JP2006118870A - Inspection condition setting method and device, and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inspection condition setting method which enables the selection of an optimum inspection condition when a flaw is detected by a Die to Die inspection system and a Ref comparative inspection system. <P>SOLUTION: In a flaw inspection system 1, a Die to Die inspection means 31a of a tuning server 30 is constituted so as to compare the predetermined regions of image data with each other to calculate a first threshold value not detecting a dummy flaw from the image data on the basis of the comparison result. A threshold value setting means 31 detects a first threshold value becoming the minimum value from the first threshold value to select the image data corresponding to the first threshold value becoming the minimum value as reference image data. A Ref comparative inspection means 31b compares the selected reference image data with other image data to calculate a second threshold value not detecting a dummy flaw at every other image data on the basis of the comparison result. Then, the threshold setting part 31 sets inspection condition information, based on the second threshold value and the reference image data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention mainly relates to an inspection condition setting method, an inspection condition setting apparatus, and a computer program for inspecting a surface defect such as a semiconductor wafer and a liquid crystal glass substrate.

Conventionally, in the inspection for detecting the presence or absence of defects on the surface of a semiconductor wafer or the like, the following two inspection methods are mainly used.
(Inspection method 1)
The first inspection method is a method called a Die to Die inspection method. In image data obtained by imaging a semiconductor wafer, two or more different semiconductor chip portions (the case where the semiconductor chip portion is called a Die) When four semiconductor chips are taken, one of the four chips may be called “Die”. Image data of a predetermined area is compared for each predetermined area, and a luminance difference is calculated from the difference image. This is an inspection method for detecting defects.
(Inspection method 2)
The second inspection method is a method referred to as a reference comparison inspection method or a Wafer to Wafer inspection method (hereinafter referred to as a Ref comparison inspection method), and an image of a good whole semiconductor wafer without defects and an image of a semiconductor wafer to be inspected. And detecting a defect by calculating a luminance difference from the difference image.

  However, each of the above two inspection methods has advantages and disadvantages. That is, the die to die inspection method has an advantage that inspection can be performed within image data obtained by imaging one semiconductor wafer, while defects are detected in comparison with each die. If there is a defect on the entire surface of the semiconductor wafer, it is difficult to detect the defect.

  On the other hand, the Ref comparative inspection method has the advantage of being able to detect defects in a wide region such as film unevenness of the entire semiconductor wafer, but on the other hand, an image of a good semiconductor wafer having no defects serving as a reference for comparison is obtained. It has to be taken as a reference image, and if the inspection conditions are not regularly reviewed, there are disadvantages in that the production line varies and cannot cope with changes over time.

Therefore, in a semiconductor wafer defect inspection method, a method of performing defect detection on a semiconductor wafer that sequentially flows through a manufacturing process by a die to die inspection method, a method of performing defect detection by a Ref comparison inspection method, a die to die inspection method, and the like. And a method for detecting defects by combining the Ref comparison inspection method and the like have been proposed (see, for example, Patent Document 1).
JP 2002-267615 A

  However, in the conventional technique, in the Die to Die inspection method and the Ref comparison inspection method, when selecting inspection conditions such as a threshold value and a reference image as a determination criterion in defect detection, it is performed manually. In many cases, it is difficult to select inspection conditions that are optimal for defect detection using a computer or the like.

  The present invention has been made in order to solve the above-described problems, and the object of the present invention is to select an optimum inspection condition when performing defect detection by the Die to Die inspection method and the Ref comparison inspection method. The object is to provide an inspection condition setting device.

  In order to solve the above-described problem, the present invention calculates a luminance difference between image data obtained by sequentially imaging a plurality of inspected objects and image data to be compared, and the luminance difference is calculated. An inspection condition setting method for setting inspection condition information used in a defect inspection method in which a portion exceeding a predetermined threshold is a defect, wherein predetermined regions of the image data are compared, and a first result is based on a comparison result. A first calculation step of calculating a threshold value for each image data; a selection step of selecting the image data corresponding to the first threshold value as a reference image; the selected reference image data and other image data; And a second calculation step of calculating a second threshold value for each of the other image data based on a comparison result of comparing the two.

  The present invention calculates a luminance difference from image data to be compared with respect to image data acquired by sequentially imaging a plurality of inspected objects, and defines a portion where the luminance difference exceeds a predetermined threshold as a defect. An inspection condition setting method for setting inspection condition information used for a defect inspection method to perform comparison between predetermined regions of the image data, and a first that does not detect a defect from the image data based on a comparison result A first calculation step of calculating a threshold value for each image data; and detecting the first threshold value that is the minimum value from the first threshold values calculated for each image data, and A defect is detected from the other image data based on a selection process of selecting the image data corresponding to the first threshold as reference image data and a comparison result of comparing the selected reference image data with other image data. Shina A second calculation step of calculating a second threshold value for each of the other image data; and an inspection condition setting step of setting the inspection condition information based on the second threshold value and the reference image data. This is an inspection condition setting method.

  According to the present invention, in the above invention, re-inspection information is stored in association with re-examination result information (for example, a review result in the embodiment) that records whether the image data has passed or failed by re-inspection. Retest information storage process stored in the means (for example, the temporary storage unit 26 and the test data storage unit 24) and the second value that is the maximum value among the second threshold values calculated by the second calculation process. A re-inspection information reading process for reading out the re-inspection result information of the image data corresponding to the second threshold value that is the detected maximum value from the re-inspection result information storage unit, and the re-inspection information A deletion step of deleting the second threshold value corresponding to the image data when the re-examination result information read out by the reading step fails. And butterflies.

  According to the present invention, in the above invention, in the inspection condition setting process, a standard deviation value of the first threshold value is calculated, and an average value of the first threshold value is calculated based on the standard deviation value. A value obtained by multiplying the average value by a predetermined coefficient (for example, “a coefficient in consideration of occurrence of pseudo defects due to factors such as variations in the manufacturing process (1.2 to 2 as an experience value)” in the embodiment) A first setting process set in the inspection condition information, a maximum value is detected from the second threshold value, and a value obtained by multiplying the maximum value by a predetermined coefficient and the reference image data are used as the inspection condition information. And a second setting process for setting.

  According to the present invention, in the above invention, a threshold value recording process for recording the second threshold value calculated in the second threshold value calculation process in the threshold information storage means, and a threshold information storage means in response to a user instruction. The second threshold value is read, the second threshold value is graphed and displayed on a screen (for example, the screen of the client terminal 11), and the test result information and the retest result information corresponding to the second threshold value are displayed. And a detection process for detecting a numerical value on the coordinate axis of the graph at a location selected by the user's operation on the screen, and the inspection condition setting process is detected by the detection process The method further includes a step of setting the inspection condition information based on the numerical value.

  The present invention calculates a luminance difference between image data obtained by sequentially imaging a plurality of inspected objects and image data to be compared, and sets a portion where the luminance difference exceeds a predetermined threshold as a defect. An inspection condition setting device (for example, tuning server 30) for setting inspection condition information used in a defect inspection device (for example, defect inspection device 12) to be inspected, comparing predetermined regions of the image data, and comparing Corresponding to the first threshold value calculated by the first calculation means (for example, the die to die inspection means 31a) for calculating the first threshold value for each image data based on the result, and the first calculation means. Selection means for selecting the image data to be used as a reference image, and a second result based on a comparison result obtained by comparing the reference image data selected by the selection means with other image data. An inspection condition setting apparatus comprising: a second calculation unit (Ref comparison inspection unit 31b) that calculates a threshold value for each of the other image data.

  The present invention calculates a luminance difference between image data obtained by sequentially imaging a plurality of inspected objects and image data to be compared, and sets a portion where the luminance difference exceeds a predetermined threshold as a defect. A computer of an inspection condition setting apparatus for setting inspection condition information used for a defect inspection apparatus to be inspected compares predetermined areas of the image data, and calculates a first threshold value for each image data based on the comparison result. A first calculating unit; a selecting unit that selects the image data corresponding to the first threshold value calculated by the first calculating unit as a reference image; the reference image data selected by the selecting unit; A computer program for causing a second threshold value to be calculated for each of the other image data based on a comparison result obtained by comparing image data.

  According to the inspection condition setting method of the present invention, predetermined regions of image data are compared, and a first threshold value is calculated for each image data from the image data based on the comparison result. Then, image data corresponding to the first threshold value is selected as reference image data, and a second threshold value is calculated for each other image data based on a comparison result obtained by comparing the selected reference image data with other image data. It is the composition to do. Therefore, inspection condition information used for the defect inspection method can be set based on the first threshold value, the second threshold value, and the reference image data. As a result, it is possible to select an inspection condition that incorporates the degree of influence of an actual manufacturing line, and a defect inspection apparatus that performs inspection based on the inspection condition information can perform suitable defect detection.

Hereinafter, an inspection condition setting apparatus for executing an inspection condition setting method according to an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a schematic block diagram showing the defect inspection system according to the present embodiment.
In FIG. 1, a defect inspection system 1 includes a client terminal 11, a defect inspection device 12, a review device 13, an image server 20, a tuning server 30, a LAN (Local Area Network) 15 connecting them, and the Internet. 16 and a gateway 14 that connects the LAN 15.

  In this defect inspection system 1, the client terminal 11 is a personal computer, for example, and is connected to the image server 20 and the tuning server 30 via the LAN 15 by an operator's operation.

  The image server 20 includes a reference information database (hereinafter abbreviated as DB) 23, an inspection data storage unit 24, a recipe file storage unit 25, a search unit 22, and a communication unit 21. In the image server 20, the reference information DB 23 uses information such as search conditions and dates for inspections performed by the defect inspection apparatus 12 as keywords, image data stored in the inspection data storage unit 24 in association with the keywords, and the like. Reference information to the information is stored. The inspection data storage unit 24 stores image data obtained by imaging a semiconductor wafer and information on an inspection result when the image data is inspected by the defect inspection apparatus 12.

  The recipe file storage unit 25 stores information on inspection conditions used when the defect inspection apparatus 12 performs inspection, reference image data, and the like. The search unit 22 monitors whether or not the predetermined data is stored in the temporary storage unit 26 that is a part of the inspection data storage unit 24. If the predetermined data is stored, the search unit 22 obtains information on the image file and the inspection result based on the data. Information is stored in the reference information DB 23 at the same time as being stored in the examination data storage unit 24 according to a predetermined folder configuration. The search unit 22 receives a search request including keywords such as inspection conditions and dates, performs a search based on information registered in the reference information DB 23, and transmits the search result to the request source. The communication unit 21 is connected to the LAN 15 and transmits / receives information to / from other devices. The communication unit 21 has a function as a web server, receives a connection request from the client terminal 11, and the client terminal 11 remotely operates the search unit 22 by an operator's operation on the web browser. Making it possible. Furthermore, the communication unit 21 also receives connections from other client terminals that connect to the Internet 16 via the gateway 14.

  The tuning server 30 corresponds to an inspection condition setting device that executes the above-described inspection condition method, and generates an inspection condition such as a threshold used in the defect inspection based on the inspection result of the defect inspection device 12. Set to 20. In the tuning server 30, the threshold setting unit 31 includes a Die to Die inspection unit 31 a and a Ref comparison inspection unit 31 b, reads a recipe file from the recipe file storage unit 25 of the image server 20, and the Die to Die inspection unit 31 a Search conditions are generated based on the information obtained by the Ref comparison inspection means 31b, and the generated inspection conditions are set in the recipe file storage unit 25. The Die to Die inspection unit 31a calculates a threshold value that does not detect a pseudo defect for each image data by the Die to Die inspection method described above. The Ref comparison inspection means 31b compares the reference image data with other image data by the Ref comparison inspection method described above, and calculates a threshold value that does not detect a pseudo defect.

  The database monitoring unit 32 detects registration of new information by monitoring the reference information DB 23 when new information is registered in the examination data storage unit 24 of the image server 20. The communication unit 33 is connected to the LAN 15 and transmits / receives information to / from other devices. Further, the communication unit 33 has a function as a web server, receives a connection request from the client terminal 11, and the client terminal 11 remotely operates the threshold setting unit 31 by an operator's operation on the web browser. It is possible to do. Furthermore, the communication unit 33 also receives connections from other client terminals that connect to the Internet 16 via the gateway 14.

  The defect inspection apparatus 12 downloads a recipe file from the recipe file storage unit 25 of the image server 20, performs defect inspection on image data obtained by imaging a semiconductor wafer based on the inspection conditions included in the recipe file, and performs defect inspection Is recorded in the temporary storage unit 26. In addition, a plurality of defect inspection apparatuses 12 (12-1 to 12-n) are installed on a line in a factory in order to perform macro inspection of a semiconductor wafer for each manufacturing process.

  The review device 13 is a device that inspects (reviews) the image data inspected by the defect inspection device 12 again. The review device 13 includes a plurality of scanning electron microscopes (SEMs) that measure the dimensions of a fine pattern formed in a semiconductor wafer manufacturing process, a review microscope, and a spectroscopic instrument that measures the spectrum of a semiconductor wafer. It is comprised by the apparatus (13-1 to 13-n). Further, the review device 13 downloads the image data inspected by the defect inspection device 12 from the inspection data storage unit 24, reviews the image data, and records the review result in the temporary storage unit 26. When the defect inspection device 12 and the review device 13 store the image and the inspection information file in the temporary storage unit 26, the search unit 22 decodes the device, the result, and the like from the inspection information file and adds them to the reference information DB 23. Then, the image and the inspection information file are moved to the inspection data storage unit 23 according to a predetermined file name and folder structure.

FIG. 2 is a diagram showing an outline of the data structure of the inspection information file stored in the inspection data storage unit 24.
The inspection data storage unit 24 stores information for each lot. Here, the lot refers to a unit of a group of products manufactured at a time in a semiconductor wafer manufacturing process, and in this embodiment, information on 25 semiconductor wafers is included in one lot. In FIG. 2, a lot information file is a file in which information of inspection results for each semiconductor wafer (hereinafter referred to as a sample) included in the lot is recorded. The sample information file is a file in which information on the inspected sample is recorded. The sample image data is image data inspected by the defect inspection device 12 and the review device 13. The sample image data is not only the image data that actually captured the sample, but also the corrected image that was subjected to image processing to avoid false defects due to differences in the characteristics of the device and the sample, and the captured image that was colored In addition, defect image data and the like displayed in an overlapping manner are also included.

  FIG. 3 shows the data structure of the lot information file 24a described in FIG. The lot information file 24a includes a header 24a-1 and data 24a-2. The header 24a-1 includes information on the date and time when the inspection was performed, a lot ID previously assigned to each lot so that the lot can be uniquely identified, a product name indicating the product type of the sample, and which process The process name indicating whether or not the sample is manufactured, the file name of the recipe file 25a used for the inspection, and the determination result of the lot are recorded.

  In addition, data 24a-2 records information on the inspection results of 25 samples. In the data 24a-2, the slot number of the wafer cassette storing each sample for one lot is recorded in “slot number”. In the “wafer ID”, an ID given in advance for each sample is recorded so that each sample can be uniquely identified. In the “number of defects”, the number of defects detected in the inspection is recorded. In the “number of defective chips”, the number of defective chips is recorded. In “Area”, the area of the defect is recorded. In “determination”, the determination of the presence / absence of a defect in the defect inspection apparatus 12 is recorded, 0 indicates that the inspection result is acceptable, and 1 indicates that the defect is not acceptable. In the “review”, the result of the review in the review device 13 is recorded, 0 indicates that the review result is acceptable, and 1 indicates that the result is unacceptable. In the “search state”, information indicating whether or not the sample is searched and used to generate the inspection condition when the inspection condition is generated by the tuning server 30 is recorded, and 0 is not searched. 1 indicates that the search has been completed, and 2 indicates that the image is set as reference image data. In “Ref threshold value”, the threshold value obtained by the Ref comparison inspection unit 31b is recorded. In the “Die threshold”, the threshold obtained by the Die to Die inspection unit 31a is recorded. Although not shown, the inspection information file also includes information on the inspected device information and inspection conditions.

  FIG. 4 is a diagram showing an outline of the data configuration of the recipe file storage unit 25. Here, a recipe is an inspection condition for detecting a defect in a sample. The recipe file storage unit 25 stores an inspection condition file, reference image data, and a reference creation image association file. In the recipe file storage unit 25, the inspection condition file is a file in which numerical data such as the amount of light when the sample is imaged in the inspection, the imaging mode, the threshold value for determining pass / fail in the Die to Die inspection or the Ref comparison inspection, and the like are recorded. It is.

  Specific inspection condition file information includes slot number information, transport method information, optical condition information, sample map information, inspection parameter information, defect extraction parameter information, pattern matching information, determination condition information, rework information, and warning condition information. Exists. The slot number information is a slot number in which a sample to be inspected is arranged. The transfer method information is information for sorting the slot number of the sample to be returned and the wafer cassette according to the inspection result. The optical condition information is information such as the type of interference filter and polarizing filter for inspection, the presence or absence of both filters, the illumination light quantity in illumination and imaging, and the sample rotation angle. The sample map information is sample map information such as the size and center coordinates of the chip divided in the sample, the number of divisions, and the coordinates and shape information of the region to be inspected.

  The inspection parameter information is information such as inspection modes such as Die to Die inspection and Ref comparison inspection, presence / absence of defect classification, and the name of image data used as a reference. The defect extraction parameter information is information on detailed parameters of inspection including a non-inspection area and an inspection threshold within the inspection area. The pattern matching information is information including an image position to be subjected to pattern matching and its parameters in order to correct a positional deviation between the inspection target image data and the reference image data. Judgment condition information is information that determines pass / fail of a sample according to parameters and classification names for defect classification, area and number of parts that failed in inspection, number of corresponding chips, defect classification results, and priority by classification It is. The rework information is information on the presence / absence of a function for counting the number of times the manufacturing process is redone. The warning condition information is information that defines a warning condition.

  Among the information included in the above inspection condition file, sample map information, inspection target area, pattern matching information, etc. are created based on reference image data, and matching between the inspection image and reference image data is performed by pattern matching during inspection. Has been taken. In addition to the above information, the inspection condition file includes information such as conditions and parameters for a measurement function using a captured image.

  The reference image data is image data used as a reference for comparison in the Ref comparison inspection. The reference creation image association file is a file in which information indicating from which image data the reference image data is stored in the examination data storage unit 24 is recorded. The inspection condition file, reference image data, and reference creation image association file are created for each process and product type, and a combination of these is collectively referred to as a recipe file 25a.

Next, the flow of the inspection process for defect inspection will be described with reference to FIGS. FIG. 5 is a flowchart showing the flow of the defect inspection process.
First, when a sample is loaded into the defect inspection apparatus 12 (step S1), a screen (reference numeral 12a) as shown in FIG. 6 is displayed on the screen of the defect inspection apparatus 12. Here, the inspection condition is specified by the product type name (symbol 12a-1) indicating the sample format and the process name (symbol 12a-2) indicating the manufacturing process. Next, the defect inspection apparatus 12 downloads the recipe file 25a from the recipe file storage unit 25 of the image server 20 using the product name and process name as keywords (step S2). And the defect inspection apparatus 12 images a sample based on the inspection conditions and reference image data contained in the downloaded recipe file 25a, and performs defect detection (step S3).

  When the defect detection is completed, the defect inspection apparatus 12 records the inspection result in the temporary storage unit 26 of the image server 20 (step S4). When new inspection result data is recorded in the temporary storage unit 26, the search unit 22 of the image server 20 generates a keyword or the like based on the newly recorded inspection result data, and stores it in the data of the reference information DB 23. Add (step S5). Further, the data is converted into a predetermined file name / folder structure, and the data in the temporary storage unit 26 is moved to the inspection data storage unit 24.

  After the inspection is completed, the operator can operate the client terminal 11 to connect to the image server 20 and refer to the inspection result by the web browser of the client terminal 11. The inspection result display screen (symbol 11a) shown in FIG. 7 is displayed on the screen of the client terminal 11 when referring to the inspection result. In the inspection result display screen, the defect (reference numeral 11a-2, reference numeral 11a-3) is displayed superimposed on the image data obtained by imaging the sample. The operator can display only captured image data or display only defects by operating buttons on the screen. Further, the table on the left side of the inspection result display screen (reference numeral 11a-4) is classified as defect information by the number and area of defects recorded in the lot information file 24a described above, and ADC (Analog Digital Converter). The classified result is displayed (step S6).

  The defect inspection apparatus 12 is also connected to a communication infrastructure in the production factory (not shown), and the inspection result of the sample is notified to the production factory management system. Based on this information, the management system issues an instruction to the transport system in the factory so as to load the sample into the review device 13. The review device 13 into which the sample has been loaded executes a review (step S7). The review device 13 stores the review result in the temporary storage unit 26 (including the image), records it in the “review” column of the lot information file 24a by the search unit 22 (step S8), and inspects the image and the inspection information file. The data is converted into a predetermined format and moved to the data storage unit 24.

Next, three embodiments relating to setting processing of inspection conditions such as threshold values and reference images performed by the tuning server 30 will be described.
(First embodiment)
The inspection condition setting process in the first embodiment will be described with reference to FIGS.
First, when the lot information file 24a is newly registered in the reference information DB 23 of the image server 20, the database monitoring unit 32 of the tuning server 30 detects that the new lot information file 24a is registered. The threshold setting unit 31 that has received the information in which the new lot information file 24 a is registered from the database monitoring unit 32 connects to the image server 20 via the communication unit 33. The threshold value setting unit 31 searches for reference information from the reference information DB 23 to the newly registered lot information file 24 a stored in the inspection data storage unit 24. Then, the lot information file 24 a is read from the search data storage unit 24. Based on the read lot information file 24a, image data corresponding to the retrieved lot information file 24a is read from the inspection data storage unit 24 as inspection data necessary for the inspection condition setting process, and corresponds to the lot information file 24a. The recipe file 25a to be read is read out from the recipe storage unit 25 (step Sa1).

  Next, the Die to Die inspection unit 31a of the threshold setting unit 31 performs Die to Die inspection on the read image data, and obtains a threshold for detecting a pseudo defect for each sample while changing the threshold. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Sa2). The threshold value setting unit 31 generates the threshold value list 40 shown in FIG. 9 on the memory, and records the threshold value for each sample obtained by the Die to Die inspection unit 31a in the threshold value list 40 (step Sa3). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data, and if not completed, repeats the processing from step Sa1 to step Sa3 (step Sa4).

  On the other hand, when the Die to Die inspection for all the image data is completed, the threshold value setting unit 31 sets the minimum threshold value among the threshold values obtained for each sample by the Die to Die inspection unit 31a from the threshold list 40. And the image data corresponding to the threshold value is registered as reference image data in the Ref comparison inspection (step Sa5). Next, the Ref comparison inspection unit 31b of the threshold setting unit 31 performs a Ref comparison inspection based on the reference image data to obtain a threshold value that does not detect a pseudo defect. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Sa6). The threshold value setting unit 31 records the threshold value for each sample obtained by the Ref comparison inspection means 31b in the threshold value list 40 in the process of obtaining the threshold value at which no pseudo defect is detected (step Sa7). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data. If the search has not been completed, the processing from step Sa6 to step Sa7 is repeated (step Sa8).

  After the above process is completed, the threshold setting unit 31 obtains the maximum threshold value registered in step Sa3 and the maximum threshold value registered in step Sa7 from the threshold list 40. Then, a threshold value that does not detect pseudo defects in all samples obtained by the Die to Die inspection unit 31a, a threshold value that does not detect pseudo defects in all samples obtained by the Ref comparison inspection unit 31b, and reference image data are used as recipes. The file is stored in the recipe file 25a corresponding to the inspection in the file storage unit 25. At this time, a threshold value multiplied by a coefficient (1.2-2 as an experience value) taking into account the occurrence of pseudo defects due to factors such as variations in the manufacturing process is registered in the recipe file 25a. Also, it is desirable to save the version of the recipe file so that it can be restored to the previous recipe file.

  Further, the threshold setting unit 31 registers the information of the threshold list 40 in the lot information file 24a. The threshold value for detecting the pseudo defect for each sample obtained by the Die to Die inspection unit 31a is recorded in the “Die threshold value” column of the lot information file 24a. Further, the threshold value for each sample obtained by the Ref comparison inspection means 31b is recorded in the “Ref threshold value” column of the lot information file 24a. In the “search state” column of the lot information file 24a, “1: inspected” is recorded for the searched image data, and “2” for the image data selected as the reference image data. : Reference image "is recorded (step Sa9).

  It should be noted that the threshold information that does not detect defects in all samples obtained by the Die to Die inspection unit 31a is not set in the recipe file 25a, but the standard deviation of all thresholds obtained by the Die to Die inspection unit 31a. An average value may be calculated from the above, and a value obtained by multiplying the average value by the above coefficient may be registered in the recipe file 25a.

FIG. 10 is a graph showing threshold values for detecting a pseudo defect obtained by the Ref comparison inspection unit 31b for each wafer ID read from the lot information file 24a, displayed on the screen of the client terminal 11.
When referring to the screen of FIG. 10 from the client terminal 11, the operator operates the client terminal 11 to connect to the image server 20 and input information of a recipe to be referred to the search unit 22 of the image server 20. The client terminal 11 that has received the response from the search unit 22 displays some information searched by the information of the recipe on the screen. When the character string of the statistical information is selected from the displayed information, the client terminal 11 that has received the operation of the operator can connect to the threshold setting unit 31 of the tuning server 30 and display the screen of FIG. it can.

  In FIG. 10, the inspection result in the defect inspection apparatus 12 of the image data corresponding to the threshold value that does not detect the pseudo defect is displayed by being classified by the mark type or color (red and blue). Further, the image data that is determined by the defect inspection apparatus 12 and the review apparatus 13 are further classified and displayed by another mark (star shape) or color (yellow). Further, a threshold line 40a for classifying image data whose image data has failed in defect inspection (defect inspection NG) and other image data is also displayed. Further, when the wafer IDs are arranged in time series, the temporal change of the inspection result can be referred to from this graph.

  By operating the client terminal 11 and selecting a mark with a pointer device such as a mouse, the operator can display an inspection result display screen (symbol 11a) on the screen. Can be confirmed. Further, when it is confirmed by visual observation by the operator that there is no defect in the image data (star shape in the figure) that has passed the review result (review OK) in the review device 13, the operator uses the client terminal. 11 is moved to move the threshold line 40a to the position 40b so that the threshold value of the image data that has passed the review result does not exceed the threshold line. The threshold setting unit 31 detects a threshold value corresponding to the threshold line 40 b and registers the threshold corresponding to the threshold line 40 b in the recipe file 25 a of the recipe file storage unit 25 in the image server 20.

  As described above, the inspection by the defect inspection apparatus 12 is executed under optimum conditions by the operator appropriately changing the threshold setting by referring to the statistical information displayed on the screen and operating the client terminal 11. Thus, the accuracy of the inspection conditions can be improved.

  In the first embodiment, the threshold setting unit 31 performs a process for obtaining a threshold that does not detect a pseudo defect in each image data. However, if the inspection time is to be shortened, the tuning server 30 is added to the defect inspection apparatus 12. You may make it provide the structure of. Further, the configuration of the tuning server 30 may be provided in the image server 20.

  With the configuration of the first embodiment, it is possible to obtain a threshold value at which no pseudo defect is detected in the Die to Die inspection. In addition, a Ref comparison inspection is performed using an image having a minimum threshold value among threshold values that do not detect a pseudo defect obtained in the Die to Die inspection, and a threshold value that does not detect a pseudo defect in the Ref comparison inspection is obtained. Image data can be extracted. As a result, by using these threshold values and reference images as inspection conditions in defect inspection in the defect inspection apparatus 12, it is possible to select inspection conditions that incorporate the degree of influence by the actual production line, and to detect defects suitably. Can be done.

  In addition, by displaying the threshold value for each sample in a graph, the operator can visually check the threshold state, and further, the operator operates the client terminal to select an appropriate threshold value in the recipe file. It becomes possible to register. Accordingly, even if an abnormal threshold value is detected, it can be deleted by an operator's operation, and a more accurate threshold value can be obtained.

(Second Embodiment)
Next, the inspection condition setting process in the second embodiment will be described with reference to FIG. The inspection condition setting process in the second embodiment is a process performed when a large number of semiconductor wafers as samples are not manufactured and the inspection conditions cannot be selected stably.

  First, when a lot information file 24a is newly registered in the reference information DB 23 of the image server 20, the database monitoring unit 32 of the tuning server 30 detects that a new lot information file 24a is registered. The threshold setting unit 31 that has received the information in which the new lot information file 24 a is registered from the database monitoring unit 32 connects to the image server 20 via the communication unit 33. The threshold value setting unit 31 searches for reference information from the reference information DB 23 to the newly registered lot information file 24 a stored in the inspection data storage unit 24. Then, the lot information file 24 a is read from the search data storage unit 24, and the recipe file 25 a corresponding to the lot information file 24 a is read from the recipe storage unit 25. The reference image data of the read recipe file 25a is deleted, and the contents of the reference creation image association file are deleted and initialization is performed (step Sb1).

  Next, the threshold value setting unit 31 reads out image data corresponding to the retrieved lot information file 24a from the inspection data storage unit 24 as inspection data necessary for the inspection condition setting process (step Sb2). Next, the Die to Die inspection unit 31a of the threshold setting unit 31 obtains, for each image data, a threshold that does not detect a pseudo defect while varying the threshold by the Die to Die inspection on the read image data. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Sb3). The threshold value setting unit 31 generates the threshold value list 40 on the memory, and records the threshold value for each sample obtained by the Die to Die inspection unit 31a in the threshold value list 40 (step Sb4). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data, and if not complete, repeats the processing from step Sb2 to step Sb4 (step Sb5).

  When the Die to Die inspection for all the read image data is completed, the threshold setting unit 31 sets the minimum threshold among the thresholds obtained for each sample from the threshold list 40 by the Die to Die inspection unit 31a. And the image data corresponding to the threshold value is registered as reference image data for the Ref comparison inspection. As a feature of the Die to Die inspection, there is a tendency that the threshold value is set lower for a sample with less unevenness or change in the entire sample, and the image data with the least unevenness among the image data retrieved by the above processing is used as reference image data. It will be registered (step Sb6).

  Next, the Ref comparison inspection unit 31b of the threshold setting unit 31 performs a Ref comparison inspection based on the registered reference image data to obtain a threshold value that does not detect a pseudo defect. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Sb7). In the process of obtaining the threshold value, the threshold value setting unit 31 determines whether the threshold value can be obtained from the image data (step Sb8). When it is determined that the image data is abnormal image data for which the threshold value cannot be obtained, the threshold value setting unit 31 displays a warning on the screen and does not record information related to the image data in the threshold value list 40 (step Sb9). On the other hand, in the case of image data for which a threshold value can be obtained, the threshold value setting unit 31 records the threshold value obtained by the Ref comparison inspection unit 31b in the threshold value list 40 (step Sb10). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data. If the search has not been completed, the processing from step Sb7 to step Sb10 is repeated (step Sb11).

  On the other hand, when the search has been completed for all the read images, the threshold value setting unit 31 detects the maximum threshold value among the threshold values obtained for each sample by the Ref comparison inspection unit 31b from the threshold value list 40. Then, image data corresponding to the threshold value is read from the inspection data storage unit 24 (step Sb12). Then, the result of review of the read image data is read from the lot information file 24a, and it is determined whether the image data has passed or failed (step Sb13). If the review result is acceptable or there is no review result, it is determined whether the threshold value of the Ref comparison inspection or the Die to Die inspection of the image data exceeds an abnormal value indicating a predetermined abnormality. To do. Here, a large value is set to such an extent that the abnormal value is clearly recognized as abnormal (step Sb14). When the abnormal value is exceeded and the review result is NG, information related to the image data is deleted from the threshold list 40 (step Sb15). Then, a warning display is displayed on the screen (step Sb16).

  Next, when the threshold values of the Ref comparison inspection and the Die to Die inspection do not exceed the abnormal values in step Sb14, the threshold value that is preset with the maximum threshold value obtained for each sample by the Ref comparison inspection is set. It is determined whether or not the value is equal to or less than the convergence value indicating the degree of convergence (step Sb17). If it is not less than the convergence value, the image data corresponding to the maximum threshold is registered in the recipe file 25a as the second reference image data. At this time, the reference image data is converted into a composite image of two pieces of image data, and information of the second piece of image data is recorded in the reference creation image association file of the recipe file 25a (step Sb18). Next, the threshold setting unit 31 determines whether or not the reference image data has been registered more than a predetermined registration number. Here, the predetermined registered number is a sufficient number for which the threshold value is stabilized (step Sb19). If it is not registered more than the predetermined number of registered sheets, the process returns to step Sb7 to extract reference image data again.

  If the maximum threshold value of the Ref comparison inspection is equal to or less than the convergence value in step Sb17, or if a predetermined number of registered sheets or more are registered in step Sb19, the threshold setting unit 31 obtains the value by the die to die inspection means 31a. The maximum threshold value information and the maximum threshold value and the reference image data among the threshold values obtained by the Ref comparison inspection unit 31b are registered in the recipe file 25a of the recipe file storage unit 25. At this time, a threshold value multiplied by a coefficient (1.2-2 as an experience value) taking into account the occurrence of pseudo defects due to factors such as variations in the manufacturing process is registered in the recipe file 25a.

  Further, the threshold setting unit 31 registers the information of the threshold list 40 in the lot information file 24a. The threshold value for detecting the pseudo defect for each sample obtained by the Die to Die inspection unit 31a is recorded in the “Die threshold value” column of the lot information file 24a. Further, the threshold value for each sample obtained by the Ref comparison inspection means 31b is recorded in the “Ref threshold value” column of the lot information file 24a. In the “search state” column of the lot information file 24a, “1: inspected” is recorded for the searched image data, and “2” for the image data selected as the reference image data. : Reference image "is recorded (step Sb20). Thereafter, the search unit 22 that has detected that the lot information file 24a of the search data storage unit 24 has been updated updates the reference information DB 23 based on the information added to the lot information file 24a.

  It should be noted that the information on the maximum threshold value among the threshold values obtained by the Die to Die inspection means 31a is not registered in the recipe file 25a, but is averaged from the standard deviations of all the threshold values obtained by the Die to Die inspection means 31a. A value may be calculated and a value obtained by multiplying the average value by the above coefficient may be registered in the recipe file 25a.

  Further, when the warning is displayed, if the threshold setting unit 31 and the client terminal 11 are connected, the warning display may be displayed on the screen of the client terminal 11.

  With the configuration of the second embodiment, it is possible to obtain a threshold value that incorporates information on the review result by the review device 13 and to extract a reference image. Accordingly, by using these threshold values and reference images as inspection conditions in defect inspection in the defect inspection apparatus 12, it becomes possible to perform defect inspection incorporating accuracy by the review apparatus 13 in addition to the degree of influence by the production line. .

(Third embodiment)
Next, the inspection condition setting process in the third embodiment will be described with reference to FIGS. The inspection condition setting process in the third embodiment is a process for setting an inspection condition by combining the Die to Die inspection method and the Ref comparison inspection method in accordance with the increase in the number of sample lots.

  First, the database monitoring unit 32 detects that information related to a new lot has been registered in the reference information DB 23 (step Sc1). The threshold setting unit 31 receives information from the database monitoring unit 32 that a new lot has been detected. Then, it is determined whether the total number of lots to be inspected is smaller than n1 set in advance (step Sc2). If the total number of lots is smaller than n1, first processing for setting a threshold based on the Die to Die inspection is executed (step Sc3). On the other hand, when the total number of lots exceeds n1, it is determined whether or not the total number of lots is equal to n1 (step Sc4).

  If the total number of lots is equal to n1, a second process for setting an inspection condition that combines the die-to-die inspection and the Ref comparison inspection is executed (step Sc5). Next, whether the total number of lots exceeds n1 and whether the total number of lots is a predetermined number (n1 + N × m: m = 1, 2, 3,...) Determine (step Sc6). If the total number of lots is not a predetermined number, the process returns to step Sc1 to monitor registration of a new lot. If the total number of lots is a predetermined number, it is next determined whether or not the total number of lots is smaller than n2. Here, n2 is a number that allows the threshold to be selected stably (step Sc7).

  When the total number of lots is smaller than n2, a third process for performing a threshold setting process by Ref inspection is executed (step Sc11). On the other hand, when the total number of lots is equal to or greater than n2, it is determined based on the lot information file 24a whether there is image data having a different review result and determination result (step Sc8). If there is no image data different from the review result, the process returns to step Sc1 to monitor the registration of a new lot. On the other hand, if there is image data different from the review result, it is next determined whether or not the number of image data different from the review result exceeds a preset stable number. Here, the stable number is a value sufficient to stabilize the threshold (step Sc9). If the number of image data different from the review result does not exceed the stable number, the process proceeds to step Sc11 and the third process is executed. On the other hand, if the number of image data different from the review result does not exceed the stable number, the second process is performed (step Sc10).

  FIG. 13 is a timing chart showing the processing flow of FIG. 12 with the total number of lots on the horizontal axis. As shown in FIG. 13, when the total number of lots is less than n1, that is, when so many semiconductor wafers are not yet inspected, the inspection conditions are set by the first process based on the die-to-die inspection, When the number reaches n1, a second process is performed in which a threshold is set by combining the Die to Die inspection and the Ref comparison inspection. Thereafter, until the total number of lots is a predetermined number (n1 + N × m: m = 1, 2, 3,...), That is, when a certain number of semiconductor wafers are inspected, the inspection conditions determined by the second process Is used. When the total number of lots reaches a predetermined number (n1 + N × m: m = 1, 2, 3,...), Next, the number of semiconductors is smaller than the total number of lots, that is, does not require a die-to-die inspection. When the wafer is inspected, a third process for performing a threshold setting process based only on the Ref inspection is performed. When the total number of lots is equal to or greater than n2, the threshold setting process is performed by selecting the second process or the third process based on the number of image data different from the review result.

Next, the first process described above will be described with reference to FIGS. FIG. 14 is a flowchart showing the flow of the first process described above.
First, the threshold setting unit 31 searches for reference information from the reference information DB 23 to the newly registered lot information file 24 a stored in the inspection data storage unit 24. Then, the lot information file 24 a is read from the search data storage unit 24. Based on the read lot information file 24a, image data corresponding to the retrieved lot information file 24a is read from the inspection data storage unit 24 as inspection data necessary for the inspection condition setting process, and corresponds to the lot information file 24a. The recipe file 25a to be read is read from the recipe storage unit 25 (step Sd1).

  Next, the Die to Die inspection unit 31a of the threshold setting unit 31 performs Die to Die inspection on the read image data, and obtains a threshold for detecting a pseudo defect for each sample while changing the threshold. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Sd2). The threshold value setting unit 31 generates the threshold value list 40 on the memory, and records the threshold value for each sample obtained by the Die to Die inspection unit 31a in the threshold value list 40 (step Sd3). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data, and if not completed, repeats the processing from step Sd1 to step Sd3 (step Sd4).

  On the other hand, when the Die to Die inspection for all the image data is completed, the threshold value setting unit 31 calculates the standard deviation of the threshold values recorded in the threshold value list 40 and obtains the optimum threshold value. For example, even when the threshold value for one lot has the relationship shown in FIG. 15, the average value of the threshold value can be obtained by calculating the standard deviation as shown in FIG. Can be selected as the optimum threshold value (step Sd5). Then, the threshold setting unit 31 multiplies the optimum threshold by a coefficient (1.2 to 2 as an experience value) that takes into account the occurrence of pseudo defects due to factors such as variations in the manufacturing process, and registers them in the recipe file 25a. .

  Further, the threshold setting unit 31 registers the information of the threshold list 40 in the lot information file 24a. The threshold value for each sample obtained by the Die to Die inspection means 31a is recorded in the “Die threshold value” column of the lot information file 24a. In the “search state” column of the lot information file 24a, “1: inspected” is recorded for the searched image data, and the process returns to the process of FIG. 12 (step Sd6).

  In the first process, by calculating the standard deviation of the threshold value for each sample of the Die to Die test and obtaining the average value, it is possible to select an appropriate threshold value while avoiding the influence of the abnormal sample.

FIG. 17 is a flowchart showing the flow of the second process described above.
First, the threshold setting unit 31 searches for reference information from the reference information DB 23 to the newly registered lot information file 24 a stored in the inspection data storage unit 24. Then, the lot information file 24 a is read from the search data storage unit 24, and the recipe file 25 a corresponding to the lot information file 24 a is read from the recipe storage unit 25. The reference image data of the read recipe file 25a is deleted, and the contents of the reference creation image association file are deleted and initialization is performed (step Se1).

  Next, the threshold value setting unit 31 reads out image data corresponding to the retrieved lot information file 24a from the inspection data storage unit 24 as inspection data necessary for the inspection condition setting process (step Se2). Next, the Die to Die inspection unit 31a of the threshold setting unit 31 obtains, for each image data, a threshold that does not detect a pseudo defect while varying the threshold by the Die to Die inspection on the read image data. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Se3). The threshold value setting unit 31 and the threshold value list 40 are generated on the memory, and the threshold value for each sample obtained by the Die to Die inspection unit 31a is recorded in the threshold value list 40 (step Se4). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data, and if not complete, repeats the processing from step Se2 to step Se4 (step Se5).

  When the Die to Die inspection for all the read image data is completed, the threshold setting unit 31 sets the minimum threshold among the thresholds obtained for each sample from the threshold list 40 by the Die to Die inspection unit 31a. And the image data corresponding to the threshold value is registered as reference image data for the Ref comparison inspection. As a feature of the Die to Die inspection, there is a tendency that the threshold value is set lower for a sample with less unevenness or change in the entire sample, and the image data with the least unevenness among the image data retrieved by the above processing is used as reference image data. It will be registered (step Se6).

  Next, the Ref comparison inspection unit 31b of the threshold setting unit 31 performs a Ref comparison inspection based on the registered reference image data to obtain a threshold value that does not detect a pseudo defect. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. (Step Se7). In the process of obtaining the threshold value, the threshold value setting unit 31 determines whether or not the threshold value can be obtained from the image data (step Se8). When it is determined that the image data is abnormal image data for which the threshold value cannot be obtained, the threshold value setting unit 31 displays a warning on the screen and does not record information related to the image data in the threshold value list 40 (step Se9). On the other hand, in the case of image data for which a threshold value can be obtained, the threshold value setting unit 31 records the threshold value obtained by the Ref comparison inspection unit 31b in the threshold value list 40 (step Se10). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data. If the search has not been completed, the processing from step Se7 to step Se10 is repeated (step Se11).

  On the other hand, when the search has been completed for all the read images, the threshold value setting unit 31 detects the maximum threshold value among the threshold values obtained for each sample by the Ref comparison inspection unit 31b from the threshold value list 40. Then, image data corresponding to the threshold value is read from the inspection data storage unit 24 (step Se12). Then, the result of review of the read image data is read from the lot information file 24a, and it is determined whether the image data has passed or failed (step Se13). If the review result is acceptable or there is no review result, it is determined whether the threshold value of the Ref comparison inspection or the Die to Die inspection of the image data exceeds an abnormal value indicating a predetermined abnormality. To do. Here, the abnormal value is set to a value large enough to be recognized as abnormal (step Se14). When the abnormal value is exceeded and the review result is NG, information related to the image data is deleted from the threshold list 40 (step Se15). Then, a warning display is displayed on the screen (step Se16).

  Next, if the review result is acceptable in step Se13, or if the threshold values for the Ref comparison inspection and the Die to Die inspection do not exceed the above abnormal values in step Se14, they are obtained for each sample by the Ref comparison inspection. It is determined whether or not the maximum value of the threshold value is equal to or less than the convergence value indicating the degree of convergence of the preset threshold value (step Se17). If it is not less than the convergence value, the image data corresponding to the maximum threshold is registered in the recipe file 25a as the second reference image data. At this time, the reference image data is converted into a composite image of two pieces of image data, and information on the second piece of image data is recorded in the reference creation image association file of the recipe file 25a (step Se18). Next, the threshold setting unit 31 determines whether or not the reference image data has been registered more than a predetermined registration number. Here, the predetermined registered number is a sufficient number for which the threshold value is stabilized (step Se19). If the predetermined number of registered sheets has not been registered, the process returns to step Se7, and reference image data is extracted again.

  If the maximum threshold value of the Ref comparison inspection is equal to or less than the convergence value in step Se17, or if a predetermined number of registered sheets or more are registered in step Se19, the threshold setting unit 31 sets the threshold value finally obtained. Information and reference image data are registered in the recipe file 25 a of the recipe file storage unit 25. At this time, a threshold value multiplied by a coefficient (1.2-2 as an experience value) taking into account the occurrence of pseudo defects due to factors such as variations in the manufacturing process is registered in the recipe file 25a.

  Further, the threshold setting unit 31 registers the information of the threshold list 40 in the lot information file 24a. The threshold value for each sample obtained by the Die to Die inspection means 31a is recorded in the “Die threshold value” column of the lot information file 24a. The threshold value for each sample obtained by the Ref comparison inspection means 31b is recorded in the “Ref threshold value” column of the lot information file 24a. In the “search state” column of the lot information file 24a, “1: inspected” is recorded for the searched image data, and “2” for the image data selected as the reference image data. : Reference image ”is recorded, and the process returns to the process of FIG. 12 (step Se20).

  FIG. 18 is a flowchart showing the flow of the third process described above. The third process is a process that is performed when a number of semiconductor wafers that do not require a Die to Die inspection are inspected, and thus the Die to Die inspection process in the second process is omitted. Yes.

  First, the threshold setting unit 31 searches for reference information from the reference information DB 23 to the newly registered lot information file 24 a stored in the inspection data storage unit 24. Then, the lot information file 24 a is read from the search data storage unit 24, and the recipe file 25 a corresponding to the lot information file 24 a is read from the recipe storage unit 25. Then, the image data corresponding to the retrieved lot information file 24a is read from the inspection data storage unit 24 as inspection data necessary for the inspection condition setting process (step Sf1).

  Next, the Ref comparison inspection unit 31b of the threshold setting unit 31 performs a Ref comparison inspection based on the reference image data of the recipe file 25a to obtain a threshold value that does not detect a pseudo defect. Here, obtaining a threshold value that does not detect a pseudo defect means detecting a minimum value of the threshold value. The threshold setting unit 31 generates a threshold list 40 on the memory, and records the threshold obtained by the Ref comparison inspection unit 31b (step Sf2). In the process of obtaining the threshold value, the threshold value setting unit 31 determines whether the threshold value can be obtained from the image data (step Sf3). When it is determined that the image data is abnormal image data for which the threshold value cannot be obtained, the threshold value setting unit 31 displays a warning on the screen and does not record information related to the image data in the threshold value list 40 (step Sf4). On the other hand, in the case of image data for which a threshold value can be obtained, the threshold value setting unit 31 records the threshold value obtained by the Ref comparison inspection unit 31b in the threshold value list 40 (step Sf5). The threshold value setting unit 31 determines whether or not the search has been completed for all of the read image data, and if not complete, repeats the processing from step Sf2 to step Sf5 (step Sf6).

  On the other hand, when the search has been completed for all the read images, the threshold value setting unit 31 detects the maximum threshold value among the threshold values obtained for each sample by the Ref comparison inspection unit 31b from the threshold value list 40. Then, image data corresponding to the threshold value is read from the inspection data storage unit 24 (step Sf7). Then, the result of review of the read image data is read from the lot information file 24a, and it is determined whether the image data has been passed or not (step Sf8). If the review result is acceptable or there is no review result, it is determined whether or not the threshold value for the Ref comparison inspection of the image data exceeds an abnormal value indicating a predetermined abnormality. Here, the abnormal value is set to a value that is clearly recognized as abnormal (step Sf9). When the abnormal value is exceeded and the review result is NG, information related to the image data is deleted from the threshold list 40 (step Sf10). If the client terminal 11 is connected, a warning display is displayed on the screen of the client terminal 11 (step Sf11).

  Next, if the review result is acceptable in step Sf8, or if the threshold values for the Ref comparison inspection and the Die to Die inspection do not exceed the above abnormal values in step Sf9, they are obtained for each sample by the Ref comparison inspection. It is determined whether or not the maximum value of the threshold value is equal to or less than a convergence value indicating a convergence degree of a preset threshold value (step Sf12). If it is not less than the convergence value, the image data corresponding to the maximum threshold is registered in the recipe file 25a as the second reference image data. At this time, the reference image data is converted into a composite image of the two image data, and the information of the second image data is recorded in the reference creation image association file of the recipe file 25a (step Sf13). Next, the threshold setting unit 31 determines whether or not the reference image data has been registered more than a predetermined registration number. Here, the predetermined registered number is a sufficient number for which the threshold value is stabilized (step Sf14). If it is not registered more than the predetermined number of registered sheets, the process returns to step Sf2, and the reference image data is extracted again.

  If the maximum threshold value of the Ref comparison inspection is equal to or less than the convergence value in step Sf12, or if a predetermined number of registered sheets or more are registered in step Sf14, the threshold setting unit 31 sets the threshold value finally obtained. Information and reference image data are registered in the recipe file 25 a of the recipe file storage unit 25. At this time, a threshold value multiplied by a coefficient (1.2-2 as an experience value) taking into account the occurrence of pseudo defects due to factors such as variations in the manufacturing process is registered in the recipe file 25a.

  Further, the threshold setting unit 31 registers the information of the threshold list 40 in the lot information file 24a. The threshold value for each sample obtained by the Ref comparison inspection means 31b is recorded in the “Ref threshold value” column of the lot information file 24a. In the “search state” column of the lot information file 24a, “1: inspection completed” is recorded for the searched image data, and for the image data selected as the reference image data, “ 2: Reference image ”is recorded, and the process returns to the process of FIG. 12 (step Sf15).

  With the configuration of the third embodiment, the threshold and reference that consistently incorporate the degree of influence by the production line and the review result by the review device 13 from when the sample is first produced on the production line until it is produced in large quantities. Inspection conditions such as images can be obtained. Thereby, it is possible to proceed with defect inspection without manually adjusting inspection conditions from the time when the production line is operated, and it is possible to improve production efficiency.

  In the first embodiment, the second embodiment, and the third embodiment described above, the inspection condition setting process is started when new information is registered by the database monitoring unit 32. The client terminal 11 may be connected to the threshold value setting unit 31 by the operation of and the inspection condition setting process may be started based on the operation of the operator.

  The tuning server 30, the image server 20, and the client terminal 11 described above have a computer system inside. The above-described inspection condition setting process is stored in a computer-readable recording medium in the form of a program, and the above process is performed by the computer reading and executing this program. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Alternatively, the computer program may be distributed to the computer via a communication line, and the computer that has received the distribution may execute the program.

It is a block diagram which shows the defect inspection system by this embodiment. It is the figure which showed the data structure of the test | inspection data preservation | save part in the embodiment. It is the figure which showed the lot information file in the same embodiment. It is the figure which showed the data structure of the recipe file storage part in the same embodiment. It is the flowchart which showed the flow of the defect inspection process in the embodiment. It is the figure which showed the screen display of the defect inspection apparatus in the embodiment. It is a screen display of the test result displayed on the client terminal in the embodiment. It is the flowchart which showed the flow of the inspection condition setting process in 1st Embodiment. It is the figure which showed the data structure of the threshold value list in 1st Embodiment. It is the figure which graphed the threshold value list displayed on the client terminal in 1st Embodiment. It is the flowchart which showed the flow of the inspection condition setting process in 2nd Embodiment. It is the flowchart which showed the flow of the inspection condition setting process in 3rd Embodiment. It is the timing chart which showed the flow of inspection condition setting processing in 3rd execution form. It is the flowchart which showed the flow of the 1st process in 3rd Embodiment. It is FIG. (1) for demonstrating the standard deviation calculation in the 1st process in 3rd Embodiment. It is FIG. (2) for demonstrating the standard deviation calculation in the 1st process in 3rd Embodiment. It is the flowchart which showed the flow of the 2nd process in 3rd Embodiment. It is the flowchart which showed the flow of the 3rd process in 3rd Embodiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect inspection system 12 Defect inspection apparatus 13 Review apparatus 20 Image server 30 Tuning server 31 Threshold setting part 31a Die to Die inspection means 31b Ref comparison inspection means

Claims (7)

A defect inspection method for calculating a luminance difference between image data obtained by sequentially imaging a plurality of inspected objects and image data to be compared, and having a portion where the luminance difference exceeds a predetermined threshold as a defect An inspection condition setting method for setting inspection condition information used for
A first calculation step of comparing predetermined regions of the image data and calculating a first threshold value for each of the image data based on a comparison result;
A selection process of selecting the image data corresponding to the first threshold as a reference image;
A second calculation step of calculating a second threshold value for each of the other image data based on a comparison result obtained by comparing the selected reference image data with other image data;
The inspection condition setting method characterized by including. A defect inspection method for calculating a luminance difference between image data obtained by sequentially imaging a plurality of inspected objects and image data to be compared, and having a portion where the luminance difference exceeds a predetermined threshold as a defect An inspection condition setting method for setting inspection condition information used for
A first calculation step of comparing a predetermined area of the image data, and calculating a first threshold value for each of the image data that does not detect a defect from the image data based on a comparison result;
The first threshold value that is the minimum value is detected from the first threshold values calculated for each of the image data, and the image data corresponding to the first threshold value that is the minimum value is selected as reference image data. The selection process,
A second calculation step of calculating, for each of the other image data, a second threshold value that does not detect a defect from the other image data based on a comparison result obtained by comparing the selected reference image data with the other image data; When,
An inspection condition setting process for setting the inspection condition information based on the second threshold and the reference image data;
The inspection condition setting method characterized by including. A re-inspection information storage process for storing re-examination result information in which re-inspection information is recorded in the re-inspection information storage means in association with the image data;
The second threshold value that is the maximum value among the second threshold values calculated by the second calculation process is detected, and the image data corresponding to the second threshold value that is the detected maximum value is detected. Re-inspection information reading process for reading out re-inspection result information from the re-inspection result information storage means;
A deletion step of deleting the second threshold value corresponding to the image data when the re-examination result information read out by the re-inspection information reading step is unacceptable;
The inspection condition setting method according to claim 2, further comprising: In the inspection condition setting process,
A standard deviation value of the first threshold value is calculated, an average value of the first threshold value is calculated based on the standard deviation value, and a value obtained by multiplying the average value by a predetermined coefficient is the inspection condition information. A first setting process to set to
A second setting step of detecting a maximum value from the second threshold, and setting a value obtained by multiplying the maximum value by a predetermined coefficient and the reference image data in the inspection condition information;
The inspection condition setting method according to claim 2, further comprising: A threshold value recording step of recording the second threshold value calculated by the second threshold value calculation step in a threshold value information storage means;
In response to a user instruction, the second threshold value is read from the threshold value information storage means, the second threshold value is graphed and displayed on the screen, and the inspection result information corresponding to the second threshold value and the second threshold value are displayed. A display process for displaying test result information;
Detecting a numerical value on a coordinate axis of the graph at a location selected by the user's operation on the screen, and
The inspection condition setting process includes:
5. The inspection condition setting method according to claim 2, further comprising a step of setting the inspection condition information based on the numerical value detected by the detection process. Defect inspection in which a difference in brightness between image data to be compared is calculated for image data acquired by sequentially imaging a plurality of objects to be inspected, and a portion where the brightness difference exceeds a predetermined threshold is inspected as a defect An inspection condition setting device for setting inspection condition information used in the apparatus,
A first calculation unit that compares predetermined regions of the image data and calculates a first threshold value for each of the image data based on a comparison result;
Selection means for selecting, as a reference image, the image data corresponding to the first threshold value calculated by the first calculation means;
Second calculation means for calculating a second threshold value for each of the other image data based on a comparison result obtained by comparing the reference image data selected by the selection means with other image data;
An inspection condition setting device characterized by comprising: Defect inspection in which a difference in brightness between image data to be compared is calculated for image data acquired by sequentially imaging a plurality of objects to be inspected, and a portion where the brightness difference exceeds a predetermined threshold is inspected as a defect A computer of an inspection condition setting apparatus for setting inspection condition information used in the apparatus compares a predetermined area of the image data, and calculates a first threshold value for each image data based on the comparison result means,
Selecting means for selecting, as a reference image, the image data corresponding to the first threshold value calculated by the first calculating means;
Second calculation means for calculating a second threshold value for each of the other image data based on a comparison result obtained by comparing the reference image data selected by the selection means with other image data;
Computer program to function as.

Images