JPH08306602A - Alignment equipment - Google Patents

Abstract

PURPOSE: To accurately detect the position of a wafer mark by reducing the influence of asymmetry of the wafer mark without accurately conforming a step-difference of the wafer mark to a special optical path length. CONSTITUTION: Laser light sources 11A-11C generate laser beams different in wavelength. Two laser light sources are so selected from the laser light sources 11A-11C and driven for lighting that the difference of wavelength becomes a specified value corresponding to the step-difference of a wafer mark. Laser beams from the two selected laser light sources are synthesized on the same axis. The synthesized luminous flux B is converted into two alignment lights having frequency difference via acoustooptic elements 14, 18, and the wafer mark is irradiated with these alignment lights through a lens 24, a beam spliter 26, etc.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an alignment device provided in an exposure device for transferring a pattern formed on a mask onto a photosensitive substrate and for aligning the mask and the photosensitive substrate.

[0002]

2. Description of the Related Art A projection exposure apparatus used for manufacturing a semiconductor element, a liquid crystal display element or the like is used as a mask especially when forming a circuit pattern on a second layer or later on a wafer (or a glass plate). An alignment device for aligning the reticle and the wafer with high precision is provided. Such an alignment apparatus includes an alignment sensor that detects the position of an alignment mark (wafer mark) on the wafer, and a control system that moves the wafer to a target movement position based on the position detected by the alignment sensor. There is.

In general, the degree of roughness of the wafer surface changes due to exposure and subsequent processes, and the step between the wafer mark and the peripheral underlying layer may differ depending on the layer (layer) on the wafer. It is difficult to accurately detect the positions of all wafer marks with the alignment sensor. Therefore, the following alignment sensors are used according to the application.

LIA (Laser Interferometric Alignm)
ent) method: This is a method of irradiating a diffraction grating-shaped wafer mark with laser light with a slightly different frequency from two directions, causing two generated diffracted lights to interfere with each other, and determining the position information of the wafer mark from the phase of the interference light. Is a sensor for detecting. This LI
The method A is effective for a wafer mark having a low step or a wafer having a large surface roughness.

FIA (Field Image Alignment) method: This is a sensor for image-processing the position of an image of a wafer mark illuminated by light having a wide wavelength band width such as a halogen lamp as a light source, and measuring the position of an aluminum layer or This is effective for measuring asymmetric marks on the wafer surface. LSA (laser step alignment) method: This is a system that irradiates a laser beam on a wafer mark and measures the position of the wafer mark by using the diffracted and scattered light. It is widely used.

Conventionally, such various alignment sensors have been used properly according to the application. Further, as an optical system of the alignment sensor, a TTR (through-the-reticle) method for aligning a reticle and a wafer through a projection optical system, and a position of a wafer mark is detected through the projection optical system. A TTL (through-the-lens) system or an off-axis system for detecting the position of a wafer mark without using a projection optical system is known, and various alignment sensors described above are combined with each of these optical systems. be able to.

For example, when an LIA type alignment sensor is used in the TTR type, a predetermined chromatic aberration occurs in the projection optical system with respect to the alignment light because the wavelength of the alignment light is different from the exposure wavelength. Therefore, for example, in Japanese Unexamined Patent Publication No. 5-160001, a chromatic aberration control member such as a phase grating is arranged at a position where alignment light passes near the pupil plane of the projection optical system, and the chromatic aberration generated by the projection optical system is corrected by this chromatic aberration control member. An alignment device adapted to correct is disclosed. In this apparatus, especially the axial chromatic aberration is corrected so that the reticle and the wafer are conjugated.

Generally, the wafer mark is a pattern of concavities and convexities (reflection type phase grating) formed on the surface of the wafer. For example, regarding a diffraction grating wafer mark to be detected by an LIA type alignment sensor, When the wavelength of the alignment light is λ, the effective optical path length of the step between the convex portion and the concave portion of the wafer mark is (λ / 4 + mλ / 2)
It is known that when (m is an integer), the diffraction efficiency is highest and a beat signal with a high SN ratio can be obtained. Similarly, in an FIA type alignment sensor or the like,
There is an optimum optical path length corresponding to the wavelength of the alignment light for the unevenness of the unevenness of the wafer mark.

[0009]

As described above, the step of the wafer mark to be detected by the alignment sensor is optimum for obtaining a detection signal with a high SN ratio according to the wavelength of the alignment light used. There is an optical path length. However, in practice, when the wavelength of the alignment light is fixed, it is difficult to form the wafer mark so that the step difference is optimized according to the wavelength. Further, especially in the peripheral portion of the wafer or the like, the concave portion of the wafer mark may be inclined to form an asymmetrical mark. Further, the steps of the wafer mark may change or the recesses of the wafer mark may become asymmetric during various processes. Further, as the upper layer is formed on the wafer, the level difference of the wafer mark tends to gradually deviate from the optimum value or the concave portion tends to be inclined.

The inconvenience when the step of the wafer mark deviates from the optimum optical path length in this way will be considered. First, FIG. 8A shows a state in which a photoresist 67 having a refractive index n is coated on a wafer mark 66 formed on the wafer 6 symmetrically with respect to the measurement direction. In the above, it is assumed that the convex portion and the concave portion of the wafer mark 66 as the reflection type phase grating have the same reflectance. Further, assuming that the step between the convex portion and the concave portion of the wafer mark is h, as described above, when the effective optical path length s (= n · h) of the step h satisfies the following equation, the diffraction of the alignment light having the wavelength λ is performed. Maximum efficiency.

2s = λ / 2 + mλ (m: integer) (1) However, when the amplitude reflectance of the convex portion of the wafer mark 66 is different from the amplitude reflectance of the concave portion (bottom portion), the effect in the equation (1) is obtained. The target optical path length s is represented by a value obtained by multiplying the phase difference between the phase of the alignment light at the convex portion and the phase of the alignment light at the concave portion by λ / (2π). When the expression (1) is satisfied in the wafer mark 66 of FIG. 8A, the intensity distribution of the reflected light (or diffracted light) from the wafer mark 66 becomes as shown in FIG. 9A. There is no asymmetry and good measurement is performed. Further, even if the effective optical path length s deviates from the equation (1) to some extent, the position is accurately detected if there is no asymmetry.

Next, as shown in FIG. 8B, the wafer 6
Consider a case where the concave portion (bottom portion) of the wafer mark 68 formed above is tilted in the measurement direction and the wafer mark 68 is asymmetric. In this case, a step at the center of the recess is used as the step h between the protrusion and the recess. And
Assuming that the equation (1) holds for the effective optical path length s (= n · h) of the step h, the wafer mark 6 in FIG.
The intensity of the reflected light from the areas 69A and 69B on the left and right of the center of the concave portion (bottom) of 8 is the same as the intensity of the reflected light from the center, and the intensity of the reflected light from the wafer mark 68 is as shown in FIG. As shown in FIG. 9 (b), there is no asymmetry as a whole. Further, although the intensity is smaller than that in the case of FIG. 9A on average, there is almost no problem in position detection.

On the other hand, when the effective optical path length s of the step h is slightly larger than the expression (1), the average optical path length in the region 69B on the right side of the recess of the wafer mark 68 approaches the expression (1) and the left side The average optical path length in the area 69A is (1)
Since it is out of the equation, the intensity of the reflected light from the region 69B becomes stronger than the intensity of the reflected light from the region 69A. Therefore,
The intensity distribution of the reflected light from the wafer mark 68 is shown in FIG.
As shown in (c), the barycentric position of the light amount is shifted to the right side compared to the barycentric position 70 in the case where the light amount distribution is symmetric, and is affected by asymmetry. On the other hand, in FIG. 8B, when the effective optical path length s of the step h is slightly smaller than that of the equation (1), the intensity distribution of the reflected light from the wafer mark 68 is as shown in FIG. 9D. The center-of-gravity position of is shifted to the left side of the original center-of-gravity position 70, and is also affected by asymmetry.

As described above, when the recessed portion of the wafer mark is inclined and the wafer mark is asymmetrical, when the effective optical path length of the step at the center of the recessed portion satisfies the expression (1), it is almost the same. It is not affected by asymmetry, but it can be seen that the effect of asymmetry appears when the effective optical path length at the center of the recess deviates from the equation (1). Incidentally, when the wafer mark is formed by etching an aluminum layer deposited on the wafer or the like, such inclination of the wafer mark in the concave portion (bottom portion) is often seen in the peripheral portion of the wafer.

In this regard, it is possible to provide a special step so that the step of the wafer mark satisfies the expression (1), but then, the throughput of the exposure step (productivity).
And the manufacturing cost increase. In view of the above problems, the present invention reduces the influence of asymmetry of the wafer mark to accurately detect the position of the wafer mark and accurately detects the wafer without adjusting the step of the wafer mark to a certain optical path length. An object of the present invention is to provide an alignment device capable of highly accurate alignment.

[0016]

A first alignment apparatus according to the present invention is provided in an exposure apparatus for transferring a mask pattern (4) onto a photosensitive substrate (6) as shown in FIGS. 1 and 5, for example. An alignment device that aligns the mask pattern (4) with the photosensitive substrate (6) based on the position of the alignment mark (48A) formed of the uneven pattern formed on the photosensitive substrate (6). Of an irradiation optical system (11A to 11C, 13A, 1) for irradiating the use mark (48A) with light of a plurality of different wavelengths
3B, 14 to 26, 2) and the alignment mark (48
A) A light receiving optical system (2, 26-) that receives a light beam from A) and generates a detection signal corresponding to the position of the alignment mark.
28, 31) and a wavelength control means (61, 64) for setting the wavelength difference of the light of a plurality of wavelengths irradiated to the alignment mark from the irradiation optical system according to the step of the alignment mark. , With.

In this case, one example of the irradiation optical system is provided with light sources (11A to 11C) which generate light of a plurality of wavelengths in an arbitrary combination selected from three or more kinds of wavelengths different from each other. , At this time, the wavelength control means
It is desirable to indicate a combination of wavelengths selected from these three or more kinds of wavelengths in the irradiation optical system according to the step of the alignment mark.

Another example of the irradiation optical system is provided with a plurality of light sources for generating lights having different and variable wavelengths. At this time, the wavelength control means controls the position of the alignment mark. It is desirable to indicate each wavelength of a plurality of wavelength-variable lights generated from a light source in the irradiation optical system according to the step. In addition, an average (meaning “geometrical mean” in the present invention) wavelength of light of two predetermined wavelengths out of a plurality of wavelengths of light irradiated to the alignment mark is λ, and a wavelength difference is Δλ, Assuming that the effective optical path length of the step of the alignment mark is s, a positive integer m and an integer n of any of 2 to 4 are used so that the wavelength difference substantially satisfies the following relationship. It is desirable to set.

Δλ = λ ² (n / 6 + m) / (2s) (2) In other words, this means setting the wavelength difference Δλ in the vicinity of any of the following. Δλ = λ ² (1/3 + m) / (2s) (3A) Δλ = λ ² (2/3 + m) / (2s) (3B) Δλ = λ ² (1/2 + m) / (2s) (3C)

The second alignment apparatus of the present invention, as shown in FIGS. 1 and 5, for example, has a plurality of different alignment marks with respect to the alignment mark in the same premise as the first alignment apparatus described above. Irradiation optical system (11A to 11C, 13A, 13B, 14 to 2 for irradiating light of wavelength
6, 2) and a light receiving optical system (2, which receives light of the plurality of wavelengths from the alignment mark and generates a detection signal corresponding to the position of the alignment mark for each wavelength.
26 to 28, 31) and the level control means (61, 6) for adjusting the level ratio of the detection signal for each wavelength output from the light receiving optical system according to the step of the alignment mark.
4) and.

[0021]

The principle of the present invention will be described. First, the alignment mark (48A) on the photosensitive substrate is a reflection type phase formed at a predetermined pitch in the measurement direction (X direction), as shown in the cross-sectional view of FIG. 4B or 4C. Can be treated as a grid. Alignment mark (48
In the case where the concave portion (bottom portion) of A) is flat (FIG. 4B), the average value of the step difference between the convex portion and the concave portion is taken as step difference h, and the concave portion is inclined in the measurement direction (FIG. 4). (C))
The average value of the steps at the center of the recess is set as the step h. Further, the alignment mark (48A) is covered with the photosensitive material (56) having a refractive index n, and the wavelength range of the light flux for alignment is set to the wavelength range which transmits the photosensitive material (56).

If monochromatic light of wavelength λ is used as the light beam for alignment, the effective optical path length s (= n · h) of the step h of the alignment mark (48A) is (1). It is desirable to satisfy the formula, but in practice it is difficult. Therefore, in the first alignment apparatus of the present invention, it is considered that the wavelength of the light beam for alignment is changed to substantially satisfy the condition of the expression (1).

FIG. 6 shows the relationship between the effective optical path length s (= n · h) of the step h of the alignment mark (48A) and the diffraction efficiency. The horizontal axis y in FIG. 6 indicates the effective optical path. The length s is converted into a phase difference, and the following relationships are established. y = 4πs / λ = 4πn · h / λ (4) Further, the vertical axis of FIG. 6 shows the value of the diffraction efficiency f (y) at the phase difference y, and the diffraction efficiency f (y) is expressed by the following equation. It

F (y) = (1 + cos y) / 2 (5) Point A in FIG. 6 shows a case where the condition of the expression (1) is satisfied. Here, when the step h increases, the diffraction efficiency moves to the point B, and when the step h decreases, the diffraction efficiency moves to the point C.
In either case, the diffraction efficiency f (y) decreases. On the contrary, for example, even if the alignment mark (48A) is asymmetric as shown in FIG. 4C, the average step h is (1).
When the condition of the formula is satisfied, the position of the center of gravity of the distribution of the diffraction efficiency is maintained at substantially the same position as when the alignment mark (48A) is symmetrical, and it is considered that there is almost no asymmetric effect.

Therefore, next, assuming that the light beam for alignment includes the monochromatic light of _two wavelengths λ ₁ and λ ₂ , the wavelength λ ₁
And the phase difference y at λ ₂ is the phase difference between points B and C in FIG. 6, respectively. In this case, when the alignment mark (48A) is symmetrical as shown in FIG. 4 (b) and the step h (average value) becomes large, the points B and C are respectively points B ′ and C to the right. As a result, the overall diffraction efficiency does not decrease. Similarly, step h (average value)
However, since the points B and C both move leftward, the overall diffraction efficiency does not decrease. Therefore,
It can be seen that the use of light of a plurality of wavelengths can reduce the decrease in the detection signal when the step h varies.

Similarly, the alignment mark (48A) is asymmetric as shown in FIG. 4 (c), and the step h with respect to the average wavelength λ (= (λ ₁ λ ₂ ) ^1/2 ) of the geometric mean. Assuming that the (average value) satisfies the expression (1), the direction of shift of the light amount centroid due to asymmetry at the wavelength λ ₁ is opposite to the direction of shift of the light amount centroid due to asymmetry at the wavelength λ _2. . Therefore, the effect of asymmetry is also reduced by using two wavelengths.

Now, let us consider the condition of the interval between the two wavelengths. First, when the step h changes, the diffraction efficiency changes in the opposite directions at the two wavelengths, so that the phase difference y on both sides of the point A where the diffraction efficiency f (y) has a peak in FIG.
Need to choose those two wavelengths so that Next, the influence of the asymmetry of the alignment mark (48A) is
For example, it depends on the difference in diffraction efficiency at both ends of the inclined concave portion (bottom portion) (58) in FIG.
The slope of the curve indicating (y) (differential with respect to the phase difference y) f ′
Proportional to (y). Further, even if the difference in the diffraction efficiency due to the inclination is large, the influence may be reduced if the average diffraction efficiency is small. After all, the effect of asymmetry is due to the diffraction efficiency f (y)
Slope of the curve f '(y) and the average diffraction efficiency f
It is considered to be given by the function g (y) of the product with (y). The function g (y) is given by the following equation.

G (y) = f (y) f '(y) (6) FIG. 7 shows a change of the function g (y) with respect to the phase difference y,
In FIG. 7, the dotted line indicates the value of the function f (y) / 2. The function g (y) becomes 0 when the following relation holds. y = mπ (m: integer) (7) That is, even if the concave portion (bottom portion) of the alignment mark has an inclination at this time, it is not affected by asymmetry. Also, the function g
(Y) is an extreme value ± (3/1 when the following relation holds.
6) Take 3 ^1/2 .

Y = 2mπ ± π / 3 (m: integer) (8) In this case, as shown in FIG. 7, the function g (x) has negative and positive extreme values at points B and C, respectively. Then, the phase difference y at the point B is (2mπ + π / 3), and the phase difference y at the point C is (2mπ−π / 3). Further, the phase difference y of the point D having a negative extreme value on the left side of the point C is {2 (m-1) π +
π / 3}, that is, (2mπ-5π / 3). Therefore,
The interval of the phase difference between two adjacent points where the function g (x) takes an extreme value is 2π / 3 or 4π / 3.

In other words, the wavelengths λ ₁ and λ at the points B and C on both sides of the point A where the diffraction efficiency f (y) has a peak in FIG.
_{If 2} is set so that the interval on the y-axis is 2π / 3 or 4π / 3 in FIG. 7, the values of the function g (y) at points B and C in FIG. Is the opposite value, and the sum is 0. That is, even if the steps change and the points B and C shift to the left and right across the peak of the diffraction efficiency, the function g (y) at the points B and C
Since the sum of the values of is almost 0, it is not affected by asymmetry. When the interval is 4π / 3, the point B is located at the point D.

In this case, point B is obtained from equations (4) and (8).
The following relationships are respectively satisfied the phase difference y ₂ at a wavelength lambda ₂ in the phase difference y _1, and point C at a wavelength lambda ₁ in. y ₁ = 4πs / λ ₁ = 2mπ ± π / 3 (9A) y ₂ = 4πs / λ ₂ = 2mπ ± π / 3 (9B) Therefore, by subtracting the expression (9B) from the expression (9A), Either relationship is established.

4πs (1 / λ ₁ −1 / λ ₂ ) = 2mπ + 2π / 3 (10A) 4πs (1 / λ ₁ −1 / λ ₂ ) = 2mπ + 4π / 3 (10B) Next, two wavelengths λ ₁ and λ ₂ difference Δλ and average wavelength λ
Is defined as follows. Δλ = λ ₂ −λ ₁ (11A) λ = (λ ₁ λ ₂ ) ^1/2 (11B) Then, by substituting the expressions (11A) and (11B) into the expression (9A), the expression (3A) is obtained. The following equation corresponding to is obtained.

Δλ = λ ² (1/3 + m) / (2s) (12A) Similarly, by substituting the equations (11A) and (11B) into the equation (10B), the following equation corresponding to the equation (3B) is obtained. The formula is obtained. Δλ = λ ² (2/3 + m) / (2s) (12B) If the difference between the two wavelengths is set to the value of the expression (12A) or the expression (12B), the sum of the two wavelengths will be obtained even if the step is changed a little. It is possible to avoid the influence of asymmetry due to the inclination of the concave portion of the alignment mark.

If it is troublesome to use the equation (12A) and the equation (12B) separately, the following (12A)
Even if the wavelength difference Δλ of the average value between the equation and the equation (12B) is used, there is no great inconvenience although it is out of the optimum value. In the following formula, the fact that the average value of 1/3 and 2/3 is 1/2 is used, and the formula below corresponds to formula (3C).

Δλ = λ ² (1/2 + m) / (2s) (12C) By the way, in the above-mentioned invention, the light intensities of the respective wavelengths need to be equal in magnitude. This light intensity is a value observed in the photoelectric conversion unit of the light receiving optical system, assuming that the diffraction efficiency is equal at each wavelength. Further, the above description has been conceptually described using the magnitude of diffraction efficiency and the intensity distribution of the mark image. However, the influence of such asymmetry will appear as a phase error in the phase detection due to the diffracted light interference of the LIA method or the like, and therefore the image processing type and the diffracted light interference type (LIA method or the like) can be similarly applied.

Next, in the second alignment apparatus of the present invention, instead of actually changing the wavelength difference Δλ in such a manner, the light intensity of the light flux of a plurality of wavelengths or the level of the detection signal of the light flux of a plurality of wavelengths. Is changing. By changing the light intensity or the level of the detection signal in this way, it is possible to reduce the influence of a step change or asymmetry as well as a wavelength difference change.

[0037]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the alignment apparatus according to the present invention will be described below with reference to FIGS. The present embodiment is a projection exposure apparatus that projects a pattern on a reticle onto each shot area on a wafer through a projection optical system.
The present invention is applied to the case of using a TTR type and LIA type alignment sensor.

FIG. 1 shows an overall schematic structure of the projection exposure apparatus of this example. In FIG. 1, the exposure illumination light of wavelength λ ₀ from the exposure illumination system 60 passes through the dichroic mirror 3 during exposure. The reticle 4 is irradiated, and under the illumination light, the pattern of the reticle 4 is reduced to, for example, 1/5 through the projection optical system 5 and projected onto each shot area of the wafer 6 coated with the photoresist. Here, the Z axis is taken parallel to the optical axis AX of the projection optical system 5, the Y axis is parallel to the plane of FIG. 1 on the plane perpendicular to the Z axis, and the X axis is perpendicular to the plane of FIG.
Take the axis.

The reticle 4 is held on the reticle stage 9, and the wafer 6 is placed on the wafer stage composed of the X stage 8X and the Y stage 8Y via the wafer holder 7. Actually, on the X stage 8X,
A Z stage or the like for positioning the wafer 6 in the Z direction is mounted. The reticle stage 9 and the wafer stage respectively position the reticle 4 and the wafer 6 on a plane perpendicular to the optical axis of the projection optical system 5. The two-dimensional coordinates of the reticle stage 9 and the wafer stage are respectively detected by an interferometer (not shown), and the detection results are supplied to the main control system 61.
2 and the wafer stage control system 63 to control the operations of the reticle stage 9 and the wafer stage, respectively. Further, the reticle 4 is provided above the periphery of the reticle 4.
Reticle alignment microscopes 39 and 40 (see FIG. 2) for aligning the center of the optical axis with the optical axis AX of the projection optical system 5 are arranged.

Next, the LIA type alignment sensor of this example will be described in detail. The alignment sensor includes an objective lens 2 above the dichroic mirror 3, an alignment optical system 1 above the objective lens 2, an alignment power supply system 64, and an alignment signal processing system 65. When performing the alignment, the main control system 61 uses the alignment power supply system 64 to perform the alignment optical system 1
Two laser light sources selected from among three laser light sources described later therein are caused to emit light, respectively.

The laser beams emitted from these laser light sources undergo predetermined frequency modulation in the alignment optical system 1 and are emitted as alignment light. This alignment light passes through the objective lens 2 and the dichroic mirror 3 and the reticle. Reticle mark 35 in the form of diffraction grating on 4
A and the light transmissive window portion (reticle window) 37A are irradiated, and the alignment light transmitted through the reticle window 37A is irradiated on the wafer mark 48A attached to the shot area of the positioning target on the wafer 6. Here, the measurement direction of the reticle mark 35A and the wafer mark 48A is the X direction.

Then, the heterodyne beam generated by the diffraction at the wafer mark 48A and the reticle mark 35.
The heterodyne beam generated by the diffraction at A returns to the alignment optical system 1 via the dichroic mirror 3 and the objective lens 2, and two light receiving systems in the alignment optical system 1 generate two beat signals. These beat signals are supplied to the alignment signal processing system 65, where the phase difference between the two beat signals is detected, and the detected phase difference is supplied to the main control system 61. The main control system 61 performs final alignment based on the detected phase difference.

Next, referring to FIGS. 1 to 4, the LI of this example
An optical path of alignment light from the A-type alignment sensor and a method of detecting a reticle mark and a wafer mark will be described. In FIG. 1, the alignment optical system 1
Respectively mean wavelength of the first and second laser beams emitted from the two laser light sources selected by inner lambda ₁ and lambda
_{Assume 2} . That is, from the alignment optical system 1, a pair of reticle alignment illumination light RB having an average wavelength λ ₁ different from the exposure wavelength λ ₀ and a frequency difference Δf (50 kHz in this example).
₁ , RB ₂ , and wafer alignment illumination light WB ₁ , W
B ₂ and a pair of reticle alignment illumination lights RB ₃ and RB _{4 having} an average wavelength λ ₂ different from the exposure wavelength λ ₀ and an average wavelength λ ₂ close to the wavelength λ ₁ and a frequency difference Δf (= 50 kHz), and wafer alignment illumination lights WB ₃ and WB. ₄ and shall be ejected.

FIG. 2 is a side view of FIG. 1 viewed in the Y direction. As shown in FIG. 2, the reticle alignment illumination lights RB ₁ and RB ₂ are focused on the reticle 4 by the objective lens 2. The diffraction grating-shaped reticle mark 35A on the lower surface of the reticle 4 is irradiated with incident angles −θ _R1 and θ _R1 . FIG. 3 shows the reticle mark 35A of the reticle 4 of this example.
3 is an enlarged view of the periphery of the reticle mark 35A in FIG. 3, and the reticle mark 35A for the X axis is a mark made of a diffraction grating formed at a pitch P _R in the X direction.
A reticle window 37A for passing alignment light traveling toward the wafer is formed inside A. The Y-axis reticle mark and reticle window are formed by rotating FIG. 3 by 90 °. And the reticle mark 35
A is irradiated with a light flux 50 composed of illumination lights RB ₁ , RB ₂ , RB ₃ , and RB ₄ , and illuminates the reticle window 37A with illumination light WB ₁ ,
A light beam 51 composed of WB ₂ , WB ₃ and WB ₄ is passing through.

Returning to FIG. 2, the incident angles −θ _R1 , θ _R1 and the grating pitch P _R of the reticle mark 35A have the following relationship, and the + _1st-order diffracted light RB ₁ ⁺¹ and the illumination light RB of the illumination light RB ¹ are given.
₂ -1-order diffracted light RB ₂ ^-1 occurs directly above each
The alignment detection light (heterodyne beam) is returned to the alignment optical system 1 via the objective lens 2.

Sin (θ _R1 ) = λ ₁ / P _R (13) Similarly, the reticle alignment illumination lights RB ₃ and RB ₄ are also focused on the reticle 4 by the objective lens 2 and then onto the reticle mark 35 A on the reticle 4. Irradiation is performed at incident angles of −θ _R2 and θ _R2 . At this time, the + 1st order diffracted light R of the illumination light RB ₃
B ₃ ⁺¹ and −1st-order diffracted light RB ₄ ^{−1 of the} illumination light RB ₄ are generated directly above and return to the alignment optical system 1 via the objective lens 2.

On the other hand, wafer alignment illumination light WB₁,
WB₂Passes through the reticle window 37A on the reticle 4 and throws
The chromatic aberration control plate 10 in the shadow optical system 5 is reached. Chromatic aberration control
Illumination light WB of the plate 10₁, WB₂The part where
Each has a diffraction grating-shaped on-axis chromatic aberration control element.
Illumination light WB₁, WB₂Is the angle −θ_G1, Θ _G1
Only bent to pair with the diffraction grating wafer mark 48A.
Angle of incidence −θ _W1, Θ_W1Is illuminated by.

FIG. 4A shows an enlarged view of the wafer mark 48A. In FIG. 4A, the wafer mark 48A is shown.
A is composed of a concave and convex diffraction grating formed at a pitch P _W in the X direction. The wafer mark for the Y axis is the wafer mark 4
It is a shape obtained by rotating 8A by 90 °. Then, the wafer mark 48A is irradiated with the light flux 51 composed of the illumination lights WB ₁ , WB ₂ , WB ₃ and WB ₄ .

Returning to FIG. 2, the incident angles −θ _W1 and θ _W1 and the grating pitch P _W of the wafer mark 48A have the following relationship.
The + _1st-order diffracted light WB ₁ ⁺¹ of the illumination light WB ¹ and the -1st-order diffracted light WB ₂ ^{-1 of the} illumination light WB ₂ are generated directly above, respectively, and these two diffracted lights become alignment detection light (heterodyne beam). .

Sin (θ _W1 ) = λ ₁ / P _W (14) At this time, since the wafer alignment illumination lights WB ₃ and WB ₄ have wavelengths close to those of the illumination lights WB ₁ and WB ₂ , on the chromatic aberration control plate 10. The passing positions are approximately the illumination light WB ₁ ,
It can be regarded as on the axial chromatic aberration control element through which WB ₂ passes. Therefore, the illumination lights WB ₃ and WB ₄ have the angles −θ _G2 and
The wafer mark 48A is bent by θ _G2 and is irradiated at incident angles −θ _W2 and θ _W2 , respectively. And the illumination light W
The + 1st-order diffracted light WB ₃ ^{+1 of} B ₃ and the -1st-order diffracted light WB ₄ ^{-1 of the} illumination light WB ₄ are generated directly above to become alignment detection light.

In this case, as shown in FIG. 1, the wafer alignment illumination light is changed by the deflection action of the chromatic aberration control plate 10.
Angle θ with respect to the wafer 6 in the non-measurement direction (Y direction)
_{Since the} light is incident at an angle of _m , the position at which each of the alignment detection lights passes on the chromatic aberration control plate 10 is different from the position at which the light is incident. The alignment detection light from the wafer mark 48A passes through another axial chromatic aberration control element on the chromatic aberration control plate 10 to correct lateral chromatic aberration,
Head to reticle window 37A. Thereafter, each detection light returns to the alignment optical system 1 again via the reticle window 37A and the objective lens 2. Further, the wafer alignment illumination light illuminates a position shifted by Δβ in the Y direction on the surface of the wafer 6 as compared with the case where the chromatic aberration control plate 10 is not arranged.

The alignment optical system 1 will be described in detail with reference to FIG. 5B is a view of the alignment optical system 1 viewed from the same direction as FIG. 1, and FIG.
Shows the alignment optical system 1 viewed from the same direction as in FIG.
FIG. 5 (c) is a bottom view of FIG. 5 (a). In FIG. 5, a first laser light source 11A, a second laser light source 11B, and a third laser light source 11C, each of which is a laser diode, are provided, and the laser light sources 11A, 11B, 1
The wavelengths of the laser beams emitted from 1C are set to 630 nm, 690 nm, and 780 nm, respectively. Then, the alignment power supply system 64 can independently drive any one or two laser light sources of the three laser light sources 11A to 11C or three laser light sources with a desired intensity. ing. The predetermined laser light source among the three laser light sources may be, for example, a He-Ne laser light source, a He-Cd laser light source, an Ar ion laser light source, or the like, and instead of the three laser light sources, for example, two laser light sources may be used. A tunable laser source may be used, 3
Instead of one laser light source, four or more laser light sources each having a different oscillation wavelength may be used. Further, when a plurality of laser light sources selected from three or more laser light sources are caused to emit light, a shutter may be used to block the laser beams of the laser light sources not selected by the shutter.

In this case, the laser beam having the wavelength λ ₁ emitted from the first laser light source 11 and the laser beam having the wavelength λ ₂ emitted from the second laser light source 12 are coaxially combined by the dichroic mirror 13A. The two laser beams thus coaxially combined and the laser beam having the wavelength λ ₃ emitted from the third laser light source are coaxially combined by the dichroic mirror 13B to become the illumination light B. Hereinafter, the illumination light B will be described as a combination of laser beams having wavelengths λ ₁ and λ ₂ , but the same applies even if the combination of the wavelengths of the laser beams forming the illumination light B is changed.

The illumination light B is incident on the acousto-optic device (hereinafter referred to as "AOM") 14 driven at the frequency F ₁ . The AOM 14 makes light incident vertically to obtain ± 1st order diffracted light by Raman-nas diffraction, that is, Raman
It is an eggplant type AOM. Since the two laser beams in the illumination light B have different wavelengths, diffracted light is generated by the AOM 14 at different angles. These + 1st-order and -1st-order diffracted lights have a frequency difference of + F ₁ and -F ₁ with respect to the original frequency, respectively.

The laser beam emitted from the AOM 14 is
After passing through the lens 15 and entering the spatial filter 16,
Only the ± 1st order diffracted light is selected by the filter 16 and selected.
The selected diffracted light passes through the lens 17 to the AOM 14.
And are rotated by 45 ° and arranged at frequency F₂Driven by
It enters so as to cross the AOM 18 that is present. AOM1
8 is a Bragg type AOM, and the incident angle is 1 of the Bragg angle.
/ Sin is set to 45 °. Therefore, the optical axis
The angle of incidence of the diffracted light is
It is a lag angle. In this case, AOM14 + F₁Frequency
Number-modulated laser beam is -1st-order diffracted by AOM18
Light is strongly generated, and the -1st-order diffracted light is -F at AOM18.
₂Is emitted from the laser light sources 11 and 12 after receiving the frequency modulation of
+ (F₁-F₂) Frequency modulation
Beam alignment beam B₁, B ₃become. Similarly, A
OM14 -F₁The laser beam that has undergone the frequency modulation of
The + 1st order diffracted light is strongly generated by the AOM 18, and the + 1st order
Origami is AOM18 -F₂Laser with frequency modulation
When emitted from the light sources 11 and 12,-(F ₁−
F₂) Alignment beam B subjected to frequency modulation of₂,
B_Fourbecome. As a result, the alignment beam B₁, B₃
And alignment beam B₂, B_FourAnd the frequency difference Δf is 2
(F₁-F₂), Δf is 50 kHz in this example
Shall be set to. Laser passed through AOM18
The beam enters the spatial filter 20 through the lens 19.
The alignment beam selected by the spatial filter 20
B₁~ B _FourIs led to the latter stage.

Alignment beam B₁~ B_FourIs lens 2
1 is focused on the field stop 22 by the reticle or window.
After the beam shape on the roof is determined, the reticle wafer
Reticle alignment by the Ha-beam separation prism 23
Illumination light RB₁~ RB_Four, And wafer alignment lighting
Light WB₁~ WB_FourIt is divided into Then alignment
The illumination light reaches the direct-view prism 25 through the lens 24.
It The direct-view prism 25 is rotatable about the optical axis,
Instruction from the main control system 61 of FIG. 1 by a motor not shown
Driven by. The wavelength when the direct-view prism 25 rotates
λ₁, Λ₂The relative angle of the two colors of illumination light changes and the wavelength λ
₁Illumination light RB₁, WB₁, RB₂, WB ₂Against
Wavelength λ₂Illumination light RB₃, WB₃, RB_Four, WB
_FourAre separated. Illumination light with a relative angle changed in this way
Goes through the beam splitter 26 to the objective lens 2 in FIG.
Go to Due to the change in the relative angle, the wavelength λ₁, Λ₂2 colors
Relative relationship of the illumination position of the illumination light on the reticle and the wafer
The staff also changes.

On the other hand, the alignment detection light from the reticle mark 35A and the wafer mark 48A in FIG. 2 returns to the alignment optical system 1 in FIG.
The light is reflected by 6 and passes through a lens 27 to be separated into reticle detection light and wafer detection light by a detection light separation prism 28 located at a position conjugate with the reticle and the wafer. Reticle detection lights RB ₁ ⁺¹ and RB ₂ ^-1 and RB ₃ ⁺¹ and RB ₄
^-1 passes through the detection light separation prism 28 and is received by the photoelectric detection element 30. Wafer detection light WB ₁ ⁺¹ and WB ₂
^-1 and WB ₃ ⁺¹ and WB ₄ ^-1 are detection light separation prisms 28.
It is reflected by and is received by the photoelectric detection element 31.
The photoelectric detection element 30 outputs a reticle beat signal S _R corresponding to the position of the reticle mark, and the photoelectric detection element 31
To the wafer beat signal S corresponding to the position of the wafer mark
_W is output.

The reticle beat signal S _R is the detection light R
B ₁ ^+1, RB ₂ ^-1, and RB ₃ ^+1, RB ₄ ^{- 1} by a sinusoidal beat signal frequency Delta] f, wafer beat signal S _W is detected light WB ₁ ^+1, WB ₂ ^-1 and WB ₃ ^{+ 1} , WB
It is a sinusoidal beat signal having a frequency Δf of ₄ ⁻¹ .
The phase difference Δφ [rad] between the two changes depending on the relative movement amount of the reticle 4 and the wafer 6 in the X direction, and the relative movement amount Δx is as shown in the following equation.

Δx (on reticle) = P_R・ Δφ / (4π) (15A) Δx (on wafer) = P_W.DELTA..phi ./ (4.pi.) (15B) Note that the Y-direction alignment sensor
Beat signal corresponding to reticle mark and wafer
can get. Reticle beat signal S_R And wafer beat signal
No. S_W Signal strength (oscillation) by amplifiers 52 and 53, respectively.
Width) is adjusted, analog / digital (A / D) converter
It is supplied to the alignment signal processing system 65 via 54 and 55.
Where both signals S_R, S _WIs detected. Inspection
The output phase difference is supplied to the main control system 61 of FIG. main
The control system 61 uses both beat signals S_R, S_WBased on the phase difference of
When the reticle 4 and the wafer mark are misaligned,
The reticle 4 and the wafer 6 are adjusted so that the target tracking value is obtained.
Align. After that, the pattern image of reticle 4 is
Projection exposure is performed on the shot area of the stack 6.

Next, an example of the operation for changing the combination of the wavelengths of the laser beams emitted from the alignment sensor to the wafer mark according to the step of the wafer mark in this example will be described in detail. First, FIG.
It is an expanded sectional view of the wafer mark 48A of this example shown in (a). The wafer mark 48A is, for example, a concavo-convex phase pattern formed at a predetermined pitch in the X direction by etching on a metal film deposited on the wafer 6. And the wafer mark 4
A photoresist 56 having a refractive index n is applied on 8A. The wafer mark 48A in FIG. 4B is a recess (bottom).
Reference numeral 57 is a flat symmetrical mark, and the rough value of the average value h of the step between the convex portion and the concave portion 57 is obtained in advance by measurement or simulation, and is used as exposure data in the storage unit of the main control system 61 in FIG. Be stored in. The wafer mark 48A may be an asymmetrical mark in which the recess 58 is inclined with respect to the measurement direction as shown in FIG. 4C, but in this case, there is a step at the center of the recess 58. A rough value of the average value h is stored in the storage unit of the main control system 61.

The value of the refractive index n of the photoresist 56 is also stored in the storage section of the main control system 61.
1 obtains the effective optical path length s (= n · h) of the average value h of the steps of the wafer mark 48A. In this case, when the wavelength of the laser beam is λ and the effective optical path length s is converted into the phase difference y by the equation (4), the diffraction efficiency f (y) at the wafer mark 48A is represented by the equation (5), The diffraction efficiency f (y) changes as shown in FIG. 6 with respect to the phase difference y.

Next, the main control system 61, among the three laser beams of wavelength lambda ₁ to [lambda] ₃ emitted from the laser light source 11A~11C 5, the wavelength difference of the two wavelengths [Delta] [lambda], the average When the value is λ, two wavelengths that best fit the formula (3A) or the formula (3B) are selected. Specifically, the wavelengths λ ₁ , λ
₂ and λ ₃ are 630 nm, 690 nm and 780 nm respectively
Assuming that the average value h of the steps of the wafer mark 48A is 0.7 μm, the refractive index n of the photoresist is 1.6, and the average wavelength λ is 660 nm and the integer m is 0,
From formula (3A), Δλ = 65 nm, and from formula (3B), Δλ = 130 nm.

The wavelength difference Δλ is 65 nm or 1
As a method of selecting any of 30 nm, the average wavelength λ
Is close to the wavelength satisfying 2s = (m + 1/2) λ (m: integer), the formula (3A), that is, 65 nm is selected, and the average wavelength λ is close to the wavelength satisfying 2s = mλ (m: integer). Selects the expression (3B), that is, 130 nm. In FIG. 6, when the average wavelength λ corresponds to the point A, the wavelengths corresponding to the points B and C are taken as two wavelengths so that the expression (3A) is satisfied, and the average wavelength λ is When it corresponds to the vicinity of the phase difference y = π, it means that the wavelengths corresponding to the vicinity of the point C and the point D are taken as the two wavelengths so that the expression (3B) is satisfied. The point is to find a combination in which the slopes of the diffraction efficiency f (y) curves in FIG.

As a result, 65 nm is selected as the wavelength difference Δλ, and 630 n is selected as the two wavelengths among the wavelengths λ _{1 to} λ _3.
m and 690 nm are selected. Therefore, the main control system 61 of FIG. 1 simultaneously drives the first and second laser light sources 11A and 11B of FIG. 5 to light up via the alignment power supply system 64. As a result, even if the level difference of the wafer mark is slightly changed, the averaging effect does not affect the asymmetry. Next, the average value h of the steps of the wafer mark is, for example, 0.3 μm.
In the case of, the average wavelength is 700 nm and the wavelength difference Δλ =
170 nm. Therefore, 630 nm and 780 nm may be selected as the two wavelengths among the wavelengths λ _{1 to} λ ₃ .
In this way, the wavelengths of the two light sources that are turned on according to the level difference of the wafer mark may be selected.

The point described above is to search for a combination in which the slopes of the curves of the diffraction efficiency f (y) in FIG. 6 are opposite at two wavelengths. By the way, when there are many light sources of the alignment sensor, a rough curve of the diffraction efficiency f (y) shown in FIG. 6 can be known from those wavelengths. Using the result, it is possible to determine a combination of two wavelengths having opposite inclinations. In this case, it is not necessary to know the value of the step of the wafer mark. By the way, it is desirable to adjust the output of the plurality of wavelengths so that the levels of the photoelectric conversion signals in the photoelectric detection element 31 of FIG. 5 become the same value, assuming that the diffraction efficiency of the wafer mark is equal at each wavelength. .

In this embodiment, the wavelength difference determined by the formula (3A) or the formula (3B) was considered, but even if these average values are set, that is, the wavelength difference determined by the formula (3C) is set. Good. In this case, the wavelength difference deviates from the optimum value, but there is no great inconvenience. This gives (3
This is convenient because it is not necessary to use the expressions A) and (3B) separately.

Next, another embodiment of the present invention will be described.
It The projection exposure apparatus of this embodiment is the projection exposure apparatus of FIGS.
Similar to the device, the configuration of the alignment optical system is also shown
The same as 5. However, the alignment of FIG.
The power supply system 64 for power supply is, for example, three laser light sources 11A to 1A.
The function to turn on 1C one by one with desired drive power one by one
Have. In this embodiment, the three laser light sources 11
Oscillation wavelength λ of A to 11C ₁~ Λ₃From the wavelength difference Δλ
Either formula (3A), formula (3B), or formula (3C) is satisfied.
Two wavelengths (for example, wavelength λ₁
And λ₂) Is determined. After that, the main control system 61 of FIG.
5 (a) through the alignment power supply system 64
Wavelength λ₁And λ₂Laser light sources 11A, 11 corresponding to
B is turned on alternately.

In this case, the photoelectric detection element 31 shown in FIG.
From, based on the beat signal S _W incorporated alternately, two position shift amount by the alignment signal processing system 65 [Delta] x ₁ and [Delta] x ₂ is obtained. Further, assuming that the amplitudes of the beat signals S _W at the wavelengths λ ₁ and λ ₂ are standardized by a predetermined average value as W ₁ and W ₂ , respectively, the alignment signal processing system 65 of this example has two positional deviations. A value obtained by weighting and adding the amounts Δx ₁ and Δx ₂ in accordance with W ₁ and W ₂ , respectively, is defined as the positional deviation amount Δx. As a result, the diffraction efficiencies at the two wavelengths become substantially the same, and the influence of the asymmetry of the wafer mark is reduced.

In FIG. 5A, a plurality of photoelectric detecting elements 31 may be provided for each wavelength so that two laser light sources are turned on at the same time. Further, the value of the ratio of the drive powers of the two laser light sources may be controlled instead of weighting the _two positional deviation amounts Δx ₁ and Δx ₂ fetched alternately or in parallel. As a result, the amount of displacement can be accurately determined.

Although the alignment sensor of the LIA method is used as the alignment sensor in the above-described embodiment, the heterodyne interference method is used. However, by applying the present invention also in the case of the homodyne interference method, the same effect as that of the above-described embodiment is obtained. Is obtained. Further, the alignment apparatus of the present invention is not limited to the above-mentioned TTR-type and LIA-type alignment sensor, but an FIA-type that measures an alignment mark illuminated by light having a wide wavelength bandwidth using a halogen lamp as a light source. The present invention can also be applied to the case where the alignment sensor or the like is used.

As described above, the present invention is not limited to the above-mentioned embodiments, and various configurations can be adopted without departing from the gist of the present invention.

[0072]

According to the first alignment apparatus of the present invention, the wavelength difference of the light of a plurality of wavelengths irradiated to the alignment mark (wafer mark) is set according to the step of the alignment mark (wafer mark). Therefore, without accurately adjusting the step of the alignment mark to a particular optical path length,
There is an advantage that the influence of the asymmetry of the alignment mark can be reduced, the position of the alignment mark can be accurately detected, and the photosensitive substrate (wafer) can be aligned with high accuracy.

In this case, the irradiation optical systems are different from each other.
A light source that emits light of a plurality of wavelengths in an arbitrary combination selected from a plurality of types of wavelengths is provided, and the wavelength control means has three or more types in the irradiation optical system according to the step of the alignment mark. When instructing a combination of wavelengths selected from the wavelengths, it is possible to set the wavelength difference of a plurality of lights according to the step of the alignment mark with a simple configuration.

Further, the irradiation optical system is provided with a plurality of light sources which generate lights of different and variable wavelengths, and the wavelength control means controls the light source in the irradiation optical system according to the step of the alignment mark. When instructing the respective wavelengths of the plurality of wavelength-variable lights to be generated, there is an advantage that the wavelength difference of the plurality of lights can be set to a predetermined value with high accuracy. Further, when the wavelength difference between the light of two predetermined wavelengths among the light of a plurality of wavelengths which is irradiated to the alignment mark is Δλ, and the effective optical path length of the step of the alignment mark is s ,
When a positive integer m and an integer n of any of 2 to 4 are used, the effect of the asymmetry of the alignment mark can be minimized when the relationship of the expression (2) is substantially established.

Next, according to the second alignment apparatus of the present invention, the detection signals corresponding to the light of a plurality of wavelengths which are irradiated to the alignment mark (wafer mark) according to the step of the alignment mark (wafer mark). As the level ratio of the alignment mark is adjusted, the influence of the asymmetry of the alignment mark is reduced without accurately adjusting the step of the alignment mark to a certain optical path length. There is an advantage that the photosensitive substrate (wafer) can be accurately aligned by accurately detecting the position.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing an entire projection exposure apparatus used in an embodiment of the present invention.

2 is a side view of a stage system and an alignment optical system of the projection exposure apparatus of FIG. 1 viewed in a Y direction.

FIG. 3 is an enlarged plan view showing a reticle mark 35A and a reticle window 37A formed on the reticle of the embodiment.

FIG. 4 is an enlarged plan view showing a wafer mark 48A attached to a shot area on a wafer of an example.

5A is a diagram showing a configuration of an alignment optical system 1 and a signal processing system in FIG. 1, FIG. 5B is a side view of the optical system of FIG. 5A, and FIG. 3] is a bottom view of the optical system of FIG.

FIG. 6 is a diagram showing the relationship between the phase difference y of the step of the alignment mark (wafer mark) and the diffraction efficiency f (y).

FIG. 7 is a diagram showing the relationship between the phase difference y of the step of the alignment mark (wafer mark) and the function g (y) representing the effect of asymmetry.

FIG. 8A is an enlarged sectional view showing a symmetrical wafer mark, and FIG. 8B is an enlarged sectional view showing an asymmetrical wafer mark.

9 is a diagram showing an intensity distribution of reflected light (diffracted light) obtained corresponding to the wafer marks in FIGS. 8 (a) and 8 (b).

[Explanation of symbols]

 1 Alignment Optical System 2 Objective Lens 3 Dichroic Mirror 4 Reticle 5 Projection Optical System 6 Wafer 7 Wafer Holder 8X X Stage 8Y Y Stage 9 Reticle Stage 10 Chromatic Aberration Control Plate 11A-11C Laser Light Source 22 Field Stop 23 Reticle / Wafer Beam Separation Prism 26 Beam Splitter 30, 31 Photoelectric detection element 35A Reticle mark 39, 40 Reticle alignment microscope 48A Wafer mark 61 Main control system 64 Alignment power supply system 65 Alignment signal processing system

Claims (5)

[Claims] 1. An exposure device for transferring a mask pattern onto a photosensitive substrate, wherein the mask pattern and the photosensitive substrate are separated from each other based on the position of an alignment mark formed of an uneven pattern formed on the photosensitive substrate. In an alignment device that performs alignment, an irradiation optical system that irradiates the alignment mark with light having a plurality of different wavelengths, and a position of the alignment mark by receiving a light beam from the alignment mark. A light receiving optical system for generating a detection signal according to the above, and a wavelength control for setting the wavelength difference of light of a plurality of wavelengths which is irradiated from the irradiation optical system to the alignment mark according to the step of the alignment mark. And an alignment device. 2. The alignment apparatus according to claim 1, wherein the irradiation optical system includes a light source that emits light of a plurality of wavelengths in an arbitrary combination selected from three or more kinds of wavelengths different from each other, The wavelength control means indicates a combination of wavelengths selected from the three or more kinds of wavelengths in the irradiation optical system according to the step of the alignment mark. 3. The alignment apparatus according to claim 1, wherein the irradiation optical system includes a plurality of light sources that emit light having different and variable wavelengths, and the wavelength control unit includes the alignment mark. An alignment apparatus which indicates each wavelength of a plurality of wavelength-variable lights generated from a light source in the irradiation optical system according to the step. 4. The alignment apparatus according to claim 1, 2, or 3, wherein an average wavelength of light having a predetermined two wavelengths out of light having a plurality of wavelengths irradiated to the alignment mark is λ. , Δ is the wavelength difference
When λ is an effective optical path length of the step of the alignment mark, a positive integer m and an integer n of 2 to 4 are used, and Δλ = λ ² (n / An alignment device, wherein the wavelength difference is set so that the relationship of 6 + m) / (2s) is established. 5. An exposure device that transfers a mask pattern onto a photosensitive substrate is provided, and the mask pattern and the photosensitive substrate are separated from each other based on the position of an alignment mark formed of an uneven pattern formed on the photosensitive substrate. In an alignment device that performs alignment, an irradiation optical system that irradiates the alignment mark with light of a plurality of different wavelengths, and receives light of the plurality of wavelengths from the alignment mark for each wavelength. The level ratio of the detection signal for each wavelength output from the light receiving optical system according to the step of the alignment mark and the light receiving optical system that generates the detection signal according to the position of the alignment mark for each An alignment device comprising: a level control unit for adjusting.