermore, when starting a cutting cycle calculation, the real-time status is input; That is, input the current cutting depth, and convert it into the percentage of cutting depth through the percentage of cutting time, so as to obtain the determination of the initial, middle and final stages of cutting. The output is based on the current batch's return line ratio and running cycle parameters. Example 1: Real-time dynamic adjustment during the S5 cutting process: Real-time status input and stage determination. At the beginning of each cutting cycle, the real-time cutting depth is input and converted into a percentage relative to the total cutting depth, thereby determining in real time which cutting stage is currently in.

[0041] Real-time wear monitoring monitors the changing trend of the main motor current and indirectly converts it into a steel wire wear rate signal.

[0042] When the system determines that the wear rate has not exceeded the preset safety threshold, it will not interfere with the current parameters; When the wear rate exceeds the threshold, the system dynamically adjusts the current stage's return ratio proportionally based on the trend of the excess, thereby achieving dynamic fine-tuning of parameters during the cutting process to mitigate abnormal wear of the steel wire.

[0043] Motion control execution calculates the specific entry and return times for the current cycle based on the final determined line running cycle and return ratio, outputs motion control commands, and drives the actuator to complete the cutting for the current cycle.

[0044] Specifically, the process involves: real-time monitoring of the main motor current change trend; converting the current change into steel wire wear value; checking whether the steel wire wear rate exceeds the set wear rate threshold; and not adjusting parameters if the wear rate does not exceed the set wear rate threshold. When the wear rate of the steel wire exceeds the set wear rate threshold, the return line ratio is proportionally reduced according to the trend. This reduces the return line ratio data and the line running cycle. The return line ratio is obtained by multiplying the corresponding running cycle and return line ratio parameters set for this batch by a proportional reduction, thus achieving dynamic setting of the line running cycle and return line ratio data. Through dynamic adjustment, the wear on the steel wire is reduced. Calculate and allocate time, calculate the entry time and return time of this cycle based on the final determined cycle and return line ratio, output motion control commands, execute the entry time and return line execution time until the cutting is completed, and determine whether the batch cyclic cutting has ended; If the current batch of cyclic cutting has not ended, a new cutting cycle calculation begins. The real-time status is input again. First, the current cutting depth is determined. At the current cutting depth, the change trend of the main motor current is monitored in real time, and wear acceleration signal and normal signal are output. Calculations are performed and time is allocated. Based on the final determined cycle and return line ratio, the entry time and return time of this cycle are calculated, and motion control commands are output. The execution time and return time are executed until the cutting is completed. It is then determined whether the current batch of cyclic cutting has ended. If it has not ended, the cycle continues, and the results are verified.

[0045] The thickness of the silicon wafers cut in this batch was tested to check for any abnormalities in the test parameters, and the current process parameters were verified.

[0046] And reduce the return ratio when the steel wire wears out quickly.

[0047] By controlling the wire running cycle and return ratio during cutting, the fluctuation of silicon wafers is reduced, thereby improving the surface morphology of the cut silicon wafers. The synergistic effect of the return ratio and wire running cycle improves the silicon wafer morphology and extends the service life of the wire. The wire operating cycle refers to the sum of the forward movement time and the reverse movement time of the wire.

[0048] The return ratio refers to the ratio of the forward movement length to the reverse movement length of the wire.

[0049] Furthermore, the initial state of the steel wire and crystal rod is input, the cutting process parameters for the corresponding cutting stage are executed, a single cutting is completed, the single cutting result is detected, the process parameters are dynamically adjusted, and cutting is executed. The single cutting is performed and dynamically adjusted. By monitoring the change trend of the main motor current in real time, dynamic adjustment is performed to effectively improve the cutting qualification rate until the overall cutting is completed. Then, the scrap rate is tested again to ensure compliance.

[0050] Example 2: Comprehensive Verification and Process Consolidation of S6 Cutting Results: A comprehensive quality inspection was performed on a complete batch of cut silicon wafers, including overall thickness, curvature, and total thickness deviation, to verify the stability of the current process parameters. If the silicon wafer scrap rate and all quality indicators are within the set threshold range, the current process parameters are considered stable and effective. These parameters are then saved and recorded as a standard cutting process for similar ingots.

[0051] If abnormal parameters are detected, such as abnormal thickness or abnormal curvature, the process of special abnormality diagnosis and parameter matching and adjustment will be initiated.

[0052] After the overall cutting of the current batch is completed, it is determined whether the scrap rate of the silicon wafers cut in that batch meets the standard. When the scrap rate is within the set threshold, the process parameters are not adjusted. The process is recorded by saving the parameters and using them as the cutting process. This process is initiated when a specific anomaly is detected during silicon wafer quality inspection. Single anomaly control is implemented, with parameter adjustments made only for one major anomaly at a time to avoid uncontrollable interactive effects introduced by simultaneous changes in multiple parameters.

[0053] The specific process is as follows: each time a new silicon wafer is cut, the initial state of the steel wire and crystal rod is detected, the detection data is input, and a similar process is selected. The parameters are loaded to start the cutting process. After the cutting is completed, the process parameters are verified, the silicon wafer thickness is detected, and the overall thickness, curvature, and total thickness deviation of the silicon wafer are judged. To measure the thickness of silicon wafers, multiple wafers were randomly selected and the following operations were performed: Overall thickness detection and judgment, measuring the thickness at the center point; Total thickness deviation was measured at multiple points, including the center point and edges of the silicon wafer. Curvature, measuring the curvature data of the silicon wafer's centerline; The collected data is classified, and the silicon wafer thickness parameter is compared with the threshold to determine whether there are any abnormalities. If there are no abnormalities, the parameters remain stable, and the next batch of cutting is performed. When parameters are abnormal, the abnormal parameters are matched according to the corresponding overall thickness abnormality, total thickness deviation abnormality, and curvature abnormality. Based on the corresponding abnormality, the corresponding parameters are adjusted to generate new parameters. Its main adjustment is to adjust the running cycle and return line ratio in the early, middle and late stages of cutting, input the adjusted new parameters, cut the next batch, and then re-detect the silicon wafer thickness. Reassess whether there are any abnormalities in the newly adjusted process parameters, compare them with any abnormal parameters, and adjust the abnormal parameters until the parameters are normal. At this point, save the process parameters and execute them as the new process. When performing abnormal parameter matching, matching is performed for overall thickness abnormality, total thickness deviation abnormality, and curvature abnormality; Perform a bending anomaly diagnosis and stress distribution analysis, since bending is an elastic deformation that occurs after the residual stress inside the silicon wafer is released due to uneven distribution.

[0054] During the cutting process, the mechanical and thermal stresses on the silicon wafer are formed at the last moment when it is completely separated from the crystal rod. At the end of the cutting process, the silicon wafer is only connected to the parent material through the remaining silicon core. Its connection area is the smallest, its structural stability is the most fragile, and it is easy to deform. At this time, any movement of the steel wire is unstable and will directly affect the final shape of the silicon wafer. Therefore, the final adjustment is given priority. Intermediate dicing is the main part of the dicing process. Intermediate parameters cause the silicon wafer to accumulate directional stress throughout the dicing process, so as to reduce the overall stress accumulation level. Secondary intermediate parameters are adjusted to match the dicing process.

[0055] When an abnormal curvature occurs, the line running cycle and return line ratio parameters at the end are adjusted first. After cutting and judging, it is first judged whether the curvature problem is solved. If it is solved, no further adjustment is needed. If it is not solved, it is judged whether the optimization is obvious. If it is obvious, the above adjustment and optimization are continued. The adjustment range is calculated and adjusted again. If it is not obvious, the intermediate cutting adjustment is performed until the curvature is within the qualified range. The adjustment direction in the late and middle stages is to reduce the line running cycle and the return line ratio. By increasing the commutation frequency, the stress concentration point moves quickly, avoiding the generation of local high stress areas at the final connection. By shortening the unidirectional stroke, the unidirectional pulling on the silicon wafer at the moment of separation is reduced, creating a high-frequency, small-amplitude cut. This allows the silicon wafer to be finally cut in a relatively uniform low-stress state, thereby optimizing the curvature.

[0056] To diagnose overall thickness anomalies and determine whether they are due to parameter anomalies, the equipment is first monitored to rule out equipment problems. The feed rate, tension, mortar concentration, and temperature parameters are then assessed to determine if there are any errors or fluctuations. If equipment anomalies are ruled out, then the problem is determined to be a parameter issue. Adjustments were made to the operating cycle and return line ratio parameters of the intermediate line. Since the intermediate cutting is the main part of the cutting, its overall thickness is abnormal, which is mainly affected by the intermediate parameters. By adjusting the intermediate parameters, a new parameter set was formed, and the next batch of cutting was executed to re-evaluate whether the thickness of this batch meets the standard. Once the target is met, save the new parameters, execute the next batch of cutting, and continuously monitor the batch thickness to promptly detect any anomalies. If the thickness test of this batch fails to meet the standard after adjustment, determine whether there is any improvement after adjustment. If there is improvement, recalculate the adjustment range to determine whether the adjustment was too much or too little. If so, increase the parameter adjustment, and follow the standard of taking the middle value of the adjustment range to avoid excessive adjustment, which may cause fluctuations in the results and affect subsequent anomaly judgment. If there is no improvement, readjust the parameter matching. After readjustment, proceed with the next batch of cutting, and re-test the thickness of this batch to see if it meets the standard, until the thickness test meets the standard, and save the new parameters.

[0057] When an abnormal total thickness deviation is detected, the parameter matching and adjustment process includes: Perform total thickness deviation diagnosis, analyze cutting force fluctuations, and adjust parameters when periodic fluctuations occur, primarily adjusting the mid-term operating cycle and secondarily adjusting the mid-term hysteresis ratio. The judgment is made based on the current medium-term operating cycle, and the current cycle is compared with the set medium-term operating range; The current operating cycle is too short, the reversal is too frequent, and the cutting force fluctuates at a high frequency, which easily forms periodic thickness ripples on the silicon wafer surface. By extending the cycle, the fluctuation frequency can be reduced, the cutting time can be lengthened, and the thickness changes can be smoothed out.

[0058] If the current operating cycle is too long, the wear effect of the steel wire will accumulate during unidirectional cutting, leading to a long-term attenuation trend of the cutting force and forming periodic thickness changes. By shortening the cycle and frequently changing direction, the cumulative effect can be reduced and the stability of the cutting state can be improved.

[0059] When a geometric tilt deviation occurs, that is, a tilted thickness change from before to after cutting, it can be preliminarily determined that there is a static position deviation, which can be divided into guide wheel problems, uneven groove spacing, non-parallel guide wheels and tilted guide wheel shafts, crystal rod installation problems, bonding tilt, and feed direction not perpendicular to the wire mesh plane. When random fluctuations occur, they are irregular and the sources of interference are random and non-periodic. It is necessary to investigate mechanical loosening issues, check for wear of guide wheel bearings, spindle runout and frame vibration, as well as whether the mortar has uneven abrasive particle distribution and unstable fluid supply.

[0060] Furthermore, when performing abnormal parameter matching, the principle of single abnormal control should be followed, that is, only one abnormal problem should be adjusted at a time to avoid simultaneous control and mutual influence.

[0061] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A balanced cutting process for reducing silicon wafer scrap rate, characterized in that, Includes the following steps: S1 Initial Parameter Setting and Loading: Detect the initial state of the steel wire and crystal rod, select the basic process scheme according to the crystal rod specifications, set the corresponding wire running cycle and return ratio parameters, and load them as the initial cutting parameters; S2 Cutting Stage Dynamic Determination: After cutting begins, the system calculates cyclically according to the cutting cycle and dynamically determines the current cutting stage based on the real-time cutting depth; S3 Stage-by-Stage Cutting Parameter Execution: Based on the cutting stage determined in step S2, the corresponding preset steel wire parameter group is called to control the cutting process. S4 batch cutting feedback: After a batch cutting is completed, the thickness uniformity of the cut silicon wafers and the wear degree of the steel wires are detected. Based on the steel wire wear difference, the return line ratio parameter of the next batch cutting is adjusted and / or the line running cycle parameter in the middle of the next batch cutting is adjusted based on the overall thickness of the silicon wafers, forming optimized new process parameters for the next batch cutting. S5 cutting process real-time dynamic adjustment: real-time status input and stage judgment. At the beginning of the cutting cycle, the real-time cutting depth is input, the cutting stage is judged in real time, and the wear rate is judged to realize the dynamic fine adjustment of parameters during the cutting process, so as to reduce the abnormal wear of the steel wire. S6 Cutting Results Comprehensive Verification and Process Consolidation: The silicon wafers after the complete batch of cutting are subjected to comprehensive quality testing to verify the stability of the current process parameters.

2. The balanced cutting process for reducing silicon wafer scrap rate according to claim 1, characterized in that, The cutting stage described in step S2 includes at least the initial cutting stage, the middle cutting stage, and the final cutting stage. The preset steel wire parameter group described in step S3 includes at least a first preset steel wire parameter, a second preset steel wire parameter, and a third preset steel wire parameter, which correspond to the initial cutting stage, the middle cutting stage, and the final cutting stage, respectively, and the wire running cycle and return ratio in each group of parameters are different.

3. The balanced cutting process for reducing silicon wafer scrap rate according to claim 1, characterized in that, The adjustment of the hysteresis ratio parameter based on the steel wire wear difference in step S4 specifically includes: Calculate the difference in steel wire wear before and after this batch of cutting; If the difference is less than the set threshold, the current hysteresis ratio parameter remains unchanged; If the difference is greater than the set threshold, the loop ratio in subsequent cuts will be reduced proportionally according to the proportion exceeding the threshold.

4. The balanced cutting process for reducing silicon wafer scrap rate according to claim 1, characterized in that, The specific parameters of the silicon wafer overall thickness adjustment line operation cycle mentioned in step S4 include: Inspect the overall thickness of this batch of silicon wafers; If the overall thickness is too thick, shorten the line operation cycle in the middle of the next batch of cutting; If the overall thickness is appropriate, then keep the current line operating cycle parameters unchanged; If the overall thickness is too thin, extend the line running cycle in the middle of the next batch of cutting.

5. The balanced cutting process for reducing silicon wafer scrap rate according to claim 1, characterized in that, Based on real-time dynamic adjustment during the cutting process in step S5: Real-time wear monitoring monitors the changing trend of the main motor current and indirectly converts it into a steel wire wear rate signal. When the system determines that the wear rate has not exceeded the preset safety threshold, it will not interfere with the current parameters; When the wear rate exceeds the threshold, the hysteresis ratio of the current stage is dynamically reduced proportionally based on the trend of the excess. Motion control execution calculates the incoming and outgoing times for the current cycle based on the final determined line running cycle and return line ratio, outputs motion control commands, and drives the actuator to complete the cutting for the current cycle.

6. The balanced cutting process for reducing silicon wafer scrap rate according to claim 1, characterized in that, Dynamic adjustment based on step S5: If the wear rate of the steel wire exceeds the set wear rate threshold, the return line ratio is adjusted proportionally, and the line operation cycle and return line ratio data are dynamically set. Perform calculations and allocate time, calculate the inbound and return times for this cycle, execute the inbound and return times, complete the cutting, and determine whether the batch of cyclic cutting has ended; If the batch of cyclic cutting has not ended, a new cutting cycle calculation begins. Real-time status is input again, and the trend of main motor current change is monitored in real time. Wear acceleration signal and normal signal are output. Calculations are performed and time is allocated. Based on the final determined cycle and return line ratio, the entry time and return time of this cycle are calculated. Motion control commands are output, and the execution time and return time are executed until the cutting is completed. It is then determined whether the batch of cyclic cutting has ended. If it has not ended, the cycle continues, and the results are verified.

7. The balanced cutting process for reducing silicon wafer scrap rate according to claim 1, characterized in that, Based on step S6, a comprehensive quality inspection is performed on a complete batch of diced silicon wafers. The inspection items include overall thickness, total thickness deviation, and curvature. Overall thickness detection and judgment, measuring the thickness at the center point; Total thickness deviation was measured at multiple points, including the center point and edges of the silicon wafer. Curvature, measuring the curvature data of the silicon wafer's centerline; The collected data is categorized, and the silicon wafer thickness parameter is compared with the threshold for judgment. If all test results are within the set threshold range, the current process parameters will be saved as the standard cutting process. If a specific anomaly is detected, a special anomaly diagnosis and parameter matching adjustment process will be initiated, and parameters will be adjusted only for one major anomaly at a time.

8. The balanced cutting process for reducing silicon wafer scrap rate according to claim 7, characterized in that, When an abnormal curvature is detected, the parameter matching and adjustment process includes: Prioritize adjusting the line running cycle and return line ratio parameters at the end of the cutting process, with the adjustment direction being to reduce the line running cycle and reduce the return line ratio; If the problem is not resolved or the optimization is not significant after adjustment, further adjust the line running cycle and return line ratio parameters during the cutting process.

9. The balanced cutting process for reducing silicon wafer scrap rate according to claim 7, characterized in that, When an abnormality in the overall thickness of the silicon wafer is detected, and after ruling out equipment problems, it is determined to be a parameter anomaly, the parameter matching and adjustment process includes: Adjust the line running cycle and return line ratio parameters during the cutting process, and execute the next batch of cutting according to the adjusted parameters; The silicon wafer thickness is re-inspected. If it meets the standard, the parameters are saved, the next batch of cutting is executed, and batch thickness is continuously inspected to detect any abnormalities in a timely manner. If the target is not met, determine whether the adjustment will improve the situation. If it does improve the situation, calculate the adjustment range to determine whether the adjustment was too much and caused a pullback, or too little and increased the parameters accordingly. If the problem doesn't improve, readjust the parameters.

10. A balanced dicing process for reducing silicon wafer scrap rate according to claim 7, characterized in that, When an abnormal total thickness deviation is detected, the parameter matching and adjustment process includes: If the deviation exhibits periodic fluctuations, prioritize adjusting the line running cycle parameters during the cutting phase; If the current running cycle is lower than the preset range, the cycle is extended; if the current running cycle is higher than the preset range, the cycle is shortened. If a geometric tilt deviation occurs, it is determined that a static positional deviation exists; If random fluctuations occur without any pattern, and the source of the interference is random and non-periodic, then mechanical problems should be investigated.

Images