A detection system, an automatic focusing method thereof, and a storage medium

Abstract

The application provides a detection system and an automatic focusing method and a storage medium thereof. The system comprises a light source assembly, an optical assembly, an imaging assembly and a processor. A surface of a to-be-detected object is provided with a feature structure for focus surface positioning. The processor is configured to coarsely adjust a relative position of the imaging assembly and the to-be-detected object, collect interference images at multiple positions, determine an initial focus surface position according to the collected interference images, then change the relative position of the imaging assembly and the to-be-detected object multiple times near the initial focus surface position and collect interference images at corresponding positions respectively, construct a sharpness index by using phase information corresponding to the feature structure, and compare index values at the positions to accurately position a best focus surface position. The application improves the focus adjustment basis from traditional image intensity analysis to phase domain analysis, utilizes high sensitivity of the phase to defocusing, realizes automatic focus adjustment with nanometer-level precision without increasing hardware, and provides reliable focus surface guarantee for high-precision semiconductor detection.

Description

Technical Field

[0001] This application relates to the field of semiconductor detection technology, specifically to a detection system and its automatic focusing method and storage medium. Background Technology

[0002] In semiconductor manufacturing, accurate inspection of wafer surface morphology is crucial for process control and yield improvement. Interferometric measurement technology, due to its advantages such as non-contact operation, high precision, and full-field measurement, is widely used in wafer surface defect detection, film thickness measurement, and morphology characterization.

[0003] However, with the continuous development of semiconductor technology, wafer sizes are gradually increasing (e.g., from 200mm to 300mm and even larger). Coupled with stress accumulation during processes such as multilayer thin film deposition, etching, and chemical mechanical polishing, wafer warpage has become a key factor affecting detection accuracy. Warpage causes significant height differences in the vertical direction across different areas of the wafer surface. Even if the stage supporting the wafer remains macroscopically horizontal, the positions of each detection point on the wafer relative to the focal plane of the imaging system will still deviate. This focal plane deviation directly causes blurred interference fringes and decreased contrast, leading to increased errors in phase information extraction, ultimately affecting the accuracy of defect detection, film thickness measurement, and topography reconstruction. In other words, the accuracy of interferometry heavily depends on the focal plane positional relationship between the imaging system and the sample under test. Only when the sample under test is precisely located at the focal plane position of the imaging system can clear interference fringes and accurate phase information be obtained, thus ensuring the reliability of the measurement results.

[0004] Traditional autofocusing methods are primarily based on image sharpness evaluation functions, such as gradient functions, frequency domain functions, and entropy functions, to locate the optimal focal plane by searching for the extreme points of these functions. While widely used in ordinary optical microscopes, these methods have the following limitations in interferometric systems: First, interferometric images contain complex fringe information, and traditional sharpness evaluation functions are easily affected by changes in fringe density and contrast, resulting in insufficient robustness. Second, the focusing accuracy of traditional methods is typically at the micrometer level, which is insufficient to meet the nanometer-level precision requirements of semiconductor detection. Third, there is a lack of effective verification mechanisms, making it impossible to objectively evaluate and correct the focusing results.

[0005] To address these issues, some existing technologies employ external auxiliary markers for focusing, such as creating alignment marks on the wafer edge and locating the marks using image recognition. However, these methods are limited by the manufacturing precision of the markers and the resolution of the recognition algorithm, making it difficult to achieve nanometer-level precision autofocus. Other technologies focus by analyzing the phase information of the interferogram, but this typically requires complex hardware modifications or additional reference optical paths, resulting in higher system complexity and cost. Summary of the Invention

[0006] This application provides a detection system, its automatic focusing method, and a storage medium, aiming to solve the technical problems of insufficient focusing accuracy and poor reliability in existing detection systems.

[0007] In a first aspect, embodiments of this application provide a detection system, including:

[0008] A light source assembly for providing a coherent illumination beam;

[0009] An optical component is used to guide the illumination beam to the surface of the object under test and to receive the beam returning from the surface of the object under test to form interference light; the surface of the object under test is provided with a feature structure for focal plane positioning.

[0010] An imaging component is used to acquire the interference light and generate an interference image;

[0011] The processor, connected to the imaging component, is configured to:

[0012] The relative position between the imaging component and the object under test is coarsely adjusted, and interferometric images are acquired at multiple locations. The initial focal plane position is determined based on the acquired interferometric images. Near the initial focal plane position, the relative position between the imaging component and the object under test is changed multiple times, and interferometric images are acquired at the corresponding locations. The focal plane state at each location is evaluated based on the phase information corresponding to the feature structures in the interferometric images acquired at each location. The optimal focal plane position is determined based on the evaluation results, and the relative position between the imaging component and the object under test is adjusted to the optimal focal plane position to begin detection. By introducing a focal plane evaluation mechanism based on phase information, the focusing criterion is elevated from traditional image intensity analysis to phase domain analysis, fully utilizing the physical characteristic of phase being highly sensitive to defocus changes in interferometry.

[0013] In some embodiments, the feature structure includes one or more stepped etched lines disposed on the surface of the object to be tested and distributed radially through it; the height of the stepped etched lines does not exceed half the wavelength of the illumination beam.

[0014] In some embodiments, the optical components include a beam splitter, a collimating lens, a reference lens, and a relay lens;

[0015] The illumination beam emitted from the light source assembly is transmitted sequentially through a beam splitter and a collimating lens before being incident on a reference mirror. A portion of the beam transmitted through the reference mirror forms a test beam that is incident on the surface of the object under test, and a portion of the beam reflected by the reference mirror forms a reference beam. The test beam is reflected by the surface of the object under test to form a reflected beam. The reflected beam interferes with the reference beam and is transmitted sequentially through the collimating lens, reflected by the beam splitter, and transmitted through the relay mirror before being incident on the imaging assembly.

[0016] In some embodiments, the imaging component is coaxially arranged with the relay mirror and is movable relative to the relay mirror along the optical axis.

[0017] In some embodiments, the detection system further includes a drive mechanism that is driveably connected to the imaging component or the stage carrying the object to be measured, and is also connected to the processor.

[0018] The processor is also configured to control the drive mechanism to move the imaging component or stage according to the determined optimal focal plane position, so that the imaging component is adjusted to the optimal focal plane position.

[0019] Secondly, embodiments of this application provide an automatic focusing method applied to a detection system, the detection system including at least a light source component and an imaging component; the automatic focusing method includes:

[0020] The relative position of the coarse-adjusted imaging component and the object under test is used to acquire interference images generated on the surface of the object under test at multiple different positions; the surface of the object under test is provided with feature structures for focal plane positioning;

[0021] Based on the interference image, determine the initial focal plane position of the imaging component;

[0022] Near the initial focal plane position, the relative position of the imaging component and the object under test is changed multiple times, and interference images are acquired at the corresponding positions.

[0023] Based on the phase information of the characteristic structures in the interferometric images acquired at each location, the focal plane state at each location is evaluated.

[0024] The optimal focal plane position is determined based on the evaluation results, and the relative position of the imaging component and the object under test is adjusted to the optimal focal plane position.

[0025] In some embodiments, the step of initially determining the initial focal plane position of the imaging component based on the interferometric image includes:

[0026] Transform the interference image to the frequency domain and extract the spatial frequency features of the interference fringes in the image;

[0027] The position with the largest characteristic amplitude in the spatial frequency features is taken as the initial focal plane position.

[0028] In some embodiments, evaluating the focal plane state at each location based on the phase information corresponding to the feature structures in the interferometric images acquired at each location includes:

[0029] Phase extraction and unwrapping are performed on the images of the feature structure regions in the interferometric images acquired at each location to obtain the continuous phase distribution at each location;

[0030] Wavefront reconstruction is performed on the continuous phase distribution at each position to obtain the corresponding wavefront phase distribution at each position;

[0031] Based on the wavefront phase distribution at each location, the gradient map of the characteristic structural region is calculated respectively;

[0032] Based on the gradient map at each location, a corresponding sharpness index is constructed, and the focal plane state at each location is evaluated according to the sharpness index. The location with the optimal sharpness index value is taken as the best focal plane location for precise positioning.

[0033] In some embodiments, calculating the gradient map of the characteristic structural region based on the wavefront phase distribution includes:

[0034] The feature structure region is segmented into an image to generate an effective region mask;

[0035] Calculate the gradient map of the effective region based on the effective region mask.

[0036] In some embodiments, the step of constructing corresponding sharpness indices based on gradient maps at each location, and evaluating the focal plane state at each location based on the sharpness indices, includes:

[0037] Based on the gradient map, extract the boundary gradient peak and the half-width at half-maximum (WHM) of the gradient distribution in the feature structure region;

[0038] A sharpness index is constructed based on the peak value of the boundary gradient and the half-width at half-maximum of the gradient distribution.

[0039] The location corresponding to the maximum value of the sharpness index is determined as the optimal focal plane position.

[0040] In some embodiments, the autofocus method further includes:

[0041] Based on the reconstructed wavefront phase of the interferometric image, the defocus term in the wavefront phase is determined.

[0042] Remove the defocus term from the wavefront phase to obtain the residual wavefront phase, and calculate the wavefront error index of the residual wavefront phase.

[0043] When there is a deviation between the focal plane position corresponding to the optimal value of the wavefront error index and the best focal plane position, the focal plane position corresponding to the optimal value of the wavefront error index shall be determined as the final best focal plane position.

[0044] Thirdly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the autofocus method as described in any embodiment of the second aspect.

[0045] The detection system and its automatic focusing method provided in this application include a light source component, an optical component, and an imaging component. Before detecting the object under test, the relative position between the imaging component and the object under test is first coarsely adjusted, and interference images are acquired at multiple locations. The initial focal plane position of the imaging component is determined based on the acquired interference images. Then, near the initial focal plane position, the relative position between the imaging component and the object under test is changed multiple times, and interference images are acquired at the corresponding locations. Subsequently, the phase information corresponding to the feature structure in the interference image is used to evaluate the focal plane state at each location. Finally, based on the evaluation results, the optimal focal plane position is determined, and the relative position between the imaging component and the object under test is adjusted to the corresponding position to begin detection. This application introduces a focal plane evaluation mechanism based on phase information, elevating the focusing basis from traditional image intensity analysis to phase domain analysis. It fully utilizes the physical characteristic that phase is highly sensitive to defocus changes in interferometry, achieving higher focusing accuracy and anti-interference capability than existing intensity analysis methods without adding any additional hardware, effectively reducing system complexity and implementation costs.

[0046] In addition, this application also provides a computer-readable storage medium that has the same beneficial effects as the autofocus method described above. Attached Figure Description

[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0048] Figure 1 This is a schematic diagram of the structure of a detection system provided in one embodiment of this application.

[0049] Figure 2 This is a schematic diagram of the feature structure on the object under test provided in one embodiment of this application.

[0050] Figure 3 This is a schematic diagram of the structure of a detection system provided in another embodiment of this application.

[0051] Figure 4 This is a flowchart of an automatic focusing method provided in one embodiment of this application.

[0052] Figure 5 A flowchart of an autofocus method provided in another embodiment of this application.

[0053] Figure 6 This is a schematic diagram of the fringe distribution on an interference pattern provided in one embodiment of this application.

[0054] Figure 7 This is a spectral distribution diagram of an interference image at different focal plane positions in the frequency domain, provided as an embodiment of this application.

[0055] Figure 8 A flowchart of an autofocusing method provided in another embodiment of this application.

[0056] Figure 9 This is a phase gradient variation diagram at different focal plane positions provided in one embodiment of this application.

[0057] Figure 10 A flowchart of an autofocusing method provided in another embodiment of this application.

[0058] Figure 11 This is a graph showing the relationship between wavefront RMS residual and defocusing amount in one embodiment of this application.

[0059] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0060] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments. Similar elements in different embodiments are referred to by related similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the present application. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to the present application are not shown or described in the specification. This is to avoid obscuring the core parts of the present application with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0061] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0062] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship. Unless otherwise specified, the terms "connection" and "linkage" used in this application include both direct and indirect connections (linkages).

[0063] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0064] Figure 1 This is a schematic diagram of the structure of a detection system provided in one embodiment of this application. The detection system provided in this embodiment is specifically applied in the field of semiconductor inspection, and aims to solve the problem of the deviation between the surface under test and the focal plane of the imaging system caused by process factors such as wafer warping. By using an automatic focusing function, it ensures that the object under test is always located at the optimal focal plane position during the interferometric measurement process, thereby obtaining clear interference fringes and accurate phase information, and ensuring the reliability and accuracy of tasks such as wafer surface morphology inspection, defect identification, and film thickness measurement.

[0065] like Figure 1 As shown, the detection system provided in this embodiment includes at least a light source assembly 10, an optical assembly 20, an imaging assembly 30, and a processor 40.

[0066] In this embodiment, the light source assembly 10 is used to provide a coherent illumination beam.

[0067] In an interferometric system, the coherence of the light source is a prerequisite for forming a stable interference pattern. Only a beam with a sufficient coherence length can ensure that the test light returning from the surface of the object under test interferes effectively with the reference light, thereby generating an interference image carrying information about the surface topography of the object under test. The light source component 10 can be a helium-neon (He-Ne) laser, a semiconductor laser, or a broadband light source with a narrowband filter, etc. In practice, appropriate light source devices and their matching drive and control circuits can be selected according to the detection accuracy, cost, and system integration requirements.

[0068] Optical component 20 is used to guide the illumination beam to the surface of the object under test and to receive the beam returning from the surface of the object under test to form interference light.

[0069] The optical assembly 20 typically includes one or more optical elements that form the main optical path of the interferometer. Its functions include shaping, guiding, and splitting the illumination beam, as well as combining the returning test beam with the reference beam to ultimately form interference light suitable for focal plane analysis. In other words, the optical assembly 20 transmits and controls the illumination beam, precisely guiding it to the surface of the object under test and collecting the beam returning from the object's surface. Through precise beam combining and optical path configuration, it superimposes the reference beam with the object beam carrying information about the object's surface, thereby forming interference light for analysis.

[0070] To facilitate accurate positioning of the focal plane of the imaging component 30 and ensure clear and accurate imaging, a feature structure for focal plane positioning is provided on the surface of the object under test. This feature structure refers to a mark with a specific geometric shape pre-fabricated on the surface of the object under test (such as a semiconductor wafer). Its core function is to provide a source in the interferometric image that can be identified for subsequent phase analysis. When the illumination beam shines on the surface of the object under test, the region where the feature structure is located will produce a modulation effect on the incident light that is different from the surrounding flat region due to height abrupt changes or differences in material properties. This will then manifest as a regular change in local phase information (such as phase jumps or gradient distribution) in the interferometric image. Based on this quantifiable phase feature, the processor 40 can accurately perceive the degree of deviation between the current focal plane position and the optimal focal plane by analyzing the phase information corresponding to the feature structure. Unlike existing technologies that rely on external independent markers or only utilize overall image intensity information, this embodiment integrates the focal plane positioning function directly into the object under test. On the one hand, this avoids the introduction of additional markers, simplifies the system structure, and reduces hardware costs. On the other hand, since the feature structure and the object under test are both rigid bodies, their focal plane state is completely consistent with the area under test, eliminating the relative position error that may be caused by the separation of the marker and the object under test, and providing a reliable analysis object for subsequent fine-tuning steps based on phase information.

[0071] Figure 2 This is a schematic diagram of a feature structure on an object under test provided in one embodiment of this application. For example... Figure 2 As shown, in some embodiments, the feature structure includes one or more stepped etched lines disposed on the surface of the object to be measured and distributed radially through it; the height of the stepped etched lines does not exceed half the wavelength of the illumination beam.

[0072] In interferometry, focal plane deflection is essentially the axial displacement between the surface of the object under test and the conjugate surface of the imaging system. On a flat surface, this displacement manifests only as a change in the overall density of the interference fringes, making it difficult to provide localized, highly sensitive positioning information. Therefore, a stepped scribe line is set on the surface of the object under test, introducing a region of abrupt change in height. When the focal plane position changes, the optical path difference at the edge of this step will change drastically, appearing as a localized steep change in the phase distribution in the interferometric image. This change has a highly sensitive monotonic relationship with the defocus amount, and thus the processor 40 can accurately sense the focal plane position by analyzing the phase information of the scribe line region.

[0073] Specifically, to ensure that the scribing covers the imaging field of view and provides identifiable boundary features regardless of whether the detection area is located at the center or edge of the wafer, the scribing is designed to be radially distributed along the surface of the object under test. In addition, the height of the scribing also needs to be considered. When the step height exceeds λ / 2, phase unwrapping may result in 2π ambiguity, which will lead to phase information distortion. Therefore, the height of the scribing should be controlled within half of the wavelength of the illumination beam, i.e., not exceeding λ / 2, to ensure that the optical path difference between the reflected light and the reference light is always within a single wavelength range, so as to obtain unambiguous phase information.

[0074] In this embodiment, the imaging component 30 is used to acquire interference light and generate an interference image.

[0075] In an interferometric measurement system, the imaging component 30 typically consists of an image sensor (such as a CCD or CMOS chip) and its associated driving circuitry and signal processing circuitry. The imaging component 30 receives the interference light generated by the optical component 20, converts the optical signal into an electrical signal, and thus acquires and generates an interference image that can be analyzed by the processor 40. The interference image contains topographic information about the object's surface and phase information for positioning. The image quality (including fringe sharpness, contrast, etc.) directly reflects the accuracy of the focal plane position and serves as the primary data source for subsequent focal plane positioning and evaluation.

[0076] In this embodiment, the processor 40 is connected to the imaging component 30 to acquire the interference images it collects. The processor 40 is configured to coarsely adjust the relative position between the imaging component 30 and the object under test, and to acquire interference images at multiple locations. The initial focal plane position is determined based on the acquired interference images. Near the initial focal plane position, the relative position between the imaging component 30 and the object under test is changed multiple times, and interference images are acquired at the corresponding positions. The focal plane state at each position is evaluated based on the phase information corresponding to the feature structures in the interference images acquired at each position. The optimal focal plane position is determined based on the evaluation results, and the relative position between the imaging component 30 and the object under test is adjusted to the optimal focal plane position to begin detection.

[0077] In the semiconductor inspection system, the processor 40, as the core control and computing unit, is mainly responsible for coordinating the collaborative work of various hardware components, including controlling the on and off of the light source component 10, adjusting the illumination intensity, triggering the imaging component 30 to acquire images, and controlling the precise positioning and motion trajectory of the stage. At the data level, the processor 40 performs routine preprocessing on the raw interferometric images acquired by the imaging component 30, such as background correction, noise filtering, and dead pixel compensation, to improve image quality. At the inspection function level, the processor 40 executes core algorithms corresponding to specific inspection objectives, such as extracting phase information from the interferometric image for wavefront reconstruction, calculating the three-dimensional morphology of the surface of the object under test, identifying the location and type of defects, and measuring the film thickness.

[0078] In this embodiment, in addition to the functions described above, the processor 40 is also configured to perform autofocus-related functions. Specifically, the autofocus process of the processor 40 is divided into two stages: coarse adjustment and fine adjustment. In the coarse adjustment stage, the processor 40 drives the relative position between the imaging component 30 and the object under test to change in steps, and acquires interference images at each changed position. Then, based on all the acquired images, it analyzes the information of the interference fringes (such as fringe contrast, frequency domain characteristics, etc.) to make a preliminary judgment on the current focal plane position of the imaging component 30, and determines an initial focal plane position, that is, quickly positioning the focal plane within a small range. Subsequently, in the fine adjustment stage, the processor 40 again changes the relative position within a small range near the initial focal plane position, and acquires interference images at each new position. Since phase information is highly sensitive to defocus changes, the processor 40 further uses each image to extract the phase information corresponding to the surface feature structure of the object under test, and quantifies the focal plane state by evaluating the phase gradient distribution at each position, and finally determines the optimal focal plane position. Finally, based on the precisely located optimal focal plane position, the processor 40 outputs a control command to adjust the imaging component 30 to that position to begin formal detection. The entire process achieves two-stage autofocus from coarse to fine, ensuring that the imaging components are always at the optimal focal plane.

[0079] In summary, the detection system provided in this embodiment divides the focusing process into two stages: coarse adjustment and fine adjustment, based on interferometric images acquired at different focal plane positions. The coarse adjustment stage utilizes frequency domain analysis to quickly locate the region near the focal plane, avoiding blind global searching. The fine adjustment stage leverages the phase information of the surface features of the object under test, using phase gradient as the evaluation criterion, elevating the focusing basis from traditional image intensity analysis to phase domain analysis, accurately perceiving the optimal focal plane position at the nanoscale. These two-stage strategies complement each other, ensuring focusing efficiency while improving positioning accuracy to the nanometer level. Furthermore, the system is entirely implemented using processor algorithms, requiring no additional hardware, significantly improving the automation level and measurement reliability of the semiconductor detection system.

[0080] Figure 3 This is a schematic diagram of the structure of a detection system provided in another embodiment of this application. Figure 3 As shown, the detection system provided in this embodiment includes a light source assembly 10, an optical assembly 20, an imaging assembly 30, a processor 40, and a driving mechanism 50.

[0081] In this embodiment, the drive mechanism 50 is drively connected to the imaging assembly 30 or the stage carrying the object to be measured, and is also connected to the processor 40. The processor 40 is further configured to control the drive mechanism 50 to drive the imaging assembly 30 or the stage to move according to the determined optimal focal plane position, so that the imaging assembly 30 is adjusted to the optimal focal plane position.

[0082] By introducing a drive mechanism 50 to adjust the imaging component 30 to the optimal focal plane position for detection, the drive mechanism 50 is mechanically connected to either the imaging component 30 or the stage used to carry the object under test, enabling it to physically move one of these two components. Simultaneously, it is electrically connected to the processor 40 to receive control commands. When the drive mechanism 50 is connected to the imaging component 30, i.e., after the processor 40 locates the optimal focal plane position based on the phase information corresponding to the interferometric image and the surface feature structure of the object under test, the object under test on the stage remains in its current position, and the processor 40 controls the drive mechanism 50 to drive the imaging component 30 to the optimal focal plane position. Conversely, when the drive mechanism 50 is connected to the stage carrying the object under test, the imaging component 30 remains in its current position, and the processor 40 controls the drive mechanism 50 to drive the stage, moving the object under test so that it maintains the optimal focal plane distance from the imaging component 30 at the new position, thus adjusting the imaging component 30 to the optimal focal plane position. By adding a drive mechanism 50 and establishing its connection with the processor 40 and the imaging component 30 or the stage, a complete closed loop is achieved, converting the determined optimal focal plane position into the actual physical execution of the imaging component 30. The adjustment can be made by moving the imaging component 30 itself via the drive mechanism 50, or by moving the stage carrying the object under test; the specific implementation method is determined by the system's mechanical structure.

[0083] like Figure 2As shown, this embodiment also provides an optical path for an optical component 20 consisting of a beam splitter 201, a collimating lens 202, a reference lens 203, and a repeater 204. Through the optical elements and their sequential positions in the optical path, the illumination beam from the light source component 10 is guided to the surface of the object under test and interference light is formed. Specifically, in the direction of the illumination beam emitted from the light source component 10, the beam splitter 201, the collimating lens 202, and the reference lens 203 are arranged sequentially, with the reference lens 203 positioned opposite to the object under test. In the imaging optical path, the repeater 204 is located between the beam splitter 201 and the imaging component 30, and the beam splitter 201 is simultaneously located between the collimating lens 202 and the repeater 204, thus forming a typical Fizeau interferometer optical path architecture. Based on this, the illumination beam emitted from the light source assembly 10 first enters the beam splitter 201, and after being transmitted through the beam splitter 201, it propagates to the collimating lens 202. The collimated parallel beam after being collimated by the collimating lens 202 enters the reference lens 203, where transmission and reflection occur and the beam is split. The transmitted part forms the test light, which continues to illuminate the surface of the object under test, while the reflected part forms the reference light as an interference reference. After being reflected by the surface of the object under test, the test light carries the surface topography information back and interferes with the reference light near the reference lens 203. The interference light is transmitted through the collimating lens 202, reflected by the beam splitter 201 (due to the change in the direction of the returning beam), and transmitted through the relay lens 204 before finally entering the imaging assembly 30. The reference lens 203 simultaneously serves as both a beam splitter and a reference surface, the collimating lens 202 ensures the parallelism of the beam to guarantee the interference quality, and the relay lens 204 accurately images the interference pattern onto the target surface of the imaging assembly 30.

[0084] Furthermore, the imaging component 30 is coaxially arranged with the relay mirror 204 and can move relative to the relay mirror 204 along the optical axis.

[0085] During setup, the optical center of the imaging assembly 30 is aligned with the optical axis of the repeater 204 to ensure that the interference light, after being imaged by the repeater 204, is accurately projected onto the target center of the imaging assembly 30. This avoids image plane shift or vignetting effects caused by optical axis deviation. Its fixed position also helps maintain the stability of the imaging magnification and aberration correction. Simultaneously, the imaging assembly 30 is configured to move relative to the repeater 204 along the optical axis, granting it adjustment freedom in the focal plane direction. After the processor 40 determines the optimal focal plane position based on interferometric image analysis, the imaging assembly 30 can be translated to that position via the drive mechanism 50 along the optical axis, thereby achieving focusing while keeping the repeater 204 fixed. The axial movement of the imaging assembly 30 directly changes the image plane position, forming a precise object-image conjugate relationship with the surface of the object under test, thus obtaining a clear interferometric image. The coaxial setup ensures that the imaging assembly 30 always receives the complete field of view from the repeater 204 throughout the entire movement process, preventing image occlusion or uneven illumination due to movement.

[0086] In some embodiments, a combination mirror group 205 is also provided between the light source assembly 10 and the beam splitter 201 to preprocess the illumination beam.

[0087] The illumination beam emitted from the light source assembly 10 first enters the combined mirror group 205. After beam expansion, collimation, filtering, and other shaping processes, it then sequentially enters the subsequent optical paths, including the beam splitter 201, collimating mirror 202, and reference mirror 203. Considering that the original laser beam is typically only 1-2 mm in diameter, which may not be able to cover the surface of the wafer under test, and that the laser itself may contain stray light or wavefront distortion, affecting the accuracy of interferometry, the combined mirror group 205 solves the problems of the original laser beam having too small a diameter and unsatisfactory wavefront quality. Through the beam expansion effect of the combined mirror group 205, the beam diameter can be increased to tens of millimeters to match the detection field of view; through collimation correction, the beam wavefront is shaped into an ideal plane wave; and through spatial filtering, stray light and high-frequency noise are effectively suppressed. The high-quality illumination beam pre-processed by the combined mirror group 205 ensures the formation of subsequent interference and the accurate extraction of phase information.

[0088] In some embodiments, the combined mirror group 205 includes at least a collimating beam expander and a compensating mirror. The collimating beam expander is located on the light-emitting side of the light source and is used to receive the original laser beam and expand it into a collimated parallel beam. The compensating mirror is located on the light-emitting side of the collimating beam expander, that is, the beam is collimated and expanded before being incident on the compensating mirror. The function of the collimating beam expander is to expand the narrow beam emitted by the laser to the required aperture and correct the beam divergence angle, making it an approximately ideal plane wave. The function of the compensating mirror is to specifically correct the residual aberrations (such as spherical aberration, chromatic aberration, or local wavefront distortion) after beam expansion and collimation, further improving the wavefront quality of the illumination beam and ensuring that the illumination light incident on the interferometer has higher flatness and uniformity.

[0089] In some embodiments, the combined mirror assembly 205 may further include a spatial filter, a filter, a waveplate or polarizer, and a beam shaping element. The spatial filter, typically composed of a microscope objective and a pinhole, can be positioned after the collimating beam expander or the compensating mirror to filter out stray light and high-frequency noise in the laser, obtaining a clean Gaussian beam. The filter can be placed at any position in the optical path to narrow the spectral bandwidth or attenuate the light intensity, improve interference contrast, or prevent detector saturation. The waveplate or polarizer is used to adjust the polarization state of the beam to adapt to the polarization characteristics of the beam splitter. The beam shaping element (such as a cylindrical mirror or a diffractive optical element) can be placed after the beam expander to convert a circular Gaussian spot into a flat-top beam or an illumination spot of a specific shape to meet the detection requirements of different objects under test.

[0090] It should be noted that the semiconductor inspection system in any of the above embodiments is only an exemplary implementation architecture. In practical applications, depending on the inspection task (such as defect detection, film thickness measurement, or morphology characterization), the system can also integrate other functional modules, such as automatic loading and unloading mechanisms, alignment mark recognition units, and multi-wavelength switching devices. The specific configuration and implementation of these modules are not limited in this application, and those skilled in the art can select them as needed based on existing technologies. The core innovation of this application lies in how the processor 40 achieves high-precision automatic focusing based on interferometric images. That is, through a two-stage strategy, it first coarsely adjusts and quickly locates the vicinity of the focal plane, and then accurately locates the optimal focal plane based on the phase information of the surface feature structure of the object under test. The automatic focusing method of this application will be described in detail step by step below with reference to the accompanying drawings.

[0091] Figure 4 This is a flowchart illustrating an embodiment of the autofocusing method provided in this application. Figure 4 As shown, the automatic focusing method provided in this embodiment is applied to a semiconductor testing system and specifically includes the following steps:

[0092] Step S410: Coarsely adjust the relative position of the imaging component and the object under test, and obtain interference images generated on the surface of the object under test at multiple different positions; the surface of the object under test is provided with feature structures for focal plane positioning.

[0093] To acquire the fundamental data for focal plane analysis, in the initial stage of autofocus, the processor 40 first drives the imaging component 30 or the stage carrying the object under test, causing their relative positions to move stepwise along the optical axis. At each stopping point, the light source component 10 and the imaging component 30 acquire an interference image generated on the surface of the object under test and convert it into digital image data. This interference image contains pre-fabricated feature structure information on the surface of the object under test. The surface of the object under test has pre-set feature structures for focal plane positioning. These feature structures include one or more stepped etched lines distributed radially across the surface of the object under test, introducing a region of abrupt change in height. When the focal plane position changes, the optical path difference at the edge of this step changes drastically, manifesting as a local steep change in phase distribution in the interference image. This change has a highly sensitive monotonic relationship with the defocus amount, allowing the processor 40 to perceive the focal plane position based on the interference image.

[0094] Step S420: Determine the initial focal plane position of the imaging component based on the interference image.

[0095] Considering that accurately locating the optimal focal plane directly from the entire interferometric image requires extensive computation and is prone to getting trapped in local extrema, and that the sharpness of the interference fringes is closely related to the focal plane position (fringes become blurry and contrast decreases when the focal plane deviates, and are sharper and denser when it is closer to the focal plane), after obtaining interferometric images at multiple different relative positions, the processor 40 analyzes all images to select an initial value representing the approximate location of the focal plane, narrowing the focal plane range to the region near the optimal focal plane. Specifically, the processor 40 extracts global features (such as the contrast of the interference fringes, spatial frequency distribution, or spectral energy) that reflect the focal plane state from each image, and by comparing the quality of these feature values ​​at each position, determines the position with the best feature performance as the initial focal plane position. This position is not the final optimal focal plane, but rather an approximate result given in the coarse adjustment stage. Its purpose is to narrow the search range for subsequent fine adjustment and avoid blindly searching throughout the entire process.

[0096] Step S430: Near the initial focal plane position, change the relative position of the imaging component and the object under test multiple times, and acquire interference images at the corresponding positions.

[0097] After coarse localization, the location is only an approximate estimate of the optimal focal plane, not a precise location. To achieve more accurate focal plane localization, this embodiment performs a fine search around the initial location. The optimal focal plane location may be slightly before or after the initial location. A single image can only reflect the focal plane state at a single current location and cannot determine the direction and extent of deviation from the true optimal focal plane. Therefore, after obtaining the initial focal plane location through coarse localization, the process moves to the fine localization stage. The processor 40 will use the initial focal plane location as the center and gradually change the relative positions of the imaging component and the object under test within a small vicinity. Each change involves acquiring one or more interferometric images at the new location. Higher-density sampling is performed within the narrowed spatial range to obtain a sequence of images covering the potential optimal focal plane range, providing more sample data for phase analysis to determine the direction of deviation of the optimal focal plane location from the current focal plane location. In the fine-tuning stage, the step size is usually smaller than that in the coarse-tuning stage. During the movement, equal or variable step sizes are used. The step size is set according to the required focusing accuracy to ensure search efficiency while ensuring that the best focal plane position is not missed due to excessive step size.

[0098] Step S440: Evaluate the focal plane state at each location based on the phase information corresponding to the feature structures in the interferometric images acquired at each location.

[0099] Considering the limitation of traditional image intensity-based evaluation methods where sensitivity tends to saturate near the focal plane, this embodiment, after coarsely adjusting the positioning and placing the imaging component 30 near the optimal focal plane, further utilizes the phase information corresponding to the surface feature structures of the object under test for a refined search. Since phase abrupt changes at the etched line boundaries are extremely sensitive to defocusing, defocusing leads to a decrease in the peak phase gradient and a widening of its width. Therefore, by extracting and analyzing the phase of the interferometric image of the feature structure region, an evaluation index capable of quantifying the degree of focal plane deviation can be constructed, perceiving the minute deviation between the current focal plane and the true optimal focal plane. Specifically, for each location of the acquired interferometric image, the processor 40 extracts the phase features contained in the feature structure region. These phase features have a definite correspondence with the current focal plane state: when the imaging component 30 is closer to the true optimal focal plane, the phase distribution of the feature structure region exhibits a specific pattern (e.g., a steeper phase abrupt change); conversely, the farther away from the focal plane, the weaker the regularity of the phase distribution. By analyzing and quantifying the phase features of each image in the image sequence, the processor 40 assigns an evaluation value that characterizes the focal plane quality to each focal plane position, thereby forming an evaluation result with focal plane position as the independent variable and focal plane quality as the dependent variable.

[0100] Step S450: Determine the optimal focal plane position based on the evaluation results, and adjust the relative position of the imaging component and the object under test to the optimal focal plane position.

[0101] After obtaining the focal plane state evaluation values ​​for all sampling positions during the fine-tuning stage, the magnitudes of these evaluation values ​​are further compared to locate the relative position with the optimal evaluation result. This position is the optimal focal plane position obtained through fine-tuning. Specifically, the processor 40 uses the focal plane quality corresponding to each position as the judgment criterion. By comparing the magnitudes of each evaluation value (such as finding the maximum or minimum value, depending on the construction method of the evaluation index), the processor 40 locates the focal plane position with the optimal evaluation result. This position is the optimal focal plane obtained through fine-tuning. After the optimal focal plane position is accurately located, the processor 40 sends control commands to the drive mechanism 50 to adjust the relative position of the imaging component and the object under test, so that it moves precisely to the optimal focal plane position. At this point, the entire autofocusing process is completed, and the imaging component 30 is precisely adjusted to the optimal focal plane state. The system can then perform subsequent formal testing in this state to ensure that the acquired interference image has the best fringe sharpness and phase accuracy, thereby guaranteeing the reliability of the test results.

[0102] In summary, the autofocusing method provided in this embodiment, based on the interferometric image generated from the surface of the object under test, employs a two-stage strategy combining coarse and fine adjustments. In the coarse adjustment stage, the overall features of the interferometric image are used to quickly narrow the search range and determine the initial focal plane position of the imaging component. In the fine adjustment stage, the high sensitivity of phase information in interferometry to defocus changes is utilized; by analyzing the phase distribution of characteristic structural regions, minute focal plane deviations are accurately detected. Without increasing hardware costs, the focusing accuracy is improved from the micrometer level of traditional methods to the nanometer level. Furthermore, this method uses the interferometric image as the sole data source and is entirely implemented using processor software algorithms, exhibiting good versatility and integrability. It can be widely applied to various interferometric detection systems, providing reliable focal plane position assurance for subsequent high-precision topography measurement, defect detection, and other tasks.

[0103] Figure 5 A flowchart illustrating an autofocus method provided in another embodiment of this application. Figure 5 As shown, in the above embodiment, step S420, determining the initial focal plane position of the imaging component based on the interference image, specifically includes:

[0104] Step S4201: Transform the interference image to the frequency domain and extract the spatial frequency features of the interference fringes in the image.

[0105] Figure 6 This is a schematic diagram of the fringe distribution on an interferogram provided in one embodiment of this application. Figure 6 As shown, the interference image appears as a pattern of alternating bright and dark stripes in the spatial domain. When the focal plane position changes, the density of the stripes (i.e., the number of stripes per unit length) changes regularly; the farther away from the focal plane, the sparser the stripes; the closer to the focal plane, the denser the stripes. To quantify this change, the processor 40 transforms the interference image to the frequency domain using a Fourier transform. Figure 7 This is a spectral distribution diagram in the frequency domain of an interferometric image at different focal plane positions, provided as an embodiment of this application. For example... Figure 7 As shown, in the frequency domain, the density of fringes corresponds to specific spatial frequency components: the denser the fringes, the higher the corresponding spatial frequency; the sparser the fringes, the lower the corresponding spatial frequency. By analyzing the spectral distribution in the frequency domain, spatial frequency features representing the current fringe density can be extracted, such as the frequency value corresponding to the spectral peak or the energy integral of a specific frequency band. Utilizing the sensitivity of frequency domain analysis to periodic signals for image transformation and extraction can eliminate the interference of random noise and local defects in the image, thereby obtaining a stable and reliable focal plane state indication.

[0106] Step S4202: Take the position with the largest feature amplitude in the spatial frequency features as the initial focal plane position.

[0107] According to physical theory, when the imaging component 30 is exactly located at the optimal focal plane, the interference fringes are the clearest and densest, and the corresponding spatial frequency components are the most prominent, manifested as the characteristic amplitude reaching its peak in the frequency domain. Conversely, when the imaging component 30 deviates from the optimal focal plane, the fringes become blurred and sparse, and the characteristic amplitude in the frequency domain decreases accordingly. Therefore, by moving the imaging component 30 within a preset search range and simultaneously acquiring the spatial frequency characteristic amplitude at each position, an evaluation result is obtained with the focal plane position as the independent variable and the characteristic amplitude as the dependent variable. The peak position of the characteristic amplitude is the initial focal plane position. This coarse-adjustment strategy, which maximizes the frequency domain characteristic amplitude, has advantages such as high computational efficiency, strong anti-interference capability, and no need to rely on specific markers, making it an effective means of achieving rapid automatic focusing.

[0108] Figure 8 A flowchart illustrating an autofocusing method provided in another embodiment of this application. Figure 8 As shown, in the above embodiment, step S430, evaluating the focal plane state at each location based on the phase information corresponding to the feature structures in the interferometric images acquired at each location, specifically includes:

[0109] Step S4301: Perform phase extraction and unwrapping on the characteristic structural regions in the interferometric images acquired at each location to obtain the continuous phase distribution at each location.

[0110] The focal plane state information carried in the interferometric image is mainly contained in the phase, rather than the fringe intensity visible to the naked eye. However, the raw phase directly obtained by conventional interferometric acquisition is wrapped in the principal value range of [-π, π] or [0, 2π], presenting a sawtooth-like abrupt distribution, which cannot be directly used for quantitative analysis. Therefore, the processor 40 first extracts the region containing the feature structure (i.e., step lines) from the interferometric image acquired at each focal plane position as the analysis object, and then extracts the wrapped phase distribution of this region through methods such as phase shifting algorithm or spatial carrier demodulation. In order to eliminate the 2π ambiguity caused by phase wrapping, a spatial or temporal unwrapping algorithm is further used to recover the physically continuous and monotonically changing phase distribution. The finally obtained continuous phase distribution directly reflects the optical path difference between the illumination beam after reflection from the surface of the object under test and the reference light. The location of the feature structure exhibits a phase jump due to the height abrupt change. The steepness of this jump is closely related to the initial focal plane position. The closer to the optimal focal plane, the sharper the phase jump. By repeating the above process at all sampling locations, a continuous phase sequence covering the entire search range was obtained.

[0111] Step S4302: Perform wavefront reconstruction on the continuous phase distribution at each position to obtain the wavefront phase distribution at each position.

[0112] Although the continuous phase distribution obtained after unpacking eliminates abrupt changes, its values ​​are still relative phase values ​​in radians and have not yet established a correspondence with the actual physical wavefront. Furthermore, the original phase data may contain noise, sampling point inhomogeneities, and inherent aberrations of the system, which could amplify errors if directly used for gradient calculation. Therefore, the processor 40 further performs wavefront reconstruction on the continuous phase distribution at each location, converting it into a wavefront phase distribution with actual physical length, allowing it to intuitively reflect the morphological features of the surface of the object under test. On the other hand, the wavefront reconstruction process typically employs techniques such as Zernike polynomial fitting or interpolation filtering to smooth and denoise the original phase data, eliminating random errors and outliers, while reconstructing the discrete phase samples into a continuous, smooth wavefront surface. After this processing, each focal plane position corresponds to a high signal-to-noise ratio, high-fidelity wavefront phase distribution, in which phase abrupt changes in characteristic structural regions are more clearly highlighted.

[0113] Step S4303: Calculate the gradient map of the characteristic structure region according to the wavefront phase distribution at each position.

[0114] The characteristic structure (rutted boundary) in the wavefront phase distribution is essentially a point of geometric height abrupt change, mathematically represented by a step change in phase value. This step change is steepest at the optimal focal plane, and gradually smooths and broadens as the defocusing amount increases. To quantify this steepness, for the wavefront phase distribution corresponding to each focal plane position, the first spatial derivative of the phase, i.e., the phase gradient, is calculated along the direction perpendicular to the rutted boundary (i.e., the normal to the rutted boundary).

[0115] The first spatial derivative of the phase can be expressed as: ;in, This represents the phase distribution of a continuous wavefront; It represents the partial differential along the normal direction (normal direction) of the scribe line boundary.

[0116] Figure 9 This is a phase gradient variation diagram at different focal plane positions provided in one embodiment of this application. For example... Figure 9 As shown, this gradient map reflects the rate of change of the wavefront phase in space. At the scribed boundary, the gradient value forms a sharp peak; in the flat areas on both sides of the boundary, the gradient value is close to zero. By calculating the gradient point-by-point across the entire feature structure region, a gradient map is obtained, with the gradient values ​​distributed in grayscale. This gradient map visually presents the steepness of the scribed boundary in numerical form: a higher gradient peak and a narrower peak width indicate a sharper boundary, and the corresponding focal plane state is closer to the optimal state; conversely, a lower gradient peak and a wider width indicate increased defocus. The gradient maps from all sampling locations constitute a quantifiable and comparable dataset, which forms the basis for constructing evaluation metrics.

[0117] In some embodiments, when calculating the gradient map of the feature structure region, the method further includes:

[0118] Step S43031: Perform image segmentation on the feature structure region to generate an effective region mask.

[0119] Step S43032: Calculate the gradient map of the effective region based on the effective region mask.

[0120] Understandably, when calculating the gradient map of a feature structure region based on the wavefront phase distribution, image segmentation is first required to ensure that the gradient calculation accurately reflects the true steepness of the scribe line boundaries and eliminates interference from the surrounding background region. This segmentation process, based on features of the wavefront phase distribution such as phase transition locations, gradient intensity thresholds, or edge detection algorithms, divides the image into a valid region containing the scribe line boundaries and a background region not involved in the analysis, generating corresponding binary masks. Pixels in the valid region are assigned a value of 1, while pixels in the background region are assigned a value of 0. Subsequently, this mask is used to spatially constrain the wavefront phase distribution, calculating the first spatial derivative of the phase only within the valid region marked by the mask, resulting in a gradient map containing only scribe line boundary information. This mask-constrained calculation method effectively eliminates noise or flat phases in the background region from contaminating the gradient results, improving the signal-to-noise ratio of the gradient peak and the accuracy of the full width at half maximum (FWHM) calculation.

[0121] Step S4304: Construct corresponding sharpness indices based on the gradient maps of each location, and evaluate the focal plane state of the corresponding location according to the sharpness indices of each location. The location with the optimal sharpness index is taken as the best focal plane location for precise positioning.

[0122] While gradient maps visually represent the gradient distribution at the scribe line boundaries, they are two-dimensional images and cannot be directly used for quantitative comparisons between different focal plane positions. Therefore, for each gradient map, a comprehensive sharpness index is constructed based on its characteristics (such as gradient concentration, peak prominence, and distribution compactness). This index condenses complex gradient distribution information into a single value reflecting the current focal plane state. The sharpness index reflects that at the optimal focal plane, the scribe line boundaries are steepest, the gradient distribution is most concentrated and prominent, and the corresponding sharpness index should be at its optimal value (e.g., the maximum value). As defocus increases, the boundaries gradually smooth out, the gradient distribution tends to diffuse, and the sharpness index deteriorates accordingly. By repeating this construction process for all sampling locations, a set of discrete data points is obtained, with focal plane position as the independent variable and the sharpness index as the dependent variable. Finally, by comparing the sharpness index values ​​at each position, the focal plane position corresponding to the optimal value (such as the maximum value) is found. This position is the best focal plane for precise positioning in the fine-tuning stage, thus completing the transformation of the abstract gradient distribution into comparable quantitative indicators.

[0123] In some embodiments, a corresponding sharpness index is constructed, specifically including:

[0124] Step S43041: Based on the gradient map, extract the boundary gradient peak and the half-width at half-maximum of the gradient distribution in the feature structure region.

[0125] Step S43042: Construct a sharpness index based on the peak value of the boundary gradient and the half-width at half-maximum of the gradient distribution.

[0126] Step S43043: Determine the position corresponding to the maximum value of the sharpness index as the optimal focal plane position.

[0127] Understandably, in constructing a sharpness index based on gradient maps to evaluate focal plane status, two key parameters are first extracted from the gradient map corresponding to each focal plane position: First, the gradient peak at the feature structure boundary, which reflects the contrast intensity of the scribe line edge in the image; a higher peak indicates a sharper boundary. Second, the half-width at half-maximum (WHM) of the gradient distribution, i.e., the spatial width corresponding to when the gradient value drops to half the peak value; it characterizes the steepness of the boundary transition; a narrower width indicates a sharper boundary. Subsequently, these two parameters are constructed into a comprehensive sharpness index according to specific mathematical relationships (such as ratios, products, or weighted combinations). This index integrates boundary contrast and sharpness information into a single quantified value. Finally, by comparing the sharpness index at different focal plane positions, the position corresponding to the maximum value is the location where the scribe line boundary is steepest and sharpest, which is also the optimal focal plane position of the optical system.

[0128] In some embodiments, the scribe line boundary is steepest at the optimal focal plane, and the gradient peak in the normal direction is highest. The full width at half maximum (FWHM) of the gradient distribution The minimum ratio of the two is chosen as the sharpness index. The ratio S is the maximum value. The ratio form has a normalization effect; as illumination increases, the gradient peak value... and half height and width They change synchronously, but their ratio remains relatively stable, suppressing external interference. As the defocus amount increases, the boundary gradually smooths out, and the gradient peak... Drop, half-width With the focal plane broadened, the ratio S decreases monotonically, making it easier to identify the optimal focal plane position.

[0129] Figure 10 A flowchart illustrating an autofocusing method provided in another embodiment of this application. Figure 10 As shown, the autofocus method provided in this embodiment, based on any of the above embodiments, further includes verifying the optimal focal plane position after determining it, specifically including the following steps:

[0130] Step S1010: Based on the wavefront phase reconstructed from the interferometric image, determine the defocus term in the wavefront phase.

[0131] After completing the fine-tuning and positioning and obtaining the optimal focal plane position, further verification of the optimal focal plane position was added to ensure the accuracy of subsequent detection. The wavefront phase can typically be represented as a linear combination of various fundamental aberrations, where the defocus term is represented as a quadratic surface distribution related to the displacement along the optical axis. In this embodiment, during verification, a polynomial fitting (such as Zernike fitting or least squares fitting) was performed on the reconstructed wavefront phase to extract the coefficients or components representing defocus, thereby quantitatively determining the defocus component contained in the current wavefront phase.

[0132] In some embodiments, phase extraction and unwrapping are performed on the images of the characteristic structural regions in the interferometric images acquired at each location to obtain the continuous phase distribution at each location, and wavefront reconstruction is performed to obtain the wavefront phase distribution.

[0133] The wavefront phase distribution is as follows: ;

[0134] The defocus term is modeled using quadratic polynomial fitting. for:

[0135] ;

[0136] in, The fitting coefficients are related to the numerical aperture and wavelength of the system.

[0137] Step S1020: Remove the defocus term from the wavefront phase to obtain the residual wavefront phase, and calculate the wavefront error index of the residual wavefront phase.

[0138] After determining the defocus term in the wavefront phase, it is subtracted from the original wavefront phase to obtain the residual wavefront phase, which contains only the topographic information of the object under test, the inherent aberrations of the system, and other higher-order aberrations. This eliminates spurious wavefront distortion introduced by focal plane deviation, preventing it from masking the true wavefront quality. Subsequently, wavefront error indices that comprehensively reflect the degree of distortion are calculated on the residual wavefront phase, such as root mean square error (RMS), peak-to-valley ratio (PV), or Strell ratio. The smaller the index, the less severe the wavefront distortion, indicating better image quality at the current focal plane position. This quantitative index transforms the abstract wavefront quality into a comparable numerical value, providing an objective basis for verifying the optimal focal plane position.

[0139] Residual wavefront phase for: ;

[0140] in, This is the true wavefront distortion after removing defocus and tilt from the wavefront phase.

[0141] Step S1030: When there is a deviation between the focal plane position corresponding to the optimal value of the wavefront error index and the best focal plane position, the focal plane position corresponding to the optimal value of the wavefront error index shall be determined as the final best focal plane position.

[0142] Ideally, the optimal focal plane position located based on the sharpness index should coincide with the position of the optimal wavefront error index (e.g., minimum RMS). However, due to factors such as noise, algorithm errors, or local defects in the feature structure, there may be slight deviations between the two. In this case, the processor 40 uses the wavefront error index as a higher confidence evaluation standard, replacing the original optimal focal plane position with the focal plane position corresponding to its optimal value, as the final determined focal plane. The verification mechanism forms a complete closed loop from coarse adjustment to fine adjustment to confirmation, effectively avoiding the random errors that may exist in a single focusing, improving the reliability and robustness of the system focusing, and ensuring that the final detection is performed under the true optimal focal plane condition.

[0143] In some embodiments, the root mean square error (RMS) is chosen as the wavefront verification metric to comprehensively assess the degree of wavefront distortion across the entire effective region. After removing the defocus term, the RMS of the residual wavefront phase mainly includes the true surface morphology of the object under test, the inherent aberrations of the system, and random noise. Figure 11 This is a graph showing the relationship between wavefront RMS residual and defocusing amount according to one embodiment of this application. Figure 11 As shown, when the focal plane deviates from the optimal position, even after fitting and filtering, some out-of-focus components will remain in the residual, leading to an increase in RMS and image blurring. Therefore, the change in RMS value is monotonically positively correlated with the amount of out-of-focus, making it an ideal indicator for verifying focal plane shift.

[0144] Specifically, the expression for RMS is: ;

[0145] Where N is the number of valid pixels (excluding invalid areas such as dark edges and defective areas).

[0146] During verification, the optimal focal plane position based on the sharpness index is taken as the zero position, and the RMS before and after this position is calculated to confirm that the position with the minimum residual is the optimal focal plane position.

[0147] In summary, the semiconductor detection system and its autofocus method provided in this application include a light source assembly, an optical assembly, an imaging assembly, and a processor. The surface of the object under test has pre-defined feature structures (such as stepped lines) for focal plane positioning. The processor is configured to perform two-stage autofocus: first, coarse adjustment of the relative position between the imaging assembly and the object under test, acquiring interference images at multiple locations, and determining the initial focal plane position of the imaging assembly based on the acquired interference images; then, near the initial focal plane position, repeatedly changing the relative position between the imaging assembly and the object under test, and acquiring interference images at the corresponding locations; subsequently, evaluating the focal plane state at each location using the phase information corresponding to the feature structures in the interference images, and finally determining the optimal focal plane position based on the evaluation results. After fine adjustment, the optimal focal plane is further verified and corrected by removing defocus terms from the wavefront error index to ensure the reliability of the final focal plane position. This application improves the focusing criterion from traditional image intensity analysis to phase domain analysis, making full use of the high sensitivity of phase to defocusing. It achieves nanometer-level precision autofocus without the need for additional hardware, effectively solving the problems of insufficient focusing accuracy and poor reliability in the prior art, and providing a reliable focal plane guarantee for high-precision semiconductor detection.

[0148] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of any embodiment of the above-described autofocus method and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0149] The processor can be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0150] Those skilled in the art will understand that all or part of the functions of the various methods in the above embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, which may include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to achieve the above functions. For example, the program can be stored in the memory of a device, and when the program in the memory is executed by the processor, all or part of the above functions can be achieved. In addition, when all or part of the functions in the above embodiments are implemented by computer programs, the program can also be stored in a server, another computer, disk, optical disk, flash drive, or external hard drive, etc., and can be downloaded or copied to the memory of a local device, or the system of the local device can be updated. When the program in the memory is executed by the processor, all or part of the functions in the above embodiments can be achieved.

[0151] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art, under the guidance of this application, can make several simple deductions, modifications or substitutions based on the spirit of this application and the scope of protection of the claims without departing from the spirit of this application and the claims. All of these are within the protection scope of this application.

Claims

1. A detection system, characterized in that, include: A light source assembly for providing a coherent illumination beam; An optical component is used to guide the illumination beam to the surface of the object under test and to receive the beam returning from the surface of the object under test to form interference light; the surface of the object under test is provided with a feature structure for focal plane positioning. An imaging component is used to acquire the interference light and generate an interference image; The processor, connected to the imaging component, is configured to: The relative position between the imaging component and the object under test is coarsely adjusted, and interference images are acquired at multiple locations. The initial focal plane position is determined based on the acquired interference images. Near the initial focal plane position, the relative position between the imaging component and the object under test is changed multiple times, and interference images are acquired at the corresponding locations. The focal plane state at each location is evaluated based on the phase information corresponding to the feature structure in the interference images acquired at each location. The optimal focal plane position is determined based on the evaluation results, and the relative position between the imaging component and the object under test is adjusted to the optimal focal plane position to begin detection.

2. The detection system according to claim 1, characterized in that, The feature structure includes one or more stepped etched lines disposed on the surface of the object to be tested and distributed radially; the height of the stepped etched lines does not exceed half the wavelength of the illumination beam.

3. The detection system according to claim 1, characterized in that, The optical components include a beam splitter, a collimating lens, a reference lens, and a relay lens; The illumination beam emitted from the light source assembly is transmitted sequentially through a beam splitter and a collimating lens before being incident on a reference mirror. A portion of the beam transmitted through the reference mirror forms a test beam that is incident on the surface of the object under test, and a portion of the beam reflected by the reference mirror forms a reference beam. The test beam is reflected by the surface of the object under test to form a reflected beam. The reflected beam interferes with the reference beam and is transmitted sequentially through the collimating lens, reflected by the beam splitter, and transmitted through the relay mirror before being incident on the imaging assembly.

4. The detection system according to claim 3, characterized in that, The imaging component is coaxially arranged with the relay mirror and can move relative to the relay mirror along the optical axis.

5. The detection system according to claim 1, characterized in that, It also includes a drive mechanism, which is connected to the imaging component or the stage carrying the object under test, and is also connected to the processor; The processor is also configured to control the drive mechanism to move the imaging component or stage according to the determined optimal focal plane position, so that the imaging component is adjusted to the optimal focal plane position.

6. An automatic focusing method applied to a detection system, said detection system comprising at least a light source assembly and an imaging assembly; characterized in that, The automatic focusing method includes: The relative position of the coarse-adjusted imaging component and the object under test is used to acquire interference images generated on the surface of the object under test at multiple different positions; the surface of the object under test is provided with feature structures for focal plane positioning; Based on the interference image, determine the initial focal plane position of the imaging component; Near the initial focal plane position, the relative position of the imaging component and the object under test is changed multiple times, and interference images are acquired at the corresponding positions. Based on the phase information of the characteristic structures in the interferometric images acquired at each location, the focal plane state at each location is evaluated. The optimal focal plane position is determined based on the evaluation results, and the relative position of the imaging component and the object under test is adjusted to the optimal focal plane position.

7. The automatic focusing method according to claim 6, characterized in that, Determining the initial focal plane position of the imaging component based on the interferometric image includes: Transform the interference image to the frequency domain and extract the spatial frequency features of the interference fringes in the image; The position with the largest characteristic amplitude in the spatial frequency features is taken as the initial focal plane position.

8. The automatic focusing method according to claim 6, characterized in that, The step of evaluating the focal plane state at each location based on the phase information corresponding to the feature structures in the interferometric images acquired at each location includes: Phase extraction and unwrapping are performed on the images of the feature structure regions in the interferometric images acquired at each location to obtain the continuous phase distribution at each location; Wavefront reconstruction is performed on the continuous phase distribution at each position to obtain the corresponding wavefront phase distribution at each position; Based on the wavefront phase distribution at each location, the gradient map of the characteristic structural region is calculated respectively; Based on the gradient map at each location, a corresponding sharpness index is constructed, and the focal plane state at each location is evaluated according to the sharpness index. The location with the optimal sharpness index value is taken as the best focal plane location for precise positioning.

9. The automatic focusing method according to claim 8, characterized in that, Based on the wavefront phase distribution, the gradient map of the characteristic structural region is calculated, including: The feature structure region is segmented into an image to generate an effective region mask; Calculate the gradient map of the effective region based on the effective region mask.

10. The automatic focusing method according to claim 8, characterized in that, The process involves constructing corresponding sharpness indices based on the gradient maps at each location, and evaluating the focal plane state at each location based on these sharpness indices, including: Based on the gradient map, extract the boundary gradient peak and the half-width at half-maximum (WHM) of the gradient distribution in the feature structure region; A sharpness index is constructed based on the peak value of the boundary gradient and the half-width at half-maximum of the gradient distribution. The location corresponding to the maximum value of the sharpness index is determined as the optimal focal plane position.

11. The automatic focusing method according to claim 8, characterized in that, Also includes: Based on the reconstructed wavefront phase of the interferometric image, the defocus term in the wavefront phase is determined. Remove the defocus term from the wavefront phase to obtain the residual wavefront phase, and calculate the wavefront error index of the residual wavefront phase. When there is a deviation between the focal plane position corresponding to the optimal value of the wavefront error index and the best focal plane position, the focal plane position corresponding to the optimal value of the wavefront error index shall be determined as the final best focal plane position.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the autofocus method as described in any one of claims 6 to 11.

Images