JPH09246175A - Position detecting method, its device, exposing method and its device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a position of all fine patters on a substrate with high preci sion. SOLUTION: By an interference alignment method by a ONdimensional detecting method, a plurality of lattice mark positions ΔXjn' on a substrate according to obtained radiation intensity signals IA, IB are detected by a circuit SA and a circuit CA (every wavelength). Further, in the circuit SA, based on changes according to relative scanning of the radiation intensity signals IA, IB and design data of the lattice mark, an undercut corresponding amount δn of each lattice mark is calculated every wavelength as surplus with respect to each semi-wavelength of incident light beams near a mark face of an undercut. In a circuit WEGA, a small weight is given to a detection position of a mark having a large detection error corresponding to the undercut corresponding amount of each lattice mark and a large weight is given to a detection position of a mark having a small error, so that positions (arrangement coordinates) of all fine patterns on the substrate are calculated by a statistical processing by a minimum square-law method.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a position detecting device and its device, an exposure method and its device, and more specifically, it detects the positions of a plurality of position measuring marks and detects the plurality of detected positions. Position detection method and apparatus for statistically calculating the position of the fine pattern area on the substrate, and overlay baking exposure while sequentially positioning each fine pattern area on the substrate to the exposure position based on the calculated array coordinates. The present invention relates to an exposure method and an apparatus therefor.

[0002]

2. Description of the Related Art Photomasks, reticles, etc. (hereinafter, "reticle" is used as an example) when manufacturing semiconductor devices, liquid crystal display devices, etc. by a photolithography process.
There is used a projection exposure apparatus for projecting the pattern image of (1) on each shot area on a wafer coated with a photosensitive material via a projection optical system. As a projection exposure apparatus of this type, in recent years, a wafer is placed on a two-dimensionally movable stage, and the wafer is stepped by this stage.
A so-called step-and-repeat type exposure apparatus, in particular, a reduction projection type exposure apparatus (stepper), which repeats an operation of sequentially exposing a pattern image of a reticle to each shot area on a wafer is frequently used.

For example, since a semiconductor element is formed by superposing a plurality of layers of circuit patterns on a wafer and exposing the same, when projecting and exposing the circuit patterns of the second and subsequent layers onto the wafer, the circuits on the wafer are already exposed. Aligning each shot area where the pattern is formed with the pattern image of the reticle,
That is, alignment between wafer and reticle
Need to do exactly. In order to perform this alignment, a position detection mark (alignment mark) formed (exposure transferred) in the previous process is formed on the wafer, and by detecting the position of this alignment mark,
The accurate position of the wafer (circuit pattern on the wafer) can be detected.

In recent years, an alignment mark on a wafer (or a reticle) is formed into a one-dimensional or two-dimensional lattice shape having irregularities, and two coherent beams symmetrically inclined in the periodic direction are projected on the lattice mark. A method of interfering two diffracted light components generated in the same direction from the grating mark to detect the position or positional deviation of the grating mark in the periodic direction (hereinafter referred to as “interferential alignment method”) is, for example, (A) It has been proposed in Japanese Laid-Open Patent Application No. 61-208220, (B) Japanese Laid-Open Patent Application No. 61-215905, and the like. Of these, publication (A) discloses a homodyne system in which the frequencies of two symmetrical coherent beams are the same, and publication (B) describes
A heterodyne system in which a constant frequency difference is provided between two symmetrical coherent beams is disclosed.

In the heterodyne system, a frequency difference between two light-transmitting beams is used as a reference AC signal, and a signal obtained by photoelectrically detecting interference light (beat light) of two diffracted light components generated from a grating mark and a reference AC signal. Of the grid mark is detected and detected as a positional deviation amount from the reference point in the periodic direction of the lattice mark.

The optical system for realizing the interferential alignment method can be roughly divided into three types depending on the method of injecting the coherent beam (laser beam) to the grating mark.

First, two laser beams are incident from ± N-order directions which are symmetric with respect to the period (= P) of the grating mark, and the grating mark interferes with an amplitude period of P / N. A diffracted light that forms a fringe and is reflected / diffracted vertically upward from this (combined light flux of ± N-order diffracted light of both incident lights: N is a natural number) is received, and the position of the grating mark is detected based on the change in the light amount. Type (hereinafter, abbreviated as “± Nth order detection method”).

Second, the two beams are made to enter symmetrically from the direction of ± N / secondary with respect to the period of the grating mark to form an interference fringe with an amplitude period of 2P / N on the grating mark, and one of the two beams is formed. A combined light flux of the 0th order diffracted light (regular reflection light) of the (first) incident light and the Nth order diffracted light of the other (second) incident light, the 0th order diffracted light of the second incident light, and the first incident light Of the N-th order diffracted light is received independently, and the average value of the two detection positions obtained from each phase of the light amount change of both light beams is used as the position of the lattice mark (hereinafter, referred to as "0N-order detection method"). "" Is abbreviated)
It is.

Thirdly, only one beam is incident on the grating mark and two ± N-order diffracted lights that are reflected / diffracted are imaged on the reference grating, and the change in the amount of transmitted light from the reference grating is analyzed. This is a type for detecting the position of a lattice mark (hereinafter, abbreviated as "reference lattice method").

In principle, there is no difference in the lattice mark detection positions by these three methods. However, ±
The Nth-order detection method and the 0Nth-order detection method can be applied to any of the heterodyne method and the homodyne method, but the reference grating method has only one incident light beam and is necessarily limited to the homodyne method.

Further, the light source of the interferometric alignment method is a plurality of beams having different wavelengths or a white beam,
The problem that the light source (detection light) is monochromatic, that is, the problem that a large position detection error occurs due to the asymmetry with respect to the step (depth) of the lattice mark (position detection mark) and the fluctuation of the resist thickness is reduced. A measure is also proposed by (C) Japanese Patent Laid-Open No. 6-82215.

Such position detection marks are exposed and transferred for each shot simultaneously with the circuit pattern in the previous exposure process. Therefore, at the time of exposure of the next circuit pattern, the position detection mark in each of these shots may be detected, and a new pattern may be superposed based on the detected position. ), The time required to detect the mark position is increased, and the processing capacity (throughput) of the exposure apparatus decreases.

Therefore, as an alignment method at the time of exposure,
In the stepper, an enhanced global alignment (hereinafter referred to as "EGA") method as disclosed in, for example, Japanese Patent Laid-Open No. 61-44429 is generally adopted. This EGA selects about ten specific position detection marks (alignment marks) formed for each shot on the wafer, detects their positions, and statistically processes the obtained detection positions. The "alignment coordinate system" of all shots is obtained, and overlay exposure is performed according to the array coordinate system.

[0014]

In general, no matter what kind of position detection sensor (for example, the above-mentioned interference type alignment sensor) is used, the detection position is accompanied by some detection error. For example, in the interferometric alignment method, the detection error is caused by the asymmetry of the lattice mark and the fluctuation of the resist thickness.

More specifically, the marks for alignment and position measurement (alignment marks) formed on the surface of a wafer or the like are generally formed with a minute step on the surface. There is some asymmetry due to the wafer process such as etching and sputtering, or uneven coating of the photoresist layer. The asymmetry causes a decrease in accuracy when detecting the mark position.

Particularly, in the interferometric alignment method in which the mutual interference light of two diffracted lights generated from the grating mark is photoelectrically detected and the photoelectric signal is used, the asymmetry of the grating mark is caused by the asymmetry of the amplitude reflectance of the mark itself. And the position detection accuracy is deteriorated. That is, when the depth (step) of the groove bottom of the line forming the grating mark has a difference in the grating period direction or there is a partial difference in the thickness of the resist layer, the absolute amplitude reflectance of the mark itself is The value and the phase become asymmetric depending on the depth of the groove bottom and the change of the resist thickness.
As a result, the diffracted light generated from the grating mark also has different intensity and phase between the positive order generated in the right direction and the negative order generated in the left direction with respect to the 0th order light, for example. Among them, the difference in intensity hardly deteriorates the position detection accuracy, but the change in phase has a great influence on the position detection accuracy.

Further, the amplitude reflectance and diffraction efficiency of the grating mark greatly vary depending on the step of the mark and the resist thickness. This is because the step difference of the mark changes the phase difference between the light reflected on the upper surface of the mark and the light reflected on the lower surface of the mark, and the multiple interference effect of the resist changes due to the change of the resist thickness. Also, when the wavelength of the detection light changes, these effects also fluctuate greatly, that is,
Even if the position of the same mark is detected, if the wavelength is different, the position detection error differs due to the influence of the step of the mark or the like.

On the other hand, according to the EGA method, it can be said that these position detection errors can be averaged and reduced.

However, the averaging by the conventional EGA method is not performed in consideration of the magnitude of the detection error included in the detection position of each mark, and the conventional position detection sensor has a large or small detection error. There is no ability to estimate the degree, and thus the averaging effect of the EGA method is naturally limited.

The present invention has been made under the above circumstances, and an object thereof is to provide a position detecting method and apparatus capable of detecting the positions of all fine patterns on a substrate with higher accuracy. It is in.

[0021]

By the way, the position detection accuracy, such as the variation of the amplitude reflectance and diffraction efficiency of the grating mark due to the variation of the mark step and the resist thickness as described above, the position detection accuracy, is the mark itself under various mark conditions. A simulation can be performed by assuming the amplitude reflectance of

An example of such a simulation result will be described with reference to FIG. This simulation result assumes that a grating mark on a wafer covered with a resist layer is irradiated with a coherent light-transmitting beam having a constant frequency difference from two symmetrical directions, and ± is generated vertically from the grating mark. Mutual interference light of first-order diffracted light,
That is, it is obtained by observing (calculating) the state (amplitude, phase, etc.) of the interference beat light.

The shape of the lattice mark on the wafer assumed in the above simulation is such that the convex portion and the concave portion are arranged at a width of 4 μm and a period P = 8 μm in the measurement direction.
The cross section of the lattice mark is shown in FIG. 9 (1), and an enlarged view of one period thereof is shown in FIG. 9 (2). The step h of the mark and the resist thickness d are used as parameters, and the concave portion of the mark has an inclination t = 0.1
% Taper. In the concave portion of the mark, the resist surface was dented by 0.3 · h, the shape of the resist surface was a sine function (period P), and the refractive indexes of the wafer and resist were 3.5 and 1.66, respectively. In addition,
In FIG. 9 (1), the number of grid marks is 5 due to space limitations.
Although the book is a book, the actual number of grid marks is generally larger than that.

In the assumed detection optical system, the wavelength of the detection laser beam is 633 nm (He-Ne laser), and the two beams are symmetrically incident from the ± first order direction with respect to the mark period P. , Which detects the position by receiving a light beam that is reflected / diffracted vertically upward (combined light beam of ± 1st-order diffracted light of both incident lights) (the above-mentioned “± 1st-order detection method” (“± Nth-order detection method N”) =
1)). However, as a detection optical system, a “01th-order detection method” in which two beams are symmetrically incident from the ± 0.5th-order direction with respect to the mark period P, or a “reference grating method” using ± first-order diffracted light is used. However, the result is exactly the same as described above.

In FIGS. 8A and 8B, the horizontal axis represents the resist thickness d [μm] and the vertical axis represents the position detection error [μm].
Is shown. In FIG. 8 (1), the mark step h is 50, 100,
In the case of 150 and 200 nm, the mark step h is shown in Fig. 8 (2).
, The simulation results for 250, 300, 350, and 400 nm are shown, respectively.

Although the detection error varies depending on the change in the resist thickness, it can be seen that when the mark step h is 100 and 300 nm, the variation is small and the absolute accuracy is good.

Since the wavelength of the detection light is 633 nm and the refractive index of the resist is 1.66, the wavelength of the detection light near the mark surface (in the resist) is 381 nm.
0 and 300 nm are approximately equal to 1/4 and 3/4, respectively. Although not shown, the step difference of the wavelength of the detection light in the resist is 5/4 and 7/4 ..., That is, (2m + 1)
As a result of simulation, it was found from the simulation result that the detection error at the mark having a step difference of / 4 times (m is an integer of 0 or more) is extremely small.

From the above simulation results, the step of the position detection mark is (2 m +
It can be seen that if 1) / 4 times, the detection error in the interference type alignment is extremely small. Therefore, if it is possible to detect the mark step amount at the same time as detecting the mark position, it is considered that the size of the detection error included in the detection position can be estimated to some extent accurately by the step amount.

According to the present invention, in the interferential alignment method using a single wavelength of illumination light, the step of the mark has a wavelength of the illumination light (more precisely, a wavelength in the resist of the illumination light) of (2 m
The present invention is conceived based on the above simulation result that the position detection error hardly occurs when it is +1) / 4 times (m is an integer of 0 or more). Adopt a method and configuration.

The invention according to claim 1 is a fine pattern,
And a plurality of (M) position detection lattice marks that repeat the concave and convex portions with a periodicity (period = P) in a specific direction.
The positions of the plurality of grid marks in the periodic direction are detected for each of the substrates, and the detected positions are statistically processed to detect the position of the fine pattern on the entire surface of the substrate. In the position detecting method, a pair of coherent light beams are respectively incident on the plurality of grating marks, the period of amplitude distribution is 2P / N (N is a natural number), and the period direction is each of the gratings. The first interference fringes are formed in the periodic direction of the marks, and the interference fringes and the grating mark are relatively scanned in the periodic direction.
A first light beam that is a combined light beam of the specularly reflected light of the first light beam and the Nth-order diffracted light of the second light beam among the reflected / diffracted light of the incident light beam by the grating marks at the plurality of locations;
And the second combined light beam, which is the combined light beam of the specularly reflected light of the second light beam and the N-th order diffracted light of the first light beam, are received sequentially and separately, and photoelectric conversion is performed. And a second step of sequentially detecting the positions of the plurality of grid marks from changes in the light amount signals of the first and second combined light fluxes sequentially obtained in the second step due to the relative scanning. And a change in the light amount signals of the first and second combined luminous fluxes sequentially obtained in the second step, which are caused by the relative scanning, and the design of the respective lattice marks. A fourth step of sequentially calculating, based on the data, the step difference of each of the grating marks as a remainder obtained by dividing the incident light beam by half the wavelength in the medium in the vicinity of the surface of each grating mark; Jth (where j is 1 to M) lattice mark The step equivalent amount, of the wavelength at the j-th grating mark near the surface of the incident light beam 1 /
When it is close to 4, a large weight is given to the detection position of the j-th grating mark, and when it is far from 1/4 of the wavelength, a small weight is given to the detection position of the j-th grating mark. And a fifth step of statistically processing the detection positions of the plurality of grid marks to calculate the position of the fine pattern on the entire surface of the substrate.

According to this, in the first to third steps, the same detection as in the interferometric alignment method by the 0N-order detection method is performed for each of the lattice marks, and the obtained 2
Two light quantity signals (light quantity signals of the first and second combined luminous fluxes)
Thus, the positions of the respective grid marks are detected as in the conventional case.

Further, in the fourth step, the step-corresponding amounts of the respective grid marks at a plurality of positions are changed by relative scanning of the light amount signals of the first and second combined light fluxes sequentially obtained in the second step, and the respective lattice marks. On the basis of the design data, the step difference of each grating mark is sequentially calculated as a remainder obtained by dividing the incident light beam by half the wavelength in the medium in the vicinity of the grating mark surface. This is to detect each wavelength as "a step difference equivalent to a remainder obtained by dividing the step difference by 1/2 of the wavelength in the resist," which becomes clear as a result of the above simulation. The step equivalent amount may be detected in any form of a step as a surplus or a phase amount of light corresponding to this. As a result, whether or not the step of the grating mark for position detection meets the condition of (2m + 1) / 4 times the in-resist wavelength of the detection light, or the grating mark for which the in-resist wavelength of the detection light should be detected It is possible to easily determine whether or not the condition of 4 / (2m + 1) times the step difference is met.

Then, in the fifth step, the step-equivalent amount of the j-th (j is 1 to M) grating mark in the plurality of places is 1 wavelength of the incident light beam near the surface of the j-th grating mark. When it is close to / 4, a large weight is given to the detection position of the j-th grating mark, and when it is far from 1/4 of the wavelength, a small weight is given to the detection position of the j-th grating mark. Then, the positions of the fine patterns on the entire surface of the substrate are calculated by statistically processing the detection positions of the plurality of grid marks. Here, 1/4 of each wavelength means the above (2m + 1)
This corresponds to the case where m = 0 of / 4, but since the step equivalent is a surplus for a half wavelength as described above, it is not necessary to consider when m> 0. Therefore, the step equivalent is the grating mark surface of the incident light beam. To be close to 1/4 of each wavelength in the vicinity means that the step equivalent is consequently close to (2m + 1) / 4 times of each wavelength in the vicinity of the grating mark surface of the incident light beam. Therefore, in the fifth step, by the new statistical processing in which the detection value of the mark having a large detection error estimated from the step equivalent amount is given a small weight, and the detection value of the mark having a small detection error is given a large weight, Since the position of the fine pattern on the entire surface of the substrate is calculated, the position (arrangement coordinates) of the fine pattern on the entire surface of the substrate can be calculated with higher accuracy than in the conventional case.

In this case, as in the invention described in claim 2, in the first step, a plurality of pairs of coherent light beams are made incident on the plurality of grating marks at different wavelengths, respectively, and the amplitude distribution is obtained. Is 2P / N (N is a natural number)
To form interference fringes whose periodic direction is equal to the periodic direction of each of the lattice marks, and to relatively scan each interference fringe and each of the lattice marks in the periodic direction. The process is performed for each wavelength to detect the position and the step equivalent amount of each grating mark for each wavelength, and in the fifth step, the position detection result of the grating mark and the step equivalent amount are detected at each wavelength. Using the result, the weighted statistical processing may be performed on the detection positions of the plurality of grid marks for each wavelength, and the position of the fine pattern on the entire surface of the substrate may be calculated. According to this, the detection light flux is converted into a plurality of wavelengths, and using the position detection result and the step equivalent amount detection result at each wavelength,
Since the new statistical processing for weighting is performed, the position (array coordinate) of the fine pattern on the entire surface of the substrate can be calculated with higher accuracy.

According to a third aspect of the present invention, a pattern formed on a mask is transferred to each of the shot areas while sequentially positioning a predetermined shot area having a fine pattern formed on a photosensitive substrate at an exposure position. An exposure method for performing overprint exposure, wherein the exposure method is performed prior to the start of exposure.
After calculating the array coordinate of each shot area on the photosensitive substrate by the position detecting method described in 1), the overprint exposure is performed while sequentially positioning the shot area at the exposure position according to the calculated array coordinate.

According to this, as described above, the arrangement coordinates of the shot areas on the photosensitive substrate are calculated with higher accuracy than before,
Overlapping exposure is performed while sequentially positioning the shot areas at the exposure positions according to the calculated array coordinates, and as a result, a more highly accurate overlay is realized.

The invention according to claim 4 is a fine pattern,
And a plurality of (M) position detection lattice marks that repeat the concave and convex portions with a periodicity (period = P) in a specific direction.
The positions of the plurality of grid marks in the periodic direction are detected for each of the substrates, and the detected positions are statistically processed to detect the position of the fine pattern on the entire surface of the substrate. And a period of amplitude distribution of 2P / on the lattice marks at the plurality of locations.
N (where N is a natural number), each of which transmits a pair of coherent light beams so as to form interference fringes whose periodic direction is equal to the periodic direction of each grating mark; and a plurality of gratings. Reflection of the incident light beam by the mark
Of the diffracted light, a first combined light beam that is a combined light beam of the specularly reflected light of the first light beam and the Nth-order diffracted light of the second light beam, the specularly reflected light of the second light beam, and the A light receiving means for separately photoelectrically converting a second combined light beam, which is a combined light beam of the first light beam and the N-th order diffracted light; and a relative relative scanning means for scanning the lattice marks and the interference fringes in the periodic direction. Scanning means; the first obtained from the light receiving means,
Position detecting means for respectively detecting the positions of the lattice marks at the plurality of locations based on changes in the light amount signal of the second combined light flux due to the relative scanning; Based on the change of the light quantity signals of the first and second combined light fluxes obtained by the relative scanning and the design data of each of the grating marks obtained by the above, the step of each of the grating marks is set to each grating of the incident light beam. Step difference detecting means for detecting each as a remainder divided by half of the wavelength in the medium near the mark surface; and the step equivalent amount of the j-th (j is 1 to M) grating mark in the plurality of places, When it is close to ¼ of the wavelength of the incident light beam in the vicinity of the surface of the j-th grating mark, a great weight is given to the detection position of the j-th grating mark, and the distance from ¼ of the wavelength If there is And a calculating means for calculating the position of the fine pattern of the j-th detected position of the grating mark with a small weight by statistically processing the detected position of the grating mark of the plurality of positions the entire surface of the substrate.

According to this, interference fringes having a period of amplitude distribution of 2P / N (N is a natural number) and a period direction of which is equal to the period direction of the grating mark are formed on the grating mark at a plurality of positions by the light transmitting optical system. In order to form each of them, a pair of coherent light beams are incident, and the relative scanning means relatively scans these interference fringes and each grating mark in the periodic direction.

Next, in the light receiving means, of the reflected / diffracted light of the incident light beam by the grating marks at a plurality of places, the first
First light flux which is a light flux of the specular reflection light of the light beam and the N-th order diffracted light of the second light beam, the specular reflection light of the second light beam, and the N-th order diffracted light of the first light beam. And a second combined light beam, which is a combined light beam of, are separately photoelectrically converted. The position detecting means detects the positions of the grid marks at a plurality of locations from changes in the light amount signals of the first and second combined light fluxes obtained by the light receiving means due to the relative scanning, and the step detecting means detects a plurality of locations of the grid marks. The step-corresponding amount of each lattice mark is made incident on the basis of the change in the light amount signals of the first and second combined light fluxes obtained by the light receiving means due to the relative scanning and the design data of each lattice mark. It is calculated as the remainder divided by half the wavelength of the light beam in the medium near the surface of each grating mark.

Then, in the calculating means, j in a plurality of places
If the step equivalent amount of the th (j is 1 to M) grating mark is close to ¼ of the wavelength of the incident light beam near the surface of the jth grating mark, detection of the jth grating mark When the position is heavily weighted and is apart from 1/4 of the wavelength, a small weight is given to the detection position of the j-th grating mark and statistical processing is performed on the detection positions of a plurality of grating marks to perform the entire substrate surface. The position of the fine pattern is calculated.

Therefore, as in the case of claim 1,
The position (arrangement coordinate) of the fine pattern on the entire surface of the substrate can be calculated with higher accuracy than in the past.

Here, the position detecting means, as in the invention described in claim 5, respectively obtains based on each phase of each light quantity signal change of the first combined light beam and the second combined light beam due to relative scanning. It is desirable to detect the average value of the detection positions as the position of the lattice mark.

Further, as in the invention described in claim 6, the step detecting means has the ratio of the widths of the concave portions and the convex portions of each lattice mark, the first combined light flux and the second combined light flux associated with the relative scanning. The step-corresponding amount of each lattice mark may be calculated based on the phase difference of each light amount signal change and the contrast. Alternatively, as in the invention described in claim 7, when a light quantity ratio measuring means for measuring the light quantity ratio between the 0th-order diffracted light reflected and diffracted by the grating mark and the Nth-order diffracted light is further provided, each step detecting means is The ratio of the widths of the concave and convex portions of the lattice mark, the phase difference between the changes in the light amount signals of the first combined light beam and the second combined light beam due to relative scanning, and the light amount ratio obtained by the light amount ratio measuring means. The step-corresponding amount of each lattice mark may be calculated based on In the latter case, since the light quantity ratio between the 0th-order diffracted light and the Nth-order diffracted light is directly measured, the step equivalent amount can be calculated with higher accuracy than in the former case in which the light quantity ratio is calculated based on the contrast. It can be calculated.

Further, as in the invention described in claim 8, in place of the light-transmitting optical system, the period of the amplitude distribution is 2P / N (N is a natural number) in sequence on the lattice marks at a plurality of positions. In order to form interference fringes that are equal in the periodic direction of each of the grating marks, a light-transmitting optical system for injecting a plurality of pairs of coherent light beams having different wavelengths is provided, The position detecting means and the step detecting means perform the respective processes for each of the plurality of different wavelengths, and the calculating means uses the position detecting result and the step detecting result of each of the grating marks at the respective wavelengths. It may be configured to perform statistical processing. In such a case, the detection light flux is converted into a plurality of wavelengths, similarly to the invention described in claim 2,
The new statistical processing for weighting is performed using the position detection result at each wavelength and the detection result of the step equivalent amount, so the position (array coordinate) of the fine pattern on the entire surface of the substrate is calculated with higher accuracy. can do.

In this case, as in the invention described in claim 9, the light-transmitting optical system time-divisions a plurality of pairs of coherent light beams having different wavelengths on the plurality of grating marks. You may make it incidentally. In such a case, the configurations of the light receiving optical system and the signal processing system are simplified. Here, the term “time-divisionally” includes a case where pairs of light beams having different wavelengths are alternately incident on the grating mark.

According to a tenth aspect of the present invention, a predetermined shot area on the photosensitive substrate on which a fine pattern is formed is sequentially positioned at an exposure position, and the pattern formed on the mask is transferred to each shot area. An exposure apparatus for performing overprint exposure, comprising the position detection device according to any one of claims 4 to 9 for detecting an array coordinate system of each shot area on the photosensitive substrate, and detecting the position. Overprint exposure is performed according to the array coordinate system.

According to this, similarly to the third aspect of the invention, the arrangement coordinates of the shot areas on the photosensitive substrate are calculated with higher accuracy than before, and the shot areas are set to the exposure position according to the calculated arrangement coordinates. Since the overprint exposure is performed while sequentially positioning, a more highly accurate overlay can be realized as a result.

[0048]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to FIGS.

FIG. 1 shows a position detecting device 10 according to the present invention.
The configuration of the main part of the exposure apparatus 200 of the embodiment including 0 is shown. The exposure apparatus 200 is a so-called step-and-repeat type reduction projection type exposure apparatus (stepper).

This exposure apparatus 200 has a projection optical system PL whose optical axis AX direction is the vertical axis direction (Z-axis direction), and a horizontal plane which is arranged above this projection optical system PL and is substantially orthogonal to the optical axis AX. A reticle stage RST for holding a reticle R as a mask therein, a wafer stage WST arranged below the projection optical system PL for holding a wafer W as a substrate, a position detector 100, and a main controller 106. I have it.

In this exposure apparatus 200, the projection optical system PL
Is a two-sided telecentric projection optical system with a predetermined reduction magnification, for example, 1/4.

The reticle stage RST is constructed so that it can be finely moved in the XY two-dimensional directions and finely rotated about the optical axis AX by a drive system (not shown).

Wafer stage WST is driven in a two-dimensional direction within a plane (XY plane) perpendicular to optical axis AX of projection lens PL by drive system 16 including a drive motor. This driving method may be either a method of rotating the feed screw by a motor or a method of directly moving the stage main body by a linear motor. The wafer W is suction-held on the wafer stage WST via a wafer holder (not shown). This wafer stage WS
The moving position (coordinate position) of T is sequentially measured by the laser interferometer 17. The measurement value of the laser interferometer 17 is input to the main controller 106 and used for feedback control of the drive system 21. In addition to this, the measurement value of the laser interferometer 17 is also input to a position detection circuit 120 described later.

A fiducial mark (reference mark) plate FP is provided on a part of the wafer stage WST. On this mark plate FP, a reference mark FG (of which the period is the same as the lattice mark MGj on the wafer which will be described later) formed by patterning lines and spaces with a chrome layer is formed on the surface of quartz glass (see FIG. 2). In the actual position detection, the position of the reference mark FG is detected prior to the position detection of the lattice mark on the wafer W to correct various errors depending on the apparatus, which will be described later in detail.

Here, in the present embodiment, the surface of the wafer W is assumed to be a conjugate surface with the reticle R with respect to the projection optical system PL in a state where the wafer W is sucked and held on the wafer stage WST via the wafer holder. Therefore, as will be described later, when the reticle R and the shot area on the wafer W are aligned, when the reticle R is illuminated by exposure light from a light source (not shown), it is formed on the pattern surface of the reticle R. The circuit pattern is projected on the wafer W through the projection optical system at a reduction ratio of 1/4, and an image of the circuit pattern is formed on the shot area on the wafer W. In this way, on the shot area. The circuit pattern of the reticle R is overprinted on the already formed fine pattern.

The main controller 106 comprises a minicomputer (or a microcomputer) including a CPU (central processing unit), ROM, RAM and the like, and controls the entire apparatus as a whole.

The position detecting apparatus 100 includes a position detecting mechanism system 110 and a position detecting circuit 120 which constitute a so-called off-axis alignment optical system in the exposure apparatus 200.
It is composed of Hereinafter, this position detection device 100
Will be described in detail.

FIG. 2 shows a specific configuration example of the position detection mechanism system 110 which constitutes the position detection apparatus 100 in an enlarged scale together with the wafer stage WST and the like. The position detecting mechanism system 110 includes a light transmitting optical system 102 and a light receiving optical system 10.
4 is provided.

Among these, the light transmission optical system 102 includes two semiconductor laser light sources LS1 and LS2, a mirror 1, a dichroic mirror 2, a mirror 4, a rotary radial grating plate RRG, a collimator lens 5, a light beam selecting member 6, and a parallel plate glass. The adjusting optical systems 7, 8, and 9, which are configured by, a beam splitter (half mirror) 10, an objective lens 11, and the like. Here, the operation of the light transmitting optical system 102 will be described together with the operation of each of the above components.

The two laser light sources LS1 and LS2 are laser beams LB1 and L having different wavelengths λ1 and λ2, respectively.
Eject B2. In this embodiment, for example, the laser light source L
S1 emits λ1 = 670 nm, and the laser light source LS1 emits λ1
= 780nm shall be emitted. Then, the laser power source 3 for driving these laser light sources (hereinafter appropriately referred to as "laser") alternately turns on two lasers at a constant frequency, and a laser lighting signal L0 indicating which laser is turned on. To occur. In addition, the repetition frequency of this alternate lighting is 1 / of the beat frequency 2 · Δf described later.
It is about 100 or less.

The two beams LB1 and LB2 are reflected by the mirror 1
Or, the beam LB of the coaxial through the dichroic mirror 2
The light enters the mirror 4 as 0, is reflected here, and then enters the rotating radial grating plate RRG. The grating plate RRG rotates at high speed around the rotation axis C0 at a constant angular velocity in one direction, and acts to increase or decrease the frequency of the diffracted light of each order diffracted by the grating plate RRG according to the angular velocity. Have.

FIG. 3 shows this rotating radial grating plate RRG.
An enlarged perspective view of is shown. The rotating radial grating plate RRG is made of a disk-shaped member, and transmission phase diffraction gratings RG are formed on the grating plate RRG at predetermined pitch intervals along the circumferential direction over 360 degrees. here,
It is assumed that the rotation axis C0 of the lattice plate RRG is set parallel to the X axis of the XYZ coordinate system. When the beam LB0 is perpendicularly incident on the grating RG of the grating plate RRG, various diffracted light is generated in addition to the 0th order light D0, but in this embodiment, the ± 1st order diffracted light is used to realize the heterodyne system. ,
In FIG. 3, only ± first-order diffracted light from the grating plate RRG is shown.

When the beam LB0 is vertically incident on the grating RG of the grating plate RRG, as shown in FIG. 3, the grating RG of the grating plate RRG produces a beam LB1 of wavelength λ1 when the laser LS1 is turned on. The generated first-order diffracted beam ± D11 (dashed line) is generated from the beam LB2 having the wavelength λ2 when the laser LS2 is turned on.
(Two-dot chain line) occurs. The diffraction angle θ of the first-order diffracted beam for each wavelength is expressed as follows.

[0064]

## EQU1 ## sin θn = λn / Prg (1) where the subscript n represents the distinction of each wavelength (n = 1, 2,
3), Prg represents the period of the lattice RG.

On the other hand, the first-order diffracted beam generated from the grating plate RRG receives a constant frequency shift Δf regardless of the wavelength. That is, the grating RG of the grating plate RRG is the beam LB0.
If the velocity across V is V, then Δf = V / Prg,
The frequency of the + 1st order diffracted beam is Δ from the frequency of the 0th order light D0.
The frequency of the -1st-order diffracted beam increases by f and the 0th-order light D
It is lower than the frequency of 0 by Δf. Thus, the rotary radial grating plate RRG functions as a kind of frequency shifter.

The light-transmitting beam + LF and the -1st-order diffracted beam -D1n composed of the + 1st-order diffracted beam + D1n (n = 1, 2) of the two wavelength components frequency-shifted as described above.
Light-transmitting beam -LF composed of (n = 1, 2) and zero-order light D
As shown in FIG. 2, the 0 is converted by the collimator lens 5 so that the principal rays become parallel to each other, and reaches the light flux selecting member 6. The light flux selecting member 6 functions as a spatial filter placed on a so-called Fourier transform surface, and here the 0th-order light D0 is blocked and the light-transmitting beams ± LF of the 1st-order diffracted light ± D1n pass.

After that, the transmitted light beams ± LF reach the beam splitter (half mirror) 10 via the adjusting optical systems 7, 8 and 9 composed of parallel flat plate glass whose tilt amount is variable.

The adjusting optical system 7 has a function of decentering the optical axis of the lens 5 without changing the distance between the light transmitting beam + LF and the light transmitting beam −LF in the Fourier space. 9 is a light-sending beam + LF and a light-sending beam -LF
It has a function of individually adjusting the positions of the optical axis and the optical axis. That is, the adjustment optical system 7 adjusts the inclination of the light-transmitting beam (± LF) with respect to the wafer W in the radiation direction centered on the optical axis of the objective lens 11, and the adjustment optical systems 8 and 9
The opening angles of the light-sending beams + LF and −LF with respect to the optical axis of the objective lens 11 on the wafer W are adjusted.

The transmitted beam ± LF is split into two by the beam splitter 10, and one (reflected light) is the objective lens 1
1, and becomes a parallel light flux, and is incident on the wafer W at different angles for each wavelength.

Here, on the wafer W, a plurality of lattice marks (position detection marks) MGi (i = 1, 2, ...) Having a sectional shape as shown in FIG. 6 are formed in advance. , In FIG. 2, a lattice mark (position detection mark) MGj
The (jth mark among the plurality of marks) is located exactly at the incident position of the light-sending beam ± LF.

Therefore, on the grating mark MGj, the interference fringes formed by the interference of the light transmission beam ± D11 of the wavelength λ1 and the interference fringes formed by the interference of the light transmission beam ± D12 of the wavelength λ2 In synchronization with the lighting of LS1 and LS2, they appear alternately in the same cycle and the same phase. Further, due to the frequency difference 2 · Δf between the light-transmitting beams + LF and −LF, the interference fringes appear to move on the lattice mark MGj in one direction (periodic direction of the lattice mark MG) at a constant velocity. To be observed. The moving speed thereof is proportional to the speed V of the grid RG of the rotating radial grid plate RRG. As is clear from this, in this embodiment, the rotary radial grating plate RRG is used.
The relative scanning means is constituted by.

As is apparent from FIG. 2, the wafer W
The surface (lattice mark MGj) and the radial lattice plate RRG are arranged so as to be conjugate (imaging relationship) with each other by a combined system of the collimator lens 5 and the objective lens 11. Therefore, ± 1 of the grid RG of the radial grid plate RRG
The diffraction image of the second-order diffracted light shows the lattice mark MGj of the wafer W.
Although formed on the upper side, since the 0th-order light D0 is shielded, a bright-dark image (interference fringe intensity distribution) that is ½ of the period of the amplitude distribution of the grating RG is formed.

In this embodiment, the period Pif of the amplitude of the interference fringes on the wafer W (twice the period of the intensity distribution) is set to 2 / N times (N is a natural number) the period Pmg of the lattice mark MGj. To do. At this time, for any of the above two wavelengths, the 0th-order diffracted light (regularly reflected light) by the grating mark MGj of one (first) incident beam (for example, + LF) and the other (second reflected light)
The Nth-order diffracted light of the incident beam (for example, -LF) by the grating mark MGj becomes a first combined light beam which is generated in the same direction and interferes with each other, and the 0th-order diffracted light of the second incident beam by the grating mark MGj. , Nth-order diffracted light of the grating mark MGj of the first incident beam is also generated in the same direction and becomes a second combined light beam that interferes with each other. The first and second combined light fluxes are beat light whose intensity is modulated at a frequency of 2 · Δf according to the frequency difference between the incident beams ± LF that are light sources.

As described above, in order to generate the 0th-order diffracted light and the Nth-order diffracted light in the same direction, from a different point of view, the focal length of the objective lens 11 is F0 and the transmitted light beam for each wavelength is used. The distance DLn on the Fourier transform plane of ± LF (near the beam splitter) is determined with respect to the detection direction of the grating mark MGj.

[Formula 2] DLn = ± NF0λn / Pmg (2) The setting of the interval DLn for each wavelength can be adjusted by appropriately determining the period of the grating RG of the rotating radial grating plate RRG and the focal length of the collimator lens 5.

Further, the light-sending beam ± LF is applied to the grating mark MG (wafer W) so that an error does not occur in the detection position even when the grating mark MGj is slightly defocused (shifted in the optical axis AX direction). Incident at an inclination angle, that is, from the optical axis AX on the Fourier transform plane to the grating mark MGj.
It is desirable to pass the positions separated by DLn / 2 in the detection directions of.

That is, the position detecting device 100 of the present embodiment.
The configuration of is close to the configuration of the conventional “0th-order method”.

By the way, if there is a slight chromatic aberration in the optical systems 5, 11, etc., the interference fringes formed on the wafer W will cause a positional shift (phase shift) and a periodic shift for each color. There is a fear. Therefore, in order to correct such a shift, the adjusting optical systems 7, 8 and 9 in FIG. 2 are used. These optical systems 7, 8 and 9 are made of parallel plate glass, and if a material having a large chromatic dispersion is used as the material thereof, the positional displacement and phase of the interference fringes formed on the wafer W for each wavelength component are mutually different. The shift can be minutely changed. Alternatively, as the adjusting optical systems 7, 8 and 9, a parallel flat plate glass having a small chromatic dispersion and a parallel flat plate glass having a large chromatic dispersion are combined, and the inclination fringes of the parallel flat plate glass having a large chromatic dispersion are adjusted so that the interference fringes of each wavelength component are mutually The overall tilt error of the light-transmitting beams ± LF on the wafer W caused by the correction can be corrected by adjusting the tilt of the parallel plate glass having a small chromatic dispersion.

Next, the light receiving optical system 104 will be described. The light receiving optical system 104 includes the objective lens 11,
Condensing lens (Fourier transform lens) 12, reference grating S
G, a spatial filter 13, photoelectric converters (detectors) 15a and 15b as light receiving means, and a photoelectric converter 14 for generating a reference signal. Here, the light receiving optical system 104 will be described together with the operation of each of the above components.

The first and second combined light fluxes A and B generated from the grating mark MGj illuminated by the interference fringes as described above pass through the objective lens 11 and the beam splitter 10 and are respectively photoelectric converters (detectors). 15a, 15
It is incident on b and the light amount signal thereof is converted into electric signals IA and IB. The photoelectric signals IA and IB are the laser light source LS.
By the alternating lighting of 1 and LS2, a signal due to the luminous flux of wavelength λ1 and a signal due to the luminous flux of wavelength λ2 appear alternately.
The interference fringes of the signal at each wavelength are the grating marks M.
While irradiating the Gj portion, the beat frequency is 2 · Δf
Becomes a sine wave that changes at the same frequency as.

On the other hand, the light beam transmitted from the collimator lens 5 through the beam splitter 10 (transmitted beam ± LF)
Enters the condenser lens (Fourier transform lens) 12. Then, irradiation is performed by superimposing it on the transmissive reference grating SG. Here again, the reference grating SG is a rotary radial grating plate R with respect to the combined system of the collimator lens 5 and the condenser lens 12.
It is arranged conjugate with RG. Therefore, one-dimensional interference fringes due to two-beam interference of the primary beams ± LF are also formed on the reference grating SG, which moves at a speed corresponding to the beat frequency 2 · Δf.

Therefore, if the period of the reference grating SG and the period of its interference fringes are appropriately determined, the ± 1st-order diffracted light generated from the reference grating SG travels in the same direction as the interference beam Bms, which is transmitted through the spatial filter 13. Transmitted through the detector 14
Received. The photoelectric signal Ims of this detector 14 is also
The sine wave has the same frequency as the beat frequency 2 · Δf, and its signal Ims becomes the reference signal of the heterodyne system. Of course, the reference signal Ims is also changed into a signal by the light flux of the wavelength λ1 and a signal by the light flux of the wavelength λ2 alternately by alternately turning on the laser light sources LS1 and LS2.

In the position detecting mechanism system 110 of this embodiment described above, a semiconductor laser is used as a light source,
In this case, a shaping optical system (a plurality of slanted parallel flat glass plates, etc.) for removing astigmatism is provided between each of the semiconductor lasers (LS1, LS2) and each of the mirrors 1, 2 and combined into one. It is preferable that the luminous flux components of the respective wavelength components of the beam LB0 have substantially the same diameter.

FIG. 4 shows an example of the position detecting circuit 120 which constitutes the position detecting device 100. The position detection circuit 120 includes a phase difference detection circuit SA, a detection position correction circuit CA, and a weighting EGA processing circuit WEGA. Here, the role of each circuit constituting the position detection circuit 120 will be briefly described.

In the case of the heterodyne system as in this embodiment, as described above, when the interference fringes irradiate the grating mark MGj portion or the reference mark FG, all the photoelectric signals IA, IB and Ims are beat. Frequency 2 · Δf
It becomes a sine wave with the same frequency as. An example of these signals is shown in FIG.

Of these, FIG. 5A shows the detector 15a.
The light intensity signal IA in the first combined light flux, which is the output of the above, is shown, and FIG. 2B shows the light intensity signal IB which is the output of the detector 15b. Further, FIG. 3C shows the reference signal Ims which is the output of the detector 14. A phase difference of ΔA and ΔB exists between these signals with reference to the reference signal Ims. The amount of these phase differences corresponds to the position and the amount of step difference of the lattice mark MGj. When the laser light source has multiple wavelengths, the phase relationship (Δ
A, ΔB) and its amplitude (maximum value AMAX, BMAX, minimum value AMI
The value of (N, BMIN) is different for each light source wavelength.

In the position detecting circuit 120 of FIG. 4, the phase difference detecting circuit SA first causes the phase differences ΔA and Δ to be detected.
B is detected. To convert this phase difference into the position of the grating mark MGj, N.times.Pmg / (2π) times (N is the period Pmg of the grating mark MGj and the period of the amplitude of the interference fringes formed on the grating mark MGj as described above. Double the ratio)
As described in the section of the description of the prior art, in the “0Nth order method”, the average value of the detection positions obtained from these two phase differences is the detection position of the grating mark MGj for each wavelength. The detection circuit SA has an average value ΔXn,

## EQU3 ## ΔXn = {ΔAnNPmg / (2π) + ΔBnNPmg / (2π)} / 2 (3) is output. Here, the subscript n represents the discrimination of the wavelength of the detection light. That is, when the laser light source has multiple wavelengths, this is detected for each wavelength (λn). Specifically, the laser lighting signal L0 generated from the laser power source 3 in FIG.
The detection position obtained from the signals IA, IB, and Ims captured during the lighting of S1 is ΔX1, and the laser LS2
Signals IA, IB, and Ims captured during the lighting of
The detected position thus obtained is ΔX2.

Further, the phase difference detection circuit SA uses the respective input signals IA, IB and the widths a, b of the concave portion and the convex portion of the lattice mark MGj inputted from the console (not shown) or the like to form the step of the lattice mark MGj. The equivalent amount δn is calculated as the phase amount of light which is the remainder obtained by dividing the true step difference by half λrn / 2 of the wavelength λrn in the vicinity of the surface of the grating mark MGj of the detected light flux of wavelength λn. Wavelength λrn near the surface of the grating mark MGj
For example, when the grating mark MGj is covered with a transparent material such as resist or glass (PSG), it means the wavelength in the transparent material (medium), and the surface of the grating mark is exposed. , It means the wavelength in air, that is, λn itself. Also, the step equivalent amount δn
The calculation algorithm of is a feature of the present invention (embodiment),
The details will be described later.

By the way, the above-mentioned detected position ΔXn is the amount of positional shift (phase shift) with respect to the reference signal Ims, and is not the position of the lattice mark MGj itself on the wafer W. Therefore, a position reference is required on the wafer W or the wafer stage WST in order to correspond the above-mentioned positional deviation amount obtained from the reference signal Ims to the position on the wafer W.

The above-mentioned reference mark FG is a mark which serves as this position reference, and prior to the position detection of the lattice mark MGj on the wafer W, the position of the reference mark FG is first detected and the above-mentioned position obtained from the reference signal Ims. The relationship between the shift amount and the position on wafer stage WST is obtained (baseline check). Specifically, first, the reference mark FG
Wafer stage WST is moved so that is positioned at the position detection beam ± LF irradiation position. This wafer stage WST
Movement of the laser interferometer 17 by the main controller 106.
Is monitored via the drive system 16.

Then, the phase difference detection circuit SA performs the above-described position detection on the reference mark FG to obtain the position ΔFXn of the reference mark FG with respect to the reference signal Ims. The detection position correction circuit CA detects the detection position ΔFXn for each wavelength and the output LFG of the laser interferometer 17 at this time.
Is stored in a memory (not shown). In this case, the position of the interference fringe at each wavelength is the wafer W (and the reference mark FG).
If the optics are tuned to match exactly above,
.DELTA.FXn becomes equal at each wavelength .lambda.n.

Next, main controller 106 moves wafer stage WST based on each design coordinate value (Xj, Yj) of lattice mark MGj, and the positions of M lattice marks MGj on wafer W are sequentially detected. However, the detection position correction circuit CA uses the laser interferometer 1 at the stage position.
The output LMGj of No. 7 and the output LFG at the reference mark FG detection position stored in the memory are added to the detection result ΔXn for each wavelength.
Then, a value Xjn 'obtained by subtracting ΔFXn is output as the detection position at each wavelength λn of the jth grating mark MGj. Also, regarding the above step equivalent amount Δn,
The j-th grating mark MGj is output as the detected level difference δjn at each wavelength λn.

The weighted EGA processing circuit WEG is based on the detected position ΔXjn ′ and the step equivalent δjn obtained as described above.
A calculates the “arrangement coordinate system” of the M lattice marks MGj on the wafer W. This array coordinate system is the conventional EGA
6 parameters Sx, Sy, Rx, Ry, O
The coordinate conversion system represented by x and Oy will be described in detail later.

In the above description, for simplification of the description, the position of each lattice mark MGj whose detection direction is the X direction is detected only in the X direction. Dimensional position detection is required, and in reality, the exposure apparatus 200 is also provided with a position detection device (not shown) similar to the one described above for the Y direction. The position of the grid mark is also detected in the Y direction.

Next, the principle and method of detecting the step amount of the lattice mark MGj, which is a feature of this embodiment, will be described.

FIG. 6 shows an enlarged example of the lattice mark MGj, where it is assumed that the width of the concave portion is a, the width of the convex portion is b, and the step between both portions is h. In FIG. 6 and the following description, for simplicity, it is assumed that the resist is not coated on the lattice mark MGj, that is, λrn = λn. Further, the amplitude reflectances for the detection light of the respective wavelengths λ (where λ is one of the above λn (n = 1, 2)) of the concave portion and the convex portion are φa and φb. At this time, the amplitude reflectances φa and φb represent the amplitudes of reflected light within the plane at the same position in the depth direction (vertical direction in FIG. 6). More specifically, assuming that this reference surface is the surface of the convex portion of the lattice mark MGj, φa is, for example, a factor corresponding to the amplitude reflectance on the concave surface and the round-trip optical path difference (phase difference) of the step h, That is, it is multiplied by exp (4πih / λ). The same applies to φb, but since the convex surface of the mark MGj is used as the reference surface here, the factor of the optical path difference (= 0) is exp (4πi0 / λ) = 1.
Therefore, if the phase difference between φa and φb is obtained, the step amount h can be obtained.

Generally, the amplitude distribution in the diffraction direction of the diffracted light generated from each of the concave portion and the convex portion in FIG.
It is represented as an nc function. For example, the amplitude distribution of the diffracted light generated from the concave portion having the width a is as follows with respect to the periodic direction of the step pattern of FIG.

## EQU4 ## ψA (u) = φa · sin (πau) / (πu) ... (4) Where u is the diffraction angle Θ

## EQU00005 ## u = sin .THETA ./. Lamda.n (5), and similarly, the amplitude distribution of the diffracted light generated from the convex portion with the width b is

## EQU00006 ## .phi.B (u) =. Phi.b.sin (.pi.bu) / (. Pi.u) (6). At u = 0 (zero-order diffracted light), these are a · φa and b · φb, respectively.

The amplitude distribution of the diffracted light from the pattern in which the concave portions and the convex portions are periodically arranged at the pitch P as shown in FIG.

[Equation 7] ψ (u) = {ψA (u) + ψB (u) · exp (πiPu)} × Pir (u) …… (7) Pir (u) = sin (lπPu) / sin (πPu) …… (8) (1 is the number of repetitions of the periodic lattice mark MGj (integer)). In the derivation of the equation (7), the center of the concave portion is used as the reference of the phase of the diffracted light, but of course, the center of the convex portion may be used as the reference.

The above Pir (u) is called the "periodic term" of the diffraction grating, and if the number of repetitions 1 of the grating mark MGj is large, u = K / corresponding to the K-order diffracted light.
It has a non-zero value l only in P (K is an integer), and is almost zero in other cases. In this embodiment, the lattice mark M
Since only the 0th-order diffracted light and the Nth-order diffracted light from Gj are used, Pir (u) may be set to a constant value (l). Also,
Exp (πiPu) in the equation (7) is 0-order diffracted light (u =
0) becomes 1, and ± N-order diffracted light (u = ± N / P)
Then, it becomes (-1) ^N. From this, the 0th and Nth order diffracted lights, ψ 0 and ψ N generated from the grating mark MGj as shown in FIG.

[Equation 8] ψ 0 = a · φa + b · φb ………… (9) ψN = a ′ · φa + (− 1) ^N · b ′ · φb ……… (10) a ′ = P · sin (Nπa / P) / π ... (11) b ′ = P · sin (Nπb / P) / π (12)

FIG. 7 shows the process in which the diffracted light amplitudes ψ 0 and ψ N are derived from the amplitude reflectances φa and φb as described above in polar coordinates of complex numbers. In addition, in FIG.
Although φa is a real number (on the Real axis) for simplification,
If the phase difference between φa and φb (4πh / λ = ω) is invariant, the result derived even if φa is a complex number does not change. Further, in FIG. 7, only the case of N = 1 is shown for simplification, but the following discussion is similarly applicable to N of 2 or more.

FIG. 7 (1) shows that the 0th-order diffracted light amplitude ψ 0 is determined from the amplitude reflectances φa and φb.
(2) indicates that the first-order diffracted light amplitude ψ1 is determined from the amplitude reflectances φa and φb. As mentioned above, amplitude reflectance φ
The phase difference between a and the amplitude reflectance φb is ω. FIG. 7 (3) shows ψ obtained from (1) and (2) of FIG.
0, ψ1 is written on the same polar coordinate, and the phase difference between ψ0 and ψ1 (the phase difference between the 0th-order diffracted light and the 1st-order diffracted light) Δ in the figure is the light quantity signal of the combined light fluxes A and B in FIG. IA, IB
Equal to half the "phase difference" (i.e., [Delta] B- [Delta] A). That is, it is an amount that can be measured from the light amount signals IAn and IBn. Of course, this is not limited to the case of N = 1, but is correct for any N (ψN).

In addition, the ratio of the magnitudes of ψN and ψ0 (| ψN
│: │ψ 0 │), the light quantity signal I
It is possible to measure from An and IBn. Both signals I
The maximum values AMAXn and BMAXn of An and IBn are the intensities in the state where ψ0 and ψN are amplitude-added in the same phase, and the minimum value A
MINn and BMINn are intensities in the case where ψ 0 and ψ N are amplitude-added in opposite phases,

## EQU9 ## AMAXn≈BMAXn≈ (| ψ0 | + | ψN |) ² (13) AMINn≈BMINn≈ (| ψ0 |-| ψN |) ² (14) The contrast γ of both signals is

[Equation 10] γ = (AMAXn-AMINn) / (AMAXn + AMINn) = 2 · | ψ0 ｜ ・ ｜ ψN ｜ / (│ψ0 │ ² + ｜ ψN │ ² ) ... (15) Therefore, the contrast γ is measured. Then, from equation (15),

(16) β = {1 ± √ (1-γ ² )} / γ (17) and the ratio β is obtained. Although ± in the equation (17) cannot be uniquely determined, since the intensity of the Nth-order diffracted light is generally weaker than that of the 0th-order diffracted light, − (minus) is used. And this gives ψ N

[Expression 12] ψ N = β · ψ 0 · exp (iΔ) It becomes possible to express as (18).

When the above Δ and β are obtained, the above process of deriving the diffracted light amplitude from the amplitude reflectance is followed in reverse, and the amplitude reflectances φa and φb of the grating mark MGj are calculated from the diffracted light amplitudes ψ 0 and ψ N. You can ask. Specifically, the simultaneous equations of the equations (9) and (10) may be solved based on the known parameters.

Generally, in the processing of a semiconductor integrated circuit, the controllability of the pattern line width is excellent. Therefore, the widths a and b of the concave and convex portions of the lattice mark MGj are almost as designed, that is, they are known. Then, a ′ and b ′ calculated from the widths a and b of the concave portion and the convex portion are also known. Therefore, in the simultaneous equations of equations (9) and (10), only unknown variables (variables that have not been measured) are φa and φb, and therefore the simultaneous equations can be solved for φa and φb. as a result,

[Formula 13] φa = {b '・ ψ0-(-1) ^N・ b ・ ψN} / C ……… (19) φb = (a ′ ・ ψ0 −a ・ ψN) / C ……… (20) (C = a * b '-(-1) ^N * a' * b) is obtained. (ΨN
Equation (18) is substituted into. ) In equations (19) and (20), the phase of ψ 0 is not known, but
Since the final result is that the phase difference (= ω) between φa and φb is known, the phase of φ0 (angle formed with the Real axis) may be any value.

From the above, from equations (19) and (20), φa
, Φb value (complex number) is obtained. Then, the respective phases (ωa, ωb) of both are obtained from the real and imaginary parts of φa and φb. That is,

Ω a = tan ⁻¹ {Im (φa) / Re (φa)} ……… (21) ωb = tan ⁻¹ {Im (φb) / Re (φb)} ……… (22) . Then, if the difference between them is ωb−ωa = ω, then φa
, Φb phase difference ω is obtained. The phase difference ω can take a value up to 4π [rad] as an absolute value, but due to the periodicity of the trigonometric function (tan ^-1 ), when the value of ω exceeds 2π, ω-2π is ω. , Ω is negative, then ω
Let ω be 2π added until becomes positive.

This phase difference ω corresponds to the round-trip optical path difference (phase difference) of the step h as described above,

## EQU15 ## ω = 4πh / λ, that is, h = ωλ / (4π) ... (23) The step difference h can be finally obtained. However, the step h here is in the range of 0 to λ / 2 according to the above method of determining ω. That is, for a true step,
The remainder of the detected light flux at half wavelength (hereinafter referred to as "h '"
).

Although the above description is for the mark not covered with the resist as shown in FIG. 6, the above logic is almost applied to the mark covered with the transparent film such as resist or glass (PSG). it can. However,
In the above logic, the amplitude reflectances φa and φb are set considering the multiple interference between the transparent film and the mark surface, and the wavelength λ in Eq. (23) is the wavelength in the transparent film (the wavelength in the air is transparent. Value divided by the refractive index of the film).

In this embodiment, according to the above principle,
The step amount of the lattice mark MGj is measured. That is, FIG.
First, the phase difference detection circuit SA in the inside detects the phase difference Δ of the respective light amount signals IA and IB with the reference signal Ims obtained at the time of position detection.
The half amount Δ of the difference between A and ΔB is calculated. Next, in the phase difference detection circuit SA, the maximum values AMAXn, B of both signals are caused by an internal peak hold circuit and bottom hold circuit (not shown).
MAXn and minimum values AMINn and BMINn are detected, and the contrast γ of each signal is calculated from this. The contrast of both signals is the same unless there is asymmetry due to the step pattern, but if they are different, the average value is adopted.

The phase difference detection circuit SA further obtains the ratio β of the magnitudes of ψ N and ψ 0 from the contrast γ and the equation (17),
From this β, the phase difference Δ, and equation (18), the exact (complex number) relationship between ψ N and ψ 0 is obtained.

The phase difference detection circuit SA has the widths a and b of the concave and convex portions of the lattice mark MGj and the period P, which are given by a means such as an operator's input from a console (not shown). Calculate the values of a'and b'using the values and equations (11) and (12), and substitute these into equations (19) and (20) to calculate the values of φa and φb (complex numbers). . Then, the phases ωa and ωb of φa and φb are calculated from the equations (21) and (22), and the difference ω is calculated as described above. In general, a plurality of lattice marks M on the wafer
The widths a and b of the concave and convex portions of Gj have the same value for all marks.

From this phase difference ω, it is of course possible to obtain the step h ′ from the equation (23) as described above. However, in the weighting EGA circuit WEGA, the weighting to the detection position ΔXjn ′ of each grating mark MGj (or each wavelength) by the detection position correction circuit CA is not the step amount itself, but the phase of each step amount to each detection wavelength. Since it is performed based on the quantity, the phase difference detection circuit SA
It is preferable that the output value from is the phase amount thereof rather than the step h ′. The phase amount δn is half (δn = ω / 2) of the round-trip phase amount ω of the step h. This phase amount Δn is also the surplus of the true phase amount with the phase difference (π) corresponding to the half wavelength of the detection light.

Of course, in the weighting EGA circuit WEGA, the detection position ΔXjn 'may be weighted on the basis of the step amount itself. In that case, the output value of the phase difference detection circuit SA is referred to as the step h'. Good. However, the step h ′ obtained from the equation (23) is affected by the refractive index of the resist or the like (refractive index near the grating mark) when the grating mark MG is covered with the resist or the like,
Since the relationship of δ = ω / 2 is not influenced by the refractive index of the resist, it is better to make the above selection based on the phase amount δn.

However, if the wavelength in vacuum is used as the wavelength λ when calculating the step h ′ by the equation (23), h ′ is not an accurate value as the step amount, but the phase amount δn is the same as the resist amount h ′. It is not affected by the refractive index.

In the above description, as an example of the lattice mark MGj, the case where the step equivalent amount is detected for a mark in which the boundary (sidewall) between the concave portion and the convex portion is vertical as shown in FIG. 6 has been described. Even with a certain mark, it is of course possible to detect a step with high accuracy. In this case, the respective widths a and b of the concave portion and the convex portion to be input are not a + b = P, but the mark step may be calculated based on the values of a and b as described above.

Next, the weighted EGA processing circuit WEGA of this embodiment will be described.

This weighted EGA processing circuit WEGA is
The "arrangement coordinate system" of the M lattice marks MGj on the wafer W is calculated based on the detected position ΔXjn ′ of the lattice mark MGj and the step-equivalent amount δjn described above. The conventional EG
The same applies to A, but in general, the array precision of each shot area of the fine pattern formed by exposure using a stepper or the like on the wafer is extremely good, and the array coordinate system of M lattice marks, that is, this M Individual lattice marks MGj
The array coordinate system of M shots including (j = 1, 2, ..., M) may be considered to be almost equivalent to the array coordinate system of all shots.

Weighted EGA processing circuit WEG of this embodiment
A is inversely proportional to the detection error estimated to be included in each detection position Xjn 'in the statistical processing of each detection position ΔXjn' of each lattice mark MGj and each design coordinate value (Xj, Yj). Weighting is performed. The estimation of the detection error is
As described above, “(2m + 1) / wavelength of detected light flux in resist
The detection error becomes extremely small when the mark level is 4 times. The grid mark MGj
The detection step equivalent amount δjn for each wavelength λn of π / 2 [ra
d] (corresponding to 1/4 of the wavelength λn of the detected light beam (m = 0: the step equivalent δjn is a half-wavelength remainder, m> 0 does not have to be taken into consideration)). The weight amount at the mark MGj is determined.

As a result, in the final array coordinate system, the detection positions of marks (and wavelengths) with smaller detection errors are reflected more, and the detection positions of marks (and wavelengths) with large detection errors are reflected. Since the (detection error) has little influence, it is more accurate than the array coordinate system obtained by the conventional method without weighting.

The weight Wjn (wavelength λn of the jth mark)
As the weight for the position detection result Xjn 'in

[Equation 16] Wjn = cos (δjn−π / 2) ………… (24),

## EQU17 ## Wjn = [1 + cos {2 (δjn-π / 2)}] / 2 (25) is used. Of course, the weight Wjn is
Any value may be used as long as jn is large in the vicinity of π / 2 and becomes smaller as it goes away from π / 2.

Generally, in the EGA, the design position (Xj, Yj of the jth grating mark MGj for measuring in the X direction as described above is used.
), And its detected position in the X direction between Xjn 'and

[Equation 18] Xjn ′ = Sx · Xj + Rx · Yj + Ox + ΔXjn (26) By applying a linear relationship, the sum of squares of residual ΔXjn (sum for j and n) at this time is minimized. (Least square method), the values of the parameters Sx, Rx, and Ox are determined. Similarly, the kth Y-direction lattice mark M for measuring
The design position of Gk is (Xk, Yk), and the detected position in the Y direction is Ykn ', and the relationship between them is

## EQU19 ## From Ykn '= Ry.Xk + Sy.Yk + Oy + .DELTA.Ykn (27), the respective values of the parameters Sy, Ry, Oy are determined so that the sum of squares of the residual .DELTA.Ykn is minimized. Where Sx, Sy
Is the amount of expansion and contraction of the mark position (that is, the wafer) in the X and Y directions, and Rx and Ry are the amount of rotation of the mark position (that is, the wafer). This rotation amount Rx, Ry
Strictly speaking, is a function of the residual rotation amount at the time of global alignment (rotational alignment) of the reticle R and the wafer W performed prior to EGA, the orthogonality error of the wafer arrangement coordinate system, and the above Sx and Sy. However, here, for the sake of simplification of description, the rotation amounts are Rx and Ry (there is no difference in results even if considered in this way).

The patterning (exposure transfer) in the previous (previous layer) exposure step is, of course, performed according to the design value for both the circuit pattern and the mark. However, due to the thermal process after exposure and the like, the wafer expands and contracts, and these linear errors occur due to the accuracy of the global alignment described above.

The sum of squares of the residual ΔXjn is

ΣΔXjn ² = Σ (Xjn′−Sx · Xj−Rx · Yj−Ox) ² (28) and the parameters Sx, Rx, Ox that minimize this
Is

[Equation 21] ∂ΣΔXjn ² / ∂Sx = 0 ………… (29) ∂ΣΔXjn ² / ∂Rx = 0 ………… (30) ∂ΣΔXjn ² / ∂Ox = 0 ………… (31) ( It can be obtained from the partial differential value of 0, that is, the minimum value. Similarly, the Y-side parameters Sy, Ry, and Oy represented by the equation (27) can be obtained.

In the conventional EGA, the detection error of each mark was not taken into consideration when determining these parameters, but in the present embodiment, the weight Wjn determined by the step equivalent amount δjn of the mark described above Each parameter is determined by considering (weighting) the mark detection error. Specifically, instead of formula (28),

[Equation 22] ΣΔXjn ² = Σ {Wjn · (Xjn′−Sx · Xj−Rx · Yj−Ox) ² } (32) is used to calculate the weighted sum of squares of the residual ΔXjn. As a result, the detected value Xjn 'having a large weight Wjn has a large "weight" when determining the parameters Sx, Rx, Ox by partial differentiation. Parameters Sx, Rx, Ox
Is calculated from the equations (29) to (31) based on this new value, and the method is the same as the conventional EGA.

The above equations (29) to (31) are the following determinants:

[0124]

(Equation 23)

Since it is equivalent to, the parameters Sx, Rx, Ox can be obtained by solving the above determinant (34), as in the conventional EGA. Of course, the point that the weight Wjn is included in the determinant is different from the conventional EGA.

Similarly, Y side parameters Sy, Ry, Oy
Is also the weighted sum of squares as above

## EQU23 ## ΣΔYkn ² = Σ {Wkn · (Xkn′−Ry · Xk + Sy · Yk + Oy) ² } (35) is determined by partial differentiation.

From the above, the parameters Sx, Sy, Rx, Ry,
Ox and Oy are determined, and the array of each exposure shot on the wafer W is estimated by these parameters. That is, when the design position of each shot area is (Xe, Ye), the actual position (Xe ', Ye') of each shot area is calculated according to the equations (26) and (27).

[Equation 25] Xe '= Sx.Xe + Rx.Ye + Ox (36) Ye' = Ry.Xe + Sy.Ye + Oy (37) is calculated (detected).

Next, the flow of processing during overlay exposure by the exposure apparatus 200 of this embodiment will be briefly described.

As a premise, on the wafer W, a predetermined shot area (fine pattern area) at the time of front layer exposure and a grid mark for X-direction position measurement and a grid mark for Y-direction position measurement which accompany each shot area are formed. It is assumed that the wafer has been already formed and, after processing such as development, the wafer has been mounted on the wafer stage WST and the global alignment described above has been completed.

First, the above-described baseline measurement is performed. Here, the baseline amount means a projection point of the center CCr of the reticle R on the wafer side (which substantially coincides with the optical axis AX) and the position detection mechanism system 1 shown in FIG.
It is nothing but the positional relationship in the X and Y directions between the detection center point Rf4 of 10 and the projection point on the wafer side. As for the positional relationship, the positional deviation amount between the corresponding mark FG of the reference mark plate FP and the projection point of the detection center point Rf4 is determined by the position detecting device 100.
And the coordinate position of wafer stage WST at that time is detected by laser interferometer 17.

Next, in a certain number of shot areas on the wafer W, for example, 8 to 15 shot areas in the X direction composed of lattice marks respectively attached to the shot areas,
The position of the alignment mark for detecting the position in each of the Y directions and the step-corresponding amount are obtained for each wavelength by using the position detecting device 100 for detecting the positions of the X and Y directions, respectively, as described above.

Next, in the position detection circuit 120 of the position detection device 100, the above-described weighted EGA processing is performed based on these detected position data, detected step equivalent amount, mark design value data, shot array design value data, and the like. Then, the array coordinate system of all the shot areas on the wafer is obtained, stored in the memory, and placed.

Then, at the time of actual exposure, stepping and exposure are repeated by the step-and-repeat method. During this stepping, main controller 106 performs two-dimensional movement of wafer stage WST via drive system 16 while monitoring the output of laser interferometer 17 in accordance with the shot array coordinates stored in the memory. Shot regions on W are sequentially positioned in the image field of the projection optical system PL.

In this case, according to the calculated array coordinates,
Each shot is aligned with reticle R simply by moving wafer stage WST two-dimensionally. And
When the reticle R is illuminated by exposure light from an exposure light source (not shown) at each positioning position, the circuit pattern of the reticle is overlaid and exposed on each shot area.

As is apparent from the above description, in the present embodiment, the phase difference detection circuit SA constitutes the step difference detection means, and the phase difference detection circuit SA and the detected position correction circuit CA constitute the position detection means. Weighting EGA circuit W
The EGA constitutes a calculation means.

As described above, according to this embodiment,
When calculating the shot array by EGA, the degree of error in the position detection of each alignment mark (lattice mark) is estimated based on the step equivalent amount of each mark, and the least squares method is performed by weighting according to the degree of this error. Since the coordinate system of the shot array is calculated by the statistical processing using, the two-dimensional movement of wafer stage WST is performed in accordance with the calculated array coordinate system, resulting in higher accuracy than in the past. Superposition is possible.

Further, according to the present embodiment, since the position of the grating mark and the amount of step difference are detected for each of a plurality of wavelengths, the position detection result and the step detection result at each wavelength are used to A weighting EGA process is performed. At this time, weighting is also performed for superiority or inferiority of each wavelength (relationship between step difference and wavelength) with respect to each grating mark, and superposition can be performed with higher accuracy.

However, the present invention is not necessarily limited to the case of using the multi-wavelength laser light source as in the above embodiment, and a light source of a single wavelength may be used. When using the detection result with a single wavelength, the weighting is based on the detection result of each lattice mark itself, that is, the processing accuracy variation of the step amount of each lattice mark in the wafer surface,
This is done for uneven coating of the resist.

In the above-described embodiment, the case where the grating mark is alternately irradiated with the light beams having a plurality of wavelengths to perform the position detection has been described. However, the present invention is not limited to this. The MG is irradiated with light fluxes of a plurality of wavelengths at the same time, the dichroic mirror is used in the light receiving section to separate the first and second composite light fluxes into respective wavelengths, and the detectors separately provided for the respective wavelengths are respectively received. You may In such a case, although the light receiving optical system and the signal processing system become somewhat complicated, there is an advantage that the measurement time can be shortened because detection can be performed for each wavelength at the same time.

Further, in the above embodiment, in the step detection, the ratio of the magnitudes of ψN and ψ0 is calculated by the light amount signal IAn.
, IBn, it was calculated from the contrast,
The present invention is not limited to this, and for example, the light quantity ratio itself of the Nth-order diffracted light and the 0th-order diffracted light may be measured and the square root thereof may be used.

As a method of detecting the light quantity ratio between the Nth-order diffracted light and the 0th-order diffracted light, for example, a shutter capable of blocking at least one of the light-transmitting beams ± LF is provided near the light flux selecting member 6 in FIG. After the position detection described above is completed or before the position detection is started, either one of ± LF is shielded by this shutter, and the intensity ratios of the light amount signals IA and IB obtained from the detectors 15a and 15b at this time may be obtained. .
At this time, since the number of beams incident on the grating mark MG is one for each wavelength, the respective light amount signals IA,
The IB has no beat and is a DC signal. When, for example, the transmitted light beam −LF is shielded by the shutter, only the 0th order diffracted light by the grating mark MG of the transmitted light beam + LF is incident on the detector 15a, and the light amount signal IA is the 0th order diffracted light at each wavelength. Only the first-order diffracted light from the transmitted beam + LF grating mark MG is incident on the detector 15b, and the light-quantity signal IB indicates the light amount of the N-th order diffracted light at each wavelength.

Incidentally, if either one of the light-transmitting beams is shielded in this way, both position detection and step amount detection become impossible. Further, since opening and closing of the shutter generally requires a considerable time, the shutter is opened and the laser light sources LS1 and LS2 are alternately turned on instead of opening and closing the shutter while each laser light source LS1 and LS2 is turned on. Signal light I
After receiving A and IB respectively, the shutter is closed and the laser is alternately turned on again to measure the light amounts of the 0th order diffracted light and the Nth order diffracted light for each wavelength.

As described above, in the method of directly measuring the light amounts of the 0th-order diffracted light and the Nth-order diffracted light, ψ N and ψ 0 The size ratio can be measured more accurately.

In the above embodiment, in order to simplify the explanation, the rotating radial grating plate RR is used as the frequency shifter.
Although the case where G is used is illustrated, the present invention is not limited to this, and two acousto-optic modulators (AOMs) are used as frequency shifters, and a first Zeeman laser light source that oscillates at a central wavelength λ1 is used. A second Zeeman laser light source that oscillates at the central wavelength λ2 may be used as the light source.
Further, various dichroic mirrors may be replaced with a dispersing element such as a prism.

Further, in the above embodiment, the case where the detection light of two wavelengths is used has been illustrated, but the detection light of two or more arbitrary wavelengths may be used.

In the above embodiment, when the heterodyne system in which the frequencies for the incident beams for position detection ± LF are made different is adopted, in other words, the relative scanning in which the grating mark and the interference fringes are relatively scanned in the periodic direction. Means for slightly differentiating the frequency of the first light beam and the frequency of the second light beam of a plurality of pairs of coherent light beams having different wavelengths from each other, and the interference fringes formed thereby are grating The case where the mark is constituted by means for moving the mark in the periodic direction at a constant speed (rotary radial grating plate) is illustrated, but the present invention is not limited to this. For example, the frequencies of the incident beams ± LF are made equal (homodyne system), and instead, a system is adopted in which the wafer stage WST is scanned in the detection direction at the time of position detection. Relative scanning means for performing relative scanning may be configured. Also in this case, each signal becomes a sine wave having an equal beat frequency proportional to the scanning speed of wafer stage WST, and the same detection as above can be performed. The homodyne method has an advantage that a frequency shifter (rotating radial grating plate RRG or the like) is not required and the light transmitting optical system is simplified.

On the other hand, in the heterodyne method, the wafer stage WST may be stopped and placed during detection. In addition to the high S / N ratio achieved by heterodyne, the wafer stage WST
Need no control mechanism for scanning at constant speed,
Further, there is also an advantage that fluctuations (mainly air fluctuations) of outputs (LFG, LMG) of the laser interference system 22 that monitors the wafer stage WST position can be averaged during the detection.

The functions of the position detection circuit 120 and the main control device 106 described in the above embodiments can be actually realized by software. Therefore, it goes without saying that they can be configured by a single computer. It is possible.

Further, in the above embodiment, the case where the position detecting device according to the present invention is applied to the off-axis alignment detecting system has been described, but other through-the-lens (TTL) alignment detecting system and through-the-reticle (T).
The position detection method and apparatus of the present invention can be similarly applied to the TR) alignment detection system.

[0150]

As described above, according to the present invention,
In "interferential alignment", it is possible to detect not only the position of the lattice mark but also the step equivalent amount thereof. Based on this step equivalent amount of the lattice mark, the detection error accompanying the detection position of the lattice mark is detected. Can be estimated. Then, a plurality of lattice marks are detected, each detection position is weighted so as to be inversely proportional to each detection error, and statistical processing is performed using design data of the mark detection position and the arrangement coordinates of the marks and the fine pattern ( Weighted EG
By performing A), there is an effect that the positions (arrangement coordinate system) of all the fine patterns on the substrate can be detected with higher accuracy than in the conventional case.

In particular, according to the third and tenth aspects of the present invention, since the overlay exposure is performed based on the detection result (calculation result) of the array coordinate system of the fine pattern, the overlay can be performed with higher precision than the conventional one. Alignment becomes possible.

Further, as in the invention described in claims 2, 8 and 9, by making the detection light flux of the interference type alignment into a plurality of wavelengths, it becomes possible to detect the position of the fine pattern with higher accuracy.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of an exposure apparatus according to an embodiment.

FIG. 2 is a diagram showing a schematic configuration of a position detection mechanism system that constitutes the position detection device of FIG.

FIG. 3 is a perspective view showing how a diffracted beam is generated by a rotating radial grating plate.

FIG. 4 is a block diagram showing an example of a position detection circuit that constitutes the position detection device of FIG.

5 (1) is a diagram showing a light amount signal IA of a light beam A which is an output of the detector 15a in FIG. 1, (2) is a diagram showing a light amount signal IB which is an output of the detector 15b, and (3) is 5 is a diagram showing a reference signal Ims which is an output of the detector 14. FIG.

FIG. 6 is an enlarged view showing an example of a lattice mark MG.

FIG. 7: Diffracted light amplitudes ψ 0 and ψ N are amplitude reflectances φa and φb
FIG. 2 is a diagram showing the process derived from the above in polar coordinates of complex numbers, (1) is a diagram showing that the 0th-order diffracted light amplitude ψ 0 is determined from amplitude reflectances φa and φb, and (2) is amplitude reflection. A diagram showing that the first-order diffracted light amplitude ψ1 is determined from the ratios φa and φb. (3) is a diagram in which ψ0, ψ1 obtained from (1) and (2) of the same figure are written on the same polar coordinates. is there.

FIG. 8 is a diagram showing an example of a result of a simulation which is a cause of the idea of the present invention, in which (1) shows a mark step h.
Where 50, 100, 150 and 200 nm are shown as abscissa resist thickness d [μm] and ordinate position detection error [μm], (2) shows mark step h of 250, 300, 350, 400
It is a figure which shows the result in the case of nm similarly to (1).

9A is a diagram showing a cross section of a lattice mark which is a premise of the simulation result of FIG. 8, and FIG. 9B is an enlarged diagram of one period thereof.

[Explanation of symbols]

15a, 15b Detector (light receiving means) 100 Position detecting device 102 Light transmitting optical system 200 Exposure device W Wafer (substrate) MG Grating mark RRG Rotating radial grating plate (relative scanning means) SAn Phase difference detecting circuit (step detecting means, position detecting means) CAn detection position correction circuit (part of position detection means) WEGA weighted EGA circuit (calculation means)

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification number Office reference number FI technical display location H01L 21/30 525W 525X 525N

Claims (10)

[Claims] 1. A substrate having a fine pattern and a plurality of (M or more) lattice marks for position detection, which have a periodicity (period = P) in a specific direction and repeat a concave portion and a convex portion. On the other hand, a position detection method for detecting the positions of the plurality of lattice marks in the periodic direction, and statistically processing the detected positions, wherein the position of the fine pattern on the entire surface of the substrate is detected, A pair of coherent light beams are respectively incident on a plurality of grating marks, and interference fringes having a period of amplitude distribution of 2P / N (N is a natural number) whose period direction is equal to that of each lattice mark are formed. A first step of forming each of the interference fringes and the grating mark relative to each other in the periodic direction; a first light of the reflected / diffracted light of the incident light beam by the plurality of grating marks Bee Of the first reflected light of the second light beam and the Nth-order diffracted light of the first light beam. A second step of sequentially and separately receiving and photoelectrically converting a second combined light beam which is a combined light beam; and a light amount signal of each of the first and second combined light beams sequentially obtained in the second step. A third step of sequentially detecting the positions of the grid marks at the plurality of locations from changes caused by the relative scanning; a first step and a first step of sequentially obtaining the step equivalent amounts of the grid marks at the plurality of points in the second step. Based on the change of each light quantity signal of the combined light flux of 2 with the relative scanning and the design data of each grating mark, the step of each grating mark is set in the medium near the surface of each grating mark of the incident light beam. 4th which is sequentially calculated as the remainder divided by half the wavelength The step equivalent amount of the j-th (j is 1 to M) grating mark in the plurality of places is 1 / the wavelength of the incident light beam near the surface of the j-th grating mark.
When it is close to 4, a large weight is given to the detection position of the jth grating mark, and when it is far from ¼ of the wavelength, a small weight is given to the detection position of the jth grating mark. And a fifth step of statistically processing the detection positions of the plurality of grid marks to calculate the position of the fine pattern on the entire surface of the substrate. 2. In the first step, a plurality of pairs of coherent light beams are made incident on the grating marks at the plurality of places, respectively, and a plurality of sets having different wavelengths are incident, and a period of amplitude distribution is 2P / N (N.
Is a natural number), and the interference fringes of which the periodic direction is equal to the periodic direction of each of the lattice marks are formed respectively, and the interference fringes and the lattice marks are relatively scanned in the periodic direction. The process of four steps is performed for each wavelength to detect the position and step equivalent amount of each grating mark for each wavelength, and in the fifth step, the position detection result and step of the grating mark at each wavelength are detected. The position of the fine pattern on the entire surface of the substrate is calculated by performing the weighted statistical processing on the detection positions of the plurality of grid marks for each wavelength using a considerable amount of detection results. The position detection method described in. 3. A predetermined shot area having a fine pattern formed on a photosensitive substrate is sequentially positioned at an exposure position,
An exposure method in which a pattern formed on a mask is transferred to each of the shot areas to perform overprint exposure, and each shot on the photosensitive substrate is detected by the position detection method according to claim 1 or 2 prior to the start of exposure. An exposure method characterized in that after the array coordinates of the areas are calculated, the shot area is sequentially positioned at the exposure position in accordance with the calculated array coordinates, and overprint exposure is performed. 4. A substrate having a fine pattern and a plurality of (M or more) grid marks for position detection, each of which has a periodicity (cycle = P) in a specific direction and which repeats recesses and projections. On the other hand, a position detecting device for detecting the positions of the plurality of lattice marks in the periodic direction, and statistically processing the detected positions, the position detecting device detecting the position of the fine pattern on the entire surface of the substrate, Amplitude distribution period is 2P on multiple grid marks
/ N (N is a natural number) and a light transmitting optical system for respectively injecting a pair of coherent light beams so as to form interference fringes whose periodic directions are equal to the periodic directions of the respective grating marks; Of the reflected / diffracted light of the incident light beam by the grating mark, a first combined light beam that is a combined light beam of the specularly reflected light of the first light beam and the Nth-order diffracted light of the second light beam, and the second light beam. Light receiving means for separately photoelectrically converting a second combined light beam, which is a combined light beam of the specularly reflected light of the light beam and the N-th order diffracted light of the first light beam; and the lattice marks and the interference fringes. And a relative scanning unit that relatively scans in the periodic direction;
Position detecting means for respectively detecting the positions of the lattice marks at the plurality of locations based on changes in the light amount signal of the second combined light flux due to the relative scanning; Based on the change of the light quantity signals of the first and second combined light fluxes obtained by the relative scanning and the design data of each of the grating marks obtained by the above, the step of each of the grating marks is set to each grating of the incident light beam. Step difference detecting means for detecting each as a remainder divided by half of the wavelength in the medium near the mark surface; and the step equivalent amount of the j-th (j is 1 to M) grating mark in the plurality of places, When it is close to ¼ of the wavelength of the incident light beam in the vicinity of the surface of the j-th grating mark, a great weight is given to the detection position of the j-th grating mark, and the distance from ¼ of the wavelength If there is Position detecting device and a calculating means for calculating the position of the fine pattern of the j-th detected position of the grating mark with a small weight by statistically processing the detected position of the grating mark of the plurality of positions the entire surface of the substrate. 5. The position detecting means calculates the average value of the detected positions based on the respective phases of the light amount signal changes of the first combined light flux and the second combined light flux associated with the relative scanning. The position detecting device according to claim 4, wherein the position detecting device detects the position of the lattice mark. 6. The step detecting means includes a ratio of widths of the concave portion and the convex portion of each of the lattice marks, and light amount signals of the first combined light flux and the second combined light flux associated with the relative scanning. The position detecting device according to claim 4 or 5, wherein the step equivalent amount of each of the lattice marks is calculated based on the phase difference of change and the contrast. 7. The reflection / diffraction 0 from each of the lattice marks
The step detecting means further includes a light quantity ratio measuring means for measuring a light quantity ratio between the second-order diffracted light and the N-th order diffracted light, and the step detecting means includes the ratio of the widths of the concave and convex portions of the lattice marks and the relative scanning. A step equivalent amount of each of the lattice marks is calculated based on the phase difference of each light amount signal change between the first combined light beam and the second combined light beam and the light amount ratio obtained by the light amount ratio measuring means. The position detecting device according to claim 4 or 5, wherein 8. Instead of the light-transmitting optical system, the periods of the amplitude distribution are sequentially set to 2P / N (N) on the plurality of grid marks.
Is a natural number), and in order to form interference fringes whose periodic direction is equal to the periodic direction of each of the grating marks, a light transmitting optical system for injecting a plurality of pairs of coherent light beams having different wavelengths is provided, and the light receiving means. The position detecting means and the step detecting means perform the respective processes for each of the plurality of different wavelengths, and the computing means uses the position detecting result and the step detecting result of each grating mark at each wavelength. The position detecting device according to any one of claims 4 to 7, wherein the statistical processing is performed by using the position detecting device. 9. The light-transmitting optical system is characterized in that a plurality of pairs of coherent light beams having different wavelengths are time-divisionally incident on the grating marks at the plurality of positions for each wavelength. 8. The position detection device according to item 8. 10. An exposure apparatus for sequentially overlaying a predetermined shot area having a fine pattern formed on a photosensitive substrate at an exposure position, and transferring a pattern formed on a mask to each shot area to perform overprint exposure. The position detecting device according to any one of claims 4 to 9 is provided for detecting the array coordinate system of each shot area on the photosensitive substrate, and the overlay printing is performed according to the detected array coordinate system. An exposure apparatus characterized by performing exposure.