JPS60245134A - Positioning device - Google Patents

Abstract

PURPOSE:To position by extracting pattern information from two local regions of the prescribed distance of chips aligned at the prescribed interval along xy coordinates, detecting and correcting a rotary error from correlation, and then correcting x, y directions in response to the remaining rotary error. CONSTITUTION:A spot light 13 is scanned in y direction in local regions spaced at a distance L in x direction, and rotary amount is identified from the displacement y3-y1 of the intensity of intensity distributions A, B and L. Then, when rotary amount theta0 is obtained at the distance (d) with respect to chips aligned in alpha-axis direction of arraying coordinates alphabeta of the chips, an error is contained due to d<L. A wafer is moved in the amount (L,theta0), scanned by a light 13c, the rotary amount theta is calculated from the positions P3, P2, and corrected once. Then, the light 13 is again scanned in y direction at two local portions on the distance L of integer times of the arraying pitch, the remaining rotary error DELTAtheta is calculated to correct the position of the wafer in xy direction. As a result, the substrate can be positioned in high accuracy.

Description

Detailed Description of the Invention (Technical Field of the Invention) The present invention relates to a positioning device for masks, reticles, wafers, etc. used in the manufacture of semiconductor devices such as ICs and LSIs.
In particular, the present invention relates to a positioning device suitable for a wafer blower or a wafer scriber.

(Background of the Invention) As an alignment device for a substrate on which a plurality of chip patterns are arranged two-dimensionally, such as a wafer, for example, Japanese Patent Laid-Open No. 58
As disclosed in Publication No. 54648, a spot of laser light is scanned on a wafer, scattered light from the pattern on the wafer is photoelectrically detected, and the waveform of the photoelectric signal is taken and stored in advance. An apparatus is known that detects a positional shift of a wafer by comparing a waveform take (ten rate) serving as a reference using putter/match/reference. In this apparatus, a template is extracted from a predetermined local area on the first wafer, and the second and subsequent wafers are two-dimensionally positioned at the same position as the first wafer. By the way, when performing such positioning, if the Ueno is placed on the device while being rotated, the desired positioning accuracy cannot be obtained, so it is necessary to correct the rotation of the Ueno by some method. Conventionally, the wafer is placed on a two-dimensional moving stage via a rotatable wafer holder. The wafer holder is then rotated to correct the rotation of the wafer, but correcting this rotation requires a certain amount of repetition and time. Accuracy may be low,
The disadvantage is that the rotation correction time takes up a considerable proportion of the overall positioning time, reducing throughput.

(Objective of the Invention) It is an object of the present invention to provide a positioning device which reduces the time required for rotational correction of a substrate having a step pattern such as a mask or the like and which improves accuracy.

(Summary of the Invention) The present invention provides an apparatus for positioning a substrate (such as a Ueno mask, etc.) on which chips of a predetermined size are two-dimensionally arranged in a predetermined hitch along an orthogonal coordinate system xy. , first information corresponding to butter/ in a first local area on the substrate, and a pattern in a second local area separated from the first local area in the direction of arrangement of the chins 7' by a distance approximately an integral multiple of the pitch. pattern information extraction means for extracting second information according to the first information;
Rotation error correction means for detecting a rotation error regarding the coordinate system xy of the substrate based on the correlation between the information and the second information and rotating the substrate so as to correct the rotation error; and a residual rotation of the substrate after the rotation correction. The entire technical point is to provide means for detecting the error and means for correctively positioning the substrate with respect to the X and X directions of the coordinate system xy in accordance with this residual rotational error.

Next, the configuration of a positioning device according to an embodiment of the present invention will be explained based on FIGS. 1 and 2. The sea urchin . . . 6 is placed on a stage 7 equipped with a position sensor 12 such as a laser interference detector. This spoon 7 is moved two-dimensionally by a drive mechanism 8 including a motor and the like. On the other hand, a convergent laser beam with an elliptical cross section is emitted from a light source including a laser oscillator 1 and a silicon/drical laser 2, and this laser beam 0 is bent by an emitting mirror 3 and emitted into a long and narrow shape on the surface of a wafer 6. The image is formed as a two-lint-like spot light 13. The laser light scattered by the uneven pattern on the surface of the wafer 6 reaches the photoelectric element 4. Light 1 of this photoelectric element 4! The signal is amplified by a predetermined amount by an amplifier 9 and then applied to the next A/D converter (hereinafter referred to as ADC) 10. Therefore A
The digital signal D2 output from the DCIO is proportional to the intensity or amount of scattered light generated from the wafer 6.

Also, a microscope 1 for observing the butter on the wafer 6
4 is provided to check whether the aligned position is correct. The above-mentioned position sensor 12 generates a position signal D1 regarding the dimensional position of the stage 70 with reference to the irradiation position of the pot light 13, and inputs it to the arithmetic processing unit 11 together with the ADClo signal D2. . The calculation processing unit 11 performs calculations necessary for alignment based on the signal DI+D, and controls the drive mechanism 8 according to the result to move the wafer 6 at a predetermined position 1.

For details about this arithmetic processing section j1, please refer to
54648, so the explanation will be omitted here, but the scattered light generated from the wafer 6 is stored in correspondence with the position (-7, a reference pattern which is stored in advance),
The purpose is to match the scattered light with the scattered light and detect the position where the matching is good.

Now, when the thickness of the spot light 13 in the longitudinal direction and the alignment target are periodically arranged like wafers or masks, the - side of the standard 1 tear) approximately equal to the size of. When the stage 7 shown in FIG. Since the scattered light generated in all directions from the uneven step portions is received over almost the entire area, the photoelectric signal hardly changes with respect to small changes in the pattern. This will be explained in more detail with reference to FIG. This figure shows the relationship between two adjacent balances 50 on the wafer 6 and the spot light 13 (θ).

As described above, the spot light 13 moves relative to the wafer 6 in a direction perpendicular to the longitudinal direction. At this time, as shown in the figure, the movement is performed along the direction in which the balance wheel 50 is arranged. Therefore, the spot light 13 moves while keeping its longitudinal direction parallel to a band width of about 50 to 100 μm, usually called a street line 53, provided between the steps 50. Inside the single chip 50, there is provided a region 51 in which actual circuit butter is formed, and a plurality of bonding bunts 52 located around the region 51. now,
Typhoid fever 50 dimensions F)X5+11711, spot light 1
Assuming that the three dimensions are 5 mm x 50 μm, when the spot light 13 irradiates the region 51, the circuit pattern has a two-dimensional and complicated shape, and scattered light is generated in all directions. Therefore, the photoelectric signal of the photoelectric element 4 has a large value on average while the spot light 13 is moving within the region 51. Furthermore, when the spot light 13 illuminates the vicinity of the bat 52, the Hakuto 5
Since the width of the bat row in which the batts 2 are lined up is usually about 100 μm, the thickness of the photoelectric signal changes as the spot light 13 moves. Furthermore, the spotlight 13 is the street line 53
When irradiated with light, the street line 53 has the characteristic of totally reflecting the normal light, that is, has the characteristic of hardly scattering it, so the photoelectric signal becomes an extremely small piece. The above-mentioned changing state of the photoelectric signal is reproduced in almost the same way at any position on the wafer 6 as long as the longitudinal direction of the spot light 13 is larger than the length of the negative side of the chip 5°.

By elongating the spotlight 13 in the form of a slit in this way, changes in butter/density on the wafer 6 (for example, the number of minute butter lines existing within the range irradiated by the spot light 13) are smoothed out. can be detected.

In addition, as shown in FIG. 3, the spotlight 13 moves along the arrangement direction of the chips 50 while remaining parallel to the street light 153. At this time, the relationship between the spotlight 13 and the wafer 16 is as shown in FIG. become that way. A plurality of chips 50 having the same pattern are formed in a matrix on the wafer 16, but one row of the chips 50, that is, the chips 50 in the left and right direction in the paper, are generally formed by linearly cutting out a part of the periphery of the wafer 6. Line up parallel to flat F. This serves as a reference end for rough positioning (referred to as pre-alignment) of the flat FU wafer 6. and,
The stage 7 has a built-in rotation correction mechanism that rotates the wafer 6 on it! One direction of movement of the flat IJ stainless steel 7 during the bridge alignment is determined, for example, parallel to the X direction shown in FIG. 4. This results in stage 7
The two mutually orthogonal moving directions X+Y and the
The two street lies 153 on 6 that are perpendicular to each other become parallel to each other. If the stage 7 is moved in the X direction or in the X direction while irradiating the wafer 6 with the spot light 13 in this Fourier aligned state, the surface of the wafer 6 will be scanned by the spot light 13. Become. At this time, as shown in FIG. 4, when the stage 7 is moved in the X direction, the spotlight 13 is in the form of a slit extending in the X direction; Shape it into a slit. For example, Siri/
This is done by rotating the generatrix direction of the trical lens 2 by 90 degrees. Now, the wafer 6 must be placed on the stage 7 in the coordinate system xy without rotational deviation. Therefore, the principle of detecting this rotational deviation will be explained based on FIGS. 5 and 6. FIG. 5 shows a state in which the wafer 6 has been positioned in the xy coordinate system with approximately no rotation in one realignment, and furthermore, two chips 50 are placed on the wafer 6 at a distance in the X direction and lined up in the X direction. shows. Here the distance is step 5
It is an integral multiple of the arrangement pitch of 0 in the X direction, and is determined to be as large as possible. First, make the spot light 13 into a slit shape extending in the
position. At this time, the signal processing section 11
The coordinate values of are read from the position sensor 12 and stored. Let that value be the position Xo + Yo. Next, signal processing section 1
1, the stage 7 is scanned in the X direction by a distance longer than the arrangement pinch of the chip 50 in the X direction, and the intensity distribution of scattered light accompanying the movement of the spot light 13 is extracted and stored.

Figure 6 (a) fd represents the forced distribution A at this time, where the vertical axis represents the intensity I of scattered light, and the horizontal axis represents the position in the X direction - to the chips on the left side of the wafer 6 At the street line 53 extending in the X direction between 50 and 50,
The intensity of the scattered light is considerably lower than the intensity of the scattered light when the spot light 13 scans the chip 50. Therefore, the intensity bottoms out (minimum value) at position y1 during scanning from position yo to y2. Next, the signal processing unit 11 returns the stage 7 to the coordinate value (Xo+yo) and moves the stage 7 from that position by a distance in the X direction without changing the y coordinate value, so that the spot light 13 is directed to the right side of the wafer 6. It is controlled so that it overlaps with the chip 50 located.

Here, the stage 7 is scanned in the X direction again to position y2,
Extract the intensity distribution of scattered light. FIG. 6(b) is a diagram showing the intensity distribution B at this time, and the vertical and horizontal axes are respectively the same as in FIG. 6(a). Here, the street lie 153 between the chips on the right side of the wafer 6 is located at y. It can be seen that it is at position y3 between y2 and y2 calls. Then, the signal processing unit 11 performs a correlation calculation with the two intensity distributions A (!:B) and calculates the deviation in the X direction of both intensity distributions A and B when the bottom of the waveform by the IJ-trine 53 is matched. Calculate the quantity, ie, Yx Y3.

In addition, the method of detecting this amount of wear (7), for example,
! J-) Extract the intensity distribution of scattered light in advance from a certain distance (a value smaller than the array pitch) before and after the line 53,
The waveform of this distribution may be stored as a reference butter/, and after calculating the degree of correlation with the reference pattern for each of the intensity distributions A and H to determine the positions y1 and ys, the entire calculation of yl-ys may be performed. .

Once the amount of deviation has been determined as described above, the amount of rotation θ of the wafer 6 can be approximated using the following equation (1).〇θ−(yl
)'3)/L... (1) Therefore, in principle, it would be good to correct the rotation of the wafer 6 by the amount of rotation θ, but if you rely only on this method, the following will occur. A problem arises. The first problem iJ is that even if the rotation amount θ is calculated as described above, if the accuracy of the mechanism for rotating the wafer 6 does not correspond to it, then after correcting the rotation of the wafer 6, the rotation of the wafer 6 is calculated again. It is necessary to repeat the operation of detecting the amount,
This is a point where rotation correction may take a long time. The second problem is that when the size of the step 50 is extremely small relative to the distance, when the stage 7 is moved a distance in the X direction, the rotation of the wafer 6 causes the spot light 13 to
This is the point where the left and right chips on the same row in the X direction are not scanned. This second problem causes significant errors. In other words, the chips on the left side and the chip on the right side of the wafer 6 are butter/matched with tips that are shifted by an integral multiple (not including zero) of the array pitch in the X direction, and the array pinch is different from the distance. The problem is that a rotation error of an integral multiple remains.

Therefore, an embodiment of the present invention that solves the above second problem will be described using FIG. 7. FIG. 7 is an extreme representation of the wafer 6 being rotated by θ with respect to the coordinate system xy[, where the coordinate system αβ is the arrangement coordinate of the chips on the wafer 6. (,7) First, one chip on the right side of wafer 6 and
The spot light 13a is positioned so that it can scan in the direction I)-TRY/. And stay ′) 7 to
Scan a predetermined distance in the direction to extract the intensity distribution of the scattered light, and compare it with the reference pattern (punch rate).
By matching /, the position y5 of the street lie / associated with the techno is detected. Next, the arrangement hip (or chip size) in the β-axis (β-axis) direction of the chip = &CP,
The rotation error of the wafer 6 that may occur depending on the accuracy of pre-alignment, etc. is ε. , a coefficient 't greater than 0 and less than or equal to 0.5
When a is an integer greater than or equal to 1 and n is (-, the signal processing unit 11 uses the following equation (2).

A distance d that simultaneously satisfies (3) is calculated.

d<a-CP/ε.・・・・・・ (2) d=n−C
P... (3) The distance d represents the distance in the X direction from the spot light 13a to the spot light 13b to be located in order to extract the intensity distribution of the scattered light next. In this way, the distance d from the spot light 13a
After positioning the spotlight 13b at the position, the stage 7 is scanned in the X direction to extract data on the intensity distribution of the scattered light. At this time, since the distance d is determined by the above formulas (2) and (3), the street line existing in the scanning range of the spotlight 13b is the same as the street line / existing in the scanning range of the spotlight 13a. It is something. That is, the intensity distribution of the scattered light in the X direction is extracted at two positions separated by a distance d with respect to a plurality of steps arranged in a line along the α axis of the coordinate system αβ. Now, from the data extracted by scanning the spot light 13b, the position y6 of the street lie/ is detected by butter/matching.

Next, the rotation amount θ0 is calculated by equation (4) based on the amount of deviation between the signal processing unit 1111''i' positions y5 and y6 and the distance d.

θo-(y5 y6)/a (4) Now, this amount of rotation θO includes an error because the distance d is shorter than the distance on the wafer 6.

Therefore, the rotation amount θ is determined by this distance d. The error (detection accuracy) in the X direction at the detected position is δ as shown in Figure 8.
. shall be. In addition, in FIG. 8, the X of the spot light 13a
The position in the direction is Pl, and the position of the spotlight 13b in the X direction is P2. Then, the spot light 13c is positioned at a position P3 that is a distance away from the position P0 in the X direction and in an angular direction determined by the rotation amount θ0 with respect to the X axis, and the intensity distribution of the scattered light in the X direction is extracted. When comparing the intensity distribution with / and calculating the amount of rubbing between the standard butter and /, and assuming that the possible amount of rubbing is 61, the distance is 8th.
It is expressed by equation (5) from the geometrical relationship shown in the figure.

L=d・(δ, /δ.) (5) Furthermore, if the deviation amount δ1 is also made smaller than 1/2 of the arrangement pinch (or step size) CP in the X direction, For example, δ] < a-CP (6).

Therefore, from the above equations (5) and (6), the signal processing unit 11 calculates an optimal distance that simultaneously satisfies the following equations (7) and (8).

L<(a-cP-d)/δo - (7) L-n-C
P... (8) Next, the signal processing unit 11 moves the stage 7 by a distance in the X direction from the position P (the position of the spot light 13a), and roughly rotates the stage 7 in the X direction by a rotation amount θ0. The amount determined by (L・
It is stopped at position P3, which is moved by θ0). Here, the spot light 13c is scanned in the X direction to extract the intensity distribution of the scattered light, and the position y7 of the street line attached to the chip in the X direction is detected by butter/matching. The street line / detected here is the same as the street line detected by the scanning of the spotlight 13a, which means that the scanning of the spot lights 13a and 13c was performed on the left and right chips in the same column.

Then, the signal processing unit 11 selects the position y5 previously found by scanning the spotlight 13a and the position y found by scanning the spot light 13c.
7, the accurate rotation amount θ of the wafer 6 is calculated using the following approximate equation (9).

θ""(y53'7)/L... (9) If you calculate the amount of rotation θ using the above method, even if the size of the tip is extremely small compared to the distance, it will be the same. Since buttering/matching can be performed on the left and right naps in the row, the second problem mentioned above is solved. Then, it may be sufficient to just correct the rotation of the wafer 6 by the amount of rotation θ by the signal processing unit 11, or if the rotation is corrected only once, the error cannot be completely corrected by the n degrees of the rotation correction mechanism, and the rotation error Δθ remains. . Is it possible to repeatedly correct this rotational error Δθ and reduce the error Δθ to zero? If so, the first problem mentioned above will occur.

An embodiment of the present invention that solves the first problem will now be described.

First, the rotation of the wafer 6 is corrected once according to the calculated rotation amount θ. Then, the signal processing unit 11 is regenerated and the X
A local area including two steps located a distance L (where L=n-CP) apart in the direction and a street line attached thereto is scanned in the X direction with the spotlight 13, for example, as shown in FIG. Then, the intensity distribution of the scattered light generated from the local portion is extracted, and the amount of positional deviation Δy in the X direction of both local portions is calculated using Butter/Match/Resolution. From the relationship with the distance, the residual rotation error Δθ is calculated using the following formula (1
0).

Δθ=Δy/L...Song (1o) Next, the signal processing unit 11 calculates the positional deviation amount Δ as shown in FIG.
A spot light 13d extending in the X direction and a spot light 13e extending in the is scanned in the X direction, and the two-dimensional intensity distribution of scattered light generated from that local portion is extracted. Note that the relative position of position P6 with respect to positions P4 and P5 is determined to be the same for all wafers to be aligned. Then, the signal processing unit 11 uses data of the two-dimensional intensity distribution of scattered light extracted in advance from a local portion corresponding to the position P6 of the first wafer as a template, and uses this template and the position of the wafer 6. Patter/matching is performed with the intensity distribution data of the local portion of P6, and the amount of deviation from the Te/Fu rate is calculated. The stage 7 is positioned so as to correct this amount of deviation, and the position P6 of the second wafer 6 is not aligned with the spot light 13 at the position corresponding to the position P6 of the first wafer. It turns out. In other words, the two-dimensional position of one wafer is reproduced for the second and subsequent wafers 6.

Then, by moving the stage 7 two-dimensionally with the reproduced nap at position P6 as the origin, a desired chip on the wafer 6 can be positioned within the observation field of the microscope 14 or directly below the 70-bar probe needle. However, position P6
The further away from the position, the larger the positional shift due to the residual rotational error Δθ.

By the way, the size or arrangement pitch of each chip in the X direction and the X direction on the wafer 6 is α, respectively. .

Speaking of β0, for example, when the array coordinate system αβ is set so that the origin is the center of Chinofu at position P6, the center of each techno is the coordinate thread αβ, and when m and n are integers, (mα0% nβ0 ) is distributed in a matrix in the coordinate values. Therefore, considering the residual rotation error Δθ, convert the position in the coordinate system αβ to the position in the coordinate system xy E~
This is expressed by the following equations (11) and (12).

x=m・α.・cos(Δθ) tenn・β.・5in(Δ
θ)...(11) y=n-μ0-cos(Δθ)-m・α0・5in(2
0 houses...(12) Since the error Δθ is extremely small, COS (Δθ) Ki1.
When 5in(Δθ) is set to Δθ, formulas (11) and (12)
are approximated as shown in equations (13) and (14), respectively.

x=m・α. +n・β.・Δθ ... (13) y=n
・β. -m・α.・Δθ ... (14) This formula (13
), according to the coordinate values (x, y) determined by (14), the chip at the position (m・α0, n・β0) in the coordinate system αβ is aligned directly under the microscope 14 or the probe needle. When positioning the stage 7, the position of the stage 7 with respect to the chip is (n・β.・Δθ, -m・α0
・Δθ) will be corrected.

This situation is shown in FIGS. 10 and 11. Fig. 10 shows the situation when the stage 7 is positioned without correcting the residual rotational error Δθ, and Fig. 11 shows the situation when the stage 7 is positioned after correcting the residual rotational error Δθ. . The rectangles indicated by dotted lines in FIGS. 10 and 11 represent the regions 30 where the probe needles are located corresponding to each naraful. If the stage 7 is positioned without correction, the tip C1 at position P6 will only have its own rotational error, and will be well aligned, but the tip C1
Regarding the steps c21 to c3 that are distant from , in the X direction (alpha direction), the positional misalignment with the area 30 increases sequentially with respect to the error Δθ. However, by making the correction as shown in FIG. 11, the center of each chip C1゜''21 C3 is located on the
When aligned, only a relative rotational error between the chips and the region 30 remains, and the same alignment accuracy can be obtained for all chips on the wafer 6.

(Effects of the Invention) As described above, according to the present invention, after detecting the rotational error of a substrate such as a wafer, it is only necessary to correct the rotation of the substrate by − times. Later, the remaining rotational error of the board is detected again and when positioning along the xy coordinate system, the position to be positioned is corrected in the xy force direction according to the residual rotational error, so for example 70-
The effect is that the needle can be aligned with the same high precision for all chips on the board.

[Brief explanation of the drawing]

Fig. 1 is a schematic configuration diagram of an alignment device suitable for an embodiment of the present invention, Fig. 2 is a perspective view of an optical system that forms a spot light, and Fig. 3 is a plane showing the relationship between the spot light and the chip. figure,
Fig. 4 is a plan view showing the two-dimensional relationship between Ueno and spot light, Fig. 5 is a diagram for explaining the invention in detail, Fig. 6 is intensity distribution of scattered light used for gebutter/matsuchi puffer fish. FIG. 7 is a diagram for explaining the present invention in detail; FIG. 8 is a diagram for explaining the method of calculating the amount of rotation of the wafer;
FIG. 9 is a diagram explaining the alignment operation, and FIG.
11A and 11B are diagrams illustrating the operation of correcting the residual rotation error. [Explanation of symbols of main parts] 5. Laser beam, 6. Wafer, 7. Stage,
11...Signal processing unit, 12...Position sensor, 13
...Spot light, 50. Cyu + C2 + C3... Chip, 53... Street Line Applicant Nippon Kogaku Co., Ltd. Agent Takashi Watanabe

Claims (1)

[Scope of Claims] An apparatus for positioning a substrate on which a plurality of chips of a predetermined size are two-dimensionally arranged at a predetermined pitch along an orthogonal coordinate system xy, comprising: a putter in a first local area on the substrate; and second information corresponding to the putter/ in a second local area that is separated from the first local area by a distance of approximately an integral multiple of the pitch in the chip arrangement direction. information extraction means; rotational error correction for detecting a rotational error of the substrate with respect to the coordinate system xy based on the correlation between the first information and the second information, and rotating the substrate so as to correct the rotational error; means for detecting a residual rotational error of the rotationally corrected substrate; and means for correcting and positioning the auxiliary substrate in the X direction and the X direction of the coordinate system xy according to the residual rotational error. A positioning device characterized by: