A self-adaptive adjustment method for warpage in wafer transfer process

Abstract

The application discloses a kind of self-adapting adjustment methods of wafer warping in wafer transmission process, specifically relates to wafer warping self-adapting adjustment technical field, and through the non-contact pneumatic sensing array of circumferential distribution, airflow back pressure and edge occlusion coordinates are collected, to avoid the measurement failure caused by reflection and transmission;According to the constructed edge flow field disturbance distribution diagram, high-resolution extraction of local warping amplitude is realized;Through calculating edge contact risk coefficient and dynamic fitting trend, the space selective adsorption mechanism from risk identification to partition pressure compensation is established;Through extracting each partition pressure tracking error to obtain attitude compensation amount, the collaborative optimization of adsorption force control and manipulator motion control is realized, so that manipulator can adaptively follow wafer warping form to carry out trajectory correction.

Description

Technical Field

[0001] This invention relates to the field of adaptive adjustment of wafer warpage technology, and more specifically, to an adaptive adjustment method for warpage during wafer transport. Background Technology

[0002] In the semiconductor manufacturing field, with the increase in the number of chip stacking layers and the widespread adoption of wafer thinning processes, wafer warpage has become a core challenge affecting yield and safety during transport. To address this issue, the industry has proposed an adaptive adjustment method based on pre-scanning. This method uses a laser displacement sensor to perform a non-contact scan of the wafer surface before the robotic arm picks up the wafer to obtain its warpage distribution, thereby dynamically planning the wafer picking posture and adjusting the multi-zone adsorption force. However, in practical applications, this method suffers from severe failures when the wafer being measured is a polished bare silicon wafer, a mirror-finished wafer after chemical mechanical polishing, or a transparent substrate such as quartz glass or sapphire. Specifically, extremely smooth surfaces cause specular reflection of the laser, preventing the reflected light from returning to the receiver along its original path; while transparent materials allow the laser beam to be directly transmitted, mistakenly measuring the support platform or the robotic arm itself on the back of the wafer. Both of these situations result in a large number of flying points or invalid signals in the acquired data, directly causing the failure of warpage modeling. Because the true spatial posture of the wafer cannot be obtained, subsequent robotic arm trajectory correction decisions lose their basis, and multi-zone pressure control cannot be executed due to the lack of a target. While theoretical alternatives such as confocal sensors or white light interferometers have the capability to overcome the measurement challenges of such materials, their high cost, large physical size, and limited scanning speed make them difficult to engineer and deploy in compact, high-speed front-end modules of equipment. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an adaptive adjustment method for warpage during wafer transfer, thereby addressing the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: An adaptive adjustment method for warpage during wafer transfer includes the following steps: The airflow back pressure value and the coordinates of the occlusion boundary of multiple sampling points at the edge of the wafer are obtained. The gap leakage rate between the wafer and the end effector is collected by a non-contact pneumatic sensor array distributed along the circumference. The edge flow field disturbance distribution map is constructed by combining the robot insertion depth and the nominal radius of the wafer. Based on the edge flow field disturbance distribution map, the local warpage amplitude of the wafer edge is extracted, and the edge contact risk coefficient is calculated based on the local warpage amplitude and the preset safe adsorption gap. Based on the edge contact risk coefficient and the horizontal feed speed of the robot arm, the dynamic bonding trend of the wafer surface relative to the adsorption surface of the end effector is determined, and the pressure compensation weight required for each adsorption zone is obtained by combining the real-time adsorption force of each adsorption zone. The predicted amount of suction adjustment for each adsorption zone is calculated based on the pressure compensation weight and the current adsorption force of the corresponding zone, so as to obtain the target pressure setting value of each adsorption zone in the next control cycle. Based on the deviation sequence between the target pressure setpoint and the actual pressure feedback value, the pressure tracking error of each zone is extracted to obtain the attitude compensation amount used to correct the robot arm's pick-up height. The trajectory of the robotic arm for picking up the piece is dynamically corrected based on the posture compensation amount.

[0005] In a preferred embodiment, the process of acquiring the airflow back pressure value and the coordinates of the obstruction boundary at multiple sampling points at the wafer edge, collecting the gap leakage rate between the wafer and the end effector through a non-contact pneumatic sensor array distributed along the circumference, and constructing an edge flow field disturbance distribution map by combining the robot insertion depth and the nominal radius of the wafer is as follows: Multiple miniature pneumatic sensing units are arranged at equal intervals along the circumference at the front end of the end effector. Each sensing unit includes a pressure sensing element and an edge detection element. During the process of the robotic arm inserting into the wafer at a preset speed, each sensing unit is triggered to collect data synchronously at a fixed sampling frequency to obtain the back pressure value and occlusion boundary coordinates at each circumferential position at each sampling moment. Based on the current insertion depth of the robotic arm and the nominal radius of the wafer, the occlusion boundary coordinates measured by each sensing unit are converted into the angular position of the wafer edge relative to the center of the end effector, and a sampling point mapping in polar coordinate system is established. And by using the pre-calibrated back pressure-gap relationship curve, the actual vertical gap between the lower surface of the wafer and the upper surface of the end effector at the sampling point is read from the back pressure value; By combining the gap values ​​of all sampling points at the same time and their corresponding angular positions, the discrete gap distribution of the wafer edge at that time can be obtained. Interpolation methods are used to fit discrete points to obtain the gap distribution function along the wafer circumference; and the deviation between the back pressure value and the theoretical reference back pressure is defined as the flow field disturbance intensity. By combining the insertion depth of the robotic arm, the flow field disturbance intensity and gap distribution are integrated to construct an edge flow field disturbance distribution map that includes spatial and temporal dimensions.

[0006] In a preferred embodiment, the process of extracting the local warpage amplitude of the wafer edge based on the edge flow field disturbance distribution map, and calculating the edge contact risk coefficient based on the local warpage amplitude and the preset safe adsorption gap is as follows: Extracting the gap distribution function from the edge flow field disturbance distribution map and the gap distribution function The minimum value is less than the preset contact threshold for the first time. The time marked as the critical time ; Gap distribution Converted to local warpage amplitude at the wafer edge The conversion formula is: ,in This represents the theoretical gap distribution for an ideally flat wafer at the same insertion depth. Based on the local warping amplitude and the preset safe adsorption gap Calculate the edge contact risk coefficient The calculation formula is: .

[0007] In a preferred embodiment, the process of determining the dynamic bonding trend of the wafer surface relative to the end effector's adsorption surface based on the edge contact risk coefficient and the horizontal feed speed of the robotic arm, and combining this with the real-time adsorption force of each adsorption zone to obtain the required pressure compensation weight for each zone is as follows: Based on the edge contact risk coefficient and the current horizontal feed speed of the robotic arm, obtain the dynamic fitting trend function; The current adsorption force of each adsorption zone is collected in real time and compared with the preset benchmark adsorption force to obtain the real-time adsorption force deviation of each zone. The dynamic bonding trend function is averaged over the circumferential range corresponding to each adsorption zone to obtain the value for each zone. The tightness of the fit ; and combined with partitioning Real-time adsorption force deviation Calculate partitions Required pressure compensation weight : ,in , These are preset weighting coefficients for the ratio of adhesion tightness, adsorption force deviation, and maximum allowable adsorption pressure, respectively. This represents the maximum permissible adsorption pressure.

[0008] In a preferred embodiment, the process of calculating the predicted suction adjustment amount for each adsorption zone based on the pressure compensation weight and the current adsorption force of the corresponding zone, and obtaining the target pressure setpoint for each adsorption zone in the next control cycle, is as follows: According to each adsorption zone Pressure compensation weight within the current control cycle and its current adsorption force Determine the adsorption partition Basic pressure regulation ; The basic pressure adjustment amount is calculated based on the product of the compensation weight and the maximum allowable adjustment range: ,in The preset maximum pressure adjustment step size for a single cycle; Combined with the current adsorption force Adsorption force with reference Adsorption force deviation Correct the adjustment amount: ; Calculate the target pressure setpoint for the next control cycle t+1. : ; The target pressure setpoint is compared with the system's allowable upper and lower pressure limits. If the calculated result exceeds the upper limit, the upper limit value is used; if it is lower than the lower limit, the lower limit value is used. ,in and These are the minimum sustained adsorption force and the maximum allowable adsorption pressure for this partition, respectively. The sequence of target pressure setpoints required for each adsorption zone in the next control cycle is obtained. ,in This represents the total number of partitions.

[0009] In a preferred embodiment, the process of extracting the pressure tracking error of each zone based on the deviation sequence between the target pressure setpoint and the actual pressure feedback value, and obtaining the attitude compensation amount used to correct the robot arm's plate-picking height, is as follows: During the current control cycle, collect the actual pressure feedback values ​​of each adsorption zone k. and the target pressure setpoint Compare and calculate the pressure tracking error of each partition. : ; Arrange the tracking errors of all partitions in order to form a pressure tracking error sequence. ; Feature extraction is performed on the pressure tracking error sequence, and the average value and gradient distribution characteristics of the error in each region of the pressure tracking error sequence are calculated. Based on the error distribution characteristics, the attitude compensation amount for the robot arm's plate-picking height is determined. : ,in The average tracking error of the pressure tracking error sequence. This represents the maximum value of the error gradient in the pressure tracking error sequence. , These are the preset weighting coefficients for the average tracking error and the maximum error gradient, respectively.

[0010] The technical effects and advantages of this invention are as follows: 1. This invention collects airflow back pressure and edge occlusion coordinates using a circumferentially distributed non-contact pneumatic sensor array. Utilizing the physical property that airflow is insensitive to material optical properties, it fundamentally avoids measurement failures caused by reflection and transmission, ensuring stable acquisition of wafer attitude data even under extreme conditions such as polished bare silicon wafers and quartz glass. Based on the constructed edge flow field disturbance distribution map, the wafer warpage morphology, which is difficult to measure directly, is converted into quantifiable flow field parameters for inversion analysis. High-resolution extraction of local warpage amplitude and overall bending morphology is achieved while maintaining a compact front-end module space. By calculating the edge contact risk coefficient and dynamic bonding trend, a spatially selective adsorption mechanism is established, from risk identification to partition pressure compensation. This mechanism actively enhances adsorption force in severely warped areas and maintains baseline pressure in well-bonded areas, ensuring reliable pickup while avoiding additional stress. By extracting the pressure tracking error of each partition to obtain attitude compensation, the collaborative optimization of adsorption force control and robot motion control is achieved. This allows the robot to adaptively follow the wafer warpage morphology for trajectory correction, maintaining optimal contact posture throughout the pickup process. Attached Figure Description

[0011] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings; Figure 1 This is a flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0012] 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.

[0013] Example: Figure 1 The present invention provides an adaptive adjustment method for warpage during wafer transfer, comprising the following steps: The airflow back pressure value and the coordinates of the occlusion boundary of multiple sampling points at the edge of the wafer are obtained. The gap leakage rate between the wafer and the end effector is collected by a non-contact pneumatic sensor array distributed along the circumference. The edge flow field disturbance distribution map is constructed by combining the robot insertion depth and the nominal radius of the wafer. Based on the edge flow field disturbance distribution map, the spatial bending trend of the wafer in the free state is analyzed, the local warping amplitude of the wafer edge is extracted, and the edge contact risk coefficient is calculated based on the local warping amplitude and the preset safe adsorption gap. Based on the edge contact risk coefficient and the horizontal feed speed of the robot arm, the dynamic bonding trend of the wafer surface relative to the adsorption surface of the end effector is determined, and the pressure compensation weight required for each adsorption zone is obtained by combining the real-time adsorption force of each adsorption zone. The predicted amount of suction adjustment for each adsorption zone is calculated based on the pressure compensation weight and the current adsorption force of the corresponding zone, so as to obtain the target pressure setting value of each adsorption zone in the next control cycle. Based on the deviation sequence between the target pressure setpoint and the actual pressure feedback value, the pressure tracking error of each zone is extracted to obtain the attitude compensation amount used to correct the robot arm's pick-up height. The robot arm's wafer-picking trajectory is dynamically corrected based on the posture compensation amount to achieve stable transport of warped wafers.

[0014] In this embodiment of the invention, the process of acquiring the airflow back pressure value and the coordinates of the obstruction boundary at multiple sampling points at the wafer edge, collecting the gap leakage rate between the wafer and the end effector through a non-contact pneumatic sensor array distributed along the circumference, and constructing an edge flow field disturbance distribution map by combining the robot insertion depth and the nominal radius of the wafer is as follows: Multiple miniature pneumatic sensing units are arranged at equal intervals along the circumference at the front end of the end effector. Each sensing unit includes a pressure sensing element and an edge detection element, which are used to measure the back pressure value of the airflow under the wafer edge and the position coordinates of the wafer edge relative to the sensing unit in real time. It should be noted that the terminator refers to the end effector installed at the end of the robotic arm for direct contact and carrying of the wafer. It is the hand of the robotic arm and is responsible for physically picking up and fixing the wafer during wafer picking, placing and transport. Common end effector types include vacuum suction type (fixing the wafer by negative pressure suction cup), Bernoulli non-contact type (using airflow to suspend the wafer) and edge clamping type, etc. The sensing unit and the robot motion controller are connected to the same time synchronization system to ensure the time consistency of the sampling data of each channel; As the robotic arm inserts itself beneath the wafer at a preset speed, each sensing unit is triggered to synchronously collect data at a fixed sampling frequency, obtaining the back pressure value at each circumferential position at each sampling moment. and occlusion boundary coordinates ,in Indicates the sensor unit number, Indicates the sampling time; Based on the current insertion depth of the robotic arm and wafer nominal radius The coordinates of the occlusion boundary measured by each sensing unit Converted to the angular position of the wafer edge relative to the center of the edge pickup. The process of establishing the sampling point mapping in polar coordinates is as follows: Establish a system with the end effector center as the origin and the robot insertion direction as... In a Cartesian coordinate system along the positive axis, the mounting positions of each sensing unit are predetermined based on the wafer size and measurement requirements during the end-effector design and manufacturing phase. Therefore, the... The fixed installation coordinates of each sensing unit in the coordinate system Given the design parameters, where This indicates the offset of the sensing unit from the center of the end effector in the direction perpendicular to the insertion direction, in millimeters; For example, with the center of the end effector as the symmetrical point, a sensing unit is arranged on each of the left and right sides at a distance of 15mm, 30mm, and 45mm from the center, respectively. Then the corresponding... The values ​​are -45, -30, -15, 15, 30, and 45, respectively. During the insertion of the robotic arm, the coordinates of the occlusion boundary are measured in real time by the edge detection element. The actual wafer edge in this coordinate system Axis coordinates, denoted as The measured value is related to the fixed installation position of the sensing unit. The relationship is: ,in For the wafer edge relative to the center of the sensing unit at The offset in the direction is directly output by the edge detection element (such as a CCD linear array or a PSD position-sensitive detector); Based on the current insertion depth of the robotic arm and wafer nominal radius According to geometric relationships, the wafer edge point is located at... With center and radius as On the circle, its The coordinates satisfy: The negative sign corresponds to the edge closest to the end effector, and the polar angle of that edge point relative to the center of the end effector is... for: ; This converts the linear coordinates of each sampling point into angular positions in polar coordinates. And through a pre-calibrated back pressure-gap relationship curve, the back pressure value... Read the actual vertical gap between the lower surface of the wafer and the upper surface of the end effector at this sampling point. This gap reflects the degree to which airflow leaks from the edge, i.e., the gap leakage rate; Interval values ​​of all sampling points at the same time and their corresponding angular positions By combining these elements, the discrete gap distribution at the wafer edge at that moment can be obtained. Interpolation methods (such as cubic spline interpolation) are used to fit discrete points to obtain the gap distribution function along the wafer circumference. ; and compare the back pressure value with the theoretical reference back pressure. The deviation of the free airflow back pressure (without wafer shielding) is defined as the flow field disturbance intensity. : ,in The continuous back pressure distribution obtained by interpolation; Combined with the insertion depth of the robotic arm By fusing the flow field disturbance intensity with the gap distribution, an edge flow field disturbance distribution map containing both spatial and temporal dimensions is constructed. : It is used to comprehensively describe the aerodynamic characteristics of the wafer edge at different circumferential positions and insertion stages; It should be noted that the edge flow field disturbance distribution map is not a simple set of measurement data, but a continuous description obtained through sensor mapping, physical inversion and interpolation fitting. Its physical significance lies in converting the spatial attitude information of the wafer edge into quantifiable flow field parameters, thereby providing high-resolution input data for subsequent back-calculation of the overall wafer warpage morphology.

[0015] In this embodiment of the invention, the process of analyzing the spatial bending trend of the wafer in a free state based on the edge flow field disturbance distribution map, extracting the local warpage amplitude of the wafer edge, and calculating the edge contact risk coefficient based on the local warpage amplitude and the preset safe adsorption gap is as follows: Extracting the gap distribution function from the edge flow field disturbance distribution map and the gap distribution function The minimum value is less than the preset contact threshold for the first time. The time marked as the critical time ; The critical moment This refers to the moment during the insertion process of the robotic arm when the wafer is about to contact the edge sensor but has not yet undergone significant deformation; the critical moment. Previously, the wafer was in a free-floating state, and its edge gap distribution It directly reflects the actual spatial height of the wafer at that circumferential position, that is, the degree of natural drooping or upward tilt of the wafer edge relative to the end-effector reference plane; Gap distribution Converted to local warpage amplitude at the wafer edge The conversion formula is: ,in The theoretical gap distribution corresponding to an ideal flat wafer at the same insertion depth can be obtained through pre-calibration. The local warpage amplitude is defined as follows: a positive value indicates that the wafer warps upwards at that location, and a negative value indicates that it is concave downwards. Based on the local warping amplitude and the preset safe adsorption gap Calculate the edge contact risk coefficient The calculation formula is: ; It should be noted that the preset safety adsorption gap The minimum wafer-end-pickup gap required to ensure reliable adsorption; the edge contact risk coefficient characterizes the risk of abnormal contact between the wafer edge and the end-pickup due to excessive warping at the current position. The larger the edge contact risk coefficient, the higher the contact risk. It is not a simple result of gap comparison, but a quantitative indicator that comprehensively considers warping amplitude, safety threshold and circumferential distribution.

[0016] In this embodiment of the invention, the process of determining the dynamic bonding trend of the wafer surface relative to the end effector adsorption surface based on the edge contact risk coefficient and the horizontal feed speed of the robotic arm, and combining this with the real-time adsorption force of each adsorption zone to obtain the pressure compensation weight required for each zone is as follows: Based on the edge contact risk coefficient and the current horizontal feed speed of the robotic arm This study analyzes the relative motion between the wafer surface and the end-effector's adsorption surface during insertion. Due to wafer warpage, different circumferential positions of the wafer approach the adsorption surface at different times and speeds as the robotic arm continues to advance, thus obtaining a dynamic adhesion trend function. for: ,in This is the characteristic distance from the front end of the end effector to the adsorption zone; The dynamic bonding trend function describes the dynamic process by which points at the edge of the wafer gradually bond to the adsorption surface under continuous feeding conditions by the robotic arm. A larger value indicates a higher urgency for fitting at that position at the current moment; And collect data on each adsorption zone in real time. Current adsorption force Compare it with the preset benchmark adsorption force By comparison, the real-time adsorption force deviation of each partition is obtained. : ; The adsorption force deviation reflects the difference between the current adsorption force and the target adsorption force. A positive value indicates insufficient adsorption force, and a negative value indicates excessive adsorption force. Dynamically fit the trend function The average value is calculated based on the circumferential range corresponding to each adsorption zone to obtain the value for each zone. The tightness of the fit ; and combined with partitioning Real-time adsorption force deviation Calculate partitions Required pressure compensation weight : ,in , These are preset weighting coefficients for the ratio of adhesion tightness, adsorption force deviation, and maximum allowable adsorption pressure, respectively, which adjust the influence of adhesion trend and real-time adsorption force deviation on the compensation weights. Maximum allowable adsorption pressure; It should be noted that, , The settings should be tailored to the specific circumstances. For example, an expert weighting method can be used, which involves inviting experts in relevant fields to determine the pre-defined weighting coefficients for each indicator through professional opinion surveys and comprehensive evaluations. , The initial value can be 0.5, 0.5; The value ranges from 0 to 1, and a larger value indicates that the partition needs stronger adsorption pressure compensation. It should also be noted that the pressure compensation weight is not a static allocation value, but is dynamically updated with the changes in the feed speed of the robot arm and the real-time feedback of the vacuum degree of each zone. Its physical meaning is to achieve spatiotemporal coordination of adsorption force distribution—for areas with high adhesion urgency and insufficient current adsorption force, the adsorption force is increased first, while for areas with low adhesion urgency or sufficient adsorption force, the adsorption force is maintained or reduced, thereby achieving adaptive pressure adjustment as the wafer gradually falls into the end effector.

[0017] In this embodiment of the invention, the process of calculating the predicted suction adjustment amount of each adsorption zone based on the pressure compensation weight and the current adsorption force of the corresponding zone, and obtaining the target pressure setpoint for each adsorption zone in the next control cycle, is as follows: According to each adsorption zone Pressure compensation weight within the current control cycle and its current adsorption force Determine the adsorption partition Basic pressure regulation ; The basic pressure adjustment amount is calculated based on the product of the compensation weight and the maximum allowable adjustment range: ,in The preset maximum pressure adjustment step size per cycle is used to prevent sudden changes in adsorption force from impacting the wafer. Combined with the current adsorption force Adsorption force with reference Adsorption force deviation Correct the adjustment amount: ; Calculate the target pressure setpoint for the next control cycle t+1. : ; The target pressure setpoint is compared with the system's allowable pressure upper and lower limits. If the calculated result exceeds the upper limit, the upper limit value is used; if it is lower than the lower limit, the lower limit value is used, ensuring that the final pressure setpoint is always within the safe operating range. ,in and These are the minimum maintaining adsorption force and the maximum allowable adsorption pressure for the partition, respectively. The former ensures adsorption reliability, while the latter prevents the wafer from generating additional stress or damage due to excessive adsorption force. Finally, the sequence of target pressure setpoints that each adsorption zone needs to achieve in the next control cycle is obtained. ,in Given the total number of partitions, this sequence represents the predicted amount of partition suction adjustment. It should be noted that the predicted amount of suction force adjustment for the partition is not calculated independently based solely on the current state, but rather integrates the spatial priority reflected by the pressure compensation weight with the real-time deviation of the current suction force. Its physical significance lies in achieving a gradual adjustment of the suction force. For partitions with high weights and insufficient current suction force, the predicted amount will guide the pressure to increase rapidly; for partitions with low weights or sufficient current suction force, the predicted amount will maintain stable pressure or moderately adjust back, thereby achieving a smooth and adaptive redistribution of suction force throughout the entire wafer picking process.

[0018] In this embodiment of the invention, the process of extracting the pressure tracking error of each zone based on the deviation sequence between the target pressure setpoint and the actual pressure feedback value, and obtaining the attitude compensation amount used to correct the robot arm's plate-picking height, is as follows: During the current control cycle, collect the actual pressure feedback values ​​of each adsorption zone k. and the target pressure setpoint Compare and calculate the pressure tracking error of each partition. : ; The pressure tracking error reflects the degree of deviation of the adsorption force of each zone from the target value within the current control cycle. A positive value indicates that the actual adsorption force is lower than the target value, and a negative value indicates that the actual adsorption force is higher than the target value. Arrange the tracking errors of all partitions in order to form a pressure tracking error sequence. ; Feature extraction is performed on the pressure tracking error sequence, and the average value and gradient distribution characteristics of the error in each region of the pressure tracking error sequence are calculated. Based on the error distribution characteristics, the attitude compensation amount for the robot arm's plate-picking height is determined. : ,in The average tracking error of the pressure tracking error sequence. This represents the maximum value of the error gradient in the pressure tracking error sequence. , These are the preset weighting coefficients for the average tracking error and the maximum value of the error gradient, respectively. It should be noted that, , The settings should be tailored to the specific circumstances. For example, an expert-empowered approach could be adopted, where experts in relevant fields are invited to determine the pre-defined proportions for each indicator through professional opinion surveys and comprehensive evaluations. , The initial value can be 0.5, 0.5; It should also be noted that the attitude compensation amount is not a simple summary of the tracking errors of each partition, but rather a spatial coupling feature mined from the collaborative control deviation matrix. Its physical meaning is to improve the uniformity of the fit between the wafer and the end effector adsorption surface by fine-tuning the height of the robot arm. When the adsorption force is generally low in a certain area due to local wafer warping, the robot arm can make the area obtain better contact conditions by slightly raising and lowering it, thereby helping the pressure of each partition to converge to the target set value more quickly.

[0019] In this embodiment of the invention, the process of dynamically correcting the robotic arm's wafer-picking trajectory based on the attitude compensation amount to achieve stable transfer of the warped wafer is as follows: attitude compensation amount Decomposed into vertical compensation And tilt angle compensation Δθti(t): ,in The effective arm length of the robotic arm is used to convert height deviation into angular deviation; The attitude compensation is superimposed on the preset reference capture trajectory Zb(t) and On b(t), the dynamically corrected chip trajectory is obtained: ,in For the robotic arm in the next control cycle The vertical position of the target at any given time. For the robotic arm in the next control cycle At any given moment, the target's angle, attitude, and position; It should be noted that the dynamic trajectory correction is not adjusted all at once, but is continuously updated in each control cycle based on the latest attitude compensation. Its physical significance is to ensure that the robot arm and the warped wafer always maintain the best relative attitude. It compensates for the overall height deviation by vertical lifting and tilting to adapt to asymmetric bending. Thus, during the entire process of the wafer gradually falling into or leaving the end effector, it minimizes contact stress and ensures that the adsorption force of each zone always works within a safe range, ultimately achieving lossless and stable transmission of the warped wafer.

[0020] This invention collects airflow back pressure and edge occlusion coordinates using a circumferentially distributed non-contact pneumatic sensor array. Utilizing the physical property that airflow is insensitive to material optical properties, it fundamentally avoids measurement failures caused by reflection and transmission, ensuring stable acquisition of wafer attitude data even under extreme conditions such as polished bare silicon wafers and quartz glass. Based on the constructed edge flow field disturbance distribution map, the wafer warpage morphology, which is difficult to measure directly, is converted into quantifiable flow field parameters for inversion analysis. High-resolution extraction of local warpage amplitude and overall bending morphology is achieved while maintaining a compact front-end module space. By calculating the edge contact risk coefficient and dynamic bonding trend, a spatially selective adsorption mechanism is established, from risk identification to zoned pressure compensation. This mechanism actively enhances adsorption force in severely warped areas and maintains baseline pressure in well-bonded areas, ensuring reliable pickup while avoiding additional stress. By extracting the pressure tracking error of each zone to obtain attitude compensation, the adsorption force control and robot motion control are synergistically optimized, enabling the robot to adaptively follow the wafer warpage morphology for trajectory correction and maintain optimal contact posture throughout the pickup process.

[0021] The above formulas are all calculated by removing dimensions and taking the numerical values. The formulas are the closest to the real situation obtained by software simulation based on a large amount of data. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0022] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0023] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An adaptive adjustment method for warpage during wafer transfer, characterized in that: Includes the following steps: The airflow back pressure value and the coordinates of the occlusion boundary of multiple sampling points at the edge of the wafer are obtained. The gap leakage rate between the wafer and the end effector is collected by a non-contact pneumatic sensor array distributed along the circumference. The edge flow field disturbance distribution map is constructed by combining the robot insertion depth and the nominal radius of the wafer. Based on the edge flow field disturbance distribution map, the local warpage amplitude of the wafer edge is extracted, and the edge contact risk coefficient is calculated based on the local warpage amplitude and the preset safe adsorption gap. Based on the edge contact risk coefficient and the horizontal feed speed of the robot arm, the dynamic bonding trend of the wafer surface relative to the adsorption surface of the end effector is determined, and the pressure compensation weight required for each adsorption zone is obtained by combining the real-time adsorption force of each adsorption zone. The predicted amount of suction adjustment for each adsorption zone is calculated based on the pressure compensation weight and the current adsorption force of the corresponding zone, so as to obtain the target pressure setting value of each adsorption zone in the next control cycle. Based on the deviation sequence between the target pressure setpoint and the actual pressure feedback value, the pressure tracking error of each zone is extracted to obtain the attitude compensation amount used to correct the robot arm's pick-up height. The trajectory of the robotic arm for picking up the piece is dynamically corrected based on the posture compensation amount.

2. The adaptive adjustment method for warpage during wafer transfer according to claim 1, characterized in that: The process of acquiring the airflow back pressure value and occlusion boundary coordinates at multiple sampling points on the wafer edge, collecting the gap leakage rate between the wafer and the end effector through a circumferentially distributed non-contact pneumatic sensor array, and constructing an edge flow field disturbance distribution map by combining the robot insertion depth and the nominal radius of the wafer is as follows: Multiple miniature pneumatic sensing units are arranged at equal intervals along the circumference at the front end of the end effector. Each sensing unit includes a pressure sensing element and an edge detection element. During the process of the robotic arm inserting into the wafer at a preset speed, each sensing unit is triggered to collect data synchronously at a fixed sampling frequency to obtain the back pressure value and occlusion boundary coordinates at each circumferential position at each sampling moment. Based on the current insertion depth of the robotic arm and the nominal radius of the wafer, the occlusion boundary coordinates measured by each sensing unit are converted into the angular position of the wafer edge relative to the center of the end effector, and a sampling point mapping in polar coordinate system is established. And by using the pre-calibrated back pressure-gap relationship curve, the actual vertical gap between the lower surface of the wafer and the upper surface of the end effector at the sampling point is read from the back pressure value; By combining the gap values ​​of all sampling points at the same time and their corresponding angular positions, the discrete gap distribution of the wafer edge at that time can be obtained. Interpolation methods are used to fit discrete points to obtain the gap distribution function along the wafer circumference; and the deviation between the back pressure value and the theoretical reference back pressure is defined as the flow field disturbance intensity. By combining the insertion depth of the robotic arm, the flow field disturbance intensity and gap distribution are integrated to construct an edge flow field disturbance distribution map that includes spatial and temporal dimensions.

3. The adaptive adjustment method for warpage during wafer transfer according to claim 2, characterized in that: The process of extracting the local warpage amplitude of the wafer edge based on the edge flow field disturbance distribution map, and calculating the edge contact risk coefficient based on the local warpage amplitude and the preset safe adsorption gap is as follows: Extracting the gap distribution function from the edge flow field disturbance distribution map and the gap distribution function The minimum value is less than the preset contact threshold for the first time. The time marked as the critical time ; Gap distribution Converted to local warpage amplitude at the wafer edge The conversion formula is: ,in This represents the theoretical gap distribution for an ideally flat wafer at the same insertion depth. Based on the local warping amplitude and the preset safe adsorption gap Calculate the edge contact risk coefficient The calculation formula is: .

4. The adaptive adjustment method for warpage during wafer transfer according to claim 3, characterized in that: Based on the edge contact risk coefficient and the horizontal feed speed of the robotic arm, the dynamic bonding trend of the wafer surface relative to the end effector's adsorption surface is determined, and combined with the real-time adsorption force of each adsorption zone, the process of obtaining the required pressure compensation weight for each zone is as follows: Based on the edge contact risk coefficient and the current horizontal feed speed of the robotic arm, obtain the dynamic fitting trend function; The current adsorption force of each adsorption zone is collected in real time and compared with the preset benchmark adsorption force to obtain the real-time adsorption force deviation of each zone. The dynamic bonding trend function is averaged over the circumferential range corresponding to each adsorption zone to obtain the value for each zone. The tightness of the fit ; and combined with partitioning Real-time adsorption force deviation Calculate partitions Required pressure compensation weight : ,in , These are preset weighting coefficients for the ratio of adhesion tightness, adsorption force deviation, and maximum allowable adsorption pressure, respectively. This represents the maximum permissible adsorption pressure.

5. The adaptive adjustment method for warpage during wafer transfer according to claim 4, characterized in that: The process of calculating the predicted suction adjustment amount for each adsorption zone based on the pressure compensation weight and the current adsorption force of the corresponding zone, and obtaining the target pressure setpoint for each adsorption zone in the next control cycle, is as follows: According to each adsorption zone Pressure compensation weight within the current control cycle and its current adsorption force Determine the adsorption partition Basic pressure regulation ; The basic pressure adjustment amount is calculated based on the product of the compensation weight and the maximum allowable adjustment range: ,in The preset maximum pressure adjustment step size for a single cycle; Combined with the current adsorption force Adsorption force with reference Adsorption force deviation Correct the adjustment amount: ; Calculate the target pressure setpoint for the next control cycle t+1. : ; The target pressure setpoint is compared with the system's allowable upper and lower pressure limits. If the calculated result exceeds the upper limit, the upper limit value is used; if it is lower than the lower limit, the lower limit value is used. ,in and These are the minimum sustained adsorption force and the maximum allowable adsorption pressure for this partition, respectively. The sequence of target pressure setpoints required for each adsorption zone in the next control cycle is obtained. ,in This represents the total number of partitions.

6. The adaptive adjustment method for warpage during wafer transfer according to claim 5, characterized in that: The process of extracting the pressure tracking error of each zone and obtaining the attitude compensation amount used to correct the robot arm's piece-picking height based on the deviation sequence between the target pressure setpoint and the actual pressure feedback value is as follows: During the current control cycle, collect the actual pressure feedback values ​​of each adsorption zone k. and the target pressure setpoint Compare and calculate the pressure tracking error of each partition. : ; Arrange the tracking errors of all partitions in order to form a pressure tracking error sequence. ; Feature extraction is performed on the pressure tracking error sequence, and the average value and gradient distribution characteristics of the error in each region of the pressure tracking error sequence are calculated. Based on the error distribution characteristics, the attitude compensation amount for the robot arm's plate-picking height is determined. : ,in The average tracking error of the pressure tracking error sequence. This represents the maximum value of the error gradient in the pressure tracking error sequence. , These are the preset weighting coefficients for the average tracking error and the maximum error gradient, respectively.

Images