JP2002342757A - Method and device for comparing and inspecting pattern - Google Patents

Abstract

PROBLEM TO BE SOLVED: To realize a pattern comparing and inspecting method and device for performing cell/cell comparison by software processing even when the array pitch R of cells is not the integral multiple of a pixel pitch P. SOLUTION: This pattern comparing and inspecting device is provided with image pickup parts 12 and 16 for generating pixel data by photographing the picture of a pattern having a plurality of basic patterns 120 repeated by a prescribed pitch R, a memory 17 for storing the picture data, and a picture processing part 18 for successively comparing the pixel data corresponding to the basic patterns on the basis of the picture data. The picture processing part 18 is provided with an integral multiple pitch setting part for setting a first integer M for defining length obtained by multiplying a prescribed pitch by the first integer M as that which is the integral multiple of the pixel pitch in terms of a resolution which is longer than a prescribed decomposed pitch P/L smaller than the pixel pitch. The pixel data corresponding to the basic patterns isolated by the first integer M set by the integral multiple pitch setting part are successively compared.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern comparison inspection method and apparatus for comparing a basic pattern with a basic pattern repeated at a predetermined pitch and inspecting the pattern for the presence or absence of a defect. The present invention relates to a visual inspection method and an apparatus for inspecting a pattern formed on a semiconductor wafer such as a semiconductor memory or a photomask pattern which is repeated by sequentially comparing neighboring cell patterns.

[0002]

2. Description of the Related Art It is widely practiced to image a formed pattern to generate image data, analyze the image data, and inspect the pattern for defects. In particular, in the field of semiconductor manufacturing, a photomask inspection device for inspecting a photomask and a visual inspection device for inspecting a pattern formed on a semiconductor wafer are widely used. The present invention can be applied to inspection of any pattern that repeats a basic pattern such as a photomask or a pattern on a wafer. In the following description, a pattern formed on a wafer is optically applied. A description will be given using image data obtained by imaging as an example.

FIG. 1 shows a state where a semiconductor chip 101 is formed on a semiconductor wafer 100. Generally, such a semiconductor chip 101 is called a die, so this term is used here. In the manufacturing process of the semiconductor device, the wafer 100
Since many layers of patterns are formed thereon, it takes a long time to complete all the steps, and even if there is a serious defect in even one layer, the die becomes defective and the yield decreases. Therefore, by analyzing the image data obtained by imaging the pattern formed in the middle of the process, the layer where a serious defect has occurred is removed and re-formed, or the defect information is fed back to the manufacturing process to improve the yield. That is being done. A visual inspection device (inspection machine) is used for this purpose.

FIG. 2 is a diagram showing a schematic configuration of a visual inspection apparatus. As shown in FIG.
1 is held. The illumination light from the light source 14 is converged by the condenser lens 15, reflected by the half mirror 13, passes through the objective lens 12, and illuminates the surface of the wafer 100. The illuminated optical image of the surface of the wafer 100 is projected onto the imaging device 16 by the objective lens 12. The imaging device 16 converts an optical image into an image signal that is an electric signal. The image signal is digitized, converted into image data, and stored in the image memory 17. The image processing unit 18
The image data stored in the image memory 17 is processed to check for defects. The control unit 19 controls each unit of the apparatus such as the stage 11, the image memory 17, and the image processing unit 18.

[0005] The pattern of a semiconductor device is very fine, and a visual inspection device requires very high resolution. Therefore, a one-dimensional image sensor is used as the imaging device, and the stage 11 is moved (scanned) in one direction, and image data is obtained by sampling the output of the imaging device in synchronization with the scanning. When the width H on the imageable wafer is smaller than the width of the die 101, for example, as shown in FIG. 1, the same portion of each die is sequentially scanned. Another part is sequentially scanned to obtain image data of all parts of each die. Accordingly, the image data is obtained by performing the scanning, and at the same time, the comparison with the image data of the corresponding portion of another die obtained by the previous scanning can be performed at the same time, so that the throughput is improved. However, the scanning method is not limited to this, and various methods have been proposed.

FIG. 3 is a diagram for explaining the operation of comparing image data between adjacent dies. Assume that dies A, B, C, and D are arranged as shown in FIG. Image data is represented in units of pixels 101. The array pitch of the pixels on the wafer surface is a value obtained by multiplying the array pitch of the image pickup device by the reciprocal of the photographing magnification in the direction perpendicular to the scanning. In the case of a one-dimensional image sensor, the scanning direction is the scanning speed and sampling. It is a value multiplied by the period. As shown in the drawing, when the comparison is performed between the dies B and C, the image data (pixel data) of the corresponding pixels of the dies B and C are compared. For example, the pixel data in the first column of the rows a of the dies B and C are compared with each other.
Therefore, it is necessary that the positions of the pixels on the dies B and C correspond to each other. A pattern is formed on each die by exposing one common photomask, and after exposing one die, the wafer is moved to expose the next die, and so on. Therefore, the positional accuracy between the dies is determined by the positional accuracy of the movement. Generally, in order to perform a sufficient comparison inspection, it is required that the displacement of a part to be compared is from a fraction of the number of pixels to one-tenth or less. Various methods have been proposed, such as correcting a shift by processing.

The comparison of pixel data between dies is as follows: A, B, B
The pixel data is generated and stored in order from the end die in the order of C and C, C and D, and the pixel data of the newly generated die is compared with the pixel data of the die generated and stored immediately before. It is common. By performing the comparison in this manner, the dies in the central portion other than the dies at both ends are compared twice with the two adjacent dies, and if the comparison results match both times, it is determined to be normal (no defect), and the comparison result is determined. If the two times do not match, it is determined that there is an abnormality (defective). The method of comparing two times in this way and determining the presence or absence of a defect based on the results of the two comparisons is called double detection. The comparison between the dies is called a die-to-die comparison.

[0008] A semiconductor memory or the like has a configuration in which a basic unit called a cell is repeated, and a pattern therefor also has a configuration in which a basic pattern corresponding to a cell is repeated.
FIG. 4 is a diagram for explaining a cell. As shown in FIG. 4, cell regions 120 each having a length (arrangement pitch) R in the scanning direction and a width S in a direction perpendicular to scanning are repeatedly arranged in two directions.
When inspecting a pattern in which such cells are arranged at a predetermined pitch, without performing the above-described die-die comparison,
The presence / absence of a defect is determined by comparing pixel data of corresponding parts between neighboring cells. This is called a cell-cell comparison.

As shown in FIG. 4, when the pixel pitch is P and R
Is an integral multiple of P, each pixel in each cell region 120 can be compared as it is, as long as each pixel in the other cell region has the same position on the pattern. Similar to the case of the die-to-die comparison, the cell-to-cell comparison sequentially compares adjacent cells while shifting the cells one by one. Therefore,
The cells in the central portion excluding the cells at both ends are subjected to double detection which is compared twice with the cells adjacent on both sides.

[0010]

However, in practice, the cell arrangement pitch R is not always an integral multiple of the pixel pitch P. FIG. 5 is a diagram showing a state in that case, and it is assumed that R and P have the following relationship. R = (K + N / M) PK, M, and N are integers, 0 <N <M, and M and N are assumed to be reduced to the maximum.

In this case, the pattern of each cell gradually shifts from the pixel by NP / M. Therefore, the pixel data cannot be directly compared. It should be noted that a deviation in the pattern allowed for comparison is defined, and this is referred to as a resolution pitch. Here, a deviation of 1 / L (L is an integer) of the pixel pitch P is permitted. Let P / L be the resolution pitch. Therefore, M ≦ L.

When the arrangement pitch R is not an integral multiple of the pixel pitch P, comparison between adjacent cells cannot be performed. Therefore,
In such a case, it is proposed to change the photographing magnification or change the scanning speed or the sampling period so that the arrangement pitch R is an integral multiple of the pixel pitch P.
There is a problem in that the cost is increased due to the need for hardware capable of changing such conditions, and the throughput is reduced due to changing the conditions.

The present invention has been made in order to solve such a problem. Even when the cell arrangement pitch R is not an integral multiple of the pixel pitch P, the cell-cell arrangement can be performed by software without changing hardware. The purpose is to allow comparison.

[0014]

FIG. 6 is a diagram for explaining the principle of the present invention. As described above, when the pixel pitch P and the array pitch R have a relationship of R = (K + N / M) P (where K, M, and N are integers and 0 ≦ N <M, M
And N are reduced to the maximum, and M ≦ L (resolution pitch). ), A cell (M
Pixels apart from each other, that is, MR = (MK + N) P
Pixels that are separated can be considered to be at the same position on the pattern. Therefore, these pixels may be compared.

That is, in the pattern comparison inspection method of the present invention, inspection is performed by sequentially comparing corresponding pixel data of the basic pattern based on pixel data of a pattern having a plurality of basic patterns repeated at a predetermined pitch. A pattern comparison / inspection method, wherein a length obtained by multiplying the predetermined pitch by a first integer (M) can be regarded as an integer multiple of the pixel pitch when expressed by a resolution equal to or greater than a predetermined resolution pitch smaller than the pixel pitch. A first integer (M) is set, and corresponding pixel data of the basic patterns separated by the first integer (M) is sequentially compared.

The above equation uses the resolution pitch P / L and has a rounding error. However, the comparison is performed between pixels separated by MR = (MK + N) P while sequentially shifting the set of pixels to be compared, so that errors do not accumulate. The comparison of the pixel data is performed sequentially while shifting one pixel along the direction in which the basic pattern is repeated.
In a central portion of a row of pixel data in the direction in which the basic pattern is repeated, except for a range obtained by multiplying a predetermined pitch on both sides by a first integer (M), a predetermined pitch is set on both sides by the first pitch.
Is compared with two pixel data separated by a length that is an integer (M) times as large as the pixel data.

In this case, for pixel data in a range (both ends) obtained by multiplying a predetermined pitch on both sides excluding the center by a first integer (M), comparison is performed only once, but 1 It does not matter if there is no problem in the comparison between the times. If the pixel data at both end portions also need to be compared twice, the predetermined pitch is further multiplied by a first integer (M) and is equal to or more than two times the length (MR = (MK + N) P). A comparison is made with another piece of pixel data separated by an integral multiple of the length. In this case, considering the accumulation of errors, it is desirable that the pixels be separated by 2 MR.

As another method of comparing pixel data at both ends with two times, when a predetermined pitch is represented by a resolution of a second resolution pitch larger than the predetermined resolution pitch, the predetermined pitch is converted to a second integer ( A second integer (T) is set so that the length multiplied by T) can be regarded as an integer multiple of the pixel pitch, and the predetermined pitch is sequentially compared with pixel data separated by the second integer (T).

The pixel data at both ends is compared with the pixel data at the central portion inspected by the double detection method in the first comparison, and is compared with the pixel data at the central portion determined to have no defect. If the comparison results match, it is determined that there is no defect. Therefore, the second and subsequent comparisons of the peripheral portion need only be performed when the comparisons do not match in the first comparison.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention for inspecting a pattern formed on a semiconductor wafer will be described below. The schematic configuration of the inspection apparatus is the same as that shown in FIG. Is used to obtain pixel data by scanning. Further, it is assumed that the pattern formed on the semiconductor wafer is a pattern of a semiconductor memory having a cell portion in which cells are repeatedly arranged at a predetermined pitch.

FIG. 7 shows an arrangement of pixel data in a cell portion in a die in the embodiment. Since the present embodiment does not cover any part other than the cell part, a die-to-die comparison is performed on the other parts. When the arrangement pitch of the pixels in the scanning direction (lateral direction) is P, the arrangement pitch R of the cells is R = (K + N / M) P as shown in FIG. 6, and K, M, R, L
(Not shown) is as described above.
Here, the arrangement pitch R is set by the operator in the image processing unit 18 in the figure based on design data and imaging conditions, is automatically calculated by the image processing unit 18 through image processing, or is semi-automatically combined with these. Is calculated.

Here, for the sake of explanation, as shown in FIG. 7, the pixel data in the first MR length range among the pixel data in the cell portion are set A, and the pixel data in the next MR length range are denoted by B. Group, the pixel data in the length range of the next MR is set to group C,
Pixel data in the length range of the last MR is set in Z sets,
The pixel data in the length range of R is set as Y sets. Each set includes M cells.

The image processing unit 18 compares the head pixel data 110 of the set A with the head pixel data of the set B, and then sequentially performs the comparison while shifting the pixel positions one by one to the right. At this time, the pixel data arranged in the direction (vertical direction) perpendicular to the scanning is also compared with the pixel data having the MR shifted in the horizontal direction. Therefore, the head pixel data of the M-th cell in the A group is compared with the head pixel data of the M-th cell in the B group. In this case, the two pixels to be compared are separated by MR = (MK + N) P, and thus are located in the same part of the cell pattern. Therefore, the pixel data may be compared as it is. If the comparison results do not match, it is likely that one of them is defective.

When the above comparison is further advanced and the comparison of all the pixel data of the sets A and B is completed, the first pixel data of the set B and the first pixel data of the set C are compared. The comparison is performed sequentially while shifting the positions one by one to the right.
Then, when the comparison of all the pixel data of the Y group and the Z group is completed, the comparison of the cell portion is completed. Therefore, double detection is performed in which all the pixel data of the group B to the group Y are compared twice with the pixel data adjacent on both sides and separated by the length of MR. On the other hand, the pixel data of the group A and the group Z are compared only once with the pixel data of the group B and the group Y. Pixel data for which double detection is performed is determined to be normal (no defect) when both detection results match, and abnormal (defective) when both detection results do not match. judge. The pixel data of the sets A and Z are compared only once, but in actuality the ratio of the pixel data of the sets A and Z is very small, so even if they are ignored, no particular problem occurs. Absent.

If it is necessary to compare the pixel data of the group A and the pixel data of the group Z twice, for example, the pixel data of the group A is compared with the pixel data of the group C. Further, the pixel data of the Z group is compared with the pixel data of the previous group of the Y group. In this case, there is no particular problem because the accumulation of rounding errors due to the resolution pitch P / L is small. That is, the pixel data of the group A and the group Z is compared with pixel data separated by a length of 2MR = 2 (MK + N) P.

As another method of comparing the pixel data of the group A and the group Z twice, there is a method of arranging the pixel data at a resolution of a second resolution pitch P / U (U <L) larger than the resolution pitch P / L. Pitch R = (K + S / T) P. At this time, S <T ≦
U. Then, the pixel data of the group A and the group Z are compared with pixel data separated by TR = (TK + S) P.

The process of comparing the pixel data of the group A and the pixel data of the group Z twice as described above may be performed only when the result of the first comparison process does not match. The pixel data of the group A and the group Z are compared with the pixel data of the group B to the group Y, and the pixel data of the group B and the group Y which are inspected by the double detection method are compared.
This is because if the pixel data of the set and the comparison result match, it is determined that there is no defect.

Further, in the above description, M ≦ L,
M = L may be set. Although the embodiments of the present invention have been described above, the present invention is applicable not only to inspection of a pattern formed on a semiconductor wafer but also to inspection of a photomask which is an original pattern of the pattern. Furthermore, in the above description, the configuration in which image data is generated by scanning using a one-dimensional image sensor has been described as an example. However, for example, when a two-dimensional image sensor is used, a comparison in any direction and both directions is performed. It is also possible to apply the present invention.

[0029]

As described above, according to the present invention,
Even when the cell arrangement pitch R is not an integral multiple of the pixel pitch P, cell-cell comparison can be performed by software processing, so that pattern comparison inspection can be performed at low cost and in a short time.

[Brief description of the drawings]

FIG. 1 is a diagram showing an arrangement of semiconductor chips (dies) formed on a semiconductor wafer and a trajectory at the time of inspection.

FIG. 2 is a diagram showing a schematic configuration of a visual inspection apparatus for inspecting a die formed on a semiconductor wafer.

FIG. 3 is a diagram illustrating a die-to-die comparison.

FIG. 4 is a diagram illustrating an example of a cell and a cell-cell comparison.

FIG. 5 is a diagram showing a case where a cell pitch is not an integral multiple of a pixel pitch in a cell-cell comparison.

FIG. 6 is a diagram illustrating the principle of the present invention.

FIG. 7 is a diagram showing an array of pixel data of a cell portion in a die according to the embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 ... Stage 12 ... Objective lens 16 ... Imaging device 17 ... Image memory 18 ... Image processing part 19 ... Control part 100 ... Wafer 101 ... Die 110 ... Pixel 120 ... Cell

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 2G036 AA25 BA00 BB12 CA06 2G051 AA51 AC21 EA12 EA14 4M106 AA01 CA39 DB04 DB21 DJ18 DJ21 5B057 AA03 CA08 CA12 CA16 CC02 CC03 DA03 DA08 DB02 DB09 DC32 5L096 AA06 BA03 JA03 DA02

Claims (12)

[Claims] 1. A pattern comparison inspection method for inspecting by sequentially comparing pixel data corresponding to a basic pattern based on pixel data of a pattern having a plurality of basic patterns repeated at a predetermined pitch, comprising: The first integer (M) is set such that a length obtained by multiplying the predetermined pitch by a first integer (M) can be regarded as an integer multiple of the pixel pitch when expressed by a resolution equal to or greater than a predetermined resolution pitch smaller than the pitch. And a method for sequentially comparing corresponding pixel data of the basic patterns separated by the first integer (M). 2. The pattern comparison inspection method according to claim 1, wherein the comparison of the pixel data is sequentially performed while shifting one pixel along a direction in which the basic pattern is repeated, and the basic pattern is repeated. The pixel data of one column in the direction of the direction is obtained by multiplying the predetermined pitch on both sides by a first integer (M) at the center portion excluding a range obtained by multiplying the predetermined pitch on both sides by a first integer (M). A pattern comparison inspection method in which two pixel data separated by a length are compared. 3. The pattern comparison inspection method according to claim 2, wherein the pixel data in a range obtained by multiplying the predetermined pitch on both sides excluding the central portion by a first integer (M) is one of the pixel data. A pattern comparison inspection method in which a predetermined pitch is compared with only one pixel data separated by a length obtained by multiplying the predetermined pitch by a first integer (M). 4. The pattern comparison inspection method according to claim 2, wherein the pixel data in a range obtained by multiplying the predetermined pitch on both sides excluding the central portion by a first integer (M) is one of the pixel data. After a first comparison with one pixel data separated by a length obtained by multiplying a predetermined pitch by a first integer (M), the length obtained by multiplying the predetermined pitch by a first integer (M) is further increased by 2 or more. A pattern comparison inspection method in which the pixel data is compared with another piece of pixel data separated by a length that is an integral multiple of. 5. The pattern comparison inspection method according to claim 2, wherein pixel data in a range obtained by multiplying the predetermined pitch on both sides excluding the central portion by a first integer (M) is included in one of the pixel data. After a first comparison with one pixel data separated by a length obtained by multiplying a predetermined pitch by a first integer (M), when represented by a resolution of a second resolution pitch larger than the predetermined resolution pitch, The second integer (T) is set such that a length obtained by multiplying the predetermined pitch by a second integer (T) is an integer multiple of the pixel pitch. The predetermined pitch is set to the second integer (T). A pattern comparison inspection method that is sequentially compared with pixel data separated by a doubled length. 6. The pattern comparison inspection method according to claim 4, wherein said predetermined pitch on both sides excluding said central portion is multiplied by a first integer (M) for a second time. The subsequent comparison is a pattern comparison inspection method performed only when the first comparison result does not match. 7. An image pickup section for taking an image of a pattern having a plurality of basic patterns repeated at a predetermined pitch to generate pixel data, a memory for storing the image data, and An image processing unit for sequentially comparing corresponding pixel data of the basic pattern, wherein the image processing unit, when represented by a resolution equal to or greater than a predetermined resolution pitch smaller than the pixel pitch, the predetermined An integer multiple pitch setting unit that sets the first integer (M) such that a length obtained by multiplying the pitch by a first integer (M) can be regarded as an integer multiple of the pixel pitch; A pattern comparison and inspection apparatus for sequentially comparing corresponding pixel data of the basic patterns separated by the first integer (M). 8. The pattern comparison inspection method according to claim 7, wherein the image processing unit sequentially compares the pixel data while shifting the pixel data by one pixel in a direction in which the basic pattern is repeated. In a central portion excluding a range obtained by multiplying one row of pixel data in the direction in which the basic pattern is repeated by the predetermined integer on both sides by a first integer (M), the predetermined pitch is set to the first integer on both sides. (M) A pattern comparison / inspection apparatus for comparing two pixel data separated by a length twice as long. 9. The pattern comparison / inspection apparatus according to claim 8, wherein the image processing unit is a pixel in a range obtained by multiplying the predetermined pitch on both sides excluding the central part by a first integer (M). A pattern comparison / inspection apparatus that compares data with only one pixel data separated by a length obtained by multiplying the predetermined pitch by a first integer (M). 10. The pattern comparison / inspection apparatus according to claim 8, wherein the image processing unit is a pixel in a range obtained by multiplying the predetermined pitch on both sides excluding the central part by a first integer (M). The data is multiplied by a first integer (M) after a first comparison of the data with one pixel data separated by a length obtained by multiplying the predetermined pitch by a first integer (M). 2 more lengths
A pattern comparison / inspection apparatus for comparing with another one pixel data separated by the integral multiple of the length. 11. The pattern comparison inspection apparatus according to claim 8, wherein the image processing unit sets the predetermined pitch to a second resolution when the resolution is expressed by a second resolution pitch larger than the predetermined resolution pitch. A second integer multiple pitch setting unit that sets the second integer (T) that can be regarded as an integer multiple of the pixel pitch, wherein the length multiplied by an integer (T) is 2; After a first comparison with pixel data in a range obtained by multiplying the pitch of the predetermined pitch by a first integer (M) and one pixel data separated by a length obtained by multiplying the predetermined pitch by a first integer (M) , The predetermined pitch,
A pattern comparison / inspection apparatus for sequentially comparing the pixel data with the pixel data separated by the second integer (T) times; 12. The pattern comparing and inspecting apparatus according to claim 10, wherein the image processing unit multiplies the predetermined pitch on both sides excluding the central portion by a first integer (M). A pattern comparison / inspection apparatus that performs the second and subsequent comparisons of the pixel data of the above only when the first comparison result does not match.