JP2008046012A - Defect detector and defect detection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily detect defects in an inspection object region, in a defect detection that uses an imaging mechanism of high resolution. <P>SOLUTION: A base plate 9, of which a predetermined unit pattern is formed on each of a plurality of dies on a principal surface, is held on a stage 2, and the imaging mechanism 3 acquires images of multiple gradations in a belt region that is narrower than the width of the dies, regarding the direction transverse to the moving direction along the principal surface of the base plate 9, by moving a line sensor 313 relatively and continuously with respect to the base plate 9 in the moving direction. In a storage part 53, a reference image is stored substantially indicating the die as a whole and as reference for defect detection. When defects are to be detected, a partial image, containing a portion corresponding to the inspection object region, is extracted from the reference image, the inspection object image indicating the inspection object region is acquired by the imaging mechanism 3, and by comparing the partial image with the image of inspecting object, the defects in the region of inspecting object are detected easily. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for detecting a pattern defect on a substrate.

  Various inspection methods have been used in the field of inspecting the appearance of a semiconductor substrate. For example, in Patent Document 1, for a plurality of dies (pellets) that are portions to be chips in a semiconductor substrate, one reference die is imaged and stored as a reference pattern, and an arbitrary position on the semiconductor substrate is stored. There has been proposed a defect inspection apparatus that detects a defect in comparison with an imaging result of a pattern formed on another die.

On the other hand, in recent years, a pattern formed on a semiconductor substrate has been miniaturized. Therefore, in Patent Document 2, one die of a semiconductor substrate is divided into a plurality of band regions (swaths), and imaging is performed for each band region while moving the imaging device relative to the semiconductor substrate to achieve high resolution. In this defect detection apparatus, a first image of one band region of a die that is a reference on a semiconductor substrate is acquired and stored. Subsequently, by detecting a second image of the corresponding band region of the die to be inspected and comparing these images, defect detection of a fine pattern is realized. Patent Document 2 also discloses a method of acquiring another first image, comparing it with the second image, and detecting a defect in the first image based on the two comparison results.
Japanese Patent Laid-Open No. 11-40638 JP 2005-351431 A

  By the way, in the defect detection apparatus of patent document 2, when the defect exists in the die | dye used as a reference | standard, even if the corresponding site | part in a to-be-tested die is not a defect, it may be detected as a defect. Further, by detecting a defect in the first image based on the comparison result between the two first images and the second image, it is possible to solve the above problem, but the processing required for defect detection is complicated. turn into.

  The present invention has been made in view of the above problems, and an object of the present invention is to easily detect a defect in a region to be inspected in defect detection using a high-resolution imaging mechanism that acquires an image of a band region.

  The invention according to claim 1 is a defect detection apparatus for detecting a defect of a pattern on a substrate, and holds a substrate on which a predetermined unit pattern is formed in each of a plurality of unit regions on a main surface. And an image pickup device, and the image pickup device is moved relative to the substrate in a moving direction along the main surface, so that a width is a width of a unit region with respect to a direction perpendicular to the moving direction. An imaging mechanism for acquiring a multi-gradation image of a narrower band area, a storage unit for storing a multi-gradation reference image that substantially shows the entire unit area and is used as a reference for defect detection; and A partial image extraction unit that extracts a partial image including a portion corresponding to a region to be inspected in a unit region that is a target of defect detection; and an image to be inspected that is acquired by the imaging mechanism and indicates the region to be inspected And said Wherein by comparing the partial image and a defect detecting section for detecting a defect in the inspection region.

  The invention according to claim 2 is the defect detection device according to claim 1, which is obtained by imaging a plurality of selected unit areas, each of which is a selected unit area on the substrate, by the imaging mechanism. In the plurality of selected images, a reference image generation unit that generates the reference image by obtaining representative values of pixel values corresponding to each other is further provided.

  The invention according to claim 3 is the defect detection apparatus according to claim 2, wherein the reference image generation unit calculates the representative value before obtaining the representative value when the plurality of selected images is 3 or more. A part of the pixel values corresponding to each other is excluded from the calculation, and the process of reducing the variation in the pixel values corresponding to each other is executed.

  The invention according to claim 4 is the defect detection apparatus according to claim 2 or 3, wherein each of the plurality of selection images indicates a plurality of band regions imaged by the imaging mechanism. It is a set of band area images, and the reference image is a set of a plurality of reference band area images corresponding to the plurality of band areas.

  The invention according to claim 5 is the defect detection apparatus according to claim 4, wherein a portion corresponding to the inspection area in the reference image straddles adjacent reference band area images, An image obtained by combining reference band region images adjacent to each other is treated as the partial image.

  A sixth aspect of the present invention is the defect detection apparatus according to any one of the first to fifth aspects, wherein the storage unit is a hard disk drive.

  A seventh aspect of the present invention is the defect detection device according to any one of the first to sixth aspects, wherein a storage device capable of communicating with the defect detection device or another defect detection device is connected to a communication network. The reference image is transmitted to the storage unit of the storage device or the other defect detection device via the communication network, or the storage device or the other of the other defect detection device is transmitted via the communication network. The apparatus further includes a communication unit that receives the reference image from the storage unit of the defect detection apparatus.

  The invention according to claim 8 is the defect detection apparatus according to any one of claims 1 to 7, wherein the display unit displays a reduced image obtained by reducing the reference image, and the inspection region corresponds to the inspection area. And a reception unit that receives an input of an area in the reduced image.

  The invention according to claim 9 is a defect detection method for detecting a defect of a pattern on a substrate, and a) preparing a substrate on which a predetermined unit pattern is formed in each of a plurality of unit regions on a main surface. And b) a step of preparing a multi-tone reference image that substantially shows the entire unit area and is used as a reference for defect detection; and c) from the reference image in a unit area that is a target for defect detection. Extracting a partial image including a portion corresponding to a region to be inspected; and d) having an imaging device, and moving the imaging device relative to the substrate in a movement direction along the main surface. By doing so, an image to be inspected that indicates the region to be inspected is acquired using an imaging mechanism that acquires a multi-tone image of a band region whose width is narrower than the width of the unit region in the direction perpendicular to the moving direction. E) the image to be inspected Wherein by comparing the partial image and a step of detecting a defect in the inspection region.

  A tenth aspect of the present invention is the defect detection method according to the ninth aspect, wherein f) a plurality of selected unit regions, each of which is a selected unit region on the substrate, are provided before the step b). The method further includes the step of generating the reference image by obtaining representative values of pixel values corresponding to each other in a plurality of selected images acquired by imaging with the imaging mechanism.

  The invention described in claim 11 is the defect detection method according to claim 10, wherein the step f) includes determining the representative values before obtaining the representative values when the plurality of selected images is 3 or more. A step of excluding a part of the corresponding pixel values to reduce the variation of the corresponding pixel values.

  The invention according to claim 12 is the defect detection method according to claim 10 or 11, wherein each of the plurality of selection images indicates a plurality of band regions imaged by the imaging mechanism. It is a set of band area images, and the reference image is a set of a plurality of reference band area images corresponding to the plurality of band areas.

  Invention of Claim 13 is the defect detection method of Claim 12, Comprising: When the part corresponding to the said to-be-inspected area | region in the said reference image straddles the mutually adjacent reference belt area | region image, In step c), an image obtained by combining the reference band area images adjacent to each other is treated as the partial image.

  A fifteenth aspect of the present invention is the defect detection method according to any one of the ninth to thirteenth aspects, wherein the unit area on the substrate is acquired by the imaging mechanism before the step d). A step of comparing a statistical value relating to a pixel value of an image showing a predetermined region in the image with a statistical value relating to a pixel value of a portion corresponding to the predetermined region in the reference image, and obtaining a comparison result; and d) step And a step of changing an imaging condition of the imaging mechanism based on the comparison result.

  A fifteenth aspect of the invention is the defect detection method according to any one of the ninth to fourteenth aspects, wherein the reduced image obtained by reducing the reference image is displayed on the display unit before the step c). And a step of receiving an input of a region in the reduced image corresponding to the region to be inspected.

  According to the present invention, in defect detection using a high-resolution imaging mechanism that obtains an image of a band area, a reference image that substantially shows the entire unit area and is used as a reference for defect detection is used. Defects in the region can be easily detected.

  In the inventions of claims 2 and 10, it is possible to accurately detect defects in the inspected region using an appropriate reference image. In the inventions of claims 3 and 11, a more appropriate reference image is generated. be able to.

  In the invention of claim 6, a large-capacity reference image can be stored, and in the invention of claim 7, the reference image can be shared with other devices.

  Further, in the inventions of claims 8 and 15, the inspected area can be easily specified, and in the invention of claim 14, defects are accurately detected by correcting the imaging condition in the imaging mechanism in accordance with the reference image. Can be detected.

  FIG. 1 is a diagram showing a configuration of a defect detection apparatus 1 according to the first embodiment of the present invention. The defect detection apparatus 1 is an apparatus for detecting a pattern defect on a semiconductor substrate (hereinafter referred to as “substrate”) 9 on which a fine pattern is formed.

  The defect detection apparatus 1 moves the substrate 9 on the stage 2 holding the substrate 9, the imaging unit 31 that captures the substrate 9 and acquires multi-gradation image data of the substrate 9, and the substrate 9 on the stage 2 along the main surface. The computer includes a stage moving mechanism 32 and a CPU that performs various arithmetic processes, a memory that stores various information, and the like. The computer 4 includes a receiving unit 41 that receives input from the operator using a keyboard and a mouse, and a display unit 42 that displays various types of information. In the computer 4, functions of an imaging control unit 43 that controls the imaging unit 31, a movement control unit 44 that controls the stage moving mechanism 32, and a calculation unit 45 that performs various calculations are realized by a CPU, a memory, and the like.

  The stage moving mechanism 32 includes an X direction moving mechanism 33 that moves the stage 2 in the X direction in FIG. 1 and a Y direction moving mechanism 34 that moves in the Y direction. In the X-direction moving mechanism 33, a ball screw (not shown) is connected to the motor 331, and the Y-direction moving mechanism 34 moves in the X direction along the guide rail 332 as the motor 331 rotates. The Y-direction moving mechanism 34 has the same configuration as the X-direction moving mechanism 33. When the motor 341 rotates, the stage 2 moves along the guide rail 342 in the Y direction by a ball screw (not shown).

  The imaging unit 31 electrically outputs an illumination unit 311 that emits illumination light, an optical system 312 that guides illumination light to the substrate 9 and is incident with light from the substrate 9, and an image of the substrate 9 formed by the optical system 312. A line sensor 313, which is an imaging device that converts signals, is provided.

  FIG. 2 is a plan view showing the substrate 9. The substrate 9 includes a plurality of unit regions (hereinafter referred to as “dies”) 91 that are diced in a subsequent process to become semiconductor chips on the main surface 90, and each of the plurality of dies 91 has a predetermined unit pattern. Is formed. 2, the illustration of the pattern formed on each die 91 is omitted for the sake of illustration (FIGS. 3, 8, 10, 11.A, 11.B and FIG. 14).

  FIG. 3 is an enlarged view of three dies on the substrate 9 (however, in FIG. 3, the three dies are denoted by reference numerals 91a, 91b, and 91c, respectively). In the defect detection apparatus 1, the stage 9 is moved in the direction perpendicular to the arrangement direction (X direction in FIG. 1) of the light receiving elements (eg, 2048 light receiving elements) of the line sensor 313 by the stage moving mechanism 32 shown in FIG. One band area 911 of a plurality of band areas (also referred to as swaths) 911 obtained by dividing the die 91 by performing image pickup by the line sensor 313 while continuously moving in the Y direction). The two-dimensional image data is acquired. That is, in the defect detection apparatus 1, the imaging unit 31 and the stage moving mechanism 32 operate as the imaging mechanism 3 in synchronization with each other, so that the line sensor 313 is moved with respect to the substrate 9 along the main surface 90 of the substrate 9. By moving relatively and continuously, it is possible to acquire a multi-tone image of the band region 911 whose width is narrower than the width of the die 91 in a direction perpendicular to the moving direction with high resolution. . In the present embodiment, one pixel in the image acquired by the imaging mechanism 3 corresponds to a rectangular area of 50 nanometer (nm) square on the substrate 9 (that is, the pixel resolution is 50 nm). ). In the defect detection of the semiconductor substrate 9, the pixel resolution in the imaging mechanism 3 is preferably 10 nm or more and 100 nm or less.

  The defect detection apparatus 1 of FIG. 1 further includes an inspection unit 51 that performs main processing in defect detection described later, a reference image generation unit 52 that generates a reference image that is a multi-tone image that is used as a reference in defect detection, and The storage unit 53 is a hard disk drive that stores reference image data (reference image data 531 in FIG. 1). The inspection unit 51 includes a partial image extraction unit 511 that extracts a partial image including a portion corresponding to a region to be inspected in the die 91 as a defect detection target from the reference image, and a defect detection unit connected to the imaging unit 31. 512. The defect detection unit 512 detects a defect in the inspection region by comparing the inspection image acquired by the imaging unit 31 and indicating the inspection region with the partial image input from the partial image extraction unit 511. A comparison unit 513 and a defect information memory 514 that temporarily stores information on defects detected by the comparison unit 513 are provided. In addition, two memories (first memory 521 and second memory 522) and the imaging unit 31 are connected to the reference image generation unit 52 (more precisely, the imaging unit 31 is connected to the reference image generation unit via the switching unit 50). 52.). In the present embodiment, the inspection unit 51 and the reference image generation unit 52 are formed by dedicated electric circuits, but these functions may be realized by the computer 4 as software.

  Next, an operation in which the defect detection apparatus 1 detects a pattern defect on the substrate 9 will be described. Here, first, a process of generating a reference image performed before the defect detection operation will be described with reference to FIG.

  When generating a reference image, first, a substrate for generating a reference image (for example, a circuit forming substrate after a predetermined process is performed) is placed on the stage 2 and prepared (step S21). ). In the present embodiment, it is assumed that the inspection target substrate 9 in FIG. 2 is also a reference image generation substrate. Subsequently, on the substrate 9, a plurality of dies arranged in the Y direction (three dies 91 a to 91 c with parallel oblique lines in FIG. 2) are selected as selection dies, and the imaging control unit 43 and the movement control unit 44 select the imaging unit. 31 and the stage moving mechanism 32 are respectively controlled, and in the plurality of selection dies 91a to 91c in FIG. 3, the (−X) -side band regions 911a, 911b, and 911c are changed to the arrows denoted by K1 in FIG. As shown, the images are sequentially taken (step S22).

  At this time, the image pickup unit 31 in FIG. 1 is connected to the reference image generation unit 52 side by the switching unit 50, and when picking up the selection die 91a, an image showing the band region 911a (hereinafter referred to as “selected band region image”). .) Is stored only in the first memory 521, and when the selection die 91 b is imaged, each pixel value of the selected band area image indicating the band area 911 b is stored only in the second memory 522. When the selection die 91c is imaged, the pixel values of the selected band area image indicating the band area 911c are input to the first memory 521, the second memory 522, and the reference image generation unit 52, respectively. Each pixel value of the selected band area image of the area 911c, the pixel value of the selected band area image of the band area 911a corresponding to this pixel value, and the pixel value of the selected band area image of the band area 911b are input to the reference image generation unit 52. Input (almost) at the same time. The selected band area image is, for example, an image having a size of about 1 gigabyte, and each of the first and second memories 521 and 522 has a storage capacity of 1 gigabyte.

  In the reference image generation unit 52, two pixel values that are close to each other are selected from the three pixel values of the selection band region images of the selection dies 91a to 91c that are input simultaneously (step S23). For example, if the three pixel values are 100, 102, and 150, respectively, the pixel values 100 and 102 are selected (ie, the pixel value 150 is excluded), and then the average value 101 of these values is 3 Calculated as a representative value of two pixel values. Thereby, one image (hereinafter referred to as “reference band area image”) having a size corresponding to one band area is generated from a plurality of selected band area images respectively indicating the plurality of band areas 911a to 911c. The band area image data is input and stored in the storage unit 53 (step S24).

  When the reference band area images corresponding to the band areas 911a to 911c are generated, the band on the (+ X) side of the band area 911c in the selection die 91c as shown by the arrow denoted by the symbol K2 in FIG. The imaging mechanism 3 sequentially captures the area 911, the (+ X) side band area 911 of the band area 911b in the selection die 91b, and the (+ X) side band area 911 of the band area 911a in the selection die 91a (step S25, S22) Among the three pixel values corresponding to each other in the selected band area image of these band areas 911, two pixel values that are close to each other are selected (step S23). Then, the next reference band area image is generated by calculating the average value of the selected pixel values, and the data of the reference band area image is stored in the storage unit 53 (step S24).

  By repeating the above steps S22 to S24 (step S25), a plurality of reference band area images respectively corresponding to the plurality of band areas 911 in the selection die 91a (or selection dies 91b, 91c) are acquired and stored in the storage unit 53. Is memorized. In the storage unit 53, a set of data of a plurality of reference band area images is handled as reference image data 531, and a reference image 61 that is a set of a plurality of reference band area images 611 is generated as shown in FIG. Actually, a large number (for example, 150 to 300) of band areas 911 are set in each of the selection dies 91a to 91c, and the reference image data 531 has a huge size (150 to 300 gigabytes).

  Here, if a set of all selection band area images is captured as a selection image in each selection die 91a to 91c, the reference image 61 is acquired by imaging the plurality of selection dies 91a to 91c with the imaging mechanism 3. It is generated by obtaining representative values of corresponding pixel values in a plurality of selected images. Therefore, the reference image 61 substantially shows the whole of one die 91 (however, a region to be removed during dicing may be removed) and shows a unit pattern with almost no defects. . Although only one reference image data 531 is shown in FIG. 1, a plurality of reference image data corresponding to a plurality of circuit formation stages is actually stored in the storage unit 53.

  Next, an operation in which the defect detection apparatus 1 detects a pattern defect on the substrate 9 will be described with reference to FIG. In the defect detection apparatus 1, first, a reference image 61, which is an image used as a reference for the substrate 9 to be inspected, is prepared (step S11). As described above, since the reference image 61 corresponding to the pattern on the substrate 9 is already stored in the storage unit 53 as the reference image data 531, the processing for preparing the reference image 61 in the present embodiment is stored. The reference image data 531 corresponding to the circuit formation stage of the substrate 9 is specified from the plurality of reference image data stored in the unit 53.

  When the reference image 61 is prepared, the computer 4 reads the reference image data 531 from the storage unit 53, and the calculation unit 45 extracts pixel values for each predetermined number of pixels in the reference image 61. 7, a reduced image 61a obtained by reducing the reference image 61 is generated and displayed on the display unit 42 (step S12). Subsequently, the operator inspects a die to be inspected (hereinafter referred to as “inspected die”) 91 and a region in the reduced image 61a corresponding to the region to be inspected (for example, a region to which parallel oblique lines are added in FIG. 7). An input specifying A1) is performed, and the input is received by the receiving unit 41, whereby the inspection area in the inspection die 91 is specified in the defect detection apparatus 1 (step S13). At this time, the operator can easily specify the region to be inspected by selecting the region A1 in the reduced image 61a by operating the mouse while referring to the reduced image 61a on the display unit 42, for example. Become.

  When the region to be inspected is specified, the partial image extraction unit 511 includes an image (that is, a portion including a portion corresponding to the region to be inspected (region A2 indicated by parallel diagonal lines in FIG. 5) in the reference image 61 in FIG. 5. Image) is extracted (step S14). Here, as described above, the reference image 61 is stored in the storage unit 53 as a set of a plurality of reference band region images 611 and corresponds to the region to be inspected in the reference image 61 as shown in FIG. When the portion A2 extends over adjacent reference band area images (two reference band area images denoted by reference numerals 611a and 611b in FIG. 5), after these reference band area images 611a and 611b are extracted Are combined and stored in a memory (for example, a memory having a storage capacity of 2 gigabytes) included in the partial image extraction unit 511. That is, the partial image extraction unit 511 treats an image obtained by combining these reference band area images 611a and 611b as a partial image.

  When the partial image is extracted, the substrate 9 to be inspected is placed on the stage 2 and prepared (step S15). Here, since the substrate 9 for generating the reference image and the substrate 9 to be inspected are the same, the process of step S15 has already been completed in step S21 of FIG. Subsequently, the imaging unit 31 moves relative to the substrate 9 near the die to be inspected (the die denoted by reference numeral 91d in FIG. 2) on the substrate 9 in FIG. When the imaging unit 31 passes over the band region 911d including the inspection region (region A3 with parallel diagonal lines in FIG. 8) of the die 91d, an inspection image indicating the inspection region A3 is acquired (step) S16). In FIG. 8, a region corresponding to the reference band region image is indicated by a two-dot chain line rectangle.

  At this time, the imaging unit 31 in FIG. 1 is connected to the defect detection unit 512 side by the switching unit 50, and the pixel value of the inspection image is transferred to the partial image extraction unit 511 and the comparison unit 513 of the defect detection unit 512. Output sequentially. In the partial image extraction unit 511, the pixel value in the partial image corresponding to the input pixel value is output to the comparison unit 513, and each pixel value of the inspected image and the corresponding pixel value of the partial image are output by the comparison unit 513. The absolute value (difference absolute value) of the difference is calculated. When the difference absolute value is larger than a predetermined threshold value, the position in the inspection area A3 corresponding to this pixel value is regarded as a defect element, and a value indicating the defect is output to the defect information memory 514, and the difference absolute If the value is less than or equal to the threshold value, the position in the inspected area A3 corresponding to this pixel value is considered normal, and a value indicating normal (non-defect) is output to the defect information memory 514. Thereby, a defect image indicating a defect in the inspection area A3 (that is, a set of adjacent defect elements) is generated, and a defect of the pattern in the inspection area A3 on the substrate 9 is detected (step S17). ). The result of the defect detection is output to the computer 4 and stored in a storage unit included in the computer 4 and displayed on the display unit 42 as necessary. When no defect exists in the inspection area A3, a defect image having a value indicating that all the pixels are non-defective is acquired. In the following description, the defect detection of the pattern on the substrate 9 includes a case where no defect is detected (that is, a defect image having a value indicating that all the pixels are non-defective is generated).

  Actually, in the defect detection apparatus 1, a plurality of dies to be inspected 91 are selected (step S18), and inspected images are similarly acquired for the other dies to be inspected 91 (step S16). And a partial image are compared to detect a defect in the inspection area (step S17). In addition, the processes of steps S15 to S18 are repeated for other substrates 9 (for example, substrates in the same circuit formation stage as the substrates 9 included in the same lot or different lots) as necessary (step S19). . Thereby, the defect detection of the pattern on a some board | substrate will be performed in a short time.

  Here, the defect detection apparatus 92 of the comparative example shown in FIG. 9 will be described. The defect detection device 92 in FIG. 9 omits the storage unit 53 and the reference image generation unit 52 in the defect detection device 1 in FIG. 1, and stores a partial image extraction unit 511 as one band area image (one swath image). A possible image memory 921 has been replaced. In the defect detection apparatus 92 of the comparative example, when inspecting the inspection die 93 shown in FIG. 10 (shown with parallel oblique lines in FIG. 10), the reference numeral 94 is shown in FIG. Detecting a defect while imaging one band area of the reference die and storing the band area image in the image memory 921 and sequentially acquiring the band area images of the corresponding band areas of the plurality of dies 93 on the substrate 95 By comparing with the band area image of the reference die 94 in the part 512, the defect of each inspection die 93 is detected (so-called Die-to-AnyDie type defect detection). However, in the Die-to-AnyDie type defect detection, if an accidental defect exists in the reference die itself, the corresponding part in the inspection die is detected as a defect even though it is not a defect. (A false alarm occurs.) Further, when inspecting a plurality of substrates in the defect detection of the Die-to-AnyDie method, an operation of imaging a reference die and storing it in the image memory 921 is performed on each substrate.

  On the other hand, in the defect detection apparatus 1 in FIG. 1, representative values of pixel values corresponding to each other are obtained in a plurality of selection images acquired by imaging a plurality of selection dies for each band region by the imaging mechanism 3. Thus, an appropriate reference image (having almost no defects) is generated and stored in the storage unit 53. Then, a partial image including a portion corresponding to the inspection region is extracted from the reference image, an inspection image indicating the inspection region is acquired by the imaging mechanism 3, and the partial image and the inspection image are converted into the defect detection unit 512. Compared. In this way, in the defect detection apparatus 1 having the high-resolution imaging mechanism 3 that acquires the image of the band region, the so-called golden template type defect detection is performed, so that defects in the inspection region can be accurately and easily performed. Can be detected.

  Further, even when a plurality of substrates 9 are inspected, the defect detection apparatus 1 uses the same reference image repeatedly for defect detection of the plurality of substrates 9 (Die-to-Any Wafer type defect detection is performed. Compared with Die-to-AnyDie type defect detection, the inspection of a plurality of substrates 9 can be performed in a short time, and the plurality of substrates 9 included in a plurality of lots can be absolutely detected. Evaluation is possible, and the degree of variation in the quality of the substrate 9 between lots can be recognized from the viewpoint of defect detection.

  Further, in another operation example (a so-called Die-to-Die type defect detection example) in the defect detection apparatus 92 of the comparative example of FIG. 9, first, the band region of the first reference die 94a on the substrate 95 of FIG. And the band area image is stored in the image memory 921. Subsequently, the corresponding band area of the inspection die 93a is imaged, and each pixel value of the band area image is input to the defect detection unit 512 and the image memory 921. Is done. Thereby, each pixel value of the band area image of the inspection die 93a and the corresponding pixel value of the first reference die 94a are input to the defect detection unit 512, and the first comparison result is acquired, and the image memory 921 is obtained. Stores the band area image of the die 93a to be inspected. Subsequently, by imaging the band region of the second reference die 94b, a second comparison result between the band region of the die to be inspected 93a and the corresponding band region of the second reference die 94b is obtained, and the first comparison result and A defect of the inspection die 93a is detected based on the second comparison result.

  On the other hand, FIG. As shown in A, in the exposure process (or other process, of course) at the time of pattern formation, predetermined process parameters (for example, focus position and exposure time) are gradually changed according to the position of the die 96. However, a substrate for process evaluation may be formed by forming a pattern on each die 96, and such a substrate for process evaluation is used when obtaining a so-called process margin that is an allowable range of parameters of an exposure process. Used for. In this case, when defect detection is performed by the Die-to-Die method, FIG. As shown in FIG. A die formed by changing the process parameter shown in A (a plurality of dies surrounded by a broken-line rectangle in FIG. 11.B, hereinafter referred to as “evaluation die 96”) under standard conditions. A substrate sandwiched between the formed dies (a die with parallel diagonal lines in FIG. 11.B, hereinafter referred to as “standard die 96a”) is prepared, and a certain evaluation die 96 is arranged in the Y direction. The defect detection of the inspection area in the evaluation die 96 is performed using the two adjacent standard dies 96a as reference dies. In this case, the number of evaluation dies 96 that can be formed on one substrate is smaller than the number of dies on the substrate, and depending on the number of process parameter changes, it is necessary to produce a plurality of substrates for process evaluation. . Further, since the plurality of evaluation dies 96 are compared with different reference dies, the evaluation results may vary.

  On the other hand, in the defect detection apparatus 1 of FIG. 1, a reference image used as a reference for defect detection is stored in the storage unit 53, and only by comparing a partial image derived from the reference image with an image to be inspected. Defects in the inspection area are easily detected. Furthermore, since it is not necessary to form a standard die on the substrate for process evaluation (that is, only the evaluation die needs to be formed), the substrate for process evaluation can be efficiently manufactured. Even when the number of changes becomes enormous, the number of substrates for process evaluation can be reduced. In addition, since a plurality of evaluation dies are compared with the same reference image, an allowable range of process parameters can be obtained with high accuracy.

  FIG. 12 is a diagram illustrating a defect detection system 80 having a plurality of defect detection apparatuses 1. In the defect detection system 80 of FIG. 12, each of the plurality of defect detection devices 1 is connected to the communication network 8 via the communication unit 46, and the communication network 8 further includes a storage device 81 that can communicate with the defect detection device 1. Connected (for example, SAN (Storage Area Network) is constructed). The communication unit 46 of each defect detection device 1 transmits the reference image to the storage unit 53 or the storage device 81 of another defect detection device 1 via the communication network 8, and the storage unit 53 of the other defect detection device 1. Alternatively, the reference image can be received from the storage device 81. Thereby, the reference image generated by one defect detection device 1 is directly transmitted to the other defect detection device 1, or the reference image is stored in the storage device 81, and then the other defect detection device. 1 can be transmitted indirectly, and the reference image can be shared among the plurality of defect detection devices 1 and the storage device 81.

  Next, an example of a defect detection operation in one defect detection apparatus 1 in FIG. 12 (hereinafter referred to as “target defect detection apparatus 1”) will be described. FIG. 13 is a diagram showing a part of the flow of the operation in which the target defect detection apparatus 1 detects a defect on the substrate 9, and shows the operation performed between step S15 and step S16 in FIG. Yes.

  In this operation example, in step S11 of FIG. 6, the pattern on the substrate 9 to be inspected is stored in the communication unit 46 of the target defect detection device 1 from the storage unit 53 or the storage device 81 of the other defect detection device 1. A reference image is received and stored in the storage unit 53. When the reference image is prepared in the target defect detection apparatus 1, subsequently, processing related to the specification of the inspection area (steps S12 and S13) and extraction of the partial image (step S14) are performed, and then on the stage 2 The substrate 9 to be inspected is placed and prepared (step S15).

  When the substrate 9 is prepared, an area in a predetermined die 91 on the substrate 9 (hereinafter referred to as “evaluation area”) is imaged by the imaging mechanism 3, and an image showing the evaluation area in the computing unit 45 of the computer 4. The average value of the pixel values is acquired (FIG. 13: Step S31). Subsequently, a portion corresponding to the evaluation area in the reference image is input to the calculation unit 45, and an average value of the pixel values is calculated (step S32). Then, a ratio between the average value of the pixel values in the evaluation region on the substrate 9 and the average value of the pixel values of the portion corresponding to the evaluation region of the reference image is acquired as a comparison result (step S33).

  In the computing unit 45, the imaging condition in the imaging mechanism 3 is further changed based on the comparison result (step S34). For example, when the comparison result indicates that the average value of the pixel values in the evaluation region is smaller than the average value of the pixel values in the reference image, the intensity of light from the illumination unit 311 in the imaging unit 31 is the comparison result. The value of the setting parameter for the illumination unit 311 in the imaging control unit 43 is changed so as to increase by the corresponding amount. In this way, the imaging conditions in the imaging mechanism 3 are corrected according to the reference image, and the brightness of the acquired image is equivalent to that of the reference image. Then, the image to be inspected is compared with the partial image while acquiring the image to be inspected by the imaging mechanism 3 (FIG. 6: Steps S16 and S17), and a defect in the inspected region of the substrate 9 is detected. If necessary, defect detection is repeated for another die on the substrate 9 or another substrate 9 (steps S18 and S19).

  As described above, in the defect detection system 80 having the plurality of defect detection devices 1, the plurality of defect detection devices 1 and the storage device 81 that can communicate with the defect detection device 1 are connected to the communication network 8. The reference image is shared among the defect detection apparatuses 1. Thereby, when a reference image necessary for a substrate 9 to be inspected (a substrate 9 in a certain circuit formation stage) is already acquired in another defect detection device 1 in another defect detection device 1. Defect detection can be performed without newly generating a reference image. Moreover, in each defect detection apparatus 1, when using the reference image acquired by the other defect detection apparatus 1, the imaging condition in the imaging mechanism 3 is corrected according to the reference image. As a result, defects on the substrate 9 can be detected with high accuracy and efficiency. In the above operation for matching the imaging condition to the reference image, other statistical values related to the pixel value of the image such as the median value of the pixel values may be used in addition to the average value of the pixel values of the image.

  In the defect detection system 80, the storage unit 53 in each defect detection device 1 is omitted, and the storage device 81 is regarded as a storage unit for each defect detection device 1, so that one storage unit in the plurality of defect detection devices 1 is obtained. (Storage device 81) may be shared (in this case, the operation of copying the reference image is not performed).

  Next, the defect detection apparatus 1 according to the second embodiment will be described. In the present embodiment, the defect detection device 1 in FIG. 1 or the defect detection device 1 in FIG. 12 is used, and the defect detection target is a display device such as a liquid crystal display device (for example, a display device mounted in an automobile). The glass substrate used in the above. As shown in FIG. 14, on the main surface 90a of the glass substrate 9a, a plurality of unit regions 91A each corresponding to the panel of the display device and having a uniform pattern are arranged in a vertical and horizontal manner with a gap. The substrate 9a is to be multi-faced according to the plurality of unit regions 91A (that is, to be cut into a plurality of portions (panels)). Moreover, in this Embodiment, the pixel resolution in the imaging mechanism 3 shall be 10 micrometers (micrometer) or less, Preferably, you may be 1-5 micrometers.

  The defect detection operation in the second embodiment is the same as that in the first embodiment. First, a reference image, which is an image used as a reference in defect detection, is prepared (FIG. 6: Step S11). Is displayed on the display unit 42 (step S12). Subsequently, an input for specifying a unit region to be inspected (hereinafter referred to as “unit region to be inspected”) and a region in the reduced image corresponding to the region to be inspected is performed (step S13). A partial image including a portion corresponding to the inspection area is extracted (step S14). When the partial image is extracted, the substrate 9a to be inspected is placed and prepared on the stage 2 (step S15), and the imaging unit 31 passes over the band region including the region to be inspected in the unit region to be inspected. Thus, an inspected image indicating the inspected area is acquired (step S16). Then, the defect detection unit 512 detects the defect in the inspection area by comparing the inspection image and the partial image (step S17). Similarly to the first embodiment, the defect detection is repeated for another unit region 91A on the substrate 9a or another substrate as necessary (steps S18 and S19).

  As described above, in the imaging mechanism 3 of the defect detection apparatus 1 according to the present embodiment, the line sensor 313 is relatively and continuously relative to the substrate 9a in the moving direction along the main surface 90a of the glass substrate 9a. By moving, the image of the band area in the unit area 91A on the substrate 9a can be acquired with high resolution. In addition, a reference image that substantially shows the entire unit area and is used as a reference for defect detection is stored in the storage unit 53, and a part that includes a portion corresponding to the inspection area from the reference image when detecting a defect. While the image is extracted, the inspection image indicating the inspection region is acquired by the imaging mechanism 3, and the defect detection unit 512 compares the partial image and the inspection image. Thereby, in the defect detection apparatus 1 having the high-resolution imaging mechanism 3 that acquires the image of the belt region, it is realized that the defect in the inspection region on the glass substrate 9a is easily detected using the reference image. The

  Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications can be made.

  In the first embodiment, when the reference image is generated, the three selection dies 91a to 91c are selected, and three selection images (a set of all selection band area images in each selection die) are acquired. In step S23 in FIG. 4, by selecting two pixel values that are close to each other among the corresponding pixel values of the three selected images and excluding other pixel values as error values, the reference image having almost no defects is selected. However, when three or more selection dies are selected and the number of selected images is three or more, for example, the maximum or / and the minimum among a plurality of pixel values corresponding to each other of these selected images is obtained. Error values may be excluded by excluding things. That is, the reference image generation unit 52 excludes some of these pixel values from the calculation and calculates the pixel values before obtaining the representative values of the corresponding pixel values when the plurality of selected images is 3 or more. Various processes for reducing the variation may be performed, which makes it possible to generate a more appropriate reference image.

  In addition, a reference image used as a reference for defect detection may be generated from design data (for example, CAD (Computer Aided Design) data) if it substantially shows the entire unit area on the substrate. Is possible. In this case, correction in consideration of the shape change of the pattern (elements) generated depending on the conditions of the manufacturing process (for example, correction of the pattern shape after etching that changes depending on the conditions in the photolithography process (etching correction)) A multi-tone reference image is generated while performing the above. Of course, the reference image derived from the plurality of selection images acquired by the imaging mechanism 3 may be corrected by the operator, and the corrected reference image may be stored in the storage unit 53 (or the storage device 81). .

  The defect detection unit 512 detects a defect in the inspected area by obtaining a difference absolute value of pixel values corresponding to each other between the inspected image and the partial image. For example, the defect absolute value of the pixel value is normalized. Then, the defect in the inspected area may be detected by other methods such as comparing with a predetermined threshold value. In other words, the defect detection unit 512 may use any technique as long as a defect in the inspection area is detected by comparing the inspection image and the partial image.

  In the defect detection system 80, only a plurality of defect detection devices 1 may be connected to the communication network 8. Further, only the storage device 81 that stores the reference image in advance and the single defect detection device 1 are included in the communication network 8. It may be connected to. That is, since the storage device 81 that can communicate with the target defect detection device 1 or another defect detection device 1 is connected to the communication network 8, the target defect detection device 1 uses a reference to another device. Images can be shared.

  The holding unit that holds the substrate is not necessarily the stage 2 on which the substrate is placed. For example, the holding unit may hold the substrate by gripping the outer edge of the substrate. Moreover, the imaging mechanism 3 acquires the image of a belt area | region by moving the line sensor 313 which is an imaging device of the imaging part 31 with respect to the fixed board | substrate continuously in the moving direction along a main surface. There may be. Furthermore, the imaging part 31 may acquire an image using an electron beam, an ultraviolet-ray, an X-ray etc., for example. When an electron beam is used, it is not easy to obtain an image with a certain quality due to a change in electron dose with time or a change in the charged state of a substrate between lots. Even when inspecting multiple substrates by correcting the imaging conditions according to the reference image, it is possible to acquire multiple inspected images with high reproducibility at the image quality that matches the reference image and to detect defects on the substrate. Can be detected with high accuracy. In addition, the process of correcting the imaging condition in accordance with the reference image is performed after the maintenance of the imaging mechanism 3 is performed regardless of whether the defect detection apparatus 1 is connected to the communication network 8 or periodically. It may be performed as calibration or the like.

  In the defect detection apparatus 1, a large-capacity reference image can be stored by using the storage unit 53 as a hard disk drive, but the storage unit 53 is a combination of a plurality of hard disk drive devices (for example, RAID ( Redundant Arrays of Inexpensive Disks) may be used. In this case, the data transfer rate in the storage unit 53 can be further improved.

  The operation flow shown in FIG. 6 may be appropriately changed within a possible range. For example, the reference image may be prepared after the substrate to be inspected is prepared.

  The substrate is not necessarily a semiconductor substrate or a glass substrate for a display device, but is a printed wiring board, a lead frame, a photomask or the like in which a unit pattern is formed in each of a plurality of unit regions on the main surface. May be.

It is a figure which shows the structure of a defect detection apparatus. It is a top view which shows a board | substrate. It is a figure which expands and shows a plurality of dies. It is a figure which shows the flow of the process which produces | generates a reference | standard image. It is a figure which shows a reference | standard image. It is a figure which shows the flow of the operation | movement which detects the defect of the pattern on a board | substrate. It is a figure which shows a reduced image. It is a figure which shows a to-be-inspected die. It is a figure which shows a part of defect detection apparatus of a comparative example. It is a figure for demonstrating the defect detection operation | movement in the defect detection apparatus of a comparative example. It is a figure for demonstrating the board | substrate for process evaluation. It is a figure which shows the board | substrate for process evaluation with respect to the defect detection apparatus of a comparative example. It is a figure which shows a defect detection system. It is a figure which shows a part of flow of operation | movement which detects the defect on a board | substrate. It is a top view which shows the board | substrate of glass.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect detection apparatus 2 Stage 3 Imaging mechanism 8 Communication network 9, 9a Substrate 41 Reception part 42 Display part 46 Communication part 52 Reference image generation part 53 Storage part 61 Reference image 61a Reduced image 81 Storage apparatus 90, 90a Main surface 91 Die 91A Unit area 91a to 91c Selection die 91d Die to be inspected 313 Line sensor 511 Partial image extraction unit 512 Defect detection unit 611, 611a, 611b Reference band area image 911, 911a to 911d Band area A3 Inspected area S11 to S17, S22 to S25 , S31-S34 steps

Claims (15)

A defect detection device for detecting a pattern defect on a substrate,
A holding unit for holding a substrate on which a predetermined unit pattern is formed in each of a plurality of unit regions on the main surface;
By having an imaging device and moving the imaging device relatively and continuously with respect to the substrate in the movement direction along the main surface, the width is greater than the width of the unit region in the direction perpendicular to the movement direction. An imaging mechanism for acquiring a multi-tone image of a narrow band region;
A storage unit for storing a multi-tone reference image that substantially indicates the entire unit area and is used as a reference for defect detection;
A partial image extraction unit that extracts a partial image including a portion corresponding to a region to be inspected in a unit region to be subjected to defect detection from the reference image;
A defect detection unit that detects a defect in the inspection region by comparing the partial image with the inspection image obtained by the imaging mechanism and indicating the inspection region;
A defect detection apparatus comprising: The defect detection apparatus according to claim 1,
In the plurality of selected images acquired by imaging the plurality of selected unit areas, each of which is a selected unit area on the substrate, by the imaging mechanism, the reference is obtained by obtaining representative values of pixel values corresponding to each other. A defect detection apparatus further comprising a reference image generation unit that generates an image. The defect detection apparatus according to claim 2,
When the plurality of selected images is 3 or more, the reference image generation unit excludes a part of the corresponding pixel values from the calculation before obtaining the representative value, and sets the corresponding pixel values. A defect detection apparatus that performs processing for reducing variation. The defect detection apparatus according to claim 2, wherein
Each of the plurality of selection images is a set of a plurality of selection band region images each indicating a plurality of band regions imaged by the imaging mechanism, and the reference image corresponds to the plurality of reference regions corresponding to the plurality of band regions. A defect detection apparatus characterized by being a set of band area images. The defect detection apparatus according to claim 4,
When a portion corresponding to the region to be inspected in the reference image straddles a reference band region image adjacent to each other, an image obtained by combining the reference band region images adjacent to each other is treated as the partial image. Defect detection device. The defect detection apparatus according to any one of claims 1 to 5,
The defect detection device, wherein the storage unit is a hard disk drive. The defect detection apparatus according to any one of claims 1 to 6,
A storage device communicable with the defect detection device, or another defect detection device is connected to a communication network, and the storage device or the storage unit of the other defect detection device via the communication network A defect detection device further comprising: a communication unit that transmits a reference image or receives the reference image from the storage unit of the storage device or the other defect detection device via the communication network. . The defect detection apparatus according to any one of claims 1 to 7,
A display unit for displaying a reduced image obtained by reducing the reference image;
A receiving unit for receiving an input of an area in the reduced image corresponding to the inspection area;
A defect detection apparatus further comprising: A defect detection method for detecting defects in a pattern on a substrate,
a) preparing a substrate on which a predetermined unit pattern is formed in each of a plurality of unit regions on the main surface;
b) preparing a multi-tone reference image that substantially shows the entire unit area and serves as a reference for defect detection;
c) extracting from the reference image a partial image including a portion corresponding to a region to be inspected in a unit region to be subjected to defect detection;
d) having an imaging device and moving the imaging device relative to the substrate in a movement direction along the main surface relative to the substrate so that the width is the width of the unit region with respect to the direction perpendicular to the movement direction; Using an imaging mechanism for acquiring a multi-tone image of a narrower band region, obtaining an inspection image indicating the inspection region; and
e) comparing the inspected image with the partial image to detect defects in the inspected area;
A defect detection method comprising: The defect detection method according to claim 9,
f) Pixels corresponding to each other in a plurality of selected images acquired by imaging a plurality of selected unit areas, each of which is a selected unit area on the substrate, by the imaging mechanism before the step b) A defect detection method further comprising a step of generating the reference image by obtaining a representative value. The defect detection method according to claim 10,
In the step f), when the plurality of selected images is 3 or more, a part of the corresponding pixel values is excluded before obtaining the representative value, thereby reducing variations in the corresponding pixel values. A defect detection method comprising the step of: The defect detection method according to claim 10 or 11,
Each of the plurality of selection images is a set of a plurality of selection band region images each indicating a plurality of band regions imaged by the imaging mechanism, and the reference image corresponds to the plurality of reference regions corresponding to the plurality of band regions. A defect detection method characterized by being a set of band area images. The defect detection method according to claim 12,
When the portion corresponding to the region to be inspected in the reference image straddles adjacent reference band region images, an image obtained by combining the adjacent reference band region images in the step c) is used as the partial image. A defect detection method characterized by being handled. A defect detection method according to any one of claims 9 to 13,
Prior to step d), a statistical value relating to a pixel value of an image obtained by the imaging mechanism and indicating a predetermined area in the unit area on the substrate, and a portion corresponding to the predetermined area in the reference image Comparing a statistical value related to the pixel value of, and obtaining a comparison result;
A step of changing an imaging condition of the imaging mechanism in the step d) based on the comparison result;
A defect detection method further comprising: The defect detection method according to any one of claims 9 to 14,
A step of displaying a reduced image obtained by reducing the reference image on the display unit before the step c);
Receiving an input of an area in the reduced image corresponding to the inspection area;
A defect detection method further comprising:

Images