A method for detecting surface defects of a quartz tube for a semiconductor oxidation furnace

Abstract

The application discloses a kind of quartz tube surface defect detection methods for semiconductor oxidation furnace, it is related to optical detection technical field, the method includes controlling quartz tube rotation and changing the direction of incidence of illumination light, polarization state and wavelength parameter, obtain the multi-time sequence optical response data of same detection area under different optical excitation conditions;Establish optical response evolution sequence, extract response change characteristics and construct optical response evolution constraint model, generate detection area response credibility distribution;According to response credibility reconstruction real defect response field, extract defect boundary continuity, internal response consistency and edge response gradient and other characteristics, establish defect evolution model, realize the joint determination of quartz tube surface defect type, size, depth and expansion trend.The application improves the distinguishing ability of real defect and abnormal optical response and the accuracy and reliability of quartz tube surface defect detection.

Description

Technical Field

[0001] This invention relates to the field of optical inspection technology, and in particular to a method for detecting surface defects in quartz tubes used in semiconductor oxidation furnaces. Background Technology

[0002] As a core component in semiconductor oxidation furnaces, quartz tubes are subjected to high temperatures, intense thermal cycling, and chemical atmospheres for extended periods, making their surfaces prone to defects such as cracks, crystallization, corrosion, and contamination. These defects not only affect the stability of the internal temperature and gas flow fields of the oxidation furnace but also lead to quartz particle shedding, process contamination, and reduced wafer yield. Therefore, regular inspection of the quartz tube surface condition is necessary.

[0003] Existing methods for detecting defects in quartz tubes mainly include manual visual inspection, microscopic imaging inspection, and automatic inspection methods based on a single optical signal. Among them, manual visual inspection relies on the experience of the inspectors, and the inspection results are easily affected by the observation angle, ambient lighting, and subjective factors of the personnel, making it difficult to achieve high-precision and consistent inspection. Although microscopic imaging methods can obtain local morphological information of the quartz tube surface, due to the high light transmittance and smooth surface characteristics of quartz material, it is easily affected by specular reflection, refraction, and internal scattering during the inspection process, making it difficult to effectively distinguish between real defects and optical interference signals.

[0004] Existing detection methods based on optical image analysis typically employ fixed light source conditions to image the surface of a quartz tube, identifying defect areas by analyzing grayscale changes, texture features, or edge information in the image. However, quartz tubes are curved, transparent materials, exhibiting varying light reflection and propagation paths at different locations. The same defect may produce significantly different optical responses under different detection angles and lighting conditions, while non-defect areas may also produce anomalous responses similar to defects due to changes in optical reflection. Therefore, relying solely on static image information under a single lighting condition is prone to missed or false detections of defects.

[0005] Furthermore, existing detection methods typically focus on the defect image features at a single moment, making judgments solely based on the spatial morphology of the defect area. They lack analysis of the changes in the optical response of the quartz tube surface under varying detection conditions, failing to effectively utilize the continuous response patterns exhibited by the defect area under different optical conditions. Simultaneously, existing methods struggle to comprehensively analyze the spatial correlation between the defect area and its surrounding areas, taking into account the curved surface structure characteristics of the quartz tube. This results in insufficient detection capabilities for microcracks, early crystallization, and low-contrast defects.

[0006] Therefore, in order to address the problems of unstable optical response, significant interference from curved surface reflection, difficulty in distinguishing between real defects and abnormal optical signals, and insufficient defect status evaluation capabilities in the surface inspection process of quartz tubes used in semiconductor oxidation furnaces, it is urgent to propose a method that can acquire multi-condition optical response information and achieve highly reliable defect detection based on the optical response variation law. Summary of the Invention

[0007] In view of the above-mentioned problems, this invention proposes a method for detecting surface defects in quartz tubes used in semiconductor oxidation furnaces. This method solves the problems of strong optical reflection interference from curved surfaces of quartz tubes, unstable defect response under single detection conditions, and difficulty in distinguishing between real and pseudo defects in the existing technology. It realizes high-precision detection of surface defects in quartz tubes and comprehensive evaluation of defect status based on multi-condition optical response evolution analysis.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] In a first aspect, embodiments of the present invention provide a method for detecting surface defects of a quartz tube used in a semiconductor oxidation furnace, comprising: controlling the quartz tube to be tested to rotate continuously around an axis, and changing the incident direction, polarization state and wavelength parameters of the irradiated light, continuously scanning the surface of the quartz tube, and acquiring multi-time-series optical response data of the same detection area on the surface of the quartz tube under different detection conditions;

[0010] Based on the multi-time series optical response data, an optical response evolution sequence of the detection area on the surface of the quartz tube is established, and the response change relationship between adjacent detection conditions in the optical response evolution sequence is extracted to generate response evolution features.

[0011] Based on the aforementioned response evolution characteristics, an optical response evolution constraint model for the quartz tube surface is established. The stability, continuity, and repeatability of the optical response change process in each detection area are analyzed, and the response reliability distribution of the detection area is generated.

[0012] The true defect response field of the quartz tube surface is reconstructed based on the response confidence distribution. The continuity of the defect boundary, the consistency of the response inside the defect and the response gradient of the defect edge are calculated using the optical response evolution law of the defect region in the true defect response field, so as to obtain the true defect characteristics of the quartz tube surface.

[0013] Based on the actual defect characteristics, a defect evolution model for the quartz tube surface is established. The defect type, defect size, defect depth, and defect expansion trend are jointly determined, and the detection results of the quartz tube surface defects are output.

[0014] As a preferred embodiment of the surface defect detection method for a quartz tube used in a semiconductor oxidation furnace according to the present invention, the method involves: changing the incident direction, polarization state, and wavelength parameters of the irradiated light to continuously scan the surface of the quartz tube, and acquiring multi-time-series optical response data of the same detection area on the surface of the quartz tube under different detection conditions, including:

[0015] Establish a synchronous control relationship between the rotational motion of the quartz tube and optical scanning, divide the outer surface of the quartz tube into multiple continuous detection areas, and set a unique position identifier for each detection area;

[0016] According to a preset optical excitation sequence, the incident direction, polarization state, and wavelength parameters of the irradiation light are combined and switched so that each detection area receives multiple sets of different optical excitations in sequence; after each set of optical excitations is completed, the reflected light response, scattered light response, and transmitted light response of the detection area are obtained.

[0017] The reflected light response, scattered light response and transmitted light response obtained under each set of optical excitations are correlated and integrated according to the position identifier, and the optical response sequence of the detection area is constructed according to the execution order of the optical excitation sequence. Each set of optical excitations and a set of optical responses together form an optical response state.

[0018] The continuity and integrity between adjacent optical response states in the optical response sequence are detected, and the optical response states that meet the requirements for continuous acquisition are stored sequentially to form multi-temporal optical response data of the detection area.

[0019] As a preferred embodiment of the surface defect detection method for quartz tubes used in semiconductor oxidation furnaces according to the present invention, the method includes: establishing an optical response evolution sequence of the detection area on the surface of the quartz tube based on multi-time-series optical response data, and extracting the response change relationship between adjacent detection conditions in the optical response evolution sequence to generate response evolution features, including:

[0020] Read the multi-temporal optical response data of each detection area, arrange the optical response states according to the execution order of the optical excitation sequence, and establish the optical response evolution sequence of the detection area;

[0021] Traverse adjacent optical response states in the optical response evolution sequence, calculate the magnitude, direction, and rate of change of reflected light response, scattered light response, and transmitted light response between adjacent optical response states, and form optical response change information;

[0022] Based on the correlation between changes in multiple consecutive optical response states, continuous evolution analysis is performed on the optical response change information to obtain the stable trend, fluctuation trend, and turning trend of optical response change.

[0023] By integrating the optical response change information, as well as the stable trend, fluctuation trend, and turning trend of the optical response change, the response evolution characteristics of the detection area are generated.

[0024] As a preferred embodiment of the surface defect detection method for quartz tubes used in semiconductor oxidation furnaces according to the present invention, the method includes: establishing a constraint model for the optical response evolution of the quartz tube surface based on response evolution characteristics, including:

[0025] The response evolution characteristics of each detection area are obtained, and the response change amplitude, change direction, change rate and continuous change trend in the response evolution characteristics are correlated to establish the evolution relationship of the optical response of the detection area with the optical excitation.

[0026] Based on the spatial adjacency relationship between different detection areas on the surface of the quartz tube, spatial correlation constraints between detection areas are constructed, and the spatial correlation constraints are fused with the evolutionary relationship to generate an optical response evolution constraint model for the surface of the quartz tube.

[0027] As a preferred embodiment of the surface defect detection method for quartz tubes used in semiconductor oxidation furnaces according to the present invention, the method includes: analyzing the stability, continuity, and repeatability of the optical response change process in each detection area, and generating a response reliability distribution for the detection area, including:

[0028] Obtain the optical response evolution constraint values ​​of each detection area, and extract the stable characteristics of response change, the continuous characteristics of spatial evolution, and the repetitive characteristics of response during multiple scans of the detection area under continuous optical excitation.

[0029] The optical response stability index of the detection area is calculated based on the response change stability characteristics, the optical response continuity index of the detection area is calculated based on the response change relationship between continuous optical excitation states, and the optical response repeatability index of the detection area is calculated based on the response consistency of the same detection area in different scanning cycles.

[0030] The optical response stability index, optical response continuity index, optical response repeatability index, and optical response evolution constraint value are fused to generate the response confidence level of the detection area.

[0031]

[0032] in, To assess the confidence level of the response in region i, For the optical response evolution constraint value, This is an indicator of optical response stability. As an indicator of optical response continuity, As an indicator of optical response repeatability, , , , For weighting;

[0033] Spatial mapping is performed based on the response confidence of each detection area to generate the response confidence distribution on the quartz tube surface. Areas whose response confidence meets the preset confidence conditions are identified as valid response areas, while areas whose response confidence is lower than the preset confidence conditions are identified as abnormal response areas.

[0034] As a preferred embodiment of the method for detecting surface defects of a quartz tube for a semiconductor oxidation furnace according to the present invention, the method for reconstructing the true defect response field of the quartz tube surface based on the response confidence distribution includes: acquiring the response confidence and optical response data of each detection area on the surface of the quartz tube, spatially mapping the spatial position, optical response intensity and response confidence of the detection area, and generating the initial response distribution of the quartz tube surface.

[0035] Based on the aforementioned response confidence level, the optical response data in the initial response distribution are subjected to confidence-weighted processing to obtain a reliable optical response distribution.

[0036] Based on the surface spatial continuity of the quartz tube, the defect response continuity of the reliable optical response distribution is reconstructed, and the response gradient changes, spatial extension directions and response connectivity between adjacent detection areas are analyzed to generate a defect response spatial correlation structure.

[0037] The defect response spatial correlation structure is fused with the reliable optical response distribution to reconstruct the true defect response field on the surface of the quartz tube, and the defect region range is determined based on the true defect response field.

[0038] As a preferred embodiment of the surface defect detection method for a quartz tube used in a semiconductor oxidation furnace according to the present invention, the method includes: calculating the continuity of the defect boundary, the consistency of the internal response of the defect and the response gradient of the defect edge by utilizing the optical response evolution law of the defect region in the real defect response field to obtain the real defect characteristics of the quartz tube surface, including: obtaining a continuous response region in the real defect response field that meets the defect response conditions, and determining the defect region boundary according to the spatial connection relationship between the detection points in the continuous response region.

[0039] Based on the response change relationship between adjacent detection points on the defect region boundary, calculate the defect boundary continuity index:

[0040]

[0041] in, This is an indicator of the continuity of the defect boundary, where N is the number of defect boundary detection points. This represents the change in defect response between adjacent boundary detection points;

[0042] The defect boundary continuity index is used to determine whether the defect boundary has continuous expansion characteristics.

[0043] Based on the defect response values ​​and optical response evolution characteristics at each detection location within the defect region, the internal response consistency index of the defect is calculated:

[0044]

[0045] in, This serves as an indicator of the consistency of internal responses to defects. This represents the response value within the defect area. The degree of response fluctuation within the defect area. ε is the average response value within the defect area, and ε is the correction parameter.

[0046] Calculate the defect edge response gradient based on the degree of response change at the defect edge location:

[0047] in, The gradient of the defect edge response. This represents the response value within the defect area. This represents the response value of the area adjacent to the defective region. The spatial distance from the defect edge;

[0048] By integrating the defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution law of the defect region, the true defect characteristics of the quartz tube surface are generated.

[0049] As a preferred embodiment of the surface defect detection method for quartz tubes used in semiconductor oxidation furnaces according to the present invention, the method includes: establishing a surface defect evolution model of the quartz tube based on actual defect characteristics, and jointly determining the defect type, defect size, defect depth, and defect propagation trend, including:

[0050] Obtain defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution characteristics from real defect features, and construct a defect state feature set.

[0051] Based on the defect state feature set, a defect spatial state description relationship is established, and the spatial distribution of the defect region, the change of response intensity, and the evolution process of optical response are integrated to form a defect evolution state model.

[0052] Based on the defect evolution state model, the spatial scale parameters, response change parameters, and boundary expansion parameters of the defect region are extracted. The comprehensive defect evaluation index is calculated. Based on the comprehensive defect evaluation index and the defect evolution state model, the defect type, defect size, defect depth, and defect expansion trend of the quartz tube surface are jointly determined, and the defect detection results of the quartz tube surface are output.

[0053] In a second aspect, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, they implement the steps of a method for detecting surface defects of a quartz tube for a semiconductor oxidation furnace as described in the first aspect of the present invention.

[0054] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, they implement the steps of a method for detecting surface defects in a quartz tube for a semiconductor oxidation furnace as described in the first aspect of the present invention.

[0055] The beneficial effects of this invention are as follows: This invention obtains multi-time-series optical response data of the quartz tube surface by changing the optical excitation conditions, and establishes a defect detection constraint relationship based on the optical response evolution law, thereby reducing the influence of quartz tube surface reflection, refraction and transmission interference on the detection results and improving the ability to distinguish between real defects and abnormal optical responses; by reconstructing the real defect response field and extracting defect evolution characteristics, a comprehensive evaluation of the location, type, size, depth and expansion trend of defects on the quartz tube surface is achieved, thereby improving the accuracy and reliability of defect detection in quartz tubes used in semiconductor oxidation furnaces. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0058] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0059] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0060] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0061] Example 1

[0062] Reference Figure 1 This is the first embodiment of the present invention, which provides a method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace, comprising:

[0063] S1: Control the quartz tube under test to rotate continuously around the axis and change the incident direction, polarization state and wavelength parameters of the irradiated light to continuously scan the surface of the quartz tube and obtain multi-time-series optical response data of the same detection area on the surface of the quartz tube under different detection conditions.

[0064] Furthermore, a synchronous control relationship between the rotational motion of the quartz tube and optical scanning is established, the outer surface of the quartz tube is divided into multiple continuous detection areas, and a unique position identifier is set for each detection area.

[0065] According to a preset optical excitation sequence, the incident direction, polarization state, and wavelength parameters of the irradiation light are combined and switched so that each detection area receives multiple sets of different optical excitations in sequence; after each set of optical excitations is completed, the reflected light response, scattered light response, and transmitted light response of the detection area are obtained.

[0066] The reflected light response, scattered light response and transmitted light response obtained under each set of optical excitations are correlated and integrated according to the position identifier, and the optical response sequence of the detection area is constructed according to the execution order of the optical excitation sequence. Each set of optical excitations and a set of optical responses together form an optical response state.

[0067] The continuity and integrity between adjacent optical response states in the optical response sequence are detected, and the optical response states that meet the requirements for continuous acquisition are stored sequentially to form multi-temporal optical response data of the detection area.

[0068] It should be noted that in this embodiment, firstly, the quartz tube to be tested is mounted on a rotating testing platform, and both ends of the quartz tube are fixed by a rotating clamping mechanism, so that the axis of the quartz tube is coaxial with the rotation center of the rotating testing platform. During the testing process, the rotating testing platform is controlled to drive the quartz tube to rotate continuously around its own axis at a preset speed, so that the outer circumferential surface of the quartz tube passes through the optical testing area in sequence.

[0069] During the rotation of the quartz tube, a synchronous control relationship is established between the rotational motion and the optical scanning process. Specifically, based on the angular position information of the rotating detection platform and the sampling period of the optical scanning device, the outer surface of the quartz tube is divided into multiple continuous detection areas along the axial and circumferential directions. Each detection area corresponds to a unique position identifier, which is used to record the spatial position of the detection area on the surface of the quartz tube. For example, the outer surface of the quartz tube is divided into multiple scanning bands arranged along the axial direction, and within each scanning band, detection units are further divided according to the circumferential rotation angle, so that each detection unit corresponds to a fixed area on the surface of the quartz tube. When the quartz tube rotates to the target angular position, the optical scanning device acquires the optical response of the current position area and records the corresponding detection area through the position identifier.

[0070] Furthermore, the optical detection module is controlled to perform multi-condition scanning according to a preset optical excitation sequence. The optical excitation sequence includes combinations of different incident directions, different polarization states, and different wavelength parameters.

[0071] In the specific detection process, the light source is first controlled to emit detection light onto the surface of the quartz tube with a first incident direction, a first polarization state, and a first wavelength parameter. The reflected light response, scattered light response, and transmitted light response of the quartz tube surface under the detection conditions are collected. Then, the optical excitation conditions are switched, such as adjusting the incident angle, changing the polarization direction, or switching the detection wavelength, and the optical response is collected again for the same detection area.

[0072] By using the above method, the same detection area is subjected to multiple sets of different optical excitation conditions in sequence, and the corresponding sets of optical response data are obtained.

[0073] Furthermore, the optical response data acquired under different optical excitation conditions are correlated. Specifically, based on the location identifier of the detection area, the reflected light response, scattered light response, and transmitted light response obtained from the same detection area under different optical excitation conditions are matched and arranged according to the execution order of the optical excitation sequence.

[0074] For example, for detection area A:

[0075] Under the first optical excitation condition, the following responses were obtained: first reflected light response; first scattered light response; first transmitted light response.

[0076] The following responses were obtained under the second optical excitation condition: second reflected light response; second scattered light response; second transmitted light response;

[0077] The above response data are then combined according to the detection order to form the optical response sequence corresponding to detection area A.

[0078] Furthermore, the continuity and integrity of the formed optical response sequence are detected.

[0079] The continuity detection is used to determine whether adjacent optical response states are acquired continuously according to a preset detection order, avoiding interruptions in the response sequence due to changes in rotational speed, optical switching delays, or acquisition anomalies. The integrity detection is used to determine whether a preset number of optical excitation acquisitions have been completed in a detection area, avoiding insufficient subsequent analysis data due to missing optical conditions. When the number of optical response states in a detection area meets the preset acquisition number and there are no abnormal intervals between adjacent optical response states, the optical response sequence corresponding to that detection area is stored.

[0080] Finally, for each detection area on the surface of the quartz tube, multi-time-series optical response data arranged in the order of optical excitation change were obtained. The multi-time-series optical response data included: detection area location identifier;

[0081] Optical excitation condition information; reflected light response information; scattered light response information; transmitted light response information; optical response acquisition time information.

[0082] The multi-time series optical response data is used to subsequently establish the optical response evolution sequence of the quartz tube surface and further analyze the optical response change law under different detection conditions, so as to distinguish between the real defects on the quartz tube surface and the optical interference response.

[0083] S2: Based on the multi-time series optical response data, establish the optical response evolution sequence of the detection area on the surface of the quartz tube, and extract the response change relationship between adjacent detection conditions in the optical response evolution sequence to generate response evolution features.

[0084] Furthermore, multi-temporal optical response data of each detection area are read, and the optical response states are arranged according to the execution order of the optical excitation sequence to establish the optical response evolution sequence of the detection area;

[0085] Traverse adjacent optical response states in the optical response evolution sequence, calculate the magnitude, direction, and rate of change of reflected light response, scattered light response, and transmitted light response between adjacent optical response states, and form optical response change information;

[0086] Based on the correlation between changes in multiple consecutive optical response states, continuous evolution analysis is performed on the optical response change information to obtain the stable trend, fluctuation trend, and turning trend of optical response change.

[0087] By integrating the optical response change information, as well as the stable trend, fluctuation trend, and turning trend of the optical response change, the response evolution characteristics of the detection area are generated.

[0088] It should be noted that in this embodiment, firstly, multi-temporal optical response data corresponding to each detection area is read. The multi-temporal optical response data includes the location identifier of the detection area, optical excitation condition information, and the corresponding acquired reflected light response, scattered light response, and transmitted light response.

[0089] Based on the execution order of the optical excitation sequence recorded during the acquisition process, multiple optical response states corresponding to the same detection area are sorted so that the optical response states acquired under different optical excitation conditions form a continuous arrangement relationship according to the detection process.

[0090] For example, for detection area A, when detection is performed sequentially using the first incident direction, the second incident direction, and the third incident direction, the following optical response states are obtained: the first optical response state; the second optical response state; and the third optical response state.

[0091] Each optical response state includes the reflected light response, scattered light response, and transmitted light response under the corresponding detection conditions.

[0092] Arrange the above multiple optical response states in the order of optical excitation change to form an optical response evolution sequence of first optical response state → second optical response state → third optical response state.

[0093] Furthermore, an ergonomic analysis is performed on adjacent optical response states in the optical response evolution sequence.

[0094] Specifically, for two adjacent optical response states, the changes in reflected light response, scattered light response, and transmitted light response are compared respectively.

[0095] For example, in the first optical response state: the reflected light response is R1; the scattered light response is S1; and the transmitted light response is T1.

[0096] In the second optical response state: the reflected light response is R2; the scattered light response is S2; and the transmitted light response is T2.

[0097] By comparing the differences between the two sets of response states, we can obtain: the magnitude of the change in reflected light response; the magnitude of the change in scattered light response; and the magnitude of the change in transmitted light response.

[0098] At the same time, the direction of change of the corresponding optical response is determined based on whether the response value increases or decreases.

[0099] For example, when the scattered light response is enhanced in the second optical response state compared to the first optical response state, the direction of the enhanced scattered light response is recorded; when the reflected light response is reduced, the direction of the weakened reflected light response is recorded.

[0100] Furthermore, based on the time interval between adjacent optical response states and the optical excitation switching sequence, the rate of change during the response change process is calculated to describe how fast the detection area responds to changes in optical excitation.

[0101] Through the above processing, information on the changes in optical response between adjacent detection conditions is obtained.

[0102] Furthermore, the relationship between the changes in multiple consecutive optical response states is analyzed.

[0103] Specifically, the correlation analysis is performed on multiple consecutive optical response changes according to the order of the optical response evolution sequence. When multiple consecutive optical response states maintain the same or similar direction of change and the magnitude of change is within a stable range, it is determined that the detection region has a stable evolution trend.

[0104] For example, if a certain detection area exhibits the following characteristics under different wavelength excitations: the reflected light response continuously decreases; the scattered light response continuously increases; and the transmitted light response shows a relatively small change, then it is considered that the area has a continuous and stable optical response change trend.

[0105] When there are large fluctuations in the magnitude of changes between adjacent response states, but the overall direction of change remains consistent, it is determined that the region has a fluctuating evolution trend.

[0106] When the direction of change in optical response is significantly reversed, for example:

[0107] If the scattered light response increases continuously in the preceding state but suddenly decreases in the subsequent state, then the region is recorded as having a response inflection trend.

[0108] Based on the above analysis, the following trends were obtained in the process of optical response change in the detection area: stable trend, fluctuating trend, and turning point trend.

[0109] Furthermore, the change information between adjacent optical response states is fused with the continuous evolution analysis results to generate response evolution characteristics of the detection area.

[0110] The response evolution characteristics include: response change magnitude characteristics; response change direction characteristics; response change rate characteristics; response stability trend characteristics; response fluctuation trend characteristics; and response turning point trend characteristics.

[0111] Finally, the above features are associated and stored according to the location identifier of the detection area to form a set of response evolution features corresponding to each detection area on the surface of the quartz tube.

[0112] S3: Based on the aforementioned response evolution characteristics, establish a constraint model for the optical response evolution of the quartz tube surface, analyze the stability, continuity, and repeatability of the optical response change process in each detection area, and generate the response reliability distribution of the detection area.

[0113] Furthermore, the response evolution characteristics of each detection area are obtained, and the response change amplitude, change direction, change rate and continuous change trend in the response evolution characteristics are correlated to establish the evolution relationship of the optical response of the detection area with the optical excitation.

[0114] Based on the spatial adjacency relationship between different detection areas on the surface of the quartz tube, spatial correlation constraints between detection areas are constructed, and the spatial correlation constraints are fused with the evolutionary relationship to generate an optical response evolution constraint model for the surface of the quartz tube.

[0115] Furthermore, the optical response evolution constraint values ​​of each detection area are obtained, and the stable characteristics of response change, the continuous characteristics of spatial evolution, and the response repetition characteristics in multiple scanning processes of the detection area under continuous optical excitation are extracted respectively.

[0116] The optical response stability index of the detection area is calculated based on the response change stability characteristics, the optical response continuity index of the detection area is calculated based on the response change relationship between continuous optical excitation states, and the optical response repeatability index of the detection area is calculated based on the response consistency of the same detection area in different scanning cycles.

[0117] The optical response stability index, optical response continuity index, optical response repeatability index, and optical response evolution constraint value are fused to generate the response confidence level of the detection area.

[0118]

[0119] in, To assess the confidence level of the response in region i, For the optical response evolution constraint value, This is an indicator of optical response stability. As an indicator of optical response continuity, As an indicator of optical response repeatability, , , , For weighting;

[0120] Spatial mapping is performed based on the response confidence of each detection area to generate the response confidence distribution on the quartz tube surface. Areas whose response confidence meets the preset confidence conditions are identified as valid response areas, while areas whose response confidence is lower than the preset confidence conditions are identified as abnormal response areas.

[0121] It should be noted that in this embodiment, the response evolution characteristics of each detection region are obtained, and the response evolution characteristics include the response change amplitude, response change direction, response change rate, and continuous change trend.

[0122] For the same detection area, the response change information formed under different optical excitation states is correlated to analyze the response evolution law of the detection area during the optical excitation change process.

[0123] For example, in a cracked region on the surface of a quartz tube, the crack causes a change in the continuity of the material's surface structure, resulting in a continuous variation in its reflection, scattering, and transmission responses as the optical excitation conditions change. However, anomalous responses generated by surface reflection typically only occur under specific detection conditions and cannot form a stable response evolution pattern. Therefore, based on the correlation between the response variation amplitude, direction, and rate, an evolutionary relationship between the optical response of the detection region and optical excitation is established.

[0124] Furthermore, spatial correlation constraints are established based on the spatial adjacency relationship between the detection areas on the surface of the quartz tube.

[0125] Specifically, based on the spatial location of the detection area on the curved surface of the quartz tube, the correlation between adjacent detection areas is determined, and the consistency of the response evolution characteristics between adjacent areas is analyzed.

[0126] When adjacent detection areas have similar response change trends, the area is considered to have strong spatial continuity; when the response change of a certain detection area is significantly abnormal, and there is no corresponding change trend in the surrounding area, the area is considered to be affected by local optical interference.

[0127] Furthermore, the spatial correlation constraints are integrated with the optical response evolution relationship of the detection area to obtain the optical response evolution constraint value of the detection area. The optical response evolution constraint value is used to characterize whether the response change of the detection area simultaneously satisfies: the continuous evolution law in the process of optical excitation change; the spatial distribution law of the quartz tube surface; and the response change characteristics of the defect area.

[0128] To comprehensively evaluate the impact of different factors on the reliability of the response in the detection area, fusion weights were set according to the importance of each factor in the quartz tube surface defect detection process.

[0129] Since the optical response evolution constraint value directly reflects whether the detection area conforms to the defect-induced response change law, it is given a high weight. Since the response stability feature and spatial evolution continuity feature respectively reflect response reliability and defect spatial continuity, and have a strong auxiliary role in judging real defects, they are given the same weight. Since the repeated scan response is used to further reduce the influence of accidental optical interference, it is given a low weight as an auxiliary verification factor. In this embodiment, the weight of the optical response evolution constraint value is set to 0.35; the weight of the response change stability feature is 0.25; the weight of the spatial evolution continuity feature is 0.25; and the weight of the response repetition feature is 0.15.

[0130] Furthermore, based on the calculated response confidence level, spatial mapping is performed on each detection area on the surface of the quartz tube.

[0131] Specifically, the response confidence value corresponding to each detection area is mapped to its position on the surface of the quartz tube, forming a response confidence distribution on the quartz tube surface. When a certain area has: a stable optical response change pattern; a spatial evolution relationship consistent with neighboring areas; and response characteristics that recur during multiple scans, then that area obtains a high response confidence and is determined as an effective response area.

[0132] When a region exhibits an abnormal response under a single optical excitation condition and lacks a continuous variation pattern or spatial correlation, its response reliability is reduced, and it is identified as an abnormal response region.

[0133] S4: Reconstruct the true defect response field of the quartz tube surface based on the response confidence distribution, and use the optical response evolution law of the defect region in the true defect response field to calculate the continuity of the defect boundary, the consistency of the response inside the defect, and the response gradient of the defect edge to obtain the true defect characteristics of the quartz tube surface.

[0134] Furthermore, the response confidence and optical response data of each detection area on the surface of the quartz tube are obtained, and the spatial location, optical response intensity and response confidence of the detection area are spatially mapped to generate the initial response distribution on the surface of the quartz tube.

[0135] Based on the aforementioned response confidence level, the optical response data in the initial response distribution are subjected to confidence-weighted processing to obtain a reliable optical response distribution.

[0136] Based on the surface spatial continuity of the quartz tube, the defect response continuity of the reliable optical response distribution is reconstructed, and the response gradient changes, spatial extension directions and response connectivity between adjacent detection areas are analyzed to generate a defect response spatial correlation structure.

[0137] The defect response spatial correlation structure is fused with the reliable optical response distribution to reconstruct the true defect response field on the surface of the quartz tube, and the defect region range is determined based on the true defect response field.

[0138] Furthermore, a continuous response region satisfying the defect response conditions in the real defect response field is obtained, and the boundary of the defect region is determined based on the spatial connection relationship between the detection points within the continuous response region.

[0139] Based on the response change relationship between adjacent detection points on the defect region boundary, calculate the defect boundary continuity index:

[0140]

[0141] in, This is an indicator of the continuity of the defect boundary, where N is the number of defect boundary detection points. This represents the change in defect response between adjacent boundary detection points;

[0142] The defect boundary continuity index is used to determine whether the defect boundary has continuous expansion characteristics.

[0143] Based on the defect response values ​​and optical response evolution characteristics at each detection location within the defect region, the internal response consistency index of the defect is calculated:

[0144]

[0145] in, This serves as an indicator of the consistency of internal responses to defects. This represents the response value within the defect area. The degree of response fluctuation within the defect area. ε is the average response value within the defect area, and ε is the correction parameter.

[0146] Calculate the defect edge response gradient based on the degree of response change at the defect edge location:

[0147] in, The gradient of the defect edge response. This represents the response value within the defect area. This represents the response value of the area adjacent to the defective region. The spatial distance from the defect edge;

[0148] By integrating the defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution law of the defect region, the true defect characteristics of the quartz tube surface are generated.

[0149] It should be noted that in this embodiment, the response reliability of each detection area on the surface of the quartz tube and the corresponding multi-time-series optical response data are obtained. The multi-time-series optical response data includes reflected light response, scattered light response, and transmitted light response acquired under different optical excitation conditions.

[0150] Based on the location markers of the detection area, the spatial location of the detection area, the optical response intensity, and the response confidence are correlated and mapped to form the initial response distribution on the surface of the quartz tube.

[0151] Because quartz tubes are transparent, highly reflective, and have a curved surface, some changes in the optical response during actual testing are caused by surface reflection, changes in the incident angle, and optical path disturbances, rather than by actual defects. Therefore, this embodiment uses the response confidence level generated in the previous step to constrain the initial response distribution.

[0152] For detection regions with stable response evolution patterns, spatial correlations, and consistent repeated scans, their response contribution is increased; for abnormal regions lacking continuous variation patterns, their response contribution is reduced, thereby obtaining a reliable optical response distribution.

[0153] Furthermore, based on the surface spatial continuity of the quartz tube, the defect response continuity of the reliable optical response distribution is reconstructed.

[0154] Specifically, based on the positional relationship between adjacent detection areas, the changes in response gradient, spatial extension direction, and response connectivity between detection areas are analyzed.

[0155] Therefore, detection areas with continuous response variations are correlated to form a spatial correlation structure for defect responses. This spatial correlation structure is then fused with a reliable optical response distribution to obtain the true defect response field on the quartz tube surface, and the defect region extent is determined based on the continuous response regions.

[0156] Furthermore, feature extraction is performed on the defect regions in the real defect response field.

[0157] First, the defect region boundary is determined based on the continuous region in the real defect response field that satisfies the defect response conditions. For the determined defect boundary, this embodiment does not directly determine the defect based on the edge position, but analyzes the response change relationship between adjacent detection points on the defect boundary to evaluate whether the defect boundary has the continuity of the real structure.

[0158] Specifically, the boundary variation pattern is described using a defect boundary continuity index. The change in response at adjacent boundary detection points within the defect boundary continuity index reflects the degree of change in optical response as the defect expands along the spatial direction.

[0159] When the defect is a real crack or structural damage, the response changes at adjacent boundary positions usually have a gradual characteristic because the defect region has a continuous spatial structure, so this index can maintain a high level.

[0160] For anomalous responses caused by localized reflections or random noise, their boundaries typically exhibit abrupt changes, leading to a decrease in the continuity index. The above method allows the continuity index of the defect boundary to distinguish between genuine defect responses and anomalous optical responses from a spatial structure perspective.

[0161] Furthermore, the consistency of response within the defective region is analyzed.

[0162] Since real defect regions usually have the same or similar material state changes, different locations inside the defect will exhibit a certain degree of consistent response pattern under various optical excitation conditions.

[0163] Therefore, this embodiment uses the internal response consistency index of the defect to describe the stability of the response distribution within the defect region. Specifically, the average response within the defect region is used to characterize the overall response level, while the response fluctuation reflects the dispersion of the response within the region.

[0164] When the internal structure of the defect area is relatively consistent, the difference in response changes at different locations is small, and the consistency index is improved; when the response changes within the area are irregular, it indicates that the area may be affected by factors such as contamination or local reflection, and the consistency index is reduced.

[0165] Furthermore, response gradient analysis is performed on the edge location of the defect area.

[0166] This embodiment utilizes the defect edge response gradient to describe the intensity of optical response changes between the defect region and the normal region. Since real defects alter the light propagation path and scattering characteristics of the quartz material surface, there is usually a significant response difference between the defect region and the surrounding normal region. The defect edge response gradient is obtained by analyzing the relationship between the response inside the defect and the response of the adjacent external region as a function of spatial distance.

[0167] Regions with larger gradients indicate significant structural differences between defective and normal regions, while regions with smaller gradients indicate slower response changes, which are considered normal optical variations. Therefore, this indicator further determines the location of defect edges and improves the accuracy of defect region localization.

[0168] Furthermore, the continuity of the defect boundary, the consistency of the internal response of the defect, the response gradient of the defect edge, and the evolution law of the optical response corresponding to the defect region are integrated to generate the real defect features of the quartz tube surface.

[0169] By jointly representing the above four types of features, the obtained true defect features simultaneously include: spatial structure information; internal state information of the region; edge change information; and optical evolution information.

[0170] Compared with existing methods that judge defects based on grayscale or texture changes in a single optical image, this embodiment avoids directly taking accidental optical anomalies as defect results. Instead, it achieves reliable extraction of real defects on the surface of quartz tubes through defect response field reconstruction and joint evaluation of multi-dimensional physical response features.

[0171] S5: Establish a quartz tube surface defect evolution model based on the actual defect characteristics, jointly determine the defect type, defect size, defect depth and defect expansion trend, and output the quartz tube surface defect detection results.

[0172] Furthermore, the defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution characteristics are obtained from the real defect features to construct a defect state feature set.

[0173] Based on the defect state feature set, a defect spatial state description relationship is established, and the spatial distribution of the defect region, the change of response intensity, and the evolution process of optical response are integrated to form a defect evolution state model.

[0174] Based on the defect evolution state model, the spatial scale parameters, response change parameters, and boundary expansion parameters of the defect region are extracted. The comprehensive defect evaluation index is calculated. Based on the comprehensive defect evaluation index and the defect evolution state model, the defect type, defect size, defect depth, and defect expansion trend of the quartz tube surface are jointly determined, and the defect detection results of the quartz tube surface are output.

[0175] It should be noted that, firstly, the defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution characteristics are obtained from the real defect features.

[0176] By associating defect boundary continuity indicators, defect internal response consistency indicators, defect edge response gradients, and optical response evolution characteristics, a set of defect state characteristics is formed.

[0177] For example, for a given detection area: if the defect boundary has high continuity, it indicates that the abnormal area has a continuous structure; if the internal response of the defect is highly consistent, it indicates that the material state changes within the area are relatively uniform; if the response gradient at the defect edge is significant, it indicates that there is a significant difference in optical response between this area and the surrounding normal quartz material; if the response changes remain stable under different optical excitation conditions, it indicates that the abnormal area has the response characteristics of a true defect. Through the combination of these multi-dimensional features, a comprehensive feature set describing the defect state is formed.

[0178] Furthermore, based on the comprehensive defect evaluation index and the defect evolution state model, the defect type, defect size, defect depth, and defect expansion trend on the quartz tube surface are jointly determined.

[0179] The specific determination process is as follows:

[0180] 1. Defect type determination: The determination is based on the spatial morphology of the defect and the evolution of its optical response. For example: if the defect region is distributed in a continuous linear pattern with high boundary continuity and obvious edge response gradient, it is determined to be a crack-type defect; if the defect region expands in a plate-like pattern with high internal response consistency, it is determined to be a crystallization-type defect; if the response change in the defect region is locally obvious but the spatial continuity is weak, it is determined to be a contamination-adhesion-type defect.

[0181] 2. Defect size determination: Based on the range of continuous defect areas in the actual defect response field, combined with spatial scale parameters, determine the defect length, width, and coverage area.

[0182] 3. Defect Depth Determination: Based on the degree of response change in the defect area under different optical excitation conditions, and combined with the response change parameters, the degree of influence of the defect on the light propagation process is determined. When the degree of response change is large, it indicates that the defect has a strong influence on light propagation, and the defect depth is determined to be large; when the degree of response change is weak, the defect is determined to be mainly located in the shallow area.

[0183] 4. Determination of Defect Expansion Trend: Based on the continuity of the defect boundary, the direction of spatial extension, and the boundary expansion parameters, it is determined whether the defect has a continuing expansion trend. If the defect boundary is continuous, the surrounding area shows continuous response changes, and the optical response evolution trend continues to increase, then the defect is determined to have an expansion trend. If the defect boundary is stable and the surrounding area shows no continuous response changes, then the defect is determined to be in a stable state. Finally, the surface defect detection results of the quartz tube are output.

[0184] This embodiment also provides a computer device applicable to a method for detecting surface defects in quartz tubes used in semiconductor oxidation furnaces, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for detecting surface defects in quartz tubes used in semiconductor oxidation furnaces as proposed in the above embodiment.

[0185] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0186] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in the above embodiments.

[0187] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace, characterized in that, include: The quartz tube under test is continuously rotated around the axis, and the incident direction, polarization state and wavelength parameters of the irradiated light are changed to continuously scan the surface of the quartz tube and obtain multi-time-series optical response data of the same detection area on the surface of the quartz tube under different detection conditions. Based on the multi-time series optical response data, an optical response evolution sequence of the detection area on the surface of the quartz tube is established, and the response change relationship between adjacent detection conditions in the optical response evolution sequence is extracted to generate response evolution features. Based on the aforementioned response evolution characteristics, an optical response evolution constraint model for the quartz tube surface is established. The stability, continuity, and repeatability of the optical response change process in each detection area are analyzed, and the response reliability distribution of the detection area is generated. The true defect response field of the quartz tube surface is reconstructed based on the response confidence distribution. The continuity of the defect boundary, the consistency of the response inside the defect and the response gradient of the defect edge are calculated using the optical response evolution law of the defect region in the true defect response field, so as to obtain the true defect characteristics of the quartz tube surface. Based on the actual defect characteristics, a defect evolution model for the quartz tube surface is established. The defect type, defect size, defect depth, and defect expansion trend are jointly determined, and the detection results of the quartz tube surface defects are output.

2. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The method involves changing the incident direction, polarization state, and wavelength parameters of the irradiated light to continuously scan the surface of the quartz tube, acquiring multi-temporal optical response data of the same detection area on the quartz tube surface under different detection conditions, including: Establish a synchronous control relationship between the rotational motion of the quartz tube and optical scanning, divide the outer surface of the quartz tube into multiple continuous detection areas, and set a unique position identifier for each detection area; According to a preset optical excitation sequence, the incident direction, polarization state, and wavelength parameters of the irradiation light are combined and switched so that each detection area receives multiple sets of different optical excitations in sequence; after each set of optical excitations is completed, the reflected light response, scattered light response, and transmitted light response of the detection area are obtained. The reflected light response, scattered light response and transmitted light response obtained under each set of optical excitations are correlated and integrated according to the position identifier, and the optical response sequence of the detection area is constructed according to the execution order of the optical excitation sequence. Each set of optical excitations and a set of optical responses together form an optical response state. The continuity and integrity between adjacent optical response states in the optical response sequence are detected, and the optical response states that meet the requirements for continuous acquisition are stored sequentially to form multi-temporal optical response data of the detection area.

3. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The step of establishing an optical response evolution sequence for the detection area on the surface of a quartz tube based on multi-time-series optical response data, and extracting the response change relationship between adjacent detection conditions in the optical response evolution sequence to generate response evolution features includes: Read the multi-temporal optical response data of each detection area, arrange the optical response states according to the execution order of the optical excitation sequence, and establish the optical response evolution sequence of the detection area; Traverse adjacent optical response states in the optical response evolution sequence, calculate the magnitude, direction, and rate of change of reflected light response, scattered light response, and transmitted light response between adjacent optical response states, and form optical response change information; Based on the correlation between changes in multiple consecutive optical response states, continuous evolution analysis is performed on the optical response change information to obtain the stable trend, fluctuation trend and turning trend of optical response change. By integrating the optical response change information, as well as the stable trend, fluctuation trend, and turning trend of the optical response change, the response evolution characteristics of the detection area are generated.

4. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The establishment of a constraint model for the optical response evolution of the quartz tube surface based on response evolution characteristics includes: The response evolution characteristics of each detection area are obtained, and the response change amplitude, change direction, change rate and continuous change trend in the response evolution characteristics are correlated to establish the evolution relationship of the optical response of the detection area with the optical excitation. Based on the spatial adjacency relationship between different detection areas on the surface of the quartz tube, spatial correlation constraints between detection areas are constructed, and the spatial correlation constraints are fused with the evolutionary relationship to generate an optical response evolution constraint model for the surface of the quartz tube.

5. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The analysis of the stability, continuity, and repeatability of the optical response change process in each detection area, and the generation of the response reliability distribution of the detection area, includes: Obtain the optical response evolution constraint values ​​of each detection area, and extract the stable characteristics of response change, the continuous characteristics of spatial evolution, and the repetitive characteristics of response during multiple scans of the detection area under continuous optical excitation. The optical response stability index of the detection area is calculated based on the response change stability characteristics, the optical response continuity index of the detection area is calculated based on the response change relationship between continuous optical excitation states, and the optical response repeatability index of the detection area is calculated based on the response consistency of the same detection area in different scanning cycles. The optical response stability index, optical response continuity index, optical response repeatability index, and optical response evolution constraint value are fused to generate the response confidence level of the detection area. in, To assess the reliability of the response in region i, For the optical response evolution constraint value, This is an indicator of optical response stability. As an indicator of optical response continuity, As an indicator of optical response repeatability, , , , For weighting; Spatial mapping is performed based on the response confidence of each detection area to generate the response confidence distribution on the quartz tube surface. Areas whose response confidence meets the preset confidence conditions are identified as valid response areas, while areas whose response confidence is lower than the preset confidence conditions are identified as abnormal response areas.

6. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The process of reconstructing the true defect response field of the quartz tube surface based on the response confidence distribution includes: acquiring the response confidence and optical response data of each detection area on the quartz tube surface, spatially mapping the spatial location, optical response intensity, and response confidence of the detection area, and generating the initial response distribution of the quartz tube surface. Based on the aforementioned response confidence level, the optical response data in the initial response distribution are subjected to confidence-weighted processing to obtain a reliable optical response distribution. Based on the surface spatial continuity of the quartz tube, the defect response continuity of the reliable optical response distribution is reconstructed, and the response gradient changes, spatial extension directions and response connectivity between adjacent detection areas are analyzed to generate a defect response spatial correlation structure. The defect response spatial correlation structure is fused with the reliable optical response distribution to reconstruct the true defect response field on the surface of the quartz tube, and the defect region range is determined based on the true defect response field.

7. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The method of calculating the continuity of the defect boundary, the consistency of the internal response of the defect and the response gradient of the defect edge by utilizing the optical response evolution law of the defect region in the real defect response field to obtain the real defect characteristics of the quartz tube surface includes: obtaining a continuous response region in the real defect response field that meets the defect response conditions, and determining the defect region boundary according to the spatial connection relationship between the detection points in the continuous response region. Based on the response change relationship between adjacent detection points on the defect region boundary, calculate the defect boundary continuity index: in, This is an index for the continuity of the defect boundary, where N is the number of defect boundary detection points. This represents the change in defect response between adjacent boundary detection points; The defect boundary continuity index is used to determine whether the defect boundary has continuous expansion characteristics. Based on the defect response values ​​and optical response evolution characteristics at each detection location within the defect region, the internal response consistency index of the defect is calculated: in, This serves as an indicator of the consistency of internal response to defects. This represents the response value within the defect area. The degree of response fluctuation within the defect area. ε is the average response value within the defect region, and ε is the correction parameter. Calculate the defect edge response gradient based on the degree of response change at the defect edge location: in, The gradient of the defect edge response. This represents the response value within the defect area. This represents the response value of the area adjacent to the defective region. The spatial distance from the defect edge; By integrating the defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution law of the defect region, the true defect characteristics of the quartz tube surface are generated.

8. The method for detecting surface defects in a quartz tube used in a semiconductor oxidation furnace as described in claim 1, characterized in that, The process of establishing a surface defect evolution model for quartz tubes based on actual defect characteristics, and jointly determining defect type, defect size, defect depth, and defect propagation trend, includes: Obtain defect boundary continuity index, defect internal response consistency index, defect edge response gradient, and optical response evolution characteristics from real defect features, and construct a defect state feature set. Based on the defect state feature set, a defect spatial state description relationship is established, and the spatial distribution of the defect region, the change of response intensity, and the evolution process of optical response are integrated to form a defect evolution state model. Based on the defect evolution state model, the spatial scale parameters, response change parameters, and boundary expansion parameters of the defect region are extracted. The comprehensive defect evaluation index is calculated. Based on the comprehensive defect evaluation index and the defect evolution state model, the defect type, defect size, defect depth, and defect expansion trend of the quartz tube surface are jointly determined, and the defect detection results of the quartz tube surface are output.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for detecting surface defects of a quartz tube for a semiconductor oxidation furnace as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for detecting surface defects of a quartz tube for a semiconductor oxidation furnace as described in any one of claims 1 to 8.

Images