JP3085292B2 - Scanning exposure equipment - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a scanning exposure apparatus for simultaneously scanning a mask and a photosensitive substrate with respect to a projection optical system when a pattern on a mask is projected and exposed on a substrate.

[0002]

2. Description of the Related Art As a conventional scanning exposure apparatus, a mask stage for holding a mask and a substrate stage for holding a photosensitive substrate (wafer) are connected to a common moving column by providing an equal-magnification reflection projection optical system. A method of performing scanning exposure at the same speed is known. This 1 × reflection projection optical system has good chromatic aberration over a wide exposure wavelength range because no refracting element (lens) is used, and has two or more emission line spectra (for example, g-line) from a light source (mercury lamp). And h-line etc.) at the same time to increase the exposure intensity and enable high-speed scanning exposure.
However, in a reflection projection system, an S (sagittal) image plane and M
(Meridional) Since the point at which both astigmatisms on the image plane are both zero is limited to the vicinity of the image height at a fixed distance from the optical axis of the reflection projection system, the shape of the exposure light illuminating the mask has a width of It is a part of a narrow orbicular zone, a so-called arc slit.

In such a 1: 1 scanning exposure apparatus (mirror projection aligner), if the pattern image of the mask projected on the wafer is not mirrored, the mask and the wafer are placed on an integral moving column. After being held in the aligned state, the exposure is completed by the moving column performing one-dimensional scanning in the width direction of the arc-shaped slit illumination light. Naturally, in a 1: 1 projection system in which the mask pattern image projected on the wafer has a mirror image relationship, it is necessary to move the mask stage and the wafer stage in opposite directions at the same speed.

Further, as a conventional scanning exposure method, it is also possible to relatively scan both the mask stage and the photosensitive substrate stage at a speed ratio corresponding to the magnification in a state where the projection magnification is enlarged or reduced by incorporating a refractive element. Are known. In this case, as the projection optical system, a combination of a reflection element and a refraction element, or a projection optical system composed of only a refraction element is used. There is one disclosed in JP-A-63-163319. A scanning exposure apparatus using this projection optical system has been announced as a step & scan type aligner (trade name: Micrascan system) by Perkin-Elmer.

Further, step-and-scan exposure is performed using a reduction projection optical system capable of full-field projection.
The two methods are also disclosed in JP-A-2-229423. Of the above-described scanning type exposure apparatuses, in an apparatus having a projection magnification other than the same magnification, it is necessary to precisely scan and move the mask stage and the wafer stage at a speed ratio corresponding to the magnification, and the scanning stage is generated during scanning. It is necessary to keep the deviation from the alignment state between the mask pattern and the pattern on the wafer within an allowable range. The allowable misalignment is roughly defined from the minimum pattern line width on the wafer. For example, if a pattern having a line width of about 0.8 μm is to be formed on the wafer, it is 1/5 to 1/10 or less. Only the amount of deviation is allowed.

Therefore, it is desirable that the relative displacement between the mask and the wafer can be constantly monitored during the scanning exposure. One such conventional example is disclosed in
As disclosed in Japanese Unexamined Patent Publication No. 1023, a plurality of U-shaped reticle marks formed on a mask (reticle);
A plurality of U-shaped wafer targets formed on the wafer are detected one after another immediately before or during scanning exposure, and the relative positional relationship between the reticle and the wafer (including magnification errors and rotation errors). Is known.

The method disclosed in Japanese Patent Application Laid-Open No. 63-41023 discloses a plurality of C-shaped mark groups RM1 and R formed around a reticle R, as shown in FIG.
M2, RM3 and a plurality of C-shaped mark groups WM1, WM2, W formed around the shot area SA on the wafer W
M3 and a slit-like illumination light beam AI arranged in a C shape
Under L irradiation, relative scanning is performed one after another. In FIG. 1, the projection optical system PL is shown in the same size for the sake of simplicity.
The reticle R and the wafer W scan and move in opposite directions as indicated by arrows. At this time, a change in the amount of specularly reflected light or scattered light reflected from the reticle marks RM1, RM2, RM3 and the wafer marks WM1, WM2, WM3 due to the irradiation of the illumination light beam AIL changes on the time axis as shown in FIG. Stipulated. FIG. 2A shows an example of a signal waveform obtained by photoelectrically detecting reflected light from a reticle mark.
(B) shows an example of a signal waveform obtained by photoelectrically detecting scattered light from a wafer mark. The alignment of the reticle R and the wafer W is such that the pulse P1 in the reticle signal waveform is temporally aligned with a pair of pulses P2 and P3 in the wafer signal waveform, and then the pulses P4 and P7 in the reticle signal waveform are respectively aligned with the wafer signal. A pair of pulses P5, P6,
This is achieved by adjusting the scanning speed and relative position of the reticle R and the wafer W so as to be temporally aligned with the pulses P8 and P9. However, if each mark is detected for the first time during actual scanning exposure, precise alignment has not been achieved at the start of scanning exposure. Therefore, pre-scanning is performed as shown in FIG. 1 before the actual exposure starts, and a position obtained by averaging the positions of the respective pulses P1, P4, and P7 in the reticle signal waveform and the respective pulses P2, P2, P3, P5, P6, P
8 and a position obtained by averaging the positions P9, and an alignment error ΔX in the scanning exposure direction can be obtained from the difference between the two average positions. When the following equation is calculated from the positions of the pulses P1 to P6, an alignment error ΔY in a direction orthogonal to the scanning direction is obtained.
I understand.

ΔY = ((P5 + P6)-(P2 + P
3))-(P4-P1)

[0009]

In the above prior art,
Since each signal waveform can be obtained only when the very thin slit-shaped illumination light beam AIL crosses each alignment mark, in order to obtain high alignment accuracy, the positions of a plurality of marks are detected and the average value is obtained. Such processing becomes indispensable. Therefore, after performing a scan for preliminary alignment, a main scan for exposure is performed. This cannot avoid the problem that the throughput is sacrificed. Further, in the above-described conventional mark shape, even if the accuracy in detecting one mark is sufficient, the size of the mark has only a very narrow width in the scanning direction. The pulse is intermittent in time, and during a period in which no pulse occurs in the signal waveform, the alignment error is not substantially detected.

For this reason, in a portion between each of a plurality of marks arranged in the scanning direction, a measurement value of a laser interferometer for measuring a position of a scanning stage for moving the reticle R or the wafer W is relied on. The present invention has been made in view of such conventional problems, and has as its object to minimize the need for alignment scanning before exposure.

According to the present invention, a mask stage and a wafer stage are precisely scanned and moved at a speed ratio corresponding to a magnification, and a deviation from a matching state between a mask pattern and a pattern on a wafer caused during the scanning is allowed. The purpose is to hold it inside.

[0012]

In order to achieve the above object, the present invention provides, for example, a method in which a mask (R) and a substrate (W) are relatively moved with respect to illumination light, respectively. In an apparatus that scans and exposes a pattern (PA) on a substrate (W), over substantially the entire exposure range on the substrate (W),
A mark detection system (XA, XB) for detecting a diffraction grating substrate mark (WMa, WMb) formed along a predetermined direction (X direction), and a mark detection system (X
(A, XB) so that the substrate mark (WMa, WMb) is continuously detected by the drive control means (moving the substrate (W) with the direction of the relative movement substantially coincident with the predetermined direction (X direction). 12).

In order to achieve the above object, the present invention provides, for example, a mask stage (6) for holding a mask (R).
And a substrate stage (9) for holding a photosensitive substrate (W). The original stage pattern (PA) of the mask (R) is exposed by synchronously moving the mask stage (6) and the substrate stage (9). In an apparatus for performing scanning exposure on a substrate (W), a mask (R) is irradiated with illumination light, and an irradiation area of the illumination light on the mask (R) is aligned with an optical axis (AX) of a projection optical system (PL). Illumination optical systems (1 to 5) whose longitudinal direction coincides with a direction (Y direction) orthogonal to the direction of synchronous movement (X direction), and a direction of relative movement (X direction). Mask stage (6)
, A plurality of alignment systems (23A, 23B,...) For detecting mask marks (RML1 to RML4, RMR1 to RMR4), which are switched and used in accordance with the positions of.
…).

According to the present invention, during the scanning and during the exposure,
An alignment error can be obtained almost continuously.

Further, the relative position between the mask and the substrate is adjusted based on the information on the alignment error, so that the deviation between the mask pattern and the pattern on the wafer caused during the scanning is within an allowable range. Can be held down.

Therefore, alignment accuracy and overlay accuracy can be improved even during scanning exposure.

[0017]

FIG. 3 shows a configuration of a step & scan exposure apparatus according to a first embodiment of the present invention, and FIG. 4 is a perspective view schematically showing a state at the time of scan exposure. FIG.
The projection optical system PL is, for example, a full-field type 1/5 reduction projection lens composed of only a conventional refractive element, and both the reticle R side and the wafer W side are telecentric.

The illumination light from the light source for exposure has a uniform illuminance distribution by a fly-eye lens or the like and irradiates a reticle blind 1 as an illumination field stop. The blind 1 is provided with a slit opening for illuminating the reticle R in a slit shape. The longitudinal direction of the slit opening coincides with the scanning direction of the reticle R and the wafer W, for example, the Y direction orthogonal to the X direction. The illumination light having passed through the slit opening of the blind 1 reaches the reticle R via the lens system 2, the mirror 3, the condenser lens 4, and the dichroic mirror (or beam splitter) 5.
Here, the blind 1 is arranged conjugate with the pattern surface of the reticle R (the surface facing the projection lens PL) with respect to the combined system of the lens system 2 and the condenser lens 4, and an image of the slit opening is formed on the reticle R.

The center of the slit opening is the projection lens P
L, and the optical axis AX of the illumination optical system (the lens system 2, the condenser lens 4, etc.). The reticle R is held by suction on a reticle stage 6 which can move at least largely in the X direction. Reticle stage 6
Is moved in the X direction on the column 7 by the motor 8 for scanning. Of course, for alignment of reticle R, Y
Although a fine movement mechanism in the direction and the θ direction is also necessary, illustration and description thereof are omitted here. The image of the pattern existing in the slit illumination area of the reticle R is formed and projected on the wafer W by the projection lens PL. The wafer W is placed on a wafer stage 9 that moves greatly in two dimensions (X and Y directions), and this stage 9 is driven by a motor 10. The laser interferometer 11 successively measures the change in the coordinate position of the wafer stage 9 and
Also outputs a speed signal relating to the moving speed in the direction and the Y direction. The drive control unit 12 controls the motor 10 with an optimal drive pattern based on the position information and the speed signal from the laser interferometer 11. In this embodiment, the scanning exposure is performed by the movement of the wafer stage 9 in the X direction, and the movement in the Y direction is used for stepping. However, it is needless to say that the reverse is also possible.

Although not shown in FIG. 3, it is assumed that the coordinate position, rotation (yawing) error and the like of reticle stage 6 are measured by a laser interferometer. Next, an example of the arrangement of the alignment marks formed on the reticle R and the wafer W will be described with reference to FIG. As shown in FIG. 4, since the reticle R and the wafer W are scanned in the opposite directions along the X direction, the reticle R and the wafer W fall within the street line area extending in the X direction around the pattern area PA on the reticle R. , Are provided with grid marks RMa and RMb in which a plurality of grid elements are arranged at a constant pitch in the X direction over the scanning range. The grid marks RMa and RMb are provided in the Y direction with the pattern area PA interposed therebetween.
It is assumed that the positional relationship in the direction is the same.

On the other hand, a plurality of pattern (shot) areas SA are formed on the wafer W, and a similar grid is formed around each shot area SA at the position of the street line area corresponding to the grid mark RMa of the reticle R. A mark WMa is formed, and a similar grid mark WMb is formed at a position in the street line area corresponding to the grid mark RMb.

The lattice mark RMa of the reticle R and the wafer W
Is detected through the alignment optical system having the optical axis AXa,
The relative displacement in the X direction between the grid mark RMb of the reticle R and the grid mark WMb of the wafer W is detected via an alignment optical system having an optical axis AXb. Each of these optical axes AXa and AXb intersects the optical axis AX at the center of the pupil plane EP of the projection lens PL.

Here, the alignment system and the control system will be described with reference to FIG. 3 again. In this embodiment, an interference fringe alignment method suitable for detecting a displacement in the pitch direction of each lattice mark of the reticle R and the wafer W is employed. An example of this interference fringe alignment method is disclosed in, for example, JP-A-63-283129 and JP-A-2-227602, and will be briefly described here.

A coherent linearly polarized laser from a laser light source 20, such as Ne--Ne, He--Cd, or Ar ions, enters a two-beam frequency shifter 21 and has two beams BL1, BL2 having a frequency difference Δf. Is made.
The upper limit of the frequency difference Δf is determined by the frequency response of a photoelectric detector that receives light from an alignment mark, and is practically 100 kHz or less, for example, 5 kHz for a semiconductor sensor.
About 0kHz is good. However, when a photomultiplier tube (photomultiplier) is used, the frequency can be set to a relatively high frequency.

The two beams BL1 and BL2 are distributed to a plurality of alignment optical systems via the light transmitting optical system 22. FIG. 3 shows an objective lens 23 and a tip mirror 24 that constitute one alignment optical system. The optical axes of the objective lens 23 are the optical axes AXa and AXb shown in FIG.
Corresponds to either one of the following. Two beams BL1, BL
Reference numeral 2 denotes a mirror 24 and a dichroic mirror 5 which are symmetrically decentered from the optical axis of the objective lens 23 and enter the objective lens 23.
Through the reticle R located at the focal position of the objective lens 23, and intersect as parallel light beams. By this intersection, the grating mark RMa or R
One-dimensional interference fringes are created on Mb. Then, the two beams transmitted through the transparent portion of the reticle R intersect on the lattice mark WMa or WMb on the wafer W via the projection lens PL to form one-dimensional interference fringes.

These interference fringes are Δf between two transmitted light beams.
Flows at a speed proportional to Δf in the pitch direction of the interference fringes. The incident directions of the two transmitted light beams are determined so that the pitch direction of each grating mark and the pitch direction of the interference fringes match, and the pitch of the grating marks and the pitch of the interference fringes have a predetermined relationship (for example, an integer). When the intersection angle of the two light transmission beams is determined so as to obtain the ratio (ratio), interference light having the same beat frequency as the frequency difference Δf is generated from each lattice mark in the vertical direction. This interference light constantly repeats light and dark at the beat frequency Δf.
Even if the relative position of the crossing area of the light transmission beams is slightly deviated in the X direction, the beat frequency Δf does not change.

Interference light from these grating marks is guided to the photoelectric detection unit 25 via the mirrors 5 and 23 and the objective lens 23, and sine-wave detection signals SR and SW are generated. The signal SR is obtained by photoelectrically detecting interference light from the grating mark RMa or RMb of the reticle R, and the signal SW is obtained by photoelectrically detecting interference light from the grating mark WMa or WMb of the wafer W. , Reticle R and wafer W
Are stationary, the frequency of both signals is Δf. However, when the lattice mark of reticle R and the lattice mark of wafer W are displaced in the pitch direction, 2
A phase difference Δφ occurs between the two signals SR and SW. The phase difference Δφ is detected by the phase difference measuring unit 27, and the amount of displacement corresponding to the detected phase difference is calculated. The detectable phase difference is usually in the range of ± 180 °, which is ± Pg / 2 (μm) or ± Pg / 4 (μm) as the amount of positional deviation when the pitch of the lattice mark is Pg (μm). Is equivalent to

The main controller 30 inputs the value of the displacement and outputs a successive correction value to the drive controller 12 of the wafer stage 9 or the drive controller 28 of the reticle stage 6. In the above-described conventional interference fringe alignment method, the main control system 30 merely drives the motor 8 or 10 until the phase difference reaches a predetermined value to finely move either the reticle stage 6 or the wafer stage 9. I just had to let it. However, another problem arises if the phase difference between the two signals SR and SW is determined even during scan exposure in which both the reticle R and the wafer W move at high speed as in the present embodiment. The reason is that the scanning exposure causes the grating mark to continue to move in the pitch direction at a speed v (mm / s) with respect to the intersecting region of the two light-transmitting beams, so that the interference light from the grating mark to be photoelectrically detected causes the Doppler effect Accordingly, the frequency of the detection signals SR and SW greatly varies from Δf. Frequency fs of signal SR and SW (kHz)
Is expressed by the following equation, where the moving speed of the lattice mark (pitch Pg) is v (mm / s).

Fs = Δf + 2v / Pg (Beat frequency Δf is 50 kHz) For example, if the speed v is -100 mm / s, the signals SR and SW
Is 25 kHz when Pg = 8 μm, and when the speed v is +100 mm / s, the frequency fs becomes 75 kHz. Therefore, the scanning speed of the reticle stage 6 and the wafer stage 9 involves some degree. For example,
The scanning speed v may be set low so that a value that does not cause a problem in the phase difference measurement can be secured as the frequency fs. In addition, the scanning exposure (the -X direction) in which the frequency fs becomes low may be avoided, and the scanning exposure may always be performed only in the + X direction.

In view of the above, the main control unit 30 of the present embodiment first sets the speed and position for driving the wafer stage 9 at a controlled constant speed in order to achieve simpler alignment during scanning. The control unit 300 includes a control unit 300, a speed and position control unit 302 for driving the reticle stage 6 at a controlled constant speed, and a tracking scan control unit 304.

In the case of ordinary positioning of a single reticle, that is, so-called reticle alignment, or alignment of a single wafer, so-called wafer global alignment (or EGA), the controller units 300 and 302 do not relate to each other, and perform the same functions as before. To achieve. At the time of scan exposure, the controller units 300 and 302 control the relative position and speed of the reticle stage 6 and the wafer stage 9 in cooperation with each other.

This cooperative control is similarly performed in the conventional apparatus shown in FIG. In the present embodiment, a tracking scanning control unit 304 is further provided so that normal cooperative control and tracking control can be switched. In this tracking control, the drive control unit 28 of the reticle stage 6 is servo-controlled so that the positional deviation amount successively output from the phase difference measurement unit 27 always becomes a constant value, and the wafer stage 9 is simply moved at a constant speed. It is controlled by. Of course, reticle stage 6 may be controlled at a constant speed, and wafer stage 9 may be controlled under tracking.

That is, in the present embodiment, attention is paid to the fact that the signals SR and SW are continuously output during the scanning exposure, in other words, the change in the relative displacement between the reticle R and the wafer W is continuously detected. Then, one of the reticle and the wafer is controlled to scan at a constant speed, and the other is controlled to follow the scanning movement. Note that in FIG.
Numeral 6 is for detecting the beat frequency Δf of the two beams BL1 and BL2, and this detection signal is input to the phase difference measuring section 27 as a sine wave reference signal SF having a frequency Δf.

The phase difference measuring section 27 determines the initial position of the reticle R from the phase difference between the reference signal SF and the detection signal SR, or calculates the initial position of the wafer W from the phase difference between the reference signal SF and the detection signal SW. It is also possible to determine the displacement. Further, a circuit for detecting a frequency change is incorporated in the phase difference measuring unit 27, and the frequency change of the detection signal SR or SW with respect to the reference signal SF is quantified, so that the reticle stage 6 or the wafer stage 9 can be measured. The speed change can be directly detected from the movement of the grid mark.

In this embodiment, as shown in FIG.
Since the set of alignment systems (optical axes AXa and AXb) detect the lattice marks RMa and RMb on both sides of the pattern area PA and the lattice marks WMa and WMb on both sides of the shot area SA, the optical axis AXa is Displacement amount ΔXa obtained from an alignment system unit having an optical axis AXb (hereinafter referred to as an alignment unit XA).
B), the relative rotation error between the reticle R and the wafer W (one shot area SA) can be calculated successively by, for example, a hardware digital subtraction circuit. Is determined immediately during the scanning exposure.

The relative rotation error also depends on the size of the pattern area PA or the shot area SA and the value of the minimum line width.
When a certain allowable amount is determined and a rotation error exceeding the allowable amount may occur, the reticle stage 6 is slightly rotated.
It is desired that the rotation error is corrected in real time during the scanning exposure by feeding back to the mechanism. In this case, Δ
The rotation center of the θ mechanism preferably coincides with the center of the slit opening image of the blind 1 projected on the reticle R.

Here, a specific example of the light transmitting optical system 22 and the photoelectric detection unit 25 in the apparatus shown in FIG. 3 will be described with reference to FIG. In FIG. 5, two light transmission beams B
L1 and BL2 intersect on the illumination field stop 40 and are limited to an illumination area of a predetermined size. The limited two transmitted light beams are passed through a lens system 41, a polarizing beam splitter 42, and a quarter-wave plate 43 to the objective lens 23.
Incident on. As is apparent from FIG. 5, the stop 40 and the grid mark RMa of the reticle R are conjugated to each other with respect to the combined system of the lens system 41 and the objective lens 23. Then, the two light transmitting beams BL1 and BL2 converge as beam waists on a Fourier transform plane (a plane conjugate with the pupil EP of the projection lens PL) in the Fourier space between the lens system 41 and the objective lens 23. At the same time, the respective principal rays of the light transmission beams BL1 and BL2 are symmetric and parallel to the optical axis AXa in Fourier space.

After the two light beams BL1 and BL2 have passed through the polarizing beam splitter 42 almost 100%,
The light is converted into circularly polarized light that rotates in the same direction by the 波長 wavelength plate 43, becomes two parallel beams again via the objective lens 23, and intersects at the position of the mark RMa. Although the arrangement of the marks RMa is shown in FIG. 4, it is assumed that the marks are actually shifted laterally in the non-measurement direction (the direction orthogonal to the lattice pitch direction) with respect to the marks WMa on the wafer side.

FIG. 6 shows the lattice mark R viewed from the reticle R side.
The relationship between Ma and WMa is shown.
0 is an aperture image. Here, the lattice marks RMa, WMa
Is 1: 1 line and space (duty 50
%) And the magnification of the projection lens PL is 1 / M,
The pitch GPr of the mark RMa on the reticle R and the pitch GPw of the mark WMa on the wafer W are GPr =
It is determined by the relationship of M · GPw. At the time of scan exposure,
The marks RMa and WMa move at the same speed v in the same direction (+ X direction in FIG. 6) on the reticle surface with respect to the area ALx. As shown in FIG. 6, the lattice marks RMa and WMa are:
When the alignment is achieved in the non-measurement direction (here, the Y direction), they are arranged so as to be shifted laterally in advance so as to keep a constant interval DS. This interval DS is determined depending on the alignment accuracy in the Y direction.

FIG. 7 shows a state in which the grating mark RMa or WMa moves at the speed + v or -v in the pitch direction, and the interference fringe IF formed on each grating mark has the speed + V.
It is assumed that the current is flowing at Vf. As shown in FIG. 7, when the grating mark moves at the speed + v, the beat frequency of the interference light BT of the ± 1st-order diffracted light generated vertically from the grating mark is Δf because the direction of the interference fringes IF coincides with the flowing direction. When the lattice mark moves at the speed -v, the direction becomes opposite to the direction in which the interference fringes IF flow, so that the beat frequency becomes higher than Δf.

Here, considering the lattice mark RMa, the relationship of the following equation is set assuming that the incident angles θ of the two light transmitting beams BL1 and BL2 are determined symmetrically with respect to the optical axis AXa. sin θ = λ / GPr Then, ± first-order diffracted light from the grating mark RMa is generated in the vertical direction. Under these conditions, the interference fringe IF
Has a relationship of Pif = １ ／ GPr.

From this, the light transmission beams BL1, BL2
Due to the relationship between the difference frequency Δf (kHz) and the grating mark speed v (mm / s), and the Doppler effect, the frequency fs (kHz) of the change in brightness of the interference light BT is fs = Δf, as described above.
+ 2v / GPr. Now, as shown in FIG. 5, the interference light BTr from the reticle grating mark RMa and the interference light BTw from the wafer grating mark WMa are transmitted through the objective lens 23, the quarter-wave plate 43, and the polarization beam splitter 42. And returns on the optical axis AXa to enter the lens system 44.
This lens system 44 acts as an inverse Fourier transform lens, and a surface conjugate with the grating mark RMa or WMa is created. The interference light beams BTr and BTw from the lens system 44 are split into two by the half mirror 45, and reach the light shielding plates 46R and 46W. The light-shielding plate 46R is arranged at a position conjugate with the lattice mark RMa, and has an opening APR arranged so as to block only the interference light BTr from the mark RMa and block other interference light BTw. Similarly, the light shielding plate 46W is arranged at a position conjugate with the lattice mark WMa, and has an opening APW arranged so as to shield only the interference light BTw from the mark WMa and shield the other interference light BTr.

The photoelectric sensor (feto diode, photomultiplier 47R, etc.) receives the interference light BTr from the opening APR and outputs a signal SR, and the photoelectric sensor 47W outputs the signal APW.
And outputs a signal SW upon receiving the interference light BTw. The processing of these signals SR and SW is as described in FIG. As described above, in the present embodiment, the wafer stage 9 is controlled at a constant speed during the scanning exposure. This is because the fluctuation of the scanning speed causes the exposure amount to be uneven in the shot area SA. Further, the blind 1 is not necessarily limited to the slit opening, and may be a regular hexagonal, rectangular, rhombic, or arcuate opening included in the circular image field of the projection lens PL.

A step-and-scan type apparatus using a blind having a regular hexagonal opening is disclosed in Japanese Patent Laid-Open No. 2-229.
No. 423, and the apparatus disclosed therein may incorporate the alignment control method of this embodiment. Next, a second embodiment of the present invention will be described. Here, the first embodiment is used as it is, and two-dimensional (X, Y directions) alignment during scanning exposure is enabled.

Assuming that the scanning exposure is in the X direction, the arrangement and structure of the grating marks on the reticle R and the wafer W are slightly polarized so that the same interference fringe alignment method can be used in the Y direction orthogonal to the X direction. In the present embodiment, the alignment system of the TTR system shown in FIGS.
Two axes are provided for the direction, and grid marks for the X direction and Y direction are provided on each street line of the reticle and the wafer.

FIG. 8 shows the arrangement of each mark on the reticle R and the arrangement of the objective lens of the alignment system. Here, assuming that the X direction is the scanning exposure direction, the reticle R extends in the X direction on both sides of the pattern area PA of the reticle R. In the street line region, grid marks RMYa and RMYb for the Y direction and grid marks RMXa and RMXb for the X direction are provided. These lattice marks are arranged as shown in FIG. 9 as an example, and the lattice mark RMYa for the Y direction is formed by extending several line-and-space patterns in the X direction. A grid mark RMXa is provided. Both sides of the lattice marks RMYa and RMXa on the reticle R are transparent parts, and the corresponding Y-direction lattice mark WMYa and X-direction lattice mark WMXa on the corresponding wafer W are transparent.
And is located.

In this embodiment, of these lattice marks, X
Direction grid marks RMXa (WMXa) and RMXb
(WMXb) is detected through the objective lenses 23Xa and 23Xb of the alignment system for the X direction, as in the first embodiment, and the grid marks RMYa (WMYa) and RMYb (WMYb) for the Y direction are Y It is detected through the objective lenses 23Ya and 23Yb of the direction alignment system.

The alignment system for the Y direction is basically
The configuration is the same as that of the direction alignment system, except that the incident angles of the two beams BL1 and BL2 with respect to the reticle R (or wafer W) are inclined in the YZ plane. The aperture stop (46R, 46W) inside the alignment system for the X direction is a grid mark RMY for the Y direction.
a, WMYa (RMYb, WMYb) are set so as to block the interference light, and the aperture stop inside the alignment system for the Y direction is a grid mark RMXa, WM for the X direction.
The setting is made such that the interference light from Xa (RMXb, WMXb) is also blocked.

Here, when the reticle R and the wafer W are relatively scanned in the X direction, the interference light (beat light) from the lattice mark RMYa for the Y direction on the reticle R and the wafer W
The frequencies of the two signals obtained by photoelectrically detecting the interference light (beat light) from the upper lattice mark WMYa for the Y direction are substantially constant irrespective of the scanning speed of the reticle R and the wafer W in the X direction. (Δf). However, when the amount of the alignment error in the Y direction changes abruptly in time, the frequency of the progress can also change accordingly, or this change is almost negligible. The alignment error amount only needs to detect the phase difference between the two signals. Also in the case of the Y direction, since the alignment error amount is successively output, the reticle stage 6 or the wafer stage 9 is slightly moved in the Y direction so that the error amount always becomes a constant value. Alternatively, a servo (feedback) control of the drive system for the Y direction of the reticle stage 6 or the wafer stage 9 may be performed based on the Y-direction alignment error signal during the scanning exposure.

In this embodiment, the respective alignment systems for the X direction and the Y direction are juxtaposed in the scanning exposure direction, but the alignment systems for the X direction and the Y direction can be aligned via a single objective lens. The beams of the book may be incident simultaneously. Although not shown in the apparatus shown in FIG. 3, a lattice mark on the reticle R and a wafer stage 9 are arranged above the reticle R under illumination light having the same wavelength as the exposure light.
A TTR type alignment system for observing the upper reference mark is provided. This is because the alignment systems shown in FIG. 3 and FIG.
Required for baseline management when using 2.

Now, a TTR alignment system using the same wavelength as the exposure light is arranged, for example, as shown in FIG. FIG.
At 0, the objective lens 23 and the mirror 24 are the same as those in FIG.
2. A TTR alignment system including a beam splitter 61, an objective lens 60, and an image sensor 63 is provided.
A reference mark plate FM is fixed on the wafer stage 9. The optical fiber 62 emits illumination light having the same wavelength as the exposure light, and the illumination light reflected by the beam splitter 61 illuminates the grid mark on the reticle R via the objective lens 60. The illumination light transmitted through the reticle R illuminates the grid mark on the reference mark plate FM via the projection lens PL. On this reference mark plate FM, a grid mark is provided at a position that can be simultaneously detected via the objective lens 23 as shown in FIG.

The image sensor 63 captures images of the lattice mark of the reticle R and the lattice mark of the reference mark plate FM,
It is used to determine the amount of displacement (ΔXe, ΔYe) between both marks. At this time, the interference fringe type alignment system is simultaneously operated via the objective lens 23, and the relative positional deviation amount (ΔXa, ΔYa) between the grid mark of the reticle R and the grid mark of the reference mark plate FM is obtained.

Thus, the baseline amount becomes (ΔXa
−ΔXe, ΔYa−ΔYe). However, in this case, the wafer stage 9 (reference mark plate FM)
Must be kept stationary, so that the wafer stage 9 is generally feedback-controlled so that the measurement value of the laser interferometer 11 becomes a constant value. However, since the laser light path of the laser interferometer 11 is open to the atmosphere, the measured value slightly fluctuates due to slight air fluctuation. For this reason, in the above-described baseline measurement, even if the wafer stage 9 is stopped at the measured value of the laser interferometer 11, drift due to air fluctuation occurs.

However, the alignment system of the interference fringe type shown in FIGS.
In addition, since the beam portion exposed to the air is slight between the projection lens PL and the wafer W, even if air fluctuation occurs, the measurement error due to the fluctuation can be almost ignored. Therefore, the reticle stage 6 or the wafer stage 9 is adjusted so that the reticle R and the reference mark plate FM are aligned using an interference fringe type alignment system.
Feedback control. Thereby, the relative displacement between the reticle R and the reference mark plate FM is driven to almost zero under the alignment system (objective lens 23) of another wavelength. Then, in that state, the amount of misalignment between the lattice mark of the reticle R and the lattice mark of the reference mark plate FM is determined using the image sensor 63. The deviation amount obtained in this way becomes the baseline amount (ΔXB, ΔYB). The base line amounts (ΔXB, ΔYB) are eigenvalues caused by chromatic aberration of the projection lens PL, and are checked each time the detection position of the grating mark on the reticle R (the image height point of the projection lens) changes.

The base line amount (ΔXB, ΔYB) is added as an offset to the relative positional deviation amount between the reticle R and the wafer W detected via the objective lens 23, and is used as a correction to a true superposition position. . In addition, since the position observed through the objective lens 60 is out of the irradiation area (slit shape) of the exposure illumination light on the reticle R, strictly speaking, a deviation may cause an inherent error. . The error is mainly due to distortion caused by the exposure wavelength of the projection lens PL. However, since the amount of distortion at each point in the projection field of view of the projection lens PL can be obtained in advance, the amount of distortion at the observation position of the objective lens 60 is stored as an offset amount unique to the apparatus, and the baseline is stored. The measured value may be further corrected.

FIG. 11 shows a grid mark arrangement according to the third embodiment of the present invention. In particular, a grid mark formed in a street line of a wafer W is formed into a two-dimensional grid to save space. . In FIG. 11, on the reticle R, a grid mark RMYa for the Y direction and a grid mark RMXa for the X direction are provided at fixed intervals in the Y direction. 2
The dimensional lattice WMxy is set to be located. The two-dimensional lattice WMxy converts a minute rectangular dot pattern into the X direction and the Y direction.
They are arranged at a predetermined pitch in both directions. At the time of actual alignment, it is better to separate the positions of the alignment system for the X direction and the alignment system for the Y direction as shown in FIG.

However, if the two beams for the X-direction alignment and the two beams for the Y-direction alignment are set in mutually complementary polarization states, the interference light generated perpendicularly from the two-dimensional grating WMxy will be polarized. Since the beams can be separated by characteristics, it is possible to irradiate four beams (two beams for the X direction and two beams for the Y direction) to the lattice mark simultaneously via the same objective lens 23.

As described above, by providing the two-dimensional grating WMxy in the entire scanning direction along the shot area on the wafer, considerable space can be saved and two-dimensional alignment correction during scanning exposure can be performed. . By the way, the general street line area on the wafer is the width (Fig. 1
1 (dimension in the Y direction) is about 70 μm. 2
If the size of the rectangular dots of the dimensional lattice WMxy is 4 μm square (that is, the pitch is 8 μm), eight rectangular dots can be formed in the Y direction, which can be expected to have practically sufficient measurement accuracy. Also, the lattice mark R on the reticle side in FIG.
MYa and RMXa can also be two-dimensional lattices similar to those on the wafer side.

FIG. 12 shows a grid mark arrangement according to the fourth embodiment of the present invention, in which a one-dimensional or two-dimensional grid is arranged in a street line area extending in the scanning exposure direction outside the pattern area PA of the reticle R. Marks RML1 to RML4,
RMR1 to RMR3 are provided at intervals in the X direction. On the wafer W, one-dimensional or two-dimensional lattice marks are provided at positions corresponding to the ones at intervals in the X direction. The marks RML1 to RML4 and the marks RMR1 to RMR3 are arranged in a nested state with each other.
Assuming that the objective lenses 23R and 23L are arranged apart from each other in the Y direction, one of the objective lenses 23R and 23L can always receive the interference light from the lattice mark during scanning exposure in the X direction. Is set. For example,
When the reticle R moves right and left from the position shown in FIG. 12, the lattice mark RML1 and the corresponding lattice mark on the wafer are aligned via the objective lens 23L, and two (or four) from the objective lens 23L are aligned. Immediately before the irradiation area of the transmitted light beam deviates from the lattice mark RML1, the lattice mark RMR1 reaches the irradiation point of the transmitted light beam from the objective lens 23R. Therefore, in the next cycle, the grid mark RMR1 and the corresponding grid mark on the wafer are aligned via the objective lens 23R. Similarly, the objective lens 23 is moved in accordance with the scanning exposure signal.
R and 23L are alternately switched for alignment.
In the present embodiment, when switching from the grating mark RML1 to the RMR1, the interference beat light obtained through the objective lens 23 and the interference beat light obtained through the objective lens 23R are separated from each other by a short time. Each mark is arranged so as to exist only at the same time.

When the grid marks are arranged as in this embodiment, another mark, for example, a global alignment (E) of the wafer, is placed between the grid marks in the X direction.
GA) marks can be arranged. FIG.
The configuration of a projection exposure apparatus according to a fifth embodiment of the present invention is different from the configuration of FIG. 3 in that a plurality of objective lenses 2 of an alignment system are arranged in a scanning direction of a reticle R (and a wafer W).
3A, 23B, 23C, and 23D are arranged. The arrangement of the reticle R and the lattice mark on the wafer W may be any of the methods shown in FIGS. 4, 8, 11, and 12.

In the case of FIG. 13, the four objective lenses 23
A to 23D each receive interference beat light generated at a different position on the lattice mark to perform alignment during the scanning movement of the reticle R and the wafer W, but depending on the scanning position, the objective lenses 23A and 23D on both sides. Sometimes only one of them can be used. Therefore, as one alignment sequence, for example, when the reticle R is scanned from the left side to the right side in FIG. 13, the alignment system used in the order of the objective lens 23A, the objective lens 23B... You can change the number and position. When a plurality of alignment systems can be used at the same time as shown in FIG. 13, the lattice mark RML1 shown in FIG.
ＲRML4 and the lattice marks RMR1 to RMR3 do not need to be nested, and signals for alignment can be obtained almost continuously.

For this purpose, for example, the objective lenses 23A to 23A
23D scanning direction (X direction) interval and grid mark RM
What is necessary is just to make the space | interval of X direction of L1-RML4 different. As described above, according to the above-described embodiment, the reticle (mask)
One of the reticle and the wafer is scanned in the speed control mode while the relative displacement between the reticle and the wafer (photosensitive substrate) is always detected by the alignment sensor. Since the scanning is performed in the servo (tracking) control mode, even if there is speed unevenness on the side scanned in the speed control mode, the magnification error in the scanning direction (expansion and contraction of the transferred image size in the scanning direction) It can be held down to the measurement accuracy determined by the alignment sensor. Therefore, even if one shot area on the wafer itself expands and contracts non-linearly in the scanning exposure direction due to the influence of the wafer process or uneven scanning during the exposure of the shot, the measurement signal from the alignment sensor is always used as a reference. As long as tracking scanning is performed, extremely high registration accuracy can be obtained over the entire surface within one shot area. FIG.
As in the embodiment shown in FIG. 2, if the grid marks are arranged in a stepping manner in the scanning direction, it is possible to save space when replacing the marks in an IC manufacturing process involving multiple layers. If the interval between the stepping stone-shaped grid marks is not made too wide, even if a dead time occurs between the end of the detection of one grid mark and the start of the detection of the next grid mark, the dead time is reduced by the air of the laser interferometer. It is possible to make the period shorter than the fluctuation period, and it is possible to obtain the same accuracy as when the grid marks are continuously provided over the range of the scanning exposure. That is, during the dead time during which no interference beat light is obtained from any of the grating marks, even if each stage is moved based on the laser interferometer for the wafer stage or the laser interferometer for the reticle stage,
There is an advantage that it is not affected by air fluctuation as compared with the conventional device. Further, even when a plurality of alignment sensors are arranged in the scanning exposure direction and the grid marks are provided in the scanning direction as in the embodiment shown in FIG. If different values are set, any alignment sensor can always detect the amount of misalignment, or the dead time during which neither alignment sensor detects a mark can be extremely short. Overlay accuracy can be improved. The above-described alignment method during scanning exposure can be applied to a conventional step-and-scan type exposure apparatus as it is.

[0063]

As described above, according to the scanning exposure apparatus of the present invention, an alignment error can be obtained almost continuously even during exposure during scanning.

Further, by performing relative positioning between the mask and the substrate on the basis of the obtained alignment error, deviation from the alignment state between the mask pattern and the pattern on the wafer during the scanning is allowed. It can be held down within the range.

Therefore, it is possible to minimize the need for scanning for alignment before exposure.

Further, the alignment accuracy and the overlay accuracy during the scanning exposure can be improved.

[Brief description of the drawings]

FIG. 1 is a perspective view illustrating an alignment method in a conventional step & scan exposure apparatus.

FIG. 2 is a waveform chart showing an alignment signal waveform obtained by the alignment method of FIG. 1;

FIG. 3 is a diagram showing a configuration of a projection scanning exposure apparatus according to an embodiment of the present invention.

FIG. 4 is a perspective view illustrating an alignment method using the apparatus of FIG.

FIG. 5 is a diagram showing a configuration of an alignment system of the apparatus shown in FIG. 3;

FIG. 6 is a diagram showing an arrangement of a reticle used in the apparatus shown in FIG. 3 and lattice marks of a wafer.

FIG. 7 is a view for explaining the operation principle of the alignment system of FIG. 5;

FIG. 8 is a plan view of a reticle having a mark arrangement according to a second embodiment.

9 is a plan view showing the mark arrangement of FIG. 8 in a partially enlarged manner.

FIG. 10 is a diagram for explaining a baseline measurement method.

FIG. 11 is a view for explaining mark arrangement according to a third embodiment.

FIG. 12 is a view for explaining mark arrangement according to a fourth embodiment.

FIG. 13 is a view for explaining an apparatus configuration according to a fifth embodiment.

[Explanation of symbols]

R Reticle RMa, RMb, RMXa, RMYa, RMXb, RM
Yb reticle grating mark WMa, WMb, WMYa, WMXa, WMxy Wafer grating mark W wafer PL Projection optical system 1 reticle blind 6 reticle stage 8 reticle stage drive motor 9 wafer stage 10 wafer stage drive motor 11 laser interferometer 12 wafer stage drive control Unit 20 Laser light source 23 Objective lens for alignment 27 Position shift amount detection unit

Claims (11)

(57) [Claims] An apparatus for scanning and exposing a pattern of the mask on the substrate by relatively moving the mask and the substrate with respect to the illuminating light, wherein a predetermined pattern is provided over substantially the entire exposure range on the substrate. A mark detection system for detecting a diffraction grating-shaped substrate mark formed along a direction, and the direction of the relative movement so that the substrate mark is continuously detected by the mark detection system during the scanning exposure. And a drive control means for moving the substrate substantially in the predetermined direction. 2. The scanning exposure apparatus according to claim 1, wherein the substrate mark includes a diffraction element whose pitch direction substantially coincides with the direction of the relative movement. 3. The scanning exposure apparatus according to claim 2, further comprising a diffraction element whose pitch direction substantially coincides with a direction orthogonal to the direction of the relative movement. 4. The mark detection system detects a diffraction grating-like mask mark formed along the direction of the relative movement over substantially the entire pattern area of the mask together with the substrate mark. The scanning exposure apparatus according to claim 1. 5. The scanning exposure apparatus according to claim 4, wherein said mask mark includes a diffraction element whose pitch direction substantially coincides with the direction of said relative movement. 6. The scanning exposure apparatus according to claim 5, wherein the mask mark includes a diffraction element whose pitch direction substantially coincides with a direction orthogonal to the direction of the relative movement. 7. A projection optical system for projecting at least a part of a pattern formed on the mask onto the substrate, wherein an irradiation area of the illumination light is within a field of view of the projection optical system.
The scanning exposure apparatus according to any one of claims 1 to 6, wherein a longitudinal direction thereof coincides with a direction orthogonal to the direction of the relative movement. 8. The projection optical system according to claim 7, wherein the projection optical system has a circular visual field, and the irradiation area is defined such that the center thereof substantially coincides with the optical axis of the projection optical system.
3. The scanning exposure apparatus according to claim 1. 9. A mask stage for holding a mask, and a substrate stage for holding a photosensitive substrate, wherein an original pattern of the mask is transferred onto the photosensitive substrate by synchronously moving the mask stage and the substrate stage. In a scanning exposure apparatus, while irradiating the mask with illumination light, an irradiation area of the illumination light on the mask includes an optical axis of the projection optical system, and a longitudinal direction thereof is orthogonal to a direction of the synchronous movement. A plurality of alignment systems for detecting mask marks, which are used by switching according to the position of the mask stage in the direction of the relative movement and which are used to detect a mask mark. A scanning exposure apparatus. 10. The apparatus according to claim 9, wherein said plurality of alignment systems are arranged apart in the direction of said relative movement. 11. The apparatus according to claim 9, wherein said plurality of alignment systems are arranged apart in a direction orthogonal to the direction of said relative movement.