JP2014093429A - Semiconductor inspection apparatus and semiconductor inspection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a semiconductor inspection apparatus which can precisely determine an uneven pattern of a semiconductor.SOLUTION: A semiconductor to be inspected is imaged by scanning charged particles to acquire an image of the semiconductor. An inspection profile is generated from the inspection image (1551), a reference profile is generated from a reference image (1541), uneven pattern information which indicates an uneven pattern of a reference portion is acquired from design information of the semiconductor to be inspected, the inspection profile and the reference profile are compared with each other (1552), and the uneven pattern of an inspection portion of the semiconductor is determined by referring to the uneven pattern information (1553).

Description

  The present invention relates to a semiconductor inspection apparatus and a semiconductor inspection method.

  Japanese Patent Application Laid-Open No. 2004-251694 (Patent Document 1) describes the characteristics of the profile waveform of an image of a concavo-convex pattern formed on a sample, for example, whether the slope of one skirt has a steep or gentle slope. Describes a method for determining irregularities from the above.

Japanese Patent Laid-Open No. 2004-251684

  The technique described in Patent Document 1 is based on the premise that the features of unevenness and profile shape are uniquely determined. However, in an actual sample, the features of the unevenness and the profile shape may not be uniquely determined depending on the material, the inclination of the unevenness, and the like, and there is a possibility that the unevenness cannot be accurately determined by the technique described in Patent Document 1.

  In order to solve the above-described problems, a semiconductor inspection apparatus according to the present invention scans a semiconductor to be inspected by scanning charged particles and acquires an image of the semiconductor, and a semiconductor from the image acquired by the image acquisition unit. And an arithmetic unit for determining the uneven pattern.

  The calculation unit acquires the inspection image of the semiconductor inspection region imaged by the image acquisition unit, generates an inspection profile from the inspection image, acquires the reference image of the semiconductor reference region imaged by the image acquisition unit, and references Generates a reference profile from the image, obtains uneven pattern information indicating the uneven pattern of the reference site from the design information of the semiconductor to be inspected, compares the inspection profile with the reference profile, and inspects the semiconductor by referring to the uneven pattern information The uneven pattern of the part is determined.

  According to the present invention, it is possible to reduce misjudgment during unevenness judgment.

The figure which shows the relationship between a positioning image, a reference image, and a test | inspection image. 1 is an overall configuration diagram of an embodiment of the present invention. FIG. 3 is a diagram illustrating profiles of a reference image and an inspection image according to the first embodiment. FIG. 3 is a diagram illustrating an example of a template according to the first embodiment. FIG. 3 is a diagram illustrating profile matching according to the first embodiment. FIG. 6 is a diagram illustrating a position of a reference image corresponding to a template according to the first embodiment. The figure in the case of using a some template by the change of Example 1. FIG. The figure at the time of selecting the unsuitable template by the change of Example 1. FIG. The figure at the time of selecting the unsuitable convex part template by the change of Example 1. FIG. The figure of the difference template reversed horizontally to the peak center by the change of Example 1. FIG. FIG. 6 is a diagram illustrating a place where an image feature amount is output in the modification of the first embodiment. The figure in the case of making a reference image into another position. The figure which shows how to select various places of a reference image. FIG. 3 is a flowchart showing an example of a diagram template showing the overall procedure of the first embodiment. FIG. 3 is a flowchart showing in detail a part of the procedure of the first embodiment. The flowchart which showed the one part procedure in detail by the change of Example 1. FIG. The flowchart which showed the one part procedure in detail by the change of Example 1. FIG. FIG. 5 is a flowchart showing a part of the procedure of Example 2 in detail. FIG. 6 is a diagram illustrating a profile and a template according to the second embodiment. The figure which shows a some template by the change of Example 2. FIG.

(1) Example 1
(1.1) Configuration FIG. 2 is a diagram showing a semiconductor inspection apparatus using a scanning electron microscope.

  The scanning electron microscope includes a stage 201, an electron gun 203, a secondary electron detection unit 204, and an imaging unit 205. The semiconductor 202 on which the resist is placed is an object to be inspected and is transported to the photographing field by the stage 201. The semiconductor 202 is manufactured based on design data, and a concavo-convex pattern is formed on the semiconductor surface by lithography. Electrons are launched by the electron gun 203, a predetermined area of the semiconductor 202 is scanned, the secondary electrons that come out are captured by the secondary electron detection unit 204, and image data of the captured image is generated by the imaging unit 205.

  The imaging unit 205 sends a control signal for operating the stage 201, the electron gun 203, and the secondary electron detection unit 204 to obtain a captured image, and appropriately positions the signals detected by the secondary electron detection unit 204 in order. To image data. That is, the imaging unit 205 is an image acquisition unit, and acquires an image of a semiconductor imaged by scanning a semiconductor to be inspected with charged particles.

  The calculation device 206 receives the image data of the captured image created by the imaging unit 205, performs predetermined image processing, and causes the display device 207 to display the result. In other words, the calculation device 206 serves as a calculation unit, and determines the semiconductor uneven pattern from the image acquired by the image acquisition unit. More specifically, the calculation device 206 includes a circuit such as a CPU (not shown), which performs a necessary calculation as a calculation unit. Note that the calculation device 206 may be built in the scanning electron microscope or may be externally attached.

  Although a scanning electron microscope using an electron gun will be described here as an example, scanning is performed with charged charged particles such as charged helium and argon, and an image is captured based on the charged particles emitted from the scanned portion. The present embodiment can also be applied to various other inspection apparatuses such as an inspection apparatus that performs the above inspection.

  In this embodiment, the semiconductor 202 with a resist is described as an example. However, the present invention can be applied to the case of inspecting a sample including a concavo-convex pattern created by other methods such as direct processing using an electron beam.

(1-2) Overview of Overall Procedure FIG. 14 shows an overview of the overall procedure of the first embodiment. A description will be given with reference to FIG. 1 showing a positional relationship between images obtained by photographing.

  Hereinafter, the outline of the procedure will be described with reference to FIG.

  Step 1401: (Capture of positioning image). Before photographing a desired inspection region of the semiconductor 202, first, a region whose position is known is photographed at a low magnification. The obtained image is called a positioning image. In FIG. 1, the positioning image is shown as 101, and is an image obtained by photographing the semiconductor 202 at a low magnification.

  Step 1402: (Taking an inspection image). Aiming at a desired inspection site from the positioning image 101, imaging is performed at a high magnification. The obtained image is called an inspection image. In the example of FIG. 1, the inspection site is indicated as 102 and the inspection image is indicated as 112. The inspection region 102 is a region where inspection is desired. The calculation device 206 looks at the form from the positioning image 101 photographed at a low magnification and looks for an examination part, and the imaging unit 205 controls each part so as to photograph this part at a high magnification. An image can be obtained. The inspection image 112 is an image obtained by photographing the inspection region 102.

  Step 1403: (photographing a reference image). A part (reference part) in which the unevenness is uniquely determined from the design data, which is a part different from the inspection part 102, is imaged with the same imaging conditions (magnification, etc.) as the inspection image 112. The obtained image is called a reference image. In the example of FIG. 1, the reference portion is indicated as 103 and the reference image is indicated as 113. The reference part is an area in the positioning image 101 photographed at a low magnification that is clear from the design data whether it is a concave pattern or a convex pattern. The reference image 113 is an image obtained by photographing the reference region 103. The reference image includes a characteristic form that can be positioned in comparison with semiconductor design data and a typical form that can create a reference profile that can be compared with others. To do so, the reference site is selected. As a result, the reference image is an image whose position is known and unevenness is determined when compared with the design data. That is, the calculation unit detects a characteristic form of the reference image, grasps the position of the reference image, and creates a reference profile in which the unevenness information is known. FIG. 1 shows an example in which, in the reference image 113, the central region is a concave portion, and the adjacent left and right regions are convex portions.

  Step 1404: (Information acquisition of reference image). Image information is acquired from the reference image 113. The detailed procedure is described in section (1-3).

  Step 1405: (determination of unevenness). With reference to the image information obtained in step 1404, the unevenness of the imaging region of the inspection image 112 is determined. The detailed procedure is described in section (1-3).

  In step 1402, the inspection image 102 is obtained by looking at the form of the positioning image 101, searching for the inspection region 102, and aiming at this region at a high magnification. At this time, the positioning image 101 and the inspection image 112 have different shooting magnifications and image centers. Therefore, the target position cannot be accurately captured, and the inspection image 112 is captured with a slight deviation from the target position. It will be. Under such circumstances, when the image pattern of the inspection image 112 is periodic and does not have a characteristic form, the unevenness cannot be known from the image pattern when the shift of the photographing position becomes larger than a half cycle. That is, it cannot be accurately determined whether the pattern of each region of the inspection image 112 shown in FIG. 1 is a concave portion or a convex portion.

  Here, by referring to the image information of the reference image 113, it is possible to realize accurate determination of the unevenness of the inspection image 112. If the irregularities are known, various desired inspections such as measuring the line widths of the convex portions and the concave portions can be performed.

  As described above, the semiconductor manufacturing apparatus according to the present embodiment scans a semiconductor to be inspected by scanning charged particles and acquires an image of the semiconductor from the image acquired by the image acquisition unit. And an arithmetic unit for determining the semiconductor uneven pattern. The calculation unit acquires the inspection image of the semiconductor inspection region imaged by the image acquisition unit, generates an inspection profile from the inspection image, acquires the reference image of the semiconductor reference region imaged by the image acquisition unit, and references Generates a reference profile from the image, obtains uneven pattern information indicating the uneven pattern of the reference site from the design information of the semiconductor to be inspected, compares the inspection profile with the reference profile, and inspects the semiconductor by referring to the uneven pattern information The uneven pattern of the part is determined.

  Details of each processing content will be described below. The processing contents described below are each performed by the calculation unit.

(1-3) Detailed Procedure FIG. 15 shows the detailed procedure of Step 1404 and Step 1405 of FIG. 3 to 6 are diagrams illustrating what is described in the procedure.

  Hereinafter, the detailed procedure will be described with reference to FIGS. Step 1541: (acquisition of reference image profile). A profile in a predetermined direction of the reference image 113 is acquired. The obtained profile is called a reference profile. In the example of FIG. 3, the reference profile of the reference image 113 is indicated as 313. For example, when a portion of a pattern with a line in the vertical direction is projected, the reference image 113 is analyzed, and the average luminance value of the image is taken in the vertical direction to create one-dimensional data in the horizontal direction. Become. The horizontal axis of the reference profile 313 shown in FIG. 3 is the horizontal position of the reference image 113, and the vertical axis is a value obtained by averaging the luminance values of the reference image 113 in the vertical direction. In the example of FIG. 3, characteristic profiles are shown for each of the level of the concave portion, the level of the convex portion, and the level of the uneven edge portion. Since the reference image 113 captures a characteristic form of one concave portion, and can be compared with the pattern of the design data from this characteristic form, whether each area of the reference image 113 is a concave portion or a convex portion. Can be determined. The design data of the semiconductor 202 is stored in, for example, a storage device provided in a computer or a scanning electron microscope, or a database provided separately.

  When the edge of the image appears unstable, a more accurate reference profile can be obtained by removing the edge of the image from the average.

  Step 1542: (Acquisition of template). As shown in FIG. 4, a predetermined part of the reference profile 313 is cut out to be a template 401. FIG. 4 shows an example in which the edge portion from the concave portion to the convex portion is cut out from the reference profile 313. Details of how to take a template will be described in section (1-5).

  Step 1551: (acquisition of inspection image profile). A profile in a predetermined direction of the inspection image 112 is acquired. The obtained profile is called an inspection profile. In the example of FIG. 3, the inspection profile of the inspection image 112 is shown as 312. As in the case of the reference profile 313, when one-dimensional data in the horizontal direction is created by taking the average in the vertical direction of the inspection image 112, this becomes the inspection profile 312. In the example of FIG. 3, profiles are shown for each of the level of the concave portion, the level of the convex portion, and the level of the edge portion of the concave and convex portions. However, it is not possible to determine which part is a concave part or a convex part only by generating the inspection profile 312. This can be achieved by template matching and unevenness determination described below. Step 1552: (Template matching). As shown in FIG. 5, one-dimensional template matching with a template 401 is performed by using an inspection profile 312 as search target data to create a one-dimensional data string of evaluation values. This is called an evaluation value data string 501. The evaluation value data string 501 is expressed as a normalized correlation value between the inspection profile 312 and the template 401 and is a value from −1 to 1. In the example of FIG. 5, the position that forms the maximum value (vmax) in the evaluation value data string 501 is the position that most closely matches the template 401. Specific contents of template matching will be described in section (1-6).

  Step 1553: (Concavity and convexity determination). The position of the inspection profile corresponding to the position of the maximum value (vmax) in the evaluation value data string 501 is the position most similar to the template 401, and the unevenness of the template 401 and the unevenness of the inspection profile 312 are the maximum value (vmax ) At the corresponding position. Since the irregularities appear alternately, the irregularities of other parts of the inspection image 112 can be determined by alternately assigning them from the previously determined ones. Thereby, the unevenness | corrugation of the place to measure can also be known. In other words, after determining the location corresponding to the template as a recess or projection, the uneven pattern is determined as the recess or projection appearing alternately in the scanning direction of the charged particle starting from the location determined as the recess or projection. To do. For example, FIG. 6 shows an example in which P2 corresponds to the peak of the template 401 among the peaks P1 to P4 of the inspection profile 312. Since the unevenness of the template 401 corresponds to the unevenness of the inspection profile 312, it is possible to accurately determine the unevenness of the pattern of the inspection image 112 by making the unevenness appear alternately starting from P2. .

  It should be noted that the template 401 shown in FIGS. 4 to 6 takes only one peak from the reference profile 313. When a template is taken so as to include two peaks, the matching mainly searches for a position having a peak interval of the inspection profile 312 that matches the peak interval. However, the peak interval between the reference profile and the inspection profile does not always coincide with each other from the semiconductor processing accuracy and the design data itself. If the two peak intervals are different, taking a template so as to include two peaks results in inaccurate position detection, and therefore increases the possibility of misjudgment of irregularities. If the template contains one peak, matching first finds a position that matches the peak position first, and then matches the shape before and after the peak. This has the effect of searching for a place and determining the unevenness without error.

  With the above procedure, the unevenness of a desired part of the inspection image 112 can be determined from the coincidence position of the unevenness information of the template 401 determined from the design data and the inspection profile 312 which is the search data.

  (1-4) Regarding the inspection image 112 and the reference image 113. The relationship between the inspection image 112 and the reference image 113 will be supplemented. As in the examples of FIGS. 1 and 3, in the case of an inspection image with a vertical line, a peak portion having a peak appears in the inspection profile 312. In general, when the sample has irregularities, the edge portion at the boundary between the irregularities is strongly reflected in the inspection image 112. The inspection profile 312 averaged in the line direction becomes a peak portion that forms a peak at a portion corresponding to the edge.

  In the technique described in Patent Document 1, the unevenness is determined from the peak shape of the inspection profile 312, but there is a case where the determination is incorrect. Here, the unevenness is not determined from the inspection profile 312, but the inspection profile 312 is compared with the inspection profile 312 using the template 401 as a portion suitable for comparison with the reference profile 313 whose unevenness is known from the design data. Judge the unevenness.

(1-5) How to Take Template As shown in FIG. 4, a peak corresponding to the uneven edge portion of the reference image 113 appears in the reference profile 313. For example, when a linear object having one recess is photographed, peaks appear at two locations corresponding to the edge of the recess (Q1, Q2). A template 401 is formed by cutting out a predetermined width d around one of the peaks.

  Of the two peaks (Q1, Q2) of the reference profile 313, the template is the peak at the rear as viewed from the direction of beam scanning with the electron gun 203 (in FIG. 4, Q2 is scanned from left to right). It is preferable to select 401 because the influence of charging at the corresponding position of the inspection profile 312 resembles that of the template 401.

  If the peak of Q1 is used as a template, there is no peak on the left side of Q1, so the profile is less affected by charging. However, the inspection profile 312 has a plurality of peaks. When scanning from the left, the leftmost peak profile is less affected by charging, and the subsequent peaks are profiles affected by charging. If the change in profile due to the effect of charging is greater than the difference in profile due to the difference in unevenness, matching will select the one with less influence of charging, and the possibility of detecting the leftmost peak position is increased. . In general, since the leftmost peak position of the inspection profile with little influence of charging and the peak position taken on the template are irrelevant, the corresponding position is erroneously recognized, and the possibility of an irregularity determination error increases.

  If the peak of Q2 is used as a template, the profile changes in such a manner that the change in profile due to the influence of charging overlaps the change in profile due to the difference in unevenness. The inspection profile 313 has a plurality of peaks, and even if the peak intervals are not exactly the same, if the intervals are similar to a certain extent, the peaks having the same unevenness as Q1 before the peaks having the same unevenness as Q2 of the inspection profile 313 If there is, the unevenness relationship and the influence of charging are the same as Q2, and the profile is similar to Q2. Therefore, in the matching, the leftmost peak that is less affected by charging is removed, and in the subsequent peaks, the same uneven part is selected, the corresponding position can be detected accurately, and the uneven part determination is accurate. There is.

  In FIG. 4, the reference profile has two peaks. However, if there are three, four, or two or more peaks, the first peak is avoided and a template including an appropriate peak later is used. Is preferable for the same reason.

  The width of the template 401 is the shortest interval (d = min (d1, d2, d3)) among the intervals (d1, d2, d3) of a plurality of peaks (P1, P2, P3, P4) of the inspection profile 312. It is preferable that they are the same (the width may be a width that is not significantly affected by the tail of the peak next to the reference image when matching with the reference image, whether it is somewhat wide or narrow). Of the inspection profile 312, the portion corresponding to the template 401 of the reference profile 313 sufficiently includes features of the shape of the peak placed at the center, and accurate matching when the width is not affected by the spread of the skirts of other peaks. This is because it becomes possible.

(1-6) Template Matching In this embodiment, template matching between the template 401 and the inspection profile 312 is performed. When normalized correlation is used for the evaluation value of matching, an evaluation value data string 501 is created based on the following equation (1) for all the places in the inspection profile 312 where a portion having the same length as the template 401 can be taken. .

J (k) = TS / (sqrt (TT) · aqrt (SS)) (1)
However,
TS = Σ (T (i) -ht) ・ (S (k + i) -hs)
TT = Σ (T (i) -ht) ・ (T (i) -ht)
SS = Σ (S (k + i) -hs) ・ (S (k + i) -hs)
ht = Σ T (i) / n
hs = Σ S (i) / n
T (i) is a template 401, i is an integer indicating its coordinates, S (k + i) is an inspection profile 312, k + i is an integer indicating its coordinates, and J (k) is an evaluation In the value data string 501, k is an integer indicating the coordinates, and n is an integer indicating the number of arrays of the template 401. , Σ indicates that i is added from 0 to n−1.

(1-7) Various variations
(A) Various calculation methods for template matching evaluation values
In section (1-6), normalized correlation is used as the evaluation value calculation method. In addition, there are various calculation methods that serve as evaluation indexes for searching for a portion similar to the template. For example, there are a method of searching for a large correlation value J2 that is not normalized as in the following equation (2), and a method of searching for a small difference sum of absolute values J3 as in equation (3).

J2 (k) = Σ (T (i) -ht) ・ (S (k + i) -hs) (2)
J3 (k) = Σabs (T (i) -S (k + i)) (3)
(B) How to take multiple templates
In section (1-5), one template is taken from the reference profile 313, but a plurality of templates can be taken.

  For example, as shown in FIG. 7, in addition to the template 401, there is a method in which a template 701 is taken around a peak different from the peak at the center of the template 401, and two templates (401 and 701) are both used. Here, an example is shown in which an edge template 701 from the convex portion to the concave portion is used in addition to the template 401 of the edge portion from the concave portion to the convex portion.

  In this case, the detailed procedure of FIG. 15 described in the section (1-3) is as shown in FIG. In FIG. 16, compared with FIG. 15, step 1542 is changed to 1642, step 1552 is changed to 1652, and step 1652b is newly added. Only the unusual steps are described below.

  Step 1642: (Acquire multiple templates). Two templates 401 and 701 are acquired. Each is cut out by a predetermined width around each peak of the reference profile 313.

  Step 1652: (Multiple template matching). Each template (401 and 701) and the inspection profile 312 are template-matched to obtain respective evaluation value data strings (501 and 702). As in the case of the evaluation value data string 501, the evaluation value data string 702 is expressed as a normalized correlation value between the inspection profile 312 and the template 701, and is a value from −1 to 1. An example of the evaluation value data string 702 is shown in FIG. Step 1652b: (Template selection).

  The maximum value vmax1 of the evaluation value data string 501 is searched for, and the type of edge at that time (there are two types of edges, that is, an edge that becomes convex from a concave and an edge that becomes convex from a convex when moving from right to left in the image). The maximum value is searched for from among the types of peaks different from the edge type having the maximum value vmax1, and the difference sa1 between them is taken.

  Similarly, the maximum value vmax2 is searched from the evaluation value data string 702, the maximum value is searched from among different types of peaks, and the difference sa2 between the two is obtained.

  It is assumed that the larger one of the two differences (sa1 and sa2) is searched, and the template at this time (whichever is selected from 401 or 701) is a template suitable for comparison with the inspection profile 312.

  Here, the template (401 or 701) having the larger difference (sa1 and sa2) is selected, but the template having the largest maximum value (max1 and max2) can be selected.

  In general, the template with the larger difference (sa1 and sa2) is more unlikely to be affected by noise, and it is more likely that the position can be detected stably by matching, and the unevenness determination is accurate. The possibility is high. However, if the difference (sa1 and sa2) is not much different, choosing the larger maximum value (max1 and max2) is less susceptible to noise than choosing the larger difference, and the unevenness judgment is In rare cases, it may be accurate. The influence of the normalized correlation value on noise, including the balance with the signal, is non-linear and complex, so it cannot be expressed in a simple form. You can also make a table that tells you if it ’s good.

  In addition, here, there are two peaks in the reference image, and two templates are prepared as a result corresponding to each peak. However, in the case where there are two or more peaks in the reference image, a plurality of appropriate peaks are similarly provided. It is also possible to prepare a plurality of templates corresponding to the peaks and similarly select a template having the largest difference in normalized correlation values.

  In step 1553, unevenness determination is performed with reference to the template selected in step 1652b (one selected in 401 and 701) and the evaluation value data string (one selected in 501 and 702) at that time. .

  As described above, by using a plurality of templates 401 and 701, it is possible to select a template that is more suitable for template matching, and it is possible to improve accuracy.

(C) How to take a reference image
In step 1403 of the section (1-2), an example is shown in which the reference image 113 is acquired by photographing the reference portion 103. Even other parts can be used as reference parts as long as the irregularities are uniquely determined from the design data. For example, the reference image 1213 may be photographed aiming at the reference part 1203 in FIG. 12, or the reference part 1303 in FIG. 13 may be photographed.

  FIG. 1 shows an example in which a reference portion 103 is used for reference, which is provided in the periphery of the semiconductor 202 and is different from the circuit pattern of the inspection portion 102. On the other hand, FIGS. 12 and 13 show an example in which a part including the same circuit pattern as the examination part is used as the reference part. For example, the reference part 1203 in FIG. 12 uses a part having the same circuit pattern as the examination part and a part having no circuit pattern outside (lateral direction) as the reference part, and the reference part 1303 in FIG. 13 has the same circuit as the examination part. An example is shown in which a pattern part and a part without a circuit pattern outside (vertical direction) are used as reference parts. Since both reference images include a portion having no circuit pattern, the reference image has sufficient morphological features to determine unevenness from design data.

(D) When the reference image 113 is photographed after the inspection image 112 In the above, the reference image 113 is photographed after the inspection image 112 is photographed. However, the inspection image 112 may be photographed after the reference image 113 is photographed. it can.

  In addition, the reference image 113 is not photographed every time, but when the same kind of semiconductor is produced, the reference image 113 is photographed first, and the template 401 is registered. When the first chip is photographed on the same wafer, the reference image 113 is photographed and the template 401 is registered, and this can be used on the same wafer.

(E) Method of using left and right difference as template with peak as axis As shown in FIG. 10, it is possible to draw a layer 1002 in which the peaks of template 401 that are symmetrical with respect to the central axis are overlapped. Then, a difference template 1012 can be newly obtained by subtracting the left and right targets from the template 401. In this way, the template matching step 1552 is not necessary, the portion including the peak portion of the inspection profile 312 is cut out for the template, the difference from the one that is symmetric about the peak is taken, and the normalized correlation between the differences is obtained. take. The peak having the highest normalized correlation with the difference from the symmetrically peaked peak can be regarded as the peak most similar to the template 401. In this case, the condition is that the peak can be detected stably and the peak position does not shift due to noise or the like.

  (F) Method of Using Feature of Predetermined Location of Reference Image The inspection profile 312, the reference profile 313, and the template 401 are not created and template matching is not performed, but a predetermined location of the inspection image 112 and the reference image 113 (for example, 11, 1101, 1102, 1103, and 1104) shown in FIG. 11 are matched to find the same unevenness and determine the unevenness.

  For example, as shown in the procedure of FIG. 17, there is a method of determining unevenness by outputting light / dark information from each average value of predetermined locations (1101, 1102, 1103, 1104). In the following description, the symbols in FIG. 17 are used as step numbers.

  Step 1741: (acquisition of reference image shading information). The average value of a predetermined place (1101, 1102) of the reference image 113, in which the unevenness is known, is obtained from the design data, and the density information of the image is acquired to determine which place has the higher or lower average value, and the unevenness and the lightness and darkness are obtained. Make the correspondence.

  Step 1751: (Acquisition of light / dark information of inspection image). An average value of predetermined locations (1103, 1104) of the inspection image 112 is calculated, and shading information is acquired as to which location has an average value larger or smaller.

  Step 1553b: (Concavity and convexity determination). Based on the correspondence between the unevenness and the light and shade found from the reference image 113, the unevenness at a predetermined place (1103 and 1104) is determined from the light and shade of the inspection image 112.

  In this case, it is a condition that the density difference between the reference image 113 and the inspection image 112 is stable (the influence of noise and the difference in the shooting location is small) to such an extent that the average value can determine the unevenness.

  In addition, there is a method of looking at the texture instead of the average value of predetermined places (1101, 1102, 1103, 1104). There are methods for evaluating the texture, such as looking at dispersion, looking at cumulants, and looking at frequency characteristics by Fourier transform. In this case, the condition is that there is a difference in texture depending on the unevenness.

(G) When the influence of charging is small When the influence of charging is small, the inspection image 112 can be taken after the reference image 113 is photographed, and the position shift when the inspection image 112 is photographed from the reference image 113 However, if the unevenness pattern period of the inspection image 112 is half or less, the template 401 and template matching are unnecessary, and the unevenness of the inspection image 112 can be directly known by referring to the design data. In addition, when the influence on the semiconductor due to the imaging can be ignored, the unevenness of the inspection image 112 is not matched by sequentially imaging another position within the range where there is little deviation of the imaging position and proceeding to the inspection site 102. You can find out at

  Similarly, after photographing the inspection image 112, another part is photographed in a range where there is little deviation in photographing, and the process proceeds to the reference region 103, whereby the unevenness of the inspection image 112 can be known without matching. .

  In addition, each variation of (A)-(G) mentioned above can also be applied to the present embodiment independently, and can be applied to the present embodiment in various combinations.

(2) Example 2
(2-1) Difference between Example 1 and Example 2 The basic configuration of Example 2 is the same as that of Example 1. However, the semiconductor 202 on which the resist is mounted is different in that it is made by a double patterning manufacturing method. Therefore, in addition to the unevenness, it is also necessary to determine the type of the recess (whether the recess is processed first or the recess processed later). However, the conventional technique has a limit in improving the inspection efficiency because it is not possible to simultaneously perform the unevenness determination and the determination of the type of the recessed portion with one template. Example 2 describes a method that can simultaneously determine the type of a recess (whether processed first or recessed processed later) in addition to the unevenness determination for a sample processed by the double patterning manufacturing method. To do.

  The outline of the procedure of the second embodiment is as shown in FIG. Although it is almost the same as the outline of the procedure of the first embodiment (FIG. 14), the step 1405 for determining the unevenness is the step 1805 for simultaneously determining the types of the unevenness and the recessed portion. The reference part 103 reflects the case of double patterning, and selects a part from which not only the unevenness but also the type of the recessed part can be determined from the design data.

  FIG. 18 shows the detailed procedure of the second embodiment. Compared with FIG. 15 of the detailed procedure of the first embodiment, the only difference is that step 1542 is changed to step 1842 and step 1553 is changed to step 1853. It is.

  Other configurations in the second embodiment are the same as those in the first embodiment unless otherwise noted.

(2-2) Detailed procedure Hereinafter, the detailed procedure of Example 2 is described.

  Step 1541: (acquisition of reference image profile). A reference profile in a predetermined direction of the reference image 113 is acquired. FIG. 19 shows a reference profile 313d as an example. Although the acquisition procedure is the same as that of the first embodiment, the reference profile 313d is characterized by being created by the double patterning manufacturing method, and not only the period of irregularities but also the order of processing (created after the previously created depression or later). It is also reflected in the difference between the recesses.

  Step 1842: (Template acquisition).

  A predetermined part of the reference profile 313d is cut out as a template 401d. Unlike the first embodiment, the template 401d is selected in the case of the double patterning manufacturing method. Since the reference profile 313d has four peaks P4a, P5a, P4b, and P5b as shown in FIG. 19, a template centered on the last peak as viewed from the scanning direction of the electron gun 203 is selected. . The width of the template is the same as the shortest interval among a plurality of peak intervals of the inspection profile 312.

  Step 1551: (acquisition of inspection image profile). A profile in a predetermined direction of the inspection image 112 is acquired to obtain an inspection profile 312d. The procedure is the same as in Example 1.

  Step 1552: (Template matching). One-dimensional template matching is performed with the template 401d using the inspection profile 312d as search target data, and an evaluation value data string 501d is created. The procedure is the same as in Example 1. The position that makes the maximum value in the evaluation value data string 501d is the position that most matches the template 401d.

  Step 1853: (Simultaneous determination of types of concave and convex portions and concave portions). In the evaluation value data string 501d, the template 401d arranged at the top of the position having the maximum value is the best match with the corresponding position of the inspection profile 312d and the template 401d. Therefore, it is determined that the unevenness and the recess type of the template 401d and the unevenness and the recess type of the inspection profile 312 are the same at the corresponding positions. Since the irregularities appear alternately and the types of the concave portions appear alternately in the concave portions, the concave and convex portions of other parts of the reference image 113 can also be assigned alternately and determined from the previously determined ones. Thereby, the kind of the unevenness | corrugation and recessed part of the place to measure can be determined simultaneously.

(1-3) In the case of taking a plurality of templates As in the first embodiment, take a plurality of templates, select the most suitable one from each evaluation data string, select the template at this time, and select the type of irregularities and recesses. It can also be determined.

  In the example of FIG. 20, two templates (401d and 701d) are taken, and template matching between each template and the inspection profile 312d is performed to create an evaluation value data string (501d and 702d). As in the case of the plurality of templates in the first embodiment, the template having the larger difference between the maximum value of the evaluation data string and the maximum value of the peak having different unevenness is selected, and the types of the unevenness and the recessed portion are selected in the same manner as in Step 1853 above. Judge simultaneously.

  In addition, the evaluation value data string (501d, 702d) has four types of peaks (peaks corresponding to the edges of the concave portions and the convex portions manufactured earlier, peaks corresponding to the edges of the convex portions and the concave portions manufactured later, and later) There are four peaks for each evaluation value data (501d or 702d) because the peak corresponding to the edge of the manufactured concave part and the convex part and the peak corresponding to the edge of the convex part and the concave part manufactured previously are alternately present. Find the average value of the peaks every other time, make the average value of each of the four types of peaks, find the maximum value among them, and find the maximum value or the difference between the maximum value and the maximum value of the unevenness is large And the corresponding template (401d or 701d) at that time can be selected as a preferred template.

  Since the reference profile 313d has four peaks, it is possible to create four templates and select one template in the same manner.

  101 ... Positioning image, 102 ... Inspection site, 103 ... Reference site, 112 ... Inspection image, 113 ... Reference image, 201 ... Stage, 202 ... Semiconductor, 203 ... Electron gun, 204 ... Secondary electron detector, 205 ... Imaging , 206 ... calculating device, 207 ... display device, 312 ... inspection profile, 312d ... inspection profile, 313 ... reference profile, 313d ... reference profile, 401 ... template, 401d ... template, 501 ... evaluation value data string, 501d ... evaluation Value data string, 701 ... Template, 701d ... Template, 702 ... Evaluation value data string, 702d ... Evaluation value data string, 801 ... Template, 901 ... Template, 1002 ... Simultaneously symmetric with respect to the center axis , 1012 ... difference template, 1101 ... predetermined place, 1102 ... predetermined place, 1103 ... predetermined place, 1104 ... predetermined place, 1203 ... reference part, 1213 ... reference image, 1303 ... reference part, 1401 ... positioning image Shooting step 1402 ... Inspection image capturing step, 1403 ... Reference image capturing step, 1404 ... Reference image information acquisition step, 1405 ... Unevenness determination step, 1541 ... Reference image profile acquisition step, 1542 ... Template acquisition step , 1551 ... Inspection image profile acquisition step, 1552 ... Template matching step, 1553 ... Concavity and convexity determination step, 1553b ... Concavity and convexity determination step, 1642 ... Multiple template acquisition step, 1652 ... Multiple template matching step, 1652b ... Template Selection step, 1741 ... acquisition information of reference image, 1751 ... acquisition information of inspection image, 1805 ... step of determining unevenness and recess type simultaneously, 1842 ... acquisition of template, 1853 ... irregularity and recess Type of simultaneous determination step, 907 ... the process of shifting the focus to the next target point candidate .

Claims (11)

An image acquisition unit that scans and scans a semiconductor to be inspected with charged particles and acquires an image of the semiconductor, and a calculation unit that determines an uneven pattern of the semiconductor from the image acquired by the image acquisition unit,
The computing unit is
Acquire an inspection image of the inspection site of the semiconductor imaged by the image acquisition unit,
Generating an inspection profile from the inspection image;
Obtaining a reference image of a reference portion of the semiconductor imaged by the image obtaining unit;
Generating a reference profile from the reference image;
Obtaining concavo-convex pattern information indicating the concavo-convex pattern of the reference site from the design information of the semiconductor to be inspected,
A semiconductor inspection apparatus that compares the inspection profile with the reference profile and determines the concave / convex pattern of the inspection portion of the semiconductor with reference to the concave / convex pattern information. The semiconductor inspection apparatus according to claim 1,
The calculation unit extracts, as a template, a location corresponding to one peak from a plurality of peaks corresponding to the uneven pattern of the reference portion of the semiconductor from the reference profile,
A semiconductor inspection apparatus that compares the inspection profile with the template and determines the concave / convex pattern of the inspection portion of the semiconductor with reference to the concave / convex pattern information. The semiconductor inspection apparatus according to claim 2,
The reference portion includes a reference pattern of one concave portion or convex portion that crosses the scanning direction of the charged particle,
The reference profile includes at least two peaks corresponding to an edge portion of the reference pattern of the one concave portion or convex portion,
The said calculating part extracts the location corresponding to one peak after the said charged particle scanning direction among the said at least 2 peaks of the said reference profile as a template, The semiconductor inspection apparatus characterized by the above-mentioned. The semiconductor inspection apparatus according to claim 3.
The said calculating part extracts the template of the location corresponding to each peak among the said at least 2 peaks of the said reference profile, respectively, The semiconductor test | inspection apparatus characterized by the above-mentioned. The semiconductor inspection apparatus according to claim 2,
The semiconductor inspection apparatus, wherein the arithmetic unit extracts a portion having a predetermined width including the one peak as the template. The semiconductor inspection apparatus according to claim 5,
The said test | inspection part is a width | variety corresponding to the shortest space | interval among the some peak space | intervals corresponding to the uneven | corrugated pattern of the test | inspection site | part of the said semiconductor. The semiconductor inspection apparatus according to claim 2,
The calculation unit determines a position corresponding to the template as a concave portion or a convex portion of the inspection portion of the semiconductor, and then scans the charged particles starting from the portion determined as the concave portion or the convex portion. A semiconductor inspection apparatus characterized in that a concave / convex pattern is determined as one in which concave portions or convex portions appear alternately. The semiconductor inspection apparatus according to claim 2,
The semiconductor is created by a double patterning manufacturing method,
The calculation unit uses a reference profile as a template corresponding to one peak corresponding to the concavo-convex pattern derived from the double patterning manufacturing method from among a plurality of peaks corresponding to the concavo-convex pattern of the reference portion of the semiconductor. Extract as
A semiconductor inspection apparatus characterized by comparing the inspection profile and the template in parallel to determine the concave / convex pattern and the type of the concave / convex pattern of the inspection portion of the semiconductor. The semiconductor inspection apparatus according to claim 8.
The type of the concavo-convex pattern is a semiconductor inspection apparatus characterized in that the concave portion of the inspection portion of the semiconductor is a concave portion created earlier or a concave portion created later in the double patterning manufacturing method. The semiconductor inspection apparatus according to claim 1,
The said calculating part acquires the said test | inspection image and the said reference image on the same imaging conditions, The semiconductor test | inspection apparatus characterized by the above-mentioned. A semiconductor comprising: an image acquisition unit that scans a semiconductor to be inspected by scanning charged particles and acquires an image of the semiconductor; and an arithmetic unit that determines an uneven pattern of the semiconductor from the image acquired by the image acquisition unit A semiconductor inspection method in an inspection apparatus,
The computing unit is
Acquire an inspection image of the inspection site of the semiconductor imaged by the image acquisition unit,
Generating an inspection profile from the inspection image;
Obtaining a reference image of a reference portion of the semiconductor imaged by the image obtaining unit;
Generating a reference profile from the reference image;
Obtaining concavo-convex pattern information indicating the concavo-convex pattern of the reference site from the design information of the semiconductor to be inspected,
A semiconductor inspection method, wherein the inspection profile is compared with the reference profile, and the concave / convex pattern of the inspection portion of the semiconductor is determined with reference to the concave / convex pattern information.

Images