JP2610300B2 - Positioning method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial application field) The present invention relates to a method of aligning two objects, and particularly a position suitable for alignment of a mask and a wafer used for pattern transfer with a wafer. Regarding the matching method.

(Prior Art) In recent years, with the miniaturization of circuit patterns of semiconductor elements such as LSIs, optical projection exposure apparatuses having high resolution performance have been widely used as pattern transfer means.
When transfer is performed using this apparatus, it is necessary to align the mask and the wafer with high accuracy before exposure.

As a method for performing alignment, another optical system (off-axis microscope) different from the projection optical system is used to detect a mark formed in advance on the wafer, position the wafer, and then position the wafer in the projection optical system. An off-axis method of indirectly aligning the mask with a precisely positioned mask by moving it to a predetermined position in the field of view with high precision, and a projection optical system for aligning a mark formed in advance on the mask and the wafer. Through the lens and directly align the mask with the wafer.

The off-axis method has the advantage that the number of alignments is small, the time required for alignment is short, and the throughput (processing speed) is large. However, it is necessary to move the aligned wafer to the position to be transferred by an accurate distance, and an absolute length measurement system must be provided. Is difficult. Therefore, in recent years, in order to perform alignment with higher precision, a method of directly detecting and detecting a mark on a mask and a wafer through a projection optical system, such as a TTL method, has become effective.

As one of the alignment methods of the TTL method, a method of superimposing two grating marks (G. Dubroe
ucq, 1980, ME.WR Trutna.Jr.1984, SPIE). This is, as shown in FIG. 13, the grating-like marks 81a,
A light for alignment is made incident on a mask 81 and a wafer 83 which are formed so as to face each other with a projection lens 82 interposed therebetween, and light diffracted by the two marks 81a and 83a is converted by a photoelectric detector 84a.
Is a method of detecting the overlapping state of the two grating marks 81a and 83a, that is, the relative position by detecting the relative position.

According to this method, the signal strength becomes maximum (or minimum) in a state where the two grating marks 81a and 83a overlap (or are shifted by half a pitch) as shown in FIG. Therefore, high-precision alignment can be performed by signal processing (for example, vibration-type synchronous detection processing or the like) that can accurately detect the maximum value (or minimum value).

By the way, using such a grating mark
The TTL alignment method has the following problems. That is, a normal projection lens is designed such that all aberrations are minimized only at the exposure wavelength, but there is chromatic aberration at other wavelengths. On the other hand, in the above-mentioned alignment method, a laser having good coherence is used as a light source, but since a mercury lamp is used as an exposure light source, it is difficult to make the wavelengths of the two light sources exactly the same. For this reason, the chromatic aberration cannot be ignored for the alignment light, and the position must be detected in a defocused state unlike the exposure wavelength at the imaging position relationship, that is, the focus position. Conventionally, when the position is detected in such a defocused state, sufficient detection sensitivity cannot be obtained and reproducibility is poor. Further, when the difference between the two wavelengths becomes large, aberration correction means such as correcting the optical path length is required, which complicates the apparatus and increases error factors. That is, since the wavelength of the alignment light cannot be made the same as the wavelength of the exposure light, there is a serious problem that the light is affected by the aberration of the projection lens.

In particular, a He-Ne laser (633 nm) is often used because the alignment light does not need to expose the resist, and the transfer exposure wavelength (λ) is generally R = Kλ.
Due to the relationship of / NA, the resolution tends to be shorter in order to increase the resolution R. At present g-line (436 nm) is the mainstream, but in the future it will change to i-line (365 nm), and excimer laser (KrF248n)
m, ArF193nm).

For this reason, the above problem cannot be ignored. Conventionally, this problem of chromatic aberration has been absorbed by providing folding mirrors 85a and 85b in the middle as shown in FIG. However, as the wavelength difference increases, the structure of the folding mirror and the lens system becomes more complicated and impractical.
Here, in the figure, 86 is a He-Ne laser for alignment, 87 is a half mirror, and 88 is a reflection mirror.

(Problems to be Solved by the Invention) As described above, conventionally, it has been difficult to align a mask / wafer with high accuracy due to a problem of chromatic aberration caused by a difference between an exposure wavelength and an alignment light wavelength. Further, when a folding mirror is used as a solution to the chromatic aberration, when the exposure wavelength and the alignment light wavelength are largely different, the mirror interval needs to be largely separated, which is not practical.

The present invention has been made in consideration of the above circumstances, and an object thereof is to accurately determine the relative position of two objects even when the wavelength of the light for alignment is different from the wavelength suitable for the optical system. It is another object of the present invention to provide a positioning method which can be easily realized without using a folding mirror or the like and can improve the positioning accuracy.

[Structure of the Invention] (Means for Solving the Problems) The gist of the present invention is that, in the TTL alignment method using the alignment mark, even if light having a wavelength different from the wavelength suitable for the optical system is used as alignment light, An object of the present invention is to improve a mark arrangement configuration, an alignment light illumination method, and an alignment information detection method so that the influence of chromatic aberration of an optical system (projection lens) can be ignored.

That is, prior to transferring the pattern formed on the mask onto the wafer (via the projection lens), the present invention uses alignment marks provided on the mask and the wafer, respectively, and optically marks these marks. In a positioning method for determining a relative position shift amount and aligning a mask / wafer in accordance with the position shift amount, a wafer mark for positioning made of a diffraction grating is provided on the wafer, and a diffraction grating is provided on the mask. First for alignment
And providing the second mask mark, the light for alignment of different frequencies f ₁ from the exposure light irradiates the first mask mark, different frequency f ₂ to the exposure light (f ₂ ≠ f ₁₎ The second mask mark is irradiated with alignment light, and a part of the diffracted light from each mask mark due to these light irradiations is condensed and interfered (through a projection lens) on the wafer mark, and Out of the reflected diffracted light of (f ₁ −
f ₂ ) The optical heterodyne detection signal having the frequency of
At the same time, the diffracted light from the mask mark, or the diffracted light without the positional information of the mask mark out of the diffracted light from the wafer mark is detected, and the optical heterodyne reference having a frequency of (f ₁ −f ₂ ) is detected. In this method, a signal is obtained, and relative positional deviation information of each mark is obtained from a phase difference between the detection signal and the reference signal.

Further, the present invention is a method wherein a one-dimensional diffraction grating is used as the mask mark, and a two-dimensional diffraction grating or a checkered diffraction grating is used as the wafer mark.

Further, according to the present invention, as the diffracted light having the position information of the mask mark, transmission + 1st-order (or -1st-order) diffracted light from the first mask mark and transmission -1 from the second mask mark-1
Utilizing reflected diffracted light from the wafer obtained by condensing and interfering the next (or + 1st) diffracted light on the wafer, the diffracted light having no positional information of the mask mark is transmitted as the diffracted light having no positional information of the mask mark. Focuses the 0th-order reflected diffracted light or the 0th-order diffracted light transmitted from the mask mark on the wafer mark.
This is a method in which reflected diffraction light from a wafer obtained by interference is used.

The present invention provides, as means for obtaining the frequency f ₁ and the alignment light f _2, the alignment light of a frequency f ₁ and the frequency shifted Delta] f ₁ by the first frequency shifting mechanism the light of frequency f In this method, the light having the frequency f is frequency-shifted by Δf ₂ by the second frequency shift mechanism to obtain the alignment light having the frequency f ₂ .

(Operation) According to the present invention, two diffracted lights from a pair of mask marks separated by a predetermined distance are condensed and interfered with a wafer mark, thereby completely eliminating the problem of a difference in imaging interval due to chromatic aberration. Alignment can be performed without it. Therefore,
It is not necessary to provide intermediate members such as a folding mirror and a lens which have been conventionally used for chromatic aberration correction, and it is possible to eliminate an error factor occurring therefrom. Further, since a folding mirror is not required, the structure is simple and the optical axis adjustment is easy. In particular, it is currently a problem with excimer laser steppers.
TTL alignment can be realized by this method.

Further, in the present invention, the transmission + first-order diffracted light from one of the pair of mask marks and the transmission-first-order diffracted light from the other are condensed and interfered with the wafer mark, and the reflected diffracted light is detected, so that the A beat signal (heterodyne detection signal) having position information can be obtained. Furthermore, a beat signal (heterodyne reference signal) having no mask / wafer position information is obtained by irradiating the transmission zero-order diffracted light from both of the pair of mask marks vertically onto the wafer and detecting the reflected diffracted light. Obtainable.
Then, it is possible to determine the amount of displacement of the mask / wafer from the phase difference between these beat signals.

Here, the phase of the light used for alignment may be disturbed by the optical path, which may also disturb the phase of the alignment detection signal. However, since the alignment reference signal is also obtained from light passing through the same optical path as the light path for obtaining the alignment detection signal, the reference signal also has the same phase disturbance as the detection signal. Therefore,
By comparing the detection signal obtained by the method of the present invention with the reference signal, it is possible to cancel the disturbance of the phase in the optical path, and it is possible to perform positioning with higher accuracy.

(Example) First, before describing an example, a basic principle of the present invention will be described. FIG. 8 is a schematic configuration diagram showing a projection exposure apparatus to which the positioning method proposed by the present inventors has been applied.

Usually, the reduction projection exposure apparatus places a reticle (mask) 10 and a wafer 30 above and below a projection lens 20. Projection lens 2
In the case of 0, the object-image interval between the mask 10 and the wafer 30 is set so that the image-forming relationship is maintained with respect to the exposure wavelength, and positioning is performed as such. In this state, light emitted from point C of the mask 10 forms an image on the wafer 30 through the projection lens 20. vice versa,
The light emitted from the wafer 30 forms an image on the mask 10 through the projection lens 20. However, when a wavelength different from the exposure wavelength, for example, a He-Ne laser or the like is used as the alignment light, the image forming relationship is broken, and the image of the wafer 30 is separated by a distance D from the virtual mask 10 'indicated by a two-dot chain line. The image will be formed above. Usually, when this amount is small, correction is performed by a folding mirror or the like as shown in FIG. However, when the wavelength difference is large such as the exposure wavelength of 248 nm and the He-Ne laser light (alignment light) of 633 nm, this amount is several m.
Therefore, it is not realistic to make correction using a folding mirror or the like.

Therefore, in this example, this problem is solved by the method described below. That is, when the mark S on the virtual mask 10 'is viewed from above the wafer 30 in FIG. 8, the mark information comes with a light beam shown by a two-dot chain line. Therefore, for example, marks 11 and 12 (mask marks) are formed at two points A and B on the mask 10, and the marks 11 and 12 are irradiated with the alignment light, and the light flux emitted therefrom is as if the marks come from above the virtual mask 10 '. If the position information is provided, this light beam will be focused on the wafer 30. That is, mark position information is created on the mask 10 as if it were on the virtual mask 10 '.

More specifically, the marks 11 and 12 of A and B are set as grid marks as shown in FIG. 5 (a), and the alignment light is focused on the entrance pupil of the lens 20 as shown in FIG. Irradiation as As a result, diffracted lights of the 0th order and ± 1st order are generated from the marks 11 and 12, respectively, but the 0th order enters the entrance pupil as it is and is incident perpendicularly to the wafer surface. The -1st-order diffracted light α from the mark 11 and the + 1st-order diffracted light β from the mark 12 have respective positional information and converge on one point on the wafer 30 through the projection lens 20. On the wafer 30, the fifth
A lattice mark 31 (wafer mark) as shown in FIG.

FIG. 9 is a diagram showing the details of FIG. 8. When the diffracted lights α and β enter the lattice mark 31 on the wafer surface, reflected diffracted light is generated. This reflected diffracted light has a spectrum pattern as shown in FIG. 6 (a) on the Fourier transform surface of the lens. Therefore, for example, the 0th-order reflected diffracted light shown in the broken line in FIG. 6A is extracted from the reflected diffracted light, guided by the mirror 43, and detected by the detector 44.

In FIG. 8, the light (frequency f) incident on the marks 11 and 12 is frequency-shifted by frequency shift mechanisms (acousto-optic modulators) 41 and 42 by Δf ₁ and Δf ₂ , respectively.
1 is irradiated with alignment light having a frequency f ₁ , and the mark 12 is irradiated with alignment light having a frequency f ₂ . Accordingly, the detector 44 detects a beat signal (heterodyne detection signal) having a frequency of Δf ₃ = | f ₁ −f ₂ | = | Δf ₁ −Δf ₂ |.
The phase of the beat signal Δf ₂ is a function of the relative position between the mask 10 and the wafer 30. Frequency shift mechanism 41, 42
Electric frequency Δ obtained from a driving circuit (not shown)
From f ₁ and Δf ₂ , the arithmetic unit 46 determines the frequency of Δf ₄ = | Δf ₁ −Δf ₂ | to obtain a beat signal (heterodyne reference signal), and a detector
The phase difference from the beat signal Δf ₃ obtained in 44 is detected by a phase detector 45.
To obtain relative position information of the mask / wafer. Then, the mask / wafer is positioned by moving a stage or the like on which the wafer 30 is placed according to the position information.

FIG. 10 is a further improved example of FIG. The two light beams incident on the mask 10 are frequency-shifted by the frequency shift mechanisms 41 and 42 at the unique frequencies Δf ₁ and Δf ₂ , respectively, and are split by the half mirrors 47 and 48 immediately before the light beams are incident on the mask. To obtain a heterodyne reference signal Δf ₄ . On the other hand, a heterodyne detection signal is obtained in the same manner as in the example of FIG. Then, by calculating the phase difference between the reference signal Δf ₄ obtained by the detector 49 and the detection signal Δf ₃ including the mask / wafer phase difference information obtained by the detector 44, the relative position of the mask / wafer is determined. Can be detected. In other words, a mask / optical heterodyne method
It is possible to detect a wafer position shift.

Here, the principle of the operation of the diffraction grating mark and the optical heterodyne method using the mark will be described with reference to FIGS.
This will be described with reference to FIG. FIG. 11 is a detailed view of the marks 11 and 12 provided on the mask 10 of FIG. 10, and FIG. 12 is a detailed view of the mark 31 provided on the wafer 30 of FIG. In FIGS. 11 and 12, the Y axis is perpendicular to the plane of the drawing.

First, the transmission diffraction grating marks 11 and 12 will be described with reference to FIG. It is assumed that there is a linear lattice in the Y direction at a constant pitch d in the X direction on the XY plane, and the I, J, and K slits of the lattice are transparent, and the other black portions do not transmit light. When parallel rays (incident rays in directions PI, QJ, RK, etc.) having an incident angle α enter this diffraction grating mark, zero-order diffracted lights (IF, JG, A first-order diffracted light (a light ray in the direction of IU, JV, KW, etc.) appears in the direction of the emission angle β.

Now, as shown in FIG. 11, a plane O _A including the y-axis rotated left by an angle α from the x-axis and a plane O including the y-axis rotated right by an angle β, taking a coordinate origin O _A which is inherently fixed with respect to the ray as shown in FIG. _A UVW
think of. Then, since the plane O _A PQR is perpendicular to the incident parallel light, the phase of the light waves of the incident light coincides with the constant wavefront. Also, since the plane O _A UVW is perpendicular to the first order diffracted light,
The phase of the light wave of the primary line light coincides with a constant wavefront.

Now, consider the phase difference of the light wave between the incident wavefront O _A PQR and 1-order diffracted wavefront O _A UVW. Assuming that the position of the slit I in the + x direction is l _A , the optical path length P → I → U of the light ray passing through the slit I is l _A (sinα + sinβ) (1) The optical path length Q → J → V of the light ray passing through the slit J is ( _1A + d) (sinα + sinβ) (2) The optical path length R → K → W of the light beam passing through the slit K is ( _1A + 2d) (sinα + sinβ) (3) PQR is the incident wavefront, and UVW is the first-order diffracted light wavefront. According to the above condition, the difference between the optical path lengths of the above equations (1) and (2) and the equations (2) and (3) must be λ = 632 nm. Therefore, the optical system of FIG. 10 is assembled so that the following equation (4) is satisfied between the incident angle α and the emission angle β of the first-order diffracted light.

sinα + sinβ = λ / d (4) The phase difference φ (rad) of the light wave between the incident wavefront PQR and the first-order diffracted light wavefront UVW is φ = 2πl _A (Equation (1), (2), (3)) sinα + sinβ) / λ This is from equation (4): φ = 2πl _A / d (5) That is, the plane O _A PQR and the plane O _A U of the fixed coordinate system with respect to the light ray
It can be seen that the phase difference φ of the light wave between VW is proportional to the x-coordinate value l _A of the grating slit from the origin O _A of the fixed coordinate system, and is shifted by 2πrad every time the diffraction grating moves by one grating pitch d in the x-axis direction. .

On the other hand, to explain the reflection type diffraction grating mark 31 shown in FIG. 12, there is a grating having a constant reflection pitch in the Y direction on the XY plane with a constant pitch D in the X direction. This is a surface that reflects light well, and other black portions do not reflect light. When a parallel ray (ray in the direction of PI, QJ, RK, etc.) with an incident angle γ is incident on this reflection type diffraction grating mark, the 0th-order diffracted light (ray in the direction of IF, JG, KH, etc.) reflected specularly is emitted. Although appearing in the direction of γ, first-order diffracted light in the vertical direction in the + Z direction (light in the direction of IU, JV, KW, etc.) also appears.

Now, taking the coordinate origin N _A unique fixed with respect to light as Figure 12, considering the plane N _A PQR and the XY plane N _A IJK including Y axis rotated to the left by an angle γ from the X-axis, plane N _A PQR is the incident wavefront, and the plane N _A IJK is the first-order diffracted light wavefront. When determining the phase difference of light waves between these wavefronts, if the + X direction position of the mirror I and m _A, the optical path length P → I is m _A sinγ ... (6) an optical path length Q → J is (m _A + D) sin [gamma] ... (7) the optical path length R → K is _{(m a + 2D) sinγ ...} (8) PQR is incident wavefront, from the conditions IJK is on the first-order diffraction wavefront, equation (6) and (7) and The difference between the optical path lengths in the equations (7) and (8) must be the wavelength γ. Therefore, the following relational expression holds.

sin γ = λ / D (9) Therefore, each of the incident γ should satisfy the relationship of the expression (9)
The optical system shown in FIG. 10 is assembled to direct the first-order diffracted light vertically in the + Z direction. Further, the phase difference of light waves between the incident wavefront PQR and 1-order diffracted wavefront IJK phi (rad) is (6), (7), (8) from equation _{φ = (2πm A sinγ) /} λ which is (9) _{φ = 2πm a / D ... (} 10) ie the equation, the light wave phase difference phi of between stationary plane N _a PQR and stationary plane N _a IJK respect rays seen to be proportional to the X coordinate value m _a of the grating mark 31 .

The operation of the TTL alignment method using the principle of the diffraction grating described above will be described in more detail with reference to FIG.

Now, if the outgoing light E of the He-Ne laser (alignment light) in FIG. 10 is represented by an electric field component of a photoelectromagnetic wave oscillating at the light frequency ω (hereinafter referred to as a light wave), E = | E | ε ^jωt . (11) where | E | is the amplitude of the light wave, j is the imaginary unit, and t is the time.

The light wave A _{1 at} the incident point of the first acousto-optic modulator 41 is Where A ₁ = k ₁ | E |, k ₁ is the attenuation coefficient, φ _A01 is the light source E
Is the phase delay of the light waves of up to A _1.

Acousto-optic modulator 41 operates as an optical frequency shifter, the excitation frequency of the modulator driver incident light A ₁ modulator 41 omega _A =
Out emitted light A ₂ which is frequency-shifted to the positive side by 2π × 80.1MHz.

The principle of an acousto-optic modulator as an optical frequency shifter is well-known. For example, the textbook of optics "Principle of optics III"
(Published by Tokai University Press) pages 876-884.

Outgoing light A ₂ is 2 minutes Hafura 47, the mirror straight light becomes incident wavefront A ₄ of the angle of incidence α of the mask mark 11.

However, | A ₄ | = k ₄ | A ₁ |, T ₁ = φ _A01 + φ _A24 , φ _A24 is A ₂
From a phase delay of the light waves of up to A _4.

Now, in FIG. 10, the coordinate origins O _A , O
_B , N _A and N _B are at fixed positions with respect to one coordinate system,
All the optical elements in FIG. 0 are fixedly arranged in this coordinate system. Therefore, the position E of the light wave of Fig. _{10, A 1 ~A 8, B 1} ~B
₈ , S is also fixed to this coordinate system. What moves with respect to this coordinate system is that the mask 10 only moves in the ± x direction and the wafer 30 moves in the ± x direction.

As shown in FIG. 10, when the mask 10 is at the regular x coordinate position,
The distance between the mark 11 and O _A and the distance between the mark 12 and O _B are l _A and l _B ,
If the mask 10 is now shifted from the normal position by a minute distance + Δx, the lattice position dimension l _{A in} FIG. 11 becomes l _A + Δx, and the dimension l _B changes to l _B -Δx. Therefore, a value is the phase delay amount φ of the equation (5) l _A was replaced with l _A + [Delta] x with respect to the wavefront A ₄ of the transmission 1 order diffraction wavefront A ₅ of the mark 11 of FIG. 10.
That is, However, | A ₅ | = k ₅ | A ₄ |, T ₂ = T ₁ + 2π (Δx + 1 _A ) / d.

Wavefront A ₅ of the exit angle β is refracted by the transfer lens 20, toward the direction of the wafer mark 31 at an incident angle gamma, the incident wave front A ₆ of the mark 31.

However, T ₃ = T ₂ + _φA56 .

As shown in FIG. 10, the wafer 30 is properly aligned X
At the coordinate position, the distances between the wafer mark 31 and N _{A and} between N _A and N _B are m _A and m _B respectively. If the wafer 30 is now shifted from the normal position by a minute distance ΔX, the wafer mark shown in FIG. position dimensions m _A to m _A + [Delta] X, dimension m _B is changed to m _B -ΔX. Therefore, the first-order reflected diffracted light A _{7 with} respect to the incident wavefront A ₆ of the mark 31
The phase delay amount φ is a value obtained by replacing the m _A + [Delta] X and m _A of equation (10). That is, _{However, | A 7 | = k 7} | A 5 |, a _{T 4 = T 3 + 2π (} ΔX + m A) / D.

The primary reflected diffracted light A ₇ is bent by the mirror 43 and the light wave A ₈
And reaches the photodetector 44.

_{However, T 5 = T 4 + φ} A78, φ A78 is the phase lag of light waves from A ₇ to A _8.

On the other hand, a portion of the emitted light A ₂ from the acoustic-optic modulator 41 is reflected by the half mirror 47 and reaches a photodetector 49 becomes light waves.

Where | A ₃ | = k ₃ | A ₁ |, T ₆ = φ _A01 + φ _A23 .

Similarly, B _{1 at} the incident point of the acousto-optic modulator 42 is Here, φ _B01 is the phase delay of the light wave from E to B ₁ .

Modulator issues emitted light B ₂ obtained by frequency shifting the frequency ω _{B =} 2π × 80.0MHz driver for example.

The incident wavefront B ₄ of the mask mark is However, it is _{U 1 = φ B01 + φ B24} .

The transmitted first-order diffracted light wavefront B ₅ of the mask mark is represented by l _{A in the} expression (5).
With l ₈ −Δx, However, it is _{U 2 = U 1 + 2π (} Δx + l B) / d.

The incident wavefront of the wafer mark is However, U ₃ = U ₂ + _φB56 .

First-order reflected diffraction light B ₇ of the wafer mark with respect to this wavefront B ₆
It is at the m _B -ΔX the m _A of equation (10), However, | B ₇ | = k ₇ | B ₅ | and U ₄ = U ₃ + 2π (ΔX + m _B ) / D.

The light wave B ₇ reaches the photodetector is a light plane B _8.

_{However, U 5 = U 4 + φ} B78, φ B78 is the phase lag of light waves between B ₈ from B _7.

On the other hand, a part of the light wave B ₂ is reflected by the half mirror 48, the light wave
B ₃ is incident on the photodetector 49.

However, | B ₃ | = k ₃ | B ₁ | and U ₆ = φ _B01 + φ _B23 .

The photodetector 49 calculates the amplitude square value of the combined incident light wave value A ₃ + B ₃ | A ₃
+ B ₃ | proportional to the ^second optical power making electrical signal I _R. | A ₃ + B
₃ | ² is obtained from the above equations (19) and (27), and I _R = | A ₃ + B ₃ | ² = | A ₃ | ² + | B ₃ | ² +2 | A ₃ || B ₃ | cos [(ω _A −ω _B ) t− (T ₆ −U ₆ )] (28) On the other hand, the photodetector calculates the amplitude square value of the incident light wave composite value A ₈ + B _8.
| Making electrical signal I _S that is proportional to the optical power of ² | A ₈ + B _8. The optical power | A ₈ + B ₈ | ² is calculated from the above equations (18) and (26) as I _S = | A ₈ + B ₈ | ² = | A ₇ | ² + | B ₇ | ² +2 | A ₇ || B ₇ | cos [(ω _A −ω _B ) t− (T ₅ −U ₅ )] (29) where T ₅ −U ₅ = 4π (Δx / d + ΔX / D) + (φ _A01 −φ _B01 ) _{+ φ dif φ dif = (φ} A24 -φ A24) + 2π (l A -l B) / d + (φ A56 -φ B56) + 2π (m A -m B) / D + (φ A78 -φ B78) ... (30 ) (28) of the electrical signal I _R of formula DC component ^{_{| a 3 | 2 + | B}} 3 | 2 except _{(ω a -ω B) = 2π} (80.1-80.0) frequency of MHz 1
The 00 KHz AC signal I _RAC is I _RAC = 2 | A ₃ || B ₃ | cos [(ω _A −ω _B ) t− (T ₆ −U ₆ )] (31) The signal is input to a first input terminal of the detection circuit 45.

On the other hand, (29) of the electrical signal I _S, the AC signal I _SAC frequency _{100kHz I SAC = 2 | A 7} || B 7 | cos [(ω A -ω B) t- (T 5 -T ₅ ) (32) is amplified by the AC amplifier and input to the second input terminal of the phase detection circuit 45.

The phase comparator calculates the phase T ₆ −U ₆ = (φ _A01 −φ _B01 ) − (φ _A23 −φ _B23 ) of the I _RAC of the equation (31) and the phase T ₅ −U ₅ = of the I _SOC of the equation (32). 4π (Δx / d + ΔX / D) + (φ _A01 −φ _B01 ) + φ _dif, and the phase difference [4π (Δx / d + ΔX / D) + φ _dif − (φ _A23 −φ _B23 ) is obtained. Since the term [φ _dif − (φ _A23 −φ _B23 )] in the equation is a constant value according to the equation (30), the phase difference φ ４ = 4π (Δx / d + ΔX / D) where the constant offset is cancelled. 33) Outputs the alignment position error signal し た proportional to. Note that such a phase comparator is well known.

As shown in FIG. 10, the mask pattern indicated by the arrow in the + x direction shows the transfer lens 2 at the exposure wavelength of 248 nm.
Exactly 1/5 multiplied by 0, indicated by arrow in -X direction
An IC pattern image is formed on the wafer 30. Therefore, when the mask is shifted from the normal position by + Δx, the imaging position on the wafer is shifted by ΔX = −Δx / 5. Therefore, wafer 3
If 0 is shifted by the normal position by ΔX = −ΔX / 5, it means that the alignment alignment is matched.

Since D and d in equation (33) have a relationship of D = d / 5 as described above, ψ in equation (33) is proportional to the alignment position error, and the sign of the value indicates the error direction. Therefore, the wafer table controller completes the alignment by controlling the position of the wafer table so that the alignment signal becomes zero.

By the way, TTL that makes such two light beams enter the mask
Alignment has the following problems. That is, since the alignment detection signal detected by the detector 44 and the alignment reference signal obtained by the drive circuits of the frequency shifters 41 and 42 have different signal detection positions, the phase fluctuation of the laser oscillator serving as the light source of the aligning light is different. And a phase change due to the influence of disturbances such as airflow in the alignment optical path have resulted in alignment errors. In other words, it is sometimes difficult to align the mask / wafer with high accuracy due to the phase fluctuation of the alignment signal due to the influence of the laser oscillator and the optical path environment. The method shown in FIG.
Although the influence of the phase fluctuation can be reduced as compared with the figure, the effect of the phase fluctuation in the optical path of the light after the mask 10 is unavoidable in this case as well. Therefore, in the present invention, a heterodyne reference signal is obtained based on the light reflected by the mask or the light reflected by the wafer through the mask.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram showing a reduction projection exposure apparatus used in the method of the first embodiment of the present invention.

Frequency shift mechanism of He-Ne laser beam split into two light beams
After shifting the frequency by Δf ₁ and Δf ₂ at 41 and 42, respectively, one-dimensional as shown in FIG. 5 (a) or FIGS. 5 (b) and (c)
Mask marks 11 and 12 composed of a two-dimensional diffraction grating as shown in FIG.
Irradiation. That is, the first mark 11 is irradiated with the alignment light having the frequency f ₁ , and the second mark 12 is irradiated with the alignment light having the frequency f ₂ . Then, a plurality of diffracted lights are generated from each of the marks 11 and 12. Of these diffracted lights, the phase of the zero-order light transmitted through the mask 10 is independent of the mask mark position, but the phase of the higher-order diffracted light depends on the mask mark position. In the present embodiment, the -1st-order diffracted light α generated at the mark 11 and the + 1st-order diffracted light β generated at the mark 12
The light is reflected and diffracted from the wafer mark 31 by the photodetector 44 and condensed and interferes with the check mark lattice-like wafer mark 31 on the wafer 30 through the light beam 20. The light diffracted by the mark 31 on the wafer surface by the α and β light shows a spectrum pattern as shown in FIG. 6A on the Fourier transform surface of the lens 20. Then, the 0th-order diffracted light shown in the broken line in FIG. 6A is extracted and guided to the photodetector 44 by the mirror 43 and the lens 51. In this detector 44, a beat signal (alignment detection signal) having a frequency difference Δf ₃ = | Δf ₁ −Δf ₂ | obtained by frequency shifting by the frequency shift mechanisms 41 and 42 is obtained. The detection of the beat signal is the same as in the examples of FIGS. 8 and 10.

The feature of the present embodiment is that the detection signal Δ
In the method of detecting the reference signal Δf ₄ to be compared with f ₃ . That is, the 0th-order light γ diffracted by the mask mark 11 and the 0th-order light δ diffracted by the mask mark 12 enter the entrance pupil and enter perpendicular to the wafer surface. The wafer mark 31 is arranged at the irradiation position of the γ and δ light and diffracts the γ and δ light irradiated on the wafer surface. FIG. 2 shows FIG. 1 in more detail.

The light reflected and diffracted by the checkered lattice mark 31 on the wafer surface by the γ and δ light forms a spectrum pattern indicated by a white circle in FIG. 6B on the Fourier transform surface of the lens.
For example, the reflected diffracted light shown within the broken line in FIG.
-1 order light (+1 order light) where light is diffracted by the lattice mark 31
And δ light are diffracted at the lattice mark 31 by −1 order light (+1 order light).

The phases of the two light beams do not have position information in the direction in which the position of the mask 10 and the position of the wafer 30 are relatively aligned. That is, for the light reflected by the mirror 43, the convex lens 5
1. By arranging the concave lens 52 as shown in FIG. 1 and combining and synthesizing the −1st order light (+ 1st order light) diffracted from the δ and γ light, it has no relative position information between the mask and the wafer, Δf ₄ =
A beat signal (alignment reference signal) having a frequency of | Δf ₁ −Δf ₂ | can be detected by the photodetector 49 as an electric signal.

The reference signal includes phase change information generated by a disturbance in which the He-Ne laser beam is split into two light beams and is applied to the mask surface. Accordingly, the phase detection circuit 45 detects the phase difference between the beat signals Δf ₃ and Δf ₄ having the relative position of the mask / wafer based on α and β obtained by the photodetector 44 as phase information. Can be reduced due to disturbance in the optical path during which the light beam is split into two light beams and reaches the mask surface, thereby causing a phase variation.

As described above, according to the method of the present embodiment, the two marks generated from the pair of mask marks 11 and 12 provided at positions separated by a predetermined distance.
By focusing and interfering the diffracted light on the wafer mark 31, the problem of the difference in the imaging interval due to chromatic aberration can be completely eliminated, and the alignment can be performed. That is, with the TTL alignment method using the alignment mark, conventionally, various problems caused by the chromatic aberration of the projection lens 20, which had been a problem when the wavelength difference was large, can be easily solved. Can be detected very accurately, and accurate positioning can be performed. Accordingly, there is no need to provide intermediate members such as a folding mirror and a lens which have been conventionally used for chromatic aberration correction, and it is possible to eliminate an error factor caused thereby. Furthermore, since a folding mirror or the like is not required, the structure is simple and the optical axis adjustment is easy. In particular, it is currently a problem with excimer laser steppers. TTL
Alignment can be realized by this method.

Also, since the optical heterodyne method is used,
As an alignment position error signal, it is proportional to the position error,
There is an advantage that a signal whose polarity changes to positive or negative depending on the direction is always obtained. Further, by comparing the detection signal with the reference signal, there is an advantage that the detection can be performed irrespective of light attenuation occurring in the optical system. Also, by obtaining diffracted light having no positional information transmitted through the mask mark and reflected from the wafer mark as a reference signal, the optical path during which the alignment light is split into two light beams and reaches the mask surface is disturbed. Alignment errors caused by phase fluctuations can be reduced.

FIG. 3 and FIG. 4 show another embodiment of the present invention. In the example of FIG. 3, the He—Ne laser light separated into two light beams is frequency-shifted by Δf ₁ and Δf ₂ by the frequency shift mechanisms 41 and 42, respectively, and then irradiated onto the mask marks 11 and 12, respectively. Here, the mask marks 11 and 12 are formed by providing a two-dimensional mark on a part of a one-dimensional diffraction grating as shown in FIG. 7 (a) or a shape of a two-dimensional diffraction grating as shown in FIG. 7 (b). It was done. Of the diffracted light from the mask marks 11 and 12, the mark 1 of the higher-order diffracted light generated by reflecting the mask surface is perpendicular to the two light flux surfaces incident on the mask 10.
In both cases, the + 1st-order light indicated by the dashed line in FIG. 3 is selected, and the two reflected light beams are combined by mirrors 61 and 62, and the reference signal Δf ₄ having no mask position as phase information is detected by the photodetector 49. Is detected as an electric signal.

Further, in the example of FIG. 4, high-order diffracted light generated by transmitting through the mask surface in a direction perpendicular to the two light beams incident on the mask out of the transmitted diffracted light generated at the mask marks 11 and 12 is referred to as As in the example shown in FIG. 3, two light beams are combined, and the reference signal Δf ₄ having no mask position as phase information is detected by the photodetector 49 as an electric signal.

The phase difference between the reference signal Δf ₄ and the beat signal Δf ₃ obtained by the detector 44 shown in the previous example is obtained by the phase detection circuit 45, so that the laser light is separated into two light fluxes and reaches the mask surface. The present invention can reduce an alignment error caused by a phase fluctuation caused by a disturbance in an optical path.

Even with such a configuration, the same effect as in the previous embodiment can be obtained. Further, in this example, it is not possible to reduce the phase disturbance in the optical path from the mask of the heterodyne detection signal to the wafer and further from the wafer to the mask, but the phase disturbance from the laser light source to the mask does not greatly affect the phase disturbance. This is an optical path, and with respect to this optical path, phase disturbance can be reliably reduced.

Note that the present invention is not limited to the above-described embodiments. For example, diffracted light from a mask that irradiates a wafer mark to obtain a heterodyne detection signal has +1 order and −
The order is not limited to the first order, and diffracted lights of the opposite orders may be selected from the pair of mask marks. Further, although a projection lens is used in the embodiment, the present invention can be applied to a transfer apparatus that does not use a projection lens, such as X-ray transfer.
In this case, a pair of diffraction grating marks are provided on the X-ray mask,
The diffracted light from these marks may be guided to the wafer mark, and the reflected diffracted light from the wafer mark may be detected by a sensor or the like. Further, the alignment irradiation light may be irradiated from above the condenser lens above the mask or directly above the mask.

Furthermore, in the embodiment, the example in which the first and second mask marks are formed separately from each other on the mask has been described, but a similar lattice mark is formed between these two marks.
As a matter of course, it may be apparently one continuous mark, and the two light beams may be applied to a position corresponding to the first and second mask mark positions on this one mark. In addition, various modifications can be made without departing from the scope of the present invention.

[Effects of the Invention] As described above in detail, according to the present invention, even when the wavelength of the light for alignment is different from the wavelength suitable for the optical system, the present invention is not limited thereto.
The relative positions of the two objects can be detected with high accuracy, and can be easily realized without using a folding mirror or the like, and the alignment accuracy can be improved. Further, by detecting the phase difference between the light having the relative position information of the mask / wafer as the phase information and the reference signal, the mask / wafer is detected.
In the TTL alignment method that aligns the wafer, the reference signal is the diffracted light generated from the mask mark or the reflected light from the wafer, and the reference signal is obtained from the light that does not have the relative position information of the mask / wafer. Can be reduced due to disturbance in the optical path during which the light beam is split into two light beams and reaches the mask surface, thereby causing a phase variation.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram showing a projection exposure apparatus used in the method of one embodiment of the present invention, FIG. 2 is a schematic configuration diagram showing a more specific version of FIG. 1, and FIGS. 3 and 4 are modified examples. 5, FIG. 5 is a plan view showing a diffraction grating mark used in the apparatus shown in FIG. 1, FIG. 6 is a schematic diagram showing a spectrum pattern of diffracted light, FIG. 7 is FIGS. FIGS. 8 to 12 are plan views showing diffraction grating marks used in the apparatus shown in the figure, respectively, for explaining the basic principle of the present invention. FIGS. 8 to 10 are schematic views showing a projection exposure apparatus. FIG. 11 is a schematic diagram showing details of a mask mark, FIG. 12 is a schematic diagram showing details of a wafer mark, and FIGS. 13 to 15 are diagrams for explaining a conventional problem. . 10 ... mask, 11,12 ... mask mark, 20 ... projection lens, 30 ... wafer, 31 ... wafer mark, 41,42 ...
Frequency shift mechanism, 43, 47, 48, 61, 62… Mirror, 44, 49
... photodetector, 45 ... phase detection circuit, 51 ... convex lens,
52 ... Concave lens.

Continued on the front page (72) Inventor Yoshitomi Sameda 1 Toshiba-cho, Fuchu-shi, Tokyo Inside the Toshiba Fuchu Plant (72) Inventor Toshikazu Yoshino 75-1 Hasunumacho, Itabashi-ku, Tokyo Tokyo Optical Machinery Co., Ltd. In-company (72) Inventor Susumu Saito 75-1 Hasunuma-cho, Itabashi-ku, Tokyo Tokyo Optical Machinery Co., Ltd. (56) References JP-A-62-261003 (JP, A) JP-A-62-172203 (JP, A) Japanese Patent Publication No. 59-38521 (JP, B2)

Claims (4)

(57) [Claims] 1. Prior to transferring a pattern formed on a mask onto a wafer, using an alignment mark provided on each of the mask and the wafer, an optical relative positional shift amount of these marks is obtained. In a positioning method for positioning a mask / wafer in accordance with the amount of positional deviation, a wafer mark for positioning comprising a diffraction grating is provided on the wafer, and at least a first and a second position of the mask are respectively provided. the first and second mask mark for alignment consisting of a diffraction grating provided, light for alignment of different frequencies f ₁ from the exposure light irradiates the first mask mark, the frequency different from the exposure light f ₂ (f ₂ ≠ f ₁ )
Irradiating the second mask mark with the alignment light of
A part of the diffracted light from each mask mark due to the light irradiation is condensed and interfered on the wafer mark, and the diffracted light having the position information of the mask mark is detected from the diffracted light reflected from the wafer mark ( An optical heterodyne detection signal having a frequency of f ₁ −f ₂ ) is obtained, and at the same time, a diffracted light having no mask mark position information among the diffracted light from the mask mark or the reflected diffracted light from the wafer mark is detected. An optical heterodyne reference signal having a frequency of (f ₁ −f ₂ ), and obtaining relative positional deviation information of each mark from a phase difference between the detection signal and the reference signal. 2. Prior to transferring a pattern formed on a mask onto a wafer via a projection lens, an alignment mark provided on each of the mask and the wafer is used and an optical relative position of these marks is used. Find the displacement amount,
In a positioning method for positioning a mask / wafer in accordance with the amount of positional deviation, a wafer mark for positioning including a diffraction grating is provided on the wafer, and at least a first and a second position of the mask are respectively provided on the mask. the first and second mask mark for alignment consisting of a diffraction grating provided, light for alignment of different frequencies f ₁ from the exposure light irradiates the first mask mark, the frequency different from the exposure light f ₂ (f ₂ ≠
irradiating the second mask mark with the positioning light of f ₁ ), and condensing and interfering a part of the diffracted light from each mask mark by the irradiation of the light on the wafer mark through the projection lens; Among the diffracted light reflected from the wafer mark, the diffracted light having the position information of the mask mark is detected to obtain an optical heterodyne detection signal having a frequency of (f ₁ −f ₂ ). At the same time, the diffracted light from the mask mark is obtained. Alternatively, among the diffracted light reflected from the wafer mark, the diffracted light having no positional information of the mask mark is detected to obtain an optical heterodyne reference signal having a frequency of (f ₁ −f ₂ ). A relative position shift information of each mark is obtained from the phase difference. 3. The diffracted light having the position information of the mask mark is transmitted through a first mask mark and is + 1st order (or -1st order).
Next) diffracted light from the wafer obtained by converging and diffracting the transmitted light from the second mask mark and the transmitted -1st (or + 1st) diffracted light from the second mask mark on the wafer mark, and the positional information of the mask mark The diffracted light having no diffraction order is the 0th-order transmitted or 0th-order diffracted light from the mask mark or the 0th-order diffracted light from the mask mark reflected and diffracted from the wafer mark obtained by converging and interfering on the wafer mark. The alignment method according to claim 1, wherein the alignment method is light. 4. Prior to transferring a pattern formed on a mask onto a wafer via a projection lens, an alignment mark provided on each of the mask and the wafer is used, and an optical relative position of these marks is used. Find the displacement amount,
In a positioning method for positioning a mask / wafer in accordance with the amount of positional deviation, a wafer mark for positioning comprising a one-dimensional diffraction grating is provided on the wafer, and at least a first mark of the mask is provided.
And first and second mask marks for positioning, each of which is formed of a two-dimensional diffraction grating or a checkered diffraction grating, are provided at a second location, and light having a frequency f different from the exposure light is frequency-shifted by Δf ₁ to obtain a frequency. with the alignment light of f ₁ is irradiated to the first mask mark, the light of the frequency f Delta] f ₂ by the frequency shift and the second alignment for the light of frequency f ₂ and
And the + 1st-order (or + 1st-order) transmitted diffracted light from the first mask mark and the -1st-order (or + 1st-order) transmitted diffracted light from the second mask mark are projected. is focused onto the wafer mark respectively through the lens, with the frequency of the reflected diffraction light from the wafer mark having a position information of the mask wafer is detected (the frequency f ₁ and f _2), respectively (f ₁ -f ₂₎ The optical heterodyne detection signal is obtained, and at the same time, the 0th-order transmitted diffracted light from the first and second mask marks is respectively perpendicularly incident on the wafer mark, and the light from the wafer mark having no mask / wafer position information is detected. The reflected diffracted light (frequency f ₁ and f ₂ ) is detected (f ₁
Obtains an optical heterodyne reference signal having a frequency of -f _2), alignment method and obtains the relative displacement information of the marks from the phase difference between the detection signal and the reference signal.