CN116978834B - Intelligent monitoring and early warning system for wafer production - Google Patents
Intelligent monitoring and early warning system for wafer production Download PDFInfo
- Publication number
- CN116978834B CN116978834B CN202311213205.3A CN202311213205A CN116978834B CN 116978834 B CN116978834 B CN 116978834B CN 202311213205 A CN202311213205 A CN 202311213205A CN 116978834 B CN116978834 B CN 116978834B
- Authority
- CN
- China
- Prior art keywords
- wafer
- dust
- early warning
- production
- equipment
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 146
- 238000012544 monitoring process Methods 0.000 title claims abstract description 46
- 239000000428 dust Substances 0.000 claims abstract description 187
- 230000007547 defect Effects 0.000 claims abstract description 170
- 238000001514 detection method Methods 0.000 claims abstract description 94
- 230000005856 abnormality Effects 0.000 claims abstract description 48
- 238000000034 method Methods 0.000 claims abstract description 39
- 230000002159 abnormal effect Effects 0.000 claims abstract description 27
- 230000003287 optical effect Effects 0.000 claims abstract description 27
- 238000004891 communication Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 25
- 238000010586 diagram Methods 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 19
- 230000002950 deficient Effects 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 11
- 239000013598 vector Substances 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000003860 storage Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 8
- 238000005286 illumination Methods 0.000 claims description 8
- 238000002360 preparation method Methods 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 238000003708 edge detection Methods 0.000 claims description 3
- 230000007257 malfunction Effects 0.000 claims description 3
- 230000001629 suppression Effects 0.000 claims description 3
- 235000012431 wafers Nutrition 0.000 abstract description 223
- 230000005540 biological transmission Effects 0.000 description 4
- 238000012546 transfer Methods 0.000 description 4
- 239000004065 semiconductor Substances 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000005764 inhibitory process Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011109 contamination Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01L—SEMICONDUCTOR DEVICES NOT COVERED BY CLASS H10
- H01L21/00—Processes or apparatus adapted for the manufacture or treatment of semiconductor or solid state devices or of parts thereof
- H01L21/67—Apparatus specially adapted for handling semiconductor or electric solid state devices during manufacture or treatment thereof; Apparatus specially adapted for handling wafers during manufacture or treatment of semiconductor or electric solid state devices or components ; Apparatus not specifically provided for elsewhere
- H01L21/67005—Apparatus not specifically provided for elsewhere
- H01L21/67242—Apparatus for monitoring, sorting or marking
- H01L21/67288—Monitoring of warpage, curvature, damage, defects or the like
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Condensed Matter Physics & Semiconductors (AREA)
- General Physics & Mathematics (AREA)
- Manufacturing & Machinery (AREA)
- Computer Hardware Design (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Power Engineering (AREA)
- Testing Or Measuring Of Semiconductors Or The Like (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses an intelligent monitoring and early warning system for wafer production, which comprises a frame, a database layer and a control layer, wherein the frame comprises an equipment layer, the equipment layer comprises input equipment and output equipment, the input equipment comprises a scanning electron microscope, an optical irradiation device and a camera, the output equipment comprises a display device, a dust abnormality early warning device and a fault abnormality early warning device, the database layer is used for storing data acquired in the working process of the system, the control layer comprises a CPU, a dust detection module, a defect detection and fault equipment prediction module, a communication module, a dust abnormality early warning module and a fault abnormality early warning module, and the dust detection module is used for monitoring whether dust abnormality increases in different production stages; the defect detection and fault equipment prediction module is used for detecting whether different equipment fails to cause wafer defects. The invention can quickly early warn abnormal production stages and equipment, and improves the quality and efficiency of producing wafers.
Description
Technical Field
The invention relates to the technical field of wafer production data processing, in particular to an intelligent monitoring and early warning system for wafer production.
Background
In semiconductor manufacturing processes, wafers are subjected to thousands of processing steps and many different types of manufacturing equipment are involved. There is an increasing need to transfer wafers between many different process modules or clusters of process modules and sometimes for production process reasons, between modules that are far apart, and in the process, the wafers are easily contaminated with dust for various reasons, thereby affecting the quality of the product, and monitoring the stage of production that leads to contamination of the wafers with dust is of great importance for improving the yield and quality of the wafers. In addition, the complex production stages and production equipment mean that finding a faulty production equipment is a challenging problem, and as the complexity of semiconductor manufacturing continues to increase, how to accurately find equipment that causes wafer defects is an important element of the wafer production process.
For example, chinese patent application publication No. CN105742216a discloses a wafer transfer system and a method for detecting the wafer transfer system, which aims to solve the problem of detecting whether the wafer contactor needs to be replaced or not during the wafer transfer process. The method comprises the steps of obtaining a detection amplitude of a wafer in a process of separating the wafer from a wafer contactor of a conveying component; acquiring the calculated amplitude of the detected amplitude after data processing, wherein the data processing is differential processing of a plurality of times or the combination of the differential processing of a plurality of times and weighting processing; and if the maximum value in the calculated amplitude is greater than or equal to a preset threshold value, sending prompt information to confirm the time for replacing the wafer contactor according to the prompt information. The wafer conveying system and the wafer conveying system provided by the invention are used for solving the defect that the wafer contactor and the wafer are easy to be adhered in the related technology, and the processing process and the precision of the wafer are affected, and can determine whether the wafer contactor needs to be replaced or not through the quantized amplitude. However, the invention does not pay attention to detecting whether abnormal dust is increased on the wafer in the wafer conveying process and carrying out early warning.
For example, chinese patent publication No. CN115193750B discloses a semiconductor wafer defect inspection system, which aims to solve the problem of distinguishing between wafers after defect inspection and non-inspected wafers. The invention comprises a defect detection table, a transmission module, a storage bin module, a positioning module, a visual detection module, a size detection module, a marking module and a defect analysis module, wherein the defect detection table is used for judging whether the wafer can be used or not by detecting particle data on the surface of the wafer, thickness data of the wafer and height data on the surface of the wafer, and when the wafer is judged to be inconsistent with a use standard, the marking module is controlled to carry out defect marking on the wafer and the transmission module is controlled to transmit the marked wafer to the corresponding storage bin so as to finish defect detection and sorting of the wafer, so that the wafer after defect detection can be effectively distinguished from the wafer which is not detected. However, the invention does not detect the equipment causing the wafer to generate defects and performs early warning.
The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person of ordinary skill in the art.
Disclosure of Invention
Aiming at the defects of the prior art, the invention mainly aims to provide an intelligent monitoring and early warning system for wafer production, which can effectively solve the problems in the background art: in the existing wafer conveying and detecting system, whether dust is abnormally increased or not in different production stages cannot be detected and early warning is carried out, and failure equipment causing wafer defects cannot be detected and early warning is carried out, so that the problem of low quality and efficiency of wafer production is caused. The specific technical scheme of the invention is as follows:
an intelligent monitoring and early warning system for wafer production comprises an equipment layer, a database layer and a control layer; the equipment layer comprises input equipment and output equipment, wherein the input equipment comprises a scanning electron microscope, an optical irradiation device and a camera, the scanning electron microscope is used for scanning the surface of a wafer and detecting defects to generate a wafer image, the optical irradiation device is used for displaying dust on the wafer, the camera is used for shooting the wafer image irradiated by the optical irradiation device, the output equipment comprises a display device, a dust abnormality early warning device and a fault abnormality early warning device, the display device is used for displaying the production stage of the wafer, the equipment for producing the wafer is abnormal, the dust abnormality early warning device is used for early warning the production stage with abnormal increase of dust, and the fault abnormality early warning device is used for early warning fault equipment causing the wafer defects;
The database layer is used for storing data acquired in the working process of the system;
the control layer comprises a CPU, a dust detection module, a defect detection and fault equipment prediction module, a communication module, a dust abnormality early warning module and a fault abnormality early warning module.
The invention is further improved in that the CPU is used for managing and controlling the operation of the system; the dust detection module is used for monitoring whether abnormal dust increase occurs in different production stages; the defect detection and fault equipment prediction module is used for detecting defects on the wafer map, predicting equipment causing defects according to the types and positions of the detected defects, and monitoring whether different equipment fails to cause the wafer defects; the communication module is used for constructing a communication network in the system to realize the mutual transmission of data in the system; the dust abnormality early warning module is used for carrying out early warning through a dust abnormality early warning device in the output equipment aiming at a certain production stage when detecting that dust abnormality increases in the production stage; and the fault abnormality early warning module is used for predicting production equipment which causes wafer defects to be generated when the wafer defects are detected, and carrying out early warning on the production equipment with faults through the fault abnormality early warning device in the output equipment.
A further improvement of the invention is that the input device comprises a scanning electron microscope for scanning the surface of the wafer, generating the wafer map, the resolution of the lens system of the scanning electron microscope being between 0.1 nm and 1 nm.
The invention is further improved in that the input device comprises an optical irradiation device, wherein the optical irradiation device is used for displaying dust on a wafer, and the optical irradiation device uses a bright field illumination system to uniformly irradiate a high-brightness light source onto the wafer to be detected after focusing through a lens.
The invention further improves that the input device comprises a camera, wherein the camera is used for shooting the front image of the wafer irradiated by the optical irradiation device, and the magnification of a lens system of the camera is 100-200 times.
The invention further improves that the dust detection module divides the wafer production process into different stages, the joint point of each stage is set as a wafer dust detection point, the wafer dust detection point irradiates the wafer by using the optical irradiation device, the front image of the wafer is shot by using the camera, the front image of the wafer shot by the camera is converted into a gray level image, and the gray level image is arranged to pass through the first stage The wafers processed in the production stage are processed by the wafer dust detection pointThe gray level is +.>Wherein->，/>Representing the total number of production stages.
The invention further improves that the dust detection module comprises a dust shape matching unit, the dust shape matching unit comprises a contour extraction subunit and a contour matching subunit, the contour extraction subunit uses a Canny edge detection algorithm to extract the contour of dust in the gray level map through noise removal, calculation of pixel gradient value and direction, non-maximum suppression, double-threshold detection and edge connection, the contour information of the dust is represented by coordinates of contour pixel points, the coordinates of the contour pixel points are stored in a clockwise sequence, and the gray level map is setThe%>The dust content is->Then->The profile information of the individual dust is +.>Wherein->，/>，/>Representation->Total number of dust in->Representing the total number of production stages.
A further improvement of the present invention is that the contour matching subunit matches the gray scale mapIs +.>The similarity of each dust in the dust-to-dust ratio is one-to-one, wherein +.>To pass->Gray scale obtained by processing the wafer processed in the production stage through the wafer dust detection point, wherein +. >，/>Representing the total number of production stages, let->Middle->The dust content is->，/>Middle->The dust content is->Wherein->，，/>，/>Representation->Total number of dust in->Representation->Total number of dust in->Representing the total number of production stages, the similarity comparison method adopts a shape context matching algorithm to set a discrimination threshold for the contour matching subunit ++>Is provided with->Is->And->If the similarity of (1) is compared with ∈>There is->Satisfy->Record->Is old dust, if->Does not have->Satisfy->Record->For new dust, the gray-scale map is traversed>After every dust in (3), counting the number of newly appeared dust +.>If->Wherein->For the first threshold value, the first ∈is displayed by the display device>And the abnormal dust increase phenomenon occurs in each production stage, and the early warning is carried out through the abnormal dust early warning module.
The invention is further improved in that the dust detection module further comprises a dust area matching unit, the dust area matching unit being at the first stage according to the dust shape matching unitThe +.sup.th extracted from the production stage>Dust aloneProfile information of->Calculation of +.>Dust->Area of (1)/(5)>Totally contain->Coordinates of pixel points of the outline are set with the +. >The coordinates of each outline pixel point are +.>First->The coordinates of each outline pixel point are；
Wherein whenWhen make->The coordinates of each outline pixel point are +.>Then->Dust->Is +.>Wherein->The first passGray level map +.>The total area of all dust in (a) is +.>Comparing said grey-scale patterns +.>Total area of all dust->And gray level map->Total area of all dust->If->Wherein->For the second threshold value, the first is displayed by the display device>And the abnormal dust increase phenomenon occurs in each production stage, and the early warning is carried out through the abnormal dust early warning module.
The invention further improves that the defect detection and fault equipment prediction module comprises a training data preparation unit, a defect detection unit and a fault equipment prediction unit, wherein the training data preparation unit uses a scanning electron microscope to detect defects of each chip on a wafer, counts the number of the defects of each chip, constructs a wafer map according to the counting result, wherein the value of each pixel point on the wafer map is the number of the defects of the chip at the corresponding position of the pixel point, and prepares the wafer map of the wafer after being processed by production equipment, and marks the type, the position and equipment for generating the defects on the wafer map according to the wafer map.
The invention further improves that the defect detection unit comprises two branches for detecting the type of defect and the position of defect respectively, each branch adopts a residual network with fifty stacked layers as a detection model, the residual network is finally connected with a full connection layer for classification, the full connection layer is activated by using a softmax function, and the detection model of the first branch for detecting the type of defect is recorded asIs provided with->The input data of (a) is the wafer map, and the wafer map is set as，/>Is vector->Wherein->，/>Is>Individual element->To predict the wafer map +.>Is category->Probability of>The representation dimension is +.>Of (2) wherein>For the number of defect types, the defect detection unit uses the wafer map +.>And corresponding actual wafer type code +.>Training->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>For the number of defect types ∈ ->The loss function of (2) is:
；
recording the detection model of the second branch for detecting the defect type as，/>Input data of (a) is the wafer map +.>，/>Is vector->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>Is>Individual element- >For predicting the wafer map +.>Coordinate information of pixel point of defect position, < >>Totally contain->A pixel dot, wherein->The coordinates of the individual pixels are denoted +.>ThenThe defect detecting unit uses the wafer map +.>And corresponding actual wafer position information +.>Training->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>The total number of coordinate values of pixel points actually used for representing the wafer position information is +.>The loss function of (2) is:
。
the invention is further improved in that the defect detection unit is trainedAnd->Then, when detecting the defect type and defect position of the wafer, the wafer map of the wafer to be detected is input +.>According toOutput->Judging the defect type with the largest probability of the wafer, if the defect type is None, namely the wafer is not defective, starting a fault abnormality early warning module to early warn, if the defect type is not None, namely the wafer is defective, starting the fault abnormality early warning module to early warn, and inputting a wafer map of the wafer to be detected>ObtainingOutput result of +.>Finally, will->And->The faulty device prediction unit is input together.
The invention further improves that the fault equipment prediction model of the fault equipment prediction unit The method comprises six linear layers, wherein the first five linear layers are all activated by RELU functions, the last linear layer outputs a prediction result through softmax function processing, and the fault equipment prediction model->Is the two branch models of the defect detection unit +.>And->Output result of +.>And->The splice result is +.>I.e. +.>WhereinThe representation dimension is +.>Of (2) wherein>For the number of defect types +.>For the total number of coordinate values of pixel points for representing wafer position information, the malfunction apparatus prediction model +.>The output result of (a) is a vectorWherein->The representation dimension is +.>Real space of>Wherein->Represents the total number of all devices in the wafer production process, wherein +.>Is>Individual element->To predict the wafer map is caused +.>The defective device in (1) is->Probability of individual devices, said faulty device prediction unit using +.>And generating the wafer map +.>Actual device coding of defects +.>Training->Wherein->The representation dimension is +.>Is used for the real space of (a),wherein->Indicating the total number of all devices during wafer production, +.>Representation->Is>Element(s)>Is>Individual element->A value of 1 indicates that the wafer map is caused +. >The defective device in (1) is->Personal device, otherwise->The value is 0->Said faulty device prediction model->The loss function of (2) is:
。
the invention is further improved in that the fault equipment prediction unit is trainedThen, when detecting the equipment causing the wafer to generate the defect, the output result of the defect detection unit is first added>And->Input->，/>Output prediction result +.>And listing fault equipment causing wafer defects according to the probability order, displaying the fault equipment causing the wafer defects through the display equipment, displaying the defect types and the positions of the defects on the wafer, and carrying out early warning through the fault abnormality early warning module.
An intelligent monitoring and early warning method for wafer production comprises the following specific steps:
a1: collecting a front image of the wafer transmitted to the wafer dust detection point, and converting the front image into a gray level image;
a2: extracting outline information of dust in the gray level diagram according to the gray level diagram in the A1;
a3: according to the contour information of the A2, comparing the similarity of the dust contour information in the gray level diagram of the wafer of the dust detection point at the current production stage with that of the dust detection point at the previous production stage, and outputting a comparison result;
a4: counting the newly appeared dust quantity of the current production stage compared with the previous production stage according to the comparison result of the A3, and if the newly appeared dust quantity exceeds a first threshold value, carrying out early warning on the current production stage;
A5: calculating the total area of dust in the gray scale according to the gray scale in A1;
a6: comparing the dust areas in the gray level diagram of the wafer of the dust detection point of the current and the previous production stage according to the total area of dust of the A5, and outputting a comparison result;
a7: counting the newly-appearing dust area increment of the current production stage compared with the last production stage according to the comparison result of the A6, and if the dust area increment exceeds a second threshold value, carrying out early warning on the current production stage;
a8: after the wafer is produced and processed, generating a wafer map;
a9: detecting the defect type on the wafer map according to the wafer map in A8;
a10: detecting the defect position on the wafer map according to the wafer map in A8;
a11: and predicting fault equipment causing wafer defects according to the defect types and the defect positions detected in the A9 and the A10, and carrying out early warning on the equipment.
A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the intelligent monitoring and early warning method for wafer production.
An apparatus, comprising:
a memory for storing instructions;
and the processor is used for executing the instructions to enable the equipment to execute the operation of realizing the intelligent monitoring and early warning method for wafer production.
Compared with the prior art, the invention has the following beneficial effects:
c1, in the process of conveying the wafer, by dividing different production stages, setting a wafer detection point to detect whether the current production stage has abnormal dust increase phenomenon or not and carrying out abnormal dust early warning, the invention can quickly find the production stage with abnormal dust increase through the abnormal dust early warning function, and the early warning mechanism greatly improves the reliability and stability of the wafer production process;
after the wafer production is finished, the wafer defect is detected according to the wafer map, and fault abnormality early warning is carried out on fault equipment which causes the defect according to the defect type and the defect position;
and C3, the invention builds a two-branch network model for detecting the wafer defect type and the wafer defect position based on the residual network, and further uses the neural network to predict the fault equipment causing the wafer defect based on the two-branch network model.
Drawings
FIG. 1 is a schematic diagram of a system for intelligent monitoring and early warning for wafer production according to the present invention.
Fig. 2 is a schematic diagram of the dust detection module of the intelligent monitoring and early warning system for wafer production according to the present invention.
Fig. 3 is a schematic diagram of the defect detection and fault prediction module of the intelligent monitoring and early warning system for wafer production according to the present invention.
Fig. 4 is a schematic diagram of a network structure of the defect detection and fault equipment prediction module of the intelligent monitoring and early warning system for wafer production according to the present invention.
Detailed Description
The following detailed description of the present invention is made with reference to the accompanying drawings and specific embodiments, and it is to be understood that the specific features of the embodiments and the embodiments of the present invention are detailed description of the technical solutions of the present invention, and not limited to the technical solutions of the present invention, and that the embodiments and the technical features of the embodiments of the present invention may be combined with each other without conflict.
The term "and/or" is merely an association relationship describing an associated object, meaning that there may be three relationships, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. The character "/", generally indicates that the front and rear associated objects are an or relationship.
Example 1
The embodiment provides an intelligent monitoring and early warning system for wafer production, which is used for solving the problems that whether dust is abnormally increased or not and early warning cannot be detected in different production stages or not and fault equipment causing wafer defects cannot be detected and early warning cannot be performed in the prior art. 1-4, an intelligent monitoring and early warning system for wafer production comprises an equipment layer, a database layer and a control layer. In terms of a device layer, the system includes an input device and an output device. The input device comprises a scanning electron microscope, an optical irradiation device and a camera. The scanning electron microscope is used for scanning the surface of the wafer and detecting defects to generate a wafer map; the optical irradiation device is used for displaying dust on the wafer; the camera is used for shooting the wafer image irradiated by the optical irradiation device. The output device comprises a display device, a dust abnormality early warning device and a fault abnormality early warning device. The display device is used for displaying whether the equipment for producing the wafer is abnormal or not in the production stage of the wafer; the dust abnormality early warning device is used for early warning the production stage with abnormal dust increase; the fault abnormality early warning device is used for early warning fault equipment causing wafer defects. The database layer is used for storing data acquired in the working process of the system. The control layer comprises a CPU, a dust detection module, a defect detection and fault equipment prediction module, a communication module, a dust abnormality early warning module and a fault abnormality early warning module. The CPU is used for managing and controlling the operation of the system, and the CPU uses a model suitable for computationally intensive tasks and supporting a deep learning accelerator for managing and controlling the operation of the wafer code recognition system. The dust detection module is used for monitoring whether abnormal dust increase occurs in different production stages. The defect detection and fault equipment prediction module is used for detecting defects on the wafer map, predicting equipment causing defects according to the types and positions of the detected defects, and monitoring whether different equipment fails to cause the wafer defects. The communication module is used for constructing a communication network in the system to realize the mutual transmission of data in the system. And the dust abnormality early warning module is used for carrying out early warning through a dust abnormality early warning device in the output equipment aiming at a certain production stage when the dust abnormality increase in the production stage is detected. And the fault abnormality early warning module is used for predicting production equipment which causes wafer defects to be generated when the wafer defects are detected, and carrying out early warning on the production equipment which has faults through the fault abnormality early warning device in the output equipment.
In this embodiment, the input device includes a scanning electron microscope, where the scanning electron microscope is a irises-4800 series model, and is used to scan a wafer surface to generate a wafer map, and a resolution of a lens system of the scanning electron microscope is between 0.1 nm and 1 nm.
In this embodiment, the input device includes an optical illumination device for visualizing dust on a wafer, and the optical illumination device uses a bright field illumination system, and is composed of a light source, an optical lens, an optical filter, a mirror, and the like. The main function of the device is to provide uniform and stable illumination, and the high-brightness light source is focused by the lens and then uniformly irradiates the wafer to be detected.
In this embodiment, the input device includes a camera, where the camera is used to capture a front image of the wafer after being irradiated by the optical irradiation device, and a lens system of the camera has a magnification of 100-200 times, and a Basler industrial camera is used.
In this embodiment, the dust detection module divides the wafer production process into different stages, the connection point of each stage is set as a wafer dust detection point, the wafer dust detection point uses the optical irradiation device to irradiate the wafer, the irradiation angle of the optical irradiation device selects 0 degree angle orthographic irradiation, the camera is used to capture a front image of the wafer, and the front image captured by the camera is converted into a gray scale image, and the gray scale image is set to pass through the first stage The gray level image obtained by the wafer after the wafer is processed in the production stage through the wafer dust detection point treatment is +.>Wherein->，/>Representing the total number of production stages.
In this embodiment, the dust detection module includes a dust shape matching unit, where the dust shape matching unit includes a contour extraction subunit and a contour matching subunit, and the contour extraction subunit uses a Canny edge detection algorithm to extract the contour of dust in the gray scale map through noise removal, calculation of pixel gradient values and directions, non-maximum suppression, dual-threshold detection, and edge connection. In the noise removing process, the Gaussian smoothing filtering is carried out on the image, so that the noise in the image is reduced; in the process of calculating the gradient value and the direction of the pixel points, a Sobel operator is used for carrying out gradient calculation on the smoothed image to obtain the gradient amplitude value and the direction of each pixel point; in the non-maximum value inhibition process, non-maximum value inhibition is carried out on each pixel point in the gradient direction, namely, a local maximum value is found in the gradient direction, and the non-maximum value point is inhibited to be zero; in the double-threshold detection process, classifying pixel points according to a preset high threshold value and a preset low threshold value, wherein the pixel points higher than the high threshold value are regarded as strong edges, and the pixel points lower than the low threshold value are regarded as weak edges; the pixel point between the two is used for determining whether the pixel point is a weak edge according to whether the pixel point is connected with the strong edge or not; in the edge connection process, whether the weak edge is considered as a final edge is determined according to whether the weak edge is connected with the strong edge or not. The contour information of dust is represented by coordinates of contour pixel points, the coordinates of the contour pixel points are stored in a clockwise order, and the gray level map is set The%>The dust content is->Then->The profile information of the individual dust is +.>Wherein->，/>，/>Representation->Total number of dust in->Representing the total number of production stages.
In this embodiment, the contour matching subunit matches the gray-scale mapEach dust and gray scale image of (a)The similarity of each dust in the dust-to-dust ratio is one-to-one, wherein +.>To pass->Gray scale obtained by processing the wafer processed in the production stage through the wafer dust detection point, wherein +.>，/>Representing production stepsTotal number of segments, provided withMiddle->The dust content is->，/>Middle->The dust content is->Wherein->，，/>，/>Representation->Total number of dust in->Representation->Total number of dust in->Representing the total number of production stages, the similarity comparison method adopts a shape context matching algorithm, wherein the shape context matching algorithm comprises the steps of feature extraction, feature vector coding, normalization, distance calculation and matching, and in the feature extraction step,selecting a group of key points as description of the shape, and calculating the relative position relation between each key point and other points; in the characteristic vector coding step, other points connected with the selected key points are represented as vectors under a polar coordinate system by taking the selected key points as centers, and the lengths and angles of the vectors are converted into characteristic vectors in a histogram form; in the normalization step, the feature vector is normalized so as not to be influenced by the scale and rotation; in the distance calculation step, the similarity of the shapes can be obtained by calculating the distance between the feature vectors of the two shapes; in the matching step, matching and recognition of the shape are performed according to the calculated similarity. Setting a discrimination threshold for the contour matching subunit >Is provided with->Is->And->If the similarity of (1) is compared with ∈>There is->Satisfy->Record->Is old dust, if->Does not have->Satisfy->Record->For new dust, the gray-scale map is traversed>After every dust in (3), counting the number of newly appeared dust +.>If->Wherein->For the first threshold value, the first ∈is displayed by the display device>And the abnormal dust increase phenomenon occurs in each production stage, and the early warning is carried out through the abnormal dust early warning module.
In this embodiment, the dust detection module includes a dust area matching unit at the first stage according to the dust shape matching unitThe +.sup.th extracted from the production stage>Dust->Profile information of (a)Calculation of +.>Dust->Area of (2)Is provided with->Totally contain->Coordinates of pixel points of the outline are set with the +.>The coordinates of each outline pixel point are +.>First->The coordinates of each outline pixel point are +.>Specially, whenWhen make->The coordinates of each outline pixel point are +.>Then->Dust->Is of the area ofWherein->Said passing->Gray level map +. >The total area of all dust in (a) is +.>Comparing said grey-scale patterns +.>Total area of all dust->And gray level map->Total area of all dust->If->Wherein->Is the second threshold value, the first is displayed by the display deviceAnd the abnormal dust increase phenomenon occurs in each production stage, and the early warning is carried out through the abnormal dust early warning module.
In this embodiment, the defect detection and fault device prediction module includes a training data preparation unit, a defect detection unit and a fault device prediction unit, where the training data preparation unit uses a scanning electron microscope to detect a defect on each chip on a wafer, counts the number of defects on each chip, constructs a wafer map according to the counted result, and sets a value of each pixel point on the wafer map as the number of defects on the chip at the position corresponding to the pixel point, and prepares the wafer map of the wafer after the wafer is processed by a production device, and marks the type, the position and the device generating the defects on the wafer map according to the wafer map. The types of defects on the wafer map include nine types in total, namely a central type, an edge ring type, an edge local type, a scratch type, a random distribution type, a hollow type, a near-full type and a None type, wherein the None type indicates that the wafer map has no defects, and the other eight types are defect types in different forms. The location of the defect on the wafer map is represented using coordinates of a plurality of pixel points. All equipment producing wafers is coded and the corresponding code is used to indicate the equipment producing the defect.
In this embodiment, the defect detection unit includes two branches, which are respectively used for detecting a type of a defect and a position of the defect, each branch uses a residual network with fifty stacked layers as a detection model, each layer of network structure of the residual network is a convolutional neural network layer including a short-circuit connection, and a formula of each layer of network structure of the residual network is:wherein->Output representing network structure of each layer, +.>Representing the output of the convolutional neural network layer in the layer network structure,/->Representing the input data, the output of the short-circuited connection is the input data +.>The residual network finally links up a full connection layer for classification, the full connection layer is activated using a softmax function, the softmax function formula is:wherein->Is a length +.>Is a vector of (a). Record the firstOne branch is used for detecting defect type with detection model +.>Is provided with->The input data of (a) is the wafer map, and the wafer map is set as +.>，/>Is vector->Wherein->，Is>Individual element->To predict the wafer map +.>Is category->Probability of>The representation dimension is +.>Of (2) wherein>For the number of defect types, the defect detection unit uses the wafer map +. >And corresponding actual wafer type code +.>Training->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>For the number of defect types ∈ ->The loss function of (2) is:
；
recording the detection model of the second branch for detecting the defect type as，/>Input data of (a) is the wafer map +.>，/>Is vector->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>Is>Individual element->For predicting the wafer map +.>Coordinate information of pixel point of defect position, < >>Totally contain->A pixel dot, wherein->The coordinates of the individual pixels are denoted +.>ThenThe defect detecting unit uses the wafer map +.>And corresponding actual wafer position information +.>Training->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>The total number of coordinate values of pixel points actually used for representing the wafer position information is +.>The loss function of (2) is:
；
in this embodiment, the defect detection unit is trainedAnd->Then, when detecting the defect type and defect position of the wafer, the wafer map of the wafer to be detected is input +.>According to->Output ofJudging the defect type with the largest probability of the wafer, wherein the wafer defects comprise nine types, including eight types of defects: central, edge ring, edge localized, scored, randomly distributed, hollow, and nearly-full, including a defect-free class Type (2): none. If the defect type is None, namely the wafer is not defective, the fault abnormality early warning module is not required to be started for early warning, if the defect type is not None, namely the wafer is defective, the fault abnormality early warning module is started for early warning, and a wafer map of the wafer to be detected is input +.>Obtain->Output result of +.>Finally, will->And->The faulty device prediction unit is input together.
In this embodiment, the fault device prediction model of the fault device prediction unitThe method comprises six linear layers, wherein the first five linear layers are all activated by RELU functions, the last linear layer outputs a prediction result through softmax function processing, and the fault equipment prediction model->Is the two branch model of the defect detection unitAnd->Output result of +.>And->Is of (a) and (b)Splicing result is +.>I.e. +.>Wherein->The representation dimension is +.>Of (2) wherein>For the number of defect types +.>For the total number of coordinate values of pixel points for representing wafer position information, the malfunction apparatus prediction model +.>The output result of (2) is vector->Wherein->The representation dimension is +.>Real space of>Wherein->Represents the total number of all devices in the wafer production process, wherein +. >Is>Individual element->To predict the wafer map is caused +.>The defective device in (1) is->Probability of individual devices, said faulty device prediction unit using +.>And generating the wafer map +.>Actual device coding of defects +.>Training->Wherein->The representation dimension is +.>Is used for the real space of (a),wherein->Indicating the total number of all devices during wafer production, +.>Representation->Is>Element(s)>Is>Individual element->A value of 1 indicates that the wafer map is caused +.>The defective device in (1) is->Personal device, otherwise->The value is 0->Said faulty device prediction model->The loss function of (2) is:
；
in this embodiment, the fault device prediction unit is trainedThen, when detecting the equipment causing the wafer to generate the defect, the output result of the defect detection unit is first added>And->Input->，/>Outputting the prediction resultAnd listing the first ten potential fault devices causing wafer defects according to the probability order, displaying the ten potential fault devices through the display device, displaying the defect types and the positions of the defects on the wafer, and carrying out early warning through the fault abnormality early warning module.
Example 2
The embodiment provides an intelligent monitoring and early warning method for wafer production, which is realized based on the intelligent monitoring and early warning system for wafer production, and comprises the following specific steps:
A1: collecting a front image of the wafer transmitted to the wafer dust detection point, and converting the front image into a gray level image;
a2: extracting outline information of dust in the gray level diagram according to the gray level diagram in the A1;
a3: according to the contour information of the A2, comparing the similarity of the dust contour information in the gray level diagram of the wafer of the dust detection point at the current production stage with that of the dust detection point at the previous production stage, and outputting a comparison result;
a4: counting the newly appeared dust quantity of the current production stage compared with the previous production stage according to the comparison result of the A3, and if the newly appeared dust quantity exceeds a first threshold value, carrying out early warning on the current production stage;
a5: calculating the total area of dust in the gray scale according to the gray scale in A1;
a6: comparing the dust areas in the gray level diagram of the wafer of the dust detection point of the current and the previous production stage according to the total area of dust of the A5, and outputting a comparison result;
a7: counting the newly-appearing dust area increment of the current production stage compared with the last production stage according to the comparison result of the A6, and if the dust area increment exceeds a second threshold value, carrying out early warning on the current production stage;
a8: after the wafer is produced and processed, generating a wafer map;
A9: detecting the defect type on the wafer map according to the wafer map in A8;
a10: detecting the defect position on the wafer map according to the wafer map in A8;
a11: and predicting fault equipment causing wafer defects according to the defect types and the defect positions detected in the A9 and the A10, and carrying out early warning on the equipment.
Example 3
The embodiment provides a computer readable storage medium, which uses a special storage server, a hard disk array or cloud service to store a computer program and data required by a wafer code identification system, and when the computer program is executed by a processor, the intelligent monitoring and early warning method for wafer production is realized.
Example 4
The present embodiment provides an apparatus comprising:
c1, a hard disk memory for storing an instruction set, and a module, a model and an algorithm of the intelligent monitoring and early warning method for wafer production;
and c2, a high-performance image processor is used for executing the instruction, so that the equipment executes the operation of realizing the intelligent monitoring and early warning method for wafer production, has parallel computing capability, and is suitable for rapidly processing image data.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, systems, and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present invention and the scope of the claims, which are all within the protection of the present invention.
Claims (17)
1. An intelligent monitoring and early warning system for wafer production is characterized by comprising an equipment layer, a database layer and a control layer;
the equipment layer comprises input equipment and output equipment, wherein the input equipment comprises a scanning electron microscope, an optical irradiation device and a camera, the scanning electron microscope is used for scanning the surface of a wafer and detecting defects to generate a wafer image, the optical irradiation device is used for displaying dust on the wafer, the camera is used for shooting the wafer image irradiated by the optical irradiation device, the output equipment comprises a display device, a dust abnormality early warning device and a fault abnormality early warning device, the display device is used for displaying the production stage of the wafer, the equipment for producing the wafer is abnormal, the dust abnormality early warning device is used for early warning the production stage with abnormal increase of dust, and the fault abnormality early warning device is used for early warning fault equipment causing the wafer defects;
The database layer is used for storing data acquired in the working process of the system;
the control layer comprises a CPU, a dust detection module, a defect detection and fault equipment prediction module, a communication module, a dust abnormality early warning module and a fault abnormality early warning module.
2. The intelligent monitoring and early warning system for wafer production according to claim 1, wherein the CPU is configured to manage and control operation of the system; the dust detection module is used for monitoring whether abnormal dust increase occurs in different production stages; the defect detection and fault equipment prediction module is used for detecting defects on the wafer map, predicting equipment causing defects according to the types and positions of the detected defects, and monitoring whether different equipment fails to cause the wafer defects; the communication module is used for constructing a communication network in the system; the dust abnormality early warning module is used for carrying out early warning through a dust abnormality early warning device in the output equipment aiming at a certain production stage when detecting that dust abnormality increases in the production stage; and the fault abnormality early warning module is used for predicting production equipment which causes wafer defects to be generated when the wafer defects are detected, and carrying out early warning on the production equipment with faults through the fault abnormality early warning device in the output equipment.
3. The intelligent monitoring and early warning system for wafer production according to claim 2, wherein the input device comprises a scanning electron microscope, the scanning electron microscope is used for scanning the surface of the wafer to generate a wafer map, and the resolution of a lens system of the scanning electron microscope is between 0.1 nm and 1 nm.
4. An intelligent monitoring and warning system for wafer production according to claim 3 wherein the input device comprises an optical illumination device for visualizing dust on the wafer, the optical illumination device using a bright field illumination system to focus a high brightness light source through a lens and uniformly illuminate the wafer to be inspected.
5. The intelligent monitoring and early warning system for wafer production according to claim 4, wherein the input device comprises a camera for capturing a front image of the wafer irradiated by the optical irradiation device, and the lens system of the camera has a magnification of 100-200 times.
6. The intelligent monitoring and warning system for wafer production according to claim 5, wherein the dust detection module divides the wafer production process into different stages, the engagement point of each stage is set to be a wafer dust detection point, the wafer dust detection point irradiates the wafer with the optical irradiation device, the front image of the wafer is photographed with the camera, and the front image of the wafer is converted into a gray scale image, and the gray scale image is set to pass through the first stage The gray level image obtained by the wafer after the wafer is processed in the production stage through the wafer dust detection point treatment is +.>Wherein->，/>Representing the total number of production stages.
7. The intelligent monitoring and early warning system for wafer production according to claim 6, wherein the dust detection module comprises a dust shape matching unit, the dust shape matching unit comprises a contour extraction subunit and a contour matching subunit, the contour extraction subunit uses a Canny edge detection algorithm to extract the contour of dust in the gray scale map through noise removal, calculation of pixel gradient values and directions, non-maximum suppression, double-threshold detection and edge connection, the contour information of the dust is represented by coordinates of contour pixels, the coordinates of the pixels of the contour are stored in a clockwise order, and the gray scale map is providedThe%>The dust content is->Then->The profile information of the individual dust is +.>Wherein，/>，/>Representation->Total number of dust in->Representing the total number of production stages.
8. The intelligent monitoring and warning system for wafer production of claim 7, wherein the contour matching subunit maps the gray scale patternIs +. >The similarity of each dust in the dust-to-dust ratio is one-to-one, wherein +.>To pass->Gray scale obtained by processing the wafer processed in the production stage through the wafer dust detection point, wherein +.>，/>Representing the total number of production stages, let->Middle->The dust content is->，/>Middle->The dust content is->Wherein->，/>，/>，/>Representation ofTotal number of dust in->Representation->Total number of dust in->Representing the total number of production stages, the similarity comparison method adopts a shape context matching algorithm to set a discrimination threshold for the contour matching subunit ++>Is provided with->Is->Andif the similarity of (1) is compared with ∈>There is->Satisfy->Record->As old dust, ifDoes not have->Satisfy->Record->For new dust, the gray-scale map is traversed>After every dust in (3), counting the number of newly appeared dust +.>If->Wherein->Is a first threshold valueThen display the +.>And the abnormal dust increase phenomenon occurs in each production stage, and the early warning is carried out through the abnormal dust early warning module.
9. The intelligent monitoring and warning system for wafer production of claim 8, wherein the dust detection module comprises a dust area matching unit, the dust area matching unit being in a first position according to the dust shape matching unit The +.sup.th extracted from the production stage>Dust->Profile information of->Calculation of +.>Dust->Area of (1)/(5)>Totally contain->Coordinates of pixel points of the outline are set with the +.>The coordinates of each outline pixel point are +.>First->The coordinates of each outline pixel point are +.>；
Wherein whenWhen make->The coordinates of each outline pixel point are +.>Then->Dust->Is +.>Wherein->Said passing->Gray level map +.>The total area of all dust in (a) is +.>Comparing said grey-scale patterns +.>Total area of all dust->And gray scale mapTotal area of all dust->If->Wherein->For the second threshold value, the first is displayed by the display device>And the abnormal dust increase phenomenon occurs in each production stage, and the early warning is carried out through the abnormal dust early warning module.
10. The intelligent monitoring and early warning system for wafer production according to claim 9, wherein the defect detection and fault equipment prediction module comprises a training data preparation unit, a defect detection unit and a fault equipment prediction unit, the training data preparation unit uses a scanning electron microscope to detect defects of each chip on a wafer, counts the number of the defects of each chip, constructs a wafer map according to the counted result, wherein the value of each pixel point on the wafer map is the number of the defects of the chip at the position corresponding to the pixel point, and the training data preparation unit prepares the wafer map of the wafer after the wafer is processed by production equipment, and marks the type, the position and equipment for generating the defects on the wafer map according to the wafer map.
11. According to claim 10The intelligent monitoring and early warning system for wafer production is characterized in that the defect detection unit comprises two branches, wherein the two branches are respectively used for detecting the type of the defect and the position of the defect, each branch adopts a residual network with fifty stacked layers as a detection model, the residual network is finally connected with a full-connection layer for classification, the full-connection layer is activated by using a softmax function, and the detection model of the first branch for detecting the type of the defect is recorded as followsIs provided withThe input data of (a) is the wafer map, and the wafer map is set as +.>，/>Is vector of output data of (a)Wherein->，/>Is>Individual element->To predict the wafer map +.>Is a category ofProbability of>The representation dimension is +.>Of (2) wherein>For the number of defect types, the defect detection unit uses the wafer map +.>And corresponding actual wafer type code +.>Training->WhereinWherein->The representation dimension is +.>Of (2) wherein>For the number of defect types ∈ ->The loss function of (2) is:
；
recording the detection model of the second branch for detecting the defect type as，/>Input data of (a) is the wafer map +.>，/>Is vector->Wherein- >WhereinThe representation dimension is +.>Of (2) wherein>Is>Individual element->For predicting the wafer mapCoordinate information of pixel point of defect position, < >>Totally contain->A pixel dot, wherein->The coordinates of the individual pixels are denoted +.>ThenThe defect detecting unit uses the wafer map +.>And corresponding actual wafer position information +.>Training->Wherein->Wherein->The representation dimension is +.>Of (2) wherein>The total number of coordinate values of pixel points actually used for representing the wafer position information is +.>The loss function of (2) is:
。
12. the intelligent monitoring and warning system for wafer production of claim 11, wherein the defect detection unit is trainedAnd->Then, when detecting the defect type and defect position of the wafer, the wafer map of the wafer to be detected is input +.>According to->Output->Judging the defect type with the largest probability of the wafer, if the defect type is None, namely the wafer is not defective, starting a fault abnormality early warning module to early warn, if the defect type is not None, namely the wafer is defective, starting the fault abnormality early warning module to early warn, and inputting a wafer map of the wafer to be detected >Obtain->Output result of +.>Finally, will->And->Input togetherThe faulty device prediction unit.
13. The intelligent monitoring and early warning system for wafer production of claim 12, wherein the fault device prediction model of the fault device prediction unitThe method comprises six linear layers, wherein the first five linear layers are all activated by RELU function, the last linear layer outputs a prediction result through softmax function processing, and the fault equipment prediction modelIs the two branch models of the defect detection unit +.>And->Output result of (2)And->The splicing result is thatI.e. +.>Wherein->Representing dimensions asOf (2) wherein>For the number of defect types +.>For the total number of coordinate values of pixel points for representing wafer position information, the malfunction apparatus prediction model +.>The output result of (2) is vector->Wherein->The representation dimension is +.>Real space of>Wherein->Represents the total number of all devices in the wafer production process, wherein +.>Is>Individual element->To predict the wafer map is caused +.>The defective device in (1) is->Probability of individual devices, saidBarrier device prediction Unit use->And generating the wafer map +. >Actual device coding of defects in a memoryTraining->Wherein->The representation dimension is +.>Real space of>Wherein->Indicating the total number of all devices during wafer production, +.>Representation->Is>Element(s)>Is>Individual element->A value of 1 indicates that the wafer map is caused +.>The defective device in (1) is->Personal device, otherwise->The value is set to 0 and the value is set to 0,said faulty device prediction model->The loss function of (2) is:
。
14. the intelligent monitoring and early warning system for wafer production of claim 13, wherein the fault device prediction unit is trainedThen, when detecting the equipment causing the wafer to generate the defect, the output result of the defect detection unit is first added>And->Input->，/>Output prediction result +.>And listing fault equipment causing wafer defects according to the probability order, displaying the fault equipment causing the wafer defects through the display equipment, displaying the defect types and the positions of the defects on the wafer, and carrying out early warning through the fault abnormality early warning module.
15. An intelligent monitoring and early warning method for wafer production, which is realized based on the intelligent monitoring and early warning system for wafer production according to any one of claims 1 to 14, and is characterized in that the method comprises the following specific steps:
A1: collecting a front image of the wafer transmitted to the wafer dust detection point, and converting the front image into a gray level image;
a2: extracting outline information of dust in the gray level diagram according to the gray level diagram in the A1;
a3: according to the contour information of the A2, comparing the similarity of the dust contour information in the gray level diagram of the wafer of the dust detection point at the current production stage with that of the dust detection point at the previous production stage, and outputting a comparison result;
a4: counting the newly appeared dust quantity of the current production stage compared with the previous production stage according to the comparison result of the A3, and if the newly appeared dust quantity exceeds a first threshold value, carrying out early warning on the current production stage;
a5: calculating the total area of dust in the gray scale according to the gray scale in A1;
a6: comparing the dust areas in the gray level diagram of the wafer of the dust detection point of the current and the previous production stage according to the total area of dust of the A5, and outputting a comparison result;
a7: counting the newly-appearing dust area increment of the current production stage compared with the last production stage according to the comparison result of the A6, and if the dust area increment exceeds a second threshold value, carrying out early warning on the current production stage;
a8: after the wafer is produced and processed, generating a wafer map;
A9: detecting the defect type on the wafer map according to the wafer map in A8;
a10: detecting the defect position on the wafer map according to the wafer map in A8;
a11: and predicting fault equipment causing wafer defects according to the defect types and the defect positions detected in the A9 and the A10, and carrying out early warning on the equipment.
16. A computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements an intelligent monitoring and pre-warning method for wafer production as claimed in claim 15.
17. An apparatus, comprising: a memory for storing instructions; a processor for executing the instructions to cause the apparatus to perform operations implementing an intelligent monitoring and early warning method for wafer production as set forth in claim 15.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311213205.3A CN116978834B (en) | 2023-09-20 | 2023-09-20 | Intelligent monitoring and early warning system for wafer production |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311213205.3A CN116978834B (en) | 2023-09-20 | 2023-09-20 | Intelligent monitoring and early warning system for wafer production |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116978834A CN116978834A (en) | 2023-10-31 |
| CN116978834B true CN116978834B (en) | 2023-11-24 |
Family
ID=88481790
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311213205.3A Active CN116978834B (en) | 2023-09-20 | 2023-09-20 | Intelligent monitoring and early warning system for wafer production |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116978834B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117316835B (en) * | 2023-11-28 | 2024-02-02 | 泓浒(苏州)半导体科技有限公司 | Environment detection and early warning method for semiconductor wafer transmission under ultrahigh vacuum environment |
| CN117539223B (en) * | 2023-11-30 | 2024-06-04 | 西安好博士医疗科技有限公司 | Intelligent electric heating control system and method for incubator |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102522350A (en) * | 2011-11-29 | 2012-06-27 | 上海华力微电子有限公司 | Method and device for detecting failed production machine platforms |
| CN112928181A (en) * | 2021-01-22 | 2021-06-08 | 徐州中辉光伏科技有限公司 | Crystal silicon battery surface treatment device with ozone purging function |
| CN215678076U (en) * | 2021-02-08 | 2022-01-28 | 合肥晶合集成电路股份有限公司 | Monitoring system and detection machine |
| CN115223881A (en) * | 2022-07-07 | 2022-10-21 | 苏州普汇达电子科技有限公司 | Wafer defect detection and position correction device and method for chip etching machine |
| CN116773548A (en) * | 2023-08-21 | 2023-09-19 | 泓浒(苏州)半导体科技有限公司 | Wafer surface defect detection method and system |
-
2023
- 2023-09-20 CN CN202311213205.3A patent/CN116978834B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102522350A (en) * | 2011-11-29 | 2012-06-27 | 上海华力微电子有限公司 | Method and device for detecting failed production machine platforms |
| CN112928181A (en) * | 2021-01-22 | 2021-06-08 | 徐州中辉光伏科技有限公司 | Crystal silicon battery surface treatment device with ozone purging function |
| CN215678076U (en) * | 2021-02-08 | 2022-01-28 | 合肥晶合集成电路股份有限公司 | Monitoring system and detection machine |
| CN115223881A (en) * | 2022-07-07 | 2022-10-21 | 苏州普汇达电子科技有限公司 | Wafer defect detection and position correction device and method for chip etching machine |
| CN116773548A (en) * | 2023-08-21 | 2023-09-19 | 泓浒(苏州)半导体科技有限公司 | Wafer surface defect detection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116978834A (en) | 2023-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116978834B (en) | Intelligent monitoring and early warning system for wafer production | |
| CN111325713B (en) | Neural network-based wood defect detection method, system and storage medium | |
| US10964004B2 (en) | Automated optical inspection method using deep learning and apparatus, computer program for performing the method, computer-readable storage medium storing the computer program, and deep learning system thereof | |
| CN111693534B (en) | Surface defect detection method, model training method, device, equipment and medium | |
| CN116188475B (en) | Intelligent control method, system and medium for automatic optical detection of appearance defects | |
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN111652085B (en) | Object identification method based on combination of 2D and 3D features | |
| CN113095438B (en) | Wafer defect classification method, device and system thereof, electronic equipment and storage medium | |
| CN111833306A (en) | Defect detection method and model training method for defect detection | |
| CN113706490B (en) | Wafer defect detection method | |
| CN117635565B (en) | Semiconductor surface defect detection system based on image recognition | |
| CN109840900A (en) | A kind of line detection system for failure and detection method applied to intelligence manufacture workshop | |
| CN113727022B (en) | Method and device for collecting inspection image, electronic equipment and storage medium | |
| CN111179250A (en) | Industrial product defect detection system based on multitask learning | |
| CN113850749A (en) | Method for training defect detector | |
| CN113516619B (en) | Product surface flaw identification method based on image processing technology | |
| CN115100104A (en) | Defect detection method, device and equipment for glass ink area and readable storage medium | |
| CN111415339A (en) | Image defect detection method for complex texture industrial product | |
| TWI747686B (en) | A defect detection method and a defect detection device | |
| CN117252815A (en) | Industrial part defect detection method, system, equipment and storage medium based on 2D-3D multi-mode image | |
| CN111707672A (en) | Method for detecting surface defects of wind power rotary supporting piece | |
| CN117197146A (en) | Automatic identification method for internal defects of castings | |
| CN114120318B (en) | Dark field image target point accurate extraction method based on integrated decision tree | |
| CN117372770A (en) | PCB micro defect detection and identification method based on photometric stereo and deep learning | |
| CN111563869A (en) | Stain testing method for quality inspection of camera module |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |