JP3199042B2 - Semiconductor device manufacturing method and exposure method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting the position of a substrate, and more particularly to a method for detecting the position of a substrate to be transferred in an exposure apparatus.

[0002]

2. Description of the Related Art Conventionally, in a semiconductor manufacturing apparatus (exposure apparatus, repair apparatus, inspection apparatus, etc.), it has been required to position a semiconductor wafer on which a chip circuit is to be formed with high precision in the apparatus. In particular, in an exposure apparatus, it is necessary to accurately detect the position of the circuit pattern on the wafer in advance in order to expose the circuit pattern of the reticle or mask on the circuit pattern on the wafer precisely.

[0003] The miniaturization of circuit patterns has reached the submicron region, and the line width rule is currently 0.4 to 0.6 µm.
Exposure apparatuses for mass production of about 16 Mbit D-RAM have been prototyped. These exposure apparatuses require alignment technology that matches the line width rule, and the accuracy of the alignment mark detection sensor alone for alignment is as follows:
About 1/10 of the line width rule must be cleared. As one alignment (mark detection) technique for obtaining such high accuracy, a diffraction grating pattern on a wafer is irradiated with a static interference fringe, and the interference fringe and the diffraction grating pattern are relatively moved to obtain a diffraction grating pattern. A method of aligning a wafer based on a change in the intensity of the generated interference light is disclosed in, for example, US Pat. No. 4,636,077.
Issue. This method basically measures displacement using the fact that the relative position change between the diffraction grating pattern and the interference fringe uniquely corresponds to the intensity change (sinusoidal level change) of the interference light. Patent No. 3,726
No. 595 is applied.

On the other hand, another position (displacement) measuring method using an interference fringe and a diffraction grating pattern is disclosed in JP-A-61-21.
As disclosed in Japanese Patent No. 5905, the interference fringes are moved at high speed in the pitch direction of the diffraction grating pattern, and a reference signal corresponding to the photoelectric signal (AC) of the interference beat light from the diffraction grating pattern and the moving speed of the interference fringes. The position shift of the diffraction grating pattern (± 交流 of the grating pitch) from the phase difference with the signal (AC)
(Or an integral multiple thereof) is also known. This method is also called a heterodyne method because it uses an optical beat signal, and the method using the above-mentioned static interference fringes is called a homodyne method. In the heterodyne method, the phase difference between two AC signals of the same frequency (measurement signal and reference signal) uniquely corresponds to the displacement of the diffraction grating pattern.
An extremely high resolution can be obtained even with a meter or the like. For example, if the pitch P of the diffraction grating pattern on the wafer is 2 μm (1
μm width line and space) and the resolution Δθ of the phase meter is ± 0.5 °, ± 1/4 of the grating pitch P is ± 18 of the phase difference in the heterodyne method under a certain condition.
Since it is proportional to 0 °, the displacement measurement resolution ΔX is obtained from the following relationship.

ΔX / Δθ = (P / 4) / 180 Therefore, under the above conditions, ΔX ≒ 0.0014 μm, and
This is a very high resolution. In addition, since the phase difference is obtained from the average of the signal waveform during a certain time (mSec order), high stability can be obtained. Furthermore, it is only necessary to find the phase change, not the signal waveform level change.
The alignment does not greatly depend on variations in signal intensity unlike the homodyne method.

[0006]

As described above, in the conventional homodyne method and the heterodyne method, the same displacement information can be obtained every 1/2 of the pitch P of the diffraction grating pattern on the substrate. In the case of a cycle, if there is a displacement of an integral multiple of the period, misalignment will be performed.
In this type of mark detection method, it is expected that the pitch P of the diffraction grating pattern becomes smaller year by year, and even if P = 2 μm as currently considered, the diffraction grating pattern can be formed within a range of ± 0.5 μm. Pre-alignment has to be performed, and the specification is almost impossible to achieve with conventional mechanical pre-alignment alone.

Accordingly, the present invention provides an alignment apparatus having an alignment system having a high resolution but a narrow range of true mark misregistration, such as the homodyne and heterodyne methods, while solving the above-mentioned problems while detecting the mark. It is an object of the present invention to minimize the lengthening of the (alignment) operation, that is, the decrease in throughput in the alignment sequence.

Further, the present invention provides a lattice pattern which can be easily detected by devising the shape of the pattern.
The purpose is to improve the throughput.

[0009]

Means for Solving the Problems To solve the above problems, according to the present invention, in order to detect the position of a semiconductor substrate, a plurality of lines and spaces are specified on the semiconductor substrate. In a method of manufacturing a semiconductor device for forming a grid mark formed by arranging in a direction, the grid mark can be automatically detected by an alignment detection system , and a center of the grid mark among the plurality of lines can be detected.
The line (M1) constituting the section intersects the arrangement direction
The other lines (M
By being formed longer than 2), the central part
The constituent line (M1) can be visually identified.
In another aspect of the present invention, a plurality of lines and spaces are formed on a substrate.
A scan detects grating mark which is configured by arranging in a predetermined direction, an exposure method for exposing a pattern of a mask onto a substrate, together with the grating mark can be automatically detected by the alignment detection system, said plurality No
Of the grid lines that constitute the center of the grid mark
(M1) is the case with respect to the direction crossing the arrangement direction.
Formed longer than other lines (M2) constituting child marks
By being, an identifiable by eyes <br/> view of the line constituting the center line constituting the center portion
(M1) to visually detect the grid mark position.

[0010]

DESCRIPTION OF THE PREFERRED EMBODIMENTS In the present embodiment, the structure or arrangement of mark patterns is determined in advance so that two mark detecting means can detect a mark pattern continuously in time. A reduction in throughput due to the use of two mark detection means is suppressed.

Here, the principle configuration and operation of the alignment system in the present embodiment will be described with reference to FIG.
In FIG. 1, a diffraction grating pattern is formed as a main mark M2 on a wafer W, and a sub mark M1 is formed at a certain distance from the diffraction grating pattern. The projection lens PL as an objective optical system is arranged between the wafer W and the reticle R, and makes the reticle R and the wafer W conjugate with each other under the wavelength of the exposure light. The alignment system includes an alignment objective lens OBJ, a beam splitter BS1, a mirror RM lens system GL, a beam splitter BS2, a photoelectric detector PED, and the like. The projection lens PL has a telecentric system on the wafer W side. Three beams LB0, LB1, LB2 having different wavelengths from the exposure light are incident on the alignment system as mark illumination light.
The beam LB0 is reflected by the beam splitter BS2 and enters the lens system GL, where the pupil (entrance pupil) E of the projection lens PL is provided.
The light passes through the center of a plane EP ′ conjugate with P and is condensed on a plane IP in space by the objective lens OBJ, and then re-images on the wafer W through the center of the pupil EP of the projection lens PL. As a result, a circular or slit-shaped spot light (first detection area) of the beam LB0 is irradiated onto the wafer W, and the spot light and the mark M1 on the wafer W are relatively scanned, so that the mark M1 Is scattered light or diffracted light ± DL
Occurs. The reflected diffracted light is reflected by the beam splitter BS1 via the projection lens PL and the objective lens OBJ, and the light receiving elements PDa, PDa on the light receiving surface of the photoelectric detector PED in which the diffracted and scattered light other than the specularly reflected light are conjugate to the pupil EP. It is detected by PDb or light receiving elements PDc and PDd. The above beam LB0, beam splitters BS1, SB2, lens system GL, objective lens OBJ, and light receiving elements PDa, PD
b, PDc and PDd constitute the first mark detecting means of the present invention.

On the other hand, two coherent beams LB1, LB
2 has a wavelength substantially equal to that of the beam LB0 (strictly, it differs by about several tens MHz as described later), but has a frequency different from each other by about several KHz to several tens KHz, and has a lens through the beam splitter BS2. The light enters the system GL from two directions, and is focused on the pupil conjugate plane EP ′ in a point-symmetric relationship with respect to the pupil center. After that, the beams LB1 and LB2 pass through the objective lens OBJ, and are converted into parallel light beams, and are converted into the plane IP.
And enters the projection lens PL. Two beams L
B1 and LB2 converge on the spot again in the pupil EP, and then intersect on the wafer W as parallel light beams symmetrically inclined with respect to the optical axis of the projection lens PL. The intersection of the two beams LB1 and LB2 corresponds to a second detection area, and an interference fringe having a pitch corresponding to the intersection angle of the two beams LB1 and LB2 is formed on the wafer W at a speed corresponding to the frequency difference. Will flow in the direction. The main mark M2 is located at the position of this interference fringe.
Is present, the first-order diffracted light vertically generated by the irradiation of the beam LB1 and the first-order diffracted light vertically generated by the irradiation of the beam LB2 interfere with each other, and the interference beam light BTL is vertically incident on the projection lens PL. . This interference beam light BTL is transmitted through the objective lens OBJ and the beam splitter BS1 to the light receiving element P on the optical axis of the photoelectric detector PED.
The light is focused on D0 as a spot light. Above, beam LB
1, LB2, beam splitters BS1 and BS2, lens system GL, objective lens OBJ, and light receiving element PD0 constitute the second mark detecting means of the present invention. Here, the surface IP is shifted from the reticle R in the optical axis direction, but this is due to chromatic aberration caused by the shift of the beams LB0, LB1, and LB2 from the exposure wavelength. The main mark M2 has a periodic structure in the left-right direction in the figure, and the diffraction grating element extends in a direction perpendicular to the plane of the drawing, and the beams LB1, LB2
Are arranged in the lattice pitch direction of the main mark M1 in the pupil EP. Further, the beam LB0 is shaped by a cylindrical lens or the like so as to be condensed in a slit shape in the pupil EP, and is incident on the beam splitter BS2. As a result, the spot of the beam LB0 on the wafer W is shifted to the pupil EP
It has a slit shape extending in a direction perpendicular to the longitudinal direction inside. The first mark detecting means is disclosed in detail in, for example, JP-A-61-128106.

Using the first mark detecting means and the second mark detecting means as described above, the first mark detecting means exclusively detects the sub-mark M1 and sets the position of the main mark M2 to an integral multiple of ± P / 4. The deviation is specified, and the second mark detecting means detects a positional deviation within ± P / 4 of the main mark M2. Here, in the present invention, the order of operation of the two mark detection units may be either order. As described above, in the present invention, the first mark detecting means as the sub-alignment system,
By combining the second mark detection means as the main alignment system, the mark detection operation can be continued, thereby preventing an extremely lower throughput and achieving a reliable main alignment. . Next, the configuration of a positioning device according to an embodiment of the present invention will be described with reference to FIGS. FIG. 2 shows an alignment system of the projection exposure apparatus. It is assumed that the reticle R is fixed on a reticle stage RST and is accurately positioned in advance with respect to the optical axis AX of the projection lens PL.
The wafer W is placed on a two-dimensionally movable stage 16, and this stage 16 is moved at high speed in a horizontal plane by a driving system such as a motor 17 and a feed screw 18. The coordinate position of the wafer stage 30 is determined by a laser light source 30, a beam splitter 31, a movable mirror 32 fixed to the wafer stage 30,
The measurement is sequentially performed by a laser interferometer (interferometer) including a fixed mirror 33 fixed to the projection lens PL and a receiver 34 for photoelectrically detecting an interference fringe. The signal from the receiver 34 is sent to the interferometer counter, and the amount of movement of the stage 16 is measured with a resolution of, for example, 0.02 μm.

On the other hand, a beam from the laser light source 1 for alignment is incident on a beam transmitting system ABO, and is converted into illumination beams LB0, LB1, LB2 suitable for the first mark detection system and the second mark detection system, respectively. The converted light is incident on the off-axis position (peripheral part of the projection field) of the projection lens PL via the beam splitter 6X, the objective lens 7X, and the mirror 13.
In FIG. 2, the two beams LB1 and LB0 are the beams L
Since they are arranged in a direction perpendicular to the paper surface with B0 interposed therebetween, they are not shown here. Now, the beam LB0 passing through the center of the pupil EP of the projection lens PL becomes a slit-like spot light for irradiating the wafer W telecentrically. At this time, for example, a beam spot 71 as shown in FIG. 3 is formed on the wafer W, and the longitudinal direction of the spot 71 is configured to be directed to the optical axis AX of the projection lens PL.

Therefore, when the stage 16 is scanned in a direction orthogonal to the longitudinal direction of the beam spot 71 under the measurement of the laser interferometer, the sub mark M1 can be moved across the spot 71 as shown in FIG. it can. In the present embodiment, the sub mark M1 is formed in a diffraction grating shape in which a plurality of minute rectangular patterns are arranged at a constant pitch in the longitudinal direction of the beam spot 71, and is orthogonal to the diffraction direction (pitch direction) of the diffraction grating pattern of the main mark M2. Direction periodic structure. At the moment when the spot 71 overlaps the sub mark M1, the sub mark M
From FIG. 1, the diffracted light (±
Primary light, ± secondary light...) ± DL is generated. These diffracted lights ± DL are reflected by the mirror 13X via the projection lens PL, are reflected by the beam splitter 6X through the objective lens 7X, and reach the photoelectric detector 8X (PED in FIG. 1).
The photoelectric detector 8X, like the PED shown in FIG. 1, includes a plurality of light receiving elements divided by a pupil conjugate plane to receive diffracted light for each of the two alignment systems. Here, the conjugate plane (plane IP in FIG. 1) of the wafer W exists between the mirror 13X and the objective lens 7X in FIG.

In FIG. 2, the alignment system and the interferometer system show only one axial direction for alignment in the X direction. However, the same alignment system and interference system are actually provided for alignment in the Y direction. I have. on the other hand,
Two beams L from the alignment beam transmitting system ABO
B1 and LB2 are the same as those described with reference to FIG.
In the above, a parallel light flux intersects in a plane perpendicular to the paper surface in FIG. 2, and forms an irradiation area (second detection area) in a range sufficiently larger than the spot 71. In the present embodiment, the spot 71 is set so as to be located substantially at the center of the cross irradiation area of the two beams LB1 and LB2. However, as long as the beam LB0 passes through the center of the pupil EP, the position of the spot 71 becomes the cross illumination. It can be anywhere in the area. However, the positional relationship must be measured in advance using another reference mark (such as a fiducial mark on a wafer stage). FIG. 5 shows an alignment system in the apparatus of FIG.
FIG. 3 is a layout view as viewed on a Y plane. A laser light source 1 such as He-Ne, Ar ion, or He-Cd oscillates a laser beam having a wavelength different from that of the exposure light. This beam is rotated by approximately 45 ° in the polarization direction by a half-rotatable half-wave plate 2 ′, and is split into two in the polarization direction by the polarization beam splitter 2. One polarized beam (for example, P-polarized light) transmitted through the beam splitter 2 is applied to a shutter S1.
Passes through the mirror 20, a beam shaping optical system 21, and a cylindrical lens 22 to enter a first mark detection system beam transmission system. The beam LB0 from the cylindrical lens 22 is amplitude-divided by the beam splitter 3, and the beam LB0 transmitted through the beam splitter 3 is divided into a lens 4X, a mirror 5X, and an X-direction alignment objective system.
Beam splitter 6X, objective lens 7X, and mirror 1
It is incident on 3X. The beam LB0 reflected by the beam splitter 3 is incident on a lens 4Y, a mirror 5Y, a beam splitter 6Y, an objective lens 7Y, and a mirror 13Y constituting an alignment objective system in the Y direction. In addition, beam LB
For 0, only the chief ray of the beam is shown. Here, the beam splitters 6X and 6Y correspond to BS1 in FIG. 1, and split the diffracted light from the sub mark M1 and the main mark M2 on the wafer W to the photoelectric detectors 8X and 8Y. Now, the polarized beam (for example, S-polarized light) reflected by the polarized beam splitter 2 passes through a shutter S2, and is composed of two frequency modulators MD1, MD2 such as an AOM (acousto-optic modulator) and a beam splitter BS3. The light enters the beam transmitting system 21 'of the two-mark detection system. The beam is split into two in a beam transmitting system 21 'that has passed through a shutter S2, and modulators MD1 and MD2 are arranged in each beam path. The two frequency-modulated beams are eccentrically combined by a beam splitter BS2. The modulators MD1 and MD2 are driven by high-frequency drive signals SF1 and SF2 (several tens of MHz) having mutually different frequencies, and the frequency difference between the signals SF1 and SF2 becomes the beat frequency (several KHz to several tens KHz). Beam splitter B
The two beams LB1 and LB2 combined in S3 are reflected by the mirror 46, and enter the beam splitter 3 through the lens 47 and the polarizing plate 2 ". The incident surface of the two beams LB1 and LB2 in the beam splitter 3 Is the beam L
The two beams LB1 and LB2, which are orthogonal to the plane of incidence of B0, are matched with the polarization direction of the beam LB0 by the polarizing plate 2 ", are amplitude-divided by the beam splitter 3, and are subjected to the X-direction alignment objective system and the Y-direction alignment objective. The two beams L that pass through the beam splitter 3 and travel to the Y-direction alignment objective system in FIG.
Illustrations of B1 and LB2 are omitted. Beam LB1
, LB2 indicate only the principal ray, and from the lens 47 to the lens 4X (or 4Y), they intersect as a parallel light flux, and between the lens 47 and the light transmission system 21 ', the principal rays become parallel to each other. I have. Therefore, the lens 47 and the lens 4X
(Or 4Y), there is a plane (intersecting position) conjugate to the wafer W (or plane IP), and an aperture for restricting the illumination area may be appropriately provided at this position. Further, the shutters S1 and S2 selectively block the beam so as to enable one of the mark detection systems.
They are not released at the same time. By the way, in this type of heterodyne method, a reference signal serving as a reference is required. In this embodiment, the two beams LB1 and LB2 branched from the beam splitter BS3 in FIG. The light is incident on the reference grid plate R for reference from two directions at a predetermined intersection angle. Interference fringes flowing at the beat frequency are formed on the reference grating plate RG, and the photoelectric element PDR receives the interference beat light interfered by the diffracted lights of the same order and outputs a reference signal (AC signal of the beat frequency) 9R. The reference signal 9R is input to the electric system unit 90, and is used at the time of mark position detection by the second mark detection system. Each output signal from the photoelectric detectors 8X, 8Y is input to the electric system unit 90 via preamplifiers 9A, 9B, 9C and the like (the preamplifier for the signal from 8X is omitted). The electric system unit 90 operates in conjunction with the switching control of the shutters S1 and S2 to operate a first mark detection system (a stage scan alignment system using the spot 71) and a second mark detection system (a heterodyne alignment system using interference fringes). ) Is used to detect the position of the sub mark M1 or the main mark M2. The mark position information detected in this way is transferred to the main controller 91, and is transmitted to the drive system (motor 1) of the stage 16 via the stage controller 92.
Used for control of 7). In the electric system unit 90, the phase difference between the output from the preamplifier 9A for amplifying the signal from the light receiving element PD0 (see FIG. 1) at the center of the photoelectric detectors 8X and 8Y and the reference signal 9R is ± 180. A digital phase meter (or phase difference calculation software by Fourier transform) or the like for measuring within the range of ° is provided, and a signal waveform from the preamplifier 9B for amplifying a signal of the light receiving elements PDa and PDb, or a light receiving element PDc,
A / D conversion for converting a signal waveform from the preamplifier 9C for amplifying the signal of the PDd into a digital value in response to a count pulse from the interferometer counter (for example, a pulse generated every time the stage 70 moves 0.02 μm). And a memory for storing its waveform.

FIG. 6 shows the arrangement of the sub mark M1 and the main mark M2 on the wafer W which are aligned by the above-mentioned apparatus. The main mark M2 is formed by arranging a plurality of lines and spaces extending in the same direction as the longitudinal direction of the sub mark M1 at the pitch P in the same direction as the direction of detecting the position of the sub mark M1. It is assumed that the mark center in the detection direction (pitch direction) of the main mark M2 is separated from the center of the sub mark M1 by d. Such a set of the main mark M2 and the sub mark M1 is formed in advance for each shot area on the wafer W. In FIG. 6, the spot 71 by the beam LB0 is shown.
And an illumination area (second detection area) DA by the two beams LB1 and LB2 are shown to be located on the right side of the main mark and the sub mark. Therefore, the wafer W is moved rightward from the positional relationship as shown in FIG. At this time, the shutter S
1 is opened, the shutter S2 is closed, and only the spot 71 is irradiated onto the wafer W. Thus, the sub-mark M1 traverses the spot 71, and a predetermined photoelectric signal waveform (light receiving element P
As soon as the signals from Da and PDb) are stored in the memory in the electric system unit 90, the open / close state of the shutters S1 and S2 is switched to form the illumination area DA on the wafer W. During this time, the wafer W continues to move rightward, and from the position where the centers of the spot 71 and the sub mark M1 coincide, d
Stop at the position just moved. FIG. 7 shows a state in which the wafer W is stopped, and the illumination area DA and the main mark M2 are aligned within approximately ± P / 4. The illumination area DA is
Here, it is a sharp rectangle, but this is the beam L
This is because a rectangular aperture is provided at a position conjugate with the wafer W in the light transmission paths of B1 and LB2. Further, here, the illumination area DA is determined to have a size including the entire size of the main mark M2, but the main mark M2 may be larger than the illumination area DA. Further illumination area D
The number of diffraction gratings of the main mark M2 included in A is such that the S / N ratio of the interference beat light becomes sufficiently good (for example, 3).
Book).

In the illumination area DA, bright and dark interference fringes extending in the longitudinal direction of the diffraction grating extend in the grating pitch direction.
The interference fringes are formed alternately at intervals of / 2, and flow in one direction as indicated by arrows in FIG. The reason why the pitch of the interference fringes is set to 1/2 of the grating pitch P is to take out the interference beat light as the strongest interference between the ± 1st-order diffracted lights, and the two beams LB1 and LB2 are applied to the wafer W. This is achieved by setting the incident angle θ to sin θ = λ / P.

The interference beat light is received by the light receiving element PD0 at the center of the photoelectric detector, and generates an AC signal having a beat frequency as a measurement signal. The electric system unit 90 detects the phase difference (± 180 °) between the measurement signal and the reference signal 9R,
As a result, ± P of the main mark M2 with respect to a predetermined reference point
A position shift amount within / 4 is obtained. This predetermined reference point is
In the present embodiment, this corresponds to the reference diffraction grating RG, and when the phase difference is zero, it means that the reference grating RG and the main mark M2 exactly match. FIG. 8 shows the marks on the street line attached to one shot area CP on the wafer W as X, Y.
FIG. 9 is a diagram showing a state of movement of a stage 16 when measuring in a direction. In each shot area CP on the wafer W,
A main mark M2x for the X direction and a main mark M2y for the Y direction are placed on each of the orthogonal coordinate axes having the origin at the shot center CC.
Are formed. A sub-mark M1x is formed on both sides of the main mark M2x in the X direction at a distance of d, and a sub-mark M1y is formed on both sides of the main mark M2y in the Y direction at a distance of d.
The reason why the sub-marks M1x and M1y are provided on both sides of the main mark is that the sign of the moving direction of the stage 16 when the mark is detected (positive or negative in the pitch direction) can be arbitrarily selected. Of course, as shown in FIG.
It may be one. On the other hand, the projected image of the reticle R via the projection lens PL is usually a rectangular area RSA centered on the optical axis AX, and is naturally included in the circular field IF of the projection lens PL. Here, when the XY coordinate system is defined with the optical axis AX as the origin, the X axis and the Y axis coincide with the length measuring axis (laser beam center line) of the laser interferometer of the stage 16, and the two beams LB1 and LB2 Illumination area DA for X direction
x is located at the periphery of the field of view IF on the Y axis, and the illumination area DAy for the Y direction is located at the periphery of the field of view IF on the X axis.
First, assuming that the field of view IF and the shot area CP are positioned in a relationship as shown in FIG. 8, the center point of the main mark M2y passes through the position P1a to the position P1b (the center of the illumination area DAy).
The stage 16 is moved according to the laser interferometer so as to stop. At the position P1a, the positional relationship is as shown in FIG. 6, and when the position is moved to the position P1b in the Y direction from FIG.
It becomes such a positional relationship. At this time, the main mark M2x moves along the locus of the position P2a, P2b, and the shot center CC advances along the locus of the position P3a, P3b. When the amount of displacement (within ± P / 4) of the main mark M2y in the Y direction within the illumination area DAy is determined, the main mark M2x then passes from the position P2b to the position P2c at P2d (the center of the illumination area DAx). The stage 16 is moved so as to stop. At the position P2c, the main mark M2x and the sub mark M1x are in the positional relationship shown in FIG.
The positional relationship is as shown in FIG.

Then, the amount of displacement of the main mark M2x in the X direction at the position P2d is obtained. Thus, one shot area C
When the position of X in the X and Y directions is specified, the stage 16 is similarly moved to measure the position of the next shot area CP. If the shot area CP is to be exposed immediately, the stage 16 is moved so that, when the amount of displacement of the main mark M2x in the X direction within the illumination area DAx is determined, the amount of displacement falls within a predetermined allowable range. In the X direction (± P
/ 4) and the stage 16 is fed by a fixed amount in the Y direction.

Immediately before this operation, the shot center CC is at the position P3d through the position P3c, and when the main mark M2x is accurately aligned in the X direction, the shot center CC at the position P3d is located on the Y axis. Will be. Therefore, after the alignment in the X direction, if the stage 16 is sent by a fixed amount in the Y direction according to the laser interferometer, the shot center CC and the optical axis AX (the center of the projection area PSA) will be exactly coincident. Exposure may be performed.

The feed amount in the Y direction depends on the main mark M
The Y coordinate value of the stage 16 at the position P1b where 2y stops is represented by Y
a, the measured positional deviation within ± P / 4 is ΔY, and the Y coordinate value of the stage 16 at the position P2d where the main mark M2x stops (the same as the Y coordinate value at the position P1d) is Yb. − (Ya−ΔY). FIG.
9 is a graph showing the mark position detection operation in FIG. 8 in relation to the moving speed of the stage 16 and time. Time 0
If the stage 16 is stopped at a certain position, it accelerates in the direction of the target main mark M2 (or sub-mark M1), and reaches a certain fixed position before the start position x1 of the signal waveform of the sub-mark M1. The signal waveform of the sub mark M1 is stored in the memory in the electric system unit 90 until the speed is reduced to the fetch end position x2. The fetch start position x1 and the end position x2 are set on the basis of the predicted position xp of the sub mark predicted based on the result of the wafer global alignment (pre-alignment). Position x
The stage 16 that has passed through 2 continues to move to the predicted position x
It is controlled to stop at the expected stop position x5 ahead of p by d. After the electric unit 90 passes through the position x2,
The signal waveform is subjected to high-speed arithmetic processing, and the actual position of the sub-mark M1 and the measured position x3 of the mark are calculated during the calculation time T1. The time T1 is set to end before the stage 16 reaches the expected stop position x5. Main controller 91
Calculates the actual stop position x6 (new target value) obtained by correcting the expected stop position x5 by the difference Δx between the predicted position xp and the actual measured position x3, and corrects and controls the vibration of the stage 16. Thus, when the stage 16 stops at the position x6, the center of the main mark M2 is positioned within ± P / 4 with respect to a predetermined reference point. Then, using the illumination area DA, ± P /
An amount of displacement within four is measured. The shutter S1
, S2 is switched between positions x2 and x6 in FIG. The time from time 0 to the position x1 is, for example, the time when the main mark M2x moves from the position P2b to the position P2c in FIG. 8, and the time from the position x1 to the position x6 in FIG. Main mark M from position P2c to P2d
2x is the time to move. As described above, in this embodiment, since the marks of the main alignment and the sub-alignment can be detected by a series of stage movements, there is an advantage that the time loss in the wafer alignment sequence is extremely small. Also, the higher the processing speed for calculating the actual measurement position of the sub-mark, the shorter the distance d between the sub-mark M1 and the main mark M2 in the measurement direction can be, and the time can be further reduced. In FIG. 9, the stage speed is set lower than the maximum speed for detecting the sub mark. However, if the processing time on hardware such as digitization of the signal waveform and storage in the memory sufficiently follows, the sub speed is maintained at the maximum speed. Obviously, the waveform of the mark M1 can be captured.

Since the sub-mark M1 is used to specify the position of the main mark M2 within ± P / 4, the accuracy of the detection system of the sub-mark M1 using the spot 71 is commensurate with it. Anything should do. For example, if the pitch P is 2 .mu.m, the position detection accuracy of the sub-mark M1 is sufficient if the margin is about. ± .0.2 .mu.m. Therefore, when the sub-mark M1 is used for pre-alignment of the main mark M2, digital sampling of signal waveforms from the light receiving elements PDa and PDb of the photoelectric detectors 8X and 8Y is roughly performed every time the stage 16 moves 0.08 μm, for example. May be. In this case, the stage 16 is moved four times as compared with the sampling by the interferometer pulse every 0.02 μm.
Double the speed (actually limited by the maximum speed of the stage).

Further, assuming that the width of the sub mark M1 is substantially equal to that of the spot 71, when the width of the sub mark M1 is about 2 μm, for example, the frequency of the interferometer pulse every 0.02 μm is divided and the pulse Each time a pulse of 0.2 μm is generated, the signal level from the light receiving elements PDa and PDb is compared with a predetermined slice level, and a pulse train where the signal level is higher than the slice level is obtained. The position where the pulse at the center of the pulse train is obtained may be set as the position of the sub mark M1.

Next, an alignment operation according to a second embodiment of the present invention will be described. The second embodiment is a method for further shortening the alignment time, and it is assumed that the main mark M2 and the sub-mark M2 have the same structure and arrangement as the first embodiment. In the first embodiment, the result of the sub-alignment is calculated during the relative movement. In the second embodiment, the main alignment is performed by positioning at an appropriate position without waiting. In other words, the main alignment operation is performed while the sub-alignment processing is being performed, and the sub-alignment calculation processing is completed before the main alignment (the deviation detection operation within ± P / 4) is completed. By feeding back the result of the sub-alignment with respect to the result of the main alignment, a correct result can be obtained even if the positional relationship between the illumination area DA for the main alignment and the main mark is shifted by ± P / 4 or more. I do. By doing so, the same effect as increasing the sub-alignment processing speed in the first embodiment can be obtained.

The above will be described with reference to FIG.
The main controller 91 and the stage controller 92 determine the main marks M2,
The stage 16 is moved based on the position information of the sub mark M1, and the sub mark M is moved between the positions x1 and x2 as in FIG.
The stage 16 is stopped at the predicted position of the main mark M2 which is separated from the predicted position xP of the sub-mark M1 by the interval d, that is, the expected stop position x5.

Immediately after passing through the position x2, the electric system unit 90 starts the calculation for obtaining the position of the sub mark M1 (x3 in FIG. 9). On the other hand, the phase
The meter obtains the phase difference between the photoelectric signal of the interference beat light from the main mark M2 and the reference signal 9R, that is, the amount of positional deviation within ± P / 4 during the time T2 immediately after the stage 16 stops at the position x5. . By the end of the time T2, the position of the sub mark M1 is specified. Here the sub mark M1
Assuming that the actual measurement position is shifted by ΔH with respect to the predicted position x p, main controller 91 sets n to an integer (including zero).
As

[0028]

(Equation 1)

Then, the amount of positional deviation within ± P / 4, which is the main alignment result, is set as ΔF, and the final amount of positional deviation is calculated as n · (P / 2) + ΔF. As described above, in the present embodiment, the position calculation of the sub-mark M1 has been completed by the end of the time T2.
Even if the time T2 ends before the end of the operation, it is possible to avoid this. However, the timing for starting the stage 16 for the next mark position measurement is based on the main mark M
2 must be after the time T2 when the phase difference detection for specifying the position is completed.

Next, a mark detecting operation according to a third embodiment of the present invention will be described. In the first and second embodiments, the order of the sub-> main alignment is used, but the order of the main-> sub-alignment may be reversed. That is, after performing the alignment using the main mark M2 in the state of FIG. 7 first, the sub-alignment is performed by moving the sub-mark M1 to the state of FIG. Then, a deviation of ± P / 4 or more during the main alignment is corrected based on the result of the sub-alignment. With this method, the sub-alignment processing time may be slightly longer. This is because when performing alignment in the order of X → Y, the time required to move to the Y alignment operation after the X alignment, the time required to move to the alignment position in the next shot area after the Y alignment, or the time required to move to the exposure position are all shown in FIG. This is because it is longer than the movement time of the interval d in 6. FIG. 11 is a graph of the above. The stage 16 stops at the predicted position x5 of the main mark M2, that is, the expected stop position. Then, a displacement within ± P / 4 of the main mark M2 during the time T2 is determined. At this time, the position x5 has a position shift depending on the accuracy of the global alignment of the wafer W, but ± 2.
It is easy to keep the distance within μm, and if the deviation is as large as this, the illumination area DA and the main mark M2 are shifted at most by one pitch, so that the S / N ratio (signal amplitude) of the interference beat light changes substantially. do not do. Therefore, in consideration of this deviation (pre-alignment or global alignment accuracy), the entire size of the main mark M2 with respect to the illumination area DA may be increased.

When the position deviation of the main mark M2 within ± P / 4 is determined, the spot 71 is irradiated to the sub mark M2.
1 in the direction of the spot 71, and the positions x1 and x
2, the signal waveform of the sub-mark M1 is fetched during the acceleration period, the position of the sub-mark M1 is specified at time T1, and the calculation of n. (P / 2) +. DELTA.F is performed as in the second embodiment. The center position of the mark M2 is accurately determined.

In this embodiment, when detecting a plurality of marks one after another, the stop period of the stage 16 only needs to be the time T2, so that the processing time is shortest (highest throughput) among the previous embodiments. There are advantages. still,
In FIG. 11, the signal waveform is captured between the positions x1 and x2 during the acceleration of the stage 16. However, as in the first and second embodiments, the signal waveform may be captured while the stage 16 is moving at a constant speed. Is based on the distance d between the main mark M2 and the sub mark M1 and the motion characteristics (acceleration, maximum speed, etc.) of the stage 16. 12A and 12B are diagrams for explaining the speeding up of the detection operation of the main mark M2. FIG. 12A shows the measurement signal waveform of the interference beat light from the main mark M2, and FIG. 12B shows the waveform of the reference signal 9R. It is a waveform. The main mark M2 in the illumination area DA
From the time t1 immediately after the precise stop of time to the time t2 after a fixed time, the waveforms of the measurement signal and the reference signal are digitally sampled by the A / D converter under the same time axis clock pulse, Store in memory. Then, the two stored waveforms are processed by software using an arithmetic technique such as Fourier transform or the like to obtain the phase difference Δθ. In this way, since the stage 16 can be moved after the time t2, the throughput can be improved by starting the stage 16 before the deviation amount of the main mark M2 within ± P / 4 is calculated.

If the frequency (beat frequency) of the two signal waveforms is set to about 20 KHz and the number of cycles on the sampled waveform is set to about 20 to 40, the time from time t1 to time t1
The time up to 2 is extremely short, 1-2 mSec. Of course, the subsequent software operation requires about 5 to 10 mSec even if a high-speed operation processor is used. However, since the time can be set in parallel with the movement of the stage 16, the apparent waiting time of 5 to 10 mSec is not shown. Does not appear in

Next, a description will be given of a mark structure and photoelectric detection according to a fourth embodiment of the present invention with reference to FIGS. FIG. 13A shows that the main mark M2 and the sub-mark M1 are brought close to each other to the same degree as the grid pitch P of the main mark M2, and the sub-mark M1 itself is set to one of the main marks M2.
It is formed as a book lattice. In this case, the width of the sub mark M1 in the measurement direction is set to be the same as the grid width of the main mark M2. In this case, the sub mark M1 can also contribute to the generation of interference beat light. FIG. 13B shows the sub mark M
1 is arranged as the central grid of the main mark M2,
The center of the sub mark M1 coincides with the center of the main mark M2. Also in this case, the sub mark M1 is set to have a pitch relationship and a line width that contribute to the generation of interference beat light.

FIG. 14 shows a modification in which the sub-mark M1 is formed as a simple line instead of a diffraction grating. FIG.
(A) shows the same mark arrangement as that shown in FIG.
FIG. 14 (B) is FIG. 13 (A) and FIG. 14 (C) is FIG.
The mark arrangement is the same as that of FIG. Here, the line width of the sub-mark M1 is made equal to the width of one grid of the main mark M2, and functions as one grid of the main mark M2.

FIGS. 13A and 13B and FIG.
According to the structures (B) and (C), there is an advantage that the area occupied on the wafer as an alignment mark can be reduced. Further, the mark structures shown in FIGS. 13B and 14C also have an advantage that the center of the main mark M2 can be easily found at the time of visual observation. By the way, in the structure of FIG.
Since the spot 71 is located at the center of the illumination area DA, after the sub mark M1 passes below the spot 71, the stage 16 is moved in the opposite direction to stop the stage 16 within ± P / 4 based on the result of the sub alignment. There is a need.

In the case of the mark structure shown in FIG. 14, when the sub mark M1 and the spot 71 move relatively, the sub mark M1
Scattered light is generated from the straight edges on both sides of the light-receiving element, and this is detected by the light-receiving elements PDc and PDd shown in FIG.
As the stage 16 moves, the digital
What is necessary is just to sample. By the way, FIG.
The state of photoelectric detection in X and 8Y is shown, and FIG. 15A shows the arrangement of the light receiving elements PDa, PDb, PDc and PDd. The light receiving element PD0 is located at the center (on the optical axis) on the pupil conjugate plane, and is positioned in the lattice pitch direction of the main mark M2.
Two light receiving elements PDc and PDd are arranged so as to sandwich 0. The light receiving elements PDa and PDb are arranged so as to sandwich the three light receiving elements PDc, PDd and PD0 in a direction orthogonal to the pitch direction.

FIG. 15B shows a case where when only the beam LB0 is irradiated on the wafer for detecting the sub mark M1, the sub mark M1 and other pattern edges do not exist in the irradiation area of the spot 71. 2 shows the distribution of reflected light on the light receiving element of FIG. In this case, since the beam LB0 is applied telecentrically to the reflection surface of the wafer W through the center of the pupil of the system, only the zero-order light 100 returns to the center of the photoelectric detector.

However, the spot 71 and FIG. 6 or FIG.
When the sub mark M1 in FIG. 3 overlaps, the primary light ± DL1 and the secondary light ± DL are placed on both sides of the zero-order light 100 as shown in FIG.
2, higher-order diffracted lights such as the third-order light ± DL3 return. These higher-order lights are received by the light receiving elements PDa and PDb. The sub mark M1 or the main mark M in FIG.
When the grating in 2 overlaps with the spot 71, the edge scattered lights 101a and 101b that spread in the longitudinal direction of the zero-order light 100 return as shown in FIG. This scattered light 101
a and 101b are received by the light receiving elements PDc and PDd.

On the other hand, if the main mark M2 does not exist in the illumination area DA formed by the two beams LB1 and LB2 and is merely a reflection surface, the regular reflection light (0th order) of the two beams LB1 and LB2 The light BL0 returns to the light receiving elements PDc and PDd as shown in FIG. When the main mark M2 is located in the illumination area DA, the interference beat light BTL returns to the position of the light receiving element PD0 as shown in FIG. These light receiving elements PD0, PDa, PDb,
PDc and PDd are photodiodes and PIN photodiodes independently formed on the same semiconductor substrate via an insulating layer.

The specularly reflected light BL of the beams LB1 and LB2
Although 0 is assumed to return to the positions of the light receiving elements PDc and PDd, this is not always necessary. Next, for some wafer alignment sequences of the present invention,
As will be described with reference to FIG. 16, the mark structure and the mark detection (stage scan) operation for this sequence are as follows.
Any of the embodiments described above may be used.

In this type of stepper, a plurality of shot areas CPn on the wafer W are sequentially stepped and the exposure is repeated. Before the exposure operation, the alignment of the wafer is always executed. This alignment sequence is roughly divided into die-by-die (or field-by-die) for detecting a mark position for each shot.
There are two types: a (by-field) alignment method and a global alignment method for detecting mark positions of several typical shots on a wafer.

In any of the die-by-die alignment method and the global alignment method, the mark position detection using the main mark M2 can be applied. In this case, the sub-alignment (± P / 4
(Positioning within) operation can also be considered. One is to collectively measure the positions of a plurality of marks on the wafer using only the sub-alignment means (first detection means), and the other is to perform sub-alignment for each shot to be main-aligned. The way to do it. The former method is
In other words, it is a global alignment method,
The latter is a die-by-die alignment method.
Therefore, two of the main alignment means and the sub-alignment means
In a stepper having one, four types of sequences can be considered as combinations. FIG. 16A shows the main alignment,
The sub-alignment also represents a global sequence, and the sub-alignment (mark position detection) using the sub-mark M1 is performed by four shots CP1, CP2,
The main alignment using the main mark M2 is performed by CP3 and CP4, and the eight shots CP1 to CP8 indicated by the circle mark on the wafer
Perform in. Here, shots CP1, CP2, CP3, C
P4 is located around the wafer on the axis of the orthogonal coordinate system, and at the same time, the eight shots CP1 to CP8 are located at substantially equal distances from the center of the wafer. The movement of the stage to each of these shots is represented by solid and dashed arrows. First, a shot CP1 close to the orientation flat, a leftmost shot CP2, an uppermost shot CP3, and a rightmost shot CP4 in the order of X, Each sub mark M1x, M in the Y direction
Measure the position of 1y. Thereafter, based on the measurement result, a statistical calculation process is performed so that the shot arrangement coordinates on the wafer and the stage movement coordinate system are associated with any point on the wafer with an accuracy of ± P / 4 or less. Then, or in parallel with the arithmetic processing, the last shots CP4 to CP5, CP3, CP6,.
The main alignment using the main mark M2 is performed counterclockwise in the order of CP1 and CP6.

The displacement amounts (± P / 4) in the X and Y directions of the eight shots CP1 to CP8 due to this main alignment.
Based on the position measurement results by the sub-alignment and the sub-alignment, the relative positional relationship between each center point of all shot areas on the wafer and the optical axis of the projection lens is precisely specified. The stage is stepped on the basis of. The sequence in FIG. 16A is the fastest. Further, since the positions of the eight shots CP1 to CP8 can be precisely determined at ± P / 4 or less, the enhanced global alignment (E.G.) disclosed in Japanese Patent Application Laid-Open No. 61-44429 is used by using these values. If the least squares approximation by the method A) is performed to determine the regularity of the shot array, extremely high alignment accuracy can be obtained.

FIG. 16B shows four shots CP1, CP2, CP3, and CP on the wafer similarly to FIG.
For 4, the sub mark M 1 is set by the sub alignment means.
Only the position is measured, and the main alignment is performed in the step and repeat exposure shot order based on the measurement result. In this case, the main mark M of each shot area
The stepping accuracy for 2 must be within ± P / 4. In FIG. 16B, the sub-alignment is used in a global manner, and the main alignment is die-by-by.
This means that the die method was used.

FIG. 16C shows a sub-alignment and a main alignment for each of the eight representative shots CP1 to CP8 on the wafer before the exposure operation. FIG. 16D shows a method of detecting both the sub mark M1 and the main mark M2 of each shot at every stepping in the step-and-repeat exposure operation, which is a complete die-by-die method.

As described above, FIGS. 16 (A), (B), (C),
The number of measurement shots shown in (D) may be changed according to each wafer process and the respective capabilities of the main alignment means and the sub-alignment means for the same. Further, since the sub-alignment means uses the spot 71 of the beam LB0, the range of the beam waist (length in the optical axis direction) is narrow. Therefore, at the time of sub-alignment, the focusing (Z in the stage 16) is performed at the time of scanning each sub-mark M1. Stage up and down movement)
However, if the main alignment means is within the intersecting region of the two beams LB1 and LB2, a clear interference fringe can be formed on any surface, so that the focal range can be relatively widened and the focus adjustment can be omitted. Can be.

FIG. 16A, FIG. 16B, FIG.
Any of the sequence software (D) is stored in the main controller 91 in advance, and any one of them can be selected by an operator. When lot management is performed on a large number of wafers, the sequence may be changed between the first few wafers and the subsequent wafers in the same lot.

FIG. 17 shows a stepper incorporating a positioning device according to another embodiment of the present invention, and the same members as those in FIG. 2 are denoted by the same reference numerals. The second mark detection means (main alignment) includes an objective lens OBJ1, a mirror RM, a second objective lens OBJ2, a beam splitter BS1, a light transmitting optical system 11 disposed above the reticle R.
A similar effect can be obtained by a TTR (through-the-reticle) system including the light receiving optical system 112 and the light receiving optical system 112. Here, the objective lens OBJ1 and the mirror RM are provided so as to be horizontally movable as indicated by arrows in the figure, and the observation position can be changed according to the position of the window on the reticle R.

In the case of the TTR system, in order to simultaneously observe the window of the reticle R and the mark WM on the wafer W in a state where the reticle R and the wafer W are in a conjugate relationship with each other via the projection lens PL, light transmission is required. It is necessary that the wavelength of the alignment beam from the optical system 110 be the same as or approximate to the wavelength of the exposure light. In the case of using an alignment beam having a wavelength different from that of the exposure light even in the TTR method, the conjugate relationship between the reticle R and the wafer W is deviated by the amount of chromatic aberration of the projection lens PL. A double focusing member may be provided in OBJ1 so that the alignment beam is double focused.

In the same TTR system, a dichroic mirror DM for separating an alignment beam of another wavelength and exposure light is inclined at 45 ° above the reticle R, and the objective lens OBJ1 is moved from outside the exposure optical path to the inside of the optical path. A main alignment system can be similarly configured with the structure to be inclined. in this case,
The objective lens OBJ1 is configured to be able to move in parallel in the vertical direction in the figure, and can continue to detect the mark (window) of the reticle R and the mark on the wafer W during the exposure operation. on the other hand,
As the first mark detection means (sub-alignment system),
An off-axis type alignment system close to the projection lens PL can be used. An off-axis alignment system is configured by the mirror PM, the objective lens OBJ4, the beam splitter BS1, the light transmitting system 113, and the light receiving system 114 which are obliquely provided at the lower end of the projection lens PL. The light transmission system 113 is provided with a system for forming a light source image having a broadband wavelength distribution on the pupil (aperture stop surface) of the objective lens OBJ4. CCD or the like). The image signal is processed by a predetermined image analysis circuit, software, or the like, and the position of the mark center is obtained. Such off-
In the Axis system, since the system does not pass through the projection lens PL, there is an advantage that the wavelength band of the illumination light for alignment can be widely used outside the exposure region and is not easily affected by the resist layer. Further, the light transmitting system 113 and the objective lens O
Even in the method of projecting a slit-shaped beam onto the wafer W via the BJ4, the beam can be made to have multiple wavelengths (a plurality of semiconductor lasers, light-emitting diodes, etc. in the red region having different wavelengths are simultaneously turned on), and the resist can be used. Less affected by
Spot scanning is possible. In both the TTR method and the off-axis method, the objective lens can be shared by the sub-alignment system and the main alignment system. As described above, in each embodiment of the present invention, the main alignment system is described by the heterodyne interference alignment method, but the same effect can be obtained by the homodyne interference alignment method. For the homodyne method, the 2 shown in FIG.
Drive signals S of the two frequency shifters MD1 and MD2
It is only necessary that F1 and SF2 have exactly the same frequency. In this case, a stationary interference fringe is formed on the wafer W, and the main mark M2 on the wafer is positioned within the range of ± P / 4 with respect to this interference fringe. Further, the light receiving element PD
0, the DC level of the photoelectric signal changes sinusoidally every time the main mark M2 moves by P / 2.
Therefore, the servo control of the wafer stage 16 is performed so that this level is stabilized at a certain value (for example, the center of the amplitude) within ± P / 4.

Even in such a homodyne system, if the coordinate position of the stage 16 when the stationary interference fringe and the main mark M2 match is stored, the sequence exactly the same as in each of the above-described embodiments is stored. Can be realized. In the case of the homodyne method, the main mark M2
Is moved, a sine-wave level change of the photoelectric signal is obtained.
It is also possible to digitally sample the photoelectric signal from the light receiving element PD0 in response to a measurement pulse (up / down pulse every ± 0.02 μm) from the laser interferometer and store the waveform once. In this case, the center of the amplitude (or the bottom or peak) is determined by digital calculation within a specific period of the stored waveform, and the stage 16 corresponding to the center of the amplitude is determined.
Is obtained. According to this method, it is possible to continuously shift from the sub alignment in which the sub mark M1 is detected by scanning the stage to the main alignment (detection of the signal waveform of the main mark M1) without stopping the stage. Of course, the reverse sequence is also possible as described in FIG. By the way, as can be said for all the embodiments, the signal from the light receiving element PD0 taken into the electric system unit 90 of the alignment signal processing system needs to be applied with an appropriate gain so as to have a strength that does not cause processing inconvenience. In the previous embodiment, since the sub-alignment operation relatively moves the spot 71 of the beam LB0 and the sub-mark M1, if the signal strength is inappropriate, the same operation must be repeated again. This is performed while the main mark M2 is stationary within the illumination area DA of the beams LB1 and LB2. On this occasion,
In order to completely stop the stage 16, the illumination area D
It takes some time after A and the main mark M2 overlap. Within this time, the measurement signal as shown in FIG. 12 is started, and by applying a gain based on the measurement signal, the time loss required for gain setting can be reduced to zero. This will be briefly described with reference to FIG. FIG. 18 shows an example of the servo characteristics exaggerating the state when the stage 16 is stopped. The horizontal axis represents time t, and the vertical axis represents the deviation from the target stop position. When the stage 16 approaches the target position and almost the entire main mark M2 enters the illumination area DA, the measurement signal (AC) from the light receiving element PD0 has a substantially stable amplitude. Therefore, the stage controller 91 issues a command to load the signal waveform of the light receiving element PD0 into the memory in the electric system unit 90 from the time Ts1 when the count value of the laser interferometer is within ± several tens counts from the target value. Output. Then, bottom and peak values (amplitude) of several waves of the signal waveform are checked, and the gain of the preamplifier 9A in FIG. 5 is switched so that the value approaches a predetermined amplitude. This gain switching is performed until time Ts2, and waveform sampling for phase difference detection is subsequently performed. Here, the time from time Ts1 to Ts2 is a time for the stage controller 91 to confirm the complete stop of the stage, and it depends on the stage.
About several tens mSec is required.

It is clear that this method can be applied to the homodyne method as it is. By the way, confirmation of the stabilization of the stage 16 can also be performed by checking the phase change between the signal from the light receiving element PD0 and the reference signal from the light receiving element PDR. In the case of the heterodyne method, when the main mark M2 moves with respect to the illumination area DA, the frequency of the signal from the light receiving element PD0 slightly changes from the original beat frequency due to the Doppler effect, and is within a certain period of time with the reference signal. In the number of signal periods. Therefore, when the difference between the frequencies falls within a certain range, it can be considered that the stage is settled.

As described above, according to the present embodiment, the detectable range is wide, but light information (scattering from inside a slit-like spot, diffracted light, The first that can detect only the light emitted by one CCD pixel)
The mark detecting means is combined with the second mark detecting means which simultaneously captures optical information from the entire mark on the substrate although the effective detectable range is narrow, and these two mark detecting means are used in a series of mark detecting operations. Since they are used sequentially or almost simultaneously, there is an effect that a time loss in the alignment sequence is extremely small and a decrease in throughput is suppressed.

[0055]

As described above, according to the present invention, a lattice mark is formed.
Of the center of the grid mark
Line (M1) intersects the arrangement direction of the grid marks
The other lines (M
Since the mark is formed longer than 2), the mark can be easily recognized visually.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of an embodiment of the present invention.

FIG. 2 is a diagram illustrating a configuration of an apparatus according to the first embodiment.

FIG. 3 is a diagram illustrating mark detection by a first mark detection unit and a structure of a sub mark.

FIG. 4 is a diagram illustrating generation of optical information from a sub mark.

FIG. 5 is a diagram showing an overall configuration of an alignment system according to the first embodiment.

FIG. 6 is a view for explaining the structure of a main mark and the state of sub-alignment and main alignment.

FIG. 7 is a view for explaining the structure of a main mark and the state of sub-alignment and main alignment.

FIG. 8 is a view for explaining the movement of the stage at the time of alignment for one shot.

FIG. 9 is a graph of speed characteristics showing the movement of a stage in the first embodiment.

FIG. 10 is a graph of speed characteristics showing the movement of the stage according to the second embodiment.

FIG. 11 is a graph of speed characteristics showing the movement of a stage according to the third embodiment.

FIG. 12 is a waveform chart showing a state of phase difference measurement in heterodyne interferometry.

FIG. 13 is a diagram showing a modification of the mark structure according to the fourth embodiment.

FIG. 14 is a diagram showing a modification of the mark structure according to the fourth embodiment.

FIG. 15 is a diagram showing a state of received light on a photoelectric detector when each mark of FIGS. 13 and 14 is detected.

16A, 16B, 16C, and 16D are views for explaining a typical wafer alignment sequence combining sub-alignment and main alignment.

FIG. 17 is a diagram showing a configuration of an apparatus according to a fifth embodiment.

FIG. 18 is a graph schematically showing characteristics when the stage is stopped.

[Description of Signs of Main Parts]

R: reticle, W: wafer, PL: projection lens, M1: sub mark, M2: main mark, LB
0, LB1, LB2: beam, DA: illumination area, BTL: interference beat light, 1: laser light source,
OBJ, OBJ1, OBJ4, 7X, 7Y: objective lens, PED, 8X, 8Y: photoelectric detector, PD0,
PDa, PDb, PDc, PDd: light receiving element, 16
···stage

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI H01L 21/30 525B (56) References JP-A-60-66820 (JP, A) JP-A-60-66819 (JP, A) JP-A-60-66818 (JP, A) JP-A-63-3416 (JP, A) JP-A-59-51529 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) H01L 21/027 G03F 7/20 G03F 9/00

Claims (5)

(57) [Claims] 1. A method of manufacturing a semiconductor device, comprising: forming a grid mark formed by arranging a plurality of lines and spaces in a predetermined direction on the semiconductor substrate in order to detect a position of the semiconductor substrate. The grid mark can be automatically detected by an alignment detection system , and among the plurality of lines, the grid
The line forming the center of the mark intersects the arrangement direction
Other lines that make up the grid mark in the direction of
The central part is formed by being formed longer than
A method for manufacturing a semiconductor device, wherein a line to be processed can be visually identified. 2. An exposure method for detecting a lattice mark formed by arranging a plurality of lines and spaces in a predetermined direction on a substrate and exposing a pattern of a mask on the substrate, wherein the lattice mark is aligned. The detection system can automatically detect , and among the plurality of lines, the grid
The line forming the center of the mark intersects the arrangement direction
Other lines that make up the grid mark in the direction of
The central part is formed by being formed longer than
An exposure method, wherein a line to be detected can be visually identified, and the grid mark position is visually detected using a line constituting the central portion . 3. The exposure method according to claim 2, wherein the center of said lattice mark can be detected by visually observing a line constituting said central portion . 4. A position of the lattice mark based on a change in intensity of the alignment light from the lattice mark obtained by irradiating the lattice mark with alignment light and moving the substrate with respect to the alignment light. 4. The exposure method according to claim 2, wherein information about the exposure is automatically obtained. 5. The method according to claim 2, wherein the grid mark is irradiated with alignment light, and information on the position of the grid mark is obtained based on positional information of the alignment light from the grid mark. 3. The exposure method according to 3.