Intelligent control method and system for surface type and flatness of thinned wafer

Abstract

The invention discloses an intelligent control method and system for the surface type and flatness of a thinned wafer, and belongs to the field of semiconductor manufacturing. The system comprises an intelligent air floating main shaft, a wafer surface type identification system, a flatness measurement system, an intelligent adjustment algorithm system, a main shaft angle adjustment device and a main shaft operation state monitoring system. By monitoring the state of the main shaft in real time and combining a machine learning optimization strategy, the system automatically adjusts the inclination angle, the air flow and the cooling water flow of the main shaft, and it is ensured that the surface type quality and the flatness meet the requirements in the wafer thinning process. The wafer surface type recognition system and the flatness measurement system measure wafer surface data in real time through a laser range finder, and a surface type three-dimensional model is generated. According to the control method and system, the working state of the main shaft can be automatically adjusted according to real-time data, manual intervention is reduced, the surface type control precision and the machining yield of the wafer are improved, and the control method and system have the advantages of being intelligent, accurate in control and high in reliability.

Description

Technical field:

[0001] The present invention relates to the field of semiconductor wafer packaging thinning, and in particular to a method and system for controlling wafer surface shape and flatness through intelligent spindle adjustment. Background technology:

[0002] With the rapid development of global electronic information technology, the integrated circuit industry has become a key pillar of my country's high-tech development. However, my country's high-end chip manufacturing faces numerous limitations, particularly in high-precision semiconductor equipment. During silicon wafer processing, wafers undergo grinding and thinning after slicing to remove unevenness caused by multiple wire saws. Wafer bonding technology, used in advanced packaging processes in the front-end process, also requires further wafer thinning. During chip packaging, the backside of the wafer is ground to achieve thinner chip packaging. Therefore, wafer thinning is used throughout the entire wafer manufacturing process, with increasingly stringent requirements placed on the wafer's thinned surface and flatness.

[0003] Wafer spin grinding is a widely used wafer thinning technology. During wafer thinning, the wafer to be thinned is vacuum-stuck on a wafer stage and rotates together. The grinding wheel's edge passes through the center of the wafer, achieving semi-contact grinding, reducing grinding heat while ensuring stable grinding forces. A wafer thinning spindle drives a diamond grinding wheel in high-speed rotation, continuously feeding the wafer axially. The spindle is secured within a sleeve via a support and adjustment end, which is guided by the Z-axis lead screw guide of the thinning equipment. During the thinning process, the operator adjusts the spindle's two adjustment ends to control the desired surface profile. The adjustment end features a differential thread, which is rotated to tilt the spindle and control the wafer's surface profile. During wafer spin grinding, wafer surface profile and flatness are two key factors in determining wafer quality. Failure to meet these requirements can negatively impact the chip fabrication process, such as debris, depth of focus, alignment issues, uneven polishing, and uneven deposition. The flatness of the wafer surface will also affect the number of stacking layers of the 3D stacked package, which in turn affects the integration, performance and life of the integrated circuit chip. At present, the thinning spindle of the wafer thinning equipment requires an experienced operator to manually adjust the spindle angle. The working space is small, and repeated grinding, wafer surface measurement and manual adjustment of the spindle angle are required. In addition, this adjustment method cannot adjust the operating state of the spindle to ensure that the wafer has qualified flatness. The operating state of the spindle of the thinning machine will change after a long period of operation, and it cannot be ensured that the spindle is in the best working state. In order to avoid repeated adjustment of the spindle, realize automatic angle adjustment and compensation, and better control the flatness of the wafer after thinning, it is crucial to develop an intelligent control method for wafer surface shape and flatness.

[0004] In the field of existing wafer thinning spindle angle adjustment methods and devices, Beijing CETC Electronic Equipment Co., Ltd. has disclosed an automatic spindle angle adjustment device for semiconductor special equipment (CN105563317A). One end of the adjustment mechanism of the device is connected to the transmission mechanism, and the other end of the adjustment mechanism is fixedly connected to the spindle. The spindle angle is adjusted by controlling the backplane feed through a motor, but the spindle has a single adjustable angle direction, and the adjustment data source only comes from the wafer thickness detected by the measuring instrument. The measuring instrument cannot accurately measure the surface shape and flatness of the wafer, there is no intelligent algorithm to accurately control the adjustment device, and the flatness of the wafer after thinning cannot be controlled; Guangdong Zhaoheng Intelligent Technology Co., Ltd. has applied for a fine-tuning mechanism and method for semiconductor processing precision equipment to adjust the position of the wafer stage; the adjustment device realizes the rotation angle and fine-tuning height of the wafer stage through two adjustment mechanisms respectively; driving mechanism. At the same time, the fine-tuning mechanism uses the transmission of the screw to convert linear motion into rotational motion. The deflection angle of the threaded sleeve and the drive motor achieves the process of converting linear motion into rotational motion. By adjusting the transmission gap between the screw and the threaded seat, zero-gap transmission is achieved, which can largely meet the precision adjustment processing of the semiconductor industry. However, this adjustment method is to adjust the height and inclination of the turntable to control the wafer surface shape, which is not as simple as controlling the inclination of the spindle. In addition, this patent does not achieve intelligence and cannot simultaneously control flatness. Among them, Ningbo Xinfeng Precision Technology Co., Ltd. proposed a grinding spindle adjustment device (CN115194191A), which uses the rotation of two threaded parts with different pitches to achieve precise adjustment. However, the angle between the first and second adjustment mechanisms of the device and the fixed end can only be evenly distributed at 120°. This distribution method cannot eliminate the problem of mutual influence between the spindle adjustment devices. In summary, the above patents for realizing spindle angle adjustment show that the existing patents for surface control of wafer thinning are still limited to manual, non-intelligent and complicated methods, and cannot realize one-click selection of process recipes for wafer surface shape and real-time online detection and feedback of thinning process. At the same time, the existing invention patents only stop at realizing fixed micro-displacement of certain points of the spindle through mechanical structure. The surface shape optimization effect is single, and the surface shape and flatness are not optimized at the same time. The thinning-measurement-adjustment process is continuously cycled through manual detection methods, and the intelligent control process of automatically completing loading-thinning-optimizing surface shape-discharging is not realized. Summary of the invention:

[0005] The purpose of the present invention is to provide a method and system for intelligent control of the surface shape and flatness of thinned wafers, which solves the problems of the prior art in adjusting the spindle tilt angle, such as the small operating space resulting in inconvenience in operation, the inability to accurately control the surface shape and flatness, the inability to automatically adjust the spindle tilt angle according to the required surface shape, and the need for operators to rely on experience to adjust the spindle tilt angle.

[0006] To achieve the above objectives, the present invention provides an intelligent control system for wafer thinning surface shape and flatness, comprising an intelligent air-floating spindle, a wafer surface shape recognition system, a flatness measurement system, an intelligent adjustment algorithm system, a spindle angle adjustment device, a spindle operation status monitoring system, and a spindle air pressure and cooling water flow control device. After wafer grinding is completed, the turntable transfers the wafer to the wafer surface shape recognition system, where an intelligent camera takes a photo and identifies the wafer surface shape, which is then classified. The wafer then enters the flatness measurement system, where the flatness of the wafer grinding is measured to obtain the wafer flatness value. The surface shape type and flatness data are transmitted to the intelligent adjustment algorithm system, which uses an AI recognition algorithm and a spindle operation status recognition algorithm to generate spindle angle adjustment, spindle air pressure, and cooling water flow data. Based on the data generated by the algorithm system, the spindle angle adjustment device begins to adjust the spindle angle to the target value, and the spindle air pressure and cooling water flow control devices begin to adjust the spindle air pressure and cooling water flow valve to the target values. The wafer enters the grinding chamber again and starts grinding according to the new spindle angle value, air pressure value and cooling water flow value. After grinding is completed, surface recognition and flatness measurement are performed again until the process requirements are met.

[0007] The intelligent spindle includes a motor, a differential thread, a water flow control valve, an air flow control valve, a temperature sensor, a vibration sensor, a displacement sensor, and a laser rangefinder. Figure 6 shown.

[0008] The lower end of the main shaft is equipped with two adjustment ends and a fixed end, and the three are evenly distributed on the circumference of the main shaft. The fixed end can also be located above the perpendicular bisector of the two adjustment ends. The upper screw sleeve of the differential thread adjustment end is fastened to the main shaft, and the lower screw sleeve of the adjustment end is fastened to the main shaft fixing plate. The upper screw sleeve of the adjustment end and the lower screw sleeve are connected by a differential screw. There are two sections of threads with different pitches and the same nominal diameter on the differential screw of the adjustment end. The lower end of the differential screw is connected to the motor, which drives the motor to rotate the differential screw, thereby tilting the main shaft, such as Figure 3 The main function of the fixed end is to realize small angle rotation without deformation, and is preferably a ball head connection, but can also be realized by a joint bearing or the like.

[0009] The water flow valve and air flow valve are installed on the water inlet pipe and the air inlet pipe respectively. After receiving the signal, they change the water flow and air supply.

[0010] The temperature sensor and vibration sensor are installed on the outside of the main shaft near the thrust bearing and on the end face of the main shaft respectively, and transmit the signals to the computer.

[0011] The displacement sensor is located above the rotor and measures the floating height of the rotor after the rotor floats.

[0012] The laser rangefinder is located above the wafer and can transmit the distance signal of each point on the wafer to the computer during grinding.

[0013] The spindle monitoring system can detect vibration, temperature, rotor lift, and air supply pressure during spindle operation. A database is created based on the detected data. If the detected values differ from those during normal spindle operation, an alarm is triggered and the cause is intelligently analyzed to assist technicians in troubleshooting, preventing further damage to the spindle and enabling targeted repairs.

[0014] The wafer surface recognition system records and analyzes the position of each point on the wafer after the computer receives the signal from the laser rangefinder, automatically generates a 3D model of the wafer, and then determines the surface shape, thickness and flatness of the wafer.

[0015] The intelligent adjustment algorithm system can be used to directly adjust the wafer thinning spindle to achieve a stable semi-contact grinding position and control the wafer surface shape. This method uses a laser rangefinder to measure the surface shape of the wafer in real time during grinding, and uses the height coordinate value of the adjustment part to replace the coordinates of the actual rotation transformation of the spindle grinding wheel grinding edge. This simplifies the tedious calculation method of rotating and calculating the height of the spindle grinding edge by the traditional method, and provides real-time feedback on the spindle angle during the grinding process. At the same time, this method is also suitable for different distribution situations of the adjustment device and has universal applicability in wafer angle adjustment. At the same time, a new type of adjustment device is proposed. The device uses inner and outer ring differential threads. The coordinates of the grinding wheel are directly obtained through the improved algorithm. The coordinate value is used to control the motor to directly drive the differential thread rotation to control the grinding wheel height adjustment, which greatly improves the angle adjustment accuracy of the equipment. The present invention can achieve precise adjustment of the spindle angle, control the wafer surface shape, and improve the wafer thinning effect. It can take points on the thinned wafer through real-time monitoring to achieve non-contact measurement of the wafer surface shape and real-time monitoring and feedback of the surface shape. It has high reliability, avoids errors caused by manual adjustment, and improves the fragility problem caused by fine threads with small pitches.

[0016] The intelligent control method includes the following steps: Figure 7 shown.

[0017] S1, monitor the spindle operation status in real time through multiple sensors to determine whether the spindle is in normal working condition;

[0018] S2. Adjust the motor rotation angle and change the spindle inclination according to the preset surface target;

[0019] S3, automatically adjust the air flow and cooling water flow according to the surface shape and flatness data of the wafer after thinning;

[0020] S4. Analyze the surface shape of the wafer through the wafer surface recognition system and determine whether the surface shape is qualified;

[0021] S5. When the surface shape is unqualified, the system readjusts the spindle inclination and operating state according to the surface shape error and performs grinding again until the surface shape meets the requirements;

[0022] S6, repeating steps S4 to S5 until the wafer meets the expected surface shape and flatness requirements;

[0023] S7, the machine learning system records the error between the results simulated according to the requirements of surface shape and flatness and the actual grinding results, so as to improve the next adjustment strategy, such as Figure 8 As shown;

[0024] Taking an 8-inch silicon wafer and a thinning grinding wheel as an example, the diameter of the thinning grinding wheel is D1, the diameter of the silicon wafer is D2, and the wafer diameter is approximately the outer edge diameter of the grinding wheel, which is recorded as R. Figure 9 As shown: A Cartesian coordinate system is established with the grinding wheel axis and the axis of the thinned silicon wafer as the Z1 and Z2 axes respectively, and the grinding plane as the XOY plane. X1Y1Z1 and X2OY2Z2 are shown in the figure, and Y1 and Y2 are collinear; during the grinding process, the grinding wheel originally in the X1OY1 plane controls the height at points A, B, and C through the spindle adjustment device to control the silicon wafer to achieve semi-contact grinding, so as to control the surface shape after grinding and reduce the heat generated during the grinding process. A`C` is the outer edge of the grinding wheel after adjusting the angle, and A`B` is the part where the grinding wheel and the wafer are semi-contact ground. a, b, and c are the adjustment displacements at points A, B, and C respectively. In order to simplify the coordinate transformation calculation, point A can be set as a fixed point, and the changing heights of points B and C can be calculated;

[0025] Since the wafer radius R1 = grinding wheel diameter R2 = R, after establishing the geometric model, it can be known that: O1C = O1B = O1A = AB = BC;

[0026] At this time in X1OY1, the coordinates of point A are The coordinates of point B are (0, R, 0), and the coordinates of point C are

[0027] Assume that under ideal conditions, during the spindle angle adjustment process, the rotation axis passes through the origin of the coordinate system X1Y1Z1, and a perpendicular line is drawn through point O1C, intersecting the grinding edge in the X1OY1 plane at point M and intersecting AB at point N. When adjusting the height of point C, points A, B, and C rotate around the line where MN is located to points A', B', and C' respectively. The unit vector with MN as the axis is According to the formula for rotation around any axis in the spatial rectangular coordinate system

[0028]

[0029] Where T 4×4 The rotation matrix R of the three-dimensional coordinate transformation for the first 3 rows and 3 columns 3×3 for:

[0030]

[0031] Where K = 1-cosθ;

[0032]

[0033] It can be obtained that the coordinates of points A', B', and C' after rotation by θ° in the X1Y1Z1 coordinate system are

[0034] At this time, it can be known from the geometric relationship that the curve with a circular radius of R where the grinding edge is located must pass through the three points A', B', and C'. At this time, the unit vector of the straight line where the X1 axis is located for:

[0035]

[0036] If point B is adjusted at this time, the entire arc will be around The straight line rotates, that is, the three points A', B', and C' rotate around Rotation, assuming the rotation angle is α, the rotation matrix R` can be obtained by the same reason 3×3

[0037]

[0038] Where K`=1-cosα.

[0039] The relationship between the coordinates and angles of the rotated points A``, B``, and C`` in the X1Y1Z1 coordinate system can be obtained.

[0040] At this time, due to the existence of the assumption, the coordinates of the center of the circle determined by the three points A``, B``, and C`` are still (0, 0, 0), and the curve is the surface determined by the three points.

[0041] AX+DY+CZ+D=0 and the cross-sectional shape of a sphere with the origin of the coordinate axes X1Y1Z1 and a radius R.

[0042] Use vector notation to represent the plane where points A``, B``, and C`` are located. Taking A`` as the base point, find Two vectors, the normal vector of the plane equation can be calculated by geometric relationship for:

[0043]

[0044] At this time, the curve equation A1 with a radius R passing through points A``, B``, and C`` is: Where M, N, and P are normal vectors The coordinate components of .

[0045] The current curve equation is in the X1Y1Z1 coordinate system. Through displacement transformation, the curve can be converted into the X2Y2Z2 coordinate system to obtain equation 2. The actual wafer surface shape is obtained by rotating equation A1 around the Z2 axis in the coordinate system X2Y2Z2. As can be seen from the figure, A2 is obtained by translation transformation of A1. A1 needs to be matrix transformed into the coordinate system X2Y2Z2 to obtain the equation of A2 as follows:

[0046]

[0047] At this time, curve 2 rotates around the Z2 axis, and the surface equation swept by the arc is the surface equation of wafer grinding under ideal conditions.

[0048] Assuming that point A is fixed at this time, the Z-axis coordinate values of points B and C are the actual adjusted heights. By controlling the range of coordinate values, the surface equations swept by the grinding arc under different adjustment heights can be obtained.

[0049] This rotational transformation process yields the relationship between the adjustment angle and the corresponding surface height curve. The relationship between the angle and the height of the corresponding adjustment end can then be used to control the wafer surface shape in real time.

[0050] Therefore, the present invention adopts a thinned wafer surface shape and flatness intelligent control method and system of the above structure, which has the following beneficial effects:

[0051] (1) Improve wafer thinning accuracy. Using an intelligent air-floating spindle and a high-precision adjustment device, the spindle inclination is adjusted through differential threads to achieve precise control of the wafer surface shape and flatness, reducing the error of traditional manual adjustment and improving processing accuracy.

[0052] (2) Automated adjustment reduces manual intervention. Traditional wafer thinning requires repeated manual measurement and adjustment. However, the present invention uses intelligent adjustment algorithms and machine learning systems to analyze wafer surface data in real time and automatically adjust the spindle tilt angle, air supply pressure, and cooling water flow, significantly reducing the complexity of manual operations and improving production efficiency.

[0053] (3) Real-time monitoring to improve stability. A multi-sensor system (temperature, vibration, displacement, air pressure, etc.) is used to monitor the spindle operating status in real time, and compared and analyzed in combination with the database to ensure that the equipment is always in the best working condition, reduce quality problems caused by changes in the spindle status, and improve the stability and life of the equipment operation.

[0054] (4) Intelligent learning and adaptive optimization. Through machine learning algorithms, the system can record the surface error of each grinding, establish a data model, and continuously optimize the adjustment strategy, thereby continuously improving the accuracy and consistency of the grinding process and achieving intelligent and adaptive process optimization.

[0055] (5) Improve semiconductor production efficiency. Since the system can automatically optimize the spindle angle and working state, it reduces the time loss caused by manual adjustment while ensuring high-precision processing, thereby improving the processing efficiency of semiconductor wafers and meeting the needs of large-scale production.

[0056] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. Description of the drawings:

[0057] Figure 1 1 is a left-view structural diagram of an embodiment of the present invention, Figure 2 1 is a schematic diagram of a top view of the structure of an embodiment of the present invention, Figure 3 1 is a cross-sectional view of the structure of the fixed end and the adjustment end of an embodiment of the present invention, Figure 4 1 is a schematic structural diagram of the fixed end of the spindle adjustment device according to an embodiment of the present invention. Figure 5 1 is a schematic structural diagram of the adjusting end of the spindle adjusting device according to an embodiment of the present invention. Figure 6 This is a schematic diagram of the differential screw at the adjusting end of the spindle adjustment device. Figure 7 This is a driving method of the automatic input device of the regulating end according to an embodiment of the present invention. Figure 8 A flowchart for machine learning based on intelligent algorithms. Figure 9 This is a schematic diagram of the geometric model for adjusting the wafer spindle angle of the present invention, showing the coordinate positions and rotational relationship of points A, B, and C.

[0058] Reference numerals

[0059] 1: Spindle fixing plate 2: Thinning spindle 3: Thinning grinding wheel 4: Spindle angle adjustment end 5: Spindle fixing end 6: Wafer 7: Wafer stage 8: Non-contact measuring instrument;

[0060] 201 Flow sensor 202 Air pressure sensor 203 Vibration sensor 204 Temperature sensor 205 Flow controller 206 Air pressure controller 207 Displacement sensor

[0061] 401: Adjustment end 402: Adjustment end 2

[0062] 5: Fixed end 501: Fixed end screw sleeve 502: Fixed end ball head 503: Fixed end washer sleeve 504 washer 505 fixing screw

[0063] 701: Ceramic suction cup 702: Porous ceramic

[0064] 40101: Upper screw sleeve at the adjustment end 40102: Lower screw sleeve at the adjustment end 40103: Locking nut at the adjustment end 40104: Differential screw at the adjustment end 40105: Differential thread zone 1 40106: Differential thread zone 2 40107: Drive motor Specific implementation method:

[0065] The present invention will be further described below in conjunction with the accompanying drawings of the embodiments:

[0066] Unless otherwise defined, the technical or scientific terms used in the present invention shall have the usual meanings understood by persons of ordinary skill in the field to which the present invention belongs. The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprise" mean that the elements or objects preceding the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0067] Example

[0068] See also Figure 1 、 Figure 2 、 Figure 3 and Figure 5 , the spindle angle adjustment end 4 and the spindle angle fixing end 5 are installed on the spindle fixing plate 1, which is used to support the thinning spindle 2. The thinning grinding wheel 3 is installed below the thinning spindle and rotates at high speed with the spindle. The wafer stage 7 is located below the thinning spindle as a whole, and the center of the wafer stage is directly below the thinning spindle grinding wheel. Before wafer thinning, the wafer 6 is adsorbed in the center position by the porous ceramic 702 in the ceramic suction cup 701 on the wafer stage 7 and rotates with the wafer stage 7. The 201 water flow sensor and the 205 water flow control valve are located on the water inlet pipe, the 206 air flow control valve and the 202 barometer are located on the air line, the 203 vibration sensor is installed on the spindle end face and placed perpendicular to each other, and the 204 temperature sensor is located on the outside of the upper and lower main thrust bearings. The 207 displacement sensor is located above the rotor.

[0069] Wafer surface recognition system: uses a laser rangefinder to perform real-time measurement to obtain the distance signal of each point on the wafer surface, and then generates a three-dimensional model of the wafer surface and calculates the wafer surface shape.

[0070] Flatness measurement system: Obtain wafer flatness data through a focusing rangefinder, measure wafer thickness changes and generate data.

[0071] Intelligent adjustment algorithm system: includes machine learning algorithms, processes surface shape and flatness data in real time, and provides adjustment instructions for spindle angle, air flow, and cooling water flow.

[0072] Spindle angle adjustment device: adjusts the spindle inclination according to the instructions output by the intelligent adjustment algorithm system to ensure the conformity of the wafer surface.

[0073] Spindle operation status monitoring system: monitors the spindle's temperature, vibration, displacement and air pressure in real time, and ensures the spindle operates in the best condition by comparing with the normal values in the database.

[0074] Wafer Fixing and Preparation: First, the wafer to be thinned is placed on the wafer stage and secured with ceramic suction cups. The wafer stage rotates with the spindle to ensure the wafer remains stable during the thinning process.

[0075] Spindle Angle Adjustment: Based on the preset wafer surface target, the intelligent air-bearing spindle's motor, driven by a differential screw drive, adjusts the spindle's inclination angle to bring the wafer's grinding surface closer to the desired profile. The intelligent air-bearing spindle regulates air pressure via an air flow control valve to ensure optimal spindle operation.

[0076] Wafer Surface Recognition and Flatness Measurement: During the grinding process, the wafer surface recognition system uses a laser rangefinder to measure wafer surface data in real time and generate a 3D surface model. The flatness measurement system then calculates the wafer's flatness. Based on the measurement results, an intelligent adjustment algorithm calculates the parameters that need to be adjusted and controls the water and air flow control valves to adjust the cooling water and air flow rates.

[0077] Intelligent Control and Feedback Adjustment: When the wafer's surface shape or flatness doesn't meet requirements, the system uses algorithms to optimize adjustment strategies based on real-time measurement data, adjusting the spindle angle, airflow, and cooling water flow rates to ensure the wafer meets the desired surface shape and flatness. The machine learning system continuously optimizes adjustment strategies and improves grinding accuracy by recording errors during each grinding pass.

[0078] Repeat grinding until the wafer meets the requirements: When the wafer surface shape and flatness meet the requirements, the grinding process ends. If the wafer still fails, the system will automatically return the wafer to the grinding position and repeat the adjustment and grinding until the surface shape and flatness fully meet the process standards.

[0079] Through the above method, this embodiment can precisely control the surface shape and flatness during the wafer thinning process, reducing the errors associated with traditional manual adjustments, improving production efficiency and wafer quality stability. The intelligent adjustment system provides real-time feedback on parameter changes during the grinding process. Through continuous learning and optimization, it further improves the accuracy of surface shape control and reduces the need for manual intervention. Furthermore, the machine learning system can optimize adjustment strategies based on historical data, reducing production issues caused by equipment failure or environmental changes.

Claims

1. An intelligent control system for thinned wafer surface shape and flatness, characterized in that: include: Intelligent air-floating spindle, which includes the air-floating spindle and the motor, differential thread, air flow control valve, water flow control valve, temperature sensor, vibration sensor, displacement sensor, etc. Wafer surface recognition system, used to obtain wafer surface data through a laser rangefinder and generate a three-dimensional model of the wafer surface; The flatness measurement system calculates the flatness of the wafer based on the generated 3D model of the wafer surface and generates relevant data; Intelligent adjustment algorithm system automatically adjusts the spindle working state according to wafer surface shape and flatness data, including motor rotation angle, spindle inclination, air pressure and cooling water flow; Spindle angle adjustment device, used to adjust the spindle's inclination to optimize the wafer surface shape; The spindle operation status monitoring system is used to monitor the spindle's temperature, vibration, displacement, air pressure, etc. in real time to ensure that the spindle is operating in normal condition; The intelligent air-floating spindle precisely controls the inclination angle through the spindle angle adjustment device, ensuring that the surface shape of the wafer can be adjusted during the thinning process. The wafer surface recognition and flatness measurement systems collect data in real time, and an intelligent adjustment algorithm analyzes and calculates optimization strategies. The spindle status monitoring system uses multiple sensors to detect spindle operation and combines flow control to ensure stability, thereby achieving high-precision thinning and improving wafer surface quality and flatness.

2. The thinned wafer surface shape and flatness intelligent control system according to claim 1, characterized in that: The lower end of the intelligent air-floating main shaft is equipped with two adjusting ends and a fixed end, and the adjusting ends can adjust the main shaft inclination angle through differential threads.

3. The thinned wafer surface shape and flatness intelligent control system according to claim 1, characterized in that: The motor drives the differential screw to rotate, thereby adjusting the spindle inclination angle. The air flow control valve and the water flow control valve respectively adjust the air supply pressure and cooling water flow of the spindle to meet the grinding process requirements.

4. The thinned wafer surface shape and flatness intelligent control system according to claim 1, characterized in that: The wafer surface recognition system acquires wafer surface data in real time through a laser rangefinder and generates a three-dimensional surface model to calculate the wafer surface shape.

5. The thinned wafer surface shape and flatness intelligent control system according to claim 1, characterized in that: The flatness measurement system includes a focusing rangefinder for measuring the flatness of a wafer and outputting relevant data.

6. A method for intelligently controlling the surface shape and flatness of a thinned wafer, applied to the system according to any one of claims 1 to 5, characterized in that: The following steps are involved: S1, monitor the spindle operation status in real time through multiple sensors to determine whether the spindle is in normal working condition; S2. Adjust the motor rotation angle and change the spindle inclination according to the preset surface target; S3, automatically adjust the air flow and cooling water flow according to the surface shape and flatness data of the wafer after thinning; S4. Analyze the surface shape of the wafer through the wafer surface recognition system and determine whether the surface shape is qualified; S5. When the surface shape is unqualified, the system readjusts the spindle inclination and operating state according to the surface shape error and performs grinding again until the surface shape meets the requirements; S6, repeating steps S4 to S5 until the wafer meets the expected surface shape and flatness requirements; S7. The machine learning system records the error between the simulated results based on the surface shape and flatness requirements and the actual grinding results, so as to improve the next adjustment strategy; 7. The method for intelligent control of thinned wafer surface shape and flatness according to claim 6, characterized in that: The intelligent adjustment algorithm records the surface shape and flatness errors of each grinding through machine learning, and optimizes the adjustment strategy to improve grinding accuracy.

8. The method for intelligent control of thinned wafer surface shape and flatness according to claim 6, characterized in that: Step S1 further includes: determining whether the spindle is in a normal working state by detecting the temperature, vibration, displacement and air supply pressure of the spindle.

9. The method for intelligent control of thinned wafer surface shape and flatness according to claim 6, characterized in that: Step S4 further includes: obtaining distance data of each point on the wafer surface by a laser rangefinder, generating a three-dimensional model of the wafer surface, and calculating the flatness of the wafer.

10. The method for intelligent control of thinned wafer surface shape and flatness according to claim 6, characterized in that: Step S5 further includes: when the wafer surface shape is unqualified, the intelligent system calculates and adjusts the angle and operating state of the spindle, and adjusts the air flow control valve and the water flow control valve to make the wafer meet the expected surface shape and flatness requirements. Constructing a mathematical model: Taking an 8-inch silicon wafer and a thinning grinding wheel as an example, the diameter of the thinning grinding wheel is D1, the diameter of the silicon wafer is D2, and the wafer diameter is approximately the diameter of the outer edge of the grinding wheel, which is recorded as R. Taking an 8-inch silicon wafer and a thinning grinding wheel as an example, the diameter of the thinning grinding wheel is D1, the diameter of the silicon wafer is D2, and the wafer diameter is approximately the diameter of the outer edge of the grinding wheel, which is recorded as R. The axis of the grinding wheel and the axis of the thinned silicon wafer are defined as the Z1 axis and the Z2 axis respectively, the grinding plane is the XOY plane, and the Cartesian coordinate systems X1Y1Z1 and X2Y2Z2 are established, where Y1 and Y2 are collinear. During the grinding process, the grinding wheel controls the heights of points A, B, and C through the spindle adjustment device to achieve semi-contact grinding, where the coordinates of point A are The coordinates of point B are (0, R, 0), and the coordinates of point C are A Cartesian coordinate system is established with the grinding wheel axis and the axis of the thinned silicon wafer as the Z1 and Z2 axes respectively, and the grinding plane as the XOY plane. X1Y1Z1 and X2OY2Z2 are shown in the figure, and Y1 and Y2 are collinear. During the grinding process, the grinding wheel originally in the X1OY1 plane controls the height at points A, B, and C through the spindle adjustment device to control the silicon wafer to achieve semi-contact grinding, so as to control the surface shape after grinding and reduce the heat generated during the grinding process. A`C` is the outer edge of the grinding wheel after adjusting the angle, and A`B` is the part where the grinding wheel and the wafer are semi-contact ground. a, b, and c are the adjustment displacements at points A, B, and C respectively. To simplify the coordinate transformation calculation, point A can be set as a fixed point, and the changing heights of points B and C can be calculated. Since the wafer radius R1 = grinding wheel diameter R2 = R, after establishing the geometric model, we can know that: O1C=O1B=O1A=AB=BC; At this time in X1OY1, the coordinates of point A are The coordinates of point B are (0, R, 0), and the coordinates of point C are Assume that under ideal conditions, during the spindle angle adjustment process, the rotation axis passes through the origin of the coordinate system X1Y1Z1, and a perpendicular line is drawn through point O1C, intersecting the grinding edge in the X1OY1 plane at point M and intersecting AB at point N. When adjusting the height of point C, points A, B, and C rotate around the line where MN is located to points A', B', and C' respectively. The unit vector with MN as the axis is According to the formula for rotation around any axis in the spatial rectangular coordinate system Where T 4×4 The rotation matrix R of the three-dimensional coordinate transformation for the first 3 rows and 3 columns 3×3 for: Where K = 1-cosθ; It can be obtained that the coordinates of points A', B', and C' after rotation by θ° in the X1Y1Z1 coordinate system are At this time, it can be known from the geometric relationship that the curve with a circular radius of R where the grinding edge is located must pass through the three points A', B', and C'. At this time, the unit vector of the straight line where the X1 axis is located for: If point B is adjusted at this time, the entire arc will be around The straight line rotates, that is, the three points A', B', and C' rotate around Rotation, assuming the rotation angle is α, the rotation matrix R` can be obtained by the same reason 3×3 Where K`=1-cosα. The relationship between the coordinates and angles of the rotated points A``, B``, and C`` in the X1Y1Z1 coordinate system can be obtained. At this time, due to the existence of the assumption, the coordinates of the center of the circle determined by the three points A``, B``, and C`` are still (0, 0, 0), and the curve is the surface determined by the three points. AX+BY+CZ+D=0 and the cross-sectional shape of a sphere with the origin of the coordinate axes X1Y1Z1 and a radius R. Use vector notation to represent the plane where points A``, B``, and C`` are located. Taking A`` as the base point, find Two vectors, the normal vector of the plane equation can be calculated by geometric relationship for: At this time, the curve equation A1 with a radius R passing through points A``, B``, and C`` is: Where M, N, and P are normal vectors The coordinate components of . The current curve equation is in the X1Y1Z1 coordinate system. Through displacement transformation, the curve can be converted into the X2Y2Z2 coordinate system to obtain equation 2. The actual wafer surface shape is obtained by rotating equation A1 around the Z2 axis in the coordinate system X2Y2Z2. As can be seen from the figure, A2 is obtained by translation transformation of A1. A1 needs to be matrix transformed into the coordinate system X2Y2Z2 to obtain the equation of A2 as follows: At this time, curve 2 rotates around the Z2 axis, and the surface equation swept by the arc is the surface equation of wafer grinding under ideal conditions. Assuming that point A is fixed at this time, the Z-axis coordinate values of points B and C are the actual adjusted heights. By controlling the range of coordinate values, the surface equations swept by the grinding arc under different adjustment heights can be obtained. This rotational transformation process yields the relationship between the adjustment angle and the corresponding surface height curve. The relationship between the angle and the height of the corresponding adjustment end can then be used to control the wafer surface shape in real time.

Images