JP2000055624A - Deviation measuring method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To highly precisely detect a relative deviation between two bodies in a short time. SOLUTION: Scale mark patterns of the first graduation SU' and the second graduation SD' of a vernier formed in two bodies respectively are uptaken into an image processor to be stored as picture elements. Lines L1, L2 perpendicular to scale lines of the graduations SU', SD' are formed respectively based on the picture elements, and center lines of the scale lines are found based on intersections between the lines L1, L2 and both-end edges of the scale lines. Distances between the center lines of the scale lines of the graduations SU', SD' with consistent graduation values are found to be squared, the graduation values are taken on the abscissa axis and the squared values are taken on the ordinate axis, a parabolic curve expressed thereby is computed, and a relative deviation between the two bodies is detected with precision of visual observation or more by finding mathematically the abscissa coordinate of the vertex of the parabolic curve.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a vernier (vernier).
The present invention relates to a displacement measurement method for detecting a relative displacement between two objects by using the method.

[0002]

2. Description of the Related Art A measuring method using a vernier is as follows.
A first graduation having a graduation line dividing the length N by M, and a length (N
+1) or a second graduation line having a graduation line dividing the length (N-1) by M in parallel, and reading the graduation line with the least deviation between them, with a reading accuracy of 1 / M. This is a measurement method that allows the relative position of both to be known.

For example, as shown in FIG. 7, a first graduation SU having 11 graduation lines obtained by dividing a length of 40 mm into ten equal parts.
And a second scale SD having 11 scale lines obtained by dividing the length of 41 mm into ten equal parts, and these are printed or drawn on different objects, respectively. The state in which the center scale lines 10 and 10 'formed longer than the other are aligned with their center lines (represented by thin vertical lines in the figure) is shifted between the two objects. This is the case where there is no (state shown).

The distance between each scale line of the first scale SU is 4 m
m, and the distance between the respective scale lines of the second scale SD is 4.1 mm. Therefore, the first value represented by the scale value +0.1
There is a deviation of 0.1 mm between the scale line of the scale SU and the scale line of the second scale SD. Similarly, the scale value +0.2,
There is a difference of 0.2 mm between the scale line of the first scale SU and the scale line of the second scale SD. Therefore, if the graduation lines of the graduation value +0.1 match, the state of FIG.
The scale SU is 0.1 to the right relative to the second scale SD.
mm. When the scale lines of the scale value +0.2 match, the first scale SU is
That is, it is shifted by 0.2 mm to the right relative to the second scale SD. This measurement is performed by visually observing which graduation lines match with an optical microscope. In this case, there is a premise that the positional accuracy of the initial contact position is sufficiently within 1 mm, and the unit of the length is not a problem, and the same applies to microns.

[0005]

SUMMARY OF THE INVENTION As in the prior art,
In the method of visually reading the scale line pattern, it is difficult to quantify the minute shift amount and express it.Moreover, the read value may differ depending on the viewing direction of the scale line, and the repeated reading accuracy is not constant due to human judgment. Such readings tend to have individual differences. Further, it takes time from reading to obtaining a result. Further, in a system in which the position is adjusted by performing feedback control based on the measured shift value, automation and unmanned operations cannot be performed in a conventional style in which a person (observer) intervenes, and the human environment and Cannot be used in remote environments.

The present invention has been made in view of the above problems, and has as its object to provide a displacement measuring method capable of detecting a relative displacement between two objects with high accuracy in a short time.

[0007]

According to the present invention, the pattern of the scale lines of the first scale and the second scale of the vernier (vernier) respectively formed on two objects is taken into an image processing system and stored as pixels. . Then, lines perpendicular to the graduation lines of the first graduation and the second graduation are formed respectively, and the center line of the graduation line is obtained from the intersection of this line and the edges at both ends of the graduation line. Scale and second
The distance between the center lines of the scale lines of the scale is obtained, and this distance is squared. Then, by calculating the parabolic curve represented by taking the scale value on the horizontal axis and taking the square value on the vertical axis, and mathematically obtaining the coordinates of the vertices of the parabolic curve, the accuracy is higher than the visual accuracy. The relative displacement between the two objects can be detected.

[0008]

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

FIG. 1 shows the pattern of the scale lines of the first scale SU 'and the second scale SD' of the vernier (vernier) according to the present embodiment. Each scale line is a rectangle, and the center line indicating the center is indicated by a thin line in the vertical direction. It should be noted that, in the drawing, the first scale S ′ is used to facilitate distinguishing the scale lines of the first scale SU ′ and the second scale SD ′.
The scale line of U 'is shown with diagonal lines. The first graduation SU ′ is composed of eleven graduation lines obtained by dividing the length of 40 μm into ten, and the second graduation SD ′ is composed of eleven graduation lines obtained by dividing the length of 41 μm into ten. . The quantitative numerical value and the unit of length shown here are properly used depending on the application. These first scale SU 'and second scale SD'
For example, printed or drawn on the boundary of a divided region on a semiconductor wafer, used for measuring the connection accuracy of adjacent circuit patterns drawn for each region, or on a wafer exposed by a different process Printed or drawn on different layers between the overlapping patterns. The drawing position is precisely specified.
The scale SU ′ and the second scale SD ′ are printed or drawn at positions where the scale lines of the scale SU ′ and the second scale SD ′ can be aligned in parallel.

FIG. 2 shows a configuration example of an apparatus used for measurement. The semiconductor wafer 2 on which the patterns of the first scale SU ′ and the second scale SD ′ are formed is positioned and set on the table 1. Above this, an optical microscope 3 is installed, and the graduation line pattern is taken into the image processing device 5 by the CCD camera 4 via the optical lens 3a. The captured image can be confirmed on the monitor 6.
These units are basic components and vary depending on the object to be measured. For example, an autofocus device, a lighting device, or the like may be used in order to clearly capture the scale line pattern. These are built in the microscope 3. Further, the image processing device 5 has an output of the external synchronization signal 7, and can also compare the measurement result with a specified value, transmit a correction signal, and the like to an external device. The entire configuration described above may be incorporated and used as one module.

The operation of the displacement measuring method according to the present embodiment configured as described above will now be described.

The pattern of the scale lines of the first scale SU 'and the second scale SD' on the wafer 2 is magnified and imaged by the optical microscope 3, which is imaged on the image pickup device by the CCD camera 4 to perform gray processing. To the image processing device 5. Then, it is digitally processed and stored as a set of two-dimensional points (pixels) having brightness (gray value) and coordinates (address) on the video frame memory. For example, width 64
It is stored in a matrix of 0 pixels (pixels) and 480 pixels (pixels) vertically. The configuration of this matrix is determined by the specifications of the image processing device 5 to be used.

Next, as shown in FIG. 1, the first scale SU '
And lines L1 and L2 orthogonal to the scale lines of the second scale SD 'are set on the frame memory. FIG.
Shows a histogram of gray values (density values) of pixels on the line L1. The horizontal direction is the direction of the line L1, and the vertical direction represents the level of the gray value. The level + X is bright and the level -X is dark. When the histogram is viewed sequentially from the left, first, the level of the gray value decreases (the straight line d).
₁ ), then rising (straight line u)
₁ ). Next, in FIG. 1, when we look from the left along the line L1, first, in the intersection g ₁ of the left end of the line t ₁ and the line L1 of the graduation line of the scale values -0.5, bright behind the There is a change in the level of the gray value, such as from the area to the dark area of line t ₁ and also to the light area out of line t ₁ . This change is a change indicated by d ₁ and u ₁ in the above-described histogram, and the position of the vertex s ₁ of the valley represented by d ₁ and u _{1 corresponds} to the region where the gray value changes rapidly and the region. boundary, which corresponds to the left end edge of the clogging lines t _1. Furthermore, histogram, small gray values from the large (change of d ₂₎ and also from a small and changes to large (the change in u _2), which lines the right end of the scale marks of the scale value -0.5 corresponds to a change of the gray value at the intersection g ₂ between t ₂ and the line L1. That is, the change in u _{2 is} a change appearing from the line t ₂ to a bright area behind, and the position of the vertex s ₂ is the scale value −0.
5 is the right edge of the scale line. In this way, the left and right edges indicating the ridge line of the scale line having the scale value of -0.5 are found. Hereinafter, similarly, for both of the scale values, the edges at both ends of the scale line are obtained along the line L1. Thus, even if the density (gray value) of the pixel varies, all 22 locations are surely obtained. , And its position (coordinate) can be obtained. Similarly, the edge of each scale line is obtained for the second scale SD '.

Next, the position of the center line (shown by a thin line in the figure) of each graduation line is calculated from the obtained edge positions at both ends of each graduation line. From this result, the distance between the center lines of the graduation lines of the first graduation SU ′ and the second graduation SD ′ having the same graduation value is obtained. This allows
For each scale value, the amount of shift between the scale lines of the first scale SU ′ and the second scale SD ′ (in the image processing apparatus 5 in pixel units) can be determined with sub-pixel accuracy. Further, a value obtained by squaring this shift amount is obtained.

For example, in the table shown in FIG. 4, the center line position of each scale line of the first scale SU 'and the second scale SD' having the scale value of -0.5 is obtained. , The square value of the shift amount between the center lines is (a−
b) Determined by ² . Obtain the other scale values in the same way,
The shift amount is calculated for all the scale values, and a numerical table is created.

If the pattern of the graduation lines of the first graduation SU 'and the second graduation SD' is accurately formed, the change of the displacement of the graduation line for each graduation value changes in an arithmetic progression. If the square value is on the vertical axis and the scale value is on the horizontal axis, a parabolic curve as shown in FIG. 5 is obtained. The approximate expression of the parabolic curve is mathematically calculated from the data of the contents of the numerical table in FIG. 4 and the coordinates of the vertex (inflection point) are obtained. The corresponding scale value can be determined to below the decimal point. That is, the abscissa coordinate value (scale value) of the vertex of the parabolic curve is a value of the relative shift between adjacent circuit patterns on the semiconductor wafer 2.
As a result, the relative displacement in units of 1/100 micron can be measured, and the measurement time can be reduced. In addition, the parabolic curve shown in FIG. 5 shows a case where the vertex is in contact with the horizontal axis as an example. However, the present invention is not limited to this, and there may be a parabolic curve as shown in FIG. In any case, there is no difference in obtaining the horizontal axis coordinates of the vertices.

In FIG. 1, the scale value +0.1 and the scale value +
At 0.3, the direction of the shift between the scale lines of the first scale SU 'and the second scale SD' is reversed, and it can be seen that there is a shift "0" position between the scale lines within this range. When read visually as in the prior art, this position is defined as the position of the scale value +0.2 (the position where the scale line 20 of the first scale SU 'and the scale line 20' of the second scale SD 'match). It is natural to read. However, the position of the shift “0” between the scale lines does not always completely match the scale value +0.2. In such a case, it is difficult to quantify and express the minute shift amount by visual measurement, and accurate reading cannot be performed, for example, there is a difference between individuals.

On the other hand, in the measuring method according to this embodiment,
If the accuracy of the scale line patterns constituting the first scale SU ′ and the second scale SD ′ is sufficiently high, the measured value is 1/100 of the visual reading.
It is obtained with an accuracy of about 10 to 1/100, and the reliability is improved. Further, highly accurate measurement can be performed at low cost. Further, if the initial setting of each device in FIG. 2 is correctly performed, it is easy to keep the viewing direction constant. Further, automation and unmanned operation can be performed, and a feedback loop between measurement and adjustment based on the measurement result can be easily constructed.

Although the embodiments of the present invention have been described above, the present invention is, of course, not limited thereto, and various modifications can be made based on the technical concept of the present invention.

In the above embodiment, the scale line is rectangular. However, the scale line may be drawn with a thin line, or may be filled with a black background, or may be represented by a single line. In any case, the pattern becomes an important pattern when detecting an edge by image processing, and an edge detection algorithm suitable for the pattern is selected according to the shape of the graduation line.

In the above embodiment, the two objects are formed as circuit patterns on the semiconductor wafer. However, the present invention is not limited to this, and has a structure in which the first scale and the second scale of the vernier scale can be printed or drawn. It can be widely applied to the positioning of parts, and can measure at least 10 times the accuracy of direct reading by visual inspection and at least 10 times the resolution of the scale of the vernier scale.

In the above embodiment, the image formed by the optical microscope 3 is captured by the CCD camera 4 into the image processing device 5, but the present invention is not limited to this, and another image sensor may be used.

[0023]

As described above, according to the displacement measuring method according to the first aspect of the present invention, the displacement can be measured with high accuracy, and the reliability of the measured value is enhanced. Further, the time required for the measurement can be reduced.

[Brief description of the drawings]

FIG. 1 shows a first scale and a second scale of a vernier scale according to the present embodiment.
It is a top view of the pattern of the scale line of a scale.

FIG. 2 is a schematic diagram showing an entire configuration of an apparatus used for displacement measurement according to the present embodiment.

FIG. 3 is a histogram showing a change in gray value of a pixel on a line L1 on a first scale.

FIG. 4 is a table showing, for each scale value, a center line position of each scale line of a first scale and a second scale, and a square value between the center lines obtained therefrom.

FIG. 5 is an example of a graph showing a change in a square value of a shift between center lines of scale lines of a first scale and a second scale with respect to a scale value.

FIG. 6 is another example of a graph showing a change in a square value of a shift between center lines of scale lines of a first scale and a second scale with respect to a scale value.

FIG. 7 is a plan view of a pattern of a scale line of a first scale and a second scale of a vernier scale of a conventional example.

[Explanation of symbols]

2 ... Semiconductor wafer, 3 ... Optical microscope, 4 ... C
CD camera, 5 ... Image processing device, L1 ... Line, L
2 ... line, SU '... 1st scale, SD' ... 2nd scale.

Claims (3)

[Claims] 1. A first graduation having a graduation line for dividing the length N into M equal parts and a second graduation having a graduation line for dividing the length (N + 1) or the length (N-1) into M equal parts. In a displacement measuring method in which a relative displacement between two objects is detected by an optical microscope from a graduation value at which a graduation line coincides, image formation of the optical microscope is image-processed, and the graduation line is obtained from a pixel. And perpendicular lines are formed, and the center line of the scale line is obtained from the intersection of this line and the edges at both ends of the scale line, and the scale values of the scale lines of the main scale and the vernier scale are coincident with each other. The distance between the center lines is obtained, this distance is squared, the scale value is plotted on the horizontal axis, and the square value is plotted on the vertical axis. From the coordinate value of the horizontal axis at the intersection with the tangent passing through Shift measuring method is characterized in that to detect the deviation. 2. The semiconductor device according to claim 1, wherein the two objects are adjacent circuit patterns on a semiconductor wafer.
The deviation measurement method described in 1. 3. The displacement measuring method according to claim 1, wherein an image formed by the optical microscope is captured by a CCD camera and subjected to image processing.