JP3368267B2 - Projection exposure equipment - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a projection exposure apparatus suitable for manufacturing semiconductor devices, and more particularly, to a relative alignment of a reticle having a conjugate relationship with a wafer and an image pickup means synchronized (aligned) with the wafer. By using the interference image based on the reflected and diffracted light generated from the periodic lattice-shaped marks (wafer mark, alignment mark) provided on the surface, it is possible to perform with high accuracy and when manufacturing a highly integrated semiconductor element. It is suitable.

[0002]

2. Description of the Related Art The recent progress in manufacturing technology of semiconductor devices is remarkable, and accompanying it, the progress of fine processing technology is remarkable.
In particular, the optical processing technology has reached the level of fine processing technology having submicron resolution at the border of the production of semiconductor devices of 1M DRAM. In many cases, a method of fixing the exposure wavelength and increasing the NA (numerical aperture) of the optical system has been used as a means for improving the resolution. However, recently, various attempts have been made to change the exposure wavelength from the g-line to the i-line and improve the resolution by an exposure method using an ultra-high pressure mercury lamp.

It is generally known that the depth of focus of a projection optical system (stepper) is inversely proportional to the square of NA. For this reason, when trying to obtain submicron resolution, the problem that the degree of focus advance becomes shallower occurs.

On the other hand, various methods have been proposed for improving the resolution by using light having a shorter wavelength, which is represented by an excimer laser. It is known that the effect of using light having a short wavelength generally has an effect that is inversely proportional to the wavelength, and the depth of focus becomes deeper as the wavelength becomes shorter.

On the other hand, in the conventional relative alignment between the wafer and the reticle in the projection exposure apparatus, that is, so-called alignment, alignment marks formed on the wafer surface are formed on the image pickup means surface with or without a projection lens, and the image pickup is performed. This is performed by observing the alignment mark on the device surface and obtaining the position information of the wafer.

At this time, as an observation method of the alignment mark on the wafer surface for obtaining the wafer position information,
The following three methods are mainly used. (A) A method that uses non-exposure light having a different wavelength from the exposure light and does not pass through the projection lens (OFF-AXIS method) (b) Method that uses light having the same wavelength as the exposure light and passes through the projection lens (exposure Optical TTL method) (C) Method of using non-exposure light having a wavelength different from that of exposure light and passing through a projection lens (non-exposure light TTL method) FIG. 9 is a main part of an OFF-AXIS optical system of (a). It is a schematic diagram. In the figure, the wafer W is irradiated with the non-exposure light from the light source 501 via the half mirror 502 and the detection lens system 503.
The alignment mark AM on the surface is illuminated. Then, the alignment mark AM is imaged on the surface of the image pickup means 504 by the detection lens system 503 via the half mirror 502.

FIG. 10 is a schematic view of the main part of the exposure light TTL type optical system shown in FIG.

In the figure, a half mirror 602 and a detection lens system 60 are sequentially formed by a light flux having the same wavelength as the exposure light from a light source 601.
3, the alignment mark AM on the wafer W surface is illuminated via the mirror 604 and the projection lens 1. Then, the alignment mark AM is sequentially applied to the projection lens 1 and the mirror 60.
4. An image is formed on the surface of the image pickup means 605 through the detection lens system 603.

FIG. 11 is a schematic view of a main part of the non-exposure light TTL type optical system of FIG. In the figure, the non-exposure light from the light source 701 is sequentially applied to the half mirror 702 and the correction lens system 703.
An alignment mark AM on the surface of the wafer W is illuminated via a mirror 704 projection lens 1 (a lens system that corrects aberrations caused by a difference in wavelength of exposure light). Then, the alignment mark AM is imaged on the surface of the image pickup means 705 through the projection lens 1, the mirror 704, the correction lens system 703, and the half mirror 702 in order.

As described above, in the conventional method shown in FIGS. 9 to 11, the position information of the wafer W is obtained by detecting the position of the alignment mark AM imaged on the image pickup means.

Further, as means for specifically detecting the position of the alignment mark from the image information obtained by each of the above observation methods, for example, the pattern matching detection method disclosed in Japanese Patent Laid-Open No. 232504/1987 or the special feature is disclosed. Kaihei 3-28
There is a phase detection method by FFT disclosed in Japanese Patent No. 2715.

[0012]

Recently, the miniaturization of semiconductor elements has been advanced, and the exposure wavelength of the projection exposure apparatus has been shortened in order to increase the integration degree of these semiconductor elements. For this reason, the g-line (436 nm) of a high pressure mercury lamp has been used as the exposure wavelength, but i-line (365 nm) or excimer laser, for example, KrF laser (248 nm) is gradually used as the exposure light. ing. As such shortening of the exposure wavelength progresses, at the same time, an important issue becomes what kind of alignment method is used. This is because the projection lens is generally aberration-corrected only for the exposure wavelength.

The OFF-AXIS method of (a) is a method of performing alignment light without passing through the projection lens, and is not affected by the exposure wavelength of the projection lens. Therefore, the design of the observation optical system is relatively easy.

However, the OFF-AXIS system is
Due to the geometrical restrictions between the observation optical system and the projection lens, the alignment position and the exposure position are generally very different.
After the alignment is completed, it is necessary to drive the XY stage to the exposure position.

There is no problem if the distance between the alignment position and the exposure position (hereinafter referred to as the baseline) is always stable, but the environment in which the apparatus is placed (temperature or atmospheric pressure, or a mechanism associated with vibration of the apparatus itself). There is a problem in that a change over time occurs due to the influence of the stability of the above).

Therefore, in the OFF-AXIS system, the baseline is generally measured and corrected at a certain fixed time interval. As described above, since the OFF-AXIS method has an error factor of “variation of baseline with time”, it takes a long time to perform the measurement and correction, and there is a problem that throughput is reduced.

Further, since the OFF-AXIS system is a system which does not pass through the projection lens, it does not follow the behavior of the projection lens (for example, magnification / focus position change due to exposure, magnification / focus position change due to atmospheric pressure, etc.). There is.

On the other hand, the TTL method of aligning the alignment light through the projection lens is advantageous from the viewpoint of following the behavior of the projection lens described above, and does not cause the problem of the baseline. OFF-AX even if it occurs again
Compared with the IS method, it is generally shorter by one digit or more and is less susceptible to environmental changes.

However, the projection lens is aberration-corrected so that the projection exposure is optimized for the exposure wavelength, and when a light beam having a wavelength other than the exposure wavelength is used as the alignment light, the aberration generated in the projection lens is very large. It becomes something.

According to the exposure light TTL method mentioned above (b),
Since the exposure light is used as the alignment light, the aberration of the projection lens is well corrected, and a good observation optical system can be obtained.

However, in many cases, a photosensitive material (resist) to which an electronic circuit pattern is transferred is applied on the wafer surface. Therefore, when observing the alignment mark formed on the wafer surface, the resist usually absorbs much at a short wavelength. Therefore, as the exposure wavelength becomes shorter as described above, it becomes difficult to detect the alignment mark on the wafer at the exposure wavelength through the resist film.

Further, when the alignment mark is observed with the exposure light, there is a problem that the resist is exposed by the exposure light and the detection of the alignment mark becomes unstable or cannot be detected.

In the non-exposure light TTL method described in (c) above, since a light beam having a wavelength other than the exposure wavelength is used as the alignment light as described above, many aberrations occur in the projection lens. Therefore, conventionally, when this non-exposure light TTL method is used, alignment is performed via a correction optical system for correcting the aberration generated in the projection lens as disclosed in, for example, Japanese Patent Laid-Open No. 3-61802. It is configured to detect the mark.

However, an excimer laser such as K
When an rF laser (248 nm) is used as the exposure light, practical glass materials that make up the projection lens are limited to quartz, fluorite, etc. due to factors such as transmittance, and as a result, the non-exposure that occurs in the projection lens. Aberration to light becomes very large. Therefore, it is difficult to satisfactorily correct the aberration generated in the projection lens by using the correction optical system, and it is difficult to obtain a sufficient NA, and further, it is impossible to construct a realistic correction system.

Further, the pattern matching detection method as a means for specifically detecting the position of the alignment mark from the image information obtained by the above-mentioned observation method includes an A / D for processing the electric signal obtained by the image pickup means. The signal becomes a discrete series by the D conversion, and the position of the detected alignment mark also takes a discrete value.

Therefore, in order to achieve the desired accuracy, it is necessary to use some kind of interpolation means, and an error due to approximation at that time causes a detection error. Further, the image of the alignment mark may be distorted due to noise such as coating unevenness of the resist applied on the upper side or uneven lighting, and S /
There is a problem that the detection accuracy is lowered due to the deterioration of the N ratio.

On the other hand, the applicant of the present invention has previously filed Japanese Patent Laid-Open No. 3-2.
The phase detection method by FFT disclosed in Japanese Patent No. 82715 has been proposed to solve this problem.

However, the bright field detection image of the alignment mark (hereinafter, wafer mark) formed on the wafer is shown in FIG.
In a structure in which a wafer mark having a cross-sectional shape as shown in FIG. 2 (c) is coated with a resist, a wafer mark image as shown in FIG. 12 (A) may be obtained depending on process conditions such as substrate reflectance and interference conditions. May be detected by the CCD, or a wafer mark image as shown in FIG. 12B may be detected by the CCD.

Therefore, when the phase detection method by FFT is used, the spatial frequency to be processed differs depending on the process conditions, and the offset (measurement value for each spatial frequency is accurate with respect to a wide range of spatial frequencies as a processing system. Difference) is required. This imposes a heavy load on the processing system and is a factor that hinders high-precision processing for a certain fixed spatial frequency.

Further, in such a bright field detection method, the S / N ratio may be poor and detection may not be possible especially in a low step process.

Further, by making the detection optical system a dark-field observation, by illuminating the edge regardless of the process conditions,
It is possible to stably detect a wafer mark image as shown in FIG.

However, dark field observation has a problem in scattering efficiency and requires a large amount of light as compared with bright field observation, and as in the case of bright field detection, particularly in a low step process, the S / N ratio is poor and detection is impossible. However, there was a problem in that

The present invention is for alignment (for alignment) provided on the wafer surface by a configuration in which each element is set appropriately.
The grid-shaped marks are illuminated with monochromatic light having a wavelength different from the wavelength of the exposure light, an interference image based on the reflected diffracted light generated at this time is formed on the surface of the image pickup means, and the video signal obtained by the image pickup means is used. By doing so, it is an object of the present invention to provide a projection exposure apparatus capable of performing alignment of a reticle and a wafer at high speed and with high accuracy, and performing projection exposure with high resolution.

[0034]

A projection exposure apparatus according to a first aspect of the present invention is a projection exposure apparatus which projects and exposes a pattern formed on a reticle illuminated with exposure light onto a wafer surface through a projection lens. Illuminating means for illuminating a lattice mark provided on the wafer surface with monochromatic light through the projection lens,
Reflection diffraction from the lattice mark passing through the projection lens
A reflecting member that guides light to the outside of the exposure light path, and
Selectively take out ± n-order light (n = 1, 2, 3 ...) Of them
Projection optical means provided with a stopper, and the extracted n-th order
Synchronized with the reticle by the projection optical means using light
And an image pickup means for forming an interference image on the reflected surface, and the stopper has a taking NA on the wafer side for each reflected diffracted light, the wavelength of the monochromatic light from the illuminating means being λ, and the grating. When the absolute value of the projection magnification of the circular mark on the surface of the image pickup means is β and the spatial frequency of the interference image formed on the surface of the image pickup means is T, 0 <NA <T · β · λ / 2 is satisfied. The one-dimensional projection integrated signal obtained by electrically or optically projecting and integrating in one direction of the two-dimensional coordinates, which is obtained from the imaging signal of the interference image formed on the surface of the imaging means, is orthogonally transformed. The spatial frequency domain is converted into a spatial frequency domain by the spatial frequency domain, and a spatial frequency component that appears uniquely from the interference image based on the periodicity of the lattice-shaped mark on the spatial frequency domain is selected from the one-dimensional projected integrated signal to determine the position of the lattice-shaped mark. By detecting, the wafer can be aligned with a predetermined position. It is characterized by doing so. According to a second aspect of the invention, in the invention of the first aspect, the projection lens is aberration-corrected with exposure light, and the wavelength of the monochromatic light from the illumination means is different from the wavelength of the exposure light. Has a correction optical system that corrects various aberrations that occur when the lattice-shaped marks on the wafer surface illuminated by the monochromatic light are projected on a predetermined surface through the projection lens. .

[0035]

1 is a schematic view of a main portion of a projection exposure apparatus as an example for explaining the present invention.

In the figure, the electronic circuit pattern on the reticle R surface illuminated by the exposure light from the illuminator IL is reduced and projected by the projection lens 1 onto the wafer W surface mounted on the wafer stage ST, and the electronic circuit is displayed. The pattern is exposed and transferred.

Next, each element of the alignment means of FIG. 1 will be described.

Reference numeral 53 is a detection optical system having the reference mark GS, 101 is an image pickup apparatus having the solid-state image pickup element 14, and GW.
Is a wafer mark (also referred to as a lattice mark or an alignment mark) provided on the wafer surface.

In FIG. 1, the relative position of the reticle R with respect to the projection lens 1, the detection optical system 53, and the image pickup device 101 is previously obtained by an appropriate detection system. And the detection optical system 5
The relative alignment between the reticle R and the wafer W is indirectly performed by detecting the positions of the projected images of the reference mark GS in 3 and the wafer mark GW of the wafer W on the imaging surface of the imaging device 101.

Next, a method for detecting the position of the wafer mark GW on the surface of the wafer W in the alignment and aligning the wafer W at a predetermined position will be described.

A light beam having a wavelength λ different from the wavelength of the exposure light emitted from the linearly polarized He-Ne laser-2 is made incident on the acousto-optic element (AO element) 3, and the lens is made by this AO element 3. The amount of light that goes to 4 is controlled so that the light is completely blocked in a certain state, for example.

The light beam passing through the AO element 3 is condensed by the lens 4, and then the illumination range is spatially limited by the field stop SIL arranged on the surface optically conjugate with the wafer W, and then the polarization beam splitter 5 is used. Is incident on. Then, it is incident on the polarization beam splitter 5. Then, it is reflected by the polarization beam splitter 5, and the λ / 4 plate 6, the lens 7, the mirror 8, the lens 19, and the projection lens 1 illuminate the wafer mark GW on the wafer W surface from the vertical direction.

The illumination light at this time is shown in FIG. 4A on the pupil plane (Fourier conversion plane of the image plane which is the wafer W plane) of the optical system composed of the projection lens 1, the lens 7 and the lens 19. ), Which is a light beam 41 as shown in FIG. (V and w in FIG. 4A are the coordinates of the pupil plane and represent the distribution of the incident angle of the illumination light with respect to the wafer W plane). The wafer mark GW on the surface of the wafer W is composed of a so-called grating mark composed of a diffraction grating with a pitch P as shown in FIG. 4 (B). The shaded area and the other areas in FIG. 4B are different in level on the surface of the wafer W, different in reflectance, different in phase, etc., thereby forming a diffraction grating.

The light beam reflected from the wafer mark GW passes through the projection lens 1, and then sequentially passes through the lens 19, the mirror 8, the lens 7, the λ / 4 plate 6, and the polarization beam splitter 5.
An aerial image of the wafer mark GW is formed at the position F through the lens 9 and the beam splitter 10.

The aerial image of the wafer mark GW formed at the position F is further passed through the Fourier transform lens 11 and the stopper-12.
Of the diffracted diffracted light from the wafer mark GW, for example, only the diffracted diffracted light beam whose angle on the wafer W corresponds to ± sin −1 (λ / P) is transmitted, and the image is picked up via the Fourier transform lens 13. An interference image of the wafer mark GW is formed on the solid-state image sensor 14 as described above.

This interference image shows a pitch P illuminated with monochromatic light.
Since it is the image of the wafer mark GW forming the diffraction grating, the signal light amount and the contrast are sufficiently high and stable as compared with the dark field image using mere scattered light.

Here, lens 19, lens 7 and lens 9
Composes a correction optical system 51, and corrects aberrations generated in the projection lens 1 with respect to the wavelength of the illumination light of the wafer mark GW, mainly axial chromatic aberration, spherical aberration and the like.

The correction optical system 51 does not need to perform aberration correction for all the captured aperture light fluxes, but only for the reflected and diffracted light fluxes passing through the stopper 12, it is necessary to perform aberration correction only. It has a simple structure.

This is because, when a so-called excimer stepper that uses light from an excimer laser is used as the exposure light, the amount of aberration generated with respect to the illumination light is larger than that of a projection exposure apparatus that uses light of the g-line or i-line. It is very effective because it is significantly larger.

On the other hand, a luminous flux from a light source 15 such as an LED, which emits a wavelength different from the wavelength of the illumination light of the wafer mark GW, is condensed by a condenser lens 16 and the reference mask 17
The reference mark GS formed on the surface is illuminated.
The reference mark GS is composed of, for example, a lattice mark similar to the wafer mark GW as shown in FIG. 5, in which the shaded areas are transparent areas and the others are opaque areas.

The light flux transmitted through the reference mark GS is the lens 1
The light beam from the LED 15 is reflected by the beam splitter 8, and the light beam from the He-Ne laser 2 is transmitted through the beam splitter 10 having an optical characteristic.
It forms an aerial image of Ku GS.

After that, the aerial image of the reference mark GS on the F surface is transformed by the Fourier transform lens 11, like the wafer mark GW.
An image is formed on the solid-state image sensor 14 by 13. The optical system 52 including the Fourier transform lenses 11 and 13 is well corrected for aberrations with two wavelengths of the illumination light to the reference mark GS and the wafer mark GW.

The wavelength of the light beam from the light source 15 may be the same as the wavelength of the illumination light of the wafer mark GW, and the beam splitter 10 may be a polarization beam splitter to combine the optical paths. At this time, the optical system 52 only needs to correct aberrations for the illumination wavelength of the wafer mark GW.

The stopper 12 acts as a pupil plane filter for the intensity distribution of the reflected light from the wafer mark GW when the illumination light is incident on the wafer W surface. As a result, the light flux incident on the solid-state image sensor 14 is only the ± n-order diffracted light from the wafer mark GW. However, n = 1, 2, 3, ...
・ It is.

Here, only the ± 1st-order diffracted light is used, and the wafer mark G as shown in FIG.
An interference image M of W is formed. The intensity distribution of the interference image M has a period T represented by T = β · P / 2 (β is an imaging magnification).
Represents a strength distribution having a positional deviation amount corresponding to the positional deviation of the wafer mark GW on the wafer W from the optical axis.

This interference image M forms a desired intensity distribution as a COS function having a period T, no matter how the step of the wafer mark GW or the film thickness of the resist covering the surface changes.

The diffracted light selected by the stopper 12 does not need to be limited to the ± 1st order diffracted light, but ± nth order diffracted light (N =
A pupil plane filter that transmits only 1, 2, 3 ...) may be used.

The fact that various orders can be selected also corresponds to changing the wavelength, and the detection rate of the wafer mark GW can be improved. For example, if the ± first-order diffracted light is small, the detection rate of the wafer mark GW can be improved by using the ± second-order diffracted light.

At this time, the coordinate position of the pupil plane filter becomes a position where the angle on the wafer W corresponds to ± sin-1 (nλ / P), and the image is formed on the solid-state image sensor 14 at the cycle T =.
It becomes a COS function of β · P / (2n).

Further, the image of the reference mark GS and the image of the wafer mark GW formed on the solid-state image pickup device 14 have the pitches Q and P of the reference mark GS and the wafer mark GW respectively. The same pitch T is obtained by determining in accordance with the imaging magnification. As a result, the FF described next
In the phase difference detection in the T processing, analysis at a fixed frequency is always possible.

In this apparatus, the projection lens 1, the detection optical system 53, and the solid-state image sensor 14 are previously detected by appropriate detecting means.
The relative position of the reticle R with respect to the image pickup apparatus 101 including

Therefore, the alignment is performed by detecting the positions of the images of the reference mark GS and the wafer mark GW on the image pickup surface of the image pickup device 101 including the solid-state image pickup device 14.

That is, the wafer mark GW on the wafer W surface
And the relative position of the reference mark GS on the surface of the reference mask 17 is detected, and the detection data at this time and the previously detected data of the relative position of the reference mark GS and the reticle R are used to make the reticle. The R and the wafer W are aligned with each other.

The image of the reference mark GS and the image of the wafer mark GW as the detection mark images formed on the solid-state image pickup device 14 are converted into a two-dimensional electric signal. FIG. 2 is an explanatory diagram on the image pickup surface including the detected mark image formed on the image pickup apparatus 101.

In FIG. 2, the interference image M is a detection mark, and when the direction for detecting the position of the pattern is the x direction in FIG. 2, the detection mark M has a cycle T in the x direction. The intensity distribution pattern is a COS function. By forming such a detection mark M, the cross-sectional shape of the pattern in the x direction has symmetry with respect to the pattern center xc to be obtained.

The pattern image converted into a two-dimensional electric signal by the image pickup device 101 is changed by the A / D conversion device 102 in FIG. 1 according to the optical magnification of the projection lens 1 and the detection optical system 53 and the pixel pitch of the image pickup surface. It is converted into a two-dimensional discrete electric signal sequence corresponding to the addresses in the XY directions of the pixels on the two-dimensional device by the determined sampling pitch Ps.

Reference numeral 103 in FIG. 1 denotes an integrating device, and FIG.
After setting a predetermined two-dimensional window containing the detection mark M in (A), pixel integration is performed in the window 20 in the y direction shown in FIG. 2 (A), and shown in FIG. 2 (B). It outputs a discrete electric signal sequence s (x) in the x direction.

Reference numeral 104 in FIG. 1 denotes an FFT operation device, which performs discrete Fourier transform of the input electric signal sequence s (x), transforms the electric signal sequence s (x) into the spatial frequency domain, and calculates its Fourier coefficient at high speed. To do. The method is known (for example, "Fast Fourier Transform" published by Science and Technology Publishing Co., E.
It is described in ORAN BRIGHAM, Chapter 10, Fast Fourier Transforms. ) N-point (N = 2r) fast Fourier transform (FFT), and when the sampling frequency is fs = 1 the complex Fourier coefficient X (k) of frequency f (k) = k / N is

[0069]

[Equation 1] Becomes

At this time, the intensity E of the spatial frequency f (k)
(K) and Θ (k) are respectively E (k) = ((Re (X (k))) 2 + (Im (X
(K)) 2)) 1/2 (2) Θ (k) = tan−1 (Im (X (k)) / Re
(X (k))) (3) (where Re (X (k)) and Im (X (k)) represent the real and imaginary parts of the complex number X (k).

FIG. 3A shows the spatial frequency intensity E (when the FFT is performed so as to include the entire detection mark with reference to the vicinity xs of the center of the detection mark M obtained by an appropriate detection means as a reference point. It represents the distribution of k).

Due to the periodicity of the pattern, the intensity of the spatial frequency f (h) = (Ps / T) .N peculiar to the pattern appearing in the one-dimensional discrete signal S (x) of the pattern becomes large,
A peak appears on the graph of FIG.

Reference numeral 105 in FIG. 1 is a frequency intensity detecting device, and the spatial frequency f (h) peculiar to the pattern is expressed by the equation (4).
The peak position Pi and the peak frequency f (Pi) are detected in the range of α near. Pi ＝ [k | max (E (k), f (h) ・ N-α <k <f (h) ・ N + α, k = 0,1,2, ・ ・ ・, N-1] (4 ) Here, α is a positive integer and is determined so that Pi becomes significant with respect to the pitch variation of the pattern.

On the other hand, in the range of α, even if the optical magnification fluctuates from a predetermined value due to some cause such as the adjustment state of the optical system, the frequency f (Pi) showing this peak is detected to detect the pattern. It is possible to correct the fluctuation of the optical magnification from the inherent spatial frequency. At the time of peak detection, if necessary, interpolation means (for example, in the spatial frequency domain)
The frequency resolution may be increased by using least square approximation, weighted averaging process, or the like.

Reference numeral 106 in FIG. 1 denotes a phase detector, which detects the peak f (Pi) of the spatial frequency intensity and the phase at the reference point xs of the frequency component in the vicinity thereof according to the equation (3).

Reference numeral 107 in FIG. 1 denotes a deviation amount detecting device,
The shift amount Δk in the real space is calculated from the phase by the equation (5) ΔK = (1 / 2π) · (N / K) · PS · ΘK (5).

FIG. 3B is a plot of the deviation amount Δk in the vicinity of the frequency region peculiar to the pattern. In general, as shown in FIG. 3B, the deviation amount Δk in the vicinity of the frequency peculiar to the mark shows a constant and stable value. Therefore, the calculated averaging amount Δk is weighted and averaged by the frequency intensity. , The deviation amount Δc of the pattern center from the reference point xs is detected according to the weighted averaging process (equation (6)).

Δc = [Σ {E (k) · Δk}] / [Σ {E (k)]] (6) f (Pi) · N−α <K <f (Pi) · N + α f ( P
i) .N-.alpha. <K <f (Pi) .N + .alpha .. FIG. 3C shows the result of this weighted averaging process. By doing so, the shift amount Δc is detected while focusing on only the signal of the detection pattern to improve the S / N ratio, reduce the calculation amount, and stabilize the detection by the weighted averaging process.

Up to now, the FFT has been used to calculate the Fourier coefficient, but by utilizing the fact that the spatial frequency peculiar to the pattern does not largely change, only the Fourier coefficient of the required frequency component is directly calculated from the equation (1). It may be done.

In this case, since the number of sampling points can be arbitrarily selected, considering that the frequency component by the discrete Fourier transform is represented by k / N, k / N becomes the closest to the frequency component to be obtained. By optimally selecting the number of sampling points N as described above, the spatial frequency peculiar to the pattern can be obtained more accurately. In addition, by reducing the frequency components used for detection, the amount of calculation can be reduced.

In this way, the wafer marker on the surface of the wafer W is
The relative position between the reference mark GW and the reference mark GS on the surface of the reference mask 17 is detected, and the detection data at this time is detected and the relative position between the reference mark GS and the reticle R detected in advance. The position data is used to align the reticle R and the wafer W.

Although the positional deviation between the wafer mark GW and the reference mark GS was detected, the reticle mark on the reticle R was connected to the solid-state image sensor 14 via the optical system instead of the reference mark GS. Alternatively, the positional deviation between the reticle mark and the wafer mark GW may be directly detected by forming an image.

Further, instead of forming the image of the reference mark GS on the solid-state image pickup device 14, the reference mark is virtually set on the image memory for recording the signal of the solid-state image pickup device 14, and the reference mark is set.
It may be used as a substitute for GS.

Further, the grid marks provided on the object to be aligned are provided in the X and Y directions, respectively, and the respective positions in the X and Y directions are detected by different optical systems, or the grid marks are arranged in a checkered pattern. It is also possible to detect them by a single optical system by forming them simultaneously or by switching the X and Y directions simultaneously. Such a thing is similarly applicable to the examples described below.

Next, a first embodiment of the projection exposure apparatus of the present invention will be described. In this embodiment, the stopper N is taken in N
The feature is that A is specified.

In the structure shown in FIG. 1, if the take-in NA (aperture shape) of the stopper 12 is set to the opening 12b as shown in FIG. 6, that is, the interference fringes formed on the surface of the image pickup means 14 are evaluated and alignment is performed. When a relatively large NA is provided in the direction orthogonal to the performing direction,
Unwanted image information in or around the lattice mark GW provided on the wafer W (object) is transmitted to the image pickup means 14, and S
The / N ratio may be reduced.

Specifically, a random detection error occurs during alignment in a process where surface roughness is severe, or a measurement error occurs when dust or defects are present in or around the lattice mark GW. It may come.

Further, in order to reduce the influence of speckle noise generated by using a monochromatic light source as illumination light, the amount of light is generally greatly lost by setting some kind of rocking means typified by a diffusion plate. Often
In some cases, the S / N ratio of the detection signal is lowered, or particularly when the condition is low such as the low step process or the low substrate reflectance, the illumination light amount may be insufficient and detection may be impossible.

Further, when the taken-in NA for the reflected diffracted light becomes large, the aberration generated in the projection lens with respect to the illumination wavelength of the non-exposure TTL non-exposure light is particularly large when the excimer laser is used as the exposure light. In some cases, the optical system for compensating for the aberration generated in the projection lens becomes complicated, and it becomes impossible to construct a realistic correction optical system in some cases.

Therefore, in this embodiment, the above-mentioned problems are solved by setting the opening condition of the stopper 12 appropriately as follows, and highly accurate alignment is realized.

In general, assuming that the angle of acceptance of each reflected diffracted light of the stopper on the wafer side is NA, the wavelength of the illumination light is λ, and the cutoff frequency (transmittable limit frequency) of the optical system is NC, NC is NC = 2 It is represented by NA / λ.

On the other hand, in the present invention, the spatial frequency of the interference fringes formed on the image pickup means 14 to be detected is T, and the wafer W is
When the absolute value of the optical magnification between the image pickup means 14 is set to β and the spatial frequency T is converted to the spatial frequency TW on the wafer W, TW = T · β.

Therefore, the cutoff frequency NC is set to the spatial frequency T
If NA is set so as to be smaller than W, it is possible to obtain a mode in which unnecessary image information in or around the lattice-shaped marks provided on the object is not transmitted to the imaging means.

Therefore, in this embodiment, the take-in NA of the stopper 12 is set as 0 <NA <T · β · λ / 2.

FIG. 7 is a plan view of the main part of the stopper 12 of this embodiment, and FIG. 8 is an explanatory view of the main part of the optical path of FIG. 1 when the stopper 12 of this embodiment is used. Figure 7
In the figure, 12a indicates the opening of the stopper 12.

In this embodiment, the wafer W to be aligned is
The wavelength λ of the illumination light for illuminating the lattice-shaped mark GW provided on the (object) with monochromatic light is λ = 633 nm, the absolute value β of the optical magnification between the wafer W and the image pickup means 14 is β = 100x,
The spatial frequency T of the interference fringes formed on the image pickup means 14 is T
= 2, the uptake NA derived from the above relational expression is N
At the wafer W side for each reflected diffracted light of the stopper 12 which selectively takes out the reflected diffracted light so that ± 1st-order reflected diffracted light selected by the stopper 12 is sufficiently passed while satisfying A <0.063. Uptake NA, NA =
It is set to 0.03.

In this embodiment, the stopper N is taken in N
By constructing A so as to satisfy the above conditional expression, unnecessary image information in or around the lattice mark GW provided on the wafer W (object) is transmitted to the image pickup means 14.
At the same time as preventing the N ratio from decreasing, some kind of oscillating means typified by a diffusing plate for eliminating the influence of speckle noise is not required, thereby preventing the loss of light quantity.

Further, since the area for correcting the aberration generated in the projection lens is limited to only the NA portion for taking in the signal light, the correction optical system is not necessary, and it is realized by a very simple structure such as a single lens. Allowed and consequently high S
The / N ratio enables stable position measurement at all times. In particular, a projection exposure apparatus using an excimer laser as exposure light can realize a highly accurate alignment method by the non-exposure TTL method. Further, in the present invention, all the image pickup means detect a two-dimensional video signal and electrically integrate them in one direction to obtain a one-dimensional projected integrated signal. For example, a cylindrical lens is provided in front of the image pickup means. It is also possible to easily configure the optical system so that it is optically integrated in one direction to obtain a one-dimensional projection integrated signal by the one-dimensional imaging means.

[0099]

According to the present invention, the lattice marks for alignment (alignment) provided on the wafer surface by the configuration in which the respective elements are appropriately set as described above are used as a single color having a wavelength different from the wavelength of the exposure light. By illuminating with light, an interference image based on the reflected diffracted light generated at this time is formed on the surface of the image pickup means, and the image signal obtained by the image pickup means is used to align the reticle and the wafer at high speed. Done with high precision,
A positioning method capable of high-resolution projection exposure and a projection exposure apparatus using the same can be achieved.

In particular, according to the present invention, the stopper N for selectively taking out the reflected diffracted light takes in N for each reflected diffracted light.
By appropriately setting A as described above, it is possible to prevent unnecessary image information in or around the lattice-like marks provided in the object form from being transmitted to the image pickup means and lowering the S / N ratio, and at the same time speckle. By eliminating the need for any oscillating means to eliminate the effect of noise, loss of light quantity is prevented, and an optical system that corrects aberrations generated in the projection lens can be realized with a very simple structure. As a result, stable position measurement is always possible with a high S / N ratio.

[Brief description of drawings]

FIG. 1 is a schematic diagram of a main part of a first embodiment of a projection exposure apparatus as an example for explaining the present invention.

FIG. 2 is an explanatory diagram related to an interference image of a lattice-shaped mark on the surface of the image pickup unit in FIG.

FIG. 3 is an explanatory diagram regarding signal processing of an interference image.

FIG. 4 is an explanatory diagram of a pupil plane and a grid mark of the projection exposure apparatus.

FIG. 5 is an explanatory diagram of a reference mark.

FIG. 6 is an explanatory view showing an opening of a stopper.

FIG. 7 is an explanatory view of a stopper according to the present invention.

FIG. 8 is a development schematic diagram of a part of an optical system of the projection exposure apparatus of the present invention.

FIG. 9 is an explanatory view of a conventional OFF-AXIS type alignment device.

FIG. 10 is an explanatory view of a conventional aligning device of exposure light TTL method.

FIG. 11 is an explanatory diagram of a conventional non-exposure light TTL alignment device.

FIG. 12 is an explanatory diagram of a wafer mark image and a wafer mark sectional shape.

[Explanation of symbols]

IL illuminator R reticle W wafer GW wafer mark (grid mark) GS standard mark 1 Projection lens 2 laser 3 AO element 4 lenses 5 Beam splitter 6 λ / 4 plate 7 lenses 8 Reflective member 10 Beam splitter 11, 13, 21 Fourier transform lens 12 stopper 14,22 Group image sensor 51 Correction optical system 52 Optical system 53 Detection optical system

─────────────────────────────────────────────────── ─── Continuation of front page (56) Reference JP-A-2-206706 (JP, A) JP-A-3-282715 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H01L 21/027

Claims (2)

(57) [Claims] 1. A projection exposure apparatus for projecting and exposing a pattern formed on a reticle illuminated with exposure light onto a wafer surface via a projection lens, wherein a grid-like mark provided on the wafer surface is exposed via the projection lens. Illuminating hand to illuminate with monochromatic light
Step and the reflection from the grid-like marks that have passed through the projection lens.
A reflecting member that guides the diffracted light to the outside of the exposure light path;
Select ± n-order light (n = 1, 2, 3 ...)
Projection optical means provided with a stopper for projecting and taking out the projection optical means
Same as the reticle by the projection optical means using the nth order light.
It has an imaging device that forms an interference image on a well-planned surface.
Then, the stopper takes in the wavelength NA of the monochromatic light from the illuminating means with respect to each reflected diffracted light on the wafer side, and the absolute value of the projection magnification of the lattice mark on the surface of the imaging means. β, 0 <NA <T · β · λ / 2 is satisfied, where T is the spatial frequency of the interference image formed on the surface of the image pickup means, and the interference image formed on the surface of the image pickup means is satisfied. A one-dimensional projection integration signal obtained by electrically or optically projecting integration in one direction of two-dimensional coordinates obtained from the image pickup signal is converted into a spatial frequency domain by orthogonal transformation, and the 1 The wafer is aligned with a predetermined position by selecting a spatial frequency component that appears uniquely from the interference image based on the periodicity of the lattice-shaped mark from the three-dimensional projection integrated signal and detecting the position of the lattice-shaped mark. Projection exposure equipment characterized by . 2. The projection lens is aberration-corrected with exposure light, the wavelength of the monochromatic light from the illumination means is different from the wavelength of the exposure light, and the projection optical means is illuminated with the monochromatic light. The projection exposure apparatus according to claim 1, further comprising a correction optical system that corrects various aberrations that occur when the lattice-shaped marks on the wafer surface are projected onto a predetermined surface through the projection lens. .