JPS60201343A - Detecting method of magnification error - Google Patents

Abstract

PURPOSE:To detect the magnification error of a projection optical system speedily with high precision by projecting images of two opening through a projection optical system and meansuring the interval between two opening edges of the basis of high-order diffracted light generated from a diffraction grating on a wafer. CONSTITUTION:A reduced image of a pattern drawn on a reticle 19 is projected on the wafer 26 through a projection lens 21 with exposure light from a light source 1. Light from an optical fiber 7L illuminates the opening 20 of the reticle 19 through a beam splitter 12L and a reflecting mirror 14L to form an image 20L' of an opening 20L on the surface of the wafer 26. Reflected light from the lattice-shaped pattern 28 provided on the wafer 26 is passed through a space filter 16L from a reflecting mirror 14L and high-order diffracted light is converged on a detector 18L to output a photoelectric signal. A control circuit including a microcomputer, etc., analyzes the intensity distribution of the diffracted light to calculate positions X1 and X2 of a two-dimensional moving stage 22 when the center lines of the patterns 28 and 27 are aligned to the center lines of projection images 20L' and 20R', and when X1=X2, variation in magnification is zero.

Description

[Detailed Description of the Invention] (Technical Field of the Invention) The present invention is based on one difference in projection magnification of a projection optical system used in an exposure device that optically transfers a pattern of an integrated circuit (IC), etc. The present invention relates to a method for detecting.

(Background of the Invention) The miniaturization of large-scale integrated circuit (LSI) patterns is progressing year by year, and a reduction projection type exposure device is used to print circuit patterns, which meets the demand for miniaturization and has high productivity. is becoming popular.

In these devices N that have been used conventionally, a reticle pattern that is several times (for example, 10 times) the size of the pattern to be printed on a silicon wafer or the like (hereinafter referred to as wafer) is reduced and projected by a projection lens. An area smaller than a square with a diagonal length of 14 mm on a wafer is covered by one exposure. Therefore, in order to completely print the pattern on the entire surface of the Uni-7, which has a diameter of about 125 mm, is it necessary to place the entire wafer on a spoon, move it a certain distance, and perform exposure? A so-called step-and-repeat method is used. In LSI manufacturing, several layers or more of patterns are sequentially formed on the surface of the sea urchin, but unless the overlay error of the patterns between different layers is kept below a certain value, the conductive or insulating state between the layers will not be as intended. Therefore, the LSI will no longer be able to perform its functions. For example, for a circuit with a minimum line width of 1 μm, the minimum line width is 0.2 μm.
Only an overlay error of about m is allowed. Among the causes of this overlay error, those generated by the exposure apparatus are (1) projection magnification error, projection distortion, curvature or inclination of the image plane, and (2) relative positional deviation between the projected image and the wafer.

Conventionally, the main focus has been on eliminating only the positional deviation (2), and little attention has been paid to (J). By the way, (J) is also becoming a problem when printing patterns of different layers using a plurality of reduction projection type exposure apparatuses. Among the errors in (1), regarding the error in projection magnification,
If the amount of error can be measured accurately, the projection magnification '(il) can be set to a predetermined value by changing the relative distance between the reticle and the projection lens, or by changing the focal length or principal point position of the lens. The overlapping of the projected reticle pattern with the projected pattern can be improved by 8 degrees. Conventionally, in order to measure magnification sound, the position of the pattern on the reticle is measured, and this reticle pattern is printed on Ueno. - The position of the upper pattern was measured to adjust the magnification.This method had the drawbacks of requiring a long time to measure and having poor accuracy.

- Several attempts have also been made to automatically perform magnification measurements. Among them, for example, JP-A-58-2024
No. 49 discloses a reticle having two apertures provided at predetermined positions, and a reticle that is accurately aligned with each of the two apertures through a projection lens when the magnification error associated with the chip is zero. A sea urchin with two linear marks like this is prepared and the magnification error is detected by photoelectrically detecting the positional deviation between these two apertures and the two linear marks? A method of measuring is disclosed. In this method, the relative positional shift between the aperture and the linear mark is determined, so a photoelectric signal that represents the relationship between the edge of the aperture and the edge of the linear mark is required. In this case, a large number of interference fringes are generated near the edge of the linear mark in parallel to the edge due to the relationship between the wavelength of the illumination light and the thickness of the photoresist. For this reason, the influence of interference fringes is superimposed on the photoelectric signal, resulting in an error in the detection of the edge position of a linear mark, or in extreme cases, a large miscalculation of the edge position, resulting in a large amount of money being incurred.

Furthermore, if the light flux that illuminates the reticle aperture is not perpendicular to the pattern plane that forms the reticle aperture, but has certain angular characteristics, in other words, the reticle side (object side) of the projection lens is not perpendicular to the pattern plane that forms the reticle aperture. In the case of a telecentric optical system, the edges of the four straight lines forming the reticle aperture, for example a rectangular aperture, are approximately orthogonal to the virtual straight lines extending radially from the center of the reticle (the point through which the optical axis of the projection lens passes). If all magnification measurements are performed using two such edges, the intensity of the reflected light (or scattered light) generated at the two edges will be asymmetrical due to the angular characteristics of the illumination light.

For this reason, the asymmetry affects the photoelectric signals representing the two edges of the aperture, and there is also the drawback that the detection of the positional deviation between the aperture and the linear mark includes an error. This also applies to a linear mark that is placed between two such edges, and the reflected light (or scattered light) on both edges (steps of unevenness) of this linear mark is the same. The intensity becomes asymmetric due to the angular characteristics of the illumination light, resulting in an error in positional deviation detection.In this way, the above method includes a potential error due to the angular characteristics of the illumination light when detecting positional deviation. However, there was a drawback in that the measurement of an originally extremely small amount of magnification error (variation in magnification) of the projection lens became extremely unreliable.

(Objective of the Invention) An object of the present invention is to solve these drawbacks and provide a method for detecting magnification errors of a projection optical system quickly and with high precision. Another object of the present invention is to provide a method for detecting a relative magnification error between a projected image of a mask (reticle) pattern and a pattern on a projected substrate (wafer) to be superimposed.

(Summary of the Invention) The present invention provides a base line that passes through all predetermined points (centers) on a mask (reticle), for example, on the X-axis of an orthogonal coordinate system The two apertures (slit apertures) are fully illuminated and the two apertures are fully projected by the projection optical system (reduction projection lens). Two aperture images are sequentially scanned in the baseline direction by a diffraction grating pattern (area grating or oblique grating) having grating elements,
The entire technical point is to photoelectrically detect the high-order diffracted light generated from the diffraction grating pattern through the projection optical system and the aperture, and to compare the photoelectric meter spacing to detect the magnification error of the projection optical system.

Furthermore, the present invention illuminates two apertures similarly formed on the mask, projects the entire image of the two apertures on the projection optical system, and arranges the images at a predetermined interval in the direction of the base line on the mask. At the same time, two diffraction grating patterns in which a plurality of grating elements are arranged in a direction intersecting the baseline direction at a predetermined angle are scanned in two aperture directions t-:tia, and the height generated from each of the two diffraction grating patterns is scanned. The next diffracted light is photoelectrically detected through the projection optical system and each of the two apertures, and the difference between the interval between the two aperture images and the interval between the two diffraction grating patterns is determined based on the electric signals of both parties and the scanning position. The key point of all technology is to measure and compare the measured difference with the design difference to detect the magnification error of the projection optical system, and thereby to distinguish between the projected image of the original pattern on the mask and the projected substrate (wafer). The relative magnification difference with the pattern above can be detected.

(Example〕 Next, the present invention will be explained in detail using examples.

FIG. 1 is an optical layout diagram of a projection exposure apparatus suitable for a first embodiment of the present invention. Light source J is a photosensitizer (photoresist)
All of the exposure light of the wavelength (for example, single wavelength light such as i-line or i-line) that is fully sensitive is generated, and the exposure light is passed through the elliptical reflecting mirror 2,
After passing through the first condenser lens 3, the light passes through the shutter 4 and reaches the second condenser lens 6. The shutter 4 is inserted into and removed from the optical path as shown by the arrow 5"j to control transmission and blocking of the exposure light, and the second condenser lens 6 illuminates the reticle (mask) 19 with uniform exposure light. Reduction projection lens (hereinafter simply referred to as a projection lens) 21 electrically or thermally reduces the image of the pattern drawn on the reticle 191c and projects the reduced image onto the wafer 26.The wafer holder 25 vacuum-chucks the wafer 26, and A two-dimensional movement stage 22 is provided so as to be able to rotate the wafer 26 for orientation and flat positioning, and move up and down for automatic focusing.On the reticle 19, the optical axis AX of the projection lens 21 Slit-shaped openings 20L and 2OR are formed in the opaque parts at both ends sandwiching the optical fiber 7L.Now, the optical fiber 7L emits illumination light having the same wavelength as the light from the light source 5, and the illumination light has a predetermined wavelength. The beam is reflected by the beam splitter 12L through the field stop p8L having an aperture 8L', the shutter ~9L, and the condenser lens 11L, and illuminates the aperture 20 of the reticle 19 by the objective lens 13L and the reflecting mirror 14L. Due to the function of lens jlL and objective lens 13L, aperture 2
An image of the field stop 8L (D aperture 8L') is formed on top of 0L.
The light d passing through the 0Li is projected onto the wafer 26 by the projection lens 21, and an aperture 20L is formed on the surface of the wafer 26. The reflected light from the wafer 26 is reflected by the projection lens 2.
1 in the opposite direction, and part of the reflected light further passes through the aperture 20I4-. The light passing through the aperture 20L is reflected by the reflecting mirror 14L, and is applied to the spatial filter 16L by the objective lens 13L to form a spatial frequency component of the light intensity on the upper surface of the reticle 19. On the wafer 26 are convex or concave minute Iiii! A lattice-like pattern 28 in which elements are arranged linearly at a constant period is formed for magnification error detection, and an image of the pattern 28 is formed on the lower surface of the reticle 19. The spatial filter 16L has a periodic structure pattern 2.
Among the light generated from B, except for the θ-order frequency component light (regularly reflected light), the other components (first-order and second-order diffracted light) are transmitted, and the spatial filter passes through the filter J6L'i. The light of these higher-order frequency components (higher-order diffracted light) is focused on the detector 18L by the lens 19L. The detector 18L performs full photoelectric conversion on the received light ':/, and outputs a full photoelectric signal corresponding to the amount of light of the high-order frequency component.

On the other hand, the aperture 2OR on the right side of the reticle 19 similarly includes an optical fiber 7R, a field stop 8R having an aperture 8R''i, a shutter 1OR, a condenser lens lIR, a beam splitter t2R9, an objective lens 13R, and a reflector 14R'.
ii- illuminated through. The light from the grid-like pattern 27 provided on the wafer 26 in the same shape as the pattern 2B is detected by the spatial filter 16R, the condensing lens 17R, and the detector 18H, and the light is A photoelectric signal is obtained.

In the above configuration, the shutters 9L and 9R are indicated by the arrow 10L in the figure.
, can move forward and backward like IOH, and has openings 20L and 20.
Open only when performing R illumination and detection of transmitted light. Also, the positions of the reflectors 14L and 14H are indicated by arrows 15L and J5H in the figure.
It can move forward and backward in a direction perpendicular to the optical axis AX of the projection lens 21, and is appropriately positioned according to the positions of the apertures 20L and 20R. Of the mutually orthogonal movement axes of the two-dimensional movement stage 22, the movement axis in the X direction is driven by a motor 23, and the two-dimensional movement axis is driven by a motor 23,
The position of the dimensional movement stage 22 in the X direction is measured. Of course, although not shown in FIG. 1, a Y-direction drive motor and a laser interference side are also provided.

Now, FIG. 2 is a plan view showing an example of the reticle 19 shown in FIG. 1. A rectangular pattern area Pa on which a desired circuit pattern is drawn is formed on the reticle 19, and the periphery thereof is backed by a light-shielding layer of chrome or the like. Then, the reticle 19Q is accurately positioned and set in the apparatus, and the origin of the orthogonal coordinate system
It is assumed that the optical axis AX coincides with the optical axis AX of No. 1. At this time, the openings 20L and 20R are each formed on the X-axis with the optical axis AX in between, at a predetermined interval Lr. The openings 20L and 20R each have a slit shape extending in the Y direction. Also the third
The figure shows a chip C formed on a wafer 26 and a pattern 27.
28 is a diagram showing an example of an arrangement of 28. Here, only specific chips C+, Ct, and Cs arranged in the X direction are shown. These chips C1hCsr''R are formed in advance by the step-and-repeat method using the projection exposure system shown in FIG. FIG. 3 shows the case where the center CCt of the chip C3 is made to coincide with the origin of the coordinate system After positioning (aligning) the two-dimensional moving stage 22 so that they match, the shutter 4 is opened for a certain period of time and exposure is repeated. Patterns 27 and 28 formed along with one chip are minute line elements extending in the X direction arranged at a constant period (pitch) in the Y direction, and the patterns 27 and 28 are arranged at intervals on the X axis. It is formed by Lw.

The distance Lw and the distance Lr between the openings 20L and 20R of the reticle 9 are determined to have a unique relationship.

That is, when the designed reduction magnification of the projection lens 21 is m, the apertures 20L and 20 are set so that m-Lr = LwK.
R and patterns 27, 28 are provided.

FIG. 4 shows the intensity distribution of the illumination light in the X direction on the aperture 20L (or 20R) on the reticle 19, with the horizontal axis representing the position and the vertical axis representing the light intensity I (i-).
L<20R)&l-The portion of the transmitted light is indicated by the shaded area. Distribution curve alt Field stop 8L (8R) + 7) Aperture 8L'(8R') (7) Width A corresponding to the width in the X direction
t, opening 20L (20R)? The transmitted light portion 32 is narrower than the width At, 1, and the light intensity is flat.

FIG. 5(a) is a cross-sectional view of the pattern 27 on the wafer 26 in the Y direction, and FIG. 5(b) is a plan view of the pattern 27. As shown in FIG. 5(b), the pattern 27 is a collection of minute line elements, and when enlarged, it is periodically formed on the surface of the wafer 26 as rectangular convex portions 27a.

Therefore, stepped edges 27e and 270' extending toward the center of the chip are formed on each convex portion 27a. When this pattern 27 is irradiated with the illumination light that has passed through the aperture 2 OR'i of the reticle 19, the illumination light has the same single wavelength as the exposure light, so it becomes 0th order diffracted light (regularly reflected light) D. In addition to the convex portion 27
The scattered light generated at edges 27e and 27e' of a strengthen and weaken each other, resulting in first-order diffracted light,
Generates higher-order diffraction light such as second-order diffraction light or third-order diffraction light.

Now, FIG. 6 is a circuit block diagram of the control system of the exposure apparatus of this embodiment. The photoelectric signals from the detectors 18L and 18R are input to analog-to-digital converters (hereinafter referred to as ADCs) 50 and 51, respectively, and converted into digital values corresponding to the amount of received light.

The digital values from the ADCs 50 and 51 are respectively stored in random access memory circuits (hereinafter referred to as RAM) 52.5.
Enter 3. RAM52.53 contains all addresses (addresses) of position information from the laser interferometer 24 for position detection in the X direction.
This information is input as information, and the address to be accessed is increased or decreased by one for each unit movement amount (for example, 0.02 μm) of the two-dimensional movement stage 22. Therefore, the digital values from the ADCs 50 and 51 are stored in the RAMs 52 and 53, respectively, in a unique correspondence with the movement position of the two-dimensional movement stage 22 and the address. That is, the RAMs 52 and 53 discretely sample and store the intensity distribution in the X direction of the higher-order diffracted light generated from the patterns 27 and 28 on the wafer 26 using the detection resolution of the laser interferometer 22. A control circuit 54 including a microcomputer or the like inputs the position information from the laser interferometer 24, detects the current position of the two-dimensional moving stage 22, and detects the intensity distribution of the diffracted light stored in the RAMs 52 and 53. Analyze and calculate the positions of patterns 27 and 28. The control circuit 54 also outputs commands to start and stop write operations in the RAMs 52 and 53. The drive circuit 55 responds to drive information output from the control circuit 54 to drive the motor 2 for driving the two-dimensional movement stage 22 in the X direction.
31i - Drive. Although not shown in Figure 1, 2
A motor 56 for driving the dimensional movement stage 22 in the Y direction is controlled by a drive circuit 57 that responds to drive information from a control circuit 54 .

Furthermore, in this embodiment, in order to correct the magnification error, among the plurality of lens elements constituting the projection lens 21, the lens barrel 2Ja
A lens element 21bi that is movable along the inner metal optical axis AX is provided. The change in magnification of the lens element 21b relative to the moving lid in the direction of the optical axis AX may be extremely small; for example, for the moving Jl lo scale, the projected 15m
It is sufficient if the corner pattern image expands and contracts by about one house.

The lens element 21b will be described here as a single lens for convenience, but as is well known in the case of zoom lenses for photography, several lenses may be moved together, or several lenses may be moved together. The lenses may be moved differently. With such a configuration, the motor 58 drives the lens element 2
1btl-move in the optical axis AX direction. Motor 58 is controlled by drive circuit 59 responsive to drive information from control circuit 54 . Therefore, if you understand the magnification error,
By moving the lens element 21b by an amount corresponding to this amount, the magnification error can be made zero, or the projection magnification can be adjusted to a constant value of B.

Next, the operation of this embodiment will be explained based on the flowchart shown in FIG.

The control circuit 54 first controls the projection lens 21 in step 300.
The optical axis AX of is, for example, the center C of the chip C3 on the wafer 26.
The drive circuits 55, 57 and the motors 23, 56 drive the 2
The dimensional movement stage 22r is moved. At this time, the exposure shutter 5 shown in FIG. Shutters 9L and 9R for illuminating the reticle 19 are both closed. The control circuit 54 then moves to the positioned position X. It is read from the gold laser interferometer 24 and stored. Further, the control circuit 54 outputs a command to open the shutters 9L and 9Rt. The entire arrangement of the chip C2 and the openings 20L and 2OR at this time is shown in FIG. In FIG. 8, the projection of the aperture 20L (tJf20L
', and the projection of the aperture 20R@2]-20R'. When the two-dimensional moving stage 22 is completely positioned in step 100, the projected image 20R' is located at a predetermined interval on the left side of the pattern 27 (in the -X direction with respect to the pattern 27), and the projected image 20L' is located on the left side of the pattern 28 (in the -X direction) -X direction) K with respect to pattern 21NC at predetermined intervals. However, optical axis AX
The amount of shift between the center CC and the center CC (-j, that is, the predetermined interval) is determined to such an extent that the projected image 20R' does not overlap the tip 10 of the neighbor 9 on the left, and the projected image 2OL' does not overlap the chip C2. . Since the wafer 26 is coated with a photosensitive agent, the projected image 20R' is projected onto the chips C, , C,.

This is because 20L' will be exposed.

Now, when positioned as shown in FIG. 8, the control circuit 54
In step 101, the two-dimensional moving stage 22' is moved by a predetermined distance in the iX direction, and the pattern 27 is scanned with the projection image 20R', and the pattern 28 is scanned with the projection image 20L'. In response to the start of scanning of the two-dimensional moving stage 22, the control circuit 54 puts the RAMs 52 and 53i into write mode.

Now, in the next step 102, the control circuit 54 reads the position information from the laser interferometer 22 and determines whether the two-dimensional moving stage 22 has moved from the position to the position Xs in the X direction. At this point #X, the projected image 20R' does not overlap the chip C8 after scanning the pattern 27, and the projected image 20R'
After 0L' scans the pattern 28, it is set at such a position that it does not overlap the chip C3 on the right of the chip C2. And in this step] 02, two-dimensional movement stage 2
2 has been reached for the position, the next step 10
At step 3, the two-dimensional moving stage 22 is stopped at the position X8, and the write mode of the RAMs 52 and 53 is prohibited and switched to the read mode.

It also outputs a command to close the shutters 9L and 9R.

Next, the control circuit 54 reads the RAM 52°53 at step ]04.
Based on the intensity distribution data of the diffracted light from the patterns 27 and 28 stored in
Compute f. At this time, the intensity distribution data stored in the RAMs 52 and 53 are as shown in FIGS. 9 and 10, respectively. Figure 9 shows the photoelectric signal of detector J8L, with the intensity DL on the vertical axis and the position in the VcX direction on the horizontal axis.
Figure 10 shows the intensity DR of the photoelectric signal from detector J8R on the vertical axis.

The horizontal axis represents the position in the X direction.

However, in the intensity distribution data stored in the RAMs 52 and 53, the position in the X direction in FIGS. 9 and 1θ uniquely corresponds to the address.

For example, the position corresponds to address 0 of RAMs 52 and 53. First, the control circuit 54 determines that the projected image 20L' is the pattern 28.
Data of intensity distribution 40 obtained when scanning '(j
It is possible to read the information from the RAM 52 and request the position Xa when the predetermined reference level Vrl matches the intensity DL of the intensity distribution 40 as shown in FIG. Next, this position Xa and xb
Calculate the position Xl of the midpoint. This position
This is the second position. The control circuit 54 continues to project @20R.
Intensity distribution 4 obtained when ' scans pattern 27
1 data 'k is read from the RAM 53, and the position Xa' l when the predetermined reference level Vrl and the intensity DR of the intensity distribution 41 match as shown in FIG. 10 can be obtained. Then, a position X11, which is the midpoint between these positions Xa' and Xb', is output. This position X is the position of the two-dimensional movement stage 22 when the center line in the X direction of the pattern 27 and the center line in the X direction of the projected image 20B' exactly match.

When the positions X, , X are determined as described above, the control circuit 5
4, all magnitude relationships between XI and X are determined. Now, as shown in FIG. 8, the projected images 20R' and 20L' move from the left side (position X) to the right side (position
, ), the position of the two-dimensional movement stage 22 in the X direction detected by the laser interferometer 24 increases. So, with respect to positions X, and X, x, =
If L, when projections 1&2OR' and pattern 27 match, projection pJt20L' and pattern 28 exactly match, LW-m-Lr, and magnification variation Δm is zero. However, for example, when XI<Xs, when the projected image 20L' and the pattern 28 match,
The projection @20R' still does not match the pattern 27, and the projected turn is projected with the chip C2↓ slightly enlarged. On the other hand, if XI > Xs, the projection image 2
When 0R' and pattern 27 match, projection @20L
' still does not match the pattern 27, which means that the projected turn was projected slightly smaller than the chip C1. Therefore, the control circuit 54 is at position X.

The difference 'ffXt When the magnification variation amount Δm is positive, it is an expansion, and when it is negative, it is a reduction.

Next, in step 105, the control circuit 54 determines whether the magnification variation amount Δm is zero or not, and if it is zero, is it the main operation? If the value is not zero, the magnification is adjusted in the next step 306.

In the next step 106, the control circuit 54 calculates the amount by which the lens element 21b should be moved in accordance with the previously determined magnification variation amount ΔmVC. At the beginning of this magnification adjustment, it is assumed that the lens element 21b is set at the origin position where the projection lens 21 has the designed magnification, and is moved up and down from the origin position in the direction of the optical axis AX. Now, the amount of magnification variation Δm mentioned earlier is the amount of expansion and contraction of the projected iron on the wafer 26, and is 1
This does not directly represent the magnification variation of the projection lens 21. Therefore, the control circuit 54 calculates the moving lid Δz of the lens element 21b from the magnification changing lid Δm. Here, lens element 2J
b Set to the origin position, and at this time set the optical axis AX on way 26.
Assume that a virtual point image is projected at a position P apart from . And lens element 21bff 2 from the origin position
Assume that when the point image is moved in the direction of the optical axis AX by the amount of the point image, the point image is projected onto the wafer 26 at a position P' from the optical axis AX. Therefore, the relationship between the amount of movement of the lens element 21b and the amount of expansion/contraction of the projection pattern is expressed by Z/(P-P') with p constant (k).
It can be determined as This constant is determined in advance at the time of manufacturing or adjustment of the device and is stored in the control circuit 54. Meanwhile, chip C on way 26! Assuming that the patterns 27 and 28 associated with are each located at the same distance from the center CC of the chip CI, and the projected images 20R' and 2OL' are each located at the same distance from the optical axis AX, the control circuit 54 is configured as follows. According to the formula, the movement amount Δzy of the lens element 21b
calculate.

Δm Δ2=□・k When this calculation is completed, the control circuit 54 moves the lens element 21bi from the origin position by Δ2 via the drive circuit 59 and the motor 58.
I can only move. Of course, it goes without saying that the direction of movement is determined so that there is no variation in magnification.

The magnification fluctuation should have been corrected in each step up to the above, but after step 106, steps 100 to 100 are performed again.
It will be more reliable if 4'i is executed and it is fully confirmed in step 105 that Δm is zero.

Further, the method of repeating steps 100 to 106 as described above is suitable for cases where the amount of movement of the lens element 21b and the amount of movement of the projected image point on the lens 26 do not correspond uniquely, that is, in the case of non-linearity. This is an extremely effective method.

As described above, in this embodiment, the chip Ctk is used to change the magnification change Δm
f was detected, but if multiple chips are similarly billed and averaged, further improvement in accuracy can be expected. Furthermore, in this embodiment, the intensity distribution data of the diffracted light is digitally memorized, but the photoelectric signal of the detector 18L (18R) is input to the analog comparator, and the reference level Vrt CV
r*), the position information from the laser interferometer 24 is latched at the rising and falling edges of the binary signal, and the position
a , Position XI, you may take precautions.

Further, the interval Lw between the patterns 27 and 28 on the wafer 26 may be made smaller (or larger) by a certain value Xd in advance than the designed interval between the projected images 2OL' and 20R'.

In this case, the magnification variation amount Δm is X.

Calculated by XI Xd.

In this embodiment, since a pair of projected images 2OR', 2OL' and a pair of patterns 27, 281- are used, the moving distance of the two-dimensional moving stage 22 in the X direction is short when detecting the diffracted light, and the measurement time is significantly reduced. It has the advantage that it can be shortened and there is little measurement error. Furthermore, since the relative magnification error between the step C on the wafer and the projected image of the pattern area Pa of the reticle can be completely corrected, lithography with extremely high overlay accuracy can be achieved.

Next, a second embodiment of the present invention will be described. The second embodiment includes a pair of openings 20L and 20R in a reticle 19 and a wafer 26.
Among the upper pair of patterns 27 and 28, the pair of openings 20L of the reticle 19 during measurement.

20Rt - When using one of pattern 27.28
A pair of patterns 27°28t-
When in use, one of the openings 20L and 20Rζ is used. So, first, pattern 27.
A case where only one of the patterns 28, for example, the pattern 28 associated with the chip C3 is used will be described with reference to FIG.

As in the first embodiment, the two-dimensional moving stage 22t is controlled so that the patterns 28 are positioned at predetermined intervals on the left side of the projected image 208'. This position t-X.

So, here's the shutter 9R'! Open to. Next, the two-dimensional moving stage 22' is scanned in the iX direction to the position X, and the RAM 53 extracts the intensity distribution of the diffracted light. After closing the shutter 98t at this position X, the projected image 20
The two-dimensional moving stage 2 is moved to a position IIX, , which is away from the position in the X direction by the designed interval m·Lr between R' and 20L'.
Position 2. At position 11Xa, pattern 28 is projection 1
120L' at a predetermined interval, the shutter 9L is opened here and the projected image 20L' is located in the pattern 28.
Scan at i position X, t.

At this time, the RAM 52 extracts the intensity distribution of the diffracted light. Next, a position X1 where the pattern 28 and the projection 920R' match, a position X where the pattern 28 and the projection image 20L' match,
Based on the data from 3. The difference between the distance between the positions XI and XI (XI XI) and the designed opening m-Lr between the projections @20R' and 20L' becomes the magnification variation amount Δm'.

This amount of variation Δm' is the difference between the chip on the wafer 26 and the reticle 1.
This does not represent a magnification deviation relative to the 90 projection pattern, but rather a deviation from the designed reduction magnification of the projection lens 21 alone. Therefore, if the lens element 21b'@l#l is adjusted so that this magnification sliding amount Δm' becomes zero,
The absolute value of the reduction magnification of the projection lens 21 (absolute magnification) can always be maintained the same as the design value.

On the other hand, the case where the magnification error is measured using the pair of patterns 27 and 28 on the wafer 26 and one of the openings 20L and 20R of the reticle 19, for example, the opening 20R't-, will be described with reference to FIG. First, after positioning the two-dimensional moving stage 22'ft so that the projection 920B' is positioned on the left side of the pattern 27 attached to the chip C snow at a predetermined interval, the shutter 9Rt- is opened. Then, the projected image 20R' is scanned in a pattern 27 from the position to X. Here RA
M53 stores the intensity distribution data of the diffracted light generated in the range from position to X. Shutter 9Rt due to position
- After closing, the two-dimensional moving stage 22t is positioned at a position X that is separated from the position of the patterns 27 and 28 by the designed interval Lv in the X direction. Here again shutter 9R
, and scan the projected image 20R'' with the pattern 28 to the position (address space).The data in these RAM 53
Based on the data, the position XI of the pattern 27 and the position X4 of the pattern 28 can be requested as in the first embodiment.

Next, find the distance (X, -X,) for position X1 and pattern 2.
If we calculate the difference between the designed spacing Lw between 7 and 28, the difference becomes the magnification variation amount Δm''. This variation amount Δm'' is the expansion/contraction error amount of the chip, and the chip Pattern 2r includes a magnification error of the projection lens when the light is emitted at 28 kN, and errors in shrinkage and enlargement of the chip depending on the expansion and contraction of the wafer due to the process.

As described above, the amount of magnification variation Δ of the projection lens 21 alone is
When the magnification variation amount Δm'' of the chip and the chip magnification variation amount Δm are determined separately, the relative magnification variation amount Δm between the chip on the wafer 26 and the projected pattern of the reticle 19 can be determined by calculating the difference between the two. . Therefore, if the lens elements 21bWrll11 are aligned according to Δm as in the first embodiment, the chips on the wafer 26 and the projected circuit pattern etc. can be accurately overlapped without any magnification error.

As described above, in this embodiment, compared to the first embodiment, it takes more time to measure the magnification fluctuation, but the aperture 2 of the reticle 19
With the interval between OL and 20H fixed, reflector 14L
, 34R can be completely fixed, which has the advantage of simplifying the structure of the device. Furthermore, the interval Lw of the patterns 27 and 28 or the openings 20L and 20H
The distance Lr between the patterns 27 and 28 does not need to be determined in a unique relationship by considering the magnification mf of the projection lens 21;
iAperture 20L.

Another advantage is that the arrangement of 20R can be freely determined.

As described above, according to the first and second embodiments of the present invention, the wafer 2
The patterns 27 and 28 attached to one chip on 6 have a plurality of minute IIj elements (lattice elements), and each minute #! i
The extension lines of the 'thickness', specifically the edges 27e and 27s of the rectangular pattern 27a shown in FIG.
It is growing towards. Therefore, the reticle 19@ of the projection lens 21 is a non-telecentric optical system, and the aperture 20L of the reticle 59.

20 Rj - Both edges 27e of the rectangular pattern 27a even when illuminated obliquely with illumination light having angle characteristics.

Reflected light (scattered light) is generated from 270' with symmetrical intensity. Therefore, by detecting the intensity distribution of the diffracted light of the patterns 27 and 28, there is an advantage that it is not affected by oblique illumination and that the measurement results do not include deviations.

Now, the present invention is not limited to the above two embodiments. For example, the pair of patterns 27 and 28 associated with one chip on the wafer 26 may be provided anywhere on a line passing through the chip center CC'fr. However, pattern 2
The design spacing between 7 and 28 must be known in advance. Furthermore, the patterns 27 and 28 have a plurality of minute line elements (edges 27 e, 27 e') extending in the X direction arranged in the t-Y direction to form a rectangular lattice shape, but at a predetermined angle (for example, 45°) with respect to the X direction. The inclined minute line elements 2 may be arranged in the y direction to form a diagonal lattice shape. Also, JP-A-56-370
A two-directional diffraction grating pattern as disclosed in Japanese Patent No. 17 may also be used. Also, patterns 27 and 28 on the wafer 26
Instead of pattern 27. , 2g may be provided on the two-dimensional movement stage 22 with a reference mark plate formed on a chrome layer. In this case, the opening 20 of the reticle 19
Projected image 20L' of L, 20R.

2OR'? Since scanning is performed using a reference mark, it is not possible to immediately correct the magnification error t relative to the pattern image of the reticle 19 projected onto the chip on the sea urchin to be superimposed. However, even if the wafer to be exposed is coated with a resist that has a low reflectance for light at the exposure wavelength, such as a multilayer resist, the projection lens can be used regardless of the wafer. Since the magnification can be measured, highly accurate magnification error measurement values can be obtained.

Further, in each embodiment of the present invention, the magnification error is corrected by adjusting the lens element 21b in the projection lens 21, but the reticle 19 or the entire projection lens 21 is adjusted so as to adjust the distance between the projection lens 21 and the reticle 190. may be moved in the direction of the optical axis AX.

Furthermore, the projection magnification can be minutely adjusted by making the space that constitutes the air gap between the plurality of lens elements in the projection lens 21 into an air-a-F structure and varying the pressure in this space.

Furthermore, in each of the above embodiments, the openings 20L and 2 of the reticle 19 are
Although the light for all OR illumination was set to have the same wavelength as the exposure light, it may also be light of a wavelength that does not sensitize the photoresist. In this case, the conjugate relationship between the apertures 20L and 20R of the reticle 19 and the wafer 26 (or the reference mark plate) is broken due to the chromatic aberration of the projection lens 21, so the reticle 19 (especially the aperture 2OL, 2
0R) and the projection lens 21, a correction optical system such as a lens mirror or a prism may be provided.

(Effects of the Invention) According to the present invention, the two
The projection optical system projects the two apertures onto a projection substrate such as a sea urchin or a reference mark plate, and scans the aperture image and the diffraction grating pattern formed in advance on the projection substrate to diffract the image. Since the high-order diffracted light (light of an order other than specularly reflected light) from the grating pattern is photoelectrically detected through the entire projection optical system and the aperture, the asymmetrically intense reflected light (scattered light) generated at the edge of the aperture is ) There is no need for full photoelectric detection, and even if the illumination light has angular characteristics, it is possible to detect magnification errors with high precision. Furthermore, there is an advantage that not only the projection magnification error of the projection optical system itself but also the relative magnification error between the pattern image of the mask and the pattern on the wafer can be detected.

[Brief explanation of the drawing]

FIG. 1 is a schematic optical layout diagram of a reduction projection type exposure apparatus suitable for an embodiment of the present invention, FIG. 2 is a plan view of a reticle, and FIG.
The figure shows an example of the arrangement of chips on a wafer, and Figure 4 shows the intensity distribution of light illuminating the entire aperture of the reticle.
Figure (a) is a cross-sectional view showing the specific shape of the diffraction grating pattern, Figure 5 (b) is a plan view of the diffraction grating pattern, and Figure N6 is a circuit diagram of the control system of the exposure apparatus shown in Figure 1. 7 is a flowchart explaining the operation of the first embodiment, FIG. 8 is a diagram showing the arrangement of an aperture image and a diffraction grating pattern when measuring magnification error, and FIGS. 9 and 10. The figure is a waveform diagram showing the waveform of a photoelectric signal corresponding to the diffracted light from the diffraction grating pattern, and FIGS. 11 and 12 are diagrams for explaining the magnification error measurement operation according to the second embodiment of the present invention. [Explanation of symbols of main parts] 16L-16R... Spatial filter, J8L.
18R...Detection g, 19...Reticle (mask). 20L/20R...Aperture, 21...Projection lens. 26...Uni, -, 27・28...Diffraction grating nokuturn Applicant: Nippon Kogaku Kogyo Co., Ltd. Agent Takashi Watanabe Figure 3 Figure 5 Figure 6

Claims (2)

[Claims] (1) Two apertures formed at a predetermined design interval between the predetermined points on a baseline passing through a predetermined point on the mask are illuminated, and *1-projection of the two apertures is performed by the projection optical system. The two aperture images are sequentially scanned in the direction of the base line by a diffraction grating pattern having a plurality of grating elements arranged in a direction intersecting the direction of the base line at a predetermined angle, and high-order diffraction light generated from the diffraction grating pattern is detected. Photoelectric detection is performed via the projection optical system and the aperture, the entire interval between the two aperture images is measured based on the photoelectric signal and the scanning position, and the measured interval and the crotch interval are compared to determine the projection optical system. A magnification error detection method characterized by detecting all magnification errors of a system. (2) Two apertures formed at a predetermined design interval between the predetermined points on the mask are illuminated on a baseline passing through all predetermined points on the mask, and the images of the two apertures are projected by the projection optical system. , are arranged at predetermined intervals in the direction of the baseline, and
The two aperture images are each scanned in the base line direction by two diffraction grating patterns each having a plurality of grating elements arranged in a direction intersecting the base line direction at a predetermined angle, and the height generated from each of the two diffraction grating patterns is The second-order diffraction light is photoelectrically detected through the projection optical system and each of the two apertures, and the two apertures are detected based on the electric signals of both parties and the scanning position.
A magnification error of the projection optical system is detected by measuring the difference between the distance between the two aperture images and the distance between the two diffraction grating patterns, and comparing the measured order difference with a design difference. magnification error detection method.