JP2610372B2 - Misalignment measurement method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a misalignment measuring method for automatically measuring misalignment of a mask or a reticle in manufacturing a semiconductor device or a liquid crystal device, for example.

[0002]

2. Description of the Related Art For example, in the manufacture of semiconductor devices and liquid crystal devices, circuit patterns are sequentially formed using a plurality of masks and reticles. If the mask (or reticle) is not aligned with a predetermined accuracy in each step of forming a circuit pattern, the formed circuit will not function sufficiently,
Defective products occur. Therefore, in general, a cross Ma shown in FIGS. 7 and 8 - click 1,2 Yaba - Niapata - with emissions, misalignment measurement and inspection has been performed by visual inspection.

In FIG . 7 , two cross marks 1 and 2 are formed, and one cross mark 1 is formed on the other cross mark.
And is located outside of the other cross mark 2. The outer cross mark 1 is the first PEP (Photo Engra
In the subsequent PEP, it becomes a reference for mask alignment. In FIG. 8 , the inner cross mark 2 indicated by oblique lines is formed after the second PEP.

If the centers of the two cross marks 1 and 2 match, it is determined that the mask has been correctly aligned, and the inner cross mark 2 is replaced with the outer cross mark. If it deviates from mark 1, it is determined that the shift amount of inner cross mark 2 has exceeded the allowable value.

In the examples shown in FIGS . 7 and 8 , the allowable value of the deviation amount is ± 1 μm, and the distance between the two cross marks 1 and 2, that is, the two cross marks 1 and 2 The size difference is 1 μm.

Further, in the example shown in FIG. 8 , the centers of the two cross marks 1 and 2 coincide with each other. In FIG. 8 , A indicates the distance between the side edges of the two cross marks 1 and 2, and B indicates the distance between the edges of the two cross marks 1 and 2. C is the line width of the outer cross mark 1, D is the line width of the inner cross mark 2, and E is the outer cross mark 2.
2 shows the length of the side edge portion. And the value of each part is A =
1 μm, B = 2 μm, C = 20 μm, D = 18 μm, E
= 20 μm.

[0007]

However, as described above, the misalignment measuring method for superimposing the cross marks 1 and 2 is a method for examining the limit of misalignment, and it is necessary to obtain a quantitative value of the amount of misalignment. Can not.

The above-described misalignment measuring method is effective as long as a person visually determines OK / NG, but for example, a plurality of marks (patterns) to be compared with each other.
Is not suitable for automatic measurement in which the image of (1) is input to the image sensor and the misalignment is automatically measured. In other words, the boundary position of the mark tends to be unclear, and it is difficult to determine OK / NG when the shift amount is near the limit value.

Further, when the misalignment is automatically measured based on the vernier pattern, a quantitative value of the amount of misalignment can be obtained, but sufficient reliability can be obtained because the shape of the pattern is too complicated. Is difficult. An object of the present invention is to provide a highly reliable misalignment measuring method which can automatically measure misalignment, hardly causes errors, and has high reliability.

[0010]

In order to achieve the above object, the present invention provides a reference pattern formed in a previous process.
An image of the pattern to be measured and a pattern to be measured formed in a later process are taken by an image sensor, and the pattern to be measured is compared with a reference pattern.
In a misalignment measuring method for measuring a misalignment at the time of forming a pattern to be measured, each line width of each pattern is set to 2 μm or more in absolute amount, 8 pixels or more on an image sensor, and each pattern is formed.
In this case, the distance between each line in the image sensor is 2 μm or more in absolute amount and 8 pixels or more on the image sensor. By doing so, the present invention enables automatic measurement of misalignment, prevents errors, and improves the reliability of misalignment measurement.

[0011]

An embodiment of the present invention will be described below with reference to the drawings.

1 to 4 show an embodiment of the present invention. FIG. 1 shows the first PEP (Photo Engraving Process).
ss) shows patterns 11, 12,... formed, and FIG. 2 shows patterns 13 to 18 after completion of all the steps.

In the first PEP, the pattern 1 shown in FIG.
Are formed, and the pattern is formed in the second PEP.
The patterns 11, 12,... Are used as reference patterns.
12 are combined with the next pattern which is the pattern to be measured. The final pattern is pattern 1 in FIG.
It becomes like 3-18.

At each step, an image of the pattern (reference pattern) formed in the previous step and the pattern (pattern to be measured) combined with this pattern is taken with an optical microscope. And is taken into an imaging device such as a CCD camera. Then, for example, the output signal of the imaging device is sample-holded, and after A / D conversion, the amount of deviation of the pattern to be measured from the reference pattern is calculated.

The pattern 11 shown in FIG. 1 is formed by arranging two large and small L-shaped patterns 19 and 20 in a substantially square shape. Each of these L-shaped patterns 19 and 20 has a length of two pieces set substantially equal. Further, in the pattern 11, two L-shaped patterns 19 and 20 are provided.
Three linear patterns 2 of approximately equal length inside
1, 21, 21 are depicted. The linear patterns 21, 21, 21 are arranged in parallel with each other and in parallel with each of the L-shaped patterns 19, 20.

In FIG. 1, five patterns 12 are formed in addition to the pattern 11 described above. These five patterns 12 are L-shaped like the L-shaped pattern 20 on the smaller side of the pattern 11, and
They are arranged at substantially equal intervals in the row and column directions together with the L-shaped pattern 20. Each of the patterns 12 has a line width and a direction substantially equal to those of the L-shaped pattern 20. In FIG. 1, the dimensions and distances of the respective parts of the patterns 11, 12,...

F indicates the length of each piece of the L-shaped pattern 19 on the large side, and G indicates the length of each piece of the L-shaped pattern 20 on the small side. Further, H indicates the line width of the L-shaped pattern 19 on the large side, and I indicates the line width of the L-shaped pattern 20 on the small side.

J indicates the line width of the linear patterns 21..., K indicates the distance between the linear patterns 21. Further, L, M and N each have two L-shaped patterns.
The shortest distance between the patterns 19 and 20, the linear pattern 21 in the direction in which the linear patterns 21 are arranged, and the L-shaped pattern on the smaller side.
The shortest distance from the linear pattern 21 in the longitudinal direction of the linear pattern 21 and the L-shaped pattern 20 on the smaller side are shown. O in FIG. 1 indicates the distance between the pattern 11 and the adjacent patterns 12 and the distance between the patterns 12 respectively. The values of the aforementioned dimensions and distances are set, for example, as follows. F = 80 μm, G = 60 μm, H = I = J = K = 8 μ
m, L = M = N = 12 μm, O = 44 μm.

Patterns 13 to 18 after all steps are completed
Is a pattern having two L-shaped patterns 19 and 20.
It has substantially the same shape as the housing 11. And each pattern
In the patterns 13 to 18, the L-shaped pattern 22 on the large side
27 and the L-shaped patterns 28 to 33 on the smaller side are arranged so as to draw a substantially square, and both L-shaped patterns 13 to 1 are arranged.
8, three parallel patterns 34 are formed on the inside of each of the lines 22 to 27.

Some of the larger L-shaped patterns 22 to 27 have small patterns 35 which are divided into a plurality of ends and are arranged at equal intervals in substantially the same shape. The number of small patterns 35 is L-shaped patterns 22 to 27.
For example, the number is set to one to five. The line width and interval of each small pattern 35 are set to 2 μm or more.

Each of the above patterns 11, 12,..., 13-1
The image 8 is captured by an image pickup device such as a CCD. The magnification of the optical microscope for guiding the images of the patterns 11, 12,...
The line width of each of the patterns constituting
The distance between these patterns is adjusted to correspond to 8 pixels or more. Each of the patterns 11, 12,..., 1
Each of the patterns 3 to 18 and the portions between these patterns are input to the image sensor using at least eight pixels of the image sensor. Here, the magnification of the optical microscope for guiding the pattern image to the image sensor is set so that one pixel of the image sensor corresponds to 1 μm.

In the misalignment measuring method as described above, the line width and distance of each pattern are set to 2 μm or more in absolute amount, and set to 8 pixels or more on the image sensor. Therefore, it is possible to automatically measure the misalignment using the imaging device. Then, the measurement error can be reduced, and the reliability of the misalignment measurement can be improved.

That is, as shown in FIG. 3 (a), for example, when two patterns 36, 37 are imaged close to each other, the dimensions of the line widths P, Q of these patterns 36, 37 are several μm.
It is known that the intensity of an image obtained through the magnifying optical system and the image pickup device is represented by a signal shown in FIG. 38 in FIG.
Is a signal representing the pattern 36 on the left side in FIG. 3A, and 39 is a signal representing the pattern 37 on the right side.

Since the two patterns 36 and 37 are separated by the distance R, no signal appears theoretically between the two patterns 36 and 37, but the signals 38 and 39 actually overlap. The part between the two signals 38 and 39 does not become zero level.

In FIG. 4, the signal 38 in FIG. 3B is enlarged. In the figure, a solid line 38a represents an actual signal, and a two-dot chain line 38b represents a theoretical signal. As shown in FIG. 4, the widths S and T of the signals 38a and 38b are different from each other, and the signals 3a and 3b at the peak position U are different.
The sizes of 8a and 38b are also slightly different.

Therefore, even if the misalignment is automatically measured based on the signals 38 and 39 as described above, the patterns 36, 3
It is not possible to determine the exact position of 7 and as a result, an error, for example, on the order of submicrons occurs during measurement. Therefore, in order to measure the misalignment with high accuracy, it is necessary to set the line widths and distances of both patterns 36 and 37 sufficiently large.

On the other hand, even if the line widths and distances of both patterns 36 and 37 satisfy the conditions for high-precision measurement, both patterns 36 and 37 are excessively thin and close on the imaging device side. If so, an error occurs as in the case described above. In particular, if the sample hold or A / D conversion of the signal is performed in the image processing, the error becomes even larger. For this reason, it is necessary to use a sufficient number of pixels for capturing the line widths and distances of both patterns 36 and 37 also on the image sensor.

The signal 38 in FIG. 3A is obtained as shown in FIG. That is, for the image of the pattern 36,
The intensity is measured at a predetermined pitch in the width direction, and a histogram is obtained. The intensity of the image of the pattern 36 has a substantially normal distribution shape, and is strongest at the center in the width direction of the pattern 36, and becomes weaker as the pattern 36 is shifted outward. Then, the least square method is used for this histogram, and a solid line 38 in FIG. 5 is created.

However, the shape of the histogram obtained by measuring the image of the pattern 36 is different from a perfect normal distribution, and the axial position of the central column 40 does not coincide with the peak position U _{1 of the} signal 38. . When the magnitude of the measurement pitch ΔP is 1, the magnitude of the deviation δ is generally about 0.2. That is, when the magnification of the optical microscope is 1 μm / pixel, an error of about 0.2 μm occurs when the solid line 38 is formed.

FIG. 6 shows a comparison between the accuracy of the optical microscope for guiding the image of the pattern 36 to the image pickup device and the measurement accuracy of the SEM.
As shown in (1), the accuracy of the optical microscope is substantially the same as each other if the object to be measured is sufficiently large. However, the accuracy of the optical microscope decreases as the size of the object decreases, and sharply decreases for a minute object of 1 μm or less.

In this embodiment, each of the patterns 11, 1
The line widths and distances of 2,..., 13 to 18 are set to 2 μm or more in absolute amount, and are set to 8 pixels or more on the image sensor. That is, an area of 0.25 μm or more is imaged per pixel.

If the misalignment measurement is performed under these conditions,
The positions of the patterns 36 and 37 can be accurately recognized without being affected by an error generated when the signal 38 is generated or the accuracy of the optical microscope. Further, the patterns do not affect each other, and it is possible to prevent an error from occurring in image capturing and image processing. Therefore, error factors in the properties of light, image processing, and the like can be largely removed, and more accurate misalignment measurement can be performed.

Here, in this embodiment, the pattern 1
The absolute amount of the line width of 1, 12,..., 13 to 18 means the distance between the edges of each line. Also, patterns 11 and 12
.., 13-18 mean the distance between the contours of the lines. The present invention can be variously modified without departing from the gist.

[0034]

As described above, according to the present invention, the reference pattern formed in the previous step and the pattern to be measured formed in the subsequent step are imaged by the image pickup device, and the pattern to be measured is obtained. In a misalignment measuring method for measuring a deviation in forming a pattern to be measured as compared with a reference pattern, each line width of each pattern is set to 2 μm or more in absolute amount and 8 pixels or more on an image sensor. At the same time, the distance between each line in each pattern was set to 2 μm or more in absolute amount and 8 pixels or more on the image sensor. Therefore, the present invention has the effects of enabling automatic measurement of misalignment, preventing occurrence of errors, and improving the reliability of misalignment measurement.

[Brief description of the drawings]

FIG. 1 is a view showing a pattern formed in a first PEP according to one embodiment of the present invention.

FIG. 2 is a view showing a pattern after all steps are completed.

FIG. 3 is an explanatory diagram showing an example of a cause of occurrence of a measurement error.

FIG. 4 is an explanatory diagram showing an example of a cause of occurrence of a measurement error.

FIG. 5 is a graph showing an intensity distribution of a pattern image.

FIG. 6 is a graph comparing the accuracy of an optical microscope with the accuracy of an SEM.

FIG. 7 is a diagram showing a conventional alignment mark.

FIG. 8 is an enlarged view of a portion surrounded by a circle V in FIG. 7;

[Explanation of symbols]

 11, 12 ..., 13-18 ... patterns.

Continuation of the front page (56) References JP-A-54-53864 (JP, A) JP-A-57-69742 (JP, A) JP-A-57-40925 (JP, A) JP-A-57-57245 (JP, A) JP-A-59-92527 (JP, A) JP-A-62-81036 (JP, A) JP-A-62-237303 (JP, A) JP-A-1-184822 (JP, A) 61-1887 (JP, B2)

Claims (1)

(57) [Claims] An image of a reference pattern formed in a previous step and a pattern to be measured formed in a subsequent step are imaged by an image pickup device, and the pattern to be measured is compared with the reference pattern. ,
In a misalignment measuring method for measuring a misalignment in forming a pattern to be measured, each line width of each of the patterns is set to 2 μm in absolute amount.
m and 8 pixels or more on the image sensor,
The distance between the lines in each of the above patterns is 2 μm in absolute amount.
As described above, the misalignment measuring method is characterized in that the number of pixels is eight or more on the image sensor.