JP3448812B2 - Mark detection apparatus, exposure apparatus having the same, and method of manufacturing semiconductor element or liquid crystal display element using the exposure apparatus - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to various manufacturing apparatuses used in the process of manufacturing semiconductor elements, liquid crystal display elements, etc., a mark detection apparatus incorporated in an inspection apparatus, an exposure apparatus having the same, and an exposure apparatus therefor. The present invention relates to a method for manufacturing a semiconductor element or a liquid crystal display element using the.

[0002]

2. Description of the Related Art Conventionally, in a manufacturing process of a semiconductor element, a liquid crystal display element, etc., an alignment mark formed on a wafer in which a semiconductor element is formed or a glass plate in which a liquid crystal display element is formed is detected. A technique for precisely positioning a wafer or a plate at a desired position, a so-called alignment technique, has become an important issue. Especially for aligners and steppers that superimpose and expose precise fine patterns on wafers and plates, dozens of times
It is necessary to perform the overlaying work about 20 times, and it is required to improve the alignment accuracy for each overlaying. The alignment accuracy is generally evaluated by the absolute error and the reproducibility, and both of them need to be sufficiently small. With an actual aligner or stepper, the mark pattern (unevenness step shape) under the resist layer is optically detected while the photoresist (thickness: 0.5 to 2 μm) is applied on the wafer or plate. ing. Usually, in processes such as wafers, a resist layer is applied on the aluminum layer deposited on the base,
It is known that the degree of unevenness of the mark pattern becomes extremely small, and therefore the accuracy of the optical mark pattern detection system, that is, the alignment apparatus, varies greatly depending on the layer structure on the wafer. Therefore, the structure of a stepper incorporating such an alignment device is shown in FIG.
Will be described with reference to.

A reticle (mask) 3 having a pattern area to be projected at a predetermined magnification by the projection lens 1 is held by a reticle holder 2 at the peripheral portion. The reticle holder 2 is provided so that it can be finely moved in the x and y directions in a plane perpendicular to the optical axis AX of the projection lens 1, and is positioned by a reticle alignment system RA so that the center point of the reticle 3 coincides with the optical axis AX as much as possible. To be done. The pattern of the reticle 3 located on the object plane side of the projection lens 1 is image-projected onto the wafer 12 located on the image plane side. At this time, the reticle illumination light for exposure is g-line (wavelength 436 nm), i-line (wavelength 365 nm), excimer laser (wavelength 248 nm, 193 n).
m) etc. ultraviolet rays are used. The wafer 12 is two-dimensional (x,
It is mounted on the stage 11 that moves in the y direction) and moves along a predetermined path during alignment or during exposure. A reference mark plate FM is fixed on the stage 11 at substantially the same height as the wafer 12 and used for various checks of the alignment system. Further, a movable mirror 10 for reflecting a beam from a laser interferometer 13 that measures the amount of movement of the stage 11 in the x direction and the y direction is fixed around the stage 11. Then, the stage 11 is two-dimensionally driven by the motor 14. The stage control unit 17 controls the motor 1 based on the length measurement value by the interferometer 13.
4 drive is controlled.

Before the exposure, it is necessary to relatively align the reticle 3 and the wafer 12 with each other. For that purpose, here, the mark on the wafer 12 (or the reference mark plate FM) is projected through the projection lens 1. Mark) is detected
A TL / alignment system 5 and an off-axis alignment system 9 fixed outside the projection optical path (lens barrel) of the projection lens 1 are provided. In general, since the projection lens 1 is aberration-corrected only for one specific wavelength,
TTL that detects the wafer mark through the projection lens 1
In the alignment system, the laser light source 6 having a narrow spectral width is used as the illumination light for alignment. The beam from the laser light source 6 passes through each optical element in the TTL / alignment system 5, is deflected by a mirror 5a obliquely provided in the space between the reticle 3 and the projection lens 1, and the projection field of the projection lens 1 is projected. Incident on the periphery of. Assuming that the projection lens 1 is telecentric on both sides, the chief ray of the beam directed from the mirror 5a to the projection lens 1 is parallel to the optical axis AX, and the beam directed from the projection lens 1 to the wafer 12 is also parallel to the optical axis AX. Become.

When a mark is formed at a specific position on the wafer 12 and the mark is irradiated with the laser beam of the TTL / alignment system 5, scattered light or diffracted light is generated from the mark according to the step shape. The scattered light and the diffracted light travel backward through the projection lens 1 to the TTL / alignment system 5, and are reflected (or transmitted) by a beam splitter that separates the light transmitting system and the light receiving system, and the photoelectric sensor (photomul,
It is received by a silicon photodiode or the like). This TT
The L / alignment system 5 includes a spatial filter that shields the light component that is normally reflected by the mark on the wafer or the peripheral portion thereof and extracts only scattered light or diffracted light. Further, here, the sectional shape of the laser beam projected onto the wafer 12 from the TTL / alignment system 5 is a slit shape according to the shape of the mark, but it may be a minute circular spot.

To detect the mark position by the TTL / alignment system 5, the stage 11 is moved in the x direction or the y direction so that the mark crosses the projection point of the laser beam, and a signal output from the photoelectric sensor 7 during that time. Is stored in the waveform memory in the alignment control unit 4, and then the waveform is analyzed. Usually, in the TTL / alignment system as shown in FIG. 1, since the mirror 5a at the tip is fixed outside the pattern area of the reticle 3 and within the projection field of the projection lens 1, the laser beam The mark detection position where the mark and the wafer mark match, and the exposure position where the projected image of the pattern region of the reticle 3 and the shot region on the wafer 12 match are deviated by a predetermined constant amount (baseline amount). There is.
Therefore, it is necessary to correct and position the stage 11 with respect to the mark detection position by a slight shift amount. Such a method is generally called a site-by-site method (or field-by-field method).

On the other hand, off-axis alignment system 9
In the wafer alignment using, the basic sequence becomes the same as the site-by-site method using the TTL alignment system, only the baseline amount increases. However, the off-axis alignment system 9
Since the wafer mark is detected independently of the projection lens 1, there is no restriction on the wavelength of the illumination light for mark detection. Therefore, of the illumination light from the halogen lamp 15, as shown in FIG. The wafer 12 is illuminated with the illumination light having the broadband wavelength characteristic. In this way, the adverse effect due to the interference fringes generated by the dry stripes between the resist layer and the wafer surface is reduced, so that highly accurate alignment (mark detection) can be expected even in the imaging method. Therefore, the off-axis alignment system 9 receives the reflected light from the local area including the mark on the wafer 12 and passes the beam splitter that separates the light transmitting system and the light receiving system to display the image of the mark on the television camera ( Camera tube or CC
D) An image is formed on the image pickup surface of 8. Since the TV camera 8 generates an image signal corresponding to the contrast of the mark image,
The mark position is detected by processing the image signal in the image analysis circuit in the alignment control unit 4.

The alignment control unit 4 includes a TTL /
Based on the mark position detection result by the alignment system 5 or the mark position detection result by the off-axis alignment 9, the coordinate position at the time of exposure of the wafer stage 11 is calculated in consideration of each baseline amount, and the calculation result is calculated. Send to the stage control unit 17. The baseline amount of the TTL / alignment system 5 is based on the difference between the coordinate positions of the stage 11 when the fiducial marks on the fiducial mark plate FM are detected by the reticle alignment system RA and the TTL / alignment system 5, respectively. The baseline amount of the off-axis alignment system 9 is determined based on the difference in the coordinate position of the stage 11 when the fiducial mark is detected by the reticle alignment system RA and the off-axis alignment system 9, respectively. It

[0009]

In the apparatus shown in FIG. 1, since the TTL / alignment system 5 is configured to detect the mark on the wafer 12 through the projection field of the projection lens 1, the baseline amount is extremely large. It can be said that it is stable.
Further, since the TTL / alignment system 5 uses the laser beam as the illumination light, it has been confirmed that the TTL / alignment system 5 has sufficient detection ability even for a mark having a very small step. On the other hand, since the off-axis alignment system 9 uses broadband (white) illumination light, excellent alignment accuracy can be obtained especially when a mark on a metal wafer (a wafer having a metal layer deposited on the surface) is detected. It is characterized by being On the other hand, since the off-axis alignment system is provided independently of the projection lens 1, the baseline amount is inevitably larger than that of the TTL / alignment system 5, and when aligning a single wafer, the TTL / alignment system is used. When the system 5 and the off-axis alignment system 9 are used together, the movement distance and the number of movements of the stage 11 for alignment processing increase,
This leads to a decrease in throughput. In particular, in the TTL / alignment system 5, the mark detection range (dimension range of the slit-shaped beam) is set in the periphery (off-axis) in the projection field of the projection lens 1, so that the mark detection direction is within the circular projection field. It is defined in a direction (tangential direction) substantially orthogonal to the line extending in the radial direction. This is when the laser light source 6 of the TTL / alignment system 5 is a laser beam in the non-photosensitive wavelength range (630 nm) such as He-Ne.
By matching the direction when the beam is slitted to the radial direction and making the mark detection direction tangential,
This is to minimize the error (asymmetry of the beam shape with respect to the mark detection direction) due to the chromatic aberration of the projection lens 1. Therefore, when the mark position detection in the x direction and the y direction is performed by the TTL alignment system 5, the TTL
Mark detection position of alignment system (slit beam)
And the mark detection position (slit-shaped beam) of the TTL / alignment system for the y direction are arranged in the visual field of the projection lens 1 so as to be separated from each other by a relationship of approximately 90 °. That is, there is a problem that the mark detection areas in the x direction and the y direction are considerably separated from each other.

In recent years, an enhanced global alignment (E.GA.) method, which determines the regularity of the array of shot areas on a wafer by a statistical method, has attracted attention. G. In method A, several typical (3
Up to 9) shot areas for the x direction, y
The respective alignment marks for directions are sequentially detected. In this case, E. G. The sample alignment shot areas (shots for mark detection) designated by the method A are considerably separated on the wafer. For this reason, when each mark in the sample alignment shot area is detected by the TTL / alignment system, the movement of the stage 11 for detecting the mark in the x direction and the movement of the stage 11 for detecting the mark in the y direction must be performed separately. If the off-axis alignment system 9 also detects each mark in the sample alignment shot area (or in the vicinity of the shot area), the number of movements of the stage 11 will increase correspondingly, and the throughput will increase. However, there was a problem that it became disadvantageous.

The present invention has been made in consideration of the above problems, and it is possible to detect a mark formed on an object such as a wafer with high accuracy and high speed, and to detect the mark. It is an object of the present invention to obtain an alignment apparatus in which the number of movements of the stage is reduced and the overall throughput is prevented from decreasing.

[0012]

In the invention according to claim 1, a projection lens (1) for projecting a mark formed on a substrate (12) onto a pattern on a mask (3) on the substrate. To a mark detection device that detects the optical system (BS1, BS1) provided outside the projection optical path of the projection optical system.
OBL, M1) and the mark on the substrate are irradiated with illumination light in the first wavelength range (broadband), the light generated from the mark by the irradiation is received through the optical system, and the received light is received. A first detection system (FIA system, waveform processing processor 69) for calculating the position information of the mark by arithmetically processing the waveform signal generated by the above, and the first wavelength range for the mark on the substrate is the wavelength range. Illuminating light of a second wavelength range (He-Ne) different from each other, and receiving the interference light of the diffracted light generated from the mark by the irradiation through the optical system, and generating the waveform signal generated by the light reception. A second detection system (LI that calculates the position information of the mark by arithmetic processing)
A system, a processor 73) for waveform processing.

[0013]

According to the first aspect of the invention, since the alignment system having different detection wavelength ranges is provided, it is possible to use the optimum illumination wavelength range according to the detection target mark.
For example, it is possible to use the optimum alignment system according to various situations, such as when detecting a mark with a small step and when detecting a mark on a metal wafer, which enables highly accurate mark detection. It will be possible.
In addition, since both are off-axis alignment systems, there is a restriction on wavelength (a restriction on chromatic aberration of the projection lens).
There is no. Therefore, the problem described in “Problem” (the condition that the detection areas of each mark in the x direction and the y direction are far apart) can be solved (there is no need to separate them), and therefore the mark detection can be performed at high speed. It will be possible. Further, as in this embodiment, if the detection ranges of two or more alignment light receiving systems having different detection principles are set in different areas in the field of view of the objective optical system, for example, the same mark pattern can be detected by different detection principles. When detecting with the alignment light receiving system, since the respective detection ranges are separated from each other but are close to each other, it is possible to immediately move the mark patterns to the respective detection ranges with a small amount of movement. The amount of movement varies depending on the size of each detection range, the size of the mark pattern, etc., but is at most about 500 μm on the substrate, and can be reduced to about 150 μm in practice.

When an observation system for observing an image is provided in the observation visual field of the objective optical system, the size of the observation region is made variable so that the detection range of each light receiving system can be observed individually. You can also observe them at the same time. At this time, if the image pickup device provided in the observation system is one light receiving system and this light receiving system is of an image detection system, when the magnification of the observation system is increased, it is within the detection range of the light detection system by the image detection system. The captured mark pattern can be detected by image analysis, and when the magnification of the observation system is lowered, the detection range of another light receiving system located near the detection range of the light receiving system of the image detection method can be observed simultaneously. can do.

[0015]

Embodiments In the embodiments of the present invention, an off-axis alignment system that is not restricted by chromatic aberration is provided with two types of alignment sensors (mark detection systems), and the objective optical system is commonly used for each alignment sensor. I decided to do it. FIG. 2 is a perspective view schematically showing the overall configuration of the alignment apparatus according to the embodiment of the present invention. In FIG. 2, the reflected light (including scattered light and diffracted light) from the mark formed on the wafer is reflected by the mirror M1 and enters the objective lens OBL, and then split by the beam splitter BS1. The light reflected by the beam splitter BS1 is incident on the FIA (field image alignment) system arranged on the upper stage, and the mark image is formed in the same manner as the conventional off-axis alignment system 9 shown in FIG. As detected by the TV camera. On the other hand, beam splitter B
The light transmitted through S1 is the LIA arranged in the lower part.
(Laser interferometric alignment) Enters the system. This LIA system will be described later in detail, but two laser beams are symmetrically inclined and irradiated to the diffraction grating mark on the wafer, and the interference of two diffracted lights generated in the same direction from the diffraction grating mark. This is a method for photoelectrically detecting light, and is based on the same principle as the technique disclosed in Japanese Patent Laid-Open No. 62-56818, Japanese Patent Laid-Open No. 2-116116, or the like. Therefore, a total of four beams, that is, two laser beams for detecting the x direction and two laser beams for detecting the y direction, are provided inside the LIA system shown in FIG. 2 through the objective lens OBL. A light transmitting system for projecting onto the wafer under optical conditions and a light receiving system for individually photoelectrically detecting the interference light from the x-direction diffraction grating mark and the interference light from the y-direction diffraction grating mark are provided. There is.

Here, the FIA is transmitted through the objective lens OBL.
An example of the observation range on the wafer detected by each of the LIA system and the LIA system will be described with reference to FIG. In FIG. 3, the circular area represents the field area IF of the objective lens OBL on the wafer, and the optical axis AXa of the objective lens OBL is located at the center point.
Passes through. Further, in FIG. 3, coordinate axes X and Y are set with the center point as the origin.
Has been set. Now, a large rectangular area PF1 and a smaller rectangular area PF2 smaller than that are virtually set in the visual field area IF. The centers of the regions PF1 and PF2 are both arranged on the optical axis AXa, and switching between the case of observing the region PF1 and the case of observing the region PF2 is performed by the variable power optical system provided inside the FIA system. On the other hand, L
The crossing areas (irradiation areas) ALx and ALy of the two beams for detecting the x direction of the IA system are the small observation area P of the FIA system.
It is set at a position deviated from the F2 in the Y direction and inside the large observation area PF1 of the FIA system. Generally, the marks formed on the wafer are scribe lines (width 100-
The diffraction grating marks for the x direction and the y direction which are detected by the LIA system are often 5 μm.
It has an area of about 0 to 80 μm square. Therefore, the areas of the irradiation areas ALx and ALy by the LIA system are also set to the same size. Further, the small observation area PF2 by the FIA system is obtained when the magnification of the FIA system television camera is increased to a high magnification.
The system uses an image signal of a wafer mark (which may be a diffraction grating mark for LIA system) existing in the area PF2 to perform F
The position shift amount with respect to the index pattern inside the IA system is detected. However, the observation range by the TV camera is the area PF2.
However, the image processing range using the image signal is limited to a range of several scanning lines (in the case of CCD, an example of an array of pixels), for example, a range of 64 lines. From the above, the size of the field of view IF of the objective lens OBL on the wafer is sufficient if the diameter is about several hundreds μm to 1000 μm, but in reality, it is the small region PF2 of the FIA system and the irradiation region ALx of the LIA system. , ALy should be separated from each other with a margin, so the field of view area IF on the wafer
Is preferably at least 1 to 5 mm. The objective lens OBL is a telecentric objective optical system so that the mark position can be accurately measured even if a slight focus error occurs during mark observation. Further, in FIG. 3, the irradiation area ALy for detecting the y direction and the irradiation area ALx for detecting the x direction of the LIA system are arranged in the X direction with the Y axis interposed therebetween, but are arranged in the Y direction with the X axis interposed therebetween. Alternatively, or across the area PF2 left and right (X direction) or up and down (Y
You may divide and arrange | position. However, LIA
Mark detection range by the system, that is, irradiation area ALx, A
If the positions of Ly are made as close as possible, the area of the mark forming regions of the diffraction grating marks for the x direction and the y direction that should be formed on the wafer becomes the smallest, and considering this fact, FIG. Arrangement is desirable.

FIG. 4 shows an example of the mark shape that can be detected by the LIA system. FIG. 4A shows the arrangement of one of the plurality of shot areas SA formed in a matrix on the wafer and the associated LIA mark area ML, and FIG. 4B shows the inside of each mark area ML. The mark MLx for the x direction and the mark MLy for the y direction formed in FIG. The mark area ML is usually arranged within a scribe line on the outer periphery of the shot area SA. Therefore, in order to avoid the positional interference with the mark area ML formed accompanying the adjacent shot area, in FIG. 4A, in one scribe line extending in the X direction for each shot area and Y The mark region ML is arranged at the center of each of the scribe lines extending in the direction. However, this is the shot area S
A case where the mark areas ML are arranged on the center line LLx passing through the center point CS of A and parallel to the X axis and on the center line LLy passing through the center point CS and parallel to the Y axis, respectively. Unless the mark areas ML are arranged at two symmetrical positions sandwiching the shot area SA on each of LLx and LLy or on a line parallel to the center line thereof, scribing of four sides of one shot area SA is performed. Mark regions ML can be formed in each of the lines.

The LIA marks MLx and MLy are
When detecting at the same time in the detection area PF2 of the FIA system,
The area PF2 shown in FIG. 3 needs to be sized to include the mark area ML. When the mark detected by the FIA system is dedicated and has a shape different from the marks MLx and MLy, the mark must be provided in the scribe line.

Next, the specific structure of the FIA system will be described with reference to FIG. Non-photosensitive broadband (bandwidth 270 nm or more) illumination light is emitted from the optical fiber 16A, and the illumination light illuminates the wafer illumination field diaphragm plate 21A with a uniform illuminance via the condenser lens 20A.
The illumination light limited by the diaphragm plate 21A is reflected by the mirror 22A, passes through the lens system 23A, and the beam splitter BS2.
Incident on. The illumination light from the optical fiber 16A split by reflection by the beam splitter BS2 is reflected by the mirror M2 and enters the beam splitter BS1. After that, the illumination light illuminates a predetermined area (for example, in a small area PF2 in FIG. 3) on the wafer via the objective lens OBL and the mirror M1. In the illumination light transmission path for the wafer, the diaphragm plate 21A includes the lens system 23A and the objective lens OB.
The composite system with L is conjugate with the wafer (imaging relationship). Therefore, the illumination area for the wafer by the FIA system is uniquely determined by the shape and size of the opening formed in the diaphragm plate 21A.

Reflected light (regularly reflected light, scattered light, etc.) is generated from the wafer irradiated with the illumination light from the optical fiber 16A, and the reflected light is reflected by the mirror M1, the objective lens OBL, the beam splitter BS1, and the mirror M2. Through the beam splitter BS2, a part (about 1/2) of the reflected light proceeds toward the detection optical system. The detection optical system includes a lens system 24, a mirror 25, an index plate 26, image pickup relay lens systems 27 and 31, a mirror 28, and a beam splitter BS3. The beam splitter BS3 detects the reflected light from the wafer in the x direction. TV camera (CCD)
8X and a television camera (CCD) 8Y for y-direction detection are separately formed, and an image (mark image) of the pattern on the wafer surface is formed on the image pickup surface of each television camera 8X, 8Y.

Here, the index plate 26 is arranged conjugate with the wafer with respect to the combined system of the objective lens OBL and the lens system 24, and the index plate 26 and each of the television cameras 8X and 8Y are conjugated with respect to the relay systems 27 and 31. Is located in.
The index plate 26 is formed by forming an index pattern with a chrome layer or the like on a transparent plate, and the portion where the image of the mark on the wafer is formed remains the transparent portion. So the camera 8X,
8Y simultaneously receives the aerial image of the wafer mark formed on the transparent portion of the index plate 26 and the image of the index pattern.

By the way, as shown in FIG.
The system A has a variable magnification function, but in the system of FIG. 5, the variable magnification optical system 30 is removably provided in the image forming optical path between the index plate 26 and the television cameras 8X and 8Y. , Corresponding to it. A wavelength filter (filter that cuts a specific band) 29 is also provided in the optical paths of the relay systems 27 and 31 so that it can be inserted and removed.
This is for cutting the wavelength component of the strong laser light reflected by the wafer when the IA system is used.

Furthermore, in the system of FIG. 5, an illumination system for independently illuminating the index plate 26 is provided. This illumination system includes an optical fiber 16B, a condenser lens 20B, an illumination field diaphragm plate 21B, a mirror 22B, and a lens system 23.
The illumination light that is composed of B and is emitted from the lens system 23B is
The beam splitter BS2 enters the beam splitter BS2 from the side opposite to the wafer illumination optical path. Therefore, the illumination light from the optical fiber 16B is emitted from the beam splitter BS.
It is reflected by 2 and reaches the index plate 26 through the lens system 24 and the mirror 25. In this system, the diaphragm plate 21B has a lens system 23.
The composite system of B and 24 is arranged conjugate with the index plate 26, and the optical fiber 16B illuminates the diaphragm plate 21B with Koehler illumination.

Here, in order to clarify the function of each system, the objective lens OB on the wafer side of the beam splitter BS1
L and the mirror M1 are called a common objective system, and the system from the beam splitter BS1 to the mirror M2, the beam splitter BS2, the lens system 24, the mirror 25, and the index plate 26 is FI.
Called A light receiving system, from the lens system 27 to the TV camera 8X,
The system up to 8Y is called the FIA detection system, and the optical fiber 16
The system from A to the beam splitter BS2 is called a wafer illumination system, and the system from the optical fiber 16B to the beam splitter BS2 is called an index illumination system. Each optical lens forming each of the common objective system, the FIA light receiving system, the FIA detection system, the wafer illumination system, and the index illumination system is coaxially arranged.

As described above with reference to FIG. 3, when the television cameras 8X and 8Y are set to a high magnification so as to observe the small detection area PF2, the index pattern and the wafer marks (MLx, MLy) on the index plate 26 are set. ) And x for image analysis
y If the low magnification is set so as to detect the misalignment in both directions and to observe the large area PF1, the image signal of only one of the TV cameras 8X and 8Y is displayed because it is only for visual observation. It may be sent to a means (CRT, liquid crystal display panel, etc.) for display.

Here, an example of the arrangement of the index patterns on the index plate 26 will be described with reference to FIG. The index plate 26 is one in which a chrome layer is vapor-deposited on a transparent glass plate and an index pattern is formed by etching. In FIG. 6, a shaded portion or a portion shown by a black portion is a light-shielding portion by the chrome layer. Since the television cameras 8X and 8Y are arranged in such a relationship that they are rotated by 90 ° so that the horizontal scanning lines are located in the X direction and the Y direction, respectively, unless the image pickup surface is a square, the television cameras 8X and 8Y are arranged as shown in FIG. The imaging area PF2x and the imaging area PF2y of the television camera 8Y do not completely match. A large observation area PF1 in FIG. 6 is formed by the television camera 8X. Now, on the index plate 26, the reference for detecting the positional deviation of the wafer mark in the X direction is 2 in the X direction.
The index pattern portions RX1 and RX2 are formed at the two places, and the reference for the misregistration detection in the Y direction is the two index pattern portions RY1 and RY2 that are formed apart in the Y direction.
The respective index pattern portions RX1, RX2, RY1, R
Y2 has the same shape and size, and a transparent slit pattern of the same shape is formed therein. Further, target marks Tx and Ty are provided so that the positions of the irradiation regions ALx and ALy by the LIA system can be recognized on the index plate 26 when switching to the large observation region PF1. And the four index pattern parts RX1 and RX
After positioning the wafer stage 11 so that the regions ML of the wafer marks MLx, MLy shown in FIG. 4 are located inside 2, RY1, RY2, the television camera 8X,
The image signal from 8Y is analyzed.

FIG. 7 shows index patterns RX1, RX2, RY.
1 and 2 are enlarged views of wafer marks MLx and M.
It also shows an arrangement example of Ly. In FIG. 7, the television camera 8X having horizontal scanning lines in the X direction has an area KX formed by n scanning lines in a specific portion of all horizontal scanning lines.
Image analysis is performed for each pattern and mark located inside. On both sides of the scanning line region KX, three transparent slits formed in each of the index pattern portions RX1 and RX2 for the X direction are arranged so as to be orthogonal to the horizontal scanning line. The marks MLx and MLy on the wafer are set so as to be located between the index pattern portions RX1 and RX2 as shown in FIG. Since the scanning region KX is for the X direction, the mark MLx having a pitch in the X direction is used as the wafer mark. At this time, the mark MLy having a pitch in the Y direction is also located in the scanning region KX, but the image signal waveform corresponding to this mark MLy is ignored during processing. Similarly, in the television camera 8Y having horizontal scanning lines in the Y direction, the instruction pattern portions RY1 and RY located in the scanning area KY by n scanning lines of a specific portion of all scanning lines.
Image analysis is performed on each transparent slit 2 and the mark MLy on the wafer.

The computer that analyzes the image signal is
The setting of the start and end points of the scanning lines necessary for the analysis of each television camera, the setting of the number of scanning lines to be analyzed (maximum n), etc. are determined in advance. Further, the position of the scanning region KX in the X direction and the position of the scanning region KY in the Y direction can be changed manually while monitoring the display screen with a joystick or the like. For that purpose, a video signal corresponding to a cursor line indicating the positions of the areas KX and KY, an enclosing frame line, an arrow-shaped display pattern for designating a range, etc. is produced on the display screen, and the video signal is produced from each of the television cameras 8X and 8Y. It is better to mix and display the signal. At this time, the cursor line, the surrounding frame line, the arrow, etc. are moved to arbitrary positions on the display screen by the joystick operation.

By the way, when the wafer is illuminated by the wafer illumination system of the system shown in FIG. 5, if the wafer is left as it is, the reflected light having the intensity depending on the reflectance of the wafer surface is used as the index pattern portion R.
It reaches X1, RX2, RY1, and RY2, and a part of the reflected light thereof passes through the transparent slit and becomes a slit image on the television camera. Therefore, if there is a complicated pattern on the wafer below the index pattern part or the reflectance is extremely low, the contrast of the slit image in the index pattern part decreases, and the position of the index pattern (slit) based on the image signal is reduced. The measurement accuracy may deteriorate.

Therefore, as shown in FIG. 5, an index illumination system (a system from the optical fiber 16B to the beam splitter BS2) is provided, and the index pattern portions RX1 and R on the index plate 26 are provided.
An opening (transparent part) is provided at a corresponding position on the illumination field diaphragm plate 21B so that only X2, RY1, and RY2 are separately illuminated.
Should be provided. FIG. 8 shows the arrangement of the transparent parts on the diaphragm plate 21B. The transparent part QX1 illuminates only the index pattern part RX1, and the transparent parts QX2, QY1, QY2 illuminate only the index pattern parts RX2, RY1, RY2, respectively. . At this time, the transparent portions QX1, QX2, QY1, QY2
Are similar to the shape of the corresponding index pattern portion and are set to have a size slightly smaller than the index pattern portion.

On the other hand, the illumination field stop 21 in the wafer illumination system.
A is a wide observation area PF as shown in FIG. 3 or FIG.
Since it is necessary to illuminate the entire area of No. 1, it is possible to use a simple rectangular opening (transparent portion) that matches the area PF1. However, the index pattern portions RX1, RX2, RY1,
When it is necessary to prevent the reflected light from the wafer surface from reaching the index pattern portion in order to keep the contrast of the slit image of RY2 constant, the transparent portion of FIG. QX1, QX2, QY1, QY
It is advisable to provide a light-shielding portion having the same shape and size as those in No. 2 above. That is, the diaphragm plate may have a light-shielding portion and a transparent portion in a complementary relationship with the diaphragm plate 21B of FIG.

FIG. 9 shows the scanning area K in the state shown in FIG.
TV camera 8 corresponding to one scanning line in X and KY
An example of the waveform of the image signal output by X and 8Y is shown in FIG.
(A) is an image signal waveform VFx of the TV camera 8X, FIG.
(B) shows an image signal waveform VFy of the television camera 8Y. First, since each pattern and mark are located in the scanning region KX for detecting the X direction as shown in FIG. 7, the signal waveform VFx
Is a waveform portion Vx1 corresponding to the three transparent slit image intensities in the index pattern portion RX1, and a waveform portion Vm corresponding to the image intensity in the direction in which the analysis grids of the wafer mark MLy are arranged.
y, a waveform portion Vmx corresponding to the image intensity in the pitch direction of the analysis grid of the wafer mark MLx, and a waveform portion V corresponding to the image intensity of three slits in the index pattern portion RX2.
x2 is included in time series. Similarly, in the signal waveform VFy,
A waveform portion Vy2 corresponding to the three slit images in the index pattern portion RY2, a waveform portion Vmy corresponding to the image contrast in the pitch direction of the analysis grid of the wafer mark MLy,
And a waveform portion Vy1 corresponding to the three slit images in the index pattern portion RY1 is included in time series.

The pitch direction of the mark MLx (X direction)
1 and the waveform portion Vmy corresponding to the pitch direction (Y direction) of the mark MLy is a repetitive waveform having a large number of bottom points as shown in FIG.
This is because the reflected light that does not return to the objective lens OBL is generated at the edge position of the grating lined up in the pitch direction of each mark as shown in 0 (A), and becomes the bottom point as shown in FIG. However, if the inclination of the edge portion of the mark lattice is gentle, the reflectance of the material forming the mark lattice is extremely lower than that of the base, or the line width of the lattice itself is small, then FIG. ), It may become the bottom point at the grid position of the mark. In principle, an algorithm for waveform processing should be used to determine whether the number of bottom points is the same as the number of mark grids or the number of bottom points is about twice the number of mark grids. For example, signal waveform processing can be performed without depending on the shapes and optical characteristics of the marks MLx and MLy.

Next, a specific configuration of the LIA system will be described with reference to FIG. The LIA system is a cross-illumination of laser beams as shown in FIG. 3 in the visual field region IF of the objective lens OBL via a common objective system including a mirror M1 of an off-axis alignment system, an objective lens OBL and a beam splitter BS1. Areas ALx and ALy are formed. In order to send a laser beam to the illumination areas ALx and ALy, linearly polarized light or circularly polarized He-Ne is used.
The laser light source 6 is provided, and the beam LB from the laser light source 6 is transmitted through the shutter 40 to the beam splitter BS8.
Divide into two. The beam transmitted through the beam splitter BS8 is appropriately incident on the heterodyne two-beam conversion unit 41X via a folding mirror. In this unit 41X, the incident beam is further divided into two, and two frequency shifters (acousto-optic modulators) for shifting each of the two divided beams by different frequency and output from each frequency shifter. And a synthesizing system for eccentrically synthesizing the two beams described above. 2 synthesized by the synthesis system
The beams of the book are LB1x and LB2x as shown in FIG.
Becomes parallel to the optical axis of the system and enters the lens system 42X. The two beams LB1x and LB2x are the lens system 42.
The rear focal plane of X intersects at a predetermined angle, and the aperture plate 43X disposed on the rear focal plane is uniformly irradiated. Therefore, two beams LB1 are formed on the aperture plate 43X.
One-dimensional interference fringes are generated by the intersection of x and LB2x, and 1 in the heterodyne two-beam conversion unit 41X is generated.
Since the drive frequencies of the pair of frequency shifters are different from each other, the one-dimensional interference fringes flow in the pitch direction at a speed corresponding to the difference in the frequencies.

The two beams limited by the aperture plate 43X are partially reflected by the beam splitter BS6, and the lens system 44X and the beam splitters BS4, BS are provided.
The light enters the objective lens OBL through 1 and reaches the irradiation area ALx on the wafer. Here, the aperture formed on the aperture plate 43X is the visual field area IF of the objective lens OBL.
Among them, it is arranged so as to be conjugate with the position of the illumination area ALx by the LIA system.

Since the two beams LB1x and LB2x are incident symmetrically with respect to the X direction in the illumination area ALx by the LIA system, one-dimensional interference fringes flow in the X direction even on the wafer. ing. Therefore, the illumination area AL
Assuming that the wafer mark MLx for X-direction alignment exists in x, the pitch dimension Pg of the mark MLx and the pitch dimension Pi of the interference fringes are set to a predetermined ratio (for example, Pg / P).
When i = 2) is set, ± 1st-order diffracted light traveling in the vertical direction from the mark MLx is generated. The + 1st order diffracted light is obtained by irradiating the laser beam LB1x, for example, and the −1st order diffracted light is obtained by irradiating the laser beam LB2x. Since the two ± 1st-order diffracted lights have the same deflection direction, they interfere with each other, and at the same time, the interference intensity changes periodically at the frequency difference between the two frequency shifters, that is, at the beat frequency. So Mark MLx
The ± first-order diffracted light that is generated vertically from is called interference beat light. This interference beat light is reflected by the mirror M1 and the objective lens OB.
L, beam splitters BS1, BS4, lens system 44X
To reach the beam splitter BS6, where it is split and reaches the light-receiving aperture plate 45X. The light-receiving aperture plate 45X is arranged conjugate (imaging relationship) with the wafer with respect to the combined system of the lens system 44X and the objective lens OBL, and transmits only the reflected light (interference beat light) from the illumination area ALx on the wafer. It has such an opening. The interference beat light that has passed through the aperture plate 45X reaches the photoelectric sensor 48X via the mirror 46X and the lens system 47X. The light receiving surface of the photoelectric sensor 48X is arranged so as to coincide with the Fourier transform surface of the lens system 47X. At the same time, the light receiving surface of the photoelectric sensor 48X is the objective lens OBL and the lens system 4
It is also conjugate with a pupil plane existing between 4X and 4X (a diaphragm position or a plane having a Fourier transform relationship with the wafer plane).

On the other hand, in the LIA system for the Y direction, of the beams from the laser light source 6, the other beam split by the beam splitter BS8 is incident, and the frequencies are different from each other.
Heterodyne two-beam conversion unit 41Y for emitting two beams LB1y, LB2y, two beams LB1y, LB
A lens system 42Y that makes 2y parallel light fluxes and intersects each other on the aperture plate 43Y, a beam splitter BS5 that divides the light transmitting / receiving system, a lens system 44Y, a light receiving aperture plate 45Y, a mirror 46Y, and a Fourier transform lens system 4
7Y and photoelectric sensor 48Y, the optical arrangement and function of each member are exactly the same as the LIA system for the X direction. However, the difference from the X-direction LIA system is that the two beams LB1y and LB2y intersecting on the wafer need to be symmetrically tilted in the Y-direction, so that the heterodyne two-beam conversion unit 41Y is arranged in the X-direction LIA. System unit 41X
Position by rotating 90 ° with respect to Y on the wafer
In order to detect the direction mark MLy in the illumination area ALy, each aperture of each aperture plate 43Y, 45Y is provided in a portion of the wafer which is conjugate with the illumination area ALy.

In the case of the heterodyne system, the intensity of the interference beat light generated from the mark changes sinusoidally at the beat frequency, and the output signals of the photoelectric sensors 48X and 48Y become sinusoidal AC signals (beat frequency). Therefore, the positional deviation of the marks MLx and MLy cannot be known only from the output signal. Therefore, as shown in FIG. 11, a reference signal generating system is provided on the remaining one surface side of the beam splitter BS4. As is apparent from the split direction of the beam splitter BS4 shown in FIG. 11, the lens system 44X
A part (1/2) of the two transmitted light beams LB1x and LB1y which have passed through the beam splitter BS4 goes straight to enter the lens system 50. The two beams LB1x and LB2x from the lens system 50 intersect each other on the transmission type diffraction grating plates 53X and 53Y via the mirror 51 and the beam splitter BS7, and create one-dimensional interference fringes there. . The grating plate 53X is arranged in a conjugate with the light-transmitting aperture plate 43X with respect to the lens systems 50 and 44X, and only the portion corresponding to the opening on the light-transmitting aperture plate 43X is a transmissive diffraction grating (similar to the mark MLx). Are formed. Therefore, if the pitch of the diffraction grating on the grating plate 53X is set to the X direction so as to match the direction of the interference fringes, the grating plate 53X
Interfering beat light whose amplitude is modulated at the beat frequency is generated from the upper grating, and is generated by the lens system 54X for Fourier transform.
By receiving light with the photoelectric sensor 55X via
A reference signal (sinusoidal AC with a beat frequency) for the X-direction LIA system is created.

On the lattice plate 53X, the LI for Y direction is used.
Although the two beams LB1y and LB2y from the A system intersect at the same time, the projection area where the beams LB1y and LB2y intersect on the grating plate 53 is simply a light-shielding portion (flat surface), so immediately. It will be cut. The reference signal generating system of the LIA system for Y direction is also composed of the same member, and the beam LB1 transmitted through the beam splitter BS7.
y and LB2y intersect at a grating portion on the transmission type diffraction grating plate 53Y having a pitch in the Y direction, and the interference beat light generated from the grating portion is received by the photoelectric sensor 55Y via the Fourier transform lens system 54Y. It Here again, the two beams LB1x and LB for the X direction are arranged on the lattice plate 53Y.
Although 2x intersects, since that portion is merely a light-shielding portion, it is prevented from being cut immediately and becoming a noise in the creation of the reference signal for the Y direction.

Among the above-mentioned LIA system configurations, the heterodyne dual-beam conversion units 41X and 41Y are disclosed in, for example, Japanese Patent Laid-Open No. Hei 2
The one disclosed in Japanese Patent No. 231504 can be used as it is. 12 (A) and 12 (B) schematically show the beam transmission path of the LIA system, in which the arrangement of members is partially changed or members unnecessary for description are omitted. Figure 1
2A shows an in-plane light transmission path including the optical axes AXa and X of the alignment system, and FIG. 12B shows the optical axes AXa and Y.
The light transmission path in a plane including the axis is shown. First, LI for Y direction
In the A system, the two beams (parallel light beams) LB1y and LB2y are incident on the aperture plate 43Y for light transmission while being symmetrically inclined in the plane of FIG. 12 (B). This aperture plate 43
Two beams LB1y and LB2 that have passed through the rectangular aperture of Y
After passing through the lens system 44Y and converging on the pupil plane EP (actually becoming a beam waist), y again becomes two intersecting beams (parallel light beams) via the objective lens OBL and illuminates the mark MLy. Two beams LB1y, LB2y
In FIG. 12B, is once condensed as a spot on the pupil plane EP at a symmetrical position with the optical axis AXa sandwiched in the Y direction. However, when viewed from the direction of FIG. 12A, the two beams LB1y and LB2y are at the center of the pupil plane EP (optical axis A).
It seems that the light is condensed as a spot at the point where Xa passes. In this type of two-beam interference type alignment system, the two beams are symmetrically tilted in the grating pitch direction of the wafer marks MLx, MLy, that is, in the measurement direction, but in the non-measurement direction (direction orthogonal to the pitch direction). Is not tilted, that is, it is kept parallel to the optical axis AXa between the objective lens OBL and the wafer. From this, according to FIGS. 12A and 12B, the interference beat light vertically generated from the mark MLy also becomes a parallel light flux, which is formed as a spot at a point where the optical axis AXa on the pupil plane EP passes. Glow.

Similarly, in the LIA system for the X direction, the two beams LB1x and LB2x become parallel light beams and enter the light transmitting aperture plate 43X at a predetermined crossing angle. At this time, the two beams LB1x and LB2x are symmetrically inclined in the plane of the paper of FIG. 12A in accordance with the pitch direction of the mark MLx. Therefore, the interference beat light vertically generated from the mark MLx becomes a parallel light beam up to the objective lens OBL, and the optical axis A
The light travels in parallel with Xa, and becomes a spot on the pupil plane EP at a point where the optical axis AXa passes to collect the light.

Therefore, the interference beat light from each of the marks MLy and MLx is photoelectric sensor 48 that is conjugate with the pupil plane EP.
When detecting with Y and 48X, since both interference beat lights are mixed as they are, the light receiving aperture plates 45Y and 45X are arranged on the surface conjugate with the wafer, and both interference beat lights are in the image plane. By utilizing the fact that they are separated, they are selectively extracted.

FIG. 13 shows the above-mentioned FIA system and LIA.
The signal processing circuits of the system are schematically shown and are provided in the alignment control unit shown in FIG. FIG.
In, FIA type TV camera (CCD) 8X, 8
Y is driven by independent drive control circuits 60X and 60Y, and outputs a composite video signal. The video signals are synchronized separation circuits 61X and 6X, respectively.
Input to 1Y, the horizontal synchronizing signal HS and the vertical synchronizing signal VS are extracted. Further, the video signals extracted from the sync separation circuits 61X and 61Y are programmable gain control circuits (gain controllers) 62X and 6X.
After a predetermined gain is given by 2Y, analog-
Input to digital waveform storage units 63X and 63Y including a digital converter and a memory (V-RAM).

On the other hand, the signals HS and VS from the sync separation circuits 61X and 61Y are input to the sampling clock generation circuits 64X and 64Y, where the timings of digital conversion and memory access to the digital waveform storage units 63X and 63Y. -The clock is generated. This clock is set so as to divide a video signal obtained during one horizontal scanning period of the CCD into, for example, 1024 pixels. Minute video waveform is captured. Here, it is assumed that a maximum of 64 video waveforms can be stored. Again 1
The capture control circuit 6 is used to specify the capture position of the video waveform on the screen (the position of the horizontal scanning line in the vertical direction).
5 is provided here, and when a horizontal scanning line serving as a capture start point comes based on the signals HS and VS, the generation circuit 64 is provided.
Storage unit 63 stores timing clocks for X and 64Y, respectively.
Instruct to output to X and 63Y. The designation of the capture start point is sent from the main control circuit 66, but it can be set visually by the operator or automatically by analyzing the video waveform.

By the way, the composite video signals from the TV cameras 8X and 8Y are mixed (screen composition) by the video controller 67, and the result is displayed on the TV monitor (CR).
T) 68 is displayed. At this time, how to mix on the screen display is performed according to an instruction from the main control circuit 66, and the FIA system detects the X direction mark and the Y direction.
When simultaneously detecting direction mark detection, the television screen is divided into two, and the images at the time of mark detection in each direction are appropriately trimmed or shifted and displayed in each of the divided areas. Further, it is possible to switch so that only one of the images of the television cameras 8X and 8Y is displayed. Further, the capture control circuit 65 creates a video signal corresponding to the cursor line (or arrow) indicating the position on the television monitor 68 based on the position information of the capture start point designated by the main control circuit 66, and outputs this. It is sent to the video controller 67 and combined with the video signal from the CCD.

Now, the digital waveform storage units 63X and 63Y
Each of the video waveforms stored in each is processed by the processor 69 for high-speed waveform processing, and the positional deviation amount of the wafer mark MLx in the X direction from the center of the index pattern and the positional deviation of the wafer mark MLy in the Y direction from the center of the index pattern center. The amount is required. Information on the amount of positional deviation is sent to the alignment processing control unit 70 via the main control circuit 66 and used as alignment data associated with the shot position on the wafer. The processor 69 also individually calculates the position of the center of the index pattern (the position in the horizontal scanning direction) and sends the result to the fetch control circuit 65 via the main control circuit 66. As a result, the control circuit 65 creates a video signal of the cursor line corresponding to the center position where the wafer mark MLx or MLy should be aligned on the television monitor 68, and the video signal
It is sent to the controller 67. This is a TV monitor 6
This is convenient when the wafer stage 11 is finely moved manually while observing 8, and the mark MLx or MLy is aligned with the index pattern. Of course, a line or a marker representing the center of the index pattern may be provided on the index plate 26, but these are the marks MLx, ML.
It is meaningless unless it is placed on y or in the immediate vicinity thereof, and since it must not exist within the signal waveform processing range, the placement of the marker is naturally restricted.

The alignment processing control unit 70 transfers the wafer stage 11 to the stage control unit 17 shown in FIG.
The positioning target value is output, which is the positional deviation amount measured by the FIA system when the stage 11 is stopped and the stop position coordinate value of the stage 11 measured by the laser interferometer 13 is taken in. And the stage stop position coordinate value at that time.

On the other hand, the signal processing system of the LIA system outputs both signals of the photoelectric sensor 55X which outputs the reference signal in the X direction and the photoelectric sensor 48X which outputs the measurement signal by receiving the interference beat light from the wafer mark MLx. Digital waveform storage 71X for input, photoelectric sensor 55Y for Y direction
, A digital waveform storage section 71Y for inputting a reference signal from the photoelectric sensor 48Y for receiving the interference beat light from the wafer mark MLy, a clock generation circuit 72,
And a high-speed waveform processing processor 73.
The waveform storage units 71X and 71Y individually digitally convert the reference signal and the measurement signal, and store the waveform (both of which are sinusoidal) in a memory (RAM) for a predetermined period. The timing of digital waveform sampling in the waveform storage units 71X and 71Y is performed in response to the pulse from the clock generation circuit 72, and the sampling of the reference signal and the measurement signal is performed at exactly the same timing. However, the storage unit 71X for the X direction and the storage unit 71Y for the Y direction
The sampling start between and does not necessarily match. Then, the processor 73 vector-calculates the reference signal waveform and the measurement signal waveform stored in each storage unit by using Fourier integration, and obtains the phase difference (within ± 180 °) of both waveforms. In the LIA method, the wafer marks MLx and ML are compared with the pitch Pi of the interference fringes generated on the wafer.
When the pitch Pg of y is doubled, the calculated phase difference ±
Δφ is 4 · ΔX / Pg = Δφ / 180 (or 4 · ΔY / with respect to the positional deviation amount ± ΔX (or ± ΔY) in the pitch direction)
Pg = Δφ / 180).

From this, the processor 73 calculates the phase difference ± Δφ, immediately calculates the positional deviation amounts ΔX and ΔY, and sends them to the alignment processing control unit 70 as alignment data. Next, how to use the FIA system and the LIA system described above will be briefly described. First, as shown in FIG. 5, when the laser cut filter 29 is placed in the optical path of the FIA observation optical system and the low magnification state is set, the observation area by the television camera 8X (or 8Y) becomes
The PF1 shown in FIG. 3 or 6 is obtained. At this time L
If the beams from the IA system irradiate the areas ALx and ALy, respectively, the areas ALx and ALy are also brightly observed on the television monitor 68. In general, the illuminance on the wafer of the beam used in the LIA system is much stronger than the illuminance of the illumination light by the FIA system.
By inserting 9 in the region PF1 in which the LIA-based beam irradiation regions ALx and ALy are wide with good contrast without causing halation on the television monitor 68, that is, on the imaging surfaces of the television cameras 8X and 8Y. Can be observed. If the LIA laser light source 6 is He-Ne, the irradiation areas ALx and ALy are strongly red, so that the TV cameras 8X and 8Y and the TV monitor 6 are provided.
When all 8 are in color, they are observed as reddish intensified in the monitor screen.

Further, the beam from the LIA laser light source 6 can be shielded by the shutter 40 as shown in FIG. 11, so that the wafer can be used for a long time at the time of mark detection by the FIA system or manual visual observation. When staying at the same position, close the shutter 40 and irradiate a strong laser beam to the areas ALx, ALx.
It prevents the resist in y from being exposed. Red light such as a He-Ne laser causes almost no problem if the illuminance on the wafer is extremely small, but if the illuminance becomes high, the sensitivity to the red light is more or less unless the sensitivity to the red light is completely zero. Photosensitization progresses with time. Therefore, a timer is provided in the stage control unit 17 shown in FIG. 1 to measure the time after the wafer stage 11 is stopped. Further, when the irradiation areas ALx and ALy of the LIA system exist on the wafer when the stage is stopped, If the timer is activated and the stage 11 does not move after a certain period of time, it is preferable to close the shutter 40 in FIG. In that case, the opening of the shutter 40 is performed after the next movement of the wafer stage 11.

When a mark is detected by a FIA system and an image is automatically aligned, a variable-magnification optical system 30 is inserted in the optical path instead of the filter 29 shown in FIG. 5 to provide a high-magnification observation system. Then, TV camera 8X (or 8
The observation area of Y) is switched to the small area PF2 as shown in FIGS. At this time, TV cameras 8X, 8Y
Since a high-quality mark image is formed using the entire wide wavelength band of the illumination light from the FIA optical fibers 16A and 16B, image processing using a video signal is extremely well performed. Furthermore, in the observation area PF2 observed by the television cameras 8X and 8Y, the irradiation areas ALx and AL of the LIA system are
Since y does not exist, it has no effect on the image processing system. When the observation area of the FIA system is set to PF2 and the positions of the wafer marks MLx and MLy are measured in the state shown in FIG. 7, when the wafer marks MLx and MLy are continuously measured by the LIA system, the wafer stage 11 is used. Is slightly translated in the Y direction, and the irradiation area AL of the LIA system is
The marks MLx and MLy may be positioned directly below x and ALy, respectively. In this case, the marks MLx, ML from the inside of the small observation area PF2 by the television cameras 8X, 8Y.
There is also an inconvenience that y is output and it cannot be observed on the TV monitor 68. Therefore, the video signal waveform is captured in the small observation area PF2 of the FIA system and stored in the storage unit 63.
Simultaneously with the completion of storing in X and 63Y, the variable power optical system 30 and the filter 29 may be switched, and the wafer stage 11 may be moved in the Y direction. The movement amount of the stage 11 at this time is about 500 μm at most. Also, depending on the TV camera, the irradiation area AL with strong laser beam
If an image of x or ALy is received even for a moment, it may be burned as an afterimage on the image pickup surface.
9 may be inserted.

When a mark is detected using only the LIA system, if a filter 29 is inserted in the optical path instead of the variable power optical system 30, the beam irradiation areas ALx and ALy of the LIA system will be generated within the wide observation area PF1. , Mark MLx, MLy
And can be observed at the same time. At this time, since the illumination light from the FIA system illuminates the marks MLx and MLy at the same time, L
The IA type photoelectric sensors 48X and 49Y also receive the reflected light. However, the interference beat light obtained from the mark by the LIA system is modulated at a constant beat frequency (several KHz to several tens KHz).
The A system is a system that detects only the phase information of the modulation period (that is, does not depend on the amplitude of the signal), and LI
Since the illumination intensity of the FIA system is much smaller than the laser beam intensity in the irradiation regions ALx and ALy of the A system, it does not substantially cause a detection error of the LIA system.

Although the embodiment of the present invention has been described above, the observation area PF2 for image analysis and the beam irradiation areas ALy and ALx for laser alignment are necessarily shown in FIG. 3 (or FIG. 6). It is not necessary to have the arrangement relationship shown in FIG. 2 and it may be located in any large observation area PF1 as long as they are separated from each other. The FIA-based index pattern portions RX1, RX2, RY1, and RY2 are arranged so as to be slightly displaced so that the position of the optical axis AXa does not become a symmetrical point as shown in FIG. In addition to the case where MLy and MLx are juxtaposed in the X direction, FIG.
Even in the case where they are juxtaposed in the Y direction as in (B), the signal waveform can be detected in the same manner.

Further, although the beam splitter BS1 shown in each of FIGS. 2, 5 and 11 is an amplitude division type, it is a polarization beam splitter of a wavefront division type and is used as a transmission beam from the LIA system (the beam in FIG. 11). Splitter BS4 to BS1
When the deflection directions of the beam sent to the FIA system and the illumination light from the FIA system (the light sent from the mirror M2 to the beam splitter BS1 in FIG. 5) are complementary to each other, the light amount of each alignment system is used. Increases efficiency.

In the LIA system shown in FIG. 11, the frequency shifter in the heterodyne two-beam conversion unit 41X for producing two beams for the X direction and the frequency shifter in the two-beam conversion unit 41Y for the Y direction are provided. , Both have the same drive frequency, that is, beam LB1x and beam LB
1y has the same frequency, beam LB2x and beam LB2y
The frequencies of the four beams are the same, but the frequencies of the four beams may be different, and the beat frequency obtained in the X direction and the beam frequency obtained in the Y direction may be set so as not to have an integral multiple relationship. With this arrangement, even if the Y-direction transmission beam or the interference beat light is mixed with the interference beat light from the X-direction mark MLx, it is almost completely separated at the stage of signal processing (Fourier integration, etc.). it can. As an example, when the frequency of the original beam LB is f0,
The beam LB1x is set to f0 +80.0 MHz and the beam LB2x is set to f0 +80.23 MHz to set the beat frequency in the X direction to 23.
It is preferable to set the beam LB1y to f0 +90.0 MHz, the beam LB2y to f0 +90.47 MHz, and the beat frequency in the Y direction to 47 KHz.

In the FIA system shown in FIG. 5, the illumination light from the wafer illumination system and the illumination light from the index illumination system have the same wavelength band, but the illumination from the optical fiber 16B of the index illumination system is used. The lights do not need to be particularly the same as long as they are within the range where the imaging optical system (27, 30, 31) up to the television camera is aberration-corrected by achromatism. Further, the combined system of the objective lens OBL of the common objective optical system and the imaging lens system 24 also needs to be achromatic according to the wavelength band of the illumination light to the wafer.

Further, the intensity of the illuminating light to the wafer and the intensity of the illuminating light to the index plate are independently adjustable, and the waveform of the image signal picked up by the television cameras 8X and 8Y is displayed on the television monitor 68. It is advisable to adjust the illumination intensity ratio of both after confirming the above. At this time, the gains of the gain controllers 62X and 62Y in FIG. 13 may be set to be optimum accordingly.

[0058]

According to the first aspect of the present invention, since the alignment system having different detection wavelength ranges is provided, the optimum illumination wavelength range can be used according to the detection target mark. For example, it is possible to use the optimum alignment system according to various situations, such as when detecting a mark with a small step and when detecting a mark on a metal wafer, which enables highly accurate mark detection. It will be possible. Moreover, since both are off-axis alignment systems, there are no wavelength restrictions (constraints regarding chromatic aberration of the projection lens). Therefore, the problem described in “Problem” (the condition that the detection areas of each mark in the x direction and the y direction are far apart) can be solved (there is no need to separate them), and therefore the mark detection can be performed at high speed. It will be possible. Further, as described in the present embodiment, if the detection ranges of two alignment systems having different detection methods are separated from each other within the field of view of the alignment objective optical system, they are detected when the mark pattern on the object is detected. Two different alignment systems can be used alternately by simply moving the object between the two detection areas. Therefore, it becomes possible to detect marks and measure displacements at high speed. When the image detection method is incorporated into one of the alignment systems as in the embodiment, the observation field of view of the image observation system can be switched between a large area and a small area within the field of the objective optical system, and In addition to enabling automatic alignment and observation using image signals, the detection range of the other alignment system is set to be located outside the small area, and the detection operation of the mark pattern by the other alignment system is performed in the large area. Since the observation and the visual alignment of the image alignment system are enabled, there is an effect that the marks can be observed efficiently during the alignment.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a conventional projection exposure apparatus.

FIG. 2 is a perspective view showing a schematic configuration of an alignment apparatus according to an embodiment of the present invention.

FIG. 3 is a plan view showing a visual field area of an objective optical system and a detection range of each alignment sensor.

FIG. 4 is a plan view showing mark arrangement and mark structure on a wafer.

FIG. 5 is a perspective view showing the configuration of a FIA system.

FIG. 6 is a plan view showing the arrangement of FIA index patterns.

FIG. 7 is a plan view showing an arrangement example of index patterns and wafer marks.

FIG. 8 is a plan view showing the configuration of the diaphragm plate of the index illumination system.

FIG. 9 is a diagram showing a waveform of an image signal detected by the FIA system.

FIG. 10 is a diagram showing a relationship between a waveform of an image signal and a mark cross section.

FIG. 11 is a perspective view showing the configuration of an LIA system.

FIG. 12 is a diagram schematically showing a light transmission system of LIA system.

FIG. 13 is a block diagram showing a configuration of each signal processing circuit of FIA system and LIA system.

[Explanation of symbols for main parts]

OBL Objective lens LIA Laser interferometric alignment system FIA Field image alignment system MLx, MLy Diffraction grating mark PF1 FIA system observation area at low magnification PF2 FIA system observation area at high magnification ALx, ALy LIA system Beam irradiation area

Claims (4)

(57) [Claims] 1. A mark detection device for detecting a mark formed on a substrate without using a projection lens for projecting a pattern on a mask onto the substrate, the mark detection device being provided outside a projection optical path of the projection optical system. Illuminating light in the first wavelength range to the optical system and the mark on the substrate.
The light generated from the mark by the irradiation
Light is received through the academic system and the waveform signal generated by the received light is played.
A first detection system that calculates the position information of the mark by arithmetic processing, and the first wavelength range for the mark on the substrate is the wavelength range.
By irradiating the illumination light of the second wavelength range with different
The interference light between the diffracted lights generated from the mark
Receives light through the system and calculates the waveform signal generated by the received light
A second detection system that processes and calculates the position information of the mark, the mark detection device. 2. Use in the first and second detection methods
Each of the illuminating light is applied to the substrate coated on the substrate.
It has a wavelength range that is non-photosensitive to the photoresist.
The mark detection device according to claim 1, which is a characteristic . 3. A fine pattern formed on a mask,
It is an exposure device that transfers onto a substrate via a projection optical system.
And the position information of the mark obtained by the mark detection device , comprising the mark detection device according to claim 1 or 2.
The mask and the substrate relative to each other based on the report.
An exposure apparatus characterized by matching . 4. Using the exposure apparatus according to claim 3,
Transfer the fine pattern on the mask onto the substrate
Semiconductor device or liquid crystal display characterized by including a step of
Device manufacturing method .