JP2010019561A - Flaw inspection device and flaw inspection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a flaw inspection device for precisely determining whether a detected flaw is a false flaw. <P>SOLUTION: The flaw inspection device is equipped with: an image comparing inspection part 1 for inspecting a flaw by comparing the image of an inspection target; a false flaw image memory part 2 for storing a preliminarily registered false flaw image; and a false flaw determining part 3 for comparing the flaw detected by the image comparing inspection part with the false flaw image stored by the false flaw image memory part and judging whether the flaw is the false flaw. Since the preliminarily registered false flaw is compared with the image to determine whether the image is the false flaw, it is precisely determined whether the flaw detected in the image comparing inspection part is the false flaw which must not originally determined as a flaw. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an appearance inspection apparatus and method for inspecting the appearance of an object to be inspected to detect defects. In particular, the present invention relates to a defect inspection apparatus and a defect inspection method for performing inspection by automatically removing from a defect candidate a pseudo defect in which a pattern that is not originally detected as a defect is detected as a defect in an appearance inspection of a semiconductor device.

  Conventionally, in appearance inspection, the appearances of objects to be inspected are compared, and a pattern defect is detected based on a difference seen between patterns that should have the same shape. In particular, in a semiconductor integrated circuit, a pattern is periodically formed on a semiconductor wafer in units of chips or shots of a stepper, and patterns are formed periodically in units of cells like a memory cell portion. Appearance inspection is performed by comparing various patterns to inspect for the presence or absence of defects. Specifically, the image captured by the image sensor is digitally processed, the difference between the corresponding pixel intensities is taken, and the inspection is performed by a procedure in which a portion where the absolute value of the difference exceeds a certain threshold is determined to be abnormal. Many. If a part where there is a difference between two patterns that are supposed to be the same (designed in the same shape) is called a defect, an image containing a defect and an area where no difference between the pattern and its surroundings (called a real image) And an image (referred to as a difference image) configured by taking a difference between pixels can be recorded. If the threshold value is reduced in the pixel intensity difference, a very large number of defects are detected, and if the threshold value is increased, the number of detected defects becomes very small. This threshold is appropriately determined according to the target and purpose. In most cases, the defect detected by the inspection apparatus under the threshold condition set according to the purpose is different from the defect detected from the viewpoint of semiconductor manufacturing. If defects detected by the inspection apparatus are called defect candidates, since there are usually a large number of defect candidates, it is determined whether or not they are defects using an automatic inspection machine such as a computer. In practice, it is very important to distinguish from these defect candidates true defects that may cause problems in semiconductor manufacturing and so-called pseudo defects that do not cause problems in semiconductor manufacturing. Further, by classifying and organizing defects, feedback to the manufacturing process is also performed. From this point of view, the following prior art is disclosed.

  Patent Document 1 describes an image comparison type defect inspection apparatus that inspects a defect by comparing image signals obtained by an image sensor in units of repetitive portions in which the same pattern such as a semiconductor chip is formed. Has been. Furthermore, the defect candidate includes a pseudo defect due to color unevenness that is not originally a defect, and it is described that the pseudo defect due to color unevenness is reduced by emphasizing a partial signal that easily causes a signal difference such as a pattern edge. Has been.

  Further, in Patent Document 2, the classification by the area for each pattern density such as the memory core part and the outer peripheral part, and the classification of the category according to the type of defect such as a defect due to foreign matter, a circuit pattern defect, or a defect due to scratches, An inspection data processing method and apparatus for enabling the selection of a fatal defect accurately and accurately are described. In addition, it is described that the defect area, the defect brightness, the defect length, and the like are numerically determined in order to determine whether the defect is fatal.

  Further, Patent Document 3 discloses an inspection data processing program for classifying defects that may occur in a post-inspection processing step by using a defect inspection result and circuit pattern design data processed in the post-inspection step. Is described.

Furthermore, in Patent Document 4, data is superimposed (die stack) for each die (chip) from information obtained by defect detection, a pseudo defect is identified from the superimposed data, and a pseudo defect is detected. It is described that a defect inspection is performed with high inspection sensitivity by setting a non-inspection area as a non-inspection area and excluding a pseudo defect from an inspection target.
JP 2002-303588 A JP 2001-305073 A JP 2005-150409 A JP 2006-126020 A

  The cause of the pseudo defect is not limited to the color unevenness described in Patent Document 1, and it is difficult to appropriately determine whether or not the defect is a pseudo defect even by the prior art document. Classification of defects as described in Patent Documents 2 and 3 is also based on the premise of determining and removing pseudo defects, and is true unless it can be appropriately determined whether or not they are pseudo defects. It is also impossible to correctly classify the defects. As described below, as in Patent Document 2, it is difficult to determine whether or not a defect is a pseudo defect simply by quantifying the area of the defect, the brightness of the defect, the length of the defect, and the like.

  According to the inventors' consideration, the reason why it is difficult to determine whether or not the defect is a pseudo defect is as follows. With the progress of miniaturization of semiconductors in recent years, in the development of new processes, test patterns incorporating several small transistors and wires are formed, and characters that can be visually recognized by an optical microscope are formed for convenience of analysis. Sometimes. In such a case, since the character pattern is not a functional element, there is no problem in evaluating the electrical characteristics even if the pattern is not completely formed in the process. That is, a semiconductor wafer has a defect that becomes a defective product if the pattern is not formed correctly, and a defect that does not become a defective product even if the pattern is not formed correctly (referred to in the present specification as a pseudo defect). .

  However, in the actual inspection, when the number of defects that affect the electrical characteristics of the semiconductor is detected and the number of defects that do not have an electrical effect is large, the purpose is to evaluate the distribution and number transition. It becomes noise and the original inspection purpose cannot be achieved.

  In addition, when a difference image with a surrounding pattern is used for detecting a defect, if the difference between the patterns itself is very small, the difference is only about one pixel. This difference may or may not be a random defect. In such a case, information that can be discarded cannot be distinguished from information that should be left behind, and it does not function for the purpose of removing unnecessary defects. Furthermore, since the inspection object has a pattern close to the resolution limit, not all patterns become abnormal, and a pseudo defect may be detected only by a single die. As described in Patent Document 4 described above, such a defect is obtained by superimposing data (die stack) for each die (chip) and identifying a pseudo defect from the superimposed data. It cannot be removed.

  In addition, as in Patent Document 4, if a place where a pseudo defect is detected is set as a non-inspection area and removed from the inspection target, a true defect is found at the same place as the place where the pseudo defect is detected. If there is, it will be overlooked. As described above, there is a demand for a defect inspection apparatus that appropriately determines whether or not a pseudo defect.

  The defect inspection apparatus according to one aspect (side surface) of the present invention compares image data of a plurality of objects to be inspected, and inspects the presence or absence of defects based on the difference image based on the presence or absence of identity. A pseudo defect image storage unit that stores a pre-registered pseudo defect image (actual image including the periphery of the defect), and the defect detected by the image comparison inspection unit and its peripheral region, the pseudo defect image storage unit And a pseudo defect determination unit that determines whether or not the defect is a pseudo defect.

  In the present invention, the “pseudo defect” refers to a defect other than the defect to be detected in the defect inspection among the defects detected in the image comparison inspection. For example, in the defect inspection, when only a specific defect is to be detected, all the defects other than the specific defect may be handled as pseudo defects.

  The defect inspection method according to another aspect of the present invention includes a step of comparing image data of a plurality of objects to be inspected to inspect for the presence or absence of defects, and a pre-registered pseudo Comparing the defect image with the defect inspected in the step of inspecting the presence or absence of the defect, and determining whether or not the defect is a pseudo defect.

  Furthermore, a defect inspection program according to another aspect of the present invention is for causing a computer to execute the defect inspection method.

  According to the present invention, a pseudo defect image is registered in advance, and it is determined whether or not the detected defect is a pseudo defect by comparing with the registered pseudo defect image. it can.

  Embodiments of the present invention will be described with reference to the drawings. The defect inspection apparatus according to an embodiment of the present invention (for example, the entire defect inspection apparatus of FIG. 1 or FIG. 7) receives image data of a plurality of objects to be inspected (for example, image data in the inspected image data storage unit 5). An image comparison inspection unit (for example, image comparison inspection unit 1) that compares and inspects for the presence or absence of defects by comparison, and a pseudo defect image storage unit (for example, pseudo defect image storage) that stores a pre-registered pseudo defect image 2) and a pseudo defect determination unit that compares the defect detected by the image comparison inspection unit with a pseudo defect image stored in the pseudo defect image storage unit and determines whether or not the defect is a pseudo defect. (For example, a pseudo defect determination unit 3).

  Further, the defect inspection method according to the embodiment of the present invention (for example, the defect inspection step S2 in FIG. 2, details are the defect inspection method in FIG. 3, FIG. 4, FIG. 6, or FIG. 8) includes a plurality of inspected objects. It is inspected in a step of comparing the image data of the target object and inspecting for the presence or absence of defects based on the presence or absence of identity (for example, image comparison inspection step S13), and a step of inspecting the presence or absence of the above-described pseudo defect image A step (for example, step S25) of comparing the defect and determining whether the defect is a pseudo defect.

  Hereinafter, it will be described in detail with reference to the drawings in accordance with embodiments.

  FIG. 1 is a diagram illustrating the configuration of the defect inspection apparatus according to the first embodiment of the present invention. First, this configuration will be described. The image recognition unit 4 recognizes an image including an object to be inspected by an image sensor, and stores the digitized image data in the inspected image data storage unit 5. The image comparison / inspection unit 1 recognizes regularly arranged images from the inspected image data stored in the inspected image data storage unit 5, and other parts of the regularly arranged images are recognized. When compared with the image, it is recognized based on the difference image whether there is an image whose regularity is impaired. The image data whose regularity is impaired is stored in the defect candidate data storage unit 6 as the defect data A. As will be described later, the defect data A includes pseudo defects that do not need to be handled as defects. The pseudo defect determination unit 3 includes a pseudo defect registration image in which a pseudo defect is not included in the defect candidates stored in the defect candidate data storage unit 6 or image data is registered in the pseudo defect image storage unit 2 in advance. Judgment is made by comparing with. The defect candidate data determined not to be a pseudo defect by the pseudo defect determination unit 3 is stored in the defect data storage unit 7 as true defect data excluding the pseudo defect. When automatically comparing using a machine, the pseudo-defect determination unit 3 uses a difference image. However, since the pattern is extracted as having a difference, an actual image is not detected in the comparison because the pattern is extracted. It is desirable to compare after processing such as blurring.

  The defect data display unit 8 displays the image data stored in the defect candidate data storage unit 6. An engineer or an operator determines whether or not the defect is a pseudo defect from the displayed image data of the defect, and if it is determined that the defect is a pseudo defect, the pseudo defect image storage unit from the pseudo defect registration unit 9 In step 2, pseudo defect data is registered.

  Here, the image data stored in the inspection image data storage unit 5 and the image data stored in the defect candidate data storage unit 6 will be described. FIG. 9 is a diagram illustrating an example of image data of an object to be inspected and a difference image detected by the image comparison inspection unit 1 according to the present invention. The defect image 24 and the comparison image 26 are a part of each image stored in the inspected image data storage unit 5. In the case where only one of the plurality of regular patterns is recognized as the defect image 24 and the other part is recognized as the comparison image 26, the image comparison inspection unit 1 causes the defect image 24 to be recognized. The defect image 24 is recognized as a defect, and the defect image 24 is stored in the defect data A together with the difference image 25 from the comparison image 26. At this time, the position where the defect image 24 is included in the entire inspection object is also stored. The defect data display unit 8 superimposes the defect image 24 and the difference image 25 on the display. If the difference image 25 is simply displayed, it is impossible to recognize what is determined to be a defect, and if only the defect image 24 is displayed, it is difficult to find out where the defect is determined to be. However, by displaying the defect image 24 and the difference image 25 in an overlapped manner, if the user is an experienced engineer or operator, should the defect be easily determined as a defect or treated as a pseudo defect? Judgment can be made. In order to make the defective part easier to understand, the difference image 25 may be blinked after being superimposed and displayed.

  In addition, in the case of inspecting a pattern using a new reticle (mask) or in the case of a new mask inspection, it is not often possible to know a pseudo defect in advance, so it is necessary to register the pseudo defect by a normal inspection method. is there. When a defect is detected by the difference image as shown in FIG. 9 and it is determined that a human sees the actual image as a pseudo defect and further registration is necessary, the defect image 24 is compared with the comparison image 26. 25 and the pseudo defect image storage unit 2 as pseudo defect data.

  Next, the main part of the defect inspection method using this defect inspection apparatus will be described with reference to the flowcharts of FIGS. FIG. 2 is a diagram showing an overall flowchart of the defect inspection method according to the first embodiment of the present invention. The defect inspection method is roughly divided into processing step S1 for registering a pseudo defect image in the pseudo defect image storage unit 2 in advance, and step S2 for performing an actual defect inspection using the pseudo defect image data registered in step S1. .

  The defect inspection in step S2 will be described on the assumption that a pseudo defect image is registered in the pseudo defect image storage unit 2 in advance. The defect inspection in step S2 includes a defect candidate extraction process that is a pre-process of the defect inspection shown in FIG. 3 and a pseudo defect exclusion process shown in FIG. FIG. 3 is a flowchart of defect candidate extraction processing that is preprocessing for defect inspection. It is assumed that the image data to be inspected is already stored in the inspected image data storage unit 5. First, as an initial setting, the defect data A in the defect candidate data storage unit 6 which is defect candidate data including pseudo defects is cleared (step S11). Next, image data is read from the inspected image data stored in the inspected image data storage unit 5 (step S12). Next, the read image data is compared with image data of an equivalent portion having periodicity (step S13). If they do not coincide with each other, data of other equivalent parts having periodicity is further read and compared to determine which one is defective (step S14). When it is determined in step S14 that the defect is present, the defect data is stored in the defect candidate data storage unit 6 as defect data A together with the position, the defect image, and the difference image from the comparison image (step S15). Whether or not there remains an inspected image that has not been inspected for defects remains. As long as an inspected image that has not been inspected remains, the process returns to step S12 to repeat the defect inspection. When the defect inspection is completed for all the images to be inspected, the defect extraction ends (step S16).

  Next, a detailed flowchart of the latter half of step S2 for determining whether or not the extracted defect is a pseudo defect will be described with reference to FIG. It is assumed that defect data is already stored in the defect candidate data storage unit 6 and pseudo defect data is registered in the pseudo defect image storage unit 2. First, as an initial setting, the defect data B in the defect data storage unit 7 which is defect data excluding the pseudo defects is cleared (step S21). Next, one piece of defect data A is read from the defect candidate data storage unit 6 (step S22). Next, the pseudo defect data to be compared with the defect data A read in step S22 is read from the pseudo defect image storage unit 2 (step S23). Next, the defect image read in step S22 and the pseudo defect image read in step S23 are compared, and it is inspected whether the defect data A read in step S22 is a pseudo defect (step S24). As a result of the inspection, if it is determined that there is no identity with the pseudo defect data, the next pseudo defect data is read and the inspection is continued. This process is continued until identical pseudo-defect data is found (Yes in step S25) or until it is determined that all the registered pseudo-defect data do not match (Yes in step S26). If no identical pseudo defect is found, it is determined that the defect is not a pseudo defect, and defect position data and defect image data are registered in the defect data B as a true defect (step S27). When a pseudo defect having the same identity is found, the defect is not registered in the defect data B as a pseudo defect. This process is repeated until all the defect data A is processed (step S28). Finally, the defect data storage unit 7 stores defect position data and defect image data of only the defects determined not to be pseudo defects in the defect data A as defect data B.

  Next, FIG. 5 is a flowchart of the pseudo defect registration process which is step S1 of FIG. First, as preprocessing for pseudo defect registration, an area on a semiconductor wafer from which pseudo defects are extracted is set (step S51). In addition, a threshold value of a difference image that is a criterion for image comparison inspection is set (step S52). In step S51 and step S52, the total number of defects to be subjected to pseudo defect extraction is narrowed down. The total number of defects is preferably selected to be about 1000 to 3000 or less in order to allow all to be seen by humans.

  Next, for the region set in step S51, the image comparison inspection unit 1 performs image comparison inspection based on the threshold value set in step S52 (step S61), and a collection of defect images obtained by this image comparison inspection. Is stored in the defect candidate data storage unit 6 as the prior defect data file C (step S62). The provisional inspection, which is the pre-processing for pseudo defect registration, is completed by the processing so far.

  Next, extraction of pseudo defect image data is started. First, a pseudo defect image registration data file D is created (step S63). Next, the defect image is read from the previous defect data file C (step S64). If an uninspected defect image remains (No in step S65), visual inspection is performed, and it is determined whether or not the defect image should be registered as a pseudo defect (step S66). If it is determined that it should be registered as a pseudo defect, the image data is registered in the pseudo defect registration data file D (step S67), and the process returns to reading the next defect image from the previous defect data file C (step S64). On the other hand, if it is determined that it is not necessary to register for the pseudo defect, the process returns to step S64 without registering for the pseudo defect (No in step S66). The processing from step S64 to step S67 is repeated for all defect images in the previous defect data file C (step S65).

  Next, the process proceeds to the verification of the validity of the registered pseudo defect (step S70, the whole in the broken line). First, the initial value of the variable Nr representing “the number of prior defects that can be removed as pseudo defects” is set to 0, and the total number of defects obtained from the prior defect data file C is set to the variable Nc (step S71). Next, a defect image is read from the previous defect data file C (step S72). It is determined whether or not the read defect image is identical to any image registered in the pseudo defect registration data file D (step S74). If the defect image read from the previous defect data file C is identical to any image registered in the pseudo defect registration data file D, it is determined as a pseudo defect. On the other hand, when there is no identity with any image registered in the pseudo defect registration data file D, it is determined that the defect image is not a pseudo defect. In this step S74, the pseudo defect determination unit 3 can automatically perform the determination.

  If it is determined that the defect is a pseudo defect, 1 is added to the “number of pseudo defects that can be removed” Nr (step S75), and the process returns to reading the next defect image (step S72). On the other hand, if it is determined that the defect is not a pseudo defect, the process directly returns to step S72 (No in step S74). The processing from step S72 to step S75 is repeated for all defect images in the previous defect data file C (step S73). When the determination of whether or not the defect is a pseudo defect is completed for all the defect images, the “total number of defects in the prior defect data file C” Nc is compared with the “number that can be removed as a pseudo defect” Nr. Then, it is determined whether or not “the number that can be removed as pseudo defects” Nr is acceptable (step S76). If it is determined that it is not acceptable, the process returns to step S64 and starts again from the process of registering a new pseudo defect in the pseudo defect registration data file D. On the other hand, if the number of registered pseudo defects is acceptable, the pseudo defect registration data file D is closed to complete the pseudo defect registration process (step S81).

  This pseudo defect registration operation is not necessary when the same product is stably flowing through the production line. Registering a pseudo defect when a new product with a different tendency to generate a pseudo defect flows on the production line or when a change occurs in the production line and a pseudo defect that has not occurred so far is generated. Is sufficient.

  In addition, the existing pattern recognition technique can be used for the process of comparing the image of the pseudo defect registered in the flowcharts of FIGS. 4 and 5 with the image of the defect to determine whether or not it is a pseudo defect. . In particular, in determining whether or not it is a pseudo defect, both the data of the registered image of the pseudo defect and the image of the normal product when there is no pseudo defect corresponding to the pseudo defect part are registered, and the defect image Is compared with the image of both the pseudo defect image and the image of the normal product, and if the defect image is included in the range of the pseudo defect image and the image of the normal product, it is treated as a pseudo defect, and the defect image is If it exceeds the range of the image and the normal image, it can be determined as a defect.

  FIG. 6 is a flowchart of the second embodiment of the present invention. The second embodiment replaces the flowcharts of FIGS. 3 and 4 in which “actual defect inspection is performed using pseudo defect image data” of the first embodiment. For the pseudo defect registration process, the flowchart of FIG. 5 can be used as it is. In addition, the configuration of the defect inspection apparatus according to the second embodiment can use the configuration of FIG. 1 as it is. In FIG. 6, steps having substantially the same processing contents as those in FIGS. 3 and 4 are denoted by the same reference numerals.

  First, as an initial setting, the defect data B in the defect data storage unit 7 is cleared (step S21), and the defect data A in the defect candidate data storage unit 6 is cleared (step S11). Next, inspection image data is read from the inspection image data storage unit 5 (step S12), image comparison inspection is performed (step S13), and defects are acquired (step S14). Steps S12 and S13 are repeated until a defect is found by image comparison inspection and one defect can be acquired.

  Next, the acquired defect is registered in the defect data A of the defect candidate data storage unit 6 (step S15), the defect candidate data is read from the defect data A (step S22), and the pseudo defect data is stored from the pseudo defect image storage unit 2. Are read (step S23) and image comparison is performed (step S24). If the defect candidate data is identical to any of the pseudo defect data stored in the pseudo defect image storage unit 2, the defect candidate data is determined to be a pseudo defect, and is identical to any of the pseudo defect data. If not, it is determined that the data is true defect data (step S25). If it is determined that the data is true defect data, the defect data B is registered (step S27), and the process returns to reading the next image to be inspected (step S12). On the other hand, if it is determined in step S25 that the defect is a pseudo defect, the process returns to step S11 without registering the defect data B. This process is repeated for all the inspected images (step S28).

  In the second embodiment, the defect data A is stored only temporarily, and each time a defect (defect candidate) is found in the image comparison inspection (step S13), the defect data A is sequentially compared with the pseudo defect image. Since the defect data A is image data, when the defect candidate extraction process (FIG. 3) and the pseudo defect exclusion process (FIG. 4) are collectively performed as in the first embodiment, the defect candidate extraction process obtains the defect data A. In addition, a large-capacity storage device for storing all the defect data A is required. However, if the defect candidate extraction (step S14) and the pseudo defect determination (step S25) are sequentially performed as in the second embodiment, there is an advantage that the capacity of the storage device for storing the defect data A can be reduced.

  Next, Embodiment 3 of the present invention will be described. In the first and second embodiments, the pseudo defect exclusion process for excluding the pseudo defects from the defect candidates is performed regardless of the number of defects. However, if the number of defects is small, it is not necessary to analyze the defects, and therefore there is a case where it is not necessary to perform the pseudo defect exclusion process that is the premise thereof. In this case, the defect inspection process can be simplified by omitting the pseudo defect exclusion process. Example 3 is an example in which the defect inspection process is simplified by determining the number of defects.

  FIG. 7 is a diagram showing the configuration of the defect inspection apparatus according to the third embodiment of the present invention. 7, parts that are substantially the same in configuration and operation as those in FIG. 1 are denoted by the same reference numerals, and description thereof is omitted. 7 is different from FIG. 1 in that a defect number determination unit 21 is provided. The defect number determination unit 21 determines whether or not the pseudo defect determination unit 3 performs the pseudo defect determination by determining the defect number data stored in the defect candidate data storage unit 6.

  Next, a processing procedure of the defect inspection method according to the third embodiment will be described. FIG. 8 is an overall flowchart of the defect inspection method according to the third embodiment. In FIG. 8, the defect candidate extraction processing step S41 is the same as the flowchart of FIG. The third embodiment is different from the first embodiment in that a defect number determination step S42 is performed after the defect candidate extraction processing step S41. The defect number determination step S42 is performed using the defect number data included in the defect data A. In the defect data A, the size of the lot to be inspected and the number of samples, and the number of defects detected as defects in the lot or sample are recorded. It is possible to determine whether or not to perform the pseudo defect removal processing step S44 by comparing this data with the past inspection record in the same process of the same product. As a determination method, assuming that the distribution of past defect number data is a normal distribution, the determination may be made based on whether or not the number of defects equal to or greater than three times the standard deviation is detected. When a defect of 3 times or more is detected, the pseudo defect exclusion process in step S44 is performed. However, when the number of defects is less than 3 times, the pseudo defect exclusion process is omitted. Here, in order to maintain the consistency of the defect data B with and without performing the pseudo defect exclusion process, the defect data A is copied to the defect data B (step S43). If this data is not used in the subsequent process, the process of step S43 can be omitted. For the pseudo defect exclusion process in step S44, the process of the flowchart of FIG. 4 which is the first embodiment or the process of the flowchart of FIG. 6 which is the second embodiment can be applied as it is. When the process of the flowchart of FIG. 6 is performed in step S44, the defect data A output in step S41 does not need to include defect candidate image data.

  In the inspection apparatuses and methods according to the first to third embodiments described above, the computer can be caused to function as the inspection apparatus according to the first to third embodiments by a computer and a program operating on the computer. Further, the computer and a program operating on the computer can cause the computer to execute the inspection methods of the first to third embodiments.

  In this case, in the configuration of the inspection apparatus shown in FIGS. 1 and 7, the image recognition unit 4 controls the image sensor connected to the computer and a program for taking in the image data output from the image sensor into the computer. And can be realized. Furthermore, the pseudo defect image storage unit 2, the inspected image data storage unit 5, the defect candidate data storage unit 6, and the defect data storage unit 7 are respectively a main storage device such as a RAM of a computer, a cache memory, and a hard disk. And other auxiliary storage devices such as an optical storage device and a magnetic storage device. The functions of the image comparison inspection unit 1, the pseudo defect determination unit 3, and the defect number determination unit 21 can be realized by a program that causes a computer to realize such functions. Further, the defect data display unit 8 can be realized by a computer screen and a program for causing such a display to be displayed on the screen. Furthermore, the pseudo defect registration unit 9 can be realized by an input device such as a keyboard or a mouse, and a program that receives an input to the input device.

  Furthermore, each process of the flowcharts of FIGS. 2 to 6 and 8 is performed by the computer and the program having the above-described configuration except that the operator performs “determination as to whether or not to register as a pseudo defect” in step S66 of FIG. realizable.

  The computer program can be installed in a computer via a semiconductor memory such as a RAM, a storage medium such as an optical storage medium or a magnetic storage medium, or the Internet, and the computer can function as a defect inspection device. The inspection method can be executed.

  The present invention has been described with reference to the embodiments. However, the present invention is not limited to the configurations of the above embodiments, and various modifications that can be made by those skilled in the art within the scope of the present invention. Of course, modifications are included.

  In the above embodiment, the defect inspection apparatus and method have been described by taking the circuit pattern of the semiconductor integrated circuit formed on the surface of the semiconductor wafer as an example. However, the present invention is not limited to the inspection target having a regular pattern. Applicable. Furthermore, even if it does not have a regular pattern, if an object without defects is known in advance in the inspection object, by comparing with the object without the defect, It is possible to inspect the inspection object for defects in appearance inspection.

  In the above embodiment, the pseudo defect registration process is performed based on the operator's judgment. However, the operator need not necessarily perform the pseudo defect registration based on the operator's judgment. For example, by building correlation data between defects in the appearance inspection results and defective products of the function and quality in the final product, the locations determined as pseudo defects in the appearance inspection results correlate with defects in the final product. In such a case, the pseudo defect is deleted from the registration data of the pseudo defect, and a defect in which no correlation between the defect in the final product and the defect in the appearance inspection result is found is automatically registered as a pseudo defect. Is also possible.

It is a figure which shows the structure of the defect inspection apparatus of one Example of this invention. It is a figure which shows the whole flowchart of the defect inspection method of one Example of this invention. In one Example of this invention, it is a flowchart which shows the detail of a defect candidate extraction process. In one Example of this invention, it is a flowchart which shows the detail of a pseudo defect exclusion process. In one Example of this invention, it is a flowchart which shows the detail of a pseudo defect registration process. It is a flowchart which shows the detail of the defect inspection method in another Example of this invention. It is a figure which shows the structure of the defect inspection apparatus of another Example of this invention. It is a figure which shows the whole flowchart of the defect inspection method of another Example of this invention. It is a figure which shows an example of the image data of the to-be-inspected target object in this invention, and the difference image detected by the image comparison test | inspection part.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image comparison inspection part 2 Pseudo defect image memory | storage part 3 Pseudo defect determination part 4 Image recognition part 5 Image data storage part 6 Defect candidate data (defect data A, defect data C) memory | storage part 7 Defect data (defect data B) Storage unit 8 Defect data display unit 9 Pseudo defect registration unit 21 Defect number determination unit 24 Defect image 25 Difference image 26 Comparison image

Claims (12)

An image comparison inspection unit that compares image data of a plurality of objects to be inspected and inspects for the presence or absence of defects based on the presence or absence of identity
A pseudo defect image storage unit for storing a pre-registered pseudo defect image;
For a defect detected by the image comparison inspection unit, a pseudo defect determination unit that compares the pseudo defect image stored in the pseudo defect image storage unit and determines whether the defect is a pseudo defect;
Defect inspection device with   The defect inspection apparatus according to claim 1, wherein the inspection object is a part of an object having a repetitive pattern, and the comparison of the image data is an image comparison inspection unit that compares image data between the repetitive patterns.   The defect inspection apparatus according to claim 1, wherein the inspection object is a part of a semiconductor wafer.   When a certain number of defects are detected by the image comparison inspection unit, the pseudo defect determination unit performs a pseudo defect determination. When the detected defects are less than a certain number, the pseudo defect determination unit The defect inspection apparatus according to claim 1, further comprising a defect number determination unit that omits the determination.   The defect inspection apparatus according to claim 1, further comprising an image recognition unit that recognizes an image of the inspection target object and generates the image data. Comparing the image data of a plurality of objects to be inspected, and inspecting for the presence or absence of defects by the presence or absence of identity;
Comparing the pre-registered pseudo defect image with the defect inspected in the step of inspecting the presence or absence of the defect, and determining whether or not the defect is a pseudo defect;
A defect inspection method comprising:   The defect inspection according to claim 6, wherein the inspection object is a part of an object having a repetitive pattern, and the step of inspecting is a step of comparing the image data between the repetitive patterns and inspecting for the presence or absence of a defect. Method.  The defect inspection method according to claim 6 or 7, wherein the object to be inspected is a part of a semiconductor wafer. Displaying an image including a portion determined to be a defect in the step of inspecting for the presence or absence of the defect;
Allowing the displayed defects to be registered as pseudo defects;
The defect inspection method according to claim 6, further comprising:   In the inspecting step, when the number of detected defects is equal to or greater than a certain number, a step of determining whether or not the defects are pseudo defects is performed, and when the number of defects is less than the certain number, the determination is performed. The defect inspection method according to claim 6, wherein the step is omitted. Comparing image data between repeated patterns for an object to be inspected having a repeated pattern, and inspecting for the presence or absence of defects between the repeated patterns; and
When one or more defects are detected in the step of inspecting for the presence or absence of the defect, for each of the detected defects, a plurality of pre-registered pseudo defect images are compared with image data including the defect, If the image data including a defect is identical to any of the plurality of pseudo defect images, determining that the defect is a pseudo defect;
Out of the detected defects, outputting a defect as a defect list excluding defects determined to be the pseudo defects;
Including defect inspection method.   A defect inspection program for causing a computer to execute the defect inspection method according to claim 6.

Images