JP2006125967A - Inspection device, inspection method, and manufacturing method of pattern substrate - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inspection device and inspection method capable of easily obtaining high detection sensitivity, and to provide a manufacturing method of a pattern substrate. <P>SOLUTION: The inspection device comprises a laser light source 101, an objective lens 102 for forming an optical spot on a sample surface, a photodetector 110 for outputting an output signal based on a reflected light reflected on the sample surface, of light beams coming into a sample 107 from the objective lens 102, a stage 108 that scans relative positions of the sample 107 and the optical spot along a scan line and scans the region where the optical spot overlaps the optical spot of the adjacent scan line so as to shine it, a defect candidate detecting section 202 for detecting a defect candidate based on the output signal from the photodetector 110, and a defect determining section 208 for determining whether it is a defect based on the distance between the defect candidates detected in the defect candidate detecting sections 202. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an inspection apparatus, an inspection method, and a pattern substrate manufacturing method using the inspection apparatus, and more particularly to an inspection apparatus and an inspection method using optical scanning.

  In recent years, with the progress of high integration of semiconductor devices, patterns formed on the surface of semiconductor wafers used for semiconductor devices have become extremely fine. In a semiconductor wafer, if a defect exists on the wafer at the time of pattern formation, the manufacturing yield is greatly reduced. For this reason, with the miniaturization of the processing dimension of the pattern of the semiconductor device formed on the semiconductor wafer, restrictions on allowable defects are becoming strict. Therefore, inspection for detecting minute defects on a semiconductor wafer is important for improving the yield and reliability of semiconductor products and their manufacture.

  As a method for detecting such a defect, by detecting the scattered light from the defect by the irradiation light from a predetermined light source as a signal light with a photodetector, the defect on the semiconductor wafer at the inspection position irradiated with the irradiation light is detected. There is a method for inspecting the presence or absence (see, for example, Patent Document 1).

  In addition, in the inspection of crystal defects such as OSF (Oxidation Induced Stacking Fault) and BMD (Bulk Micro Defect) of semiconductor wafers, detection is performed using the number of generated defects and the generation density as criteria for improving the defect detection accuracy. There has been proposed a method for determining whether or not a defect is a true defect (see, for example, Patent Document 2).

Furthermore, a surface defect inspection apparatus is disclosed that calculates the distance between each defective portion and obtains the total value of the areas of a plurality of defective portions whose distance is less than a predetermined value (see, for example, Patent Document 3). ).
JP 2002-181726 A JP 2002-365235 A JP-A-8-101130

  In general, a defect inspection of a semiconductor wafer is generally performed by performing processing on the intensity of an output signal generated by irradiating a semiconductor wafer with light such as a laser and receiving scattered / reflected light. That is, the generated output signal is supplied to the defect detection circuit and compared with a positive / negative threshold value (slice level). If the output signal exceeds this slice level, it is detected as a defect.

  However, noise such as thermal noise or shot noise is randomly generated in this output signal. Therefore, even a portion that does not actually have a defect may be detected as a pseudo defect if it exceeds the slice level due to noise. In order to improve the defect detection sensitivity, it is necessary to separate the output signal due to the defect and the output signal due to the noise.

  In order to perform an effective inspection on such a semiconductor wafer, it is necessary to lower the slice level. However, when the slice level is lowered, there is a problem that false detection due to noise increases.

  The problem as described above also occurs in the surface inspection other than the semiconductor wafer such as the defect detection of the mask blank for manufacturing the semiconductor device.

  The present invention has been made in the background of such circumstances, and an object thereof is to provide an inspection apparatus and an inspection method that can easily obtain high detection sensitivity, and a method of manufacturing a pattern substrate using the inspection apparatus. .

  An inspection apparatus according to a first aspect of the present invention includes: a light source that generates a light beam; an objective lens that collects the light beam emitted from the light source to form a light spot on a sample surface; Of the light beam incident on the sample, the reflected light reflected from the sample surface, the scattered light from the defective part of the sample, or the transmitted light transmitted through the sample is received, and an output signal based on the amount of the received light is output. And a scanning means for scanning a relative position between the sample and the light spot along a scanning line, wherein the light spot illuminates a region overlapping with a light spot of an adjacent scanning line Based on a distance between defect candidates detected by the defect candidate detection unit, a defect candidate detection unit that detects a defect candidate based on an output signal from the photodetector, In which and a determining defect determining section whether Recessed. Thereby, a defect can be detected with high detection sensitivity.

  In the inspection apparatus according to the second aspect of the present invention, in the above-described inspection apparatus, the scanning unit sets the pitch between the scanning lines of the light spots to half or less of the spot diameter of the light spots. Thereby, a defect can be detected with high sensitivity on the entire surface of the sample.

  An inspection apparatus according to a third aspect of the present invention further includes a diffraction grating that converts the light beam emitted from the light source into a plurality of sub-beams in the above-described inspection apparatus, and the trajectory of the sub-beams on the sample surface. A plurality of light spots are formed so as to overlap the trajectory of the adjacent sub beam, and the photodetector has a plurality of light receiving elements corresponding to the plurality of sub beams. Thereby, an inspection can be performed at high speed.

  An inspection apparatus according to a fourth aspect of the present invention is the above-described inspection apparatus, wherein the inspection apparatus includes a plurality of the light sources, and the locus of the light beam from the light source is the locus of the light beam from another light source on the sample surface. A plurality of light spots are formed so as to overlap with each other, and the photodetector has a plurality of light receiving elements corresponding to light beams from the plurality of light sources.

  The inspection apparatus according to a fifth aspect of the present invention is the inspection apparatus according to the above-described inspection apparatus, wherein the defect determination unit detects the defect candidate detected by the light spot corresponding to the first scanning line among the scanning lines. Whether there is a defect or not is determined by whether or not there is a defect candidate detected by the light spot corresponding to the second scanning line within the defect determination distance from the upper position. Thereby, erroneous detection due to noise can be prevented.

  An inspection apparatus according to a sixth aspect of the present invention is the inspection apparatus described above, wherein the defect candidate detection unit detects the defect candidate at a position where an output signal exceeds a slice level.

  The inspection apparatus according to the seventh aspect of the present invention is the inspection apparatus described above, wherein, in the defect candidate detection unit, there are two or more positions within the defect determination distance that exceed the slice level in one scanning line. One defect candidate is detected. This can prevent erroneous detection of defect candidates.

  An inspection apparatus according to an eighth aspect of the present invention is the above-described inspection apparatus, wherein the defect candidate is detected in a standard sample in which the number of defects actually present on the sample surface is known, and the number of defect candidates is detected. The number of erroneously detected defects erroneously determined as the defect is calculated based on the above, and the slice level is set based on the number of erroneously detected defects. As a result, high-sensitivity detection can be performed in a range where the number of erroneously detected defects is less than or equal to an allowable value.

  An inspection method according to a ninth aspect of the present invention includes: emitting a light beam; condensing the light beam to irradiate the sample so as to form a light spot on a sample surface; and Scanning a relative position of the sample and the light spot along the scan line so as to illuminate a region overlapping with a light spot of an adjacent scan line; among the light beams incident on the sample, the sample Detecting reflected light reflected from the surface of the sample, scattered light from a defective portion of the sample or transmitted light transmitted through the sample, and outputting an output signal based on the detected reflected light, scattered light or transmitted light A step of detecting a defect candidate based on the output signal, and a step of determining whether the defect is a defect based on a distance between the detected defect candidates. That. Thereby, a defect can be detected with high sensitivity.

  In the inspection method according to the tenth aspect of the present invention, in the above-described inspection method, in the scanning step, the pitch between the scanning lines of the light spot is scanned with a half or less of the spot diameter of the light spot.

  An inspection method according to an eleventh aspect of the present invention is the inspection method described above, wherein a plurality of the light spots are formed on the sample surface, and the plurality of light spots are formed by a plurality of light receiving elements corresponding to the plurality of light spots. It is to detect. Thereby, an inspection can be performed at high speed.

  An inspection method according to a twelfth aspect of the present invention is the inspection method described above, wherein in the step of detecting the defect candidate, the defect candidate is detected at a position where the output signal exceeds a slice level. It is.

  In the inspection method according to the thirteenth aspect of the present invention, in the above-described inspection method, when there are two or more positions within the predetermined distance in one scanning line that exceed the slice level, they are detected as one defect candidate. Is. This can prevent erroneous detection of defect candidates.

  The inspection method according to a fourteenth aspect of the present invention is the above inspection method, wherein the defect candidate is detected in a standard sample in which the number of defects actually existing on the sample surface is known in advance, and the standard sample Estimating the number of erroneously detected defects that are erroneously determined as the defect based on the number of defect candidates in, and setting the slice level so that the number of erroneously detected defects is less than or equal to the allowable number of erroneously detected defects Further comprising the step of:

  A pattern substrate manufacturing method according to a fifteenth aspect of the present invention includes a step of detecting a defect of the substrate that is the sample by the above-described inspection method, and a step of correcting the defect detected by the step of detecting the defect. And a step of forming a pattern on the substrate in which the defect is corrected. Thereby, the productivity of the pattern substrate can be improved.

  ADVANTAGE OF THE INVENTION According to this invention, the inspection apparatus and inspection method which can obtain a high detection sensitivity simply, and the manufacturing method of a pattern board | substrate using the same can be provided.

  Hereinafter, embodiments to which the present invention can be applied will be described. The following description is to describe the embodiment of the present invention, and the present invention is not limited to the following embodiment. For clarity of explanation, the following description and drawings are omitted and simplified as appropriate. Moreover, those skilled in the art can easily change, add, and convert each element of the following embodiments within the scope of the present invention. In addition, in each drawing, the same code | symbol is attached | subjected to the same element and the duplication description is abbreviate | omitted as needed for clarification of description.

  The configuration of the inspection apparatus according to the present invention will be described with reference to FIG. FIG. 1 is a schematic diagram showing the configuration of an example of a defect inspection apparatus 100 according to the present invention. 101 is a laser light source, 102 is a diffraction grating, 103 is a first relay lens, 104 is a second relay lens, 105 is a beam splitter, 106 is an objective lens, 107 is a sample, 108 is a stage, 109 is a third relay Reference numeral 110 denotes a photodetector, and reference numeral 200 denotes a processing device.

  A laser beam generated from the laser light source 101 enters the diffraction grating 102. The diffraction grating 102 is, for example, a one-dimensional diffraction grating, and converts an incident laser beam into n sub beams. These sub beams are assumed to be aligned in a line. The n sub-beams emitted from the diffraction grating 102 pass through the first relay lens 103 and the second relay lens 104 and enter the objective lens 106 through the beam splitter 105.

The objective lens 106 forms n light spots on the sample 107 to be inspected by focusing the incident n light beams into a minute spot shape. The pitch of the n light spots is approximately equal to half the spot diameter of the light spot. That is, the distance between the centers of the adjacent light spots on the sample 107 is half of the light spot diameter. Here, for example, a light spot having a spot diameter (width at an intensity of 1 / e ² with respect to the peak intensity) of about 0.4 μm can be used.

  The stage 108 on which the sample 107 is placed is an XY stage having an XY drive mechanism, and can move the sample 1077 relative to the light spot. By moving the stage 108 in the horizontal direction using stage control means such as a controller (not shown), it is possible to irradiate a light spot at an arbitrary position of the sample 107 placed on the stage 108. Further, the sample 107 can be illuminated over the entire surface by moving the stage 108 using a controller and scanning each light spot along the scanning line with n light spots. That is, after scanning a plurality of light spots along the scanning line, the stage 108 is moved in a direction perpendicular to the scanning line. This is repeated to scan and illuminate the entire sample surface. The pitch of each scanning line is preferably less than half the diameter of the light spot. For example, when the light spot diameter is 0.4 μm, the scanning line pitch is preferably 0.2 μm or less. As a result, scanning is performed so that the light spots overlap in adjacent scanning lines.

  The reflected beam from the light spot formed on the sample 107 is reflected by the beam splitter 105 again through the objective lens 106. The reflected beam reflected by the beam splitter 105 enters the photodetector 110 through the third relay lens 109. The photodetector 110 is, for example, a linear image sensor having a plurality of light receiving elements arranged in a line. As the linear image sensor, for example, a photodiode array or a CCD line sensor can be used. Then, the reflected beam from each light spot on the sample 107 is incident on the corresponding light receiving element of the photodetector 110. The light receiving elements of the linear image sensor are separated from each other by a light shielding member, and the light shielding member or its frame functions as a pinhole that regulates light incident on the light receiving element. Therefore, the linear image sensor itself has a spatial filter having a pinhole.

  An output signal from each light receiving element is amplified by an amplifier (not shown) and supplied to the processing device 200. The processing apparatus 200 compares the output signal from each light receiving element of the photodetector 110 with a positive / negative threshold value (slice level), and determines a defect from the defect candidate detection unit and the defect candidate. It has a defect determination unit. The processing device 200 will be described later.

  Here, the defect inspection method according to the present embodiment will be described with reference to FIG. FIG. 2 is a diagram showing an output signal generated when a portion where the actual defect 114a exists on the sample 107 is scanned. Note that the actual defect 114a is a defect in which foreign matter or the like adheres to the surface of the sample and the reflectance changes greatly with respect to a normal location. Here, the case where there are four light spots 111 is illustrated. The four light spots 111 are arranged side by side in the vertical direction. As shown in FIG. 2, they are aligned in a line so that a part of the locus of the adjacent light spot 111 overlaps. In other words, the trajectory of the second light spot 111b and the trajectory of the third light spot 111c are partially overlapped so that the trajectory of the first light spot 111a and the trajectory of the second light spot 111b partially overlap. In addition, the trajectories of the respective light spots are arranged in a line so that the trajectory of the third light spot 111c and the trajectory of the fourth light spot 111d partially overlap each other. These light spots 111 are scanned in the horizontal direction, that is, along the scanning lines, as indicated by arrows in FIG. In FIG. 2, an arrow indicating a scanning line indicates a trajectory through which the center of the light spot passes.

  By using these four light spots 111, the sample 107 is scanned in the direction of the arrow in FIG. The pitch of the locus of each light spot coincides with half the spot diameter. That is, the illumination areas obtained by scanning each light spot 111 are overlapped by half. Specifically, the upper half of the area illuminated by scanning the light spot 111b overlaps the lower half of the area illuminated by scanning the light spot 111a. On the other hand, the lower half of the area illuminated by scanning the light spot 111b overlaps the upper half of the area illuminated by scanning the light spot 111a. The same applies to the other light spots 111. In this way, the areas illuminated by scanning adjacent light spots overlap in half. In the above description, it has been described that the light from the laser light source 101 is branched by the diffraction grating 102 so that the sub beams overlap. However, the embodiment of the present invention is not limited to this. For example, it is not necessary to form a light spot where two adjacent sub-beams among the sub-beams branched by the diffraction grating 102 overlap. That is, scanning may be performed so that any one of the light spots 111 formed by the sub-beams branched at a predetermined interval overlap. In other words, it is only necessary to scan so that the trajectories of the scanned light spot 111 overlap. For example, when the light beam is branched from the laser light source 101 into two sub beams, one sub beam may form the light spot 111a and the other sub beam may form the light spot 111c. In this case, scanning is sequentially performed so that the light spot 111a and the light spot 111b are formed by one sub-beam, and the light spot 111c and the light spot 111d are formed by the other sub-beam. Furthermore, when only one light beam is used without branching the light beam, scanning may be performed so that the trajectories of the one light beam overlap.

  A portion where the output signal obtained by this scanning exceeds the slice level is detected as a defect candidate 112. The slice level is indicated by a solid line in the horizontal direction in FIGS. Furthermore, the slice level is set above and below the standard reference level. That is, when the reflectance on the sample surface increases and the amount of light received by the photodetector 110 increases, the output signal exceeds the slice level above the reference level. On the other hand, when the reflectance at the sample surface decreases and the amount of light received by the photodetector 110 decreases, the output signal exceeds the slice level below the reference level. At these times, the position of the defect candidate is specified by the signal from the stage 108, and the defect candidate 112 is detected. If there are two or more defect candidates 112 detected by the adjacent light spot 111 within a predetermined distance, it is determined that a defect exists at that location. A spot determined to have a defect is defined as a defect spot 114. On the other hand, it is determined that no defect exists at a location where only one or less defect candidates 112 detected by the adjacent light spot 111 exist within a predetermined distance. A location where it is determined that no defect exists is defined as a normal location 113. Here, among the defective portions 114, a portion that is actually defective due to foreign matter or the like adhering to the sample is defined as an actual defect 114a, and no foreign matter or the like is attached to the sample, but within a predetermined distance due to noise or the like. A location where two or more defect candidates 112 exist is defined as an erroneously detected defect 114b (pseudo-defect location). In other words, there are two types of defect locations 114: actual defects 114a and false detection defects 114b. A predetermined distance between defect candidates 112 for determining whether or not this defect exists is defined as a defect determination distance.

  First, an output signal in the actual defect 114a in which a foreign substance actually exists on the sample will be described. As shown in FIG. 2, a case is considered in which foreign matter adheres between the scanning lines of the second light spot 111b and the third light spot 111c. The region irradiated by the second light spot 111b and the region irradiated by the third light spot 111c overlap. For this reason, when a foreign substance actually exists in a region where the second light spot 111b and the third light spot 111c overlap, both the light receiving elements corresponding to the second light spot 111b and the third light spot 111c respectively. The output signal exceeds the slice level. That is, the defect candidate 112b is detected in the light receiving element corresponding to the second light spot 112b (FIG. 2B), and the defect candidate 112c is detected in the light receiving element corresponding to the third light spot 112c. (FIG. 2 (c)). Thus, since the actual defect 114a exceeds the slice level at substantially the same position in the scanning line direction, another defect candidate 112c is detected in the vicinity of the defect candidate 112b by the overlapping light spot. In this case, since the distance between the defect candidates 112b and 112c is within the defect determination distance, the portion where these defect candidates 112 exist is determined as the defect location 114.

  Next, the output signal at the normal location 113 will be described. Since noise is generated randomly, the place where it is generated is usually remote. Even if the defect candidate 112 is detected by one light spot, the probability that another defect candidate 112 is detected within the defect determination distance from the position where the defect candidate 112 is detected by the adjacent light spot is very low. This probability changes depending on the setting of the slice level. For example, even if the defect candidate 112a is detected by the light receiving element corresponding to the first light spot 111a (FIG. 2A), the light receiving element corresponding to the second light spot 111b is near the defect candidate 112a. No defect candidates are detected. (FIG. 2 (b)). Therefore, a portion where the output signal exceeds the slice level due to noise or the like is usually determined as a normal portion 113.

  Further, an output signal in the false detection defect 114b will be described. As described above, the probability that two or more defect candidates due to noise occur within the defect determination distance is extremely low. However, since it is difficult to set this probability to 0, two or more defect candidates due to two noises are detected within a defect determination distance with a certain low probability. When a defect candidate 112 due to noise is detected within the defect determination distance, it is determined as a defect location. However, since there is no defect on the sample, this becomes the false detection defect 114b. That is, the false detection defect 114b is a location where noise exceeding the slice level has occurred in the vicinity in an adjacent scanning line. Specifically, when two or more defect candidates 112 due to noise are generated within the defect determination distance in adjacent scanning lines, the location becomes an erroneously detected defect 114b. The probability of occurrence of the false detection defect 114b varies depending on the slice level. That is, when the slice level is lowered, the density of the defect candidates 112 increases, so that the probability of occurrence of the false detection defect 114b increases.

  The pitch of the scanning line is set to be equal to or less than half the spot diameter when the light beam is focused on the sample 107. Accordingly, since the entire surface of the sample is illuminated by two or more light spots, it is possible to perform defect determination on the entire surface of the sample.

  In the case of the present embodiment, if a defect candidate 112 due to another light spot is detected within a predetermined defect determination distance from the position of the detected defect candidate 112 on the sample 107, the defect candidate 112 is The configuration is such that an existing portion is determined as the defective portion 114. For example, FIG. 2 shows an example in which the defect candidate 112b and the defect candidate 112c are detected at the same position in the direction of the scanning line by the adjacent light spots, but the defect candidate detected by the second light spot 111b. If the position of 112b on the sample 107 and the position of the defect candidate 112c by the third light spot 111c on the sample 107 are not the same position in the scanning line direction but are within the defect determination distance, that defect candidate. Is determined as the defective portion 114. As a result, erroneous detection due to noise can be prevented more reliably.

  In the detection of a defect waveform in this embodiment, a band pass filter may be used for noise removal. In this case, the signal waveform for determining the defect is obtained by differentiating the signal waveform shown in FIG. At this time, the signal waveform at the defective portion is, for example, as shown in FIG. FIG. 3 schematically shows a signal waveform of a defective portion where the amount of detection light decreases. When the band pass filter is used, the signal waveform becomes a differential waveform of the change in the detected light amount, and therefore exceeds the upper slice level after exceeding the lower slice level at the above-described defect location. Therefore, when the bandpass filter is used, the slice level is exceeded at two positions at the defect position. Even if the distance between the two locations is within the defect determination distance, this location is not determined as the defect location 114. That is, when there are two locations that exceed the slice level within a predetermined distance in one scanning line, the locations are collectively set as one defect candidate 112. In this case, as indicated by the dotted line in FIG. 3, the position from the position exceeding the lower slice level to the position not exceeding the upper slice level is determined as one defective portion 112. As described above, when a portion exceeding the slice level is within a predetermined distance in one scanning line, it is collectively determined as one defect candidate 112. This predetermined distance is determined based on the time constant of the bandpass filter. By performing this processing, defect candidates can be accurately determined even when a bandpass filter is used. Of course, the above processing may be performed even when no bandpass filter is used or when other noise filters are used. Thus, when two or more defect candidates are detected within a predetermined distance in one scanning line, it is preferable to detect them as one defect candidate.

  The defect determination distance is greater than or equal to the scanning line pitch. Furthermore, it is preferable that the defect determination distance is determined so as not to fail to detect the actual defect 114a in consideration of the scanning line pitch, the positional accuracy of the light beam during scanning, and the like. Thus, by detecting whether or not the defect candidate is within the defect determination distance, it is possible to prevent detection of a pseudo defect due to noise.

  However, if the defect determination distance is too large, the probability that the false detection defect 114b will occur increases. As described above, the erroneously detected defect 114b is a defect in which the distance between defect candidates due to noise is within the defect determination distance in adjacent scanning lines. As shown in FIG. 2, the defect candidate 112d detected by the first light spot 111a and the defect candidate 112e detected by the fourth light spot 111d exist within the defect determination distance. Therefore, although there is actually no defect on the sample, it should be determined as the defective portion 114. That is, the defect candidate 112d and the defect candidate 112e are generated due to noise, but if they exist within the defect determination distance, it is determined as the defect location 114. Therefore, this position becomes the false detection defect 114b.

  When the defect determination distance is large, the defect candidate 112e is included within the defect determination distance of the defect candidate 112d. Therefore, a portion where these defect candidates 112d and 112e exist is determined as the defect location 114. For this reason, it is necessary to set the defect determination distance to an optimum value so as not to fail to detect the actual defect and to reduce the number of erroneous detection defects. That is, the upper limit of the defect determination distance is set so as to reduce the number of erroneously detected defects. By making the defect determination determination distance smaller than the distance between the defect candidate 112d and the defect candidate 112e, the defect candidate 112d and the defect candidate 112e detected in the vicinity thereof can be determined as the normal location 113. As a result, the number of erroneously detected defects 114b can be reduced. For example, the defect determination distance is a distance obtained by adding a position error of the light beam to the radius of the light spot 111. Specifically, a distance obtained by adding the positional deviation amount of the light beam by scanning to the pitch of the scanning line can be set as the defect determination distance.

  Next, an example of a slice level determination method will be described with reference to FIG. In the present embodiment, the maximum sensitivity can be obtained when the density of the erroneously detected defects 114b is set to the permissible limit. A method for performing such setting will be described. That is, since the probability of detecting the defect candidate 112 depends on the slice level, the probability of detecting the false detection defect 114b also depends on the slice level. When the absolute value of the slice level is increased, the probability that the defect candidate 112 and the erroneously detected defect 114b are detected decreases, but the detection sensitivity of the actual defect 114a decreases. On the other hand, when the absolute value of the slice level is reduced, the foreign substance detection sensitivity is improved, but the probability that the defect candidate 112 and the false detection defect 114b are detected is increased. Therefore, when the allowable limit of the density of the false detection defect 114b is set to a predetermined value, the slice level at that time can be determined. Thereby, since the slice level can be lowered, a defect can be detected with high sensitivity. In the following description, the density of the false detection defects 114b is defined as a false detection defect density, and the number of false detection defects 114b is defined as the number of false detection defects.

  Specifically, the optimum value of the slice level for defect candidate determination is determined. The smaller the absolute value of the slice level, the higher the detection sensitivity. However, if the value is too small, the number of defect candidates increases and the number of erroneously detected defects 114b increases. Therefore, it is necessary to determine an optimum slice level in order to improve detection sensitivity. Therefore, when actually inspecting, it is determined how many false detection defects are allowed for one sample. If this is the total number of allowable erroneous detection defects, the density of erroneous detection defects can be obtained from the total number of allowable erroneous detection defects.

  The number of erroneously detected defects is determined from the ratio of the total area of the circle having the radius of the defect determination distance to the total inspection area and the total number of defect candidates. However, in this case, it is necessary to use a defect-free sample or a sample whose number of actual defects 114a is obtained in advance. A sample whose number of actual defects 114a is known in advance is used as a standard sample. Note that a defect-free sample is also included in the standard sample because the number of actual defects 114a is known to be zero. Furthermore, even if there is a region where the number of actual defects is not known in advance in the sample, if there is a region where the number of actual defects is known in advance, the sample is also included in the standard sample. That is, the standard sample only needs to have a region where the number of actual defects is known in advance. The number of actual defects 114a can be obtained based on the location where defect candidates are detected at the same position by scanning the light spot a plurality of times.

  Here, a case where the determination of the slice level is performed using a defect-free sample will be described. In order to determine whether or not there is an actual defect 114a, the same part may be scanned a plurality of times and the defect candidate 112 may not be detected at the same position. When using a sample for which the number of actual defects 114a is obtained in advance, the slice level may be determined by excluding the signal of the actual defect 114a. FIG. 4 illustrates the distribution of defect candidates when a defect-free sample is used. In FIG. 4, small white circles indicate defect candidates 112, and large white circles around them indicate circles having a defect determination distance as a radius. This large white circle is defined as a defect determination area 117.

  As described above, defect candidates 112 are randomly generated due to noise such as thermal noise and shot noise. Assuming that the defect candidates 112 occur randomly, the error defect detection density is calculated from the probability that the two defect candidates 112 occur within a certain defect determination distance, that is, the probability that the defect candidate 112 is determined as the defect location 114. Can be sought.

である。欠陥候補検出数Ｃとは、実欠陥１１４ａ、誤検出欠陥１１４ｂ、正常箇所１１３を含めた全欠陥候補１１２の総数である。ここでは、試料上の実欠陥１１４ａのない部分を走査しているため、実欠陥１１４ａの数は０となる。
また、欠陥判定距離をｒとすると、欠陥判定領域１１７の面積Ｓ_０は、

である。全ての欠陥候補１１２が同様の確率で、欠陥判定領域１１７内に入るとすると、欠陥候補１１２が欠陥判定領域１１７内に入る確率は、

となる。 When the area of the total inspection region 116 of the standard sample inspected for determining the slice level is S and the number of defect candidate detections detected in the total inspection region 116 is C, the number of defect candidate detections per unit area, that is, defect candidates. Density D is

It is. The defect candidate detection number C is the total number of all defect candidates 112 including the actual defect 114a, the false detection defect 114b, and the normal portion 113. Here, since the part without the actual defect 114a on the sample is scanned, the number of the actual defects 114a is zero.
When the defect determination distance is r, the area S ₀ of the defect determination region 117 is

It is. If all defect candidates 112 enter the defect determination area 117 with the same probability, the probability that the defect candidate 112 enters the defect determination area 117 is

It becomes.

  Here, C is the number of defect candidates detected, and F is the number of erroneously detected defects. The number of erroneously detected defects F is an erroneously detected defect in which there are two or more defect candidates 112 due to noise within the defect determination distance and there are no actual defects 114a within the area S, and the defect location 114 is determined. 114b is the total number.

誤検出欠陥１１４ｂはある程度発生するが、面積Ｓにおいて許容することができる誤検出欠陥数Ｆの最大値を許容誤検出欠陥数Ｆ_０とする。式（３）から、許容欠陥候補数Ｃ_０を以下のように求めることができる。

From the above probability, the number of erroneously detected defects F is estimated as shown in Equation (4).

Is erroneously detected defect 114b to some extent occur, the permissible false detection defect number F ₀ the maximum value of the erroneous detection number of defects F can be tolerated in the area S. From equation (3), the allowable defect candidate number C ₀ can be obtained as follows.

Here, the allowable defect candidate number C ₀ is the maximum number of defect candidates that can be allowed in the area S. Here, if the number of defect candidates 112 detected in a specific area S is the allowable defect candidate number C ₀ , the number of erroneous detection defects 114 b is the allowable erroneous detection defect number F ₀ . For example, the allowable number of erroneously detected defects can be set to such a number that one erroneously detected defect occurs in 1 mm square. When false detection defect 114b acceptable false detection defect number F ₀ the number of, it is possible to set the lowest slice level. Therefore, defects can be detected with high sensitivity, and the number of erroneously detected defects can be within a range that is allowed. Thus, a particular defect candidate detection number C in the area S may be set in the slice level so that the acceptable defect number of candidates C _0. That is, the number of detected defect candidate sets the slice level so that the acceptable defect number of candidates C ₀ when it examines the standard sample. Acceptable false detection number of defects F ₀ may, depending on the sample to be inspected is set to an appropriate value. Thereby, a slice level suitable for the sample to be inspected can be set, and highly sensitive detection can be performed.

Specifically, the slice level is set by calculating the number of erroneously detected defects F for each of a plurality of slice levels set in advance. That is, the defect candidate 112 is detected based on each slice level for the output signal when the area S is scanned. Then, the number of erroneously detected defects, that is, the number of erroneously detected defects F is calculated from the defect candidates 112. FIG. 5 shows a graph in which the horizontal axis is the slice level and the vertical axis is the number of erroneously detected defects F at the slice level. When the relationship between the calculated number of erroneous detection defects F and the slice level is plotted on this graph, an inversely proportional graph as shown in FIG. 5 is obtained. Here, the intersection of the predetermined number of allowable false detection defects F ₀ and the inversely proportional graph is determined as a suitable slice level. Of course, the relationship between the slice level and the number of erroneously detected defects F is not limited to inverse proportion.

となる。
このように、許容誤検出欠陥密度Ｄ_０からスライスレベルの決定を行う方法では、試料の全面を検査しなくても試料の一部を検査するだけで、様々な試料に対して簡単に好適なスライスレベルを決定することができる。よって、高感度検出を行うことができるスライスレベルの決定にかかる時間を短縮することができる。好適なスライスレベルを決定することで、高感度で欠陥検出を行った場合でも、誤検出欠陥数を許容値以下とすることができる。 From the above, the allowable false detection defect density D ₀ is

It becomes.
As described above, the method of determining the slice level from the allowable false detection defect density D ₀ is suitable for various samples simply by inspecting a part of the sample without inspecting the entire surface of the sample. The slice level can be determined. Therefore, it is possible to reduce the time required to determine the slice level at which high sensitivity detection can be performed. By determining a suitable slice level, even when defect detection is performed with high sensitivity, the number of erroneously detected defects can be reduced to an allowable value or less.

  A configuration of a processing apparatus for performing the above processing will be described with reference to FIG. FIG. 6 is a block diagram illustrating a configuration of the processing apparatus 200. The processing device 200 is a normal personal computer or the like, and includes a central processing unit (CPU) that performs predetermined processing, and a storage device such as a memory and a hard disk that stores processing results and signal waveforms. Furthermore, a display device for displaying the processing result and signal waveform on the screen and an input device for executing predetermined processing such as a mouse and a keyboard are provided. The processing device 200 includes a slice level storage unit 201, a defect candidate detection unit 202, an erroneous detection defect calculation unit 203, an allowable erroneous detection defect number storage unit 204, a slice level calculation unit 205, an inter-defect candidate distance calculation unit 206, and a defect determination distance. A storage unit 207 and a defect determination unit 208 are included.

First, processing for determining the slice level will be described. In the slice level storage unit 201, several slice levels are input and stored in advance. The plurality of slice levels can be arbitrarily determined. The defect candidate detection unit 202 receives an output signal output from the photodetector 110 while scanning the standard sample. The position where the output signal exceeds the slice level is the defect candidate position. Then, the defect candidate detection unit 202 detects the defect candidate 112 for each slice level in one scan. That is, the position of the defect candidate on the sample is specified for each slice level stored in the slice level storage unit 201. At this time, when there is a portion exceeding the slice level within a predetermined distance in one scanning line, the defect candidate detection unit 202 determines that it is one defect candidate. This is determined by the position (coordinates) exceeding the slice level. Thereby, it is possible to prevent erroneous detection of defect candidates even when a bandpass filter is used. The defect candidate distance calculation unit 206 calculates the defect candidate distance based on the position (coordinates) of the defect candidate 112 detected by the defect candidate detection unit 202. In the erroneous detection defect calculation unit 203, the number of erroneous detection defects F is calculated based on the defect candidate 112 and the defect determination distance. That is, the distance between the defect determination distance stored in the defect determination distance storage unit 207 and the defect candidate 112 calculated by the inter-defect candidate distance calculation unit 206 is referred to in determining whether or not the defect portion 114 is present. Based on the allowable erroneous detection defect number F ₀ stored in the allowable erroneous detection defect number storage unit 204 and the erroneous detection defect number F calculated by the erroneous detection defect calculation unit 203, the slice level calculation unit 205 is Determine the slice level. At this time, as shown in FIG. 5, a relationship between the slice level and the number of erroneously detected defects F with respect to the slice level may be automatically obtained, and a suitable slice level may be set based on the relationship. That may be determined slice level when the allowable false detection number of defects F ₀ on the basis of this relationship. This slice level may be stored in the slice level storage unit 201.

  As described above, the light spot is scanned with respect to the sample 107 to be inspected using the determined slice level. The defect candidate detection unit 202 detects the defect candidate 112 by comparing the output signal from the photodetector 110 with a suitable slice level preset in the slice level determination process. At this time, when there is a portion exceeding the slice level within a predetermined distance in one scanning line, the defect candidate detection unit 202 determines that it is one defect candidate. This is determined by the position (coordinates) exceeding the slice level. This can prevent erroneous detection of the defect candidate 112 even when a noise filter is used. The defect candidate distance calculation unit 206 calculates the distance between defect candidates detected by the defect candidate detection unit 202. The defect determination unit 208 compares the calculated distance between the defect candidates and the defect determination distance stored in the defect determination distance storage unit 207 to determine whether the defect candidate is the defect part 114 or the normal part 113. That is, if the distance between defect candidates is shorter than the defect determination distance, the portion of the defect candidate is determined as the defect location 114.

  Note that the method for determining the slice level is not limited to this, and a method of calculating the average value of the electric signal values for each region and performing defect detection based on the average value (see Japanese Patent Laid-Open No. 60-135707). Or any other method such as a method of calculating the received light amount / light receiving position and improving the detection sensitivity by adding and subtracting the electric signals (see Japanese Patent Laid-Open No. 60-69539). Also good.

  In the present embodiment, the laser beam emitted from one laser light source 101 is configured to form a plurality of sub-beams by the diffraction grating 102. However, the present invention is not limited to this, and may be a multi-beam including a plurality of light sources. Good. A plurality of light spots obtained by focusing a plurality of beams may be scanned all at once, or may be scanned independently. Even if scanning is performed independently, the position of each light spot can be stored and the defect can be determined by converting the coordinates to the position on the sample. Further, even if scanning is not performed so that the light spots by the light beams from the adjacent light sources overlap, it is sufficient to scan so that the trajectories of the light spots overlap.

  Further, even if a plurality of light spots are not used, scanning may be performed so that the scanning regions overlap with one light spot. What is necessary is just to make the pitch of the scanning line which scans a light spot into half or less of the spot diameter of a light spot. As a result, scanning can be performed so that the light spots overlap in adjacent scanning lines, and detection sensitivity can be improved.

  The scanning method may be a raster scan in which the laser beam is scanned in the ± X direction with respect to the surface of the sample, or the sample may be rotated to scan the laser beam concentrically or spirally. Scanning may be performed by setting the beam pitch so that adjacent light spots overlap each other. In this case as well, the coordinate system may be converted onto the sample, and the positions of the defect candidates on the sample may be compared.

  In the above description, the reflected light reflected by the sample has been described. However, defect detection can be performed similarly for the transmitted light that passes through the sample. That is, the transmitted light is detected by a photodetector and the same processing is performed based on the output signal. Thereby, it can test | inspect based on the transmitted light. Furthermore, you may detect the scattered light scattered in the defect location of a sample. Thereby, a defect inspection can be performed accurately.

  By using a semiconductor wafer in which a defect is detected by the defect detection method according to the present invention, it is possible to improve the manufacturing yield of a semiconductor device or the like. That is, a defect of the semiconductor wafer is detected by this defect detection method, and the defect is corrected based on the detection result. Then, a desired pattern is formed on the semiconductor wafer whose defect has been corrected, and a semiconductor device is manufactured. This pattern is formed by a widely known process such as a thin film deposition process, an etching process, an oxidation process, or an ion implantation process. Thus, a pattern substrate such as a semiconductor device is formed. A desired pattern can be formed on a semiconductor wafer from which a defect has been detected using the defect detection apparatus or the defect detection method of the present invention, and an exposure process in the manufacture of a semiconductor device can be performed. With the defect detection method of the present invention, it is possible to improve the manufacturing yield of not only a semiconductor device but also a pattern substrate. Therefore, the productivity of the pattern substrate can be improved.

It is the schematic which shows the structure of an example of the test | inspection apparatus concerning this embodiment. It is a figure explaining determination of the real defect of this embodiment. In determination of the real defect of this embodiment, it is a figure which shows an example of the signal waveform in a defect location. It is a figure explaining an example for determining a slice level. In the inspection method concerning this invention, it is a figure which shows the relationship between and a slice level. It is a block diagram which shows an example of a structure of the processing apparatus used for the test | inspection apparatus concerning this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Defect detection apparatus, 101 Laser light source, 102 Diffraction grating, 103 1st relay lens, 104 2nd relay lens, 105 Beam splitter, 106 Objective lens, 107 Sample, 108 Stage, 109 3rd relay lens, 110 Light Detector,
111 light spot, 112 defect candidate, 113 normal defect, 114 actual defect location,
114a actual defect, 114b false detection defect, 116 inspection area, 117 defect determination area,
200 processing apparatus, 201 slice level storage unit, 202 defect candidate detection unit,
203 erroneous detection defect calculation unit, 204 allowable erroneous detection defect number storage unit,
205 slice level calculation unit, 206 defect candidate distance calculation unit,
207 Defect determination distance storage unit, 208 Defect determination unit

Claims (15)

A light source that generates a light beam;
An objective lens for condensing the light beam emitted from the light source to form a light spot on the sample surface;
Of the light beam incident on the sample from the objective lens, the reflected light reflected by the sample surface, scattered light from the defect portion of the sample or transmitted light transmitted through the sample is received, and the amount of the received light is changed. A photodetector that outputs an output signal based thereon;
Scanning means for scanning a relative position between the sample and the light spot along a scanning line, wherein the light spot scans so as to illuminate a region overlapping with the light spot of an adjacent scanning line. When,
A defect candidate detector for detecting a defect candidate based on an output signal from the photodetector;
An inspection apparatus comprising: a defect determination unit that determines whether or not a defect is based on a distance between defect candidates detected by the defect candidate detection unit.   The inspection apparatus according to claim 1, wherein the scanning unit sets a pitch between scanning lines of the light spot to be half or less of a spot diameter of the light spot. A diffraction grating for converting the light beam emitted from the light source into a plurality of sub-beams;
On the sample surface, a plurality of light spots are formed so that the trajectory of the sub beam overlaps the trajectory of the adjacent sub beam,
The inspection apparatus according to claim 1, wherein the photodetector has a plurality of light receiving elements corresponding to the plurality of sub beams. A plurality of the light sources;
On the sample surface, a plurality of light spots are formed such that the trajectory of the light beam from the light source overlaps the trajectory of the light beam from the other light source,
The inspection apparatus according to claim 1, wherein the photodetector has a plurality of light receiving elements corresponding to light beams from the plurality of light sources.   The defect determination unit corresponds to a second scan line within a defect determination distance from a position on the sample of the defect candidate detected by the light spot corresponding to the first scan line among the scan lines. The inspection apparatus according to claim 1, wherein it is determined whether or not the defect candidate is detected based on whether or not the defect candidate detected by the light spot exists.   The inspection apparatus according to claim 1, wherein the defect candidate detection unit detects the defect candidate at a position where an output signal exceeds a slice level.   The inspection apparatus according to claim 6, wherein the defect candidate detection unit detects, as one defect candidate, when there are two or more positions within the defect determination distance that exceed the slice level in one scanning line.   In a standard sample in which the number of defects actually present on the sample surface is known, the defect candidates are detected, and the number of erroneously detected defects erroneously determined as the defects is calculated based on the number of defect candidates. The inspection method according to claim 7, wherein the slice level is set based on the number of the false detection defects. Emitting a light beam;
Condensing the light beam to form a light spot on the sample surface and irradiating the sample;
Scanning the relative position of the sample and the light spot along a scan line so that the light spot illuminates a region overlapping with the light spot of an adjacent scan line;
Detecting a reflected light reflected from the surface of the sample, a scattered light from a defect portion of the sample, or a transmitted light transmitted through the sample, among light beams incident on the sample;
Outputting an output signal based on the detected reflected light, scattered light or transmitted light;
Detecting defect candidates based on the output signal;
And a step of determining whether or not the defect is based on a distance between the detected defect candidates.   The inspection method according to claim 9, wherein in the scanning step, scanning is performed with a pitch between scanning lines of the light spot being equal to or less than half of a spot diameter of the light spot. A plurality of the light spots are formed on the sample surface,
The inspection method according to claim 9 or 10, wherein the plurality of light spots are detected by a plurality of light receiving elements corresponding to the plurality of light spots.   12. The inspection method according to claim 9, wherein the defect candidate is detected at a position where the output signal exceeds a slice level in the step of detecting the defect candidate.   13. The inspection method according to claim 12, wherein when there are two or more positions within the predetermined distance in one scanning line, the position is detected as one defect candidate. Detecting the defect candidates in a standard sample in which the number of defects actually present on the sample surface is known in advance;
Estimating the number of erroneously detected defects that are erroneously determined as the defect based on the number of defect candidates in the standard sample; and
The inspection method according to claim 12 or 13, further comprising the step of setting the slice level so that the number of the erroneously detected defects is equal to or less than the allowable number of erroneously detected defects. Detecting a defect of the substrate, which is the sample, by the inspection method according to claim 9;
Correcting the defects detected by detecting the defects;
Forming a pattern on the substrate in which the defect is corrected.

Images