A method for automatic flattening and measuring step synchronization

Abstract

This invention discloses a method for automatically synchronizing leveling and step distance measurement. The method includes: starting from wafer loading, a path planning module reads the coordinate set of leveling points and step distance reference points, maps them to the same wafer coordinate system, and establishes a control point model; the calibration process is divided into four stages, generating a staged access sequence and converting it into a timing motion command containing motion posture and Z-axis movement; the chuck moves sequentially to each target point according to the command, the vision system captures and outputs pixel coordinates, and records the mechanical coordinates; real-time acquisition of deviation and contact status feedback is performed, and when the deviation exceeds a threshold or contact is abnormal, the subsequent path is adjusted online; after traversal is completed, the leveling matrix and step distance correction coefficient are calculated in parallel based on the same batch of data and written to the motion controller in one go. This invention integrates leveling and step distance measurement into a single traversal, reduces invalid chuck movement, achieves visual data reuse and synchronous parameter activation, shortens preparation time, and improves calibration accuracy and detection stability.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor testing equipment technology, specifically relating to a method for automatically leveling and measuring step distance simultaneously. Background Technology

[0002] In semiconductor wafer testing, automated probe stations must complete two key calibration tasks before formal testing: wafer leveling (i.e., alignment of the wafer coordinate system with the equipment's mechanical coordinate system) and step distance measurement (i.e., measurement and correction of the deviation between the actual movement distance of the chuck and the theoretical step distance). The accuracy of these two calibrations directly affects the accuracy of subsequent probe placement and the reliability of the test results.

[0003] In existing technologies, solutions related to semiconductor wafer testing mainly focus on the following areas: Direction 1: Vision-based probe station image positioning and alignment technology. For example, Chinese patent CN105513990A discloses a probe station image positioning device and a vision alignment method. This method uses an industrial camera to acquire images of wafer surface markings or pads, extracts feature point pixel coordinates using image processing algorithms, and controls the motion stage for position compensation. However, this technology focuses on a single alignment step, failing to reuse visual recognition data for step measurement or to provide a unified plan for the flattening and step measurement paths.

[0004] Direction Two: Probe Station Positioning Compensation Calibration Technology. Chinese Patent Publication No. CN115902327B discloses a calibration method and probe station for probe station positioning compensation. Position calibration is performed by obtaining the compensation offset of a preset reference point and the key positions of critical solder joints. This method still primarily follows a phased approach of "alignment first, calibration later," requiring the wafer to undergo separate step measurement and movement after flattening, resulting in numerous chuck reciprocations.

[0005] Direction 3: Probe angle and height calibration technology. Chinese patent CN117476529B discloses a probe calibration method, focusing on offline or online calibration of probe angle and height, but does not involve path fusion and data reuse between wafer flattening and motion step measurement.

[0006] Direction 4: Wafer leveling (flattening) technology. Chinese patent CN118737919A discloses a wafer leveling method that uses an iterative cycle of "matching-rotation-re-movement" to gradually reduce the deviation between the wafer and the standard angle. This technology is only for a single wafer leveling task and does not involve step measurement throughout the process.

[0007] Direction 5: Wafer Probe Station Inspection Path Planning Technology. Chinese Patent Publication No. CN118553635A discloses a wafer probe station inspection path planning method for planning the pin addressing path of the die under test during parallel testing of multiple workstation probe cards. This technology plans the die under test rather than the flattening markers or step reference feature points used for coordinate calibration, and does not involve pre-test wafer flattening calibration or step measurement.

[0008] Direction Six: Probe Deviation Compensation and Synchronization Calibration Technology. Some solutions calibrate the probe position by obtaining the deviation between the target measurement position and the actual probe insertion position, or calibrate the synchronization of multi-channel testing systems, but none of them integrate wafer coordinate system flattening and motion axis step distance measurement as a unified process at the path level.

[0009] The aforementioned existing technologies share common drawbacks: flattening and step measurement are performed in stages, with each stage involving movement to different target points for imaging, requiring the chuck to travel back and forth on the wafer multiple times, resulting in a long preparation time; the images and coordinate data acquired during the flattening stage are discarded after calculation and are not used for step correction; the step measurement stage requires re-movement and data acquisition, leading to resource waste; existing path planning technologies only address die pinning during the formal testing stage and do not cover the joint planning of two types of heterogeneous control points in the pre-test calibration stage; some solutions use offline calibration methods, which cannot respond promptly to real-time error changes caused by temperature drift, mechanical wear, etc. Summary of the Invention

[0010] This invention provides the following technical solution: a method for automatically leveling and measuring step distance simultaneously, characterized in that: From the moment the wafer is loaded into the chuck, the path planning module reads the wafer recipe and obtains the set of flattening point coordinates and the set of step pitch reference point coordinates; wherein, the flattening point is a process control point used to determine the contact plane between the probe and the wafer, and the step pitch reference point is a reference control point used to calibrate the step pitch of the motion coordinate system. The flattening point and the step pitch reference point are mapped to the same wafer coordinate system, and the spatial association, functional priority and access restriction conditions between each control point are established based on the wafer map, the die arrangement relationship and the preset process constraints. The single wafer calibration process is divided into a pre-flattening stage, a coarse pitch calibration stage, a fine flattening stage, and a compensation measurement stage. For each stage, the corresponding sub-planning strategy is called to generate a staged access sequence. The phased access sequence is converted into probe station motion control instructions, which include the XY plane access sequence, the motion posture between adjacent control points, the Z-axis pre-lift and pre-lower action timing, and the contact and separation process triggering conditions between the probe and the wafer. The chuck moves sequentially to each target point according to the motion control command; at each target point, the vision system captures an image of the wafer surface and identifies feature points, outputs the pixel coordinates of the feature points in the image coordinate system, and records the mechanical coordinates of the chuck at the current target point. Real-time data collection of leveling deviation, contact status, step distance measurement error, and motion mechanism status feedback; in response to actual deviations exceeding set thresholds or contact anomalies, online adjustments are made to the sequence, density, and access method of subsequent unvisited control points. After all target points have been visited, the least squares method is used to fit the rotation angle and translation amount of the wafer coordinate system relative to the machine coordinate system based on the pixel coordinates and machine coordinates of all flattening points to generate the flattening matrix. At the same time, the step correction coefficient of each axis is calculated based on the actual machine displacement and theoretical step distance between adjacent step distance reference points. The two calculation processes are independent and executed in parallel. The flattening matrix and step size correction coefficient are written into the motion controller at once, and subsequent tests are performed based on the corrected coordinate system and step size parameters for addressing.

[0011] Preferably, the pre-flattening stage includes: prioritizing geometric coverage, selecting sparse flattening points within the effective area of ​​the wafer to form a large-span, low-density initial access sequence.

[0012] Preferably, the coarse step calibration stage includes: generating a reciprocating scanning sequence with a fixed direction and a fixed step increment around the flattening point or in a dense area of ​​step reference points.

[0013] Preferably, the fine flattening stage includes: based on the wafer warpage data fed back by the pre-flattening, refining the local flattening points, and generating a local fine-repair path with small step size and low impact.

[0014] Preferably, the compensation measurement stage includes: dynamically adding compensation access points based on the measured step distance error and the flattening residual deviation, and inserting them into the remaining path.

[0015] Preferably, the least squares fitting of the rotation angle and translation amount of the wafer coordinate system relative to the machine coordinate system based on the pixel coordinates and machine coordinates of all flattened points includes: Define the coordinates of the k-th target point in the mechanical coordinate system as (x_{k,m}, y_{k,m}), and the theoretical coordinates in the wafer coordinate system as (x_{k,w}, y_{k,w}). The transformation relationship from the wafer coordinate system to the machine coordinate system satisfies: x_{k,m} = x_{k,w}·cosθ − y_{k,w}·sinθ + t_x + ε_{k,x} y_{k,m} = x_{k,w}·sinθ + y_{k,w}·cosθ + t_y + ε_{k,y} Where θ is the rotation angle of the wafer coordinate system relative to the machine coordinate system, (t_x, t_y) is the translation amount, and ε_{k,x} and ε_{k,y} are the fitting residuals; Solve for θ, t_x, and t_y by minimizing the sum of squared residuals at all flattened points using the least squares method: min_{θ,t_x,t_y} Σ_{k∈A} (ε_{k,x}² + ε_{k,y}²) In the formula, A is the set of flattening point indices.

[0016] Preferably, the step correction coefficients for each axis calculated based on the actual mechanical displacement and theoretical step distance between adjacent step distance reference points include: The step size correction coefficient α_x in the x-direction satisfies: α_x = Σ(Δx_{actual,j} · Δx_theory) / Σ(Δx_theory²) The step correction coefficient α_y in the y-direction satisfies: α_y = Σ(Δy_{actual,j} · Δy_theory) / Σ(Δy_theory²) Where Δx_theory and Δy_theory are the theoretical step distances, and Δx_{actual,j} and Δy_{actual,j} are the actual mechanical displacements of the j-th pair of adjacent step distance reference points.

[0017] Preferably, the subsequent test, which performs addressing based on the corrected coordinate system and step parameters, includes: The motion controller converts the theoretical target coordinates (x_target, y_target) into corrected coordinates corresponding to the actual transmitted pulses: x_cmd = (x_target − t_x) / α_x y_cmd = (y_target − t_y) / α_y Where (x_cmd, y_cmd) is the target position written to the motion controller, α_x and α_y are the step correction coefficients in the x and y directions, respectively, and (t_x, t_y) is the translation amount.

[0018] The beneficial effects of this invention are as follows: 1. By merging the flattening point and the stepping reference point into a single traversal path, the original two independent wafer movement journeys are compressed into one, significantly reducing the number of chuck round trips and lowering the preparation time before testing; only one image needs to be captured at the same target point to simultaneously provide input data for flattening calculation and stepping correction calculation, avoiding repeated shooting and image processing overhead; the calculation of the flattening matrix and the stepping correction coefficient are performed independently and in parallel, and the parameters are written to the motion controller at once after the calculation is completed, avoiding the latency and inconsistency risks caused by step writing.

[0019] 2. The path planning module automatically generates a set of target points and a traversal path based on information such as diameter and notch type in the wafer recipe, eliminating the need to develop separate path programs for different wafer types. Flattening and step measurement are completed in the same traversal, and the correction parameters can simultaneously reflect real-time errors such as current temperature drift and mechanical status, resulting in better dynamic adaptability. Multi-stage planning avoids mechanical impact caused by high-density reciprocating motion throughout the process, and the dynamic correction mechanism can adjust subsequent paths in a timely manner in case of anomalies, reducing hard contact and repeated positioning errors. Attached Figure Description

[0020] Figure 1 This is a block diagram of the system modules in an embodiment of the present invention; Figure 2 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0022] In the description of this invention, it should be noted that the terms "upper", "lower", "inner", "outer", "front end", "rear end", "both ends", "one end", "the other end", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0023] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installed," "equipped with," "connected," etc., should be interpreted broadly. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Simultaneously, when an component is referred to as "fixed to" or "equipped with" another component, it can be directly on the other component or may have an intervening component present. When an component is referred to as "connected to" another component, it can be directly connected to the other component or may have an intervening component present. When an component is referred to as "fixedly connected to" another component, it can be a common fixed connection method such as welding, bolting, or gluing. In summary, those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0024] This embodiment provides a system that automatically levels and measures step distances simultaneously. Based on this system, the method in the present invention can be implemented, such as... Figure 1 As shown, the system specifically includes the following modules: The visual recognition module, including a high-precision industrial camera and an image processing unit, is used to capture images of the wafer surface after the chuck moves to each target point, identify flattening marker points and step distance reference feature points, and output the pixel coordinates of the feature points in the image coordinate system. In this invention, the flattening marker points and step distance reference feature points have different functional attributes and recognition requirements: the flattening marker points are used to determine the rotation and translation of the wafer coordinate system relative to the machine coordinate system, and must be evenly distributed along the edges and interior of the effective area of ​​the wafer to ensure geometric coverage; the step distance reference feature points are used to calibrate the step distance deviation of the motion axis, and must be regularly arranged along the x and y axes to facilitate the measurement of the actual displacement between adjacent points. Both types of feature points can use the same feature extraction algorithm (such as template matching, edge detection, or blob analysis) during image processing, but the output data is sent to the flattening parameter calculation module and the step distance correction calculation module respectively, achieving the effect of single-shot, dual-channel multiplexing. The typical range of image resolution for visual recognition modules is 0.5μm / pixel to 5μm / pixel. The specific value is determined by a comprehensive trade-off between the required recognition accuracy and the field of view. Higher resolution (lower value) is more conducive to sub-pixel localization of fine feature points, but the field of view of a single frame is correspondingly reduced, which may increase the number of target points. In practical applications, for routine testing of 300mm wafers, the image resolution is usually set to 1μm / pixel to 2μm / pixel.

[0025] The motion control module, including an XYZθ four-axis controller and servo drive unit, receives motion commands from the path planning module, controls the chuck to move along a uniform path, locks the pose at each target point and feeds back the chuck's current mechanical coordinates, and writes the leveling matrix and step correction coefficients to the controller register after parameter calculation. The chuck positioning repeatability directly affects the reliability of the fitting between the step correction coefficients and the leveling matrix. In this system, the chuck positioning repeatability is configured to ≤±1μm. This accuracy level is the standard configuration for high-precision automatic probe stations, ensuring that the measurement error of the actual mechanical displacement between adjacent step reference points is controlled within an acceptable range, preventing significant deviations in the estimation of α_x and α_y.

[0026] The path planning module is used to perform joint planning of control points in the synchronous process of wafer prober leveling and pitch measurement, and to generate access command sequences coupled with motion posture, Z-axis movement, and contact flow. This module specifically includes the following sub-units: The control point reading unit is used to read the list of flattening point coordinates and the coordinates of the step pitch reference points from the wafer recipe. The wafer recipe also records process parameters such as wafer diameter, notch type (Notch or Flat), die arrangement, and center-to-center distance between adjacent dies. The path planning module can automatically adapt to the calibration requirements of wafers of different specifications based on this. For example, for Notch type wafers, the flattening points can be distributed with reference to the Notch direction. For Flat type wafers, the flattening points need to be selected to avoid the Flat edge area to ensure that all flattening points are located within the effective area of ​​the wafer.

[0027] The control point modeling unit is used to map the flattening point and the step pitch reference point to the same wafer coordinate system, and establish the spatial association, functional priority and access restriction conditions between each control point based on the wafer map, the die arrangement relationship and the preset process constraints. The multi-stage planning unit is used to divide a single wafer calibration process into a pre-flattening stage, a coarse pitch calibration stage, a fine flattening stage, and a compensation measurement stage, and calls the corresponding sub-planning strategy for each stage to generate a staged access sequence. The instruction coupling unit is used to convert the phased access sequence into probe station motion control instructions. The motion control instructions include the XY plane access sequence, the motion posture between adjacent control points, the Z-axis pre-lift / pre-lower action timing, the contact / separation process triggering conditions between the probe and the wafer, and the sampling enable timing of the flattening sensor and the step measurement sensor. The dynamic correction unit is used to collect data on leveling deviation, contact status, step distance measurement error, and motion mechanism status feedback in real time during execution. In response to actual deviations exceeding set thresholds or contact anomalies, it adjusts the order, density, and access method of subsequent unvisited control points online. The leveling deviation threshold can be set to ±50μm, and the step distance measurement error threshold can be set to ±0.05% of the theoretical step distance. When the deviation exceeds the limit or a contact anomaly occurs, the dynamic correction unit triggers path replanning, adding compensation access points or adjusting the access density in the remaining path, and marking data that has exceeded the tolerance as invalid, to be re-acquired during the compensation measurement phase.

[0028] The flattening parameter calculation module is used to generate a flattening matrix by fitting the rotation angle and translation amount of the wafer coordinate system relative to the device coordinate system using the least squares method based on the pixel coordinates output by the vision recognition module and the chuck mechanical coordinates fed back by the motion control module.

[0029] The step distance correction calculation module is used to calculate the step distance correction coefficient for each axis by comparing the actual mechanical displacement and theoretical step distance between adjacent reference points in the same batch of target point data, and output it to the motion control module.

[0030] like Figure 2 The flowchart shown in this embodiment illustrates the automatic preparation of a 300mm diameter silicon wafer with a notch for electrical testing using an automated probe station. The process will be described in detail below. First, path planning: After the wafer is loaded into the chuck, the path planning module reads the wafer recipe and obtains 4 flattening points (distributed in different directions on the edge of the wafer) and 6 step reference points (3 pairs each distributed along the x and y directions); at the same time, it reads the wafer map and the relationship between the die arrangement.

[0031] The control point modeling unit maps 10 target points to the same wafer coordinate system and establishes spatial associations and access restrictions between the points: flattening points within the same flattening area are accessed consecutively to ensure spatial consistency of the data points used for fitting; the step distance reference point is accessed in a fixed posture before the pre-flattening stage and after the fine flattening stage, that is, the access of the step distance reference point maintains the same chuck θ-axis angle to ensure the comparability of the displacement measurement results of the two stages; the movement distance between adjacent control points does not exceed the mechanical stroke threshold.

[0032] The multi-stage planning unit divides the calibration process into four stages: pre-leveling, coarse step calibration, fine leveling, and compensation measurement. Pre-flattening stage: Select 4 edge flattening points to generate a large-span, low-density initial access sequence; Coarse step size calibration stage: Generate a reciprocating scan sequence around the flattening points with a fixed direction and fixed step size increment, covering 6 step size reference points; The goal of this stage is to obtain an initial estimate of the step size deviation. The step size reference points are arranged along the x and y directions according to the theoretical step size Δx_theory=10mm and Δy_theory=10mm, respectively. The access sequence is advanced pair by pair along the same direction (e.g., accessing the 1st, 2nd, and 3rd pairs sequentially along the positive x direction). To ensure that the actual displacement measurement between adjacent points is not affected by the backlash; Fine flattening stage: Based on the wafer warpage data from the pre-flattening feedback, determine whether it is necessary to densify the local flattening points. If the warpage exceeds the threshold and the height difference between the wafer edge and the center exceeds ±50μm, add 2-4 fine flattening points in the central region. The selection of additional points is determined based on the warpage distribution, with priority given to areas with large warpage gradients to generate a small-step, low-impact local fine-tuning path; Compensation measurement stage: Based on the measured steppage error and the residual flattening deviation, determine whether to add compensation access points. If the steppage error exceeds ±0.05% of the theoretical steppage (i.e., the error exceeds ±5μm under a 10mm theoretical steppage), add 1-2 compensation measurement points in the direction of maximum error; if the residual flattening deviation exceeds ±30μm, add 1-2 compensation points in the area of ​​maximum deviation. The added compensation access points are inserted into the remaining path and executed first.

[0033] The instruction coupling unit converts the phased access sequence into motion control instructions that include the XY access order, motion posture, Z-axis pre-lift or pre-descent action timing, probe contact or separation process trigger conditions, and sensor sampling enable timing.

[0034] During the data acquisition phase, the chuck moves sequentially to each target point according to a phased access sequence. At each target point, the vision system captures an image and identifies feature points, outputting the pixel coordinates of the feature points in the image coordinate system, while simultaneously recording the mechanical coordinates of the chuck at the current target point. Pixel coordinates and mechanical coordinates are stored in pairs. For composite target points that simultaneously perform both leveling and step distance measurement functions, the vision system executes two sampling modes sequentially at the same location: first, it captures a leveling mark image with a field of view suitable for global recognition, and then switches to a high-magnification local field of view to capture a step distance reference feature image. During the two captures, the chuck remains in a locked pose to ensure accurate correspondence between pixel coordinates and mechanical coordinates.

[0035] In this invention, the key parameters involved in each module and process step are determined based on conventional operating conditions in the semiconductor testing field. Specifically, the high-precision industrial camera used in the vision recognition module typically has an image resolution ranging from 0.5 μm / pixel to 5 μm / pixel. The specific value is determined by a comprehensive trade-off between the required recognition accuracy and the field of view size—higher resolution (smaller value) is more conducive to sub-pixel positioning of fine feature points, but the field of view per frame is correspondingly reduced, potentially increasing the number of target points. In practical applications, for conventional testing of 300mm wafers, the image resolution is usually set to 1 μm / pixel to 2 μm / pixel. The chuck positioning repeatability of the motion control module directly affects the fitting reliability of the step correction coefficient and the flattening matrix. In this system, the chuck positioning repeatability is configured to ≤±1 μm. This accuracy level is the standard configuration for high-precision automatic probe stations, ensuring that the measurement error of the actual mechanical displacement between adjacent step reference points is controlled within an acceptable range, preventing significant deviations in the estimation of α_x and α_y. The total number of target points K in a single traversal is determined comprehensively based on wafer size, accuracy requirements, and testing efficiency, with a typical range of 4 to 20 points. Among them, the number of flattening points is usually 3 to 8, which must ensure sufficient geometric coverage within the effective area of ​​the wafer. The number of step pitch reference points is usually 2 to 12, with 1 to 6 pairs distributed along the x and y directions respectively. The theoretical step pitches Δx_theory and Δy_theory between adjacent reference points are determined based on the arrangement period of the grains on the wafer, with typical values ​​ranging from 5 mm to 20 mm. The flattening points and step pitch reference points together constitute the target point set for a single traversal. The step correction coefficients α_x and α_y are dimensionless scaling factors that reflect the combined influence of screw pitch error, thermal expansion, and other effects of the mechanical transmission system on the deviation between the actual and theoretical displacements. Typical values ​​range from 0.999 to 1.001. The fitted wafer rotation angle θ is determined by the initial attitude deviation during wafer loading, typically ranging from -5° to +5°. The magnitude of the translation (t_x, t_y) is related to the wafer diameter and loading offset, usually ranging from several millimeters to tens of millimeters, and is specifically determined by the fitting results of the flattening point. The above parameters can be adaptively configured by the operator in the wafer recipe panel according to different probe station models, wafer specifications, and precision requirements. This invention does not limit their specific values.

[0036] The system collects data in real time on leveling deviation, contact status, step distance measurement error, and motion mechanism status feedback. If the actual deviation exceeds a set threshold or a contact anomaly occurs, the system returns to the path planning stage to adjust the order, density, and access method of subsequent unvisited control points online. When the deviation exceeds the limit, additional access points are added to the excess area. When a contact anomaly occurs, the chuck Z-axis descent speed is reduced and the access density in that area is increased for re-acquisition, while the already acquired abnormal data is marked as invalid.

[0037] After the traversal is completed, the flattening parameter calculation module uses the least squares method to fit the rotation angle θ and translation amount (t_x, t_y) of the wafer coordinate system relative to the machine coordinate system based on the pixel coordinates and machine coordinates of all flattening points; the step correction calculation module calculates the step correction coefficient α_x in the x direction and the step correction coefficient α_y in the y direction based on the actual displacement and theoretical step distance between adjacent step reference points.

[0038] The specific calculation formula is as follows: The transformation relationship from the wafer coordinate system to the machine coordinate system satisfies: x_{k,m} = x_{k,w}·cosθ − y_{k,w}·sinθ + t_x + ε_{k,x} y_{k,m} = x_{k,w}·sinθ + y_{k,w}·cosθ + t_y + ε_{k,y} Minimize the sum of squared residuals for all flattened points using the least squares method: min_{θ,t_x,t_y} Σ_{k∈A} (ε_{k,x}² + ε_{k,y}²) Step correction factor: α_x = Σ(Δx_{actual,j} · Δx_theory) / Σ(Δx_theory²) α_y = Σ(Δy_{actual,j} · Δy_theory) / Σ(Δy_theory²) The two calculation processes are executed independently in the flattening parameter calculation module and the step correction calculation module, respectively, without dependence on each other, making full use of multi-core processor resources to achieve parallel acceleration.

[0039] The flattening matrix (θ, t_x, t_y) and step correction coefficients (α_x, α_y) are written into the motion controller at once. During subsequent testing, the motion controller converts the theoretical target coordinates into corrected coordinates corresponding to the actual transmitted pulses. x_cmd = (x_target − t_x) / α_x y_cmd = (y_target − t_y) / α_y Formal testing then commenced. The simultaneous one-time writing ensured that the flattening parameters and step correction parameters took effect at the same moment, avoiding the mismatch between the coordinate system and step correction caused by step-by-step writing.

[0040] Table 1. Example of parameter settings for the embodiment:

[0041] This parameter setting example is based on a typical pre-test calibration scenario for a 300mm (12-inch) silicon wafer, providing a specific implementation reference for simultaneous flattening and pitch measurement. Four edge flattening points ensure sufficient geometric coverage for fitting the wafer coordinate system rotation angle θ and translation (t_x, t_y). Six pitch reference points (three pairs each in the x and y directions) are distributed with a theoretical pitch of 10mm, sufficient to support reliable estimation of the pitch correction coefficients α_x and α_y for each axis. The fitted rotation angle of 0.05° and translation (-0.12, 0.08)mm are exemplary calculation results, reflecting common small positional deviations after wafer loading. Pitch correction coefficients of 1.0002 and 0.9999 indicate that mechanical scaling errors are typically on the order of one-thousandth, consistent with the actual working conditions of a high-precision probe station. During the fine leveling stage, 0-4 additional densification points are dynamically added based on the wafer warpage data from the pre-leveling feedback. During the compensation measurement stage, 0-4 additional compensation access points are dynamically added based on the step error and residual leveling deviation, demonstrating the core characteristics of this invention: multi-stage planning and dynamic correction. The above parameters are illustrative and can be adjusted by the operator according to different wafer specifications, accuracy requirements, and equipment status in practical applications. They should not be construed as any limitation on this invention.

[0042] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for automatically leveling and measuring step distance simultaneously, characterized in that, include: From the moment the wafer is loaded into the chuck, the path planning module reads the wafer recipe and obtains the set of flattening point coordinates and the set of step pitch reference point coordinates; wherein, the flattening point is a process control point used to determine the contact plane between the probe and the wafer, and the step pitch reference point is a reference control point used to calibrate the step pitch of the motion coordinate system. The flattening point and the step pitch reference point are mapped to the same wafer coordinate system, and the spatial association, functional priority and access restriction conditions between each control point are established based on the wafer map, the die arrangement relationship and the preset process constraints. The single wafer calibration process is divided into a pre-flattening stage, a coarse pitch calibration stage, a fine flattening stage, and a compensation measurement stage. For each stage, the corresponding sub-planning strategy is called to generate a staged access sequence. The phased access sequence is converted into probe station motion control instructions, which include the XY plane access sequence, the motion posture between adjacent control points, the Z-axis pre-lift and pre-lower action timing, and the contact and separation process triggering conditions between the probe and the wafer. The chuck moves sequentially to each target point according to the motion control command; at each target point, the vision system captures an image of the wafer surface and identifies feature points, outputs the pixel coordinates of the feature points in the image coordinate system, and records the mechanical coordinates of the chuck at the current target point. Real-time data collection of leveling deviation, contact status, step distance measurement error, and motion mechanism status feedback; in response to actual deviations exceeding set thresholds or contact anomalies, online adjustments are made to the sequence, density, and access method of subsequent unvisited control points. After all target points have been visited, the least squares method is used to fit the rotation angle and translation amount of the wafer coordinate system relative to the machine coordinate system based on the pixel coordinates and machine coordinates of all flattening points to generate the flattening matrix. At the same time, the step correction coefficient of each axis is calculated based on the actual machine displacement and theoretical step distance between adjacent step distance reference points. The two calculation processes are independent and executed in parallel. The flattening matrix and step size correction coefficient are written into the motion controller at once, and subsequent tests are performed based on the corrected coordinate system and step size parameters for addressing.

2. The method according to claim 1, characterized in that, The pre-flattening stage includes: prioritizing geometric coverage, selecting sparse flattening points within the effective area of ​​the wafer to form a large-span, low-density initial access sequence.

3. The method according to claim 2, characterized in that, The coarse step calibration stage includes: generating a reciprocating scanning sequence with a fixed direction and a fixed step increment around the flattening point or in a dense area of ​​step reference points.

4. The method according to claim 3, characterized in that, The fine flattening stage includes: based on the wafer warpage data fed back from the pre-flattening, refining the local flattening points, and generating a local fine-repair path with small step size and low impact.

5. The method according to claim 4, characterized in that, The compensation measurement stage includes: dynamically adding compensation access points based on the measured step distance error and the flattening residual deviation, and inserting them into the remaining path.

6. The method according to claim 5, characterized in that, The rotation angle and translation amount of the wafer coordinate system relative to the machine coordinate system, which are fitted using the least squares method based on the pixel coordinates and machine coordinates of all flattened points, include: Define the coordinates of the k-th target point in the mechanical coordinate system as (x_{k,m}, y_{k,m}), and the theoretical coordinates in the wafer coordinate system as (x_{k,w}, y_{k,w}). The transformation relationship from the wafer coordinate system to the machine coordinate system satisfies: x_{k,m} = x_{k,w}·cosθ − y_{k,w}·sinθ + t_x + ε_{k,x} y_{k,m} = x_{k,w}·sinθ + y_{k,w}·cosθ + t_y + ε_{k,y} Where θ is the rotation angle of the wafer coordinate system relative to the machine coordinate system, (t_x, t_y) is the translation amount, and ε_{k,x} and ε_{k,y} are the fitting residuals; Solve for θ, t_x, and t_y by minimizing the sum of squared residuals at all flattened points using the least squares method: min_{θ,t_x,t_y} Σ_{k∈A} (ε_{k,x}² + ε_{k,y}²) In the formula, A is the set of flattening point indices.

7. The method according to claim 6, characterized in that, The step correction coefficients for each axis, calculated based on the actual mechanical displacement and theoretical step distance between adjacent step distance reference points, include: The step size correction coefficient α_x in the x-direction satisfies: α_x = Σ(Δx_{actual,j} · Δx_theory) / Σ(Δx_theory²) The step correction coefficient α_y in the y-direction satisfies: α_y = Σ(Δy_{actual,j} · Δy_theory) / Σ(Δy_theory²) Where Δx_theory and Δy_theory are the theoretical step distances, and Δx_{actual,j} and Δy_{actual,j} are the actual mechanical displacements of the j-th pair of adjacent step distance reference points.

8. The method according to claim 7, characterized in that, The subsequent tests, which perform addressing based on the corrected coordinate system and step size parameters, include: The motion controller converts the theoretical target coordinates (x_target, y_target) into corrected coordinates corresponding to the actual transmitted pulses: x_cmd = (x_target − t_x) / α_x y_cmd = (y_target − t_y) / α_y Where (x_cmd, y_cmd) is the target position written to the motion controller, α_x and α_y are the step correction coefficients in the x and y directions, respectively, and (t_x, t_y) is the translation amount.

Images