JP2008004863A - Appearance inspection method and device therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technology that improves a substantial sensitivity by suppressing a misreport, while keeping the entire defect capture rate at a high rate in an appearance inspection. <P>SOLUTION: In the appearance inspection, a preliminary inspection is conducted to acquire image information (background information) per at least one die concurrently with the detection of a fault candidate, the acquired image information and coordinate position information inside the die of the fault candidate are superimposed and indicated, and a misreport group is estimated to set a non-inspection region or a sensitivity-deteriorated region based on these information (similarity of backgrounds and density of fault candidates). <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention is directed to thin pattern devices such as semiconductor wafers, TFTs, and photomasks. Fine pattern defects and foreign matter based on an image of a substrate to be inspected obtained by using lamp light, laser light, or an electron beam. More particularly, the present invention relates to an appearance inspection method and apparatus suitable for performing an appearance inspection of a semiconductor wafer.

  Thin film devices such as semiconductor wafers, liquid crystal displays, and hard disk magnetic heads are manufactured through a number of processing steps. In the manufacture of such a thin film device, appearance inspection is performed for each of a series of processes for the purpose of improving and stabilizing the yield. In the appearance inspection, areas corresponding to two patterns originally formed to have the same shape are patterned defects or foreign matter based on the reference image and inspection image obtained by using lamp light, laser light, electron beam or the like. Detect defects such as. That is, the difference between the reference image and the inspection image is calculated and a difference is calculated, and a portion where the difference is larger than a separately defined threshold value is detected as a defect or a foreign object.

  As a threshold calculation method, Japanese Patent Application Laid-Open No. 2000-105203 (Patent Document 1) discloses, for example, a false report on a variation (standard deviation) of a detection signal at or near a portion originally formed to have the same shape. A method of calculating by multiplying a multiple m1 corresponding to the occurrence probability is disclosed.

  In such an inspection, in order to detect a minute defect, it is necessary to make a determination by setting a threshold value low. However, if the threshold value is lowered, many false alarms occur due to sampling errors, minute differences in patterns such as roughness and grain, or uneven brightness due to uneven film thickness. Considering the purpose of the inspection, the false information is completely unnecessary. Therefore, in many cases, the inspection is performed at the sacrifice of sensitivity by setting the threshold value high enough that the ratio of the false information becomes sufficiently small in the whole wafer.

  As a method for improving the sensitivity, Japanese Patent Application Laid-Open No. 2004-79593 (Patent Document 2) divides the entire wafer or die into regions and performs defect determination using different threshold values for each region. A method of performing is disclosed. According to this method, it is possible to suppress the occurrence of false information without reducing the sensitivity of the entire wafer.

JP 2000-105203 A JP 2004-79593 A JP 2004-117229 A

  However, when using the method disclosed in Patent Document 2, it is necessary to perform preliminary inspection, review the result of the preliminary inspection, classify it into real defects and false reports, and divide the area according to the density of false reports was there. For this reason, there is a problem that the review takes a long time.

  It is also possible to divide the area without reviewing and perform defect determination using different threshold values for each area, but in that case it is necessary to determine the area without a guideline, and appropriate area setting There was a problem that it was difficult.

  The first object of the present invention is to estimate the false alarm in a short time based on the result of the preliminary inspection, and to make it possible to set an appropriate non-inspection area or sensitivity reduction area based on the result. It is to provide a method and apparatus thereof.

  In addition, a second object of the present invention is to provide an appearance inspection method and apparatus capable of setting an appropriate non-inspection area or sensitivity reduction area without manual intervention based on the result of preliminary inspection. It is to provide.

  In order to achieve the above object, according to the present invention, in an appearance inspection method and an appearance inspection apparatus using an appearance inspection apparatus, preliminary inspection is performed, and at least one die of image information (inspection image palm) is detected simultaneously with detection of a defect candidate. Background information representing a background image including a threshold image or the like is acquired, and the acquired image information (background information) and defect candidate coordinate information in a die are superimposed and displayed, and the displayed information (background And a non-inspection area or a sensitivity reduction area is set by estimating a group of false reports based on the similarity of the above and the density of defect candidates).

  Further, according to the present invention, in an appearance inspection method and an appearance inspection apparatus using an appearance inspection apparatus, preliminary inspection is performed twice with different sensitivities, and image information (inspection image and threshold value) for at least one die is detected simultaneously with detection of a defect candidate. (Background information representing a background image including a value image) is acquired, and the acquired image information (background information) and position information of defect candidates that increase when the sensitivity is increased are displayed in an overlapping manner. A non-inspection area or a sensitivity reduction area is set by estimating a group of false information based on such information (information on a defect candidate that increases when the sensitivity is increased).

  The present invention also provides an appearance inspection method and an appearance inspection apparatus using an appearance inspection apparatus, and performs preliminary inspection to detect image information (inspection image, threshold image, etc.) for at least one die simultaneously with detection of a defect candidate. Background information representing the background image including), classify based on the in-die coordinate information of defect candidates, select a small number of defect candidates for each classified group, and check whether they are false information, The acquired image information and the in-die coordinate information of the defect candidate for the group containing a large amount of the confirmed false information are superimposed and displayed, and a non-inspection area or a sensitivity reduction area is set based on the displayed information. It is characterized by that.

  Further, according to the present invention, in the appearance inspection method and the appearance inspection apparatus using the appearance inspection apparatus, preliminary inspection is performed, and image information (inspection image or the like) for at least one die is acquired simultaneously with the detection of the defect candidate. A feature amount representing background information is calculated, classification is performed based on the calculated feature amount, a small number of defects are selected for each of the classified groups, and whether or not they are false information is obtained, the acquired image information And the coordinate information in the die of the defect candidate for the group including the confirmed false information are displayed in an overlapping manner, and the false information group is estimated based on the displayed information, and the non-inspection area or the sensitivity reduction area is determined. It is characterized by setting.

  Further, the present invention is an appearance inspection method and appearance inspection apparatus using an appearance inspection apparatus, performs preliminary inspection over the entire wafer surface, calculates the density of defect candidates for each region divided by concentric circles or square lattices, It is tested whether there is a significant difference between the areas in the calculated defect candidate density. If there is a significant difference as a result of the test, a small number of defect candidates are selected from the high-density area and whether they are false reports. If a large amount of false information is included as a result of the confirmation, the region is set as a non-inspection region or a sensitivity reduction region.

  According to the present invention, it is possible to easily estimate a group of false alarms or an area with many false alarms, so that an appropriate non-inspection area or sensitivity reduction area can be set. As a result, the defect capture rate can be increased over the entire wafer or the entire die. Generation of false information can be suppressed without lowering, and a high-sensitivity inspection can be made substantially possible.

  Further, according to the present invention, it is possible to set an appropriate non-inspection area or sensitivity reduction area without manual intervention by automatically setting the sensitivity for each area based on the false alarm area estimation.

  Based on an image of an object obtained by using lamp light, laser light, electron beam, or the like for thin film devices such as semiconductor wafers, TFTs, and photomasks according to the present invention, fine pattern defects, foreign matters, etc. Embodiments of an appearance inspection method and apparatus for performing defect detection and defect classification will be described with reference to the drawings.

[First Embodiment]
A first embodiment of an appearance inspection method and apparatus according to the present invention will be described in detail with reference to FIGS.

  FIG. 1 is a block diagram showing a first embodiment of an appearance inspection apparatus according to the present invention. As a first embodiment of an appearance inspection apparatus, a case of an optical appearance inspection apparatus having a semiconductor wafer as an inspection object will be described. Of course, the appearance inspection apparatus is not limited to the optical type, and can also be applied to an electron beam type appearance inspection apparatus. The optical appearance inspection apparatus includes an inspection object (inspected substrate) 11 such as a semiconductor wafer, a stage 12 that is controlled to move by a mechanical controller 120, a light source 101 that emits irradiation light, and an emission from the light source 101. An irradiation optical system configured to have a condensing optical system 102 for condensing the irradiated light and irradiating the inspection object 11 with the irradiation light condensed by the condensing optical system 102 of the irradiation optical system. An objective lens 103 that condenses an optical image obtained by reflection from the irradiated object 11, an imaging lens (not shown) that forms an optical image condensed by the objective lens 103, and the imaging lens The detection unit 13 includes a detection optical system that includes the image sensor 104 that converts the optical image formed in (1) into an image signal according to brightness (lightness / darkness). Here, the light source 101 is composed of, for example, a lamp light source or a laser light source, and the image sensor 104 is composed of, for example, a CCD linear sensor, a TDI sensor, or a photomultiplier.

  In the present invention, even if the irradiation optical system is high-angle irradiation (bright-field irradiation) of 40 degrees or less with respect to the normal direction of the inspection object 11, the normal direction of the inspection object 11 is used. It may be low-angle irradiation (dark field irradiation) of 40 degrees or more, and is not limited to the irradiation angle. Further, the detection angle with respect to the inspection object 11 in the detection optical system is not limited.

The image processing unit 14 converts an input signal from the image sensor 104 of the detection unit 13 into a digital signal, and performs image correction such as shading correction and dark level correction on the AD converted digital signal. A comparison is made between the preprocessing unit 106 to be performed and the reference image g _n (i, j) detected from the corresponding position (i, j) of the adjacent die and the detected image f _n (i, j). A defect determination unit 107 that outputs a portion where the difference value is larger than a separately set threshold value TH (i, j) as a defect candidate, and the position of the defect candidate determined by the defect determination unit 107 is represented by wafer coordinates and A feature amount (brightness (tone value), contrast, pattern edge density, etc.) representing background information of the defect candidate determined by the coordinate conversion unit 108 that converts to the die coordinate and the defect determination unit 107 is calculated. Feature amount calculation unit 119 representing background information, position information of defect candidates converted into wafer coordinates and die coordinates by the coordinate conversion unit 108, and defect candidate calculated by the feature amount calculation unit 119 representing the background information. Feature value representing background information, difference value ΔS _n−1 (i, j), ΔS _n (i, j), which is a result of comparison in the defect determination unit 107, the smaller absolute value, the defect determination unit Data for storing the standard deviation σ (i, j) indicating the variation between the dies calculated in 107, the set threshold value Th (i, j), the ID of the die including the detected defect candidate, and the like. Based on the coordinates of the detected defect determined by the storage unit 109 and the defect determination unit 107, the threshold value is reset according to the criteria determined for each region in the inspection object, and the comparison by the defect determination unit 107 is performed. Threshold with difference value of reset The area-specific defect determination unit 111 that outputs a larger portion as a defect and the position of the detected defect determined by the area-specific defect determination unit 108 are stored in the data storage unit 109 in a predetermined size (stored). Image cutout unit 112 that cuts out the detected image, the reference image, and the threshold image that have been cut out, and the feature amount of the defect (gray value (gradation value), number of detected pixels (area), and each axial direction from the cut out image A feature extraction unit 113 that calculates the maximum or minimum projection length), a defect classification unit 114 that classifies defects based on the calculated feature amount of the defect, and a user interface unit (GUI unit) in preliminary inspection or the like And an image storage unit 115 that stores an image of an area in the specified inspection object when specified using the inspection recipe input unit 116.

  The overall control unit 15 divides the entire wafer or die into regions based on at least one of position information and background information of defect candidates obtained in the data storage unit 109 by performing preliminary inspection, which is a feature of the present invention, An area-specific sensitivity setting unit 110 having an area-specific sensitivity setting GUI for setting a threshold for area-specific threshold calculation so as to have different sensitivities is connected, and an inspection recipe (created for each product type and process, at least , Product name, process name, wafer size, inspection direction, die size, die matrix information, threshold parameter (s, k, o) and irradiation condition of irradiation optical system (irradiation angle, irradiation intensity, irradiation wavelength or irradiation) Inspection recipe that inputs detection conditions (including detection angle, detection magnification, detection wavelength, detection polarization condition, spatial filter condition, etc.) And an inspection recipe storage unit 118a for storing the inspection recipe, and information on detected defects (for example, defects determined by the region-specific defect determination unit 111) are connected to the power unit (which may be executed by the GUI unit 117) 116. Position information (position coordinates), image information cut out by the image cutout unit 112, feature amount information of defects extracted by the feature extraction unit 113, classification class information classified by the defect classification unit 114, and the like are stored. A storage device 118 having an information storage unit 118b for detected defects is connected, and a user interface unit 117 for receiving inspection recipe changes from the user and displaying detected defect information is connected to the image processing unit 14. The mechanical controller 120 is based on a control command from the overall control unit 15. There is for driving the stage 12 or the like. Although not shown, the image processing unit 14, also detecting unit 13 or the like is driven by a command from the overall control unit 15.

  Next, a defect detection method using the optical appearance inspection apparatus shown in FIG. 1 will be described.

  As shown in FIG. 2, the semiconductor wafer 11 that is an object to be inspected regularly has a large number of dies 21 having the same pattern. The detected images are compared at the same position of two adjacent dies, for example, the region 22 in FIG. 2 and the chip region 23 adjacent thereto, and a portion having a difference between the two is detected as a defect.

  Explaining the operation, the overall control unit 15 continuously moves the semiconductor wafer 11 as the inspection object in the direction opposite to the direction of the scan A shown in FIG. In synchronization with the continuous movement of the stage 12, the image sensor 104 of the detection unit 13 sequentially detects the optical image of the inspection object (inspected substrate) 11 in the direction of scan A, and the chip image is detected by the detection unit. 13 to the image processing unit 14. The image processing unit 14 first converts an input analog signal into a digital signal by the AD conversion unit 105, and performs shading correction, dark level correction, and the like by the preprocessing unit 106. The defect determination unit 107 determines defect candidates by a method described later.

  The region-specific defect determination unit 111, which is a feature of the present invention, uses the region-specific sensitivity setting unit 110 to preliminarily store the wafer coordinates and die coordinates converted by the coordinate conversion unit 108 for each defect candidate detected by the defect determination unit 107. It is determined where the set region is included, and a reference (for example, a region-specific threshold parameter (A (s), A (k), A (o) ))), The threshold value for each area is reset as A (Th (x, y)), and the defect determination unit 107 determines the reference image and the inspection image for the defect candidate stored in the data storage unit 109. A region whose difference value is larger than the reset region-specific threshold A (Th (x, y)) obtained by the following equation (1) is output as a defect. Since the area-specific threshold A (Th (x, y)) is reset according to the area on the wafer, it basically has a wafer coordinate system (x, y). On the other hand, the standard deviation σ (i, j) indicating the variation in brightness is represented by die coordinates (i, j) as described later. Here, A (s) is the lowest threshold value for each area, A (k) is a coefficient for multiplying σ (i, j) for each area by a constant, and A (o) is the offset for each area.

画像切出部１０９では、データ記憶部１０９に記憶された欠陥候補の検出画像、参照画像及びしきい値画像から、領域別欠陥判定部１０８で検出された欠陥の位置を中心として予め定められたサイズで、該欠陥と判定した検出画像、参照画像およびしきい値画像を切出す。特徴抽出部１１０では、複数の欠陥候補各々について、切出した検出画像および参照画像に基づいて欠陥のサイズを表す特徴量、欠陥の明るさを表す特徴量、欠陥の形状を表す特徴量、背景の情報を表わす特徴量などの欠陥分類向けの特徴量を算出する。欠陥分類部１１１では、予め設定された条件を用いて分類を行い各欠陥のクラス情報を出力する。分類方法としては、公知の識別方法のどれを用いてもよい。領域別欠陥判定部１０８から出力される欠陥の位置情報、画像切出部１０９から出力される画像情報、特徴抽出部１１０から出力される欠陥の特徴量および欠陥分類部１１１から出力される欠陥クラス情報は、記憶装置１１８の検出欠陥の情報記憶部１１８ｂに保存され、また、ユーザインターフェース部１１７を介して、ユーザに提示される。

In the image cutout unit 109, the defect position detected by the region-specific defect determination unit 108 is determined in advance from the detected defect candidate image, reference image, and threshold image stored in the data storage unit 109. A detected image, a reference image, and a threshold image determined to be the defect are cut out by size. In the feature extraction unit 110, for each of a plurality of defect candidates, a feature amount that represents the size of the defect, a feature amount that represents the brightness of the defect, a feature amount that represents the shape of the defect, a background amount, A feature amount for defect classification such as a feature amount representing information is calculated. The defect classification unit 111 performs classification using preset conditions and outputs class information of each defect. Any known identification method may be used as the classification method. Position information of defects output from the defect determination unit by area 108, image information output from the image cutout unit 109, feature amounts of defects output from the feature extraction unit 110, and defect class output from the defect classification unit 111 The information is stored in the information storage unit 118b of the detected defect of the storage device 118, and is presented to the user via the user interface unit 117.

A first embodiment 107a of the defect determination unit 107 will be described with reference to FIG. As described in Patent Document 1, the defect determination unit 107a includes an inspection target die image signal output from the preprocessing unit 106, that is, a detected image signal f _n (i, j) and one of the inspection target dies. The image signal of the previous die, that is, the reference image signal g _n (i, j) is input as a set. Here, n is a die number, and (i, j) is an in-die coordinate. The reference image signal g _n (i, j) is obtained by being delayed by the delay memory 301 for a time that the stage moves by one die. The misregistration detection unit 302 calculates the misregistration amount due to stage vibration between two images that are continuously input. At this time, the detected image signal f _n (i, j) and the reference image signal g _n (i, j) are continuously input, but the calculation of the positional deviation amount is performed with a specific length as one processing unit. Sequentially performed for each unit. The difference calculation unit 303 calculates the difference image ΔS _n (i, j) after aligning the detected image and the reference image using the calculated displacement amount. The difference image ΔS _n (i, j) is temporarily stored in the data storage unit 304. Next, a threshold value Th (i, j) is calculated for each pixel using a difference image over a plurality of dies. The difference image ΔS _n (i, j) over a plurality of dies is input to the maximum / minimum removal unit 305. The maximum / minimum removal unit 305 obtains and removes the maximum value and the minimum value from a plurality of input data. From these input data, the sum of squares of the difference signal (ΣΔS _n ² ) is calculated by the square sum calculator 306, and the number of input data m is calculated by the data count unit 307. The threshold value calculation unit 308 uses the square sum ΣΔS _n ² and the number m of input data to indicate the standard deviation σ (i, j) = √ (ΣΔS _n (i, j) ² / m) is calculated, and the threshold value Th (i, j) is calculated by the following equation (2) using the threshold parameters s, k, and o.

ここでｓは最低しきい値、ｋはσ（ｉ，ｊ）を定数倍する係数、ｏはオフセットである。σ（ｉ，ｊ）を計算する時に最大と最小を除外するのは、欠陥が存在するときにその影響でばらつきが大きく算出されるのを防ぐためである。また、異なるダイの同じ画素に欠陥が複数の欠陥が存在する確率は非常に低いため、これ以上除外する必要はない。以上の処理によりダイ間の対応する位置（ｉ，ｊ）毎にσ（ｉ，ｊ）、すなわち明るさのばらつきに応じたしきい値Ｔｈ（ｉ，ｊ）が設定される。このようにしきい値を設定することにより、ばらつきの大きい箇所にあわせてしきい値を設定する必要がなくなり、ばらつきの小さいところでは高感度に欠陥検出できることになる。

Here, s is a minimum threshold value, k is a coefficient for multiplying σ (i, j) by a constant, and o is an offset. The reason why the maximum and minimum values are excluded when calculating σ (i, j) is to prevent a large variation due to the influence of defects when they exist. Further, since the probability that a plurality of defects exist in the same pixel of different dies is very low, it is not necessary to exclude any more. With the above processing, σ (i, j), that is, a threshold value Th (i, j) corresponding to the variation in brightness is set for each corresponding position (i, j) between the dies. By setting the threshold value in this way, it is not necessary to set the threshold value in accordance with a location where the variation is large, and defect detection can be performed with high sensitivity where the variation is small.

The comparison unit 309 sequentially inputs the difference images ΔS _n−1 (i, j) and ΔS _n (i, j) from the data storage unit 304 and compares them with the threshold value Th (i, j) to calculate the absolute difference. A region having both large values is output as a defect candidate. ΔS _n (i, j) is can not be determined either on whether there is a defect of a comparison of only _f n (i, j) and _g n (i, j).

As described above, the defect determination unit 107a compares the difference images ΔS _n−1 (i, j) and ΔS _n (i, j) obtained by die comparison with the threshold Th (i, j). The defect candidate is detected by.

After the defect determination by the defect determination unit 107a, the position of the detected defect candidate is converted into wafer coordinates (x, y) and die coordinates (i, j) by the coordinate conversion unit 108 and stored in the data storage unit 109. Of the difference values ΔS _n−1 (i, j) and ΔS _n (i, j), the standard deviation σ (i, j), and the threshold value Th (i, j). ) Is stored in the data storage unit 109 together with the ID of the die including the defect candidate. Further, the data storage unit 109 stores the detection image, the reference image, and the threshold image in a predetermined size with the position of the defect candidate determined by the defect determination unit 107 as the center.

The difference calculation unit 303 may calculate the absolute difference value | ΔS _n (i, j) |. In that case, in order to exclude the influence when a defect exists, the maximum / minimum removal unit 305 removes the maximum value and the second highest value instead of the minimum value.

  The threshold value calculation unit 308 uses the threshold parameter (s, k, o) stored in the inspection recipe storage unit 118a of the storage device 118.

Next, the operation of the region-specific sensitivity setting unit 110 having the region-specific sensitivity setting GUI, which is a feature of the present invention, will be described with reference to FIGS. FIG. 4 shows a first embodiment (A1) in which the region of the wafer 11 of the region-specific sensitivity setting GUI 401 is divided by concentric circles. The area-specific sensitivity setting unit 110 having the area-specific sensitivity setting GUI 401 has an inspection recipe (at least the product name, process name, wafer size, inspection direction, die size, die matrix information, and irradiation) in which items have been set in advance. It is assumed that the irradiation condition of the optical system and the detection condition of the detection optical system) are selected, and that information is reflected on the screen described below. The sensitivity setting method for each region setting set in the sensitivity setting unit 110 for each region having the GUI 401 includes a method (A1) of dividing a region on the wafer 11 given by the wafer coordinate system (x, y) by concentric circles, and a wafer coordinate system. A method (A2) for designating a rectangular area on the wafer 11 given by (x, y), a method (A3) for designating each die given by a die ID (x, y), and a die coordinate (x, y) There are four types (A4) of specifying a rectangular area on a given die 21. Note that (x, y) indicates position coordinates in the matrix. These four types are selected by tabs 402a to 402d in the sensitivity setting screen 401 for each region. The map 403 displays the area division on the wafer 11 given in the wafer coordinate system. Here, an embodiment (A1) in which the region is divided by concentric circles is shown. By the way, the film thickness distribution normally formed on the wafer changes in the radial direction, and the difference value ΔS _n-1 (i, j) detected by the defect determination unit 107 often changes in the radial direction. It is effective to change the sensitivity (threshold parameter) by dividing the region as described above. The table 405 includes area numbers, radii (r) given in the wafer coordinate system, and first area threshold parameters (A1 (s), A1 (k), A1 (o)). Respectively correspond to the regions 404a to 404d. For example, 404b is a region having a radius r1 or more and less than r2. The sensitivity of the corresponding region can be set by inputting (A1 (s), A1 (k), A1 (o)). To change the range of the region, click the boundary line of the region on the map 403 to move it, or change the numerical value in the radius column of the table 405. If one of the map 403 and the table 405 is changed, the other is also changed in association with it. When it is desired to add a division of the area, after pressing the new button 406, a position on the map 403 where the division is desired to be added is clicked. Concentric boundary lines are added to the map 403, and rows are added to the table 405 so that the radii are arranged in ascending order of radius, and the numerical value of the radius is displayed. In (A1 (s), A1 (k), A1 (o)), the setting value for each region of the entire wafer is displayed first, and is changed as necessary. In order to delete the division of the area, the user clicks the delete button 407 and clicks the boundary line on the map 403. The corresponding boundary line is deleted from the map 403, and the corresponding row is deleted from the table 405. The automatic button 408 is for automatically setting the area based on the preliminary inspection result by a method described later, and the return button 409 is for returning to the state before the automatic setting. In this example, the sensitivity setting for each region is shown in the middle of setting, but in the initial state, no boundary line is displayed on the map 403, and the table 405 shows the radius in the region 1 row. Only the maximum value and the first threshold parameter (A1 (s), A1 (k), A1 (o)) set for the entire wafer are displayed.

  When the inspection result superimposing button 410 is pressed on the sensitivity setting screen 401 for each region, the immediately previous inspection result determined by the region-specific defect determination unit 111 is reflected on the map 403. That is, as shown in FIG. 5, marks are plotted at the defect positions determined by the area-specific defect determination unit 111. The results determined by the region-specific defect determination unit 111 are also reflected on the region-specific sensitivity setting screens selected by the other tabs 402b to 402d. Here, the inspection result to be reflected is the previous one, but a list of inspection results may be displayed and selected from the list.

  When the registration button 411 is pressed, the sensitivity for each region (threshold parameter for each region (A1 (s), A1 (k), A1 (o)) set by the above-described four methods on the sensitivity setting screen for each region 401 is displayed. (A4 (s), A4 (k), A4 (o))) are all added to the inspection recipe being edited and stored in the inspection recipe storage unit 118a of the storage device 118. That is, whether or not each method has been set is stored in the inspection recipe storage unit 118a. In this method, the number of regions, the radius corresponding to the number of regions, the sensitivity for each region (the threshold parameter for each region (A1 (s), A1 (k), A1 (o))). After saving the inspection recipe, the screen 401 is closed. When the cancel button 412 is pressed, the screen 401 is closed without changing and saving the inspection recipe.

  FIG. 6 shows a second embodiment (A2) of the method of designating a rectangular area on the wafer 11 of the GUI 401 for sensitivity setting for each area. This is a screen selected by the tab 402b on the sensitivity setting screen 401 for each region. The map 403 displays the wafer 11 given in the wafer coordinate system and rectangular areas 601a and 601b for sensitivity setting. The table 602 is composed of an area number, a starting point and an ending point of a rectangular area given by the wafer coordinate system, and second area-specific threshold parameters (A2 (s), A2 (k), A2 (o)). 1 and 2 correspond to the rectangular regions 601a and 601b, respectively. Here, the start point is the wafer coordinates (xs, ys) in the lower left corner, and the end point is the wafer coordinates (xe, ye) in the upper right corner. In the initial state, only the wafer 11 is displayed on the map 403 and nothing is displayed on the table 602. To add a rectangular area, a new button 406 is pressed and a rectangle is drawn on the map 403. In order to delete the rectangular area, after pressing the delete button 407, the user clicks on the line of the rectangle to be deleted. To change the rectangular area, click the corner or side of the rectangle on the map 403 to move it, or change the numerical value of the start point or end point in the table 602. If one of the map 403 and the table 602 is changed, the other is also changed in association with the other. Sensitivity is set by entering numerical values in the A2 (s), A2 (k), and A2 (o) fields of Table 602. The functions of the automatic button 408 and the return button 409 are the same as described above. When the registration button 411 is pressed, in this method, the number of areas and the start point, end point, A2 (s), A2 (k), and A2 (o) of the rectangle for the number of areas are stored in the inspection recipe storage unit 118a.

  FIG. 7 shows a third embodiment (A3) of the method for setting the sensitivity for each die of the GUI 401 for sensitivity setting for each region. This is a screen selected by the tab 402c on the sensitivity setting screen 401 for each region. A map 403 displays a wafer 11 in the wafer coordinate system and a die matrix 701 given by a die ID. The table 702 includes area numbers, display colors, and third area threshold parameters (A3 (s), A3 (k), A3 (o)). The dies of the map 403 are colorless or displayed in the display color shown in Table 702. The sensitivity of each die is the third threshold parameter (A3 (s), A3 (k) corresponding to the display color. , A3 (o)). However, colorlessness corresponds to the threshold parameter for each region set for the entire wafer. In the initial state, all dies are displayed in the map 403 as colorless, and nothing is displayed in the columns A3 (s), A3 (k), and A3 (o) of the table 702. To change the designation of each die, the selection state is changed by pressing the row in which the parameters (A3 (s), A3 (k), A3 (o)) are set in Table 702, or the new button 406 or the delete button 407. Then click on the die you want to change. When the table 702 is clicked, the corresponding display color is selected, and the die clicked on the map 403 until the next change of the selection state is displayed in the selected display color. When the new button 406 is pressed, the display color corresponding to the top (area 3 in the example in the figure) of the row in which the parameter for each area in the table 702 is not set is selected. At that time, threshold parameters set for the entire wafer are temporarily set in the A3 (s), A3 (k), and A3 (o) columns. When the delete button 407 is pressed, the die clicked on the map 403 until the next change of the selected state is displayed in colorless. If the die corresponding to a certain display color disappears due to the work, the parameter of the line is deleted, and the parameter is shifted so that the parameter is aligned so that the correspondence between the parameter and the display color does not change. The display color of the map 403 is changed.

  When the registration button 411 is pressed, in this method, the number of areas, (A3 (s), A3 (k), A3 (o)) for the number of areas and information on the dies included in the areas are stored in the inspection recipe storage unit 118a. Saved. The die information includes the number of dies included in the region and the die IDs corresponding to the number of dies. The die ID is information that can uniquely identify the die position, such as an (x, y) address of a matrix.

  FIG. 8 shows a fourth embodiment (A4) of a method for specifying a rectangular area on the die 21 of the GUI 401 for setting sensitivity for each area. This is a screen selected by the tab 402d on the sensitivity setting screen 401 for each region. The map 403 displays the die 21 and rectangular regions 801a to 801e in the die given by the die coordinate system. Table 802 shows the area number, the start point (xs, ys), the end point (xe, ye) of the rectangular area given by the die coordinate system, and the fourth area threshold parameters (A4 (s), A4 (k), A4). (o)), and regions 1 to 5 correspond to the rectangular regions 801a to 801e, respectively. Here, the start point is the die coordinate of the lower left corner, and the end point is the die coordinate of the upper right corner. When the inspection result overlay button 410 is pressed, the defect determined by the region-specific defect determination unit 111 based on the region-specific threshold parameters (A4 (s), A4 (k), A4 (o)). The position is expressed in die coordinates and displayed for all dies in an overlapping manner. The method for changing the rectangular area is the same as that for specifying the rectangular area on the wafer 11. The functions of the automatic button 408 and the return button 409 and the saved contents when the registration button 411 is pressed are also the same as when a rectangular area is designated on the wafer 11.

  The areas are set by the four types of methods described above, and the threshold parameters for each area (A1 (s), A1 (k), A1 (o)) to (A4 (s), A4 (k), The sensitivity can be partially lowered by setting A4 (o)). However, it is possible to reduce the sensitivity by, for example, significantly increasing the region-specific minimum threshold A (s) to 255 in the region 1 shown in FIG. It is also possible to make an inspection area.

  Next, a method for determining a defect in the defect determination unit 116 for each region based on the threshold parameter for each region set by the sensitivity setting unit 115 for each region will be described in detail with reference to FIG.

  That is, the area-specific defect determination unit 116 performs processing for each defect candidate determined by the defect determination unit 107 (S91). First, the defect determination unit 116 for each region determines the ID (x, y), wafer coordinates (x, y), die coordinates (i, j), and difference value ΔS () of the defect candidate determined by the defect determination unit 107. i, j), brightness variation σ (i, j), and threshold value Th (i, j) are read from the data storage unit 109 (S92).

  Next, the region-specific defect determination unit 116 checks the defect candidates determined by the defect determination unit 107 for each of the four types of region-specific sensitivity setting methods set by the region-specific sensitivity setting unit 115 (S93). In the first time, when the position coordinates (x, y) of the first defect candidate are set for the sensitivity by region by concentric circles on the wafer as in the first embodiment (A1) (S94), the coordinate conversion unit Based on the wafer coordinates (x, y) converted in 108, a radius region including defect candidates is examined, and first threshold parameters (A1 (s), A1 (k) corresponding to the radius region are obtained. ), A1 (o)) and the brightness variation σ (i, j) at the die coordinates (i, j), the first threshold value A1 (Th (x, y)) is calculated (S95). When the first threshold value for each region is larger than the threshold value Th (i, j) in the defect determination unit 107 (S96), the first threshold value for the first region is reset. Another threshold value is set (S97). When the newly reset first area-specific threshold value A1 (Th (x, y)) is larger than the absolute value of the difference value ΔS (i, j) of the defect candidate (S98), the first defect candidate is set. Is not a defect and is excluded from the defect candidates (S99).

  In the second time, when the position coordinates (x, y) of the first defect candidate is set for sensitivity by region by specifying a rectangle on the wafer in the second embodiment (A2) (S94), It is checked whether or not the wafer coordinates (x, y) of the defect candidate are included in the designated rectangular area. If they are included, the corresponding second area-specific threshold parameters (A2 (s), A2 ( k), A2 (o)), and the second threshold value A2 (Th (x, y)) for each region is calculated by the above equation (1) (S95). If the defect candidate wafer coordinates are included in a plurality of rectangular regions, a plurality of second region-specific threshold values are calculated using a corresponding set of second region-specific threshold parameters. And select the largest of them. If it is large as described above, the threshold value A2 (Th (i, j)) for each region is updated (S96, S97), and the updated threshold value for each region is the difference value ΔS (i , J) is larger than the absolute value (S98), it is determined that the first defect candidate is not a defect and is excluded from the defect candidates (S99).

  In the third time, when the position coordinates (x, y) of the first defect candidate are set for each area according to designation for each die as in the third embodiment (A3) (S94), the defect Using the third region threshold parameter (A3 (s), A3 (k), A3 (o)) corresponding to the ID of the candidate die, the third region threshold according to the above equation (1) A3 (Th (x, y)) is calculated (S95). If it is large as described above, the region-specific threshold value A3 (Th (x, y)) is updated (S96, S97), and the updated region-specific threshold value is larger than the absolute value of the difference value ΔS. In the case (S98), it is determined that the first defect candidate is not a defect and is excluded from the defect candidates (S99).

  In the fourth time, when the position coordinates (x, y) of the first defect candidate is set for sensitivity by region by specifying a rectangle in the die as in the fourth embodiment (A4) (S94), It is checked whether or not the coordinates are included in the designated rectangular area. If included, the corresponding fourth area-specific threshold parameters (A4 (s), A4 (k), A4 (o)) are set. Then, the fourth threshold value A4 (Th (x, y)) for each region is calculated by the above equation (1). The processing in the case of being included in a plurality of rectangular areas is also the same as the case where the sensitivity setting for each area is set by specifying the rectangle on the wafer (S95). If it is large as described above, the region-specific threshold A4 (Th (x, y)) is updated (S96, S97), and the updated region-specific threshold is larger than the absolute value of the difference value ΔS. In the case (S98), it is determined that the first defect candidate is not a defect and is excluded from the defect candidates (S99).

  As described above, the region-specific defect determination unit 116 first determines that the first defect candidate determined by the defect determination unit 107 is not a defect based on the four types of region-specific threshold values. The defect candidate is determined to be a defect, and the position coordinates of the first defect candidate and its size (area, projection length on each axis, etc.) are transmitted to the image cutout unit 112 (S101).

  Furthermore, the region-specific defect determination unit 116 repeats the processing of S91 to S99 over all the defect candidates determined as defects by the defect determination unit 107, and is determined not to be a defect based on the four types of region-specific threshold values. If not, the defect candidate is determined to be a defect, and the position coordinates of the defect candidate and its size (area, projection length on each axis, etc.) are transmitted to the image cutout unit 112 (S101). Then, the region-specific defect determination unit 116 ends when all defect candidates determined as defects by the defect determination unit 107 are processed (S100).

  Next, in the appearance inspection method and apparatus according to the present invention, an area-specific sensitivity setting method (A4) by specifying a rectangular area in the die will be described. Basically, a region with many false alarms is estimated based on defect information obtained by preliminary inspection, and a fourth region-specific threshold parameter is set so that non-inspection or sensitivity is low. As a method for estimating a region with many false reports, the following five methods can be considered.

  Next, the flow of the first method will be described with reference to FIG. First, an inspection image f (i, j) for at least one die is acquired in the image storage unit 115 while performing a preliminary inspection (S111). The preliminary inspection method is almost the same as the above-described defect detection method. The image acquired by the image sensor 104 while the semiconductor wafer 11 is moved by the stage 12 is used as an input, and the defect is detected via the AD conversion unit 105 and the preprocessing unit 106. The determination unit 107 performs defect determination and stores defect candidate position information in the data storage unit 109. At the same time, the image storage unit 115 stores an image for one die (background information representing a background image such as an inspection image). The die for storing images is designated in advance, or is the first die or the die closest to the center of the inspection area. As inspection recipes, product type, process name, wafer size, inspection direction, die size, die matrix information, threshold parameters (s, k, o) set for the entire wafer, irradiation conditions of the irradiation optical system, and detection optics Use system detection conditions already set. However, the inspection range may be the entire wafer or only one or several rows in the stage scan direction. Next, the area-specific sensitivity setting section 110 displays an area-specific sensitivity setting screen 401, and selects a method by specifying a rectangular area in the die shown in FIG. Therefore, the inspection result superimposing button 410 is pressed to display the preliminary inspection result superimposed on the map 403. As shown in FIG. 11A, the acquired inspection image and defect candidates determined by the defect determining unit 107 are determined. The position information is displayed in an overlapping manner (S112). Next, as shown in FIG. 11 (b), a rectangular area is set in the portion that is estimated to be false information, and the fourth area in Table 802 is set so that the sensitivity of the corresponding area becomes low or becomes a non-inspection area. The threshold parameters (A4 (s), A4 (k), A4 (o)) are adjusted (S113). This is a manual operation. Then, by pressing the registration button 411, the sensitivity setting information for each rectangular area in the die is output to the inspection recipe stored in the inspection recipe storage unit 118a, and the process ends (S114). According to the present embodiment, since the position information of defect candidates and the inspection image (background information representing the background image) are displayed in an overlapping manner in the sensitivity setting unit 110 for each region, it is possible to easily estimate the false information. For example, since it is considered that it is false information empirically when it occurs side by side in the same background, it is only necessary to set a rectangular area in such a portion so that the sensitivity is lowered. Even if no defect has occurred, the rectangular area can be set to the place where the same background continues, and more precise sensitivity setting is possible.

  It should be noted that the image to be displayed superimposed on the position information of the defect candidate is not an inspection image as long as the background information, that is, the pattern density (pattern edge density (pattern edge distance)) or brightness distribution is known. Good. For example, the threshold Th (i, j) calculated by the defect determination unit 107 can be stored for one die. Also, pixel merging based on the inspection image (average of multiple pixels is made to be one pixel) or processing time cannot be secured, sampling and reducing the data amount of the image and storing it, or A method of creating an image representing the distribution of the pattern density of the die from the layout data is also conceivable.

  Next, the flow of the second method will be described with reference to FIG. In the second method, the preliminary inspection is performed twice while changing the sensitivity. First, in the first preliminary inspection, the defect determination unit 107 performs defect determination of the defect candidate using the high-sensitivity fourth area threshold parameter, and stores the position information of the defect candidate in the data storage unit 109. Further, an inspection image for at least one die is acquired and stored in the image storage unit 115 (S121). Next, the second preliminary inspection is performed using the low-sensitivity fourth threshold parameter for each region, the position information of the defect candidates is stored in the data storage unit 109, and an inspection image for at least one die is acquired. The image is saved in the image saving unit 115 (S122). In the second time, the wafer may be actually scanned, and the fourth region-specific threshold value is set using the second region-specific threshold parameter based on the defect variation σ detected in the first time. It may be recalculated and judged by comparing with the difference value ΔS of the defect. Next, the sensitivity setting unit for each area 110 displays the position information of the defect candidates detected by the two preliminary inspections and the inspection image in an overlapping manner (S123). At this time, the defect candidate detected for the first time but not detected for the second time and the defect candidate detected for the second time are displayed in different colors or marks. Based on the displayed information, a false alarm generation area is manually estimated, a rectangular area is set in that area, and a fourth threshold parameter for each area is set so that the sensitivity is lowered by increasing the threshold for each area. Is adjusted (S124). Then, by pressing the registration button 411, the adjusted inspection recipe is stored in the inspection recipe storage unit 118a, and the process ends (S125). According to the present embodiment, by performing the inspection twice while changing the sensitivity, the fact that the ratio of the false information included in the defect candidates that increase when the sensitivity is increased is larger than usual, the false information estimation is highly reliable. It is possible to do in degrees.

  Next, the flow of the third and fourth methods will be described with reference to FIG. In the third method, preliminary inspection is performed (S131), and the region-specific sensitivity setting unit 110 performs classification (repeated defect, dense defect, arc-shaped distribution defect) based on the position information of the defect candidate determined by the defect determination unit 107. , Grouped into radial distribution defects, linear distribution defects, ring / bulk distribution defects, random defects, etc.) (S132). Further, the area-specific sensitivity setting unit 110 selects a small number of defects for each group, for example, on the GUI screen, confirms whether the defect is a defect or a false report by review or visual inspection, and stores the defect in the data storage unit 109 (S133). ). In the case of review, a small number of selected defects are executed by a review device (not shown), and the result is stored in, for example, the data storage unit 109. Next, the sensitivity setting unit 110 for each area displays the position information of the defect candidates detected by the preliminary inspection. In step S133, the defect candidates of the group with many false reports and the group with many defects are displayed in different colors or marks. It is displayed (S134). Furthermore, the sensitivity setting unit 110 for each region sets a rectangular region in a place where it is estimated to be false information based on the displayed information, and increases the threshold value for each region so that the sensitivity is lowered. The threshold parameter is adjusted (S135). Then, by pressing the registration button 411, the adjusted inspection recipe is stored in the inspection recipe storage unit 118a, and the process ends (S136).

  In addition, as a classification method based on the position information of defect candidates in the sensitivity setting unit 110 for each region, for example, the defect disclosed in Patent Document 3 is based on the idea that defects occurring at substantially the same position on the die are false information. It is conceivable to group defects by connecting defects whose distance between them is equal to or less than a predetermined threshold value. Or, in the idea that the defects arranged in one row seem to be false information, they are sorted in the order of x coordinate or y coordinate, and n or more defects inside the width h using the distance parameter h and the number parameter n specified in advance. If there is a group, grouping is possible by connecting them sequentially.

  Here, when the sensitivity setting unit 110 for each region confirms whether the defect is a false report or not by visually checking the inspection image, the defect clicked on the map 403 is preferably displayed in an enlarged manner. Alternatively, the inspection image of the selected defect may be displayed as a list by group.

  In the fourth method, in the feature amount calculation unit 119 representing background information, the defect determination unit 107 performs defect determination while performing preliminary inspection, calculates a feature amount representing background information of the determined defect candidate, and stores the data. The information is stored in the unit 109 (S131). The background information is, for example, brightness, contrast, pattern edge density, and the like. The area-specific sensitivity setting unit 110 classifies similar items into the same group based on the feature amount representing the background information stored in the data storage unit 109 (S132). The subsequent processing is the same as in the third embodiment. As a classification method, a method in which learning is performed in advance using a test sample and classification is performed using some pattern recognition technique is conceivable. Further, a method of dividing the feature amount space by a threshold value along the feature amount axis and assigning one group to each can be considered. At that time, it may be divided equally using one feature amount axis, or a binarization threshold may be determined by discriminant analysis for each of two or three feature amount axes, and divided into four or eight portions. Moreover, you may combine the classification | category by the positional information on the defect candidate shown by the 3rd method.

  As described above, the region-specific sensitivity setting unit 110 classifies the defects using the position information and background information of the defect candidates determined by the defect determination unit 107, so that only a small number of defects are confirmed. The defect can be estimated with high reliability. Since it is not necessary to acquire images for all defect candidates determined by the defect determination unit 107 during preliminary inspection, the image processing unit requires less memory than the first and second methods. In addition, images of all defect candidates determined by the defect determination unit 107 at the time of preliminary inspection may be acquired in the data storage unit 109. In this case, it is further easier to estimate false reports and defects.

  Next, the flow of the fifth method will be described with reference to FIG. In the fifth method, an inspection image for at least one die is acquired in the image storage unit 115 while performing a preliminary inspection (S141). Next, the region-specific sensitivity setting unit 110 segments the region for one die stored in the image storage unit 115 based on the background information stored in the data storage unit 109 (S142). Then, the sensitivity setting unit 110 for each region calculates a defect density for each segmented region, and tests whether there is a significant difference between the regions (S143). Further, when there is a significant difference (S144), the sensitivity setting unit 110 for each region selects a small number of defects from the high-density region and confirms whether or not they are false alarms (S145). The confirmation method is the same as the third and fourth methods. If many false alarms are included (S146), a low-sensitivity fourth threshold parameter for each area set in advance to set the area as a non-inspection area or a sensitivity reduction area, or a false alarm group The fourth threshold parameter for each region is adjusted so that the defect density becomes lower than a preset allowable value (S147). By pressing the registration button 411, the adjusted sensitivity setting information for each rectangular area within the die is output as an inspection recipe to the inspection recipe storage unit 118a, and the process is terminated (S148). Here, the sensitivity may be set so that all high-density regions are non-inspection regions or sensitivity-reduced regions. In that case, step S145 for confirming whether or not it is a false alarm can be omitted, and sensitivity setting for each region can be performed without manual intervention.

  Note that step S113 in FIG. 10, step S124 in FIG. 12, and step S135 in FIG. 13 are manual rectangular area setting and parameter setting for each area. However, automatic setting may be performed by pressing the automatic button 408 here. Is possible. In this case, an example of processing executed in the sensitivity setting unit for each area 110 will be described with reference to FIGS. 15 and 16. FIG. 16 is an example showing a part of a defect image 1501 superimposed on an inspection image.

  When performing the preliminary inspection once, the area-specific sensitivity setting unit 110 first connects defects arranged in a line based on positional information of defect candidates, by connecting defects whose distance between defects is equal to or less than a predetermined threshold value. Grouping is detected (S151). For example, when there are n or more defects inside the width h using the distance parameter h and the number parameter n specified in advance, the detection method for detecting that they are arranged in a line is sorted in the order of x coordinate or y coordinate. Connect sequentially. When using the defect candidate background information (S152), it is confirmed whether the defect background in the group is similar for each defect group using the distance in the feature amount space or the like. Those not present are removed from the group (S153). Next, as shown in FIG. 16, the area-specific sensitivity setting unit 110 automatically sets a rectangular area 1502 surrounding all the defects in the group for each group having a sufficiently large number of defects (S154). When the background information of the defect candidate is used (S155), when a portion where the background is not similar is included in the rectangular area, it is removed from the rectangular area (S156). If it is divided into two or more, it is treated as a separate area thereafter. Next, the area-specific sensitivity setting unit 110 checks whether the background of the adjacent area for each rectangular area that has been set is similar to the background in the rectangle, and if so, widens the rectangular area (S157). . Here, the adjacent region is a small region 1503 that is in contact with the rectangular region 1502 in the longitudinal direction with the same width. In this embodiment, the same process is repeated until the background of the adjacent area is no longer similar to the background in the rectangle until the area 1504 indicated by the dotted line is expanded. Finally, the region-specific sensitivity setting unit 110 sets a fourth region-specific threshold parameter that adjusts the sensitivity for the set region (S158). As the fourth area-specific threshold parameter, a threshold parameter set for the entire wafer is set as a default. Alternatively, a predetermined low sensitivity threshold parameter may be set.

  When the preliminary inspection is performed twice as shown in FIG. 12, the region-specific sensitivity setting unit 110 performs the same process as described above only for defects that increase when the sensitivity is increased.

  Moreover, when the sensitivity setting part 110 according to area | region confirmed whether it is a false report or a defect about a few defects for every group as shown in FIG. 13, it is the same as the above for only the defect of the group which contains many false reports. Perform the process.

  By the method described above, the area automatically set by the area-specific sensitivity setting unit 110 is displayed on the map 403 on the area-specific sensitivity setting screen 401. Accordingly, it is possible to modify the area by an operation on the map 403, delete the area by the delete button 407, and add an area by the new button 406.

  Here, the sensitivity setting unit 110 for each area sets the area by pressing the automatic button 408, confirms and corrects it, and then presses the registration button 411 to output the recipe. Based on the background information and position information of defect candidates obtained based on the result, a false information group is estimated, a rectangular area surrounding the defects in the group is set for each group, and a predetermined low sensitivity fourth area Another threshold parameter may be set. Further, the fourth area threshold parameter may be adjusted so that the defect density of the false alarm group is lower than a preset allowable value. In this way, it is possible to set the sensitivity for each region with high reliability without any manual intervention.

  The above description has been made on the premise that preliminary inspection is performed, but normal inspection is performed by applying threshold parameters (s, k, o) set on the entire wafer, and the sensitivity setting unit 110 for each region The false information group is estimated from the background information and position information of the defect detected by the defect determination unit 107, and a rectangular area surrounding the defect in the group is set for each group, and the fourth area is classified so as to have low sensitivity. Even if the threshold parameter is set and the fourth threshold value parameter set for each region is applied to the defect detected by the defect determination unit 107, the determination by the region defect determination unit 108 performs the determination again. good. In this way, even when the sensitivity for each region is not set in the inspection recipe, it is possible to inspect with high sensitivity while suppressing false information.

  Next, in the appearance inspection method and apparatus according to the present invention, the region-specific sensitivity setting method (A1) (A2) by concentric circle region designation or rectangular region designation on the wafer in the region-specific sensitivity setting unit 110 will be described with reference to FIGS. Will be described. First, a preliminary inspection is performed over the entire surface of the wafer (S181). Next, as shown in FIG. 17, the sensitivity setting unit 110 for each region calculates a defect density for each region divided by concentric circles or square lattices (S182). Then, the sensitivity setting unit 110 for each region tests whether or not there is a significant difference between the regions by a method such as a chi-square test. If there is a significant difference (S183), a small number of defects are detected from the high-density region. Is selected and it is confirmed whether or not they are false reports (S184). If many false alarms are included (S185), is the first low-sensitivity first or second threshold parameter for each area set in advance so that the area is set as a non-inspection area or a sensitivity reduction area? The first or second region-specific threshold parameter is adjusted so that the defect density of the false alarm group is lower than a preset allowable value (S186). By pressing the registration button 411, the adjusted sensitivity setting information for each concentric region or rectangular region in the wafer is output to the inspection recipe storage unit 118a as an inspection recipe, and the process is terminated (S187). Here, the sensitivity may be set so that all high-density regions are non-inspection regions or sensitivity-reduced regions. In this case, step S184 for confirming whether the information is false or not can be omitted, and sensitivity setting for each region can be performed without manual intervention.

[Second Embodiment]
The appearance inspection method and apparatus according to the present invention are not limited to the appearance inspection apparatus described in the first embodiment.

  For example, even when the detection system 13 is a dark field type or SEM type, it is possible to set sensitivity for each region as a similar configuration.

  A second embodiment according to the present invention will be described with reference to FIG. FIG. 19 is a diagram showing a configuration in the case where the detection system 13 is a dark field optical system that irradiates light obliquely and detects scattered light from the inspection object 11 upward. The light emitted from the laser light source 1801 is formed into a slit shape by a beam expanding optical system composed of a concave lens 1802 and a convex lens 1803, a curved lens 1804 that approximates a cone, and a mirror 1805, and the inspection object 11 that is a wafer from an oblique direction. Irradiate. The reason for making the irradiation light slit is to inspect at high speed. A detection optical system that detects scattered light on the surface of the inspection object 11 includes a Fourier transform lens 1806, a spatial filter 1807, an inverse Fourier transform lens 1808, an ND filter 1809, an optical filter 1810 such as a polarizing plate, and the image sensor 104. The A spatial filter 1807 is placed on the Fourier transform plane and shields diffracted light from a repetitive pattern on the inspection object. On the other hand, scattered light from the defect spreads irregularly on the Fourier transform plane, so that most of the scattered light is received by the image sensor 104 without being shielded. Therefore, the S / N is improved, and it becomes possible to detect defects with high sensitivity. A signal detected by the image sensor 104 becomes an input to the image processing unit 14, and defect detection is performed by the same processing method and processing unit as in the first embodiment. The sensitivity setting method for each region is the same as that in the first embodiment.

[Third Embodiment]
Even when the defect determination method is different, the sensitivity setting method for each region according to the present invention is applicable. A third embodiment according to the present invention will be described with reference to FIG. FIG. 20 is a diagram illustrating a second example 107 b of the defect determination unit 107. The defect determination unit 107b includes an image signal of the inspection target die output from the preprocessing unit 106, that is, a detected image signal f (i, j), and an image signal of the previous die of the inspection target die, that is, a reference image signal g ( i, j) are input as a set. Let (i, j) be in-die coordinates. The reference image signal g (i, j) is obtained by being delayed by the delay memory 301 for the time that the stage moves by one die. The misregistration detection unit 302 calculates the misregistration amount due to stage vibration between two images that are continuously input. At this time, the detected image signal f (i, j) and the reference image signal g (i, j) are continuously input, but the calculation of the positional deviation amount is performed with a specific length as one processing unit. To sequentially. The brightness correction unit 310 performs the alignment of the brightness after aligning the detected image and the reference image using the calculated displacement amount. The method will be briefly described. First, the image is decomposed into several categories based on the feature amount of the pixel of the reference image. Among them, those having a large number of pixels are regarded as normal categories. For each decomposed category, a scatter diagram is created by plotting the reference image signal g on the horizontal axis and the detected image signal f on the vertical axis, and a correction coefficient for adjusting the brightness of the reference image to the detected image is calculated. Correct the reference image signal using the correction coefficient calculated for the normal category and the correction coefficient calculated for the normal category with a close distance in the feature space for the other categories, and adjust the brightness to the detected image. . The difference calculation unit 303 calculates a difference image ΔS (i, j) between the detected image f (i, j) and the reference image g (i, j) after alignment and brightness correction. On the other hand, the threshold value calculation unit 311 calculates the threshold value Th so that the normal part is not determined to be defective based on the scatter diagram after brightness correction of all normal categories. For example, a method is conceivable in which the maximum value of the absolute value of the difference between the detected image and the reference image is calculated, or an envelope that wraps the scatter diagram is calculated. The comparison unit 309 detects a pixel whose absolute value of the difference value ΔS is larger than the threshold value Th as a defect candidate. In the present embodiment, by adjusting the brightness of the detected image and the reference image, the difference value of the normal portion can be kept small, so that the threshold value can be set low and highly sensitive defect detection is possible. It becomes. Therefore, even if the threshold value calculation unit 311 is not provided and the threshold value is constant over the entire wafer, the effect of increasing the sensitivity can be obtained, and in this case, the calculation amount can be suppressed.

  After the defect determination by the defect determination unit 107b, the position of the detected defect candidate is converted into wafer coordinates (x, y) and die coordinates (i, j) by the coordinate conversion unit 108 and stored in the data storage unit 109. In addition, the difference value ΔS (i, j) and the threshold value Th are stored in the data storage unit 109 together with the ID of the die including the defect candidate. Further, the data storage unit 109 stores the detection image, the reference image, and the threshold image in a predetermined size with the position of the defect candidate determined by the defect determination unit 107 as the center.

  Next, the region-specific defect determination unit 108 recalculates the region-specific threshold A (Th) according to the information set in advance by the region-specific sensitivity setting unit 115, and the stored difference value is larger. Is detected as a defect.

  Further, the image cutout unit 109 cuts out the detected image and the reference image stored in the data storage unit 109 around the position of the defect detected by the region-specific defect determination unit 108, and the feature extraction unit 110 extracts the detected image from the detected image. A feature amount for defect classification is calculated. The defect classification unit 111 performs classification using preset conditions and outputs class information of each defect.

  The sensitivity setting method for each region in the third embodiment is almost the same as that in the first embodiment. However, the sensitivity adjustment is not performed using the threshold parameters (s, k, o), but using the region-specific minimum threshold S and the region-specific coefficient K (1 or more). That is, the larger one of the region-specific minimum threshold value S and the constant multiple (Th × K) of the threshold value Th stored in the data storage unit 109 is set as a new region-specific threshold value. By setting the minimum threshold value S to the maximum value of the pixel value, it is possible to make the corresponding region non-inspected.

[Fourth Embodiment]
In addition, the appearance inspection by comparing wafers on which patterns having the same shape have been formed has been described as an example, but the region-based sensitivity setting method using concentric circles on the wafer according to the present invention is suitable for inspection of unpatterned wafers. Is also applicable. That is, a fourth embodiment according to the present invention will be described with reference to FIG.

  FIG. 21 is a diagram showing an example of the configuration of an appearance inspection apparatus for a patternless wafer. That is, the semiconductor wafer 11 that is the object to be inspected is held on the rotary stage 2001. The illumination optical system 2002 includes a laser light source 2003, a beam expander 2004, mirrors 2005 and 2006, and a condenser lens 2007. The detection optical system 2008 includes condensing lenses 2009a and 2009b and photoelectric conversion elements 2010a and 2010b. The signal processing system 2011 includes threshold processing units 2012a and 2012b, a defect determination unit 2013, and a threshold setting unit 2014 for each region.

  Next, an inspection method using the inspection apparatus shown in FIG. 21 will be described. The laser beam emitted from the laser light source 2003 is expanded by a beam expander 2004, and then the beam diameter is reduced by a condenser lens 2007 via mirrors 2005 and 2006, and irradiated onto the wafer 11. If a defect such as a foreign substance exists at the beam irradiation position on the surface of the wafer 11, strong scattered light is generated. The scattered light is condensed on the light receiving surface of the photoelectric conversion element 2010a by the condensing lens 2009a, and photoelectrically converted by the photoelectric conversion element 2010a. Scattered light is similarly detected by the condenser lens 2009b and the photoelectric conversion element 2010b arranged at different angles. Since the direction in which strong scattered light is generated differs depending on the size and shape of the defect, sensitivity can be improved by detecting from a plurality of angles in this way. Further, it is better to increase the number of detection systems and cover all directions.

  The outputs of the photoelectric conversion elements 2009a and 2009b are compared with threshold values in the threshold processing units 2012a and 2012b, respectively, and when larger than the threshold values, they are detected as defects. Both detection signals are input to the defect determination unit 2013, and unevenness determination and defect size determination are performed based on the magnitude and ratio of each signal. In order to inspect the entire surface of the wafer 11, it is necessary to scan the irradiation position of the laser beam relative to the wafer 11. By rotating the wafer 11 linearly so as to approach the irradiation position while rotating the wafer 11 by the rotation stage 2001, spiral scanning can be performed. Although not shown, when the scanning of the laser beam in the radial direction of the wafer 11 is combined, it is possible to move straightly by the scanning width of the laser beam during one rotation, so that the time for inspecting the entire wafer surface can be shortened. Instead of scanning the laser beam, a slit-like beam that is long in the radial direction of the wafer 11 may be formed and irradiated using a cylindrical lens or a curved lens that approximates a cone, although not shown.

  The threshold values used in the threshold processing units 2012a and 2012b are set in advance for each region by the region-specific threshold setting unit 2014. In the case of a non-patterned wafer, there is no factor that causes the signal intensity to fluctuate in the x direction and the y direction. Therefore, the region setting method using concentric circles shown in FIG. 4 may be applied.

  The present invention is applicable to an appearance inspection apparatus for thin film devices such as semiconductor wafers, TFTs, and photomasks.

It is a schematic structure figure showing a 1st embodiment of an appearance inspection device concerning the present invention. 1 is a plan view of a semiconductor wafer to be inspected according to the present invention. It is a figure for demonstrating one Example of the defect determination processing method which concerns on this invention. It is explanatory drawing of the 1st Example (method to divide | segment the area | region on a wafer with a concentric circle) of the sensitivity setting GUI classified by area | region in the external appearance inspection apparatus which concerns on this invention. FIG. 5 is a diagram showing an embodiment in which position information of defect candidates is displayed in a concentric circle area in the first embodiment shown in FIG. 4. It is explanatory drawing of the 2nd Example (method by the rectangular area designation | designated on a wafer) of the sensitivity setting GUI classified by area | region in the external appearance inspection apparatus which concerns on this invention. It is explanatory drawing of the 3rd Example (method by die specification on a wafer) of the sensitivity setting GUI classified by area | region in the external appearance inspection apparatus which concerns on this invention. It is explanatory drawing of the 4th Example (method by the rectangular area specification on die | dye) of the sensitivity setting GUI classified by area | region in the external appearance inspection apparatus which concerns on this invention. It is a flowchart which shows one Example of the defect determination according to area | region in the defect determination part according to area | region based on the threshold parameter according to area | region set in the sensitivity setting part according to area | region which concerns on this invention. It is a flowchart which shows the 1st method of the sensitivity setting method according to rectangular area | region according to this invention. It is a figure which shows the Example which accumulated and displayed the inspection image on the die | dye which concerns on this invention, the positional information on a defect candidate, and the rectangular area to set. It is a flowchart which shows the 2nd method of the sensitivity setting method according to the rectangular area | region according to this invention. It is a flowchart which shows the 3rd and 4th method of the sensitivity setting method according to the rectangular area | region according to this invention. It is a flowchart which shows the 5th method of the sensitivity setting method according to rectangular area | region according to this invention. It is a flowchart for demonstrating one Example of the specific process of the sensitivity setting according to the rectangular area | region in the die in the 1st thru | or 4th method of the sensitivity setting method according to the rectangular area | region which concerns on this invention. It is explanatory drawing of the method of setting a rectangular area automatically using the background information which concerns on this invention. It is a figure which shows the Example which divides the area | region of the wafer which concerns on this invention by a concentric circle, and calculates the density of a defect candidate. It is a flowchart which shows one Example of the sensitivity setting method according to the concentric in-wafer or rectangular area based on this invention. It is a figure which shows the detection system which is 2nd Embodiment of the external appearance inspection apparatus which concerns on this invention. It is a figure which shows the defect determination part which is 3rd Embodiment of the external appearance inspection apparatus which concerns on this invention. It is a schematic block diagram which shows 4th Embodiment of the external appearance inspection apparatus which concerns on this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 11 ... Inspection object (inspection board | substrate, semiconductor wafer), 12 ... Stage, 13 ... Detection part, 14 ... Image processing part, 15 ... Control part, 101 ... Light source, 102 ... Condensing optical system, 103 ... Objective lens, DESCRIPTION OF SYMBOLS 104 ... Image sensor, 105 ... AD conversion part, 106 ... Pre-processing part, 107, 107a, 107b ... Defect determination part, 108 ... Coordinate conversion part, 109 ... Data storage part, 110 ... Sensitivity setting part according to area, 111 ... Area Another defect determination unit, 112 ... Image extraction unit, 113 ... Feature extraction unit, 114 ... Defect classification unit, 116 ... Inspection recipe input unit, 117 ... User interface unit, 118 ... Storage device, 118a ... Inspection recipe storage unit, 118b ... information storage unit of detected defect, 119 ... feature amount calculation unit representing background information, 120 ... mechanical controller, 301 ... delay memory, 302 ... misregistration detection unit, 03 ... Difference calculation unit, 304 ... Data storage unit, 305 ... Maximum / minimum removal unit, 306 ... Sum of squares calculation unit, 307 ... Data count unit, 308 ... Threshold calculation unit, 309 ... Comparison unit, 310 ... Brightness Correction unit, 311... Threshold value calculation unit, 401... Sensitivity setting GUI for each region, 403... Map, 402 a to 402 d ... tab, 405, 602, 702, 802 ... table, 1801 ... laser light source, 1802 ... concave lens, 1803 DESCRIPTION OF SYMBOLS ... Convex lens, 1804 ... Conical curved lens, 1805 ... Mirror, 1806 ... Fourier transform lens, 1807 ... Spatial filter, 1808 ... Inverse Fourier transform lens, 1809 ... ND filter, 1810 ... Optical filter, 2001 ... Rotation stage, 2002 ... Illumination optics System 2003 ... Laser light source 2004 ... Beam expander 2005, 20 6 ... Mirror, 2007 ... Condensing lens, 2008 ... Detection optical system, 2009a, b ... Condensing lens, 2010a, b ... Photoelectric conversion element, 2011a, b ... Signal processing system, 2012a, b ... Threshold processing unit, 2013 ... Defect determination unit, 2014 ... Threshold setting unit for each region.

Claims (16)

An irradiation optical system for irradiating light or an electron beam on the substrate to be inspected, a detection optical system for detecting light or electrons obtained from the irradiated substrate to be inspected to acquire an image signal, and the acquired image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus provided with a defect detection unit that detects a defect candidate by determining with a predetermined determination threshold for a signal,
Preliminary inspection process for obtaining position information and background information on a substrate to be inspected for defects detected from the defect detection unit in preliminary inspection using the appearance inspection apparatus;
An area-specific sensitivity setting that sets an area on the inspected substrate based on positional information and background information of defect candidates acquired in the preliminary inspection process, and sets sensitivity parameters for the set areas Process,
A defect inspection process for each area in which a defect is determined based on a sensitivity parameter for each area set in the sensitivity setting process for each area for a detection signal of a defect candidate detected by the defect detection unit. Method. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus equipped with a defect detection unit for detecting
Preliminary inspection for acquiring background image representing background information for at least one die acquired by the detection optical system in preliminary inspection using the appearance inspection apparatus and in-die coordinate information of defect candidates detected from the defect detection unit Process,
The background image acquired in the preliminary inspection process and the in-die coordinate information of the defect candidate are displayed in an overlapping manner, an area is set in the die based on the displayed information, and the set in-die Sensitivity setting process for each area that sets sensitivity parameters for each area,
A defect determination process for each area in which a defect is determined based on a sensitivity parameter for each area in the die set in the sensitivity setting process for each area for the difference image of the defect candidate detected by the defect detection unit. Appearance inspection method.   The appearance inspection method according to claim 2, wherein the background image acquired in the preliminary inspection process is the inspection image acquired by the detection optical system.   The background image acquired in the preliminary inspection process is obtained by performing pixel merging or sampling on the inspection image acquired by the detection optical system to reduce the data size. Appearance inspection method.   The appearance inspection method according to claim 2, wherein the background image acquired in the preliminary inspection process is a determination threshold image used in the defect detection unit.   3. The appearance inspection method according to claim 2, wherein the background image acquired in the preliminary inspection process is an image representing a pattern density distribution calculated based on layout data in advance. The preliminary inspection process includes a process of acquiring position information and background information on a substrate to be inspected for defect candidates detected from the defect detection unit,
3. The area-specific sensitivity setting process sets an area in the die based on position information and background information of defect candidates obtained in the preliminary inspection process on a substrate to be inspected. Appearance inspection method The sensitivity setting process for each region is as follows:
The defect candidates acquired in the preliminary inspection process are classified based on the in-die coordinate information, and defect candidates not similar to the background information among the defect candidates in the group are classified for each classified group. Classification process to remove from the group,
For each group classified in the classification process, a rectangular area surrounding the defect candidates in the group is set, a portion where the background information is not similar in the set rectangular area is excluded from the rectangular area, and further background information A rectangular area setting process for integrating adjacent areas similar to the background information in the rectangular area into the rectangular area;
The visual inspection method according to claim 7, further comprising a low sensitivity setting step of setting a predetermined low sensitivity parameter in the rectangular region set in the region setting step. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus equipped with a defect detection unit for detecting
A high-sensitivity preliminary inspection process for detecting defect candidates based on a high-sensitivity determination threshold value from the defect detection unit in the high-sensitivity preliminary inspection using the appearance inspection apparatus and obtaining in-die coordinate information of the defect candidates;
A low-sensitivity preliminary inspection process for detecting defect candidates based on a low-sensitivity determination threshold value from the defect detection unit in the low-sensitivity preliminary inspection using the appearance inspection apparatus and obtaining in-die coordinate information of the defect candidates;
A background image acquisition process for acquiring a background image representing background information for at least one die from the detection optical system in the preliminary inspection using the appearance inspection apparatus;
Overlay the background image acquired in the background image acquisition process, in-die coordinate information of defect candidates acquired only in the high sensitivity preliminary inspection process, and defect candidates acquired in both the high sensitivity and low sensitivity preliminary inspection processes Display in such a manner that it can be distinguished from the in-die coordinate information, set a region based on the displayed information, and set a sensitivity parameter for each region,
A defect determination process for each area in which a defect is determined based on a sensitivity parameter for each area in the die set in the sensitivity setting process for each area for the difference image of the defect candidate detected by the defect detection unit. Appearance inspection method. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus equipped with a defect detection unit for detecting
A preliminary inspection process for acquiring position information of defect candidates from the defect detection unit in the preliminary inspection using the appearance inspection apparatus;
The defect candidates are classified based on the position information of the defect candidates acquired in the preliminary inspection process, and a small number of defect candidates are selected for the group having a large number of defect candidates from the classified groups to confirm whether it is a false report or a defect. In addition, the in-die coordinate information of the defect candidates of the group including a lot of false information as a result of the confirmation is displayed so as to be distinguishable from the in-die coordinate information of the other defect candidates, and the area is determined based on the displayed information. Set sensitivity parameters for each area, and set sensitivity parameters for each area,
And an area-specific defect determination process for determining a defect based on an area-specific sensitivity parameter set in the area-specific sensitivity setting process for the difference image of the defect candidate detected by the defect detection unit. Inspection method. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus equipped with a defect detection unit for detecting
A preliminary inspection process for calculating a feature amount representing background information of a defect candidate detected by the defect detection unit in the preliminary inspection using the appearance inspection apparatus or both the feature amount and position information of the defect candidate;
The defect candidate is classified based on the feature amount representing the background information of the defect candidate calculated in the preliminary inspection process, or both of the feature amount representing the background information and the position information of the defect candidate, and the defect among the classified groups Select a small number of defects for a large number of groups to check whether they are false or flawed, and as a result of the confirmation, distinguish the in-die coordinate information of the defect in the group that contains a lot of false information and the in-die coordinate information of other defects Display as possible, set a region based on the displayed information, and set a sensitivity parameter for each region,
And an area-specific defect determination process for determining a defect based on an area-specific sensitivity parameter set in the area-specific sensitivity setting process for the difference image of the defect candidate detected by the defect detection unit. Inspection method. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus equipped with a defect detection unit for detecting
A background image acquisition process for acquiring a background image representing background information for at least one die from the detection optical system in the preliminary inspection using the appearance inspection apparatus;
The background image acquired in the background image acquisition process is divided into regions based on the characteristics of the background image, and the density of defect candidates detected by the defect detection unit in the preliminary inspection is calculated for each of the divided regions. , Whether or not there is a significant difference between the regions in the calculated defect candidate density, and if there is a significant difference as a result of the test, select a small number of defects for a high density of defect candidates Confirming whether it is a false report or a defect, and a sensitivity setting process for each area that sets a sensitivity parameter so as to be non-inspected or low-sensitivity for an area that contains many false reports as a result of the confirmation,
And an area-specific defect determination process for determining a defect based on an area-specific sensitivity parameter set in the area-specific sensitivity setting process for the difference image of the defect candidate detected by the defect detection unit. Inspection method. An irradiation optical system for irradiating light or an electron beam on the substrate to be inspected, a detection optical system for detecting light or electrons obtained from the irradiated substrate to be inspected to acquire an image signal, and the acquired image A visual inspection method for inspecting a defect on a substrate to be inspected using a visual inspection apparatus provided with a defect detection unit that detects a defect candidate by determining with a predetermined determination threshold for a signal,
A preliminary inspection process for detecting a defect candidate from the defect detection unit by performing a preliminary inspection using the appearance inspection device;
The substrate to be inspected is divided into regions by concentric circles or square lattices, and the density of defect candidates detected in the preliminary inspection process is calculated for each of the divided regions. Test whether there is a significant difference, and if there is a significant difference as a result of the test, select a small number of defects for areas with high defect density to check whether they are false or defective. Sensitivity setting process for each region that sets the sensitivity parameter so that it is non-inspected or low sensitivity for
A defect inspection process for each area in which a defect is determined based on a sensitivity parameter for each area set in the sensitivity setting process for each area for a detection signal of a defect candidate detected by the defect detection unit. Method. An irradiation optical system for irradiating light or an electron beam on the substrate to be inspected, a detection optical system for detecting light or electrons obtained from the irradiated substrate to be inspected to acquire an image signal, and the acquired image A visual inspection apparatus including a defect detection unit that detects a defect candidate by determining with a predetermined determination threshold for a signal,
Furthermore, acquisition means for acquiring position information and background information on the inspection target substrate of defect candidates detected from the defect detection unit in preliminary inspection;
An area-specific sensitivity setting unit that sets an area on the inspected substrate based on positional information and background information of defect candidates acquired by the acquisition means, and sets sensitivity parameters for each set area When,
And an area-specific defect determination unit that determines a defect based on a region-specific sensitivity parameter set by the region-specific sensitivity setting unit with respect to a defect candidate detection signal detected by the defect detection unit. Inspection device. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection apparatus comprising a defect detection unit for detecting
Further, an acquisition means for acquiring a background image representing background information for at least one die acquired by the detection optical system in preliminary inspection and in-die coordinate information of a defect candidate detected from the defect detection unit;
The background image acquired by the acquisition unit and the in-die coordinate information of the defect candidate are displayed in an overlapping manner, and an area is set in the die based on the displayed information. A sensitivity setting section for each area that sets sensitivity parameters for each area,
An area-specific defect determination unit that determines a defect based on the area-specific sensitivity parameter in the die set by the area-specific sensitivity setting unit with respect to the difference image of the defect candidate detected by the defect detection unit. A visual inspection device. Irradiation optical system for irradiating each of a plurality of dies arranged on a substrate to be inspected with light or an electron beam, and light or electrons obtained from each of the plurality of dies arranged on the substrate to be inspected. A detection optical system that detects and acquires a comparison target inspection image and a reference image, and a defect candidate using a determination threshold for a difference image between the acquired comparison target inspection image and a reference image A visual inspection apparatus comprising a defect detection unit for detecting
Further, the substrate to be inspected is divided into regions by concentric circles or square lattices, a preliminary inspection is performed for each of the divided regions, and the density of defect candidates detected from the defect detection unit is calculated, and is calculated for each region. A sensitivity setting unit for each area that sets sensitivity parameters for each area according to the density of the defect candidates
And an area-specific defect determination unit that determines a defect based on a region-specific sensitivity parameter set by the region-specific sensitivity setting unit with respect to a defect candidate detection signal detected by the defect detection unit. Inspection device.

Images