JP2012042390A - Defect inspection device and defect inspection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a low-cost defect inspection device.SOLUTION: The defect inspection device of an embodiment includes: detection means 40 which detects a plurality of defects generated on a substrate surface or a film surface in a manufacturing process; clustering means 41 which classifies the plurality of defects detected by the detection means into defect groups on the basis of their positions; identification information giving means 42 which gives identification information to the defect groups and the defects; and storage means 34 which holds position information of the defect groups and the defects together with the identification information given by the identification information giving means. Furthermore, the defect inspection device of the embodiment includes: grouping means 42 which draws a line between two selected from the defect groups and the defects and, if the difference of inclinations of the lines is within a prescribed range, determines that the defect groups and the defects on these lines belong to the same group to give the same label to these defect groups and defects and holds the label in the storage means together with their position information and identification information; and deletion means 42 which keeps position information about two or more defect groups or defects out of defect groups and defects belonging to each group and deletes position information about the other defect groups and defects in the group from the storage means.

Description

  Embodiments described herein relate generally to a defect inspection apparatus and a defect inspection method for a semiconductor device.

  In the manufacturing process of a wafer (semiconductor substrate) on which a semiconductor device is formed, inspection of foreign matter and defects has a great influence on product yield. In particular, grasping the type and distribution of defects on the entire wafer surface is a necessary factor for setting the yield prediction accuracy and priorities for defect countermeasures. In general, a defect inspection is performed by a comparative inspection method using a plurality of acquired images.

JP 2008-268232 A Japanese Patent No. 29986868

  However, when a large number of defect elements are discontinuously present at a plurality of locations, they cannot be recognized as the same type of defect only by the distance. It is necessary to determine the coordinates of the distribution of the coordinates and to count those defects as other types of defects. Therefore, the number of defects increases, the amount of calculation for image processing becomes enormous, and the processing time becomes longer. Therefore, there is a problem that a large-scale image processing system must be constructed at a high cost for defect inspection.

  The defect inspection apparatus according to the embodiment includes a detection unit that detects a plurality of defects generated in a manufacturing process on a surface of a semiconductor substrate or a film formed on the substrate, and the plurality of the detection units detected by the detection unit. Clustering means for classifying defects into defect groups based on positions, identification information giving means for giving identification information to the defect groups and the defects, and position information for the defect groups and the defects are given by the identification information giving means Storage means for holding the identification information together. Furthermore, the defect inspection apparatus of the embodiment selects two lines from the defect group and the defect, draws a straight line, and the straight line having a difference in inclination within a predetermined range among the plurality of straight lines drawn. The defect group and the defect above belong to the same group and the same group label is attached to the storage means together with the position information and the identification information, and for each group, the group belonging to the group Deletion means for deleting the other position information from the storage means while leaving the position information about two or more defect groups or the defect from among the defect group and the defect.

1A and 1B are diagrams for explaining a defect detection technique based on a difference image. FIG. 1A is an image to be inspected, FIG. 1B is a reference image, FIG. 1C is a difference image, and FIG. ) Shows a diagram in which a plurality of defects are determined to be the same defect. FIG. 2 is a diagram showing scratch defects on the wafer. FIG. 3 is a diagram showing a configuration of the defect inspection apparatus according to the present embodiment. 4A and 4B are diagrams for explaining image comparison. FIG. 4A shows an image α, FIG. 4B shows an image β, and FIG. 4C shows a difference image (α−β). FIG. 5 is a diagram illustrating a pixel number distribution for each luminance difference of the difference image. FIG. 6 is a diagram illustrating an internal configuration of the signal processing unit. FIG. 7 is a diagram showing the distribution of a plurality of detected defects (groups) in the inspection area. FIG. 8 is a flowchart showing the cluster processing executed by the cluster processing unit B.

  The technology that is the premise of the embodiment described below will be described first. The procedure is shown in FIG. 1 as an example. FIG. 1 (a) is a difference image in which pixels having a pixel value larger than a predetermined threshold are obtained by performing a process for each pixel of the defect 21 of the image to be inspected in FIG. 1 (a) and the reference image in FIG. 1 (b). The defect 22 is detected as c).

  However, due to the surface roughness of the wafer (semiconductor substrate) and inspection device noise (stage vibration, electrical noise of the detector, etc.), even if a general averaging filter is executed before processing, a single defect is in close proximity. It may be detected as a defect. Further, in recent years, the process becomes more difficult due to miniaturization, and the specific gravity of systematic defects is increasing for defects generated on the surface of a semiconductor substrate and the surface of a film formed on the substrate. As a characteristic of the systematic defect, there is a case where the defect is distributed in part in the distribution of the position of the defect or the position is localized on the outer periphery of the wafer.

  When a systematic defect occurs, only one defect type is detected as a large amount of defects. Therefore, since the number of defects increases, the load on the inspection equipment increases, and the observation work becomes sampling observation from the total number of defects, and there is a small proportion of other fatal defect types, so there is a risk of missing in sampling observation. is there. In that case, it is not possible to efficiently grasp the number of each defect type and to observe the defect (a system using an optical microscope or an electron microscope is used). Conventionally, as the above solution, if a plurality of defects detected as separate ones are within a predetermined distance, the plurality of defects are determined as the same defect 23 as shown in FIG.

  However, as shown in FIG. 2, when the defect elements such as the scratch defect 24 of the CMP process on the wafer are linearly present in large numbers and discontinuously at a plurality of locations, the same kind of the same type is required only by the distance. It cannot be recognized as a defect. Alternatively, even if defects are distributed linearly for the same reason, they cannot be recognized as one defect because they cannot be detected linearly due to variations in accuracy of the inspection apparatus.

  Hereinafter, a defect inspection apparatus and a defect inspection method according to embodiments of the present invention will be described in detail with reference to the accompanying drawings. Note that the present invention is not limited to these embodiments.

(Embodiment)
FIG. 3 shows a configuration of the defect inspection apparatus 100 of the present embodiment. A stage 26 on which a wafer 25 to be inspected is mounted; an irradiation unit 28 that irradiates the wafer 25 with inspection light 27 (or electron beam); a detection unit 30 that detects reflected light 29; a signal processing unit 31 that processes the detection signal; It comprises a monitor 32 for displaying the processed signal, a control PC 33 for controlling the apparatus, and a storage unit 34 for storing inspection recipes and inspection results.

  Although omitted in FIG. 3, an optical lens (electromagnetic lens in the case of an electron beam) is formed between the irradiation unit 28 and the wafer 25. Next, a procedure for acquiring an image with the defect inspection apparatus 100 having this configuration will be described. An inspection recipe for each wafer type stored in advance from the type of wafer to be inspected by the operator is selected from the storage unit 34 by the control PC 33. Next, the defect inspection apparatus 100 irradiates the wafer 25 with inspection light 27 by moving the stage 26 to the position of the wafer 25 to be inspected in advance based on the inspection recipe stored in the storage unit 34.

  The reflected light 29 reflected on the wafer 25 is detected as a signal by the detection unit 30. Next, the stage 26 is moved to the next inspection location and the same operation is repeated. At this time, the operation of irradiating the inspection light while moving the stage 26 can be continuously repeated without stopping the stage 26. The signal that has entered the detection unit 30 is transferred to the signal processing unit 31 to perform predetermined processing, and is displayed on the monitor 32 as an inspection result. And it is memorize | stored in the memory | storage part 34 as a test result.

  Next, image comparison will be described in detail. FIG. 4 shows an image obtained by acquiring images of two similar patterns by the above method. When comparing the image α in FIG. 4A and the image β in FIG. 4B with the two images, the difference image in FIG. 4C is the difference in luminance between pixels of the image β from the image α. Is calculated. Here, it is assumed that the image α has a defect 35.

  In the difference image of FIG. 4C, if a difference in luminance occurs only at the position of the defect 35 and there is no noise, the difference image is exactly the same as the image α in FIG. The difference image of FIG. 4C is a graph showing the number of defects for each absolute value of the luminance difference, the absolute value of the luminance difference on the horizontal axis, and the number of absolute values of the luminance difference of each pixel on the vertical axis. 5 can be created. Then, the defective portion 38 is separated on the graph from the non-defective portion 36. The defect 35 can be detected by providing a threshold 37 of the absolute value of the luminance difference that detects the defective portion 38. According to this method, even when any film is formed on the wafer 25, defects on the film surface can be detected in the same manner, and the following description is also the same. Further, even for a minute change in shape due to pattern variations, the variations are not erroneously detected by comparing with adjacent patterns.

  Next, the internal configuration of the signal processing unit 31 will be described with reference to FIG. The signal processing unit 31 includes a signal filter unit 39, a defect detection unit 40, a cluster processing unit A41, a cluster processing unit B42, and a defect classification unit 43. The signal sent from the detection unit 30 into the signal processing unit 31 is subjected to preprocessing for removing noise by the signal filter unit 39. As the filter method, a general averaging process or the like in image processing can be selected according to the image. General processing includes differentiation processing, binarization, smoothing processing, etc., and an optimal processing may be used according to noise. As described above, the signal from which the noise has been removed is sent to the defect detection unit 40, and the defect is detected by the image comparison method.

  As an image to be compared at this time, an image in which a predetermined portion in the inspection wafer or a portion of the same pattern on another wafer is stored, or a design pattern may be used. The detected defect is subjected to clustering processing in the cluster processing unit A41 so that defects having a predetermined distance or a distance equal to or less than the number of pixels are the same defect (group). In parallel with the clustering process, the defect classification unit 43 can classify the defect based on the feature amount (luminance, number of pixels, shape, etc.). If classification can be performed, information on defects that are not necessary can be deleted from the storage unit 34, or can be performed based on the feature amount of defects when clustering.

  FIG. 7 shows a state of the defect (group) after the cluster processing unit A41 performs the clustering process. FIG. 7 shows the distribution of a plurality of detected defects (groups) in the inspection area. Thereafter, the cluster processing described below is performed in the cluster processing unit B42 according to the flowchart shown in FIG. First, numbering (identification information addition) processing for assigning defects (groups) 1, 2, 3,..., 12 (= n) and defect numbers (identification information) is performed on all defects (groups) (step) S1). The position information of these defects (group) is held in the storage unit 34 together with the assigned defect number. Next, the initial value of i is initialized to i = 1 (step S2).

  Next, in step S3, a straight line is drawn from the defect (group) 1 to the defect (group) 2 to 12, and the inclination is calculated. For example, the straight line may be drawn so as to pass through the center of gravity of a defect group and the point of a single defect as a point. In step S3, generally, only a defect having no group number among the defects (groups) is drawn to calculate its inclination. When i = 1, there is no defect (group) to which the group number is assigned yet, so a straight line is drawn to all the defects (groups) 2 to 12 and the inclination is calculated.

  Next, the process proceeds to step S4, where the difference (in absolute value) between the slopes of the respective straight lines obtained in step S3 is obtained, and the straight lines satisfying such a condition are found if they are within a predetermined range. Are in the same group. In the example of FIG. 7, the defect (group), (defect (group) 1, 3, 6) and (defect (group) 7, 8, 9, 10, 11, 12) connected by a thick line are the same. Belongs to the group. Each defect (group) on the straight line belonging to the same group is regarded as a defect (group) due to the same cause, and is given the same group number (label) indicating that it belongs to the same group. This group number (label) is added to the position information of each defect (group) together with the defect number assigned in step S1, and is held in the storage unit 34.

  Then, the process proceeds to step S5, and two or more representative defects (groups) are left out of the coordinates of the defect (group) of the same group to which the same group number (label) is assigned, and other information is stored. Erase from part 34. For example, information other than the defect (group) having the smallest defect number and the defect (group) having the largest defect number is erased from the storage unit 34.

  Thereafter, i = i + 1 and i are incremented (step S6), and the process proceeds to step S7. If i is less than n (the maximum value of the defect number) in step S7 (No in step S7), the process proceeds to step S3, and the defects i to 12 (= n) are the same starting from the defect i. A line is drawn only for the defect having no group number, and its inclination is calculated.

  The above loop is repeated until i becomes n (the maximum value of the defect number) or more (step S7, Yes) in step S7, and the grouping is completed by performing all the defects (groups). When calculating the inclination between defect groups, it is possible to determine that the inclination along the main directionality of the pattern is the same defect using the design data. The main directionality referred to here indicates a direction that can be automatically calculated because, for example, in the case of a line pattern, there are many pattern outlines in a certain direction.

  Even if a large number of defects (groups) due to the same defect type generated due to systematic causes distributed substantially along a straight line are detected according to the above-described embodiment, the defects (groups) are grouped out from the storage unit 34 after being grouped. Therefore, the storage capacity of the storage unit 34 can be reduced. For this reason, it is possible to prevent oversight of a small number of defects other than the same defect type that occurs frequently.

  In other words, the defect inspection apparatus using the above defect inspection method can inspect by avoiding an overflow that exceeds the storage capacity of the storage means due to an increase in the number of defects (group) due to the same type of defect (group). Therefore, the distribution of defects on the entire surface of the wafer (or the entire surface of the film formed on the wafer) can be obtained without imposing a load on the storage unit or the image processing unit of the inspection apparatus.

  As described above, a wide range of defects due to systematic defects are often distributed linearly mainly due to the cause of occurrence. This is because the miniaturization advances and defects exist along the wiring and holes, so that the linearity increases. Accordingly, when the coordinates where the defect occurs are processed to some extent, it is possible to predict where the same defect will appear next. In the embodiment described above, by utilizing this fact, it is possible to determine a defect within a predetermined error range as the same defect without performing complicated image processing calculation using a simple algorithm.

  As described above, according to the above embodiment, a large number of systematic defects can be classified (hereinafter categorized) at high speed and simply. Therefore, it is possible to quickly and easily classify defects generated by the same cause in the defect inspection apparatus using the defect coordinates.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  1 to 12, 21 to 24, 35 Defect (group), 25 wafer, 26 stage, 27 inspection light, 28 irradiation unit, 29 reflected light, 30 detection unit, 31 signal processing unit, 32 monitor, 33 control PC, 34 storage Part, 36 non-defect part, 37 threshold value, 38 defect part, 39 signal filter part, 40 defect detection part, 41 cluster processing part A, 42 cluster processing part B, 43 defect classification part, S1-S7 steps.

Claims (5)

Detecting means for detecting a plurality of defects generated in the manufacturing process on the surface of the semiconductor substrate or the surface of the film formed on the substrate;
Clustering means for classifying the plurality of defects detected by the detection means into defect groups based on positions;
Identification information giving means for giving identification information to the defect group and the defect;
Storage means for holding the defect group and the position information of the defect together with the identification information given by the identification information giving means;
A line is drawn by selecting two from the defect group and the defect, and the defect group and the defect on the straight line in which a difference in inclination is within a predetermined range among the plurality of the drawn straight lines are the same Grouping means for attaching the same group label as belonging to a group and holding it in the storage means together with the position information and the identification information;
Deletion means for leaving the position information about two or more defect groups or the defect from among the defect group and the defect belonging to the group, and deleting other position information from the storage means for each group When,
A defect inspection apparatus comprising: The defect inspection apparatus according to claim 1, wherein the detection unit performs threshold processing on a luminance difference due to image comparison. 2. The defect inspection apparatus according to claim 1, further comprising: a defect classification unit that classifies the plurality of defects detected by the detection unit based on luminance, the number of pixels, or a shape, or based on design data. . A detection step of detecting a plurality of defects generated in the manufacturing process on the surface of the semiconductor substrate or the surface of the film formed on the substrate;
A clustering step of classifying the plurality of defects detected by the detection step into defect groups based on positions;
An identification information giving step for giving identification information to the defect group and the defect after the clustering step;
Storing the defect group and the position information of the defect in the storage unit together with the identification information given in the identification information giving step;
A line is drawn by selecting two from the defect group and the defect, and the defect group and the defect on the straight line in which a difference in inclination is within a predetermined range among the plurality of the drawn straight lines are the same A grouping step of storing the same group label as belonging to a group in the storage means together with the position information and the identification information;
For each group, leaving the position information about two or more defect groups or the defect from among the defect group and the defect belonging to the group, and deleting other position information from the storage means; ,
A defect inspection method comprising: The defect detection method according to claim 4, wherein the detection step is performed by performing threshold processing on a luminance difference by image comparison.

Images