JPS61278137A - Alignment method for semiconductor exposure device - Google Patents

Abstract

PURPOSE:To reduce the number of alignment optical systems by automatically detecting positions of the (x) direction and (y) direction of each chip on a wafer by one alignment pattern and one alignment detection optical system. CONSTITUTION:Beam light emitted from an Ar laser 45 is reflected by a mirror section 37, and lights an alignment pattern 61 on a wafer 3 through a reduction projection lens 2. Reflected beams from the wafer 3 are reflected by the mirror section 37 on the lower surface of a reticle, and image-formed at a position 39 on just a half mirror 40 on the surface lower than the reticle only by the chromatic aberration section of the reduction projection lens 2 to the wavelength of the Ar laser 45. The image is reflected by a mirror 41, and formed at the front-side focal position 85 of a Fourier transformer lens 19 by a magnifying lens 47. The magnified image of the front-side focal position 85 is complex Fourier-transformed optical by the Fourier transformer lens 19, and complex amplitude distribution thereof is overlapped on an interference pattern 25 on an optical correlation filter 26, thus forming correlation. Complex amplitude distribution is complex Fourier inverse-transformed optically by a Fourier transform lens 28, and an optical peak is detected by a solid-state image pickup element 51.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to an alignment method for a semiconductor exposure apparatus, and particularly to an alignment method suitable for detecting the position of a 72-dimensional pattern on a wafer with high precision.

[Background of the invention]

The present invention will be described in detail by taking a reduction projection exposure apparatus as an example of a semiconductor exposure apparatus.

As the miniaturization of semiconductor integrated circuits progresses, higher and higher alignment precision between a reticle and a wafer is required during exposure using a reduction projection exposure apparatus. Therefore, TTL (T
The through Th@Lens) alignment method has become mainstream in the manufacture of highly integrated circuits.

FIG. 17 shows an example of the TTL alignment method. The circuit pattern of reticle L is the reduction projection lens 2
The wafer 3 is exposed one to several chips at a time. 4 is a wafer steno.

First, the reticle 1 is set at the initial position by the reticle alignment optical system 5.5'. Then, the alignment turn 14.14' on the wafer 3 is imaged onto the alignment/IP turn 13.13' on the reticle 1 through the reduction projection lens 2, and both patterns are detected by the wafer alignment and alignment detection optical systems. , align. The wafer alignment detection optical system includes a mirror 6.6', lenses 7.7' and 8.8', a movable slit 9.9', a photomultiplier tube 10.10', and alignment illumination light having the same wavelength as the exposure light. It consists of optical fibers 11 and 11' that emit light. 18(a) to 18(,) show a method for detecting the center position of the alignment pattern 14 on the wafer 3. FIG. The figure (&) shows the alignment/I on the wafer 3 imaged within the window-like alignment/quantity turn 13 on the reticle.
The P-turn 14 is shown, and FIG. 3' is an 81 substrate, 3' is StO for forming alignment/9 turns 14, a step, and 15 is a photoresist. FIG. 6(c) shows detection signals for both alignments and nine turns output from the photomultiplier tube 10 as the movable slit 9 moves. 16 is a detection signal of the alignment pattern 13 on the reticle 1, and 17 is a detection signal of the alignment pattern 14 on the wafer 3. Center position IW of alignment groove 14 on wafer 3
is obtained by finding the center of symmetry of the waveform of the detection signal 17, and the shear alignment amount is obtained by the difference Δ from the center position ①R of the alignment turn 13 on the reticle. When the alignment is completed, exposure light is irradiated by the exposure system 12 (FIG. 17). A related device of this type is JP-A-55-41739.

Problems with this alignment method include the following points. First, a wafer alignment detection optical system in both directions is required, and a total of four alignment optical systems including two reticle alignment optical systems are located around the reticle, making the apparatus complicated and large. Also, “s”
Since the alignment cuts on the wafer in the V-force direction are detected and processed separately, the alignment time is long and the throughput is reduced.

Furthermore, since the reflected light from the alignment cutter on the wafer is directly imaged and the center position of the alignment pattern is determined from its intensity distribution, the intensity distribution is greatly affected by multiple interference of illumination light inside the resist. ing.

That is, as shown in FIG. 19(,), the intensity of the reflected light 98 from the surface of the resist 15 at the alignment non-turn portion is 11, the intensity of the reflected light 98' from the underlayer 3' is I,
Then, as a result of the interference of both reflected lights, the intensity I of the interference light at the alignment/9 tan part is obtained where n is the refractive index of the resist, λ is the wavelength of the illumination light, and d is the film thickness of the resist. On the other hand, suppose that the intensity of the reflected light 99 from the surface of the renost 15 and the intensity of the reflected light 99' from the step layer 3 in the vicinity of the alignment pattern are 11 and 11, respectively.

Then, similarly, the intensity (2) of the interference light in the vicinity of the alignment pattern portion is given by the above equation (1). Then,
The difference between the Lennost film thickness 100 at the alignment point and turn part and the Lennost film thickness 101 near the alignment cutter 7 is exactly (λ/2 ny)" n (n = Or
1 +2,...), the intensity of the interference light near the alignment pattern part and the alignment pattern part is not equal, and the detection signal has a low contrast as shown in Figure 19(b). However, there is a problem in that the alignment accuracy decreases.

Furthermore, when using a resist containing a light absorbing agent or a multilayer resist, it is necessary to use light with a wavelength different from that of the exposure light as the alignment pattern illumination light. and turns are reticle alignment) z? The image is formed at a position away from the turn by the amount of the envelope aberration difference. Therefore, the reticle alignment nine turns and the wafer alignment turn must be detected separately, making the detection optical system increasingly complex.

[Purpose of the invention]

SUMMARY OF THE INVENTION An object of the present invention is to provide an alignment method for a semiconductor exposure apparatus that solves the problems of the above-mentioned conventional techniques, has a simple configuration, is quick and accurate, and can perform alignment in both I and 7 directions simultaneously. .

[Summary of the invention]

In order to achieve the above object, in the present invention, the four turns of the pliment on the wafer are illuminated with coherent light, the complex amplitude distribution of the reflected light from the four turns is optically complex Fourier transformed, and the above-mentioned An optical correlation filter having a complex amplitude distribution conjugate to the complex amplitude distribution obtained by complex Fourier transform is installed, both complex amplitude distributions are optically superimposed, and the complex amplitude distribution obtained by the superposition is optically The alignment pattern position on the wafer is detected from the maximum amplitude position in the complex amplitude distribution obtained on the complex Fourier transform plane.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to FIGS. 1 to 16.

First, the present invention will be explained in detail with reference to FIGS. 13 to 17. FIG. 13 is a diagram showing optical complex Fourier transform of the complex amplitude distribution of reflected light obtained from the alignment pattern 61 placed near the chip/turf 18 on the wafer. Alignment of the detection optical system (not shown) on the wafer/Fourier transform lens (Fourier transform lens) e19 at a position 86 distant from the focal length f from the imaging position 85 of the 4' turn 61
Set up. When the alignment pattern 61 on the wafer is illuminated with coherent light (not shown), a complex Fourier transform image 20 of the alignment pattern 61 is obtained at the rear focal position 87 of the Fourier transform lens 19.

That is, the complex amplitude distribution of the reflected light from the wafer is expressed as w(x
, y), the complex amplitude distribution of the reflected light from the alignment pattern 61 is p(x, y), the complex amplitude distribution of the reflected light from other areas is s(x, y), and the position of the wafer turn is (xy * yw), then v(x, y) = e(:c, y) + p(x-xw, y-y
w) −(2). Therefore, the back focus position 87
Complex Fourier transform image 20 of reflected light from the wafer at
The complex amplitude distribution W(μ, ν) of W(μ, ν) =? (=Cx, y))=f (-(
x, y) + p(x-x-, y-yW) )"' E
(μ, S+P(μ, y)*exp(-j2π(μazw
+ν・yw))...A3) However, 7"(): Operator μ, ν representing complex Fourier transform: Spatial frequency E(μ, S:・Complex Fourier transform image P(μ) of (x, y)
, S: becomes a complex Fourier transform image of p(x,y). In this embodiment, the alignment pattern 61 on the wafer is a square pattern of 6 μm square (on the wafer).

FIG. 14 is a diagram showing a method of creating an optical correlation filter for detecting the position of the alignment pattern 61. First, at the front focal position 88 of the Fourier transform lens 23, a chromium pattern (opaque body) 95 having exactly the same shape and size as that at the imaging position 85 in FIG.
A model mask 21 having the following characteristics is installed. Further, a photographic plate 26 is installed at the rear focal position 90. Then, this model mask 21 is illuminated with coherent light 22 having the same wavelength λ used for detecting the alignment pattern. Then, the same m as in Fig. 13, 7-IJ-! : The complex amplitude distribution of the transmitted light from the model mask 21 is subjected to complex Fourier transform by the conversion lens 23, and a complex Fourier transform image is obtained at the rear focal position 90. At this time, a plane wave 24 having the same wavelength as the illumination light 22 is made incident as a reference light at an incident angle θ with respect to the optical axis of the light 22, optically interferes with the complex Fourier transformed image, and this interference pattern 25 is printed onto a photographic plate 26. Now, the complex amplitude distribution from the model mask 21 is f(x, y), and the complex amplitude distribution of the plane wave 24 is the unit amplitude UB=axp(j2π
α, the transmittance M(μ, ν) of the interference pattern 25 printed on the photographic plate 26 is M(μ, ν)=17−(f(x,y))+Uml”=I
F(μ, ν)+exp(j2Kas12=: 1 +
IP(A, S12 IF(μ, ν)−@XP(j2παν)+F′″(μ,
ν) eexp(j2Kccν)...---(4
), where F(μ, ν): complex Fourier transform of f(x, y), F9(μ, ν) and conjugate complex function α of F(μ, ν)
= image θ/λ (λ: wavelength). The above is the method for creating the optical correlation filter used in this example. In the following, the photographic plate 26 on which the interference pattern 25 is printed will be referred to as an optical correlation filter 26.

FIG. 15 is a diagram showing the principle of detecting the position of the on-wafer alignment noreen 61 by optical correlation using this optical correlation filter 26. First, as in FIG. 13, the on-wafer alignment of the detection optical system (not shown) is performed. A Fourier transform lens ('19) is installed at a position 86 that is a focal length f away from the imaging position 85 of the turn 61.

Also, the above-mentioned optical correlation filter 26 is located at the rear focal position 87.
At the same time, a Fourier transform lens arm is installed at a position 91 that is a focal length f away from the rear focal position 87. Now, when the on-wafer alignment noturn 61 is illuminated with coherent light of the same wavelength as that used to create the optical correlation filter 26 (not shown), the complex amplitude distribution of the reflected light from the wafer is as follows. A complex Fourier transform is optically performed by the Fourier transform lens 19, and as a result, the complex amplitude distribution immediately before the optical correlation filter 2G is expressed by the above equation (3).

Therefore, the complex amplitude distribution Q(μ, ν) immediately after the optical correlation filter 26 is Q(μ, ν)=W(μ, ν)−M(μ, ν)=[E(μ
, S+P(μ, ν)@xp(-j2π(μ・Xy+ν'
yw))]−(1+IF(μ,ν)I”+F(μ
,y)txp(-j2παri1F"(μ, ν)11・x
p(j2παν)] = E(μ, S+E(μ, S・IF(
μ,su12+E(μ,ν)−F(μ,suexp(−j
2παs+1i: (, ci, u)・F”(μ, ν)ax
p(j2παν)+p(μmν)axp(−jzπ(μ
oz, +highw))+p(μ,su*IF(μ,sl”e
@xp(-j2π(μ・xW+ν・yw))+P(μ,
ν)11F′(μ, ν)exp[−j2π(μ*xw+
ν*yy+asu)+p(μ, Lee F*(μ, ν)exp
(−j2π(μ・xW+shi117w −αs),...
...(5) becomes. This complex amplitude distribution Q(μ, ν) is further optically inversely complex-Fourier-transformed by a Fourier-transform lens path. By inverse transformation, the complex amplitude distribution r(X, 7) obtained at the rear focal position 31 of the Fourier transform lens 4 is r(x, y)='y''''1(Q(μ, ″″8
(W(μ, ν)・M(μ, ν))=e(xIy) +e(x, y) fist f(x, y)Of(:, y)+e (
X + 3') book f(x, y) * δ(X+7-α
)+・(x,y)Of(x,y)umbrella δ(x,y+α)+
p(x,y)*δ('E ``W,'!)'w)+p(
x, y) of (x, y) of (x, y) umbrella δ (X-Xw
*7)'w) + p(x, y) books f(x, y) books δ
(X Xy, (7-72(ri)+p(x,y)Of(
xeF)"δ(+c-xw-(y yw+cri)
・＊+ (6) However, “(”+7)”(”, y):
Convolution integral of e (x, y) and f (x, y) δ (x, y): delta function (however, the amplitude is 1) 8 (
X1y)Of(x,y): a(x,y) and f(x
, y). Here, the light intensities of the first to third terms and the fifth to seventh terms are very small, so they can be ignored. On the other hand, the fourth term is the complex amplitude distribution e(x,
y) and the interference pattern 5 on the optical correlation filter 26 created based on the alignment pattern 61. Alignment zJ? When there is a pattern with a similar shape near the turn 61, a low optical peak 30 appears at a position (0, α) close to the optical axis at the rear focal position 31, as shown in FIG. The eighth term is the correlation between the alignment pattern 61 and the interference pattern 25 on the optical correlation filter 26 to be determined.

Since the interference pattern 25 is created based on the alignment pattern 61, this eighth term shows a very high correlation, and at the rear focal position 31, the coordinates (XW+ 3'W) and shifted +α in the y direction ("iFe
A sharp correlated light peak 4 is produced at yw+α).

Therefore, if a pattern with a similar shape is always prevented from being formed in the vicinity of the alignment/return 61 within the detection field of view, the optical peak 30 will always be the same as the optical peak 2.
The position of the alignment lid 761 can be detected with an extremely high signal-to-noise ratio of less than 9. Note that since the shift amount α in the X direction is always fixed, it can be easily corrected.

FIG. 16 shows the rear focus position 31 of the Fourier transform lens path.
This figure shows how a solid-state image sensor is installed as a light-receiving element and a correlated light peak is received on the element surface 96. After storing the detection signal in the image memory etc.,
By reading the signal as a signal in the X direction and finding the center of symmetry of each signal, or by binarizing it with an appropriate threshold value and finding the center position of the optical peak 32 in the X direction and the optical beak in the Find the center position of 61 in the X and X directions. The above describes on-wafer alignment using an optical correlation filter according to the present invention.
This is the principle of turn position detection.

FIG. 1 shows a Ueno alignment detection optical system constructed based on this principle. In this example, as the on-wafer alignment/cutan illumination light, we used Ar with a wavelength of 514°5 nm, which is different from the exposure light, from the viewpoint that it is compatible with light-absorbing agent-containing silenost and multilayer resist, and it must be coherent light. It uses Rede. The beam light emitted from the Ar radar 45 is reflected by the mirror 44 and the half mirror 46, and then the shutter 43
, relay lens 42, mirror 41 and half mirror 40
After that, the light is reflected by a mirror portion 37 formed of chrome CR on the lower surface of the reticle, and illuminates the alignment pattern 61 on the wafer 3 through the reduction projection lens 2. wafer 3
The reflected light passes through the reduction projection lens 2 again, is reflected by the mirror portion 37 on the lower surface of the reticle, and is reflected by the monochromatic aberration of the reduction projection lens 2 relative to the wavelength of the Ar radar 45, just above the half mirror 40 on the lower surface of the reticle. The image is formed at position 39. This image is reflected by the mirror 41, passes through the relay lens 42, the shutter 43, and the half mirror 46, and is focused on the front focal position 85 of the Fourier transform lens 19 by the magnifying lens 47. Thereafter, based on the principle diagram in FIG. Figure 15) overlaps and correlations are taken. Then, the complex amplitude distribution immediately after the optical correlation filter 26 is optically subjected to inverse complex Fourier transform using the Fourier transform lens 28, and a light peak as shown in FIG. 15 is detected using the solid-state imaging device 51. The detection signal is processed by a computer, and as shown in FIG. 16, the alignment/turn position (W+3'W) is calculated from the position of the optical peak 29 (Wr3'W+α). Note that at this time, the shutter 59 is closed.

FIG. 2 shows the arrangement of the alignment pattern 61 on the wafer 3 in this embodiment.

According to the invention, one alignment no! Since position detection in the X and X directions is possible with the turn, the alignment mother turn 61 may be provided close to either the X or one side of the X direction of the nine turns 18.

On the other hand, the position of the reticle is detected as follows. 1st
In the figure, shutter 43 is closed and shutter 59 is opened. The light beam emitted from the Ar radar 45 passes through a mirror 44,
Half mirror 46, half mirror 55, relay lens 5
4, the light is reflected by a half mirror 40, and illuminates a reticle alignment turf 38 provided on the lower surface of the reticle. This reticle alignment pattern 38
As shown in the figure, it is provided outside the circuit pattern 60 of the reticle 1 adjacent to the /4' turn of the mirror portion 37. In addition, reticle alignment/parallel turn 38 is
A reticle alignment turf 38' for direction detection is also provided.

FIG. 4 is an enlarged view of the main turn 37 and the reticle alignment turn 38 of the mirror portion 37. Reference numeral 63 indicates a spot of illumination light and reflected light on the alignment or turn on the wafer. The reticle alignment data turn 38 is a nine-turn series in which a hyperbolic Fresnel mother-turn 97 is connected with an approximate mother-turn 62. Therefore, when this alignment mother turn 38 is illuminated with the light beam of the Ar lede, the hyperbolic Fresnel no J? Due to the turn, the reflected light is diffracted and condensed in a straight line at the position shown in FIG. 1 53 . This image is enlarged and formed on the element surface 58 of the solid-state image sensor 57 via the relay lens 54, half mirror 55, and magnifying lens 56.

FIG. 5 shows the diffraction image on the element surface 58 and the detected signal waveform. xR is reticle alignment pattern 3B
is the center position. In the y direction, the center position y of the reticle alignment/turf 38' is determined in the same manner. The y-direction alignment detection optical system is for detecting only the reticle alignment/4 turn 38', so the Ar radar 45, mirror material, half mirror 39°55, relay lens 54,
It is sufficient to use only the magnifying lens 56, the li body fi [corresponding to the E element 57], and the configuration is simplified. and,
The position of the wafer alignment parable turn found earlier (xw
The alignment amount is determined from the difference between +)'w).

As described above, according to this embodiment, the position of each chip on the wafer in the X direction and the y direction is divided into one 7 alignment i? It is possible to detect the turn with a single alignment detection optical system, and it is possible to simplify the area around the reticle and downsize the reduction projection exposure apparatus, while also shortening the alignment time. In addition, like the conventional alignment method, the reflected light from the nine alignment turns on the wafer is imaged as it is, and from the intensity distribution, the alignment no. Compared to the method of determining the center position of the turn, this method optically performs a complex Fourier transform on the complex amplitude distribution of the reflected light from the alignment pattern, thereby obtaining not only the amplitude distribution information of the reflected light but also the phase distribution information, that is, Alignment no! Phase information based on the difference in resist film thickness between the turn portion and its vicinity is used. As a result, even if the intensity distribution of the reflected light has a low contrast, a highly correlated light peak appears on the output surface, making it possible to detect the alignment position with high precision.

FIGS. 6 and 7 are diagrams showing a second embodiment of the present invention. In the above example, nine turns of a 6 μm square (on the wafer) were used as the alignment mother turn on the wafer, but in this case, the alignment between each chip and the optical correlation filter in the detection optical system was It is necessary to perform angle correction in the direction perpendicular to the optical axis between the two. In other words, in FIG. 15, if there is an angular deviation between the alignment 9 turns 61 and the interference 9 turns 25 on the optical correlation filter 26, the peak of the correlated light at the output surface 31 will become low and the detection accuracy will decrease. descend. Therefore, in the second embodiment, as shown in FIG. 6, a circular nine-turn 68 having a diameter of 6 μm is used as the nine alignment turns on the wafer. That is, since the complex Fourier transform image of the circular/arm turn is rotationally/axially symmetrical with respect to the optical axis, even if an angular shift occurs between the alignment pattern 68 and the interference pattern 69, the output surface 31
The height of the correlated light peak at is always constant, eliminating the influence of multiple angle shifts. Furthermore, if this alignment master turn is detected in two chips on the wafer, the angular deviation between the wafer and the optical correlation filter can be detected from the respective X and Y coordinates, and this can be used to correct the angular deviation. be able to.

FIG. 7 shows the arrangement of the alignment/data turn 68 on the wafer 3 in this embodiment.

In this embodiment, as in the previous embodiment, position detection in the X and y directions is possible with one alignment turn. It may be provided close to.

8 to 12 are diagrams showing a third embodiment of the present invention. In this embodiment, first, as shown in FIG.
Similar to the previous two embodiments, reticle alignment, 4
A turn 73 is provided on the outside of the circuit turn 60 of the reticle 1, adjacent to the mirror line 37.

FIG. 9 is an enlarged view of the mirror pattern 37 and the reticle alignment/4' tar 773. In this embodiment, the position detection of the reticle 1 is performed using hyperbolic @ as in the previous two embodiments.
7 on the wafer rather than using a 7 Renel pattern.
A Cr zQ turn 74 having the same shape and size as the alignment pattern is used, and the alignment detection optical system and optical correlation filter used for wafer alignment/tonolean detection are used as they are.

FIG. 1G shows the alignment detection optical system in this embodiment. Since the detection of wafer alignment and pattern is exactly the same as in the first embodiment, the explanation thereof will be omitted. When detecting the fourth turn of reticle alignment, first the shutter 43 is closed and the shutter 59 is opened. The light beam emitted from the Ar radar 45 passes through a mirror 44, a half mirror 46, a shutter 59, a half mirror 75, a relay lens 79, and a half mirror 40.
After that, the reticle alignment turn 73 is illuminated. The reflected light passes through the half mirror 7-40, the relay lens 79, the half mirror 75, the mirror 76, and the magnifying lens 77, and forms an enlarged image at a position 85 on the half mirror 78. At this time, the size of the pattern 74 on the reticle alignment groove 773 is enlarged so that the enlarged image of the reticle alignment pattern 73 on the half mirror 78 and the enlarged image of the wafer alignment pattern 61 are the same size. Adjust the magnification of the lens 77. Below, as shown in Fig. 1f, the sign of the coordinates ("11*7R) of the position of the pattern 74 is inverted on the output surface 31 according to the principle of optical correlation in the first embodiment (Fig. 15). , position +α shifted in the X direction (XH, -yR+
A sharp correlated optical peak 80 occurs at α). Although the wafer alignment groove 61 and its correlated light peak 29 are shown in FIG. 11 for comparison, in reality, both are detected one by one by opening and closing the shutters 43 and 59. FIG. 12 shows an image 80 of the correlated light peak on the imaging surface 96 of the solid-state imaging device 51 and its corresponding X direction and
Direction detection signals 81 and 83 are shown. The weak image 30 and its signal peaks 82 and 84 interfere with areas other than the reticle and flat turns as in the previous embodiment.
This is caused by a correlation with 5, but the light intensity is very low, so it does not pose a particular problem.

As explained above, according to each embodiment, it is possible to detect not only each chip on the wafer but also the position of the reticle in the X direction and the X direction using the same alignment detection optical system and one optical correlation filter. This enables the periphery of the reticle to be simplified and the reduction projection exposure apparatus to be more compact, while improving system throughput by shortening alignment time.

Although the present invention has been described based on embodiments, the present invention is applicable not only to reduction projection exposure apparatuses but also to X-ray exposure apparatuses, etc. It is fully applicable as an alignment method set for a roximity exposure apparatus, a contact type exposure apparatus, a reflection type projection exposure apparatus, etc.

〔Effect of the invention〕

As explained above, according to the alignment method of the present invention, the position of each chip on the wafer in the X direction and the
Two alignments! Turns can be automatically detected using one alignment detection optical system, the number of alignment optical systems can be reduced, the area around the reticle can be simplified, and the reduction projection exposure apparatus can be downsized. Moreover, as a result, there is an effect of simplifying and shortening the time required for assembling and adjusting the device. Further, according to the present invention, since alignment in the X direction and the X direction is possible at the exposure position, the alignment time can be shortened and throughput can be improved, that is, semiconductor productivity can be improved.

[Brief explanation of the drawing]

FIG. 1 is a configuration diagram of a wafer alignment detection optical system according to a first embodiment of the present invention, FIG. 2 is a layout diagram of an alignment pattern on a wafer in the first embodiment, and FIG. 3 is a reticle diagram in the first embodiment. A layout diagram of the mirror 4 turns and reticle alignment 9 turns provided on the lower surface, FIG. 4 is a partially enlarged view of FIG. 3, and FIG. 5 shows the reticle alignment ? FIG. 6 is a diagram showing the relationship between the diffraction image of a turn and its detection signal. FIG. 6 is a diagram showing the principle of detecting the position of a mother turn on a circular wafer using an optical correlation filter in the second embodiment of the present invention. FIG. FIG. 8 is a diagram showing the arrangement of the alignment/detailed turns on the wafer in the embodiment. FIG. 8 shows the Milano master turns and reticle alignment/! 9 is a partial enlarged view of FIG. 8, FIG. 10 is a configuration diagram of the wafer alignment detection optical system in the third embodiment, and FIG. 11 is a diagram showing the use of the optical correlation filter in the third embodiment. Fig. 12 is a diagram showing the position detection principle of the reticle alignment/turns in the third embodiment. FIG. 13 is an explanatory diagram showing the optical complex Fourier transform of the complex amplitude distribution of the reflected light from the wafer alignment/turn in the present invention, and FIG. 14 is a diagram for detecting the position of the wafer alignment/turn. FIG. 15 is an explanatory diagram showing the method for creating an optical correlation filter. FIG. 15 is an explanatory diagram of the principle of detecting the position of wafer alignment and turns using the optical correlation filter. FIG. Relationship diagram between the image of the correlated light peak for 9 turns and its detection signal in the X direction and the X direction, 1st
Figure 7 is a configuration diagram showing an example of the conventional TTL alignment method, Figure 18 (a) is a relationship diagram between the reticle alignment turn and wafer alignment nozzle in the conventional method, and Figure 18 (b) is a diagram showing the relationship between the reticle alignment turn and the wafer alignment nozzle in the conventional method. Figure 19 (C) is a diagram showing the relationship between the detection signals of the reticle alignment pattern and the wafer alignment pattern.
Figure (a) shows multiple interference within the resist, Figure (b)
) is a waveform diagram of a detection signal with reduced contrast of wafer alignment/toha turn. 1... Reticle, 2... Reduction projection lens, 3...
Wafer, 9...Movable slit, 10...Photomultiplier tube, 14.14', 61.68...Wafer alignment mother turn, 20...Complex Fourier transform image of 4 turns of wafer alignment , 26.70... Optical correlation filter, 29.71... Correlation light peak for wafer alignment/turn, 37... Mirano/turn, 38.38', 73... Reticle alignment/turn , 45 -- Rede oscillator. Agent Patent Attorney Tadashi Akimoto Jitsutsu 2 Figure 3 Figure 4. Figure 5 Figure 7 Figure 8 Figure 9 Figure I2 Figure I H
Fig. 13 Fig. 14 Fig. te Q Fig. 17 Fig. Ivr I scale

Claims (1)

[Claims] 1. In a semiconductor exposure apparatus that exposes a circuit pattern on a reticle onto a wafer to form the same circuit pattern as the circuit pattern on the wafer, an alignment pattern on the wafer is illuminated with coherent light. , an optical correlation filter that optically performs complex Fourier transform on the complex amplitude distribution of the reflected light from the alignment pattern, and has a complex amplitude distribution that is conjugate to the complex amplitude distribution obtained by the complex Fourier transform on the complex Fourier transform surface. is installed to optically superimpose the complex amplitude distribution and the complex amplitude distribution of the reflected light, optically perform complex Fourier inverse transform on the complex amplitude distribution obtained by the superposition, and transform the complex Fourier inverse transform surface. An alignment method for a semiconductor exposure apparatus, characterized in that the position of the alignment pattern on the wafer is detected based on the maximum amplitude position in the complex amplitude distribution obtained by the above method, and the alignment pattern on the wafer and the alignment pattern on the reticle are aligned. 2. The alignment method for a semiconductor exposure apparatus according to claim 1, wherein the alignment pattern on the wafer is a two-dimensional pattern spread in two-dimensional directions.