JP3421522B2 - Pattern inspection apparatus, pattern inspection apparatus using electron beam, and method therefor - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection apparatus for detecting a defect in a pattern of an object to be inspected such as a wafer in which a large number of identical pattern groups are regularly arranged, and more particularly to the appearance of a semiconductor memory device. The present invention relates to an electron beam pattern inspection apparatus suitable for inspection, a method thereof, and a process abnormality diagnosis system.

[0002]

2. Description of the Related Art As the first conventional technique, Japanese Patent Laid-Open No. 59-
As described in No. 192943, while an inspected object having a repeating pattern such as a semiconductor memory is moved at a constant speed, an image of the inspected object is optically detected and detected by an image sensor such as a line sensor. The optical image signal is compared with the optical image signal whose time is delayed by the pitch of the repeated pattern, and the mismatch is recognized as a defect. That is, the detection images of adjacent cells or adjacent chips on the same wafer are compared with each other, and different portions are recognized as defects (hereinafter, these are compared in order with adjacent cell comparison, adjacent chip Called comparison).
As the second conventional technique, not the adjacent pattern but the optical detection image of the sample known to be defect-free in advance is stored as a standard image, and this is compared with the optical detection image of the inspection object. Methods are also known. Japanese Patent Laid-Open No. 1-287449 discloses an inspection apparatus having a function of storing a standard image and then correcting (adding, deleting, etc.) the standard image on the screen of a CRT, if necessary. ing. Note that these are all related to the inspection technology based on the optical system.

[0003]

In the first prior art described first, since the adjacent cells or chips on the same wafer are compared and inspected with each other, defects which are not uniformly formed on the wafer are irregular. Can be detected, but there is a problem in that defects that uniformly occur on the entire surface of the wafer cannot be detected. For defects that occur uniformly on the entire wafer surface,
For example, there are defective pattern resolution due to improper setting of exposure conditions, interlayer short-circuit or open caused by misalignment between layers, and defective pattern resolution due to defocus during exposure. It is difficult to detect these defects not only by the adjacent cell comparison method but also by the adjacent chip comparison method, because the conditions are similar at the sites on the same wafer that are close in distance. In a semiconductor process using a conventional wafer pattern inspection apparatus of the adjacent cell comparison system or the adjacent chip comparison system, defects generated on the entire surface of the wafer are overlooked, which causes a problem that the yield is reduced. Further, in the second conventional technique, since the optical detection image is compared with the standard image obtained from the defect-free sample,
It should be possible to detect defects that occur non-uniformly and irregularly on the wafer, or defects that occur uniformly on the entire surface of the wafer, but this is not the case. The comparison with the standard image obtained from the defect-free sample means the comparison with the detection image of another wafer. If the wafer is different, the film thickness and the positional relationship with the lower layer pattern are slightly different.Therefore, when the standard image and the optical detection image obtained from the wafer to be inspected have the brightness or the lower layer pattern can be seen through (actually, In the optically detected image, in many cases, the lower layer pattern and the edges of the lower layer pattern appear in the detected image). The positional relationship between the lower layer pattern and the upper layer pattern is different. Therefore, when the two images are compared, there is a problem that mismatches due to different wafers frequently occur, and it is impossible to distinguish from mismatches due to defects. In other words, if the mismatches due to different wafers are not erroneously recognized as defects and the defects with a certain mismatch amount or less are not judged as defects, there is a problem that only some defects can be detected.

An object of the present invention is to solve the above problems.
A large number of identical pattern groups are regularly arranged on a target object to be inspected, such as a wafer, to detect defects that occur non-uniformly and defects that occur uniformly on the entire surface of the target object at high speed. It is an object of the present invention to provide a pattern inspection apparatus, an electron beam pattern inspection apparatus, and a method thereof that can perform inspection with high reliability. Another object of the present invention is to provide a process abnormality diagnosis system capable of performing abnormality diagnosis of a process for manufacturing an object to be inspected such as a wafer in which a large number of identical pattern groups are regularly arranged. To do.

[0005]

In order to achieve the above object, the present invention provides an electron gun for emitting an electron beam and an electron beam emitted from the electron gun, wherein a plurality of identical pattern groups are regularly arranged. A lens system for focusing the arrayed inspection target, a deflector for scanning the focused electron beam, a detector for detecting electrons generated from the inspection target, and an electron obtained from the detector. An A / D converter that converts an image signal into a digital image signal, and a feature amount of a pattern at a specific location on the inspection target is calculated from the digital image signal obtained from the A / D converter, and the calculated feature amount is calculated. Calculation to judge conformity / nonconformity to the standard of non-defective products based on the feature amount of the pattern, and further compare the image signals according to the pattern group obtained from the detector to extract the non-coincidence portion and detect the defect Means and Is a pattern inspection apparatus using an electron beam, characterized in that there was e. The present invention also provides an electron gun for emitting an electron beam, a lens system for focusing the electron beam emitted from the electron gun on an object to be inspected in which a plurality of identical pattern groups are regularly arranged, and a focusing system. A deflector for scanning the generated electron beam, a detector for detecting electrons generated from the object to be inspected, an A / D converter for converting an image signal by the electrons obtained from the detector into a digital image signal, From the digital image signal obtained from the A / D converter, the feature amount of the pattern at the specific location on the inspection target is calculated, and based on the calculated feature amount of the pattern, conformity / nonconformity with the standard of non-defective product is obtained. By an electron beam characterized by including a calculation means for determining the defect, and a defect determination means for detecting a defect by comparing the image signals corresponding to the pattern group obtained from the detector to extract a mismatched portion. Patta It is an emissions inspection apparatus.

Further, the present invention is characterized in that the pattern inspection apparatus using the electron beam further comprises an acceleration voltage control means for controlling acceleration of the electron beam incident on the inspection object. Further, according to the present invention, in the calculation means in the pattern inspection apparatus using the electron beam, the designation of the specific portion on the inspection target is made to correspond to the correlation between the digital image signal obtained from the A / D converter and a desired standard pattern. It is characterized by performing based on. Further, the present invention is characterized in that, in the calculation means in the pattern inspection apparatus using the electron beam, the calculation of the feature amount of the pattern based on the detection image at the predetermined location uses a detection image having a higher magnification than in the comparison inspection. . Further, according to the present invention, an electron beam emitted from an electron gun is focused and irradiated by a lens system to an object to be inspected in which a plurality of identical pattern groups are regularly arranged, and the electron is generated from the object to be inspected. A step of detecting an image signal by means of an electron detecting a detector, and converting the detected image signal by means of an electron into a digital image signal by means of an A / D converter, and converting the converted digital image signal onto the object to be inspected. The feature amount of the pattern at a specific location is calculated, and the conformity / non-conformance with the standard of the non-defective product is determined based on the calculated feature amount of the pattern, and the image signals according to the converted pattern group are compared with each other. And an inspecting step of detecting a defect by extracting a non-coincidence portion with each other.

According to the present invention, a detector for detecting an image signal from an object to be inspected in which a plurality of identical pattern groups are regularly arranged, and an image signal obtained from the detector are converted into a digital image signal A. A / D converter, and a feature amount of a pattern at a specific portion on the inspection object is calculated from a digital image signal obtained from the A / D converter, and a non-defective standard is calculated based on the calculated feature amount of the pattern. Comparing with the calculation means for determining conformity / non-conformity, and defect determining means for comparing the image signals corresponding to the pattern groups obtained from the detector to extract the non-matching portion and detecting the defect. A characteristic pattern inspection apparatus. Further, the present invention is
An electron gun that emits an electron beam, a lens system that focuses the electron beam emitted from the electron gun on an object to be inspected in which a plurality of identical pattern groups are regularly arranged, and a focused electron beam A deflector for scanning, a detector for detecting electrons generated from the object to be inspected, an A / D converter for converting an image signal by the electrons obtained from the detector into a digital image signal, and the A / D conversion Calculating means for calculating the feature amount of the pattern at the specific portion on the inspection object from the digital image signal obtained from the instrument, and judging conformity / non-conformity with the standard of non-defective product based on the calculated feature amount of the pattern. A pattern inspecting device using an electron beam provided with, and further systematically accumulating the feature amount of the pattern calculated by the calculation means of the pattern inspecting device, and monitoring the variation of the feature amount for abnormality diagnosis of the process An abnormality diagnosis system of a process which is characterized in that it comprises a device.

The present invention is also directed to a wafer in which a plurality of identical pattern groups are regularly arranged, and a pattern inspection method for scanning a wafer and detecting a pattern defect using a detection image obtained by an image pickup means. Defects that occur irregularly and irregularly on the wafer are extracted by comparing detected images of a plurality of pattern groups on the same wafer and searching for a non-matching portion, and similarly occur all over the wafer. The defect is a pattern inspection method characterized by extracting by calculating the feature amount of the pattern of the detected image at a predetermined location. Further, in the pattern inspection method of the present invention, a comparison inspection in which detected images of a plurality of pattern groups on the wafer are compared and a mismatched portion is pointed out as a defect is performed in a wide range on the wafer to detect the predetermined portion. The feature amount calculation of the pattern based on the image is performed at several points on the wafer.
The feature is that the feature amount is calculated in a narrow range near the pattern. Further, according to the present invention, the output signal from the image pickup means is branched to compare the detection images of a plurality of pattern groups on the same wafer and to calculate the feature amount of the pattern using the detection images at a predetermined location in parallel. It is characterized by performing.
Further, the present invention, the inspection by comparing the detection images of a plurality of pattern groups on the wafer is performed by dedicated hardware,
The calculation of the feature amount of the pattern based on the detected image of the predetermined portion is performed by software processing by a computer.

As described above, according to the above-mentioned configuration, defects and non-uniformly generated defects are generated on the inspection object such as a wafer in which a large number of identical pattern groups are regularly arranged, and the entire inspection object. It is possible to inspect at high speed and with high reliability for defects that uniformly occur in the above. That is, based on an image detected by an electron beam that has a good material contrast, that is, it is easy to set rules, the adjacent cell comparison method or the adjacent chip comparison method can accurately detect defects that occur irregularly and irregularly in the wafer. In addition, by calculating the feature amount such as the pattern size and positional relationship using the detection image of a predetermined position on the wafer, the resolution of the pattern may be poor, and the positional deviation between the upper layer pattern and the lower layer pattern may occur. It is possible to detect with high reliability the defects that similarly occur on the entire surface of the wafer. further,
If the accelerating voltage is properly selected, the penetration depth of the irradiation electron beam can be made shallow so that it does not reach the lower layer pattern, so it is possible to make the detection image a simpler pattern that includes only the uppermost layer pattern. Become. Further, by selecting an accelerating voltage such that the irradiation electron beam penetrates to the layer below the uppermost layer, it becomes possible to easily obtain the characteristic amount relating to the positional relationship between the uppermost layer and the layer below it.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of a pattern inspection for a semiconductor substrate such as a wafer in which a large number of the same pattern groups forming a semiconductor memory device according to the present invention are regularly arranged will be described with reference to the drawings. To do. In a semiconductor substrate such as a wafer in which a large number of the same pattern groups forming a semiconductor memory device or the like are regularly arranged, a defect that occurs nonuniformly without a regularity and a defect that occurs uniformly over the entire surface occur. . Factors that cause defects that occur uniformly over the entire surface are, for example, poor resolution of patterns due to improper setting of exposure conditions, interlayer shorts and opens caused by misalignment between layers, and defocus during exposure. There is a poor resolution of the pattern due to. Also, in the case of semiconductor substrates such as wafers, since they are formed with a thin film multilayer pattern, in the optical detection image, it is possible to see through to the lower layers below, or the steps appear dark, resulting in a complicated pattern. Tend to be. On the other hand, in the case of an image detected by an electron beam, since the amount of the electron beam generated basically depends on the composition (atomic number), different substances have different brightness, which is a good material contrast. An image that is easy to set can be obtained.

First, a pattern inspection on a semiconductor substrate such as a wafer using an electron beam according to the present invention will be described. Figure 1
The principle of pattern inspection on a semiconductor substrate such as a wafer by an electron beam according to the present invention will be described. Reference numeral 1 indicates a semiconductor substrate such as a wafer on which a pattern to be inspected is repeatedly formed. Reference numeral 25 denotes an image pickup device (detector) that detects electrons generated from the semiconductor substrate 1 by irradiation with an electron beam and obtains an electron beam image of a scanning region based on the intensity change. Reference numeral 8 denotes an A / D converter that converts an electron beam image detected by the image pickup device (detector) 25 into a digital image signal. 9 is A /
This is a delay circuit for storing the digital detection image of the adjacent cell or the adjacent chip in the digital detection image by the electron beam obtained from the D converter 8 and delaying the interval between the cells or the chips. The comparison / defect determination circuit 10 is an adjacent cell comparison method for comparing the digital detection image of the adjacent cell or the adjacent chip obtained from the delay circuit 9 and the digital detection image obtained from the A / D converter 8 or the adjacent chip comparison method. Is performed to detect defects that occur irregularly and irregularly in the wafer. In this adjacent cell comparison method or adjacent chip comparison method, since the distances on the same wafer are close to each other, there is almost no difference between the detected images due to the difference in film thickness or the positional relationship with the lower layer pattern. Absent. Therefore, since the difference between the detected images to be compared is basically only the defective portion, the defect can be detected with high accuracy.

Further, the image processing computer 15 calculates the feature quantity of the pattern by using the detected image of the predetermined portion on the wafer selected by the predetermined portion selecting means 13,
Defects that occur uniformly on the entire surface of the wafer are detected. That is, the image processing computer 15 uses the detected image of a predetermined portion on the wafer to calculate the feature amount such as the dimension and the positional relationship of the pattern, so that the resolution of the pattern is poor and the upper layer pattern and the lower layer pattern are separated. It is possible to detect a defect that similarly occurs on the entire surface of the wafer, such as displacement.

In the calculation of the feature quantity of this pattern, it is difficult to set a rule for calculating the feature quantity if the pattern is complicated. Therefore, the image pickup device (detector) 25 is optimized to detect the detected image. Try to have a simple pattern. Specifically, the wafer is scanned with an electron beam, the electrons generated from the wafer by the irradiation of the electron beam are detected, and an image of the scanned portion (hereinafter referred to as an electron beam image) is obtained based on the intensity change. To An optically detected image of a thin-film multi-layer pattern such as a wafer tends to be a complicated pattern because the lower layers under many layers can be seen through or dark steps can be seen. On the other hand, in the case of electron beam images, the amount of generated electrons basically depends on the composition (atomic number). An image that is easy to set can be obtained. Furthermore, if the accelerating voltage is appropriately selected, it is possible to make the penetration depth of the irradiation electron beam shallow so that it does not reach the lower layer pattern, so it is possible to make the detection image a simpler pattern containing only the uppermost layer pattern. It will be possible. Alternatively, by selecting an accelerating voltage such that the irradiation electron beam penetrates to a layer below the uppermost layer, it becomes possible to easily obtain the characteristic amount related to the positional relationship between the uppermost layer and the layer below it.

Next, a specific description will be given with reference to FIGS.

FIG. 2 (a) shows that a semiconductor substrate such as a wafer is scanned with an electron beam to detect electrons generated from the semiconductor substrate by the irradiation of the electron beam, and an electron beam image of the scanned portion is detected based on the intensity change. 1 is a configuration diagram showing an embodiment of an apparatus for obtaining a pattern and performing a pattern inspection using an electron beam image. Figure 2
FIG. 3B is a perspective view showing the surface of a semiconductor substrate such as a wafer on which a pattern inspection is performed using an electron beam image. Figure 2 (a)
, The electron beam emitted from the electron gun 21 is
After 2 and 23, the beam diameter is narrowed to a pixel size on the sample surface. When irradiated with electron beam, sample (wafer 1)
Generates electrons. A two-dimensional electron beam image of the sample is detected by detecting electrons generated from the sample in synchronization with repeated scanning of the electron beam in the X direction by the deflector 24 and continuous movement of the sample in the Y direction by the stage 7. Is obtained. The electrons generated from the sample are captured by the detector 25, amplified by the amplifier 26, and then converted into a digital signal S11 by the A / D converter 8. The digital signal S11 is
The delay circuit 9 electrically delays a semiconductor substrate such as a wafer by one chip or one cell,
In the comparison / defect determination circuit 10, the original signal S11 obtained from the A / D converter 8 is compared. 2a, 2b
2 is an adjacent chip on the semiconductor substrate 1 such as a wafer as shown in FIG. 2B. However, if the time delayed by the delay circuit is set to the time required for the semiconductor substrate such as a wafer to move by one chip, Signal S11 of and the output signal S1 of the delay circuit 9
2 corresponds to image signals of adjacent chips, for example, 2a and 2b, which is equivalent to comparing detected images of 2a and 2b.

The comparison / defect determination circuit 10 has a logic realized by an electric circuit, and detects a mismatched portion by pipeline processing. For example, if the difference image is obtained after the edges of the image signals S11 and S12 are extracted and aligned, the non-coincidence portion can be detected. In addition, various algorithms have been proposed to detect the non-coincidence portion, and a suitable algorithm may be implemented as hardware. The reason why the processing is performed by hardware instead of software processing on a computer is to increase the inspection speed. Because,
The purpose of this comparative inspection is to detect defects that occur irregularly and irregularly. Since it is advantageous to inspect a wider area to detect defects that occur irregularly and irregularly, the inspection speed must be high. This is because it must be done. Therefore, compare
Even if the defect determination circuit 10 does not obtain defect information, the mismatch portion (defect candidate) is temporarily stored in the memory, and the defect candidate stored in the memory is subjected to detailed analysis in the computer to determine a true defect (it is necessary to determine the defect). Since it can be inspected at high speed, it may be realized by software processing in a computer. The defect information obtained here is used as a system control computer 11 for controlling the entire apparatus.
Sent to.

On the other hand, when the wafer reaches the coordinates specified in advance (the system control computer 11 manages the coordinate data), the detected image (image signal S12) is taken into the storage device 14a or 14b (stored). Switching to the devices 14a and 14b is performed by controlling the input switching circuit 13 by the system control computer 11.) The image processing computer 15 reads the detected image stored in the storage device 14a or 14b, calculates the pattern feature amount (dimension, positional relationship, etc.), and determines whether the feature amount conforms to the conforming product standard. It should be noted that the two storage devices 14a and 14b are provided because while the image processing computer reads and processes an image from 14a,
When 4b is used for image capturing and image reading and processing are completed, this time, 14a is switched to image capturing and 14b is switched to image reading, and the image processing computer 15 stores the image already stored in 14b. By using the method of reading and starting
This is to eliminate the inconvenience that the semiconductor substrate such as a wafer cannot move to the next detection position unless the processing of the image processing computer 15 is completed. When the amount of processing in the image processing computer 15 is large, or when there are many locations where the feature amount should be calculated and this configuration is too late, it is possible to increase the number of image processing computers and storage devices. The calculation of the characteristic amount can be realized by hardware (depending on the content of the characteristic amount), but in the present embodiment, it is performed by the software processing on the image processing computer 15 as described above. . The reason for this is that the purpose of feature amount calculation is to detect defects that occur uniformly over the entire surface of the wafer, so even if the entire surface of a semiconductor substrate such as a wafer is not targeted, for example, even at a frequency of about one location per chip. This is because the software processing capable of flexibly changing the image processing content is more suitable than the high-speed processing because the effect can be expected sufficiently. For example, “one place” is 50 μm * 50 μ.
The range of m is one image detected with a pixel size of 0.1 μm. The method of calculating the feature amount will be described later with reference to FIGS.

Further, a grid electrode installed on the upper surface of the stage 7 and the surface of the inspection object (semiconductor substrate such as wafer) 1 or the position close to the surface of the inspection object (semiconductor substrate such as wafer) 1 The effective acceleration voltage is optimized by applying a voltage from the high-voltage power supply 27 during this period. Even if the voltage applied to the electron gun 21 is changed, the accelerating voltage can be changed, but automatic adjustment is difficult because the electron optical system needs to be readjusted. Here, since it is desired to change the accelerating voltage depending on the material and structure of the object to be inspected, the above-described configuration is possible in which automatic control is possible. The effect of optimizing the acceleration voltage is schematically shown in FIG. The object to be inspected is a multi-layer pattern composed of three layers, in which a layer, b layer, and c layer are laminated in this order. By lowering the acceleration voltage,
By making the penetration depth of the irradiation electron beam shallow so as not to reach the lower layer pattern, a pattern in which the c layer on the right side of FIG. 3 is exposed can be obtained. When the pattern of the c-layer can be revealed, not only the inspection of the c-layer can be performed with high accuracy, but also the feature amount of the pattern described later can be easily calculated.

Even in a pattern inspection apparatus for a semiconductor substrate such as an electron beam type wafer, the purpose of the comparison inspection is to detect defects that are non-uniformly and irregularly generated on an object to be inspected (semiconductor substrate such as a wafer). Therefore, a wider range of inspection is required to increase the defect capture rate. High-speed detection of an electron beam image is a key to performing a wide range inspection in a realistic inspection time. Hereinafter, constituent elements suitable for high-speed detection of an electron beam image will be described with reference to (a) to (d). (A) As a method for obtaining a two-dimensional electron beam image, there is also a method in which the stage is fixed and the electron beam is two-dimensionally scanned like a scanning electron microscope. However, because the step and repeat of the stage is required to move the image detection point,
Large time loss due to stage acceleration / deceleration. Therefore, the one-dimensional electron beam scanning + continuous stage movement method is more advantageous. (B) Deflector 24 that repeatedly scans the electron beam in the X direction
An electrostatic deflector is used for. Magnetic field deflectors commonly used in scanning electron microscopes are not suitable due to their slow deflection speed. (C) As the electron gun 21, a thermal field emission type (TFE: Th
ermal Field Emission) Use an electron gun. The TFE electron gun is a field emission type (F type) commonly used in scanning electron microscopes.
E: Field Emission) Since the electron beam current can be increased as compared with the electron gun, the electron beam irradiation time can be shortened = the electron beam scanning speed can be increased (the S / N of the image is per unit area). It is proportional to the 1/2 power of the electron beam irradiation.Therefore, in order to obtain the necessary S / N value, there is a relation that the irradiation time must be lengthened if the electron beam current is small). Further, since the TFE electron gun has a smaller time variation of the electron beam current than the FE electron gun, the detection images at different detection times are compared with each other. Minute or the time of moving one cell is compared with an image detected in the past.) (D) As the detector 25, a semiconductor detector that can be driven at high speed is used.

Next, the calculation of the pattern feature amount performed by the image processing computer 15 will be described. Since the calculation of the feature amount of the pattern is intended to detect a defect that similarly occurs on the entire surface of the wafer, it is necessary to select an appropriate feature amount according to the pattern shape and the type of defect to be detected. . Some examples are shown below. For example, if the purpose is to detect a resolution defect that occurs due to an inappropriate exposure condition of the exposure device, the resolution defect often causes a change in the pattern dimension. One of the guidelines is d1 and the pattern interval d2. By image processing,
There are various possible methods for calculating the dimensions from the detected image,
For example, the absolute values of d1 and d2 can be known by separating the pattern area and the background area according to the difference in brightness, searching for the minimum width portion, and counting the number of pixels of that width. Alternatively, the rough situation can be grasped only by examining the area ratio (pixel number ratio) of the pattern region and the background region after the region separation. Even if the values of d1 and d2 deviate from the standard, if adjacent cells and chips also deviate from the standard, it cannot be detected by the comparison inspection. In this way, using the detected image, d1 and d2 The occurrence of defective resolution can be detected by examining the absolute value of.

Alternatively, if a resolution defect occurs in the step of forming the contact hole, the diameter of the contact hole changes. FIGS. 4A and 4B show contact holes for connecting the upper layer (solid line) 31 and the lower layer (dotted line) 32. The diameter (diameter d1) of the contact hole 33 shown in FIG. 4 (a) is smaller than that of the normal contact hole (diameter d0) 33 shown in FIG. The calculation of the diameter of the contact hole in the image processing computer 15 is performed by the storage devices 14a, 14
The contact hole portion 3 is obtained by a process of searching for a closed circular portion after the edge is extracted from the detected image read from any one of b, for example, by performing a differentiation process.
Select 3. After that, the frequencies 34 and 35 of the number of pixels projected onto the X axis and the Y axis are taken, and the frequency 34 in the Y axis direction is taken.
And the frequency 35 in the X-axis direction may be averaged to obtain the diameter, or the number of pixels (area) S inside the ring may be counted to obtain the diameter d and back calculated based on √ (4S / π). Is also good. Such an abnormality in the contact hole diameter often occurs in the same manner in a wide area on the wafer, and therefore cannot be detected by the adjacent cell comparison method or the adjacent chip comparison method. The occurrence of defective resolution can be detected by examining the absolute value of the diameter in the image processing computer 15.

Further, FIGS. 5A and 5B show a state of misalignment. Although FIG. 5B shows a normal state, in FIG. 5A, the c layer and the d layer are relatively displaced with respect to the a layer and the b layer, and as a result, the interlayers of the a layer and the d layer are displaced. There is a short circuit. To detect this, in the same way as in the above-described embodiment, the image processing computer 15
, Finds a great circle or a small circle in the detected image read from either the storage device 14a or 14b, calculates the center coordinates of each circle, and calculates the distance between the center coordinates of the great circle and the small circle. You should ask. Then, the image processing computer 15 determines that the value is normal if the value is within the specified value and abnormal if the value exceeds the specified value. Further, in order for the image processing computer 15 to detect the resolution defect more directly, the degree of edge sag is determined by using the high-magnification detection image of the resist pattern stored in any of the storage devices 14a and 14b. You can also look up. 6A and 6B show a case where the edge of the resist 51 is small, FIG. 6A shows a detection image in that case, and FIG. 6B shows A- shown in FIG. A'shows a resist cross section viewed from the arrow. 6C and 6D show a case where the edge of the resist 51 has a large sag, FIG. 6C shows a detected image in that case, and FIG. 6D shows B- shown in FIG. 6C.
B'shows a resist cross section viewed from the arrow. As shown in FIGS. 6 (c) and 6 (d), when the edge largely deviates, the width of the edge portion indicated by the halftone dot 52 in the detected image widens, so that the image processing computer 15 determines the halftone dot based on the detected image. If the width dimension g of the portion 52 is obtained, the quality of the resolution state can be determined. However, since the resist pattern is often transparent, the inspection itself is difficult with an optical inspection apparatus, and it is possible only with an electron beam inspection apparatus.

Next, the search for the feature quantity calculation pattern performed by the system control computer 11 with respect to the image processing computer 15 will be described. The image acquisition position for calculating the feature amount is managed by the system control computer 11, and the image is detected by the detector 25 when the semiconductor substrate such as the wafer reaches the designated coordinates. Therefore, there is an error in the position on the image of the pattern for calculating the feature amount. For example, if the pixel size of the detected image is 0.1 mm and the position accuracy of the stage is ± 1 mm, the position of the pattern for which the feature amount is calculated will have an error of ± 10 pixels on the image. It does not matter if the feature amount of any pattern on the image can be calculated (such as when the same pattern is repeated), but if the feature amount of a specific pattern needs to be calculated, Positional error is a problem. To meet this need, the image processing computer 15
A standard pattern in which a calculation point is designated is stored in advance, and the calculation point is obtained by the correlation between the standard pattern and the detected image. In this way, the image processing computer 15 compares the standard pattern of the calculation location stored in the memory with the detected image detected by the A / D converter 8 and calculates the correlation point by calculating the correlation. Points, and the detected images at the calculated points are stored in the storage devices 14a and 14a.
The image processing computer 15 may read out from any one of b and calculate the feature amount. Next, the calculation of the feature amount of the pattern width performed in the image processing computer 15 will be described with reference to FIG. FIG. 7A shows a sectional shape in which a resist 71 is formed on a semiconductor substrate 72 such as a wafer.
FIG. 7B shows a digital detection image by an electron beam detected from the resist pattern shown in FIG.
That is, as shown in FIG. 7B, the digital detection image at a predetermined location is stored in either of the storage devices 14a and 14b. The image processing computer 15 uses a plurality of scanning lines in the width direction of the resist pattern (JJ ′ shown in FIG. 7B,
KK ′) digital detection images are added, and FIG.
The digital image signal shown in is obtained. In this way, in the image of the electron beam, the edge portion is detected bright, which is generally called the edge effect. Next, the image processing computer 15
Of the digital image signal shown in FIG. 7C is sequentially cut into, for example, a 3 × 3 pixel group, and the cut 3
FIG. 7 shows the smoothing process performed by averaging the digital image in the × 3 pixel group and outputting the digital image signal as the central pixel of the 3 × 3 pixel group.
A digital image signal from which the noise component shown in (d) is removed is obtained. Then, the image processing computer 15 is shown in FIG.
MAX value (maximum value) and MI of the digital image signal shown in
The N value (minimum value) is calculated, and based on the calculated MAX value (maximum value) and MIN value (minimum value), for example, (M
AX value + MIN value) / 2 is calculated to obtain the threshold value Th,
As shown in FIG. 7E, the coordinates P1, P2, P3, P4 of four points that cross the threshold Th are calculated. Further, the image processing computer 15 determines that the line width W, which is the characteristic amount of the resist pattern,
Can be calculated based on the following equation (1).

W = ((P3 + P4)-(P1 + P2)) / 2 (Equation 1) When the calculated line width W is equal to or larger than the allowable value with respect to a normal line, the image processing computer 15 is abnormal. Can be determined as As described above, the feature amount of the pattern calculated in the image processing computer 15 is connected to the process abnormality diagnosing device (composed of a computer or the like) 16 for each wafer or lot, for example. By accumulating in the device 17, the process abnormality diagnosis device 16 can perform the process abnormality diagnosis for each wafer or lot. Further, the process abnormality diagnosing device 16 inputs the defect information by inputting the information of defects which are generated irregularly and irregularly in the wafer, which are obtained from the comparison / defect determination circuit 10, for each wafer or lot. In consideration of this, it is possible to perform process abnormality diagnosis on a wafer-by-wafer basis or on a lot-by-lot basis. Further, in the process abnormality diagnosis device 16, the input means 18
By inputting a wafer unit or a lot unit, it is possible to output and display the change information of the pattern feature amount and the change information of the abnormality (defect) based on the feature amount on the display means 19 and display the regularity. It is possible to output and display map information on the wafer showing the occurrence frequency of defects that occur unevenly, information on changes in the occurrence of defects, and the like. In this way, the process abnormality diagnosis device 16 can perform the abnormality diagnosis of the specific process by displaying and outputting the inspection result information on the display unit 19 or the like for each wafer or each lot. In particular, it is possible to perform abnormality diagnosis in the exposure / development process based on the characteristic amount of the pattern.

In the case of a semiconductor substrate such as a wafer, since it is formed of a thin film multi-layer pattern, in an optical detection image, even lower layers, which are many layers below, can be seen through, or steps may appear dark, which is complicated. However, it can also be applied to an optical detection image having such a complicated pattern. In particular, with respect to a wiring pattern that is opaque to the light existing on the surface, it is possible to inspect an optical detection image in the same manner as an electron detection image. FIG. 8 shows the configuration of an optical wafer pattern inspection device according to the present invention. In FIG. 8, the wafer 1 to be inspected is illuminated by the illumination lamp 84 through the half mirror 85, and the reflected light from the wafer 1 is imaged on the image sensor 86 by the objective lens 83. . As an image sensor, for example, TDI (Ti
meDelay & Integration) line sensors are preferred.
Then, by moving the wafer 1 in the direction orthogonal to the scanning of the image sensor, that is, in the X direction, the stage 7 can detect a two-dimensional image of the pattern to be inspected. The output signal S10 of the image sensor 86 is A / D
It is converted into a digital signal S11 by the converter 8.
The digital signal S11 is electrically delayed by the delay circuit 9 for the time required to move the wafer by one chip or one cell, and is compared with the original signal S11 in the comparison / defect determination circuit 10. 2a and 2b are adjacent chips on the wafer 1, but if the time delayed by the delay circuit is the time for the wafer to move by one chip, the original signal S1
1 and the output signal S12 of the delay circuit 9 correspond to image signals of adjacent chips, for example, 2a and 2b, which is equivalent to comparing detected images of 2a and 2b.

The comparison / defect determination circuit 10 has a logic realized by an electric circuit, and detects a mismatched portion by pipeline processing. For example, if the difference image is obtained after the edges of the image signals S11 and S12 are extracted and aligned, the non-coincidence portion can be detected. In addition, various algorithms have been proposed for detecting the non-coincidence portion, and a suitable algorithm may be implemented as hardware. The reason why the processing is performed by hardware instead of software processing on a computer is to increase the inspection speed. This is because the purpose of this comparative inspection is to detect defects that occur irregularly and irregularly, and it is advantageous to inspect a wider area to detect defects that occur irregularly and irregularly. This is because the speed must be high. The defect information obtained here is sent to the system control computer 11 that controls the entire apparatus. On the other hand, when the wafer reaches the coordinates designated in advance (coordinate data management is performed by the system control computer 11), the detected image (image signal S12) is captured in the storage device 14a or 14b (14a, b).
Switching is performed by the system control computer 11 controlling the input switching circuit 13). The image processing computer 15 reads out the detected image stored in the storage device 14a or 14b, calculates the pattern feature amount (dimension, positional relationship, etc.), and conforms the feature amount to the conforming product standard.
Judge nonconformity. Two storage devices 14a and 14b are provided because while the image processing computer reads and processes an image from 14a, 14b is used for image capturing, and the image reading and processing ends. Then,
This time, 14a is switched for image capturing and 14b for image reading, and the image processing computer 15 has already switched to 14b.
This is to eliminate the inconvenience that the wafer cannot be moved to the next detection position unless the processing of the image processing computer is completed by using the method of starting reading and processing of the image stored in. Image processing computer 1
When the amount of processing in 5 is large, or when there are many places where the feature amount should be calculated and this configuration is too late, it is possible to increase the number of image processing computers and storage devices.

The calculation of the feature amount in the image processing computer 15 can be realized by hardware (depending on the content of the feature amount), but in the present embodiment, as described above, the image processing computer is used. This is done by the software processing above. The reason is that the purpose of feature amount calculation is detection of defects that occur uniformly on the entire surface of the wafer. Therefore, even if the entire surface of the wafer is not targeted, for example, the frequency can be expected to be sufficient even at a frequency of about one location per chip. Therefore, the software processing capable of flexibly changing the image processing content is more suitable than the high-speed processing. For example, “one place” is one image in which a range of 50 μm * 50 μm is detected with a pixel size of 0.1 μm. In addition,
The method of calculating the characteristic amount is performed as described above.

[0028]

According to the present invention, by inspecting both a defect that occurs irregularly and irregularly in a semiconductor substrate such as a wafer and a defect that similarly occurs throughout the wafer, the wafer etc. It becomes possible to inspect all the defects generated on the semiconductor substrate, and as a result, it is possible to further improve the yield of the semiconductor substrate such as the wafer and to manufacture the semiconductor substrate such as the wafer having high reliability. Play. Further, according to the present invention, there is an effect that a process abnormality diagnosis can be performed by systematically accumulating pattern feature amounts. According to the invention,
By using an image of an electron beam, an image signal of a simple pattern is formed for a thin film multilayer pattern having an optically transparent pattern, and both a defect that occurs non-uniformly and a defect that occurs in the same manner It becomes possible to inspect all the defects that occur on the semiconductor substrate such as a wafer by performing the inspection of (1), and as a result, the yield of the semiconductor substrate such as the wafer is further improved and the reliability of the wafer such as that having high reliability is improved. It is possible to manufacture a semiconductor substrate.

Particularly in the case of an electron beam image, since the amount of generated electrons basically depends on the composition (atomic number), different materials have different brightness, which has a good material contrast.
That is, it becomes possible to obtain an image in which rules can be easily set. Furthermore, if the accelerating voltage is appropriately selected, the penetration depth of the irradiation electron beam can be made shallow so that it does not reach the lower layer pattern, so that the detection image can be a simpler pattern that includes only the uppermost layer pattern. In addition to the above, it is also possible to make it easier to find the feature amount related to the positional relationship between the uppermost layer and the layer below it by selecting an accelerating voltage so that the irradiation electron beam enters the layer below the uppermost layer. Becomes

[Brief description of drawings]

FIG. 1 is a diagram showing a basic principle according to the present invention.

FIG. 2 is a configuration diagram showing an embodiment of an electron beam type wafer pattern inspection apparatus according to the present invention.

FIG. 3 is a schematic diagram showing an effect of optimizing an acceleration voltage in the inspection apparatus shown in FIG.

FIG. 4 is a diagram showing a first embodiment for calculating a feature amount of a pattern in the inspection apparatus shown in FIG.

5 is a diagram showing a second embodiment for calculating a feature amount of a pattern in the inspection device shown in FIG.

6 is a diagram showing a third embodiment for calculating a feature amount of a pattern in the inspection device shown in FIG.

7 is a diagram showing a fourth embodiment for calculating a feature amount of a pattern in the inspection device shown in FIG.

FIG. 8 is a diagram showing a configuration showing an embodiment of an optical wafer pattern inspection device according to the present invention.

[Explanation of symbols]

1 ... Wafer (object to be inspected), 7 ... Stage, 8 ...
A / D converter, 9 ... Delay circuit, 10 ... Comparison / defective determination circuit, 11 ... System control computer, 12 ... Stage control circuit, 13 ... Input switching circuit (predetermined position selecting means), 14, 14a, 14b ... Storage device, 15 ...
Image processing computer, 16 ... Process abnormality diagnosis device (computer), 17 ... External storage device, 18 ... Input means, 19 ... Display means (output means), 21 ... Electron gun,
22 ... Magnetic field lens, 23 ... Magnetic field lens, 24 ... Deflector, 25 ... Detector (imaging device), 26 ... Amplifier,
27 ... High-voltage power supply (accelerating voltage applying means), 2 ... Chip

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Aritoshi Sugimoto 2326 Imai, Ome City, Tokyo, Hitachi, Ltd. Device Development Center (56) References JP-A-8-128976 (JP, A) JP-A-64 -59750 (JP, A) JP 62-22010 (JP, A) JP 2-216037 (JP, A) JP 7-239309 (JP, A) (58) Fields investigated (Int.Cl) . ^7, DB name) G01N 23/00 - 23/227 G01N 21/84 - 21/958 H01L 21/66 G06T 7/00

Claims (6)

(57) [Claims] 1. An electron gun that emits an electron beam, and a lens system that focuses the electron beam emitted from the electron gun on an object to be inspected in which a plurality of identical pattern groups are regularly arranged.
A deflector for scanning a focused electron beam, a detector for detecting electrons generated from the object to be inspected, and an A / D converter for converting an image signal by the electrons obtained from the detector into a digital image signal. , A feature amount of a pattern at a specific portion on the inspection target is calculated from a digital image signal obtained from the A / D converter, and conforms to a standard of good product based on the calculated feature amount of the pattern. A pattern inspection by an electron beam, which comprises a calculating means for judging nonconformity and further comparing image signals corresponding to a pattern group obtained from the detector to extract a mismatched portion to detect a defect. apparatus. 2. An electron gun that emits an electron beam, and a lens system that focuses the electron beam emitted from the electron gun on an object to be inspected in which a plurality of identical pattern groups are regularly arranged.
A deflector for scanning a focused electron beam, a detector for detecting electrons generated from the object to be inspected, and an A / D converter for converting an image signal by the electrons obtained from the detector into a digital image signal. , A feature amount of a pattern at a specific portion on the inspection target is calculated from a digital image signal obtained from the A / D converter, and conforms to a standard of good product based on the calculated feature amount of the pattern. An electron beam comprising a calculation means for judging nonconformity, and a defect judgment means for comparing the image signals corresponding to the pattern groups obtained from the detector to extract a mismatched portion and detect a defect. Pattern inspection equipment by. 3. The pattern inspection apparatus using an electron beam according to claim 1, further comprising an acceleration voltage control means for controlling acceleration of an electron beam incident on the inspection object. 4. The calculation means specifies the specific portion on the inspection object based on the correlation between the digital image signal obtained from the A / D converter and a desired standard pattern. The pattern inspection apparatus using an electron beam according to claim 2. 5. An electron generated from an object to be inspected by focusing and irradiating an object to be inspected in which a plurality of identical pattern groups are regularly arranged by a lens system with an electron beam emitted from an electron gun. A step of detecting an image signal by an electron for detecting the
A digital image signal is converted by a D / D converter, a characteristic amount of a pattern at a specific location on the inspection object is calculated from the converted digital image signal, and a non-defective product is obtained based on the calculated characteristic amount of the pattern. An inspection process for determining conformity / nonconformity with a standard, comparing image signals according to the converted pattern groups, extracting a mismatched portion, and detecting a defect. Pattern inspection method. 6. A detector for detecting an image signal from an object to be inspected in which a plurality of identical pattern groups are regularly arranged, and an A / D conversion for converting the image signal obtained from the detector into a digital image signal. And a digital image signal obtained from the A / D converter to calculate a feature amount of a pattern at a specific location on the inspection target, and based on the calculated feature amount of the pattern, a standard of good product is obtained. It is characterized by comprising a calculation means for judging conformity / nonconformity and a defect judgment means for comparing the image signals corresponding to the pattern groups obtained from the detector to extract a non-coincidence portion and detecting a defect. Pattern inspection device.