KR100541272B1 - Method of detecting mark position on substrate, position detection device by the method and exposure apparatus using the position detection device - Google Patents

Abstract

A method for detecting the position of a grating mark for position detection formed along an extremely small stepped structure on the surface of a flat object, wherein the illumination beam is different in the step of irradiating the grating mark with an illumination beam at a predetermined incident angle. It includes a plurality of interference beams consisting of the above n wavelengths λ ₁ , λ ₂ , λ ₃ , ..., λ _n , and the function of each wavelength is λ ₁ <λ ₂ <λ ₃ , ..., <λ when the _n, the n number of wavelength _{(1 / λ 1 - 1 /} λ 2) = (1 / λ 2 - 1 / λ 3) = ... = (1 / λ _n-1- / λ _n ) is set such that the relationship is approximately satisfied within a range of about ± 10%, and the grating upon irradiation of an illumination beam composed of the n wavelength components Photoelectric detection of a change in the amount of light of diffracted light occurring in a specific direction from the mark and determining a position of the grating mark based on a signal obtained according to the photoelectric detection. .

Description

A method of detecting a mark position on a substrate, a position detecting device by the method, and an exposure apparatus using the position detecting device

(Field of invention)

The present invention relates to a technique for aligning a relative position of a mask pattern and a photosensitive substrate, for example, applied to an exposure apparatus used in a photolithography process for exposing a mask pattern onto a photosensitive substrate when manufacturing a semiconductor device or the like. A technique for detecting a mark pattern on a photosensitive substrate.

(Related Background Art)

For example, in a photolithography process for manufacturing a semiconductor device, a liquid crystal display device, a thin film magnetic head, or the like, an image of a photomask or a reticle (hereinafter simply referred to as a reticle) having a transfer pattern is formed. Background Art An exposure apparatus that transfers onto a photoresist-coated wafer (or a photosensitive substrate such as a glass plate) by a projection exposure method or a projection exposure method through a projection optical system is used.

In such an exposure apparatus, it is necessary to accurately perform alignment of the reticle and the wafer prior to exposure. In order to perform this alignment, a position detection mark (alignment mark) formed by exposure-transferring and etching in a previous step is formed on the wafer, and the exact position of the wafer (circuit pattern on the wafer) is detected by detecting the position of the alignment mark. can do.

Recently, a lattice mark on a wafer (or reticle) is made into a one-dimensional or two-dimensional lattice, and two coherent beams symmetrically inclined in the pitch direction are projected on the lattice marks, The method of detecting the position and the position offset of the grating mark in the pitch direction by interfering two diffracted light components occurring in the same direction from the grating mark is described, for example, in (A) Japanese Patent Application Laid-Open No. 61-208220 (corresponding USP 4,828,392). And (B) Japanese Patent Application Laid-Open No. 61-215905 (corresponding USP 4,710,026). Publication (A) discloses a homodyne scheme in which the frequencies of two symmetric interference beams are equal, and publication (B) discloses a heterodyne scheme with a constant frequency difference between two symmetric interference beams. do.

Moreover, although the heterodyne position detection apparatus was applied to the through-the-reticle alignment system or the through-the-lens alignment system in a reduction projection exposure apparatus, (C) Unexamined-Japanese-Patent No. 2-227602. (Corresponding USSN 483,820, filed February 23, 1990), (D) JP-A-3-2504 (corresponding USP 5,118,953) and the like. In the heterodyne systems disclosed in these publications (C) and (D), a He-Ne laser beam is incident simultaneously on two acousto-optic modulation elements (AOM), and each AOM has a frequency difference of about 25 kHz, for example. It is driven by a high frequency drive signal (80 MHz on one side and 79.975 MHz on the other), giving a 25 kHz frequency difference between the diffracted beams emitted from each AOM. These two diffraction beams are a pair of incident light beams for irradiating a lattice mark on a wafer or a reticle at a predetermined crossing angle.

In the heterodyne system, a signal obtained by photoelectric detection of interference light (beat light beam) of two diffracted light components generated from a lattice mark is used as a reference alternating current signal with a frequency difference (25 kHz) between two incident light beams. The phase difference between and the reference AC signal is measured and detected as a position offset amount from a reference point with respect to the pitch direction of the grid mark.

In the heterodyne system described above, the more monochromatic the two incident light beams illuminating the lattice marks, the more the detection accuracy and resolution of the position offset are improved, and the position detection and alignment of the nanometer order is possible. However, the good monochromaticity of the two incident light beams means that the phase of the wavelength order between various diffracted lights generated from the grating marks tends to be sensitively changed depending on the asymmetry of the grating marks, the resist layer, and the like.

The influence of the double resist layer is a fatal problem in wafer alignment in the exposure apparatus, and unless a special technique of locally removing the resist in the mark portion is not used universally or unless the optical mark detection technique is abandoned, It is an unavoidable problem.

Thereby, the heterodyne system which reduces the influence by the resist layer or the influence of the asymmetry of the cross-sectional shape of the mark to enable more accurate position detection is disclosed in (E) Japanese Patent Application Laid-Open No. 6-82215 (July 14, 1993). Proposed USSN 091,501). In the publication (E), a plurality of beams having different wavelengths or white beams are used, and two diffraction beams obtained by irradiating the fixed diffraction grating on the first diffraction grating are incident on the first stage AOM, and the zeroth order diffracted by this AOM. By relaying the beam, the + 1st order diffraction beam, and the -1st order diffraction beam within the second stage AOM, for example, a pair of incident light beams by the first wavelength and a pair of incident light beams by the second wavelength are generated. The method of making and projecting these two pairs of incident light beams simultaneously to the grating mark on a wafer is disclosed.

At this time, the interference bit light generated from the grating mark and photoelectrically detected includes a first wavelength component and a second wavelength component, but they are photoelectrically detected in an added form as light quantity on the light receiving surface of the photoelectric element. For this reason, the phase difference of the interference bit light for each wavelength component by the influence of the thin film interference of a resist layer or the asymmetry of a mark cross-sectional shape is intensity-averaged, and more accurate position detection is attained.

Whether homodyne or heterodyne, the light source suitable for multiplexing the illumination beam generally has high coherence in itself, such as a gas laser light source or a semiconductor laser light source, and has sufficient light intensity. What can be obtained is selected. For this reason, the wavelength selected during conventional multi-wavelength determination is determined so that the oscillation center wavelength is dropped by an appropriate amount, for example, 20 to 40 nm, from among a light source suitable for practical use (a laser light source, etc., which has been used in the past). It was.

On the other hand, although the resist layer is apply | coated to the surface of the grating mark formed on the wafer with a substantially constant thickness (for example, about 0.5-1.5 micrometers), the thickness becomes gap with respect to some center value. In addition, the thickness of the resist layer varies depending on the position on the wafer, and the cross-sectional shape of the lattice mark (such as the minute step amount of the lattice groove) also varies slightly depending on the position on the wafer in accordance with the wafer processing process. The overall amplitude reflectance including the resist layer on the surface may vary greatly with respect to the illumination beam which has only been multi-wavelength with a specific wavelength component.

However, in the wavelength selection by the multi-wavelengthization, the lattice mark in which the amplitude reflectance can fluctuate greatly is not optimized so that high-precision position detection is necessarily achieved.

In addition, when the illumination beam used for position detection is made into the beam which has a some wavelength or predetermined wavelength bandwidth, and simultaneously receives the interference light containing the some wavelength component which generate | occur | produces from a grating mark with the same photoelectric element, illumination If there is a wavelength component having a high intensity in the beam, the interference light from the lattice mark also becomes strong at the portion of the wavelength component, which may be a problem in that averaging efficiency is obtained. Moreover, even if each wavelength component in an illumination beam is the same intensity | strength, each wavelength component of the interference light from a grating mark is influenced by the surface state (resistance thickness of a resist, degree of asymmetry of a grating mark, etc.) of photosensitive substrates, such as a wafer. A big difference may occur in every strength.

For this reason, even if the interference light including the plurality of wavelength components generated from the lattice mark is received by a single photoelectric device, it may not be possible to obtain good position detection accuracy depending on the surface state of the substrate.

(Summary of invention)

Then, an object of this invention is to provide the position detection method and apparatus which are hard to affect the surface state of the lattice form periodic pattern (lattice mark) formed in board | substrates, such as a wafer.

In addition, the present invention, when the illumination beam for detecting the grating mark on the substrate is multi-wavelength into a relatively spectral narrow coherent light beam emitted from each of the plurality of light sources, the cross-sectional shape of the grating mark (for example, the groove It is an object of the present invention to provide a position detection method and apparatus which reduces the temporal coherence of multi-wavelength illumination beams in consideration of the change in the depth of the light and the like and the change in resist thickness.

It is also an object of the present invention to provide a projection exposure apparatus that is multi-wavelength using a plurality of coherent light sources, produces an incoherent illumination beam, and uses it for alignment of grating marks on the photosensitive substrate. It is done.

Moreover, an object of this invention is to solve the problem at the time of photoelectric detection mentioned above, and to provide the position detection method or apparatus which is hard to be influenced by the surface state of board | substrates, such as a wafer. Moreover, an object of this invention is to provide the position detection method or apparatus which is hard to be influenced by the difference of the light intensity for every wavelength component, even when the grating pattern (mark) is illuminated by the illumination beam containing a some wavelength component. . In addition, the present invention, when measuring the position of the grating pattern by irradiating the grating pattern on the substrate with an illumination beam containing a plurality of wavelength components, a high-precision position that reduced the position measurement error of the grating pattern depending on the state of the substrate surface An object of the present invention is to provide an alignment device.

In order to achieve the above object, this invention gratings a pair of coherent light beams to the position detection grating mark MG formed in the surface of the flat-shaped object W, such as a semiconductor wafer or a glass plate with a micro step structure, A method of detecting the position of the lattice mark in the periodic direction based on a signal obtained by irradiating the mark with a symmetrical incident angle with respect to the pitch direction of the mark and photoelectrically detecting a change in the amount of light of diffracted light generated in a specific direction from the lattice mark, or Applies to the device.

In the present invention, the coherent light beam ± LF that irradiates the grating marks symmetrically has n (n = 3 or more) wavelengths λ ₁ , λ ₂ , λ ₃ . When multiple wavelengths of n pairs of light beams (± D _1n ) consisting of λ _n and the magnitude relation of each wavelength are λ ₁ <λ ₂ <λ ₃ ... <λ _n , (1 / λ _{The relationship between 1} -1 / λ ₂ ) = (1 / λ ₂ -1 / λ ₃ ) ... = (1 / λ _n-1 -1 / λ _n ) is approximately within the range of about ± 10% N wavelengths or a plurality of coherent light sources LS1, LS2, and LS3 were selected to be satisfied.

In a preferred configuration of the present invention, a pair of multiwavelength interfering light beams are heterodyned to have a predetermined frequency difference between two paired light beams, and are perpendicular to the lattice pitch direction from the grating mark MG. Photoelectric detection detects that the mutually interfering light of two primary diffracted light generated in the direction is intensity modulated at a bit frequency.

In addition, as a preferred application of the present invention, such a multi-wavelength position detecting device is integrated into the projection exposure apparatus as alignment means (mark detection means) of the TTR method, the TTL method, or the off-axial method. It was made.

In order to solve the problem of photoelectric detection, the present invention projects illumination light onto a diffraction grating MG formed on a substrate (wafer W or fiducial mark plate FG) to be position-detected, and the diffraction grating It applies to the method of detecting the position of a board | substrate by photoelectric detection of the diffracted light from (MG). First, (a) project an illumination beam (e.g., beams diffracted with reference grating RG ± D ₁₁ , ± D ₂₂ ) containing different wavelength components λ ₁ , λ ₂ on the diffraction grating MG and And a plurality of diffraction beams including respective wavelength components from the diffraction grating MG, and (b) an order difference (+ 1st order) consisting of a first wavelength component λ ₁ of the generated plurality of diffraction beams. And a first interference beam produced by the interference of two diffraction beams having -1st order or 0th order and 2nd order) at the first photoelectric element, and at the same time a second wavelength component (λ ₂₎ of the plurality of diffraction beams. The second photoelectric element receives a second interference beam made by interference of two diffraction beams each having aberrations (+1 and -1 or 0 and 2). And then, (c) calculating the photoelectric signals (I _m1) first location information (△ X ₁₎ a circuit unit (CU ₃₎ relates to the period direction of the diffraction grating (MG) based on from the first opto-electronic device , and calculates the photoelectric signal in the second position information (△ X ₂₎ the circuit unit according to the period direction of the diffraction grating (MG) to the base (1 _m2) (CU ₄₎ from the second opto-electronic device. And finally, (d) the weighting is changed in accordance with the amplitude value of the photoelectric signal from the first photoelectric element and the amplitude value of the photoelectric signal from the second photoelectric element to convert the first position information and the second position information into the circuit unit (CU). _The weighted average calculation was used to determine the position of the substrate on which the diffraction grating was formed.

Generally, the alignment and position measurement marks formed on the surface of a wafer or the like have a small step on the surface thereof, but due to wafer processes such as etching or sputtering in a semiconductor processing process or uneven application of a photoresist layer, It has some asymmetry. The asymmetry results in a drop in accuracy at the mark position detection.

In the interference alignment method using photoelectric detection of two diffracted light generated from the grating mark, and using the photoelectric signal, the asymmetry of the grating mark is an asymmetry of the amplitude reflectance of the mark itself, which acts on the deterioration of position detection accuracy. do. That is, when the depth of the groove bottom of the line constituting the lattice mark or the like has a difference in the lattice pitch direction or when there is a partial difference in the thickness of the resist layer, the absolute value and phase of the amplitude reflectance of the mark itself is determined by the bottom of the groove. The asymmetry is caused by the change in depth or resist thickness. As a result, the intensity and phase of the positive order diffracted light generated from the lattice mark in the right direction with respect to the 0th order light are different from the intensity and phase of the negative order diffracted light component generated in the left direction with respect to the 0th order light. It becomes. Among these, the difference in intensity hardly contributes to the deterioration of the position detection accuracy, but the change in phase greatly affects the position detection accuracy.

Therefore, the simulation results of the position detection accuracy in the heterodyne system using conventional illumination light of a single wavelength will be described with reference to FIGS. 1 and 2. This simulation assumes the case of irradiating a coherent incident light beam having a constant frequency difference from two directions symmetrical to the grating mark MG on the wafer covered by the resist layer PR as in the second diagram, and ± generated vertically from the grating mark MG. It is obtained by observing the mutually interfering light of the primary diffracted light, that is, the state (amplitude, phase, etc.) of the interference bit light by changing the wavelength.

2 schematically shows a one-dimensional lattice MG such as a wafer assumed in the simulation and a partial enlarged cross section of the resist layer PR applied to the surface thereof. Therefore, the pitch Pmg of the grating MG is set to 8 m, the duty is 1: 1, the step difference (or depth) T2 of the groove is 0.7 m, and the bottom of the grating MG is tapered (tilt) ΔS in the pitch direction. Asymmetry of 0.1% was set. The resist layer PR covering such a lattice MG has a thickness T ₁ from the surface of the upper part of the lattice MG as 0.9 µm, and a recess amount on the surface of the resist layer corresponding to the position of each bottom of the lattice MG. T was assumed to be ΔT ≒ 0.3T ₂ (0.21 μm). Such a grating structure of FIG. 2 is called a grating having an asymmetric amplitude reflectance.

1 shows the wavelength λ (μm) of the interference light synthesized with the illumination light or the ± 1st order diffracted light on the horizontal axis, and the relative amplitude of the variation (AC component) of the signal according to the change in the amount of light of the interference light on the vertical axis and the error of position detection. A vehicle (μm) was taken. In the simulation result of FIG. 1, the second component is set so that the alternating current component of the photoelectric signal according to the interference light received by the heterodyne system is exactly zero, that is, the wavelength lambda composed of only the direct current component matches the wavelength 0.663 mu m of the He-Ne laser. The lattice mark structure and the resist layer conditions of FIG.

As is apparent from this, when a laser light having a wavelength of 0.663 占 퐉 is used, it is understood that the detection error of the mark position becomes considerably large near the wavelength (about +20 nm). This is natural in the heterodyne system, because the phase difference measurement itself becomes impossible unless the alternating current component corresponding to the bit frequency is included in the photoelectric signal to be subjected to the phase difference measurement. The same applies to the case where the position detection is performed by the homodyne method under the lattice mark structure and the resist layer under the same conditions.

By the way, the amplitude reflectance of the grating mark itself varies not only with the depth of the mark or the thickness of the resist but also with the wavelength component of the illumination beam. For this reason, the variation of the amplitude reflectance of the grating mark may also be said to depend on the temporal coherence of the illumination beam.

Therefore, in the first aspect of the present invention for solving the problem on the illumination light side, the temporal coherence of the illumination beam is reduced and incoherence is achieved in a practical range with respect to the actual step height structure of the mark or the thickness state of the resist layer. As a precondition for this, three or more different center wavelengths λ ₁ , λ ₂ , λ ₃ ,. , multiple wavelengths are performed by a plurality of coherent light beams having? _n .

In addition, in the first aspect of the present invention, when considering a set of coherent light beams adjacent to each other in order of wavelengths as a basic condition of the non-coherence, the wave number value of two light beams obtained for each set ( Each wavelength is determined so that the difference of 1 / wavelength) is about ± 10% of each other.

Thus, as an example, the temporal addition of the illumination beam when multiple wavelengths are made with three coherent light beams (the wavelength width is sufficiently narrow) of the center wavelengths λ ₁ = 0.633 μm, λ ₂ = 0.690 μm, and λ ₃ = 0.760 μm By simulating coherence, the result is equal to the third degree. The abscissa in FIG. 3 represents the thickness (µm) of the resist layer applied on the silicon substrate, and the ordinate represents the reflectance obtained by adding the intensity of reflected light of each wavelength generated from the silicon substrate by irradiation of the multi-wavelength illumination beam. Is displayed. The amplitude of the reflectance variation with the change in resist thickness indicates the temporal coherence of the illumination beam. In addition, temporal coherence can be calculated by Fourier transform of the spectral distribution of the illumination light.

From the simulation results of FIG. 3, it is found that the variation amplitude of the reflectance is small and the temporal coherence is small, that is, the more incoherent in the resist thickness (or mark step) in the range of 0.5 µm to 1.7 µm. Able to know. In this way, the coherence can be reduced in a specific optical thickness or in a range of microsteps (0.5 μm to 1.7 μm in resist thickness), and the basic condition representing the relationship between the wavelengths of multiple wavelengths is ± 10%. Because it is made from within.

That is, when the wavelength λ ₁ is 0.633 μm of the He-Ne laser light, the wavelength λ ₂ is 0.690 μm of the commercially available semiconductor laser light, and the wavelength λ ₃ is 0.760 μm of the other commercially available laser light.

Δλ ₁₂ = 1 / λ ₁ -1 / λ ₂ = 0.1305

Δλ ₂₃ = 1 / λ ₂ -1 / λ ₃ = 0.1335

This is because the deviation is Δλ ₂₃ / Δλ ₁₂ = 1.023 (2.3% error). Therefore, in the case of an illumination beam multi-wavelength with three or more wavelengths deviating from such conditions, it is difficult to obtain a good incoherent state by reducing the fluctuation amplitude of the reflectance in a specific range as in the third degree.

Thus, the condition Δλ ₁₂ ≒ Δλ ₂₃ . By satisfying Δλ _{(n-1) n} (± 10%), a state close to incoherence can be satisfactorily obtained in a practical range, so that the resist thickness or the depth of the mark step is incoherent (for example, For example, the mark detection position obtained by photoelectric detection of multi-wavelength interference light obtained by the synthesis of ± 1st-order diffraction light generated from the mark, even if it changes slightly within 0.5 to 1.7 mu m), is multi-wavelength using light of three or more wavelengths. It means little change due to averaging by the sum.

Therefore, by making it multi-wavelength with interfering light of three or more wavelengths satisfying the conditions defined in the first invention, the case is less affected by the asymmetry caused by the fluctuation of the mark step and the variation of the resist thickness, and always Good position detection accuracy can be maintained.

As described above, the amplitude reflectance of the grating mark itself varies not only with the depth of the mark and the resist thickness but also with the wavelength of the illumination light (detection light). When the wavelength of the detection light is plural (or wideband), the amplitude reflectance of the mark itself is different for each wavelength component, and the position detection result is also different. Therefore, the position detection accuracy can be simulated by assuming the amplitude reflectance of the mark itself under each mark condition.

According to a second aspect of the present invention, when irradiating a grating mark with illumination light containing only a specific wavelength and photoelectrically detecting diffracted light generated from the grating mark, the intensity of the photoelectric signal (relative scanning of the light beam and the grating mark) When the amplitude of the accompanying signal change) becomes very small, it is conceived based on the simulation result that the position detection accuracy generally deteriorates. From the simulation results, in the second invention of the present application, the position detection result is weighted average of the position detection results for each wavelength using illumination light of different wavelengths in combination, even in a grating mark condition in which the amplitude of the photoelectric signal becomes very small when the illumination light of a single wavelength is used. This is to prevent extreme deterioration of detection accuracy.

Therefore, even under the same conditions as those in FIGS. 1 and 2, if a semiconductor laser having a wavelength? Of about 0.670 µm or 0.725 µm is used as the illumination beam, the detection error of the mark position can be sufficiently reduced. From this, using a two-color illumination beam of different wavelengths, such as a He-Ne laser and a semiconductor laser, the detected mark position (or based on the beam irradiation of the wavelength inducing a large amplitude in the detected signal (AC component) (or Attention (selection or weighting) of the position offset amount) becomes effective.

Alternatively, not only the interference light of two primary diffraction lights traveling in a specific one direction is detected but also the photoelectric detection of the interference light of zero order light and the second diffraction light traveling in other directions, and the mark position determined based on the signal. There is also a method to consider. The fourth turning the diffraction grating mark MG wavelength λ 2 of the irradiation beam ± L1 to the wavelength λ ₂ 2 of the irradiation beam is incident to ± L2, and the grating marks wavelength on the MG λ ₁ and λ _2, the pitch of the same intensity distribution of the _first to The beam incidence conditions in which interference fringes with Pif are generated are used, and then, when the pitch PmG of the lattice mark MG is regarded as Pmg = 2Pif, the diffraction light of 0th order light, ± 1st order, ± 2nd order It illustrates the occurrence.

In FIG. 4, the interference beam BM of the first-order diffracted light ± D1n traveling perpendicular to the grating mark MG includes components of both wavelengths λ ₁ and λ ₂ . Since the incidence angles of the 0th order light (normal reflected light) are slightly different with the beams ± L ₁ and ± L ₂ , four of ± D ₀₁ and ± D ₀₂ go in different directions corresponding to the respective beams ± L ₁ and L ₂ . . Thus, the first subscripts of D ₀₁ and D ₀₂ represent diffraction orders and the second subscripts represent wavelengths λ ₁ and λ ₂ .

By the way, the secondary light -D ₂₁ generated by the irradiation of the beam ± L ₁ proceeds in the direction of reversing the optical path of the beam + L ₁ and interferes with the zero-order light + D ₀₁ of the beam -L ₁ . Similarly, the other secondary light + D ₂₁ , -D ₂₂ , + D ₂₂ also proceed in the same direction as the corresponding zero-order light -D ₀₁ , + D ₀₂ , -D ₀₂ , respectively. The interference light of these zero-order light and the secondary light also changes in intensity in accordance with the relative displacement of the grating MG and the interference fringe similarly to the interference beam BM of the ± first-order light.

Therefore, considering only the wavelength λ ₁ , photoelectric detection of the primary component (interference beam BM of primary light ± D ₁₁ ) is performed to obtain the position (or position offset) of the mark, and at the same time, two secondary components (zero-order light). Photoelectric detection of each of + D ₀₁ and secondary light -D ₂₁ interfering light and 0 order light -D ₀₁ and secondary light + D ₂₁ interfering light), respectively, using individual signals of the two secondary components The value obtained by averaging the mark positions obtained by is obtained as the mark position. According to the magnitude relationship between the amplitude value of the signal of the primary component and the amplitude average value of each signal of the secondary component, either one of the mark position detected using the primary component and the offset of the mark position detected using the secondary component is determined. The method of selecting or performing a weighted average becomes effective.

As described above, since the direction of the diffracted light generated from the grating MG varies depending on the degree, the amplitude of the intensity change of the interference light of the order component traveling in the given direction decreases, so that the order of the diffracted light used for the mark detection varies. This is because the amplitude of the intensity change of the interference light of the order component advancing in the other direction does not decrease as much even when the deterioration of the deterioration occurs, and the detection accuracy may not be deteriorated.

This is also confirmed from the simulation results shown in FIGS. 5A and 5B. 5A and 5B show the amplitude and the position detection error of the variation (AC component) of the signal using the He-Ne laser having a wavelength of 0.633 μm as the irradiation beam and the step T ₂ of the grating MG in FIG. ₂ as a parameter. a simulation graph of the relationship, the pitch Pmg = 8㎛, duty 1: 1, the amount of taper (rate) △ S = 0.1% is one as the thickness T ₁ of the upper surface grid of the resist layer PR as 1.15㎛. FIG. 5A is a simulation in the case of the interference beam BM of the primary component (primary light ± D ₁₁ ), and FIG. 5B is an interference of the secondary component (0 order light ± D ₀₁ and the secondary diffraction light ± D ₂₁ ). Simulation of the light case.

As understood from FIG. 5A and FIG. 5B, the amplitude component of the signal obtained by exposing and detecting each interference light of the primary component and the secondary component largely changes according to the subtle change in the shape of the grating mark (step T ₂ ). do. For example, in Fig. 5A, when the step T ₂ of the grating is 0.86 mu m, the amplitude of the intensity change of the interference light of the primary component becomes extremely small, and as a result, the position detection error also increases rapidly. However, when the step T ₂ is 0.86 占 퐉 in Fig. 5B, the intensity change of the interference light of the secondary component is relatively large, and the deterioration of the position detection error is small. In addition, although the amplitude of the signal change part in FIG. 5A and 5B is all represented by the relative value, the scale is matched with FIG. 5A and 5B.

Thus, the above-mentioned algorithm also employs the detection of the position of the grating mark using the interference light of the primary component and the position detection of the grating mark using the interference light of the secondary component, and employs any one of the above results. As is clear from the simulation of FIG. 1, it is preferable to weight-average the detection position (or position offset) obtained by the illumination light of the several wavelength component using wavelength dependency.

As described above, by plural wavelengths of the detection light and averaging the mark position information obtained for each wavelength component, it becomes possible to detect the position with higher accuracy than conventionally. As shown in FIG. 1, when the amplitude (change component) of the light quantity signal of the diffracted light (interfering light) of a given wavelength is small, the probability of the positional detection accuracy using the diffracted light of that wavelength deteriorates. High simulation results can also be obtained. Therefore, when detecting diffracted light (interfering light) of a plurality of wavelength components, a mark position detected for each wavelength component is given a small weight for the small amplitude of the signal variation and a large weight for the large amplitude. Average In this way, only a small weight is automatically added to the detection result of the mark position using the diffracted light of the wavelength component with a high possibility of including a large error, and the final mark position detection result also maintains its precision exactly.

In addition, when detecting signals of secondary components (interfering light of 0th order light and secondary diffracted light), diffraction light (interfering light) There is no fear that the mark position cannot be detected due to the cancellation effect of each wavelength as described later when receiving light.

Further, in the present invention, since the incident light beams for each wavelength are sequentially switched and irradiated at the time of detecting the lattice mark, the photoelectric element which detects the interference light by the primary component or the interference light by the secondary component is applied. It is not necessary to provide a plurality of sets of photoelectric elements for photoelectric detection for each wavelength component, and the wavelength selection means can be omitted because the interference light is discriminated for each wavelength component.

(Detailed Description of the Preferred Embodiments)

Next, a first embodiment of the present invention will be described with reference to FIGS. 6 and 7, but a position detection apparatus using a heterodyne system is illustrated. In Figure 6, the three laser light sources LS _1, LS _2, LS ₃ is emitted, each with a different wavelength λ _1, λ _2, the laser beam LB1, LB2, LB3 of λ _3. As an example, the laser light source LS ₁ is set to a He-Ne laser light source of λ ₁ = 0.633 μm, the light source LS ₂ is a semiconductor laser light source of λ ₂ = 0.690 μm, and the light source LS ₃ is a semiconductor laser light source of λ ₃ = 0.760 μm; , The relationship between the wavelengths is selected as λ ₁ <λ ₂ <λ ₃ .

These laser light sources are generally commercially available, but due to the problem of heat generation and miniaturization of the laser tube, when all three light sources are to be used as semiconductor laser light sources, the second harmonic beam is emitted from a semiconductor laser light source having an oscillation wavelength of 1.20 µm. A beam having a wavelength of λ ₁ = 0.600 μm may be made incident on the generator SHG, and the beam may be used instead of the beam LB ₁ from the He-Ne laser light source LS _{1 in} FIG. 6. At this time, the light source LS ₂ of FIG. 6 becomes a semiconductor laser light source having a wavelength λ ₂ = 0.640 μm, and the light source LS ₃ becomes a semiconductor laser light source having a wavelength λ ₃ = 0.690 μm.

Also in this case, Δλ ₁₂ = 1 / λ ₁ -1 / λ ₂ = 0.1042

Δλ ₂₃ = 1 / λ ₂ -1 / λ ₃ = 0.1132

Become,

Since DELTA lambda ₂₃ / DELTA lambda ₁₂ = 1.087 (8.7% error), the conditions specified in the present invention are satisfied within the range of ± 10%.

The result of the simulation of the state of temporal coherence in this case like FIG. 8 is shown in FIG. As distinguished from the graph of FIG. 8, since the three wavelengths are shifted to the shorter wavelength side as a whole than in the case of FIG. 3, the non-coherent region is shifted to a range (0.7 to 2.0 µm) in which the resist thickness (or mark step) becomes large as a whole. do. However, it can be seen that in the range (0.7 to 0.2 mu m), the variation in reflectance is extremely small, and incoherence is satisfactorily performed as in the case of FIG.

Returning to the description of FIG. 6 again, the three beams LB ₁ , LB ₂ , LB ₃ are synthesized into one coaxial beam LB0 via mirror MR, dichroic mirrors DCM ₄ , DCM ₅ , reflected from the mirror MR and rotated radially. It enters the grid plate RRG. The diffraction grating PRG rotates at a high speed around the isotropic rotation axis C ₀ in one direction, and acts as a frequency modulator (frequency shifter) that increases and decreases the diffracted light frequency of each order diffracted by the grating RRG by an amount corresponding to each speed. do.

7 is an enlarged perspective view of the rotating radial grating plate RRG, so that the rotation axis C0 is set parallel to the Z axis of the XYZ coordinate system, and the circular grating plate RRG is formed with a transmissive phase diffraction grating RG over 360 °. When the beam LB0 is incident perpendicularly to the grating RG of the grating plate RRG, various diffracted light is generated in addition to the 0th order light D0. In this embodiment, the heterodyne system is realized by using the ± 1st order diffracted light. FIG. 6 and FIG. 7 show only the ± 1st order diffracted light from the lattice RRG.

In addition, the grating RG of the diffraction grating RRG, the beam of wavelength λ ₁ of the beam first-order diffracted beam produced from the beam LB ₂ on LB 1st-order diffracted beam produced from the ₁ ± D _11, the wavelength λ ₂ ± D _12, and the wavelength λ ₃ LB _A first diffraction beam, ± D ₁₃ , made of _three occurs. The diffraction angle θ _n of the primary diffraction beam for each wavelength is expressed as follows.

sin θ _n = λ _n / Prg

So n represents the number of wavelengths and Prg represents the pitch of the grating RG.

On the other hand, if the first diffraction beam receives a constant frequency shift Δf irrespective of the wavelength, and the speed at which the lattice RG of the lattice RRG crosses the beam LB0 is V, it is expressed as Δf = V / Prg, and the + first order diffraction △ f beam is high, as for the frequency of the zero-order light D0, -1 order diffracted beam is lowered by △ f for the frequency of the zero-order light D _0. As a result, the rotating radial grid RRG acts as a frequency shifter.

As shown in FIG. 6, the incident light beam ± LF consisting of the three- _dimensional diffraction beams ± D _1n (n = 1, 2, 3) and the zero-order light D ₀ are generated by the collimator lens 10. The chief rays of light are changed to be parallel to each other, reaching the beam selection member 12. This beam selection member 12 functions as a spatial filter placed on a so-called Fourier transform surface, so that the 0th order light D ₀ is blocked, and the incident light beam + LF by the first diffraction light + D _1n passes.

The incident light beam ± LF then arrives at the beam splitter (half mirror) 20 through adjustment optics 14, 16, 18 composed of parallel flat glass of varying inclination amount. The adjustment optical system 14 has a function of moving the incident light beam ± LF with respect to the optical axis of the lens 10 without changing the distance in the Fourier space of the incident light beam + LF and the incident light beam -LF, and the adjustment optical system 16 18 has a function of individually adjusting the positions of the incident light beam + LF and the incident light beam -LF with respect to the optical axis. The incident light beam ± LF is divided into two in the beam splitter 20, one of which enters the objective lens 22 as the irradiation optical system, and the other of the incident light beam ± LF of the incident light beam ± LF through the wavelength selection filter 24 Only the primary beam, and thus the primary beam ± D ₁₂ of wavelength λ ₂ is selected and enters the condenser lens (Fourier transform lens) 26 through the filter 24.

On the other hand, the incident light beams ± LF incident on the objective lens 22 become parallel beams, respectively, and simultaneously irradiate the grating MG on the wafer W at different angles. Thereby, on the grating MG, an interference fringe produced by the interference of the incident light beam ± D ₁₁ of wavelength λ _1, an interference fringe produced by the interference of the incident light beam ± D ₁₂ of wavelength λ ₂ , and an incident light beam ± D ₁₃ of wavelength λ ₃ . Three of the interference fringes created by the interference of the two layers appear at the same pitch and in the same phase. Further, because of the frequency difference 2 · Δf between the incident light beams ± LF and -LF, the interference fringe is observed to move the constant velocity on the grating MG in one direction. The moving speed is proportional to the speed V of the rotating radial grating RRG grating RG.

Further, as is apparent from FIG. 6, the wafer W surface (lattice MG) and the radial grating plate RRG are arranged so as to be conjugate to each other by the combined optical system of the collimator lens 10 and the objective lens 22. Therefore, the diffraction image by the +/- 1st-order diffraction light of the grating RG of the radial grating plate RRG is formed on the grating MG of the wafer W according to the magnification of the synthetic optical system. Since the zeroth order diffracted light component D ₀ is shielded, the diffraction image (interference intensity distribution) projected onto the wafer W is formed at one half of the pitch of the grating RG. The pitch Pif on the wafer W of the interference fringe is set to 1/2 of the pitch Pmg of the lattice MG.

In the above configuration, the incident angle φ _n of the incident light beams ± D ₁₁ , ± D ₁₃ of each wavelength component emitted from the objective lens 22 and irradiating the lattice mark MG of the wafer W (n is the number of wavelengths 1, 2, 3) Is set to the relationship expressed by the following expression.

sinφ _n = ± λ _n / Pmg

When the above relationship is satisfied, the first-order diffracted light in the grating MG is generated vertically by irradiation of the incident light beam ± LF. That is, the interference beam BM which the primary diffraction light which generate | occur | produced perpendicularly by irradiation of the incident light beam + LF and the primary diffraction light which generate | occur | produced vertically by irradiation of the incident light beam -LF generate | occur | produces. The interference beam BM is a bit light intensity-modulated at the frequency 2Δf. As described above, in order to generate the ± first-order diffracted light (interference beam BM) in the same direction, the angle of incidence of the objective lens 22 is F0 and the incident angle of each wavelength beam of the incident light beam is φ _n . The interval DL _n from the optical axis on the Fourier transform plane of the incident light beam ± LF for each wavelength

_{DL n = F 0 · sinφn =} ± F 0 · λ n / Pmg (n = 1, 2, 3)

Set to. The setting of the interval DL _n for each wavelength can be adjusted by appropriately determining the pitch of the grating RG of the rotational radial grating plate RRG and the focal length of the collimator lens 10.

In addition, since the interference fringe formed on the wafer W is coupled by the diffraction image of the lattice RG of the radial lattice RRG, in principle, the interference fringe of one of the three wavelengths λ ₁ , λ ₂ , and λ ₃ is obtained. If the pitch and the pitch of the lattice mark MG of the wafer W are integer multiples, then the pitch of the interference fringes by other wavelength components is also in that relationship itself, and then the interference fringes for each wavelength component are completely matched with each other. Phase offset, does not cause position offset.

However, in practice, interference fringes for each wavelength component produce position offsets, position offsets, and pitch offsets, depending on the degree of chromatic aberration of the optical system such as the objective lens 22 and the collimator lens 10.

Thus, the adjustment optical systems 14, 16, and 18 in Fig. 6 are used to correct this offset. These optical systems 14, 16, and 18 are made of parallel flat glass, and when the material having a large color dispersion is used, the position offset and the phase offset of mutual interference fringes formed on the wafer W for each wavelength component are minute. It is possible to change. Alternatively, as the adjusting optical systems 14, 16, and 18, a parallel plate glass having a small color dispersion and a parallel plate glass having a large color dispersion are combined to adjust the inclination of the interference fringes for each wavelength component by inclination adjustment of the parallel plate glass having a high color dispersion. With respect to the overall inclination error on the wafer of the incident light beam ± LF generated by the correction, a parallel flat plate with small color dispersion can be corrected by adjusting the inclination of the glass.

The interference beam BM generated vertically from the grating MG illuminated by the interference fringe as described above passes through the objective lens 22 and the beam splitter 20 to reach the spatial filter 28. The spatial filter 28 is disposed at or near the Fourier transform plane with respect to the objective lens 22, and has an opening for transmitting only the interference beam BM (± first order diffracted light) in this embodiment. The interference beam BM that has passed through the spatial filter 28 is converted into a parallel beam in the lens system (inverse Fourier transform lens) 30 and then reflected by the mirror 32 and received by the photoelectric element 36A.

The photoelectric element 36A simultaneously receives an interference beam BM including three wavelengths λ ₁ , λ ₂ , and λ ₃ , and the interference beam BM is intensity modulated at a bit frequency 2Δf. For this reason, the photoelectric signal I _m1 of the photoelectric element 36A becomes an alternating current waveform level changing to a sine wave shape at the same frequency as the bit frequency 2Δf while the interference beam BM from the grating mark is present.

On the other hand, the primary beam ± D ₁₂ selected by the wavelength selective filter 24 and incident on the condenser lens 26 is weighted and irradiated onto the transmission-type reference grating SG. Nevertheless, the reference grating SG is disposed in conjugate with the rotational radial grating plate RRG with respect to the composite optical system of the collimator lens 10 and the condenser lens 26. For this reason, the one-dimensional interference fringe by the two-beam interference of the primary beam +/- D ₁₂ is formed also on the reference grating SG, and it moves at a speed corresponding to the bit frequency 2Δf.

Thus, if the pitch of the reference grating SG and the pitch of the interference fringe are appropriately determined, the ± first-order diffracted light generated from the reference grating SG proceeds to be the interference beam B _ms in the same direction, which passes through the spatial filter 38 to transmit the photoelectric element. 40 is received. The photoelectric signal I _ms of the photoelectric element 40 becomes a waveform which level changes to a sine wave shape at the same frequency as the bit frequency 2Δf, and the signal I _ms is a reference signal of the heterodyne system.

In the above configuration, the reference lattice SG is formed by depositing a chromium layer on a glass plate, and etching the chromium layer so that transparent lines and light shielding lines are alternately formed. Thus, at least the asymmetry of the lattice mark MG on the wafer W and the resist layer An ideal grating without problems, ie amplitude transmittance, is made as a symmetric grating. For this reason, the pair of incident light beams irradiated to the reference grating SG can obtain sufficient precision only by the incident light beams corresponding to one of the three wavelengths λ ₁ , λ ₂ , and λ ₃ . Of course, all three primary beams ± D ₁₁ , ± D ₁₂ , and ± D ₁₃ included in the incident light beam ± LF may be simultaneously irradiated to the reference lattice SG to form a multicolor interference fringe similar to the lattice mark MG on the wafer. .

If a multi-color interference fringe is formed on the reference grating SG as described above, and the interference beam B _ms generated from the reference grating SG is configured to be photoelectrically detected for each wavelength, the reference signal according to the wavelength λ ₁ and the wavelength λ ₂ Since the reference signal and the reference signal corresponding to the wavelength λ ₃ can be obtained separately, the position measurement of the grating mark MG can be performed for each wavelength. In addition, even if the interference fringes generate constant position offsets (phase offsets) for each of three wavelength components formed on the wafer W, it is possible to measure the offset amount in advance. It will be described later in detail.

By the way, the wafer W shown in FIG. 6 is located on the wafer stage WST which moves two-dimensionally in the surface (XY plane) perpendicular | vertical to the optical axis of the objective lens 22. FIG. The two-dimensional movement on this stage WST is performed by a drive source 42 including a drive motor, and may be either a method of rotating a feed screw by a motor or a method of directly moving the stage body by a linear motor. good. Next, the polar coordinate position of the stage WST is measured by the laser interferometer 44. The measured value of the laser interferometer 44 is used for feedback control of the drive source 42.

In addition, the reference mark plate FG is provided in a part of wafer stage WST. On the mark plate FG, a reflective intensity grating (pitch is the same as the lattice MG on the wafer) formed by patterning a line and space with a chromium layer on the surface of the quartz glass. Because of this, the intensity grating is not asymmetrical, unlike the grating mark MG formed with irregularities on the wafer W, and has a characteristic that the diffraction efficiency does not depend on the wavelength of the illumination light (or the detection light), i. There is no characteristic. Next, the reflectance of the chromium layer also does not change in the wavelength band (normally 0.5 to 0.8 mu m) of the illumination light for position detection.

In the configuration of FIG. 6, a semiconductor laser is used as the light source, but in this case, a stereoscopic optical system for removing astigmatism between the semiconductor lasers LS ₂ and LS ₃ and the dichroic mirrors DCM ₄ and DCM ₅ It is good to provide a plurality of parallel flat glass or the like) and to make the beam component the same diameter for each wavelength component of the beam LB ₀ synthesized in one piece. Also in other cases, it is preferable to provide a beam shaping optical system in which the diameter of the beam LB ₀ after synthesis is constant for each frequency component.

In FIG. 6, for the sake of simplicity, the rotational radial lattice plate RRG is used as the frequency shifter, but only the first zebra which uses two other acousto-optic modulators (AOMs) or oscillates at the center wavelength lambda ₁ and the center Only the second agent oscillating at the wavelength λ ₂ may be used as the light source. In the case of the laser only, the bit frequency of the interfering beam for photoelectric detection increases accordingly because it generally oscillates two laser beams having complementary polarization directions and gives a frequency difference of several hundred kiloHz between the two beams. 36A and 40 use a PIN photodiode, a photomar, or the like with high responsiveness.

In addition, the various dichroic mirrors shown in FIG. 6 may be replaced by dispersing elements such as prisms. In this case, one prism has the same function as, for example, a combination of two dichroic mirrors DCM ₄ and DCM ₅ .

An example of a position detection and position control circuit suitable for the apparatus of FIG. 6 will be described with reference to FIG. In the heterodyne scheme of FIG. 6, while the lattice mark MG on the wafer W or the interference beam BM from the reference mark plate FG is occurring, the signals I _m1 and I _ms from the photoelectric elements 36A and 40 are set to 10A. Fig. 10B also shows a sinusoidal alternating current waveform.

FIG. 10B shows a temporal intensity change of the signal I _ms serving as a reference signal, and FIG. 10A shows an example of the temporal intensity change of the signal I _m1 when receiving the interference beam BM from the lattice mark MG on the wafer W. FIG. . Therefore, on the basis of the phase of the signal I _ms , it is assumed that the phase of the signal I _m1 is shifted by -ΔΨ ₁ with respect to the signal Ims. In addition, the amplitude (peak to peak of the alternating current component) of the signal I _m1 is assumed to be E ₁ .

In the circuit block shown in FIG. 9, each signal I _m1 , I _ms is input to an analog-to-digital conversion (A / D converter) circuit unit 50, and thus a clock signal from the sampling clock generation circuit 52. In response to the (pulse) C _ps , the instantaneous intensity level of each signal is converted into a digital value. The frequency of the clock signal C _ps is determined higher enough than the bit frequency of the signal I _m1, I _ms, the clock signal C _ps is sent also to the waveform memory circuit unit 54, a digital from the A / D converter 50 It is used to update the memory address when storing a value (data).

Therefore, in the waveform memory circuit unit 54, two waveform data shown in FIGS. 10A and 10B are digitally sampled over a predetermined period (e.g., 10 cycles or more) of the signals I _m1 and I _ms . do. At this time, since the two signals I _m1 and I _ms are simultaneously sampled by the common clock signal C _ps , it is assumed that each waveform data in the waveform memory circuit unit 54 has no offset on the time axis. In addition, when the rotational radial grating plate RRG is used, since the bit frequency is the upper limit of several kHz, the clock signal C _ps may be about tens of kHz. In addition, when using a frequency shifter in which two AOMs are arranged in series as in (E) JP 6-82215 A, the bit frequency is determined to be twice the frequency difference of the high frequency modulated signal applied to each AOM. It is relatively free to decide.

Further, each waveform data in the memory circuit unit 54 is read into the calculation circuit unit 56 of the phase difference ΔΨ and the position offset ΔX, so that the phase difference ΔΨ ₁ as shown in FIGS. 10A and 10B is digitally calculated. It is calculated by the Fourier Integral Method. As previously assumed, if the pitch Pmg of the lattice mark MG of the wafer W and the pitch Pif of the interference fringe irradiated thereon are set to Pmg = 2Pif, one period of each waveform of FIGS. 10A and 10B is equal to Pmg / 2. It corresponds.

In general, since the phase difference measurement is performed in a range of ± 180 degrees, the calculation circuit 56 converts the calculated phase difference ΔΨ ₁ in the range of ± Pmg / 4 according to the conversion equation ΔX = Pmg · △ Ψ / 4π. Is converted to ΔX. This offset amount ΔX represents an offset within ± Pmg / 4 of the grating mark MG with respect to the reference grating SG. Therefore, if the resolution of the phase difference measurement is about 0.2 °, the resolution of the offset amount is (0.2 / 180) Pmg / 4, and if the pitch Pmg is 4 µm, the practical range is about 0.002 µm (2 nm). Can be.

The offset amount ΔX thus obtained is an offset in the pitch direction of the grating mark MG with respect to the reference grating SG, and the data is sent to the position controller (display controller) 62 and the wafer W is aligned in real time (position). Is sent to the servo control circuit unit 64 as well.

The servo control circuit unit 64 has two functions, one of which is a function (direct servo mode) for feedback control of the drive source 42 until the offset amount ΔX becomes a predetermined value. In the case of this function, the operations of the A / D converter circuit 50, the memory circuit unit 54, and the offset amount calculating circuit unit 56 are repeated in sequence, and the offset amount ΔX for every extremely short time (for example, msee.). The value of is calculated.

On the other hand, one function of the servo control circuit unit 64 is a function (interferometer servo mode) which moves the wafer stage WST based on the measured value of the laser interferometer 44. This function is, for example, positioning the grid of the reference mark plate FG on the stage WST or the grid mark MG on the wafer W directly below the objective lens 22, or on the basis of the position of the detected grid mark MG on the wafer W. It is used when positioning an arbitrary point just below the objective lens 22.

In this interferometer servo mode, the target position information of the wafer stage WST is output from the position controller 62 to the servo control circuit unit 64, and the control circuit unit 64 reads the current state of the stage WST read from the laser interferometer 44. The drive source 42 is feedback controlled so that the deviation between the position and the target position falls within a predetermined allowable range (e.g., ± 0.04 mu m).

When the direct servo mode is executed following the interferometer servo mode, the servo enable range in the direct mode is ± Pmg / 4 with respect to the pitch Pmg of the grid mark MG. This is because if it is shifted more than this, positioning is performed with half offset of one pitch of the grid mark MG. Therefore, the positioning allowable range of the stage WST in the interferometer servo mode is not normally set to ± 0.04 占 퐉, and the allowable range is set only when the lattice mark MG (or the reference mark plate FG) is detected. may be switched to [alpha]].

For example, when the pitch Pmg is 4 占 퐉, the allowable range is about 0.5 占 퐉, so that positioning servo can be performed with a looser accuracy than the conventional allowable range (占 -0.04 占 퐉), so that the final time is shortened. Then, if it enters the loose allowable range (± 0.5 占 퐉), it is immediately switched to the direct servo mode, thereby enabling high-precision positioning (alignment) at high speed.

The position controller 62 also has a function of displaying the coordinate position of the grid mark MG and the obtained offset amount [Delta] X, unlike the above-described servo mode switching instruction.

Although the position detection apparatus and the detection method according to the first embodiment of the present invention have been described above, the lattice marks formed on the semiconductor wafer due to the alignment during the photolithography process are heterodyne (or homodyne) as in the present embodiment. In the case of adopting the interference mark detection method, the detection method of the grating mark may be made as a relatively small low step mark.

In the interference mark detection method, since the mark itself is composed of about 10 to 20 line and space patterns, and the diffraction light generated from the entirety of those lines and space patterns is photoelectrically detected, even a low level grating mark is sufficient to detect. Mark position detection of the amount of light is possible.

FIG. 11 shows a partially enlarged cross-sectional structure of the pitch direction of the lattice mark MG 'of the low step, and the step of the lattice mark MG' is the height difference T ₂ between the mark surface S _om and the bottom of the lattice groove. . The model used in the above-described simulation of FIG. 1 was lattice mark MG of the cross-sectional structure shown in FIG. ₂ , and the stepped amount T ₂ assumed therein was 0.7 µm. On the other hand, in the low step mark, the step amount T ₂ in FIG. 11 is 0.5 µm or less, and asymmetry according to the wafer processing process hardly occurs. Therefore, since fluctuations in amplitude reflectance due to at least asymmetry are suppressed little, it is considered to contribute to improving the accuracy of mark position detection.

In addition, the average thickness (thickness to be applied) T ₀ (or T 1) of the resist layer PR in FIG. 11 is usually set in the range of 0.7 µm to 1.2 µm. For this reason, the change range T ₁ to T ₃ of the substantial resist thickness when the lattice mark MG 'having a low level is used is recessed by the surface of the resist layer S _or in the groove portion of the lattice by ΔT = 0.3T ₂ , and T _{Since 3} ≒ T ₂ + (T ₁ -ΔT) = T ₁ + 0.7T ₂ , the step T ₂ of the lattice mark MG 'is 0.5 μm, and is 0.7 μm to 1.5 μm.

This range coincides well with the non-coherent region of the illumination beam ± LF multi-wavelength with three wavelengths λ ₁ , λ ₂ , and λ ₃ , as is apparent from the simulation results of FIGS. 3 and 8.

Thus, in this embodiment, even if the lattice mark is made low in level and almost eliminates the structural asymmetry of the lattice mark, the influence (resistance of the amplitude reflectance) due to the resist layer remaining as a subsequent problem can be reduced.

FIG. 12 shows the configuration of the position detection device according to the second embodiment, whereby the relative position offset amount in the pitch direction (X direction) between two diffraction grating marks RG and MG is measured in a homodyne manner. To illustrate. The beams LB ₁ , LB ₂ , LB ₃ as illumination beams are emitted from the laser light sources LS ₁ , LS ₂ , LS ₃ shown in FIG. 6 at different wavelengths λ ₁ , λ ₂ , λ ₃ and coaxial as parallel beams. Are combined and then irradiated perpendicular to the grating mark RG through the mirror MR ₁ .

From the grating mark RG, a plurality of diffracted light is generated by irradiation of the beams LB ₁ , LB ₂ , LB ₃ (parallel beam), but the grating mark RG is a one-dimensional grating having a transmission duty of 1: 1, and the pitch If the direction is a left-right direction in the ground of FIG. 12, each of these diffraction beams (diffraction beams) is bent at a predetermined diffraction angle in the ground of FIG.

Claim 12 is also in these diffracted beam as a wavelength λ _1, the beam LB _1, is generated a first order diffracted beam + D _11, -D _11, wavelength of the 1st-order diffraction beam generated from the beam LB ₂ of λ ₂ + D _12, in the - D _12, illustrates the wavelength λ ₃ of the beam of first-order diffracted beam produced from the D + LB _{3 13,} -D _13, D _0, and the zero-th order beam. Of course, further higher diffracted light also occurs for each of the beams LB ₁ , LB ₂ , LB ₃ at each wavelength, but so only the first diffraction beam is shown for simplicity of explanation.

In addition, each diffraction beam enters an imaging optical system (projection optical system for projection exposure, alignment optical system for mark detection, etc.) divided into a front lens system G ₁ and a rear lens system G ₂ . The grating mark RG is disposed at the position of the front focal length fla of the front lens group, and the phase of the rear focal length f1b of the front lens system G ₁ and the position of the front focal length f2b of the rear lens system G ₂ are substantially coincident with the Fourier transform plane (the pupil plane). ) When EP is formed, each primary diffraction beam crosses (images) at the position of the rear focal length f2a of the rear lens group G ₂ . However, in the imaging optical system composed of the lens systems G1 and G2, it is assumed that chromatic aberration is corrected for three wavelengths λ ₁ , λ ₂ , and λ ₃ . As shown in FIG. 12, a small mirror MR ₂ is fixed at the center of the Fourier transform surface EP, and the zero-order beam D ₀ from the grating MG is shielded by this mirror MR ₂ , and enters the rear lens group G ₂ . It is prevented. When also be irradiated from the respective first-order diffracted beam grating mark RG, the beam LB _1, LB _2, but is a similarly parallel beam LB _3, front group lens system by the action of G ₁ is the beam west at the location of the deflection plane Fourier EP converged do.

Thus, the pitch Prg of the grating mark RG, wavelength λ beam (angle of the zero-order beam D ₀₎ 1-order diffracted beams diffracted angle of ± D ₁₁ generated by the LB ₁ of ₁ θ _1, the beam LB ₂ having a wavelength λ ₂ the first-order diffraction beam of ± D ₁₂ generated by diffraction angle θ _2, the wavelength λ ₃ of the beam LB ₃ 1-order diffracted beams generated by the diffraction angle θ of ± ₁₃ D ₃ is represented by the formula below, respectively.

sin θ ₁ = λ ₁ / Prg (1)

Sin θ ₂ = λ ₂ / Prg (2)

sin θ ₃ = λ ₃ / Prg (3)

Thus, when λ ₁ <λ ₂ <λ ₃ , θ ₁ <θ ₂ <θ ₃ , and in the Fourier transform plane EP shown in FIG. 12, the first-order diffraction beam ± D ₁₁ is the first-order diffraction beam ±. D is the inside of the passage ₁₂ (the zero-order beam D ₀ side), and passed through the first-order diffracted beams D ₁₂ ± 1-order diffracted beams inside (first-order diffracted beams D ± ₁₁ side) of ± ₁₃ D of the side.

In addition, each primary diffraction beam is superimposed on the reflective lattice mark MG for measurement formed in a concave-convex shape on the wafer W through the rear lens group G ₂ , respectively, and overlaps with each other. At this time, the pitch direction of the grating mark MG also coincides with the X direction, and on the grating mark MG, a one-dimensional interference fringe of the wavelength λ ₁ (the pitch direction is the X direction) is generated by two-beam interference of the first-order diffraction beam ± D ₁₁ , 1st-order diffracted beams ± by a two-beam interference of D ₁₂ wavelength one-dimensional interference fringes of λ _2, and (pitch direction is X direction) is generated, the first-order diffracted beam of ± D ₁₃ 2-beam by the interference wavelength λ ₃ of the first A dimensional interference fringe (pitch direction is X direction) is created.

At this time, since the light of wavelength λ _1, the light of wavelength λ _{2, and} the light of wavelength λ ₃ are different wavelengths, interference does not occur between each of the first-order diffraction beams ± D ₁₁ , ± D ₁₂ , and ± D ₁₃ . . And importantly, an interference fringe of wavelength λ ₁ produced by the first diffraction beam ± D ₁₁ , an interference fringe of wavelength λ ₂ produced by the first diffraction beam ± D ₁₂ , and an interference fringe produced by the first diffraction beam ± D ₁₃ Each of the interference fringes having the wavelength λ ₃ is made to have the same pitch of the respective intensity distributions, and appears as a single interference fringe.

The pitch Pif of the intensity distribution of the interference fringe is determined by the pitch Prg of the grating mark RG and the magnification M of the imaging optical systems G ₁ and G ₂ , and is represented by Pif = M · Prg / 2. For example, if the pitch Prg is 4 µm and the magnification M is 1/4 (the pattern size of the grid RG is reduced to 1/4 at the lattice mark MG side), the pitch Pif of the interference fringe is 0.5 µm. Therefore, if the pitch Prg of the grating MG for measurement is determined by the relationship of Prg = 2Pif, that is, the relationship of Pmg = M · Prg, the first-order diffraction beam ± D _1n (n = 1, 2, 3) from the grating mark Re-diffracted light is generated for each wavelength using each of the incident light beams.

For example, one re-diffracted light generated from the grating mark MG using the first-order diffraction beam + D ₁₁ as the incident light beam is -first-order diffraction light (wavelength λ ₁ ) running perpendicularly from the grating mark MG, and the first order. One re-diffracted light generated from the grating mark MG using the diffraction beam -D ₁₁ as the incident light beam is + 1st order diffracted light (wavelength λ ₁ ) running perpendicularly from the grating mark MG. The ± first order diffracted light of the wavelength lambda ₁ traveling vertically has interference intensity according to the phase state of each other and reaches the mirror MR ₂ which becomes the interference beam BM.

Similarly, re-diffracted light is generated from the grating mark MG using the primary diffraction beams ± D ₁₂ and ± D ₁₃ as incident light beams, but is generated from the grating mark MG by irradiation of the primary diffraction beam + D12 (+ D13). The first-order diffracted light of wavelength λ ₂ (wavelength λ ₃ ) is perpendicular to the grating mark MG, and the wavelength λ ₂ (generated from the grating mark MG by irradiation of the first-order diffraction beam -D ₁₂ (-D ₁₃ ). The + 1st order diffracted light of wavelength λ ₃ ) runs perpendicular to the grating MG. These ± first-order diffraction light beams of wavelengths λ ₂ and λ ₃ that run vertically have interference intensities according to mutual phase states, and reach mirror MR ₂ as interference beam BM.

That is, the interference beam BM has a wavelength λ ₁ of the interference beam B _{m1, m2} of the wavelength λ ₂ interference beam B and the wavelength λ ₃ of the interference beam B _m3 is included on the same axis.

This interference beam BM is reflected by the mirror MR ₂ and reaches the photoelectric element DT through the lens system G ₃ constituting the photoelectric detection system. Thereby, the interference beams B _m1 , B _m2 , B _m3 for each wavelength in the interference beam BM are simultaneously received by the photoelectric element DT, and the photoelectric element DT ₃ receives the photoelectric signal I _m1 at the level corresponding to the intensity of the interference beam BM. and outputs to the circuit unit CU _3.

The circuit unit CU ₃ is configured in the same manner as the signal processing circuit shown in FIG. 9 above, but is a homodyne system, and thus the stage for placing the second lattice mark MG, that is, the wafer W, with respect to the interference fringe in the mark position detection. The WST is slightly moved in the lattice pitch direction, and at that time, the level change (waveform) of the signal I _m1 output from the photoelectric element DT is changed so as to be measured in correspondence with the movement position of the stage. The side of the incident light beam ± D _1n may be moved without changing the crossing angle of the paired incident light beams ± D _1n so that the side of the interference fringe moves in the pitch direction.

Therefore, in the present embodiment, the configuration of the signal processing circuit is an A / D converter instead of the sampling pulse C _ps in the circuit block shown in FIG. 9 for the pulse signal for position measurement output from the interferometer 44 shown in FIG. Supplied to the circuit 50 and the waveform memory circuit 54, so that the signal I _m1 of the photoelectric element DT is waveform-sampled by a pulse signal of the interferometer 44 (for example, 1 pulse for a movement of 0.02 mu m of the stage WST). We can realize by changing. However, since this embodiment is a homodyne system, it is not necessary to treat the reference signal I _ms like the processing circuit of FIG.

The circuit unit CU ₃ in FIG. 12 includes the same configuration as each of the A / D converter circuit 50, the waveform memory circuit 54, and the calculation circuit 56 in FIG. The calculation method of the position offset amount ΔX is different from the case of the heterodyne method. That is, in the homodyne system, the sinusoidal sampling waveform stored in the waveform memory circuit 54 becomes a function of position rather than a function of time, and the waveform itself shows the relative positional relationship with the lattice mark MG.

In addition, since the homodyne method is employed in the present embodiment, the intensity of the interference beam BM changes in accordance with the change in the relative position of the grating marks RG and MG in the X direction. For example, if the grating marks RG and MG are stopped in a certain state, Each level of I _m1 takes a certain value. Therefore, the interference fringe on the grid mark MG generated by the grid mark RG and the grid mark MG are relatively scanned in a X direction in a certain amount (distance equal to or more than the pitch Pif minutes of the interference fringe), and the level change of the sinusoidal shape of the signal I _m1 generated during that time. by sampling at the peak value and the beam tomgap the circuit unit CU ₃ in, and to calculate a mark position offset amount △ X after a certain amplitude value from the difference value.

Therefore, the change of the signal I _m1 according to the change in the positional relationship between the interference fringe and the grating mark MG will be described with reference to FIGS. 13A to 13D. Since the interference fringes of the pitches Pif are the two-beam interferences in FIGS. 13A, 13B, and 13C, they have a good sinusoidal intensity distribution and are set to Pmg = 2Pif for the pitch Prg of the grid mark MG. When the interference fringe moves leftward with respect to the grating MG in the order of FIGS. 13A, 13B, and 13C, the level of the signal I _m1 changes to a sinusoidal shape as in the 13D diagram. At the position where each peak of the interference fringe overlaps each edge of the lattice mark MG as in the 13B diagram, the signal I _m1 becomes the bottom level as point B. Thus, the level of the point A in FIG. 13D shows the case of the positional relationship of FIG. 13A, and the level of the point C shows the case of the positional relationship of FIG. 13C.

In this manner, the signal I _m1 changes level periodically whenever the interference fringe and the grating mark MG move by Pmg / 2 in the X direction. For this reason, it is impossible to determine the peak level and bottom level of the detected signal I _m1 unless the interference fringe and the lattice mark MG are moved preliminarily. For this reason, the level change of the signal I _m1 while the interference fringe and the grating MG are relatively moved, for example, over a distance of about 5 to 10 times the pitch of the grating MG, to the waveform memory circuit 53 in the circuit unit CU ₃ . Remember

In addition, the circuit unit CU ₃ processes the signal waveform stored in the waveform memory circuit 54 by the calculation circuit 56, and based on the amplitude value of the signal I _m1 and a preset function or conversion equation F (I _m1 ), The position offset amount ΔX in the X direction of the interference fringe and the grating mark MG is calculated. The position offset amount ΔX is determined as a value within a range of ± Pmg / 4 from the peak point or bottom point of the signal I _m1 in Fig. 13D as a reference (origin), for example.

The function (or expression) F (I _m1 ) uses a sine function or a sine function because the signal I _m1 has a sinusoidal shape. As an example, if the peak level of the signal I _m1 described above is E _p1 , the bottom level is E _b1 , and the level of the signal I _m1 is e ₁ at the detection position,

(Ep1 + Eb1) / 2 + [(Ep1-Eb1)] sinΨ1] / 2 = e1

The radian Ψ1 satisfying the following formula is obtained, and substituted into the following exchange equation using the value of the pitch Pmg, the offset amount λX from the reference point can be obtained.

ΔX = Pmg Ψ / 4π (4)

The offset amount [lambda] X calculated in this way is the position offset amount with respect to the grating RG of the grating MG finally calculated | required.

As described above, in the present embodiment, when the lattice marks RG and MG are irradiated with the multi-wavelength illumination beam including the beams LB1, LB2, and LB2 having three different wavelength components at predetermined light intensity ratios, the angles of the three beams are determined. Since the relationship between the center wavelengths λ ₁ , λ ₂ and λ ₃ is set to 1 / λ ₁ -1 / λ ₂ = 1 / λ ₂ -1 / λ ₃ (± 10%), the interference beam BM In addition to multiple wavelengths, more reliable position detection results are obtained. In the optical arrangement shown in FIG. 12, when the grating RG is a grating mark on a mask, the grating MG is a mark on a wafer, and the imaging systems G ₁ and G ₂ are projection lenses on a wafer of a mask pattern, Alignment apparatus can be realized.

FIG. 14 shows a schematic configuration according to the third embodiment, and the same reference numerals are given to those having the same function as the members, beams, etc. in FIG. In the third embodiment, the three beams LB ₁ , LB ₂ , LB ₃ for illumination are incident on the small mirror MR2 disposed in the center of the Fourier transform plane EP of the imaging optical system G ₁ , G ₂ via the lens system G4. The beams LB ₁ , LB ₂ , LB ₃ bent downward in the small mirror MR2 are irradiated perpendicularly to the grating mark MG with the parallel beam through the rear lens system G2.

And the diffraction wavelength λ _1, first-order diffracted beams ± D _11, the wavelength λ ₂ of the first-order diffracted beams ± D _12, the wavelength of the 1st-order diffraction beam ± D ₁₃ of λ ₃ lens system G _1, G ₂ in the grating mark MG Cross (image) on the lattice RG. Since the grating RG is a transmission type, the ± first-order diffracted light of the re-diffracted light generated from the grating RG by the irradiation of the first-order diffraction beam ± D ₁₁ forms an image perpendicular to the grating RG which becomes the interference beam B _m1 (including BM). Proceeding in the opposite direction to the optical system, the spatial filter 28 is reached through the mirror MR ₃ and the lens system G ₅ , so that unnecessary diffraction components are removed and received by the photoelectric element DT.

Similarly, the first-order diffraction light of the re-diffracted light generated from the grating RG by the irradiation of the first-order diffraction beams ± D ₁₂ and ± D ₁₃ at wavelengths λ ₂ and λ ₃ includes the interference beams B _m2 , B _m3 (BM ) And proceed vertically from the grating RG, and are received by the photoelectric element DT through the mirror MR ₃ , the lens system G ₅ , and the spatial filter 28. These interfering beams B _m1 , B _m2 , B _m3 occur coaxially as one interfering beam BM, but since the light sources that make each beam are individual, it is difficult for each of the interfering beams B _m1 , B _m2 , B _m3 to interfere. no.

The present embodiment has a configuration in which the relationship between the incident light and the light reception of the beam is opposite to that in FIG. 12, but this configuration forms the grating mark MG on the semiconductor wafer, the grating mark RG as the reticle (mask), and the lens system G ₁ ,. one of G ₂ to a reduction projection lens of the projection exposure of the reticle pattern (F) can be applied to Unexamined Patent Publication No. 3-3224 of the device.

However, in the apparatus disclosed in the publication F, a small lens for refraction of the primary diffraction beam by only a small amount is provided in the pupil surface EP of the projection lens to correct chromatic aberration generated in the projection lens, but the embodiment of FIG. When applied, a small lens (e.g., an optimal correction) for each of three sets of first-order diffraction beams ± D ₁₁ , ± D ₁₂ , ± D ₁₃ (three pairs of two-beam illumination beams) slightly different in wavelength It is necessary to provide a base material of a printing system with large color dispersion). In the third embodiment, the illumination beams LB ₁ , LB ₂ , and LB ₃ are configured to be incident directly on, for example, the lattice mark MG on the wafer, and are more than the intensity of the first-order diffraction beam (interfering beam BM) generated from the lattice mark MG. You can usually increase it.

By the way, when fixing the reference mark plate FG which has the well-known reflectance chromium surface on the wafer stage WST, the mark plate FG can be used for the measurement of various base line quantities, and the measurement of a focus state. The base line amount basically means a measured value for determining the relative positional relationship between the projection point of the center of the mask (reticle) mounted on the projection exposure apparatus and the detection center point of various alignment systems.

FIG. 15 is a fourth embodiment of the present invention, which shows a schematic arrangement of an alignment system of a projection exposure apparatus that requires measurement of a baseline amount, and a reticle R is adsorbed onto a reticle stage RST and dichroic from an exposure illumination system ILX. Irradiated uniformly by ultraviolet rays (i-ray or excimer laser beam) emitted through the mirror DCM. The pattern image of the reticle is configured to be projected and exposed to a predetermined shot area on the wafer W through an equal magnification or reduced projection optical system PL.

In Fig. 15, the mark group detectable by the alignment system TTRA of the through reticle (TTR) method, the mark group detectable by the reticle alignment system RA, and the through lens on the surface of the reference mark plate FG on the wafer stage WST. A group of marks detectable by the (TTL) alignment system TTLA and a group of detectable marks are formed by the alignment system OFA of the off-access system fixed to the ridge of the projection optical system PL.

These mark groups are some commonly used. Each of the alignment systems RA, TTRA, TTLA, and OFA is provided with direct or indirect detection center points R _f1 , R _f2 , R _f3 , and R _f4 serving as marks for mark detection.

In the case where the same position detecting device as the sixth degree described above is applied to each alignment system, the detection center points R _f1 , R _f2 , R _f3 and R _f4 are defined by the reference grid SG. In the reticle alignment system RA, however, the reticle alignment mark (lattice mark) RM around the reticle R and the corresponding lattice mark on the reference mark plate FG are irradiated with illumination light having the same wavelength as the illumination light for projection exposure of the pattern PR. In the case where the reticle stage RST is moved so that the mark has a predetermined positional relationship, the detection center point Rf1 is not required.

The same applies to the alignment system TTRA, for die-by-die (D / D) alignment formed at the periphery of the corresponding mark on the reference mark plate FG or the mark on the wafer W and the pattern PR of the reticle R. In the case of imaging the mark as an image and detecting the position offset on both marks, it is not necessary to particularly configure the detection center point Rf2.

Thus, the base line amount is X between the projection point (substantially coincides with the optical axis AX) of the center CCr of the reticle R and the projection point toward the wafer side of each detection center point R _f1 , R _f2 , R _f3 , R _f4 . This is not different from the positional relationship in the Y direction. The positional relationship detects the position offset amount between the corresponding mark group of the reference mark plate FG and the projection point of each of the detection center points R _f1 to R _{f4 by} the alignment systems RA, TTRA, TTLA, and OFA itself, and at the same time, the wafer stage It can obtain | require by detecting the coordinate position of WST by the laser interferometer 44 (refer FIG. 6).

However, when the multi-wavelength incident light beam ± LF is used, the interference fringe generated on the wafer W for each incident light beam of each wavelength component is slightly different (for example, about 0.05 μm) in its pitch or pitch direction. There is. This slight offset can ideally be zeroed by the adjustment, but even with such an effort, it is not very realistic considering the occurrence of drift over time. Therefore, a calibrating function may be provided in which the relative position (phase) offset, i.e., the color offset of the interference fringe, can be sometimes realized for each wavelength component of the multi-wavelength interference fringe. This function will be described later in detail.

The pupil surface EP in the projection optical system PL shown in FIG. 15 is equivalent to the Fourier transform surface EP shown in FIG. 12 described above. The objective optical axes of the alignment systems RA, TTRA, and TTLA, which detect objects on the wafer stage WST (mark of the wafer W or the mark of the reference mark plate FG) through the projection optical system PL, are all provided on the wafer stage WST side. It is set to be parallel to.

Also, when not only the wafer side of the projection optical system PL but also the reticle side is loaded with a telecentric system (in case of FIG. 15), the optical axis of the objective lens of each alignment system is also determined by the reticle side of the projection optical system PL. It is parallel to the optical axis AX. The optical axis of the objective lens passes through the center of the pupil surface EP of the projection optical system PL (the portion through which the optical axis AX passes). The effective diameter of the pupil surface EP corresponds to the numerical aperture NA that determines the resolution (minimum resolution line width) of the projection lens PL, and a projection lens with a NA of about 0.5 to 0.7 is currently developed.

FIG. 16 shows an example of the main part of the alignment system TTLA among the alignment systems shown in FIG. 15, and shows a pair of multicolored incident light for detecting the grating mark MG or the reference mark plate FG on the wafer. Beam ± LF (corresponds to beam ± LF and beam-LF in FIG. 16) is corrected optics CG, polarized beam splitter (PBS; function equivalent to half mirror 20 in FIG. 4), quarter wave plate QW, objective Incident on the projection lens PL via the lens OBJ (corresponding to the objective lens 22 in FIG. 6) and the two mirrors MR.

At this time, the surface FC of the wafer W and the conjugate plane FC are formed between the two mirror MRs, and a pair of beams LF intersect in the plane FC. The beam LF is relayed by the projection lens PL, and intersects on the wafer to irradiate the grating beam MG.

In addition, in this embodiment, the beam ± LF incident on the polarizing beam splitter PBS is a direct ray, and the incident light beam efficiently reflected on the polarizing beam splitter PBS rotates in one direction when passing through the quarter wave plate QW. And the lattice mark MG on the wafer passes through the objective lens OBJ and the projection lens PL.

The interference beam BM generated perpendicularly from the grating mark MG passes through substantially the center of the pupil surface EP of the projection lens PL, and passes through two mirrors MR, an objective lens OBJ, and a quarter wave plate QW, and the polarization beam splitter PBS. Is reached. At this time, since the interference beam BM is linearly polarized orthogonal to the polarization direction of the incident light beam, the polarization beam splitter PBS efficiently passes through and reaches the photoelectric element 36A.

In such an alignment system (TTLA), the incident light beam ± LF includes a plurality of wavelength components (about 30 to 40 nm apart from each other), and is affected by the chromatic aberration (axial and magnification) of the projection lens PL, or by the objective lens OBJ. Due to chromatic aberration, the intersecting area of the beam LF irradiated onto the wafer may slightly shift in the Z direction or the XY direction for each wavelength component. Thus, as shown in FIG. 16, a correction optical system CG is provided to correct an error caused by chromatic aberration in the optical path of the incident light beam ± LF. The correction optical system CG is composed of a convex lens, a concave lens, a combination lens thereof, a parallel plate lens, or the like, and the adjustment optical systems 14, 16 and 18 shown in FIG. 6 may be used.

In the case of the alignment system TTRA in Fig. 15, the D / D alignment mark DDM on the reticle R is used as a diffraction grating, and the lattice mark MG on the wafer W corresponding to the mark DDM and the relative position offset are determined by the fourth degree. When detecting by the same heterodyne system, as described in (G) JP-A-302504, a transparent flat plate correction plate PGP is provided on the pupil surface EP of the projection lens PL, and the incident light beam ± LF is placed on the correction plate PGP. (B) Transmissive phase gratings (etched with uneven lines at predetermined pitches on the surface of the compensator PGP) can be formed only at the position where the interference beam BM passes, and the influence of chromatic aberration on the axial and chromatic aberration on the magnification can be reduced.

17A and 17B show the configuration of the projection exposure apparatus according to the fifth embodiment in which such a correction plate PGP is placed in a part of the alignment system TTRA, and FIG. 17A shows a pitch having a pitch in the X direction (measurement direction). The optical path between the incident light beam ± LF and the interference beam BM in the case of detecting the grating mark MG is seen in the XZ plane, and FIG. 17B is the optical path in FIG. 17A as viewed in the YZ plane intersected with it.

From the objective lens OBJ (corresponding to the objective lens 22 in FIG. 6) of the alignment system TTRA, the incident light beam ± LF is emitted slightly eccentrically from the optical axis AXa, reflected by the mirror MR, and the peripheral window RW of the pattern region of the reticle R Enters the projection lens PL through. The pair of incident light beams ± LF is multi-wavelength and transmits through the window RW in a symmetrical inclination as in the XZ plane in the XZ plane, and in the YZ plane in relation to the optical axis AXa of the objective lens OBJ in the YZ plane. It is inclined to penetrate the window RW.

The pair of incident light beams ± LF passes through two phase-type diffraction gratings (hereinafter referred to as phase gratings) PG1 and PG2 respectively on the correction plate PGP disposed on the pupil plane EP of the projection lens PL. . At this time, by the action of the phase gratings PG1 and PG2, each of the incident light beams LF is emitted from the projection lens PL by changing the inclination by a predetermined amount in a predetermined direction like a solid line from a broken line in the drawing. In the XZ plane, the incident light beam ± LF irradiates the wafer W and the grating mark MG at symmetrical incidence angles as in the 17A diagram, and slightly in the Y direction with respect to the grating mark MG in the Y-2 plane as the 17B diagram. Inclined obliquely.

As a result, the interference beam BM which is slightly inclined in the Y direction from the grating mark MG again enters the projection lens PL, and passes through positions different from the phase gratings PG ₁ and PG ₂ on the pupil surface EP. At that position, the optical path of the interference beam BM is corrected to pass through the projection lens PL to the window RW of the reticle R so that the interference beam BM becomes a solid line from the broken line in FIG. 17B.

The interference beam BM passing through the window RW is directed to the same light receiving system as the fourth degree via the mirror R and the objective lens OBJ. At this time, the interference beam BM passes through the window RW of the reticle R in a state inclined slightly in the non-measurement direction with respect to the optical axis AXa of the objective lens OBJ.

In the case of using such a correction plate PGP, if the incident light beam ± LF is multi-wavelength, it is slightly shifted in the X direction on the correction plate PGP for each wavelength component of the incident light beam ± LF. For this reason, the phase gratings PG1 and PG2 are also formed symmetrically and large in the X direction. In addition, the use of such a correction plate PGP is naturally possible for the alignment system TTLA shown in FIG. For example, a projection lens (reflective element and refractive lens) may be combined using quartz or fluorite stone as a material for the refraction lens and using ultraviolet light (excimer laser light source) having a wavelength of 180 to 300 mm as exposure light. In the case of an exposure apparatus using a), chromatic aberration with respect to the wavelength of a beam from a He-Ne laser or a semiconductor laser becomes very large, and the wafer conjugate plane FC shown in FIG. 16 is spaced several tens of centimeters or more from the projection lens. . Thus, using the correction plate PGP, the incident light beam ± LF is to correct so that the intersecting wafer conjugate plane FC is close to the projection lens.

As described above, the correction plate (PGP) is a phase grating for light incident on the PG _1, PG ₂ is a multi-wavelength beam Chemistry + LF, -LF phase grating passes through, but then for the entire wavelength components used for PG _1, PG ₂ It is difficult to optimize the lattice structure. For this reason, the phase grating structures of the phase gratings PG ₁ and PG ₂ are set to be optimized to certain specific wavelength components, and in the incident light path of the incident light beam ± LF (generally the light source side of the objective lens OBJ), each wavelength component It is advisable to provide an adjusting optical member so that the incident light beams are compensated in advance only in the direction deviation or position deviation caused by the difference in diffraction effect accommodated by the phase gratings PG ₁ and PG ₂ .

That is, the interference fringes formed on the lattice mark MG of the wafer W (or the reference plate FG) due to the interference of the pair of incident light beams ± LF do not generate extreme position offsets or pitch offsets for each wavelength component. It is very important to adjust the position of the adjustment optical system 14, 16, 18 or the correction lens CG in FIG.

Next, a sixth embodiment of the present invention will be described with reference to FIG. In this embodiment, the polarization direction between a pair of transfer beams + LF and -LF that irradiates the lattice mark MG for measurement (alignment) on the wafer W (or reference plate FG) via the objective lens 22 is determined. It is a complementary relationship. That is, in the case of linearly polarized light, the polarization directions of the incident light beams + LF and -LF are crossed, and in the case of circularly polarized light, the incident light beams + LF and -LF are set to reverse rotation polarizations. For this reason, the two incident light beams + LF do not interfere with each other as they are, and the first order diffracted light BH for each wavelength λ ₁ , λ ₂ , λ ₃ generated vertically from the grating mark MG does not interfere with each other. Therefore, when photoelectric detection of the +/- 1st order diffracted light BM is carried out through the objective lens 22 and the small mirror MR2, the polarizing beam splitter PBS as an analyzer (analyzer) is used. With this arrangement, the polarization beam splitter PBS by ± 1-order light (BM) passing through is a first interference beam BP ₁ to interfere with each other, is to interfere the ± 1-order light (BM) reflected by the polarization beam splitter PBS first 2 becomes the interference beam BP2.

Although the interference beams BP ₁ and BP ₂ are complementary to each other, if the interference beams are heterodyned, the interference beams BP ₁ and BP ₂ are intensity modulated in a sinusoidal shape according to the bit frequency. In addition, the phases of the intensity modulation of the interference beams BP ₁ and BP ₂ differ by 180 degrees.

In addition, the half wave plate HW shown in FIG. 18 has a case where the linearly polarized light direction in which the incident light beam ± LF and the first-order diffracted light BM are orthogonal to each other is different from the polarization separation direction of the polarized light beam splitter PBS (rotating). Is provided for the purpose of correcting the linearly polarized light direction between the ± first order diffracted light BM. For this reason, when the linear polarization directions orthogonal to each other between the ± first-order diffraction light BM coincide with the polarization separation direction of the polarization beam splitter PBS from the beginning, or when the incident light beams + LF and -LF are circularly polarized in reverse rotation, It is not necessary to use the half wave plate HW. So the light reception in the present embodiment, the interference beam BP photoelectric element through the mirror 32 and received by the light-36A _1, an interference beam BP ₂ through the mirror 32 photoelectric elements (36A _2). In addition, each of the output signals I _a1 and I _a2 of the photoelectric elements 36A ₁ and 36A ₂ is subtracted by the differential amplifier to become the photoelectric signal I _m1 .

The use of the differential amplifier in this way is because the output signal of the photoelectric element 36A ₁ and the output signal of the photoelectric element 36A ₂ are out of phase with each other (180 ° difference), and thus the in-phase noise included in both outputs. This is because the component (common-mode noise) is canceled by subtraction, so that the substantial S / N ratio of the signal I _m1 is improved.

In addition, the objective lens 22 shown in FIG. 18 or FIG. 6 of the present embodiment, or the objective lens OBJ shown in FIG. 16, has at least an axial shape among various chromatic aberrations generated in the wavelength ranges λ ₁ to λ _{3 used} . It is preferable that the chromatic aberration be compensated to some extent. For example, when the band of wavelengths λ ₁ to λ ₃ to be used is 100 nm or less, such axial chromatic aberration may be selected by selecting a base material of a plurality of lens elements constituting the objective lens 22 or by using a lens element having a different refractive index or dispersion ratio. Combination becomes possible to some extent. Of course, such chromatic aberration does not need to be completely corrected in the objective lenses 22 and OBJ, and is corrected by the adjustment optical system 14, 16, 18 shown in FIG. 6 or the correction optical system CG shown in FIG. It is also possible to correct.

Next, a seventh embodiment of the present invention will be described with reference to FIG. 19, but here the multi-wavelength interference beam (BM) generated from the lattice mark MG on the wafer W or the reference plate FG is ± 1st order. The position detection apparatus which adds the function which can automatically continue the offset amount by the color of the interference fringe for each wavelength produced | generated by 2-beam interference by photoelectric detection for each wavelength component) is illustrated.

The configuration of this embodiment shown in FIG. 19 is a part of the configuration of FIG. 6, specifically, the photoelectric detection system of the interference beam BM from the grating mark MG is changed, and therefore, the member having the same function as the member in FIG. Give the same sign. Incident light system 100 in FIG. 19 includes light sources LS ₁ , LS ₂ , LS ₃ , mirror MR, dichroic mirror DCM4, DCM5, radial grating RRG as a frequency shifter, lens 10, It consists of the spatial filter 12, the adjustment optical system 14, 16, 18, etc., and emits a pair of incident light beam + LF, -LF.

Incident light beams ± LF including respective components of wavelengths λ ₁ , λ ₂ , and λ ₃ are partially reflected by the half mirror 20 to be incident to the objective lens 22, and a part is incident to the reference light receiving system 110. do. The reference light receiving system 110 is composed of the wavelength selective filter 24, the lens 26, the reference grating SG, and the spatial filter 38 of FIG. 6, and guides the reference light Bms to the photoelectric element 40.

Further, when the grating MG on the wafer W via the objective lens 22 is irradiated by the incident light beam ± LF, the interference beam BM of the ± first-order diffracted light is generated vertically, and zero in the direction of travel of each incident light beam. Various interference beams of secondary to secondary light are generated. This interference beam reaches the spatial filter 28 disposed on the Fourier transform plane of the objective lens 22 through the objective lens 22 and the half mirror 20, where only the interference beam BM of the ± first-order diffraction light is spaced. It is selectively transmitted through filter 28 and split into two in beam splitter 29.

The interference beam BM reflected by the beam splitter 29 is received by the photoelectric element 36A through the lens system 30 such as the sixth degree, and the interference beam BM transmitted through the beam splitter 29 is the spectrometer 34. Incident. The spectrometer 34 reflects only the interference beam B _m1 having a wavelength λ ₁ from the center of the interference beam BM received by the lens system 31, and the dichroic mirror DCM ₁ and the dike induced in the photoelectric device DT ₁ . Loic transmitted through the mirror (DCM ₁₎ interference beam B _m2 (wavelength λ ₂₎ and B _m3 (wavelength λ ₃₎ by reflecting only the interference of beam B _m2 was converted to the photoelectric device DT _2, and transmits only the interference beam B _m3 It is composed of a dichroic mirror (DCM ₂ ) leading to the photoelectric element (DT).

And from each photoelectric element DT _1, DT _2, DT _3, the photoelectric signal I _p1, I _p2, I _p3 is output which varies in a sine wave shape to the foot level, frequency 2 · △ f. These photoelectric signals I _pn (n = 1, 2, 3) are combined with the photoelectric elements I _ms from the photoelectric elements 40 which receive the reference light B _ms and the signal processing circuits 70A, 70B, 70C shown in FIG. , 70D). These signal processing circuits all include circuit elements such as the A / D converter circuit 50, the waveform memory circuit 54, and the operation circuit 56 shown in FIG. 9, and the phase offset for each photoelectric signal ( ΔΨ) is calculated. The information of the phase offset (ΔΨ) of each photoelectric signal calculated in this way is processed by a computer in the position offset calculating circuit 72, and the amount of minute position offset of the interference fringes of each wavelength component, that is, each Offset amounts are determined from each other with respect to the pitch direction of the interference fringe.

In the above configuration, each A / D converter circuit 50 and the waveform memory circuit 54 in the signal processing circuits 70A to 70D are clock pulses from a common sampling lock generation circuit 52 (see FIG. 6). In response to C _ps , the waveform of each photoelectric signal is digitally sampled at the same time. In addition, the calculation circuits 56 in the signal processing circuits 70A to 70D perform Fourier integration operations based on the sinusoidal data sinωt and the cosine wave data cosωt previously stored therein, and each photoelectric signal I _p1 , I _p2 , I _p3, P _ms of data sinωt (or cosωt) a phase difference △ one relative to the _{Ψ p1, △ Ψ p2, △} Ψ p3, △ Ψ ms of data sinωt (or cosωt) a phase difference △ Ψ _p1 relative to, △ Ψ _p2 , _ΔΨp3 , ΔΨ _ms are respectively calculated. The frequency ω of the sinusoidal data and the sine wave data is determined to be ω = 2π (2ΔF) in relation to the bit frequency 2ΔΔf.

Next, the position offset calculation circuit 72 performs position offset amounts ΔX ₁ and ΔX _{2 for} each wavelength of the grating mark MG based on the phase differences ΔΨ _P1 , ΔΨ _p2 , ΔΨ _p3 , and ΔΨ _ms , respectively. , ΔX ₃ is calculated by the following operations. However, k is k = ± Pmg / 4π.

ΔX ₁ = k (ΔΨ _ms -ΔΨ _p1 )

ΔX ₂ = k (ΔΨ _ms -ΔΨ _p2 )

ΔX ₃ = k (ΔΨ _ms -ΔΨ _p3 )

The position offset amounts calculated in this way are almost the same value when the grid mark MG on the reference plate FG shown in FIGS. 6 and 15 is detected, but the position offset amounts ΔX ₁ , ΔX ₂ , Δ If there is a deviation between X ₃ , it means that a mutual offset in the pitch direction occurs between the interference fringes for each wavelength. Therefore, before the operation of aligning the wafer W, the grating mark MG on the reference plate FG is detected to confirm the mutual offset between the interference fringes for each wavelength, and if the mutual offset is larger than the allowable range, the adjustment optical system 14, 16 of FIG. 18) or part 16 of the correction optical system CG in FIG. 16 to adjust the mutual offset within the allowable range.

According to the present embodiment described above, the interference fringes generated by the multi-wavelength two beams can be kept in a good state not always offset to each other, and the error in detecting the lattice marks of the wafer W can be kept small.

According to each of the above first to seventh embodiments, three or more illumination beams from a plurality of light sources with strong interference are originally synthesized to produce multi-wavelength illumination light, and the periodicity of a grid mark or the like is irradiated by the illumination light. When detecting a pattern, the relationship between the wavelengths of three or more illumination beams to be multi-wavelength is approximately ± 10% for any difference in the wave number (1 / λn) between two adjacent illumination beams on the wavelength axis. Since it is made to have the allowable range of, sufficient non-interference can be achieved even if light with strong interference is used.

For this reason, compared with the past, position detection, such as a grating mark, is attained with high precision. The position detection apparatus disclosed in each embodiment of the present invention can be easily applied not only to the projection exposure apparatus but also to various alignment systems in the proximity type exposure apparatus.

Next, each Example regarding the photoelectric processing in a multi-wavelength alignment system with respect to 2nd invention of this application is demonstrated.

FIG. 21 shows the configuration of the position detecting apparatus according to the eighth embodiment of the present invention, and the basic configuration is similar to that of the twelfth diagram, but the pitch direction between the two diffraction gratings RG and MG (it is X direction). The case where the relative position offset of is measured by the homodyne method is illustrated. The beams LB ₁ , LB ₂ as illumination beams are respectively emitted from different laser light sources at different wavelengths λ ₁ , λ ₂ and are irradiated perpendicularly to the grating RG through the beam splitter BS, mirror MR ₁ , coaxially synthesized. The beam splitter BS is amplitude-divided a portion of the beams LB ₁ , LB ₂ (a few%) and is led to the photoelectric elements DT ₁ , DT ₂ through the dichroic mirror DCM ₁ . Dichroic mirror DCM ₁ to dichroic DT is sent to the photoelectric element ₁ and the beam LB ₁ having a wavelength λ ₁ at least 90% transmission, to the beam LB ₂ reflected with a wavelength λ ₂ greater than 90% and sends it to the photoelectric element DT _2. Each photoelectric element DT ₁ , DT ₂ outputs a signal I _r1 representing the intensity value of the received beam of wavelength λ ₁ and a signal I _r2 representing the intensity value of the beam of wavelength λ ₂ .

Although a plurality of diffraction beams are generated from the grating RG by irradiation of the beams LB ₁ and LB ₂ (parallel beams), the grating RG is a one-dimensional grating having a transmission duty of 1: 1, and the pitch direction is shown in FIG. In the left and right directions in the ground, each of these diffraction beams is bent at a predetermined diffraction angle in the ground in FIG.

21 in FIG. 1 from the generated beam LB1 having a wavelength λ ₁ as they diffraction order diffracted beam ratio + D _11, -D _11, the wavelength λ ₂ of the beam LB ₂ The 1st order diffracted beam + D _12, -D ₁₂ produced from And 0th order beam D ₀ . Of course, higher order diffracted light is also generated for each beam LB ₁ , LB ₂ of each wavelength, but only the first order diffraction beam is shown for simplicity of explanation.

However, each diffraction beam enters the imaging optical system divided into the front lens group G ₁ and the rear lens group G ₂ . Grating RG is placed at the position of the front focal length f1a of the front group lens G _1, front group lens G rear focal distance position and a rear group lens system G ₂ in the front focal plane position and the matching by applying a Fourier transform of the distance f2b EP formation of f1b ₁ Each primary diffraction beam intersects (images) at the position of the rear focal length f2a of the rear lens group G2. However, the lens system G _1, G ₂ are the two wavelengths λ _1, λ _2, it is assumed that the chromatic aberration correction for two wavelengths λ _1, λ _2.

As shown in FIG. 21, a small mirror MR ₂ is fixed to the center of the Fourier transform surface (the pupil surface) EP, the zero-order beam D ₀ from the grating RG is shielded by the mirror MR ₂ , and the rear lens system G _{2 is provided.} Incident to is prevented. When the first diffracted beam is emitted from the grating RG, the beam becomes a parallel beam similarly to the beams LB ₁ and LB ₂ , but converges as a beam waist at the position of the Fourier transform surface EP under the action of the global lens system G ₁ .

Thus, if the pitch of the grating RG as Prg, the wavelength λ beam LB 1-order diffracted beams ± diffraction (angle with respect to the zero-order beam D _0), each of D ₁₁ generated by the ₁ θ ₁ and wavelength λ ₂ beam LB ₂ The diffraction angle θ2 of the first diffraction beam ± D ₁₂ generated by the equation is expressed by the following equation, respectively.

sinθ ₁ = λ ₁ / Prg (5)

sinθ ₂ = λ ₂ / Prg (6)

Therefore, when λ ₁ <λ ₂ , θ ₁ <θ ₂ , and as shown in FIG. 21, in the Fourier transform plane EP, the first diffraction beam ± D11 is the inner side of the first diffraction beam ± D ₁₂ . (0th order beam D ₀ group).

In addition, each primary diffraction beam overlaps each other on a reflective grating MG for measurement, which is formed in an irregular shape on the object side through the rear lens group G2, respectively. At this time, the pitch direction of the grating MG also coincides with the X direction, and on the grating MG, one-dimensional interference fringes having a wavelength λ ₁ (the pitch direction is the X direction) are generated by two-beam interference of the first-order diffraction beam ± D ₁₁ . , the first-order diffracted beam by a two-beam interference ± D ₁₂ of wavelength λ ₂ of a one-dimensional interference fringes (the pitch direction is X direction) is generated. At this time, since the light of wavelength λ _{1 and} the light of wavelength λ ₂ are different wavelengths, no interference fringe occurs between the first-order diffraction beams ± D ₁₁ and ± D ₁₂ . And importantly, the pitch is exactly the same as the interference fringe of wavelength λ ₁ generated by the first diffraction beam ± D ₁₁ and the interference fringe of wavelength λ ₂ generated by the first diffraction beam ± D ₁₂ and furthermore a single interference fringe. Appear as

The pitch Pif of the intensity distribution of the interference fringe is determined by the pitch Prg of the grating RG and the magnification M of the imaging optical systems G1 and G2, and is expressed by Pif = M · Prg / 2. For example, if the pitch Prg is 4 µm and the magnification M is 1/4 (the pattern size of the grid RG is reduced to 1/4 at the lattice MG side), the pitch Pif of the interference fringe is 0.5 µm. Thus, if the relationship between the pitch Pmg = 2Pif of the grating MG for measurement is determined, that is, the relationship of Pmg = M.Prg, one grating generated from the grating MG having the first diffraction beam + D ₁₁ as the incident light beam from the grating MG. The diffracted light is -first-order diffraction light (wavelength λ1) running perpendicularly from the grating MG, and one re-diffracted light generated from the grating MG with the first-order diffraction beam -D ₁₁ as the incident light beam travels perpendicularly from the grating MG. Is + 1st order diffracted light (wavelength λ ₁ ). These first-order diffracted light of the wavelength lambda ₁ running vertically have interference intensity according to mutual phase states. It is the interference beam BM and reaches the mirror MR _2.

On the other hand, although re-diffracted light also arises from the grating MG using the primary diffraction beam ± D ₁₂ as the incident light beam, -first-order diffraction light generated from the grating MG by irradiation of the first diffraction beam + D ₁₂ (wavelength λ ₂ ) Travels perpendicular to the grating MG, and the + 1st order diffracted light (wavelength? ₂ ) generated from the grating MG by irradiation of the first diffraction beam -D ₁₂ also proceeds perpendicular to the grating MG. These first-order diffraction light beams of wavelength lambda ₂ running vertically also have interference intensities according to mutual phase states, and become an interference beam BM to reach the mirror MR ₂ . That is, the interference beam BM has interference beam B _m2 of the interference beam of wavelength λ ₁ and wavelength λ ₂ B _m1 is contained coaxially.

The interference beam BM is reflected by the mirror MR ₂ and reaches the photoelectric elements DT2 and DT4 through the lens system G ₃ and the dichroic mirror DCM2 constituting the photoelectric detection system. The dichroic mirror DCM ₂ divides the wavelengths λ ₁ and λ ₂ , and substantially the same as the dichroic mirror DCM1 is used. Therefore, the interference beam B _m1 of the wavelength λ _{1 in} the interference BM is received by the photoelectric element DT ₃ , and the interference beam Bm ₂ of the wavelength λ ₂ is received by the photoelectric element DT4.

The photoelectric element DT ₃ outputs the photoelectric signal I _m1 at the level corresponding to the intensity of the interference beam B _m1 to the circuit units CU ₁ and CU ₃ , and the photoelectric element DT ₄ is the photoelectric signal I _m2 at the level corresponding to the intensity of the interference beam B _m2 . Is output to the circuit units CU ₂ and CU ₄ . The circuit unit CU ₁ obtains the ratio C ₁ of the signal I _r1 from the photoelectric element DT ₁ to the amplitude value of the photoelectric signal I _m1 by the calculation of I _m1 / I _r1 , and the circuit unit CU ₂ is obtained from the photoelectric element DT ₂ . The ratio C ₂ between the signal I _r2 and the amplitude value of the photoelectric signal I _m2 is obtained by calculation of I _m2 / I _r2 . The data of these ratios C ₁ and C ₂ are output to the circuit unit CU ₅ which calculates the weighted average described later.

In addition, in the present embodiment, the homodyne method is adopted, and the intensity of the interference beams B _m1 and B _m2 changes according to the change in the relative position of the gratings RG and MG in the X direction, for example, when the signal is stopped in the states of gratings RG and MG. The levels of I _m1 and I _m2 each take some constant value. Therefore, the interference fringes on the grid MG generated by the grid RG and the grid MG are relatively scanned in the X direction by a predetermined amount (more than the pitch Pif of the interference fringes), and the level change of the sinusoidal shapes of the signals I _m1 and I _m2 generated therebetween. In the above, the peak value and the bottom value are sampled, and the value is used as the amplitude value for the calculation of the circuit units CU ₁ and CU ₂ , respectively.

Therefore, the level change of the signal I _m1 (also I _m2 ) according to the change in the positional relationship between the interference fringe and the grating MG will be described with reference to FIGS. 22A to 22D. 22A, 22B, and 22C are the same as 13A, 13B, and 13C, respectively, and since the interference fringe of the pitch Pif is two-beam interference, it has a good sinusoidal intensity distribution, and Pmg to the pitch Pmg of the grating MG. = 2Pif is set. When the interference fringe moves in the right direction with respect to the grating MG in the order of FIGS. 22A, 22B, and 22C, the level of the signal Im1 changes to a sinusoidal shape as in the case of 22D. At the position where each peak of the interference fringe overlaps each short edge of the grating MG as shown in the 22B diagram, the signal I _m1 becomes a normal level as point B. Thus, the level of the point A in the 22D shows the case of the positional relationship of FIG. 22A, and the level of the point C shows the case of the positional relationship of the 22C.

As such, the signal I _m1 periodically changes level whenever the interference fringe and the grating MG move by Pmg / 2 in the X direction. Therefore, the peak level or normal level of the detected signal I _m1 cannot be obtained unless the interference fringe and the grating MG are moved preliminarily. The same applies to the signal I _m2 . Signal Im2 are ± 1-order diffracted due to interference of light indicate the intensity of the beam B _m2, the 22D also as shown by the imaginary line in, even if the level and the larger difference between the signal I _m1, the phase of the signal I _m1 is It is not extremely misaligned (but it may be misaligned by a few% due to resist interference or asymmetry of the mark). Therefore, even if each level of I _m1 and I _m2 is sampled at an arbitrary positional relationship where the interference fringe and the grating MG stop, theoretically, the ratio C ₁ and C ₂ can be calculated by the circuit units CU ₁ and CU ₂ . Do. However, as apparent from Fig. 22D, the sampling of each level is advantageous in that the signals I _m1 and I _m2 become peaks in terms of various noise problems and detection accuracy.

As shown in FIG. 21, the circuit units CU ₃ and CU ₄ respectively interfere based on the amplitude values of the signals I _m1 and I _m2 and based on a preset function or a conversion equation F (I _m1 ) and F (I _m2 ). The position offset amounts ΔX ₁ , ΔX ₂ in the X direction of the fringe and the grating MG are calculated. Position offset amount △ X _1, △ X ₂ are obtained, for example the 22D is also a value in the range of the signals I _m1, to a peak point or beam tomjeom of I _m2 on the basis (origin) therefrom ± Pmg / 4 in .

The functions (or expressions) F (I _m1 ) and F (I _m2 ) use a sine function or a sine function because each of the signals I _m1 and I _m2 has a sinusoidal shape. As an example, when the peak level of the signal I _m1 described above is E _p1 , the bottom level is E _b1 , and the level of the signal I _{m1 at} the position to be detected is e ₁ (E _p1 + E _b1 ) / 2. By calculating the radian Ψ that satisfies + {(E _p1 -E _b1 ) sinΨ ₁ } / 2 = e ₁ and substituting it into the following exchange equation using the value of pitch Pmg, the offset amount ΔX from the reference point can be known. .

ΔX = Pmg Ψ / 4π (7)

The data of the offset amounts ΔX ₁ , ΔX ₂ thus calculated is sent to the circuit unit CU ₅ which performs the weighted average operation, and the following calculation is performed using the ratios C ₁ and C ₂ obtained as the weighting coefficients.

ΔX = (C ₁ · ΔX ₁ + C ₂ · ΔX ₂ ) / (C ₁ + C ₂ ) (8)

The offset amount ΔX obtained by this calculation is the position offset amount with respect to the grid RG of the grid MG finally obtained.

As is apparent from this calculation formula, the offset amount ΔX is determined by weighting more than the measurement result of the position offset amount using the interference beam of the wavelength component of the higher intensity in the interference beam BM. As described above, in the present embodiment, the interference beams BM to be received by irradiating each grating RG and MG using _two beams LB ₁ and LB ₂ having two different wavelength components are also photoelectrically detected for each wavelength, and the interference beams B _{m1 for} each wavelength. , B _m2 is individually weighted and averaged according to the amplitude of the received light for each wavelength, so that a more reliable position detection result can be obtained.

The algorithms of the signal processing systems (circuit units CU ₁ to CU ₅ ) shown in FIG. 21 above are common in other embodiments described below, and are individually changed and improved in realizing the functions of the circuit units. If so, explain it at that time. In the optical arrangement shown in FIG. 21, when the grating RG is a grating mark on a mask, the grating MG is a mark on a wafer, and the imaging systems G ₁ and G ₂ are projection lenses on a wafer of a mask pattern, the projection exposure apparatus Alignment device can be realized.

23 shows a schematic configuration according to the ninth embodiment, the basic configuration of which is the same as that of the fourteenth diagram, but the configuration of the light receiving portion is different from that of the fourteenth diagram. In Fig. 21, the same reference numerals are given to those having the same function as the members, beams, and the like. In the ninth embodiment, the two beams LB ₁ and LB ₂ for illumination are incident on the mirror MR ₂ disposed in the center of the pupil plane of the imaging optical system G ₁ and G ₂ through the lens system G ₄ , and down from this mirror MR ₂ . The bent beams LB ₁ and LB ₂ are irradiated perpendicularly to the grating MG with the light beam beam through the rear lens group G ₂ . The first diffraction beam D ₁₁ of wavelength λ ₁ diffracted at the grating MG and the first diffraction beam ± D ₁₂ of the wavelength λ ₂ are crossed (imaged) on the grating RG through the lens systems G ₁ and G ₂ . Since the grating RG is a transmission type, the ± first-order diffraction light of the re-diffracted light generated from the grating RG by the irradiation of the first-order diffraction beam ± D ₁₁ proceeds in the opposite direction to the imaging optical system perpendicular to the grating RG, and the mirror MR ₃ and the dike It becomes the interference beam B _m1 through the wing mirror DCM ₃ and is received by the optoelectronic device DT ₃ . The first-order re-diffracted light generated by irradiation of the first-order diffraction beam ± D _{12 also} becomes the interference beam B _m2 , passes through the same optical path as the interference beam B _m1, and is selected by the dichroic mirror DCM ₃ to reach the optoelectronic device DT ₄ . do. The rest of the configuration is the same as in the twenty-first degree.

The present embodiment is a configuration in which the relationship between the incident light and the light reception of the beam is inverse to that of FIG. 21, but this configuration forms the grating MG on the semiconductor wafer, the grating RG on the reticle (mask), and the lens system G ₁ ,. one of G ₂ to a reduction projection lens of the projection exposure of the reticle pattern (F) can be applied to Unexamined Patent Publication No. 3-3224 of the device. However, in the apparatus disclosed in the publication (F), a small lens for refraction of the primary diffraction beam by only a small amount is provided on the pupil surface EP of the projection lens, and the chromatic aberration generated in the projection lens is corrected, but the embodiment of FIG. 23 is applied. In this case, it is necessary to install a small lens (for example, a material of a high color dispersion printing system) so that an optimal correction can be made for each of two sets of first diffraction beams ± D ₁₁ and ± D _{12 that} are slightly different from each other. have.

As described above, in the ninth embodiment, the beams LB ₁ and LB ₂ for illumination are configured to directly enter the grating MG on the wafer, for example, and the respective intensities of the first diffraction beams ± D ₁₁ and ± D ₂₁ generated from the grating MG are adjusted. It is possible for each intensity of the primary diffraction beams ± D ₁₁ , ± D ₁₂ generated from the grating MG in FIG. 21 to be high through the diffraction beam (interfering beam BM) generated from the grating MG in FIG. 21.

Next, a tenth embodiment of the present invention will be described with reference to FIGS. 24 and 25, but the basic configuration is the same as that of the apparatus of FIG. 6 described above, and the configuration of the light receiving system is slightly different. So instead of homodyne, he uses heterodyne. 25. The method of claim 24 also, the three laser light sources LS _1, LS _2, LS ₃ is emitted, each with a different wavelength λ _1, λ _2, λ ₃ the laser beam LB _1, LB _2, LB ₃ a. As an example, the laser light source LS ₁ is set to a He-Ne laser light source of λ ₁ = 0.633 μm, the light source LS ₂ is a semiconductor laser light source of λ ₂ = 0.690 μm, and the light source LS ₃ is a semiconductor laser light source of λ ₃ = 0.760 μm. , The relationship between the wavelengths is to be selected as λ ₁ <λ ₂ <λ ₃ .

These three beams LB ₁ , LB ₂ , LB ₃ are synthesized into one coaxial beam LB ₀ via mirror MR, dichroic mirrors DCM ₄ , DCM ₅ , and are reflected at the mirror MR and enter the rotating radial grating plate RRG. . As shown in FIG. 7, the grid RRG rotates at high speed about the rotation axis C _{0 at} an equiangular velocity in one direction, and increases and decreases the frequency of the diffracted light of each order diffracted by the grid RRG by the amount corresponding to the angular velocity. Has

In this embodiment, ± 1st-order diffraction light from the radial grating plate RRG is used to implement the heterodyne system. In FIG. 24, only ± 1st-order diffraction light ± LF from the grating plate RRG is shown.

As shown in FIG. 21, from the grating RG of the grating plate RRG, the first diffraction beam ± D ₁₁ made of the beam LB1 of the wavelength λ ₁ , the first diffraction beam ± D ₁₂ made of the beam LB ₂ of the wavelength λ ₂ , and the wavelength A first diffraction beam, ± D ₁₃ , made of a beam LB ₃ of λ ₃ occurs. The diffraction angle θ of the primary diffraction beam for each wavelength is expressed as follows.

sinθ _n = λ _n / Prg

Where n represents the number of wavelengths and Prg represents the pitch of the grating RG of the rotating grating plate PRG.

On the other hand, if the primary diffraction beam has a constant frequency shift Δf depending on the wavelength, and the speed at which the lattice RG of the lattice plate RRG crosses the beam LB ₀ is V, it is expressed as Δf = V / Prg and the + 1st order The diffraction beam is lowered by Δf relative to the frequency of the zeroth order light. For this reason, the rotational radial grating PRG acts as a frequency shifter.

Incident light beams ± LF and zero-order light D ₀ each consisting of the first-order diffraction beams ± D _1n (n = 1, 2, 3) of the three wavelength components are applied to the collimator lens 10 as shown in FIG. By this, the chief rays of light are converted to be parallel to each other, and reach the beam selection member 12. The beam selecting member 12 functions as a spatial filter placed on a so-called Fourier transform surface, so that the 0th order light D ₀ is blocked, and the incident light beam + LF by the first diffraction light + D _1n passes.

Then, the incident light beam ± LF reaches the beam splitter (half mirror) 20 through the adjustment optical systems 14, 16, 18 made of parallel flat glass of which the tilt amount is variable. The adjustment optical system 14 has a function of eccentricity with respect to the optical axis of the lens 10 without changing the distance in the Fourier space between the incident light beam + LF and the incident light beam -LF, and the adjustment optical systems 16 and 18 have the incident light beam. It has the function of adjusting the position with respect to each optical axis of + LF and incident light beam -LF separately.

The incident light beam ± LF is divided into two in the beam splitter 20, one of which is incident on the objective lens 22, and the other of the incident light beam ± LF is of a specific wavelength in the divided incident light beam ± LF. the primary beam, i.e., is selected only the first order beam of the ± ₁₂ D λ ₂ incident on the condenser lens (Fourier transformation lens) 26.

On the other hand, the incident light beams ± LF incident on the objective lens 22 become parallel beams, respectively, and simultaneously irradiate the lattice MG on the wafer? At different angles. Thereby, on the grating MG, an interference fringe produced by the interference of the incident light beam ± D ₁₁ of wavelength λ _1, an interference fringe produced by the interference of the incident light beam ± D ₁₂ of wavelength λ ₂ , and an incident light beam ± D ₁₃ of wavelength λ ₃ . Three of the interference fringes created by the superimposition appear in the same pitch and in the same phase. Further, because of the frequency difference 2 · Δf between the incident light beams + LF and -LF, the interference fringe is observed as if it is moving at a constant velocity on the grating MG in one direction. The moving speed is proportional to the speed V of the grating RG of the rotating radial grating RRG. In addition, as is apparent from FIG. 24, the wafer W surface (lattice MG) and the radial lattice plate RRG are arranged so as to be conjugated to each other by the synthesis system of the collimator lens 10 and the objective lens 22. Therefore, the diffraction image by the ± 1st order diffracted light of the grating RG of the radial grating plate RRG is formed on the grating MG of the wafer W, but since the 0th order light D ₀ is shielded, the diffraction image of 1/2 of the pitch of the grating RG (interference fringe Intensity distribution) is formed. The pitch Pif on the wafer W of the interference fringe is set to 1/2 of the pitch Pmg of the grating MG as in the previous embodiment.

When the above relationship is satisfied, the first-order diffracted light is generated vertically from the grating MG by irradiation of the incident light beam ± LF. In other words, the first-order diffracted light generated vertically by the irradiation of the incident light beam + LF and the first-order diffracted light generated vertically by the irradiation of the incident light beam LF generate an interference beam BM. The interference beam BM becomes beam light intensity-modulated at the frequency 2 · Δf. As described above, since the first-order diffraction light (interfering beam BM) is generated in the same direction, from another viewpoint, the focal length of the objective lens 22 is set to F ₀ on the Fourier transform plane of the incident light beam ± LF for each wavelength. The interval DL _n from the optical axis,

_{DLn = F 0 · sinθ n =} ± F 0 · λ n / Pmg (n = 1, 2, 3)

Set to. The setting of the interval DL _n for each wavelength can be adjusted by properly determining the pitch of the grating RG of the rotational radial grating plate RRG and the focal length of the collimator lens 10.

In addition, since the interference fringe formed on the wafer W is formed as a diffraction image of the grating RG of the radial grating plate RRG, in principle, the pitch of the interference fringe by one wavelength component among three wavelengths λ ₁ , λ ₂ , and λ ₃ When the pitches of the lattice marks MG of the wafer W become integer multiples, the pitches of the interference fringes due to different wavelength components are also related to each other, and the interference fringes of the respective wavelength components are completely matched to each other so that the phase offset and position are mutually correct. It did not cause an offset. In practice, however, interference fringes for each wavelength component cause position offsets, phase offsets, and pitch offsets according to the degree of chromatic aberration of the optical system such as the objective lens 22 and the collimator lens 10.

Therefore, in order to correct this offset, the adjusting optical systems 14, 16, and 18 in Fig. 24 are used. These optical systems 14, 16, and 18 are made of parallel flat glass, and when the material having a large color dispersion is used, the mutual position offset and phase offset of the interference fringe formed on the wafer W for each wavelength component are minutely reduced. Can change. Alternatively, as the adjustment optical systems 14, 16, and 18, a parallel plate glass having a small color dispersion and a parallel plate glass having a high color dispersion are combined to adjust the inclination of the interference fringes for each wavelength component by tilt adjustment of the parallel plate glass having a high color dispersion. It is possible to correct the relationship and correct the inclination of the parallel plate glass with small color dispersion with respect to the overall inclination error on the wafer of the incident light beam ± LF generated by the correction.

The interference beam BM generated vertically from the grating MG illuminated by the interference fringe as described above passes through the objective lens 22 and the beam splitter 20 to reach the spatial filter 28. This spatial filter 28 is disposed on or near the Fourier transform surface of the objective lens 22, and has an opening for transmitting only the interference beam BM (± first order diffracted light) in this embodiment. The interference beam BM transmitted through the spatial filter 28 is converted into a parallel beam in the lens system (Fourier transform lens) 30 and then each of the first dichroic mirror 32 and the second dichroic mirror 34. The wavelength is selected by.

In addition, the beam B _m1 of the component of the wavelength λ ₁ of the interference beam BM is 90% or more from the dichroic mirror 32 to the dichroic reflection is received by the photoelectric elements (36A). The beam B _m2 of the component of wavelength λ ₂ of the interference beam BM passes through the dichroic mirror 32 and is then reflected by the dichroic mirror 34 to 90% or more to be received by the photoelectric element 36B. The beam B _m3 of the component of the wavelength λ _{3 in} the BM passes through the dichroic mirrors 32 and 34 and is received by the photoelectric element 36c. These photoelectric elements 36A, 36B, 36C have the same function as the photoelectric elements DT3, DT4 in FIG. 21, except that the interference beams B _m1 , B _m2 , B _m3 to be received are each of the intensity at the bit frequency 2Δf. It only differs in that it is modulated. In addition, depending on the intervals of the wavelengths λ ₁ , λ ₂ , and λ ₃ to be used, the wavelength division by the dichroic mirrors 32 and 34 may be insufficient, so that each of the light receiving elements 36A, 36B, and 36C An interference filter (low pass filter) may be disposed just before. It goes without saying that the dichroic mirrors 32 and 34 are the same as the dichroic mirrors DCM5 and DCM4 of the incident light measurement, respectively.

The photoelectric signals I _m1 , I _m2 , and I _m3 of each photoelectric element 36A, 36B, 36C are level-changed in a sinusoidal shape at a frequency equal to the bit frequency 2 · Δf while the interference BM from the lattice mark MG is present. It becomes a waveform.

On the other hand, the primary beam ± D ₁₂ selected by the wavelength selection filter 24 and incident on the condenser lens 26 is irradiated on the transmissive reference grating SG. Nevertheless, the reference grating SG is disposed in conjugation with the rotational radial grating plate RRG (frequency modulator) with respect to the synthesis system of the collimator lens 10 and the condenser lens 26. For this reason, the one-dimensional interference fringe by the two-beam interference of the primary beam +-D12 is formed also on the reference grating SG, and it moves at a speed corresponding to the bit frequency 2? Δf.

Thus, if the pitch of the reference grating SG and the pitch of the interference fringe are appropriately determined, the ± first-order diffracted light generated from the reference grating SG becomes the interference beam B _ms in the same direction, and it passes through the spatial filter 38 The light is received by the photoelectric element 40. The photoelectric signal I _ms of the photoelectric element 40 becomes a waveform which level changes to a sine wave at the same frequency as the bit frequency 2Δf, and the signal I _ms is a heterodyne reference signal.

In the above configuration, the reference lattice SG is made by depositing a chromium layer on a glass plate and etching the chromium layer so that transparent lines and light shielding lines are alternately formed, so that at least an asymmetrical, resist layer such as lattice mark MG on the wafer W is formed. An ideal grating with little problem of, i.e., a grating symmetrical in amplitude transmittance. For this reason, the pair of incident light beams irradiated to the reference grating SG can obtain sufficient accuracy only by the primary beam corresponding to one of the three wavelengths λ ₁ , λ ₂ , and λ ₃ . Of course, all three primary beams ± D ₁₁ , ± D ₁₂ , and ± D ₁₃ included in the incident light beam ± LF may be irradiated simultaneously with the reference grid SG to form a multicolor interference fringe similar to the grid mark MG on the wafer. good.

If a multi-color interference fringe is formed on the reference grating SG as described above, and the interference beam Bms generated from the reference grating SG is configured to be photoelectrically detected for each wavelength, the reference signal according to the wavelength λ ₁ and the reference according to the wavelength λ ₂ Since the signal and the reference signal corresponding to the wavelength λ ₃ can be obtained separately, the position measurement of the grating mark MG can be performed for each wavelength. In addition, even if the interference fringes generate mutually constant position offsets (phase offsets) for each of the three wavelength components formed on the wafer W, it is possible to measure them as an offset amount in advance. It will be described later in detail.

By the way, the wafer W shown in FIG. 24 is located on the wafer stage WST which moves two-dimensionally in the surface (XY plane) perpendicular | vertical to the optical axis of the objective lens 22. FIG. The two-dimensional movement on this stage WST is performed by the drive source 42 including a drive motor, and may be either a method of being sent by the motor to rotate the feed screw or a method of directly moving the stage main body by a linear motor. Next, the coordinate position of the stage WST is measured in order by the laser interferometer 44. The measured value of this laser interferometer 44 is used for feedback control of the drive source 42. Next, a part of wafer stage WST is provided with reference mark plate FG. On the mark plate FG, a reflective intensity grating (pitch is the same as the lattice MG on the wafer) formed by patterning a line and space with a chromium layer on the surface of quartz. Because of this, the intensity grating is characterized in that, unlike a phase grating such as lattice mark MG formed by irregularities on the wafer W, the diffraction efficiency without asymmetry does not depend on the wavelength of the illumination light (or detection light), that is, the asymmetry in amplitude reflectance There is no characteristic. Next, the reflectance of the chromium layer also does not change in the wavelength band (generally 0.5 to 0.8 mu m) of the illumination light for position detection. Therefore, the reference mark plate using the intensity grating on the FG if can be accurately obtained, thereby obtaining the photoelectric signal I _m1, I _m2, the ratio of the change in the amplitude of the I and trade of each _m3 for each wavelength.

In the above-described apparatus of FIG. 24, a semiconductor laser is used as a light source, but in this case, a standard optical system for removing the non-score difference between the semiconductor lasers LS ₂ and LS ₃ and the dichroic mirrors DCM ₄ and DCM ₅ (tilted providing a plurality sheets of the parallel flat glass plates, etc.), it is appropriate that the beam component of each wavelength component of the beam LB ₀ synthesized by one in the same aperture. Also in other cases, it is preferable to provide a beam shaping optical system for each wavelength component of the diameter of the beam LB ₀ after synthesis.

In FIG. 24, in order to simplify the description, the rotational radial grating plate RRG is used as the frequency shifter, but only the first zebra laser light source and the center wavelength lambda that use two other acoustic optical modulators (AOM) or oscillate at the center wavelength lambda 1 _{Only the second} agent oscillating at 2 may use the laser light source as the light source. However, in the case of the Zeeman laser, since the two laser beams generally having complementary polarization directions oscillate and give a frequency aberration of several hundred kilo Hz to several mega Hz between the beams, the bit frequency of the interference beam for photoelectric detection is also As it increases, the photoelectric elements 36A, 36B, 36C, and 40 use PIN diodes or photomultipliers having high reliability.

The various dichroic mirrors shown in FIG. 24 may be replaced with a dispersion element such as a prism. In this case, one prism has the same function as, for example, two dichroic mirrors DCM ₄ , a set of DCM ₅ , or a set of dichroic mirrors 32, 34.

Next, an example of a position detection and position control circuit suitable for the apparatus of FIG. 24 will be described with reference to FIG. In the heterodyne scheme of FIG. 25, the signals I _m1 , I _m2 from each photoelectric element 36A, 36B, 36C, 40 while the interference beam BM is generated from the lattice mark MG on the wafer W or the reference mark plate FG. , I _m3 , and I _ms are sinusoidal alternating current waveforms as shown in FIGS. 26A to 26D.

FIG. 26D shows the temporal change in intensity of the signal I _ms which became the reference signal, and FIGS. 26A, 26B and 26C show the signal I _m1 when receiving the interference beam BM from the lattice mark MG on each wafer W. FIG. An example of the temporal intensity change of, I _m2 and I _m3 is shown. So signal when a phase reference of I _ms, the signal phase of the I _m1 is a signal for a I _ms - △ Ψ shifted by _one, and the signal phase of the I _m2 is for the signal I _ms - eogeutnamyeo than △ Ψ _2, signal I _m3 The phase of is assumed to be shifted by -ΔΨ ₃ with respect to the signal I _ms . In addition, the amplitude of the amplitude (a peak-to-peak AC component) of the signal I _m1 is the amplitude of E _1, E ₂ is the signal I _m2, I signal _m3 is assumed that E _3.

In the circuit block shown in FIG. 25, each signal I _m1 , I _m2 , I _m3 , I _ms is input to an analog-to-digital conversion (A / D converter) circuit unit 50, so that the sampling clock generation circuit ( In response to the clock signal (pulse) C _ps from 52), the instantaneous intensity level of each signal is converted into a digital value. The clock signal frequency of the C _ps signal I _mn determined (n = 1, 2, 3 ), higher enough than that of I _ms bit frequency, the clock signal C _ps is sent to the waveform memory circuit unit (54), A / D It is used to update the memory address when storing the digital value (data) from the converter 50. Accordingly, in the waveform memory circuit unit 54, the 26A [deg.] To 26D also four waveform data showing the predetermined period of the respective signals I _mn, I _ms to be the digital sampling over a (e.g., 10 cycles or more) . At this time, since the four signals I _mn and I _ms are sampled simultaneously by the common clock signal C _ps , it is assumed that each waveform data in the waveform memory circuit unit 54 has no offset on the time axis. In addition, when the rotational radial grating plate RRG is used, since the bit frequency is the upper limit of several kHz, the clock signal C _{ps is} also good as several tens of kHz. In addition, in the case of using a frequency shifter in which two AOMs are arranged in series, such as (E) Japanese Patent Laid-Open No. 6-82215, since the bit frequency is determined to be twice the frequency difference of the high frequency modulated signal applied to each AOM, it is relatively free. It is possible to decide.

Further, each waveform data in the memory circuit unit 54 is read into the detection circuit unit 56 of the phase difference ΔΨ _n (n = 1, 2, 3) and the position offset ΔX _n (n = 1, 2, 3). Therefore, each phase difference ΔΨ ₁ , ΔΨ ₂ , ΔΨ _{3 as} shown in FIG. 7 is calculated by digital calculation (Fourier integral method). As previously assumed, if the pitch Pmg of the lattice mark MG of the wafer W and the pitch Pif of the interference fringe irradiated thereon are set to Pmg = 2Pif, one period of each waveform of FIGS. 26A to 26D corresponds to Pmg / 2. do. In addition, since the phase difference measurement is generally performed in the range of ± 180 degrees, the detection circuit 56 sets the calculated phase differences ΔΨ ₁ , ΔΨ ₂ , and ΔΨ ₃ in the range of ± Pmg / 4 according to the previous equation (7). It is converted to a position in the offset amount _{△ X 1, △ X 2,} △ X 3. This offset amount ΔX _n shows the offset within ± Pmg / 4 of the grating mark MG with respect to the reference grating SG.

Therefore, if the resolution of the phase difference measurement is about 0.2 degrees, the resolution of the offset amount is (0.2 / 180) Pmg / 4, and if the pitch Pmg is 4 µm, the practical range is about 0.002 (2 nm). have.

On the other hand, the signal amplitude and amplitude ratio detection circuit unit 58 reads out each waveform data such as those shown in Figs. 26A to 26D stored in the waveform memory circuit unit 54, and the amplitude values E ₁ , E ₂ , and E for each waveform. ₃ is detected by digital calculation. The detection circuit unit 58 has the amplitude of each of the photoelectric signals I _m1 , I _m2 , and I _m3 obtained when the interference beam BM generated from the grating of the FG of the reference mark plate is received by the photoelectric elements 36A, 36B, and 36C in advance. The values A ₁ , A ₂ and A ₃ are stored.

That is, before measuring the lattice mark MG on the waiter W, the lattice mark of the reference mark plate FG is moved below the objective lens 22, so that the signal as shown in Figs. 26A to 26D from each photoelectric element 36A, 36B or 36C. Is generated and stored in the waveform memory circuit unit 54, and the amplitude detection circuit 58 detects and stores the amplitude values A ₁ , A ₂ , and A ₃ . At this time, the stop position of the stage WST where the reference mark plate FG is detected is read out from the laser interferometer 44 and stored, and at the offset amount detecting circuit unit 56, the position offset amounts ΔX _b1 and ΔX _b2 for each wavelength. ,? X _b3 can also be used as data at the time of baseline determination.

In addition, the base line referred to here means mutual minute error when the position offset amounts ΔX _b1 , ΔX _b2 , and ΔX _b3 of the lattice marks on the mark plate FG measured for each wavelength differ only in extremely small amounts. . Originally, in the incident light system shown in FIG. 24, the interference fringes for each wavelength generated on the reference mark plate FG by the respective beams of wavelengths λ ₁ , λ ₂ , and λ ₃ closely match each other, thereby detecting photoelectricity for each wavelength. If the electrical responsiveness and distortion characteristic of a system are fully provided, each value of the position offset amount (DELTA) _Xb1 , (DELTA) _Xb2 , (DELTA) _Xb3 of mark plate FG will correspond completely.

However, as a practical matter, when the resolution is about 2 mm, it is difficult to adjust the incident photometer or the detection system such that the position offset amounts DELTA X _b1 , DELTA X _b2 , and DELTA X _b3 are constant at the resolution degree. Therefore, the mutual difference between the position offset amounts ΔX _b1 , ΔX _b2 , and ΔX _b3 measured by the mark plate FG remains as an offset (baseline error) inherent in the alignment system shown in FIG. 24.

The base line error is a position offset amount ΔX _b1 obtained by first detecting each of the position offsets ΔX _b1 , ΔX _b2 , and ΔX _b3 for each wavelength determined by the detection circuit 56 by detecting the lattice mark MG on the wafer W. , ΔX _b2 , ΔX _b3 , respectively. By way of example, since the limit in claim 24 degrees device an interference beam B _ms obtained from the reference grid SG to the wavelength λ _1, based on the position offset amount of the measured reference mark plate _{FG △ X b1 △ X b2 -} △ X b1 = _{ΔX b21} , _{ΔX b30} - _{ΔX b31} = _{ΔX b31} is calculated and stored. And correcting and calculating the value of ΔX ₂ such that the measured position offset amounts ΔX ₁ , ΔX ₂ , ΔX ₃ for ΔX ₂ -ΔX ₁ = ΔX _b2 1 measured on the lattice mark MG on the wafer W , ΔX _{3 -ΔX 1} = ΔX _b3 1 may be calculated by correcting the value of ΔX ₃ .

Of course, when the interference B _ms obtained from the reference grid SG is configured to include the respective wavelengths λ ₁ , λ ₂ , and λ ₃ , and when the reference signal is generated by separately detecting the interference beam for each wavelength, each reference signal (each The position offset amounts ΔX _b1 , ΔX _b2 , and ΔX _b3 of the reference mark plate FG are determined for each wavelength), and the position offset amounts ΔX ₁ , ΔX ₂ , ΔX ₃ of the grid marks MG on the measured wafer. _{X 1 - △ X b1, △} X 2 - may be calculated as the correction _{△ X b2 - △ X b2,} △ X 3.

Next, the amplitude non-detection circuit unit 58 measures each ratio of the amplitude values A ₁ , A ₂ , A ₃ stored in advance and the amplitude values E ₁ , E ₂ , E ₃ obtained when the grid mark MG on the wafer W is detected. C ₁ , C ₂ , C ₃ are calculated as C ₁ = E ₁ / A ₁ , C ₂ = E ₂ / A ₂ , C ₃ = E ₃ / A ₃ . This ratio C ₁ . C ₂ and C ₃ correspond to the weighting coefficients described in the embodiment of FIG. 21.

The data of the position offset amounts ΔX ₁ , ΔX ₂ , ΔX ₃ and the ratios C ₁ , C ₂ , C ₃ obtained as described above are sent to the weighted averaging calculation circuit unit 60, where the weighted grid mark MG The offset amount ΔX of is calculated. The operation is performed by the following equation.

ΔX = (C ₁ · ΔX ₁ + C ₂ · ΔX ₂ + C ₃ · △ X ₃ ) / (C ₁ + C ₂ + C ₃ )

The offset amount ΔX thus obtained is an offset in the pitch direction of the grating mark MG with respect to the reference grid SG, and the data is sent to the position control and the display 62, and the wafer W is aligned in real time (positioning). Is also sent to the servo control circuit unit 64.

As shown in Fig. 9, this servo control circuit unit 64 has two functions, one of which is a function of feedback control of the drive source 42 until the offset amount? X becomes a predetermined value (direct servo Mode). In the case of this function, the operations of the A / D converter circuit 50, the memory circuit unit 54, the offset amount detection circuit unit 56, and the averaging circuit unit 60 are repeated one after another, for a very short time (for example, Every several msec), the value of the offset amount DELTA X is calculated. In addition, the amplitude detection circuit unit 58 calculates the ratio C _1, C ₂ C ₃ may be the manyirado by one of the first, it may be carried out every time for each calculation of the offset amount △ X. Ratio C _1, C _2, that is, if every time of performing the calculation of the C ₃ every calculation of the offset amount △ X by the weighted averaging circuit unit 60 is non-C _1, the value of C _2, C ₃ if some change Of course. In the non-C _1, C _2, while detecting the same grating marks MG after that if the output of C ₃ first only once or multiple times with the same ratio of the value is used.

On the other hand, one function of the servo control circuit unit 64 is a function (interferometer servo mode) which moves the wafer stage WST based on the measured value of the laser interferometer 44. This function locates, for example, the lattice of the reference mark plate FG on the stage WST or the lattice mark MG on the wafer W directly under the objective lens 22, and based on the position of the detected lattice mark MG, Is used when positioning a point just below the objective lens 22. In this interferometer servo mode, target position information of the wafer stage WST from the position controller 62 is output to the servo control circuit unit 64, and the control circuit unit 64 reads the stage WST read from the laser interferometer 44. The drive source 42 is feedback-controlled so that the deviation between the current position and the target position of the laser beam falls within a predetermined allowable range (for example, ± 0.04 µm).

Also in the present embodiment, in the ninth embodiment, the direct servo mode can be executed following the interferometer sub mode.

The position control and indicator 62 also has a function of displaying the coordinate position of the grid mark MG and the obtained offset amount ΔX different from the above-described servo mode switching instruction. In some cases, the values of the ratios C ₁ , C ₂ , and C ₃ , which serve as weighting coefficients when the grid mark MG is detected, are also stored and stored. In this case, the same grating mark MG in a number of locations on the wafer W is formed, and when the sequentially detects the position of their marks MG, ratio C _1, C _2, C ₃ is also sequentially remember leave a mark of a certain part on the wafer W It can be verified that MG has a problem due to asymmetry or a resist layer. The portion where the weighting coefficients (ratio C ₁ , C ₂ , C ₃ ) are greatly changed may be graphically displayed on the wafer W. FIG. At this time, if the wafer prior to applying the resist layer through a chemical process such as a diffusion process or an etching process is mounted in the apparatus of FIG. 24 to obtain a change in weighting coefficient, the influence on the wafer surface caused by the chemical process can be indirectly investigated. It may be. In addition, when the resist layer is applied to the wafer to obtain the same change in weighting coefficient, and compared with the change in weighting coefficient before coating, the influence of the resist layer can be indirectly investigated.

In the tenth embodiment described above, the reference mark plate FG is provided on the stage WST, and the rate of change of the signal amplitude, that is, the ratios C ₁ , C ₂ , and C ₃ , is determined for each wavelength by using the reference mark plate FG. 21 degrees), it is not necessary to prepare photoelectric elements DT ₁ and DT ₂ for directly detecting the light intensities of the incident light beams LB ₁ and LB ₂ . On the contrary, in the eighth (or ninth) embodiment, when the reference mark plate FG, which is a reference, is juxtaposed with the lattice MG, the ratio C ₁ and C ₂ can be detected without providing the photoelectric elements DT ₁ and DT ₂ . It means you can.

27 shows the configuration of the signal processing circuit according to the eleventh embodiment, where the wavelength selective filter 24 shown in FIG. 24 is omitted, and the interference beam B _ms from the reference grid SG is replaced by a dichroic mirror or the like. It is _assumed that three photoelectric elements t40A, 40B, 40C are separated into beams B _ms1 , B _ms2 , B _ms3 for each of the three wavelengths λ ₁ , λ ₂ , and λ ₃ , and photoelectrically detected by them. In this case, the signals I _m1 , I _m2 and I _m3 from the photoelectric elements 36A, 36B and 36C for measurement are respectively referred to by reference signals I _ms1 , I _ms2 and I _ms3 from the photoelectric elements 40A, 40B and 40C. Phase difference detection is performed between and. In other words, this position is offset (△ X ₁₎ of the grating mark MG at the time when using the incident light beam of wavelength λ ₁ by obtaining the phase difference △ Ψ ₁ of the reference signal I _ms1 As for the measurement signal I _m1 is obtained.

In such a configuration, since the number of signals to which waveform data are to be inserted is large, three sets of waveform sampling circuits (the A / D converter 50 in FIG. 25 and the clock) corresponding to each wavelength are shown in FIG. It has a function of the generation circuit 52 and the memory circuit 54; 80A, 80B, and 80C are provided. Since the internal configurations of the circuits 80A, 80B, and 80C are the same, a detailed configuration of only the circuit 80A is shown in FIG. 27, and detailed descriptions thereof are omitted in the other circuits 80B and 80C.

In the present embodiment, as shown in the circuit 80A, the photoelectric element 40A that receives the signal I _m1 from the photoelectric element 36A that receives the interference beam B _m1 for measurement and the interference beam B _ms1 for reference. Input signal I _ms1 from the S / H circuits 800 and 802, respectively, and the signal level from the S / H circuits 800 and 802 through the analog multiplexer 804. (ADC) 806 is input.

The digital value converted by the ADC 806 is written to the accessed address of the rewritable memory (RAM) 808. The RAM 808 generates an address value by the address counter 810, and the address value is incremented (or decremented) in response to the clock signal C _ps . In this case, however, the address counter 810 has a special function, and the clock signal C _ps is supplied as a flag to one of the specific upper bits of the address counter. Thus, the address space of the RAM 808 is divided into two pages. The address space of the first page is accessed between the clock signal C _ps and the logic &quot; 0 &quot;, while the clock signal C _ps is the logic &quot; 1 &quot;. Two pages of address space are accessed.

The clock signal C _ps is also supplied to the S / H circuit 802. The clock signal C _ps is also supplied to the timing circuit 814 and used to generate signals (pulses) for the timing of digital conversion of the ADC 806 and the data write timing of the RAM 808.

Therefore, when the clock signal C _ps is "1", the S / H circuit 802 is held, and the level at that time of the signal I _m1 of the photoelectric element 36A is changed through the analog multiplexer 804 through the ADC 806. The digital value corresponding to the level is stored in one address position in the address space of the second page of the RAM 808. In contrast, when the clock signal C _ps is "0", the S / H circuit 802 is held and the level at that time of the signal I _ms1 of the photoelectric element 40A is supplied to the ADC 806 through the multiplexer 804. The digital value corresponding to the level is stored in one address position in the address space of the first page of the RAM 808.

The above operation is repeated at a high speed for a predetermined period (for example, 10 cycles or more) of the signal I _m1 (or I _ms1 ), and the waveform data of the reference signal I _ms1 is stored in the first page of the RAM 808, and 2 The waveform data of the measurement signal I _m1 is stored in the page. In this way, the set of waveform data stored in the RAM 808 responds to the address value set in the address counter 810 from data ABS such as a microprocessor, and is read out to the data bus DBS of the microprocessor. The microprocessor processes each waveform data by a program which achieves the same function as the respective detection circuits 56 and 58 shown in FIG. _6, and calculates the position offset amounts DELTA X ₁ , DELTA X ₂ , DELTA X ₃ .

The configuration and operation of the above sampling circuit 804 are completely the same in the sampling circuits 80B and 80C, and the circuit 80B temporarily _stores respective waveform data of the measurement signal I _m2 and the reference signal I _ms2 , and the circuit 80C. ) Temporarily _stores the waveform data of the measurement signal I _m3 and the reference signal I _ms3 .

Since the A / D converter is not provided for each of the measurement signal I _mn and the reference signal I _msn in the eleventh embodiment, sampling at the same time in a microsecond order is difficult, but as a practical matter, the bit frequency of the interference beam is several tens of kHz. If it is below, the concurrency on the order of microseconds is not very necessary. It is advantageous to reduce the hardware cost of the signal processing circuit by simplifying the circuit configuration by halving the number of A / D converters and the like.

By the way, when the reference mark plate FG which has the well-known reflectance chromium surface on the wafer stage WST is fixed, it can use for the measurement of the baseline amount which has the said mark plate FG, and the measurement of a focus state as mentioned above. The base line amount basically means a measurement operation for determining the relative positional relationship between the projection point of the center of the mask (reticle) mounted on the projection exposure apparatus and the detection center point of various wafer alignment systems.

Therefore, in the case where the position detecting device as shown in FIG. 24 is applied to each alignment system of the projection exposure apparatus shown in FIG. 15, the detection center points R _fi , R _f2 , R _f3 and R _f4 of each alignment system are determined by the reference grid SG. It is prescribed. However, in the RA of the reticle alignment system, the mark (lattice pattern) RM for the reticle alignment around the reticle R and the corresponding lattice mark on the reference mark plate FG are irradiated with illumination light having the same wavelength as the illumination light for the projection exposure of the pattern PR. When the reticle stage RST is moved so that both marks are in a predetermined positional relationship, the detection center point R _f1 is not required.

This fact is the same also in alignment system TTRA, and the mark for alignment for every mark by die on the reference mark plate FG or the mark on the wafer W, and the die-by die (D / D) formed in the periphery of the pattern PR of the reticle R Is taken as an image and the detection center point Rf2 need not be defined separately in the case of a system for detecting the position offset on both marks.

Here, the baseline amount is in the X and Y directions between the projection point (substantially coincides with the optical axis AX) of the center CCr of the reticle R and the projection point toward the wafer side of each detection center point Rf1, Rf2, Rf3, and Rf4. It is not different from positional relationship. The positional relationship detects the position offset amount between the mark group corresponding to the reference mark plate FG and the projection points of the respective detection center points Rf1 to Rf4 in each of the alignment systems RA, TTRA, TTLA, and OFA itself, and at the time, the wafer stage WST. Can be obtained by detecting the coordinate position of the laser interferometer 44 (see FIG. 24).

Thus, when the alignment system is equipped with a heterodyne position detection device such as the 24th degree, the grating of the reference mark plate FG is detected during such baseline measurement operation, and the photoelectric elements 36A, 36B, and 36C shown in FIG. The amplitude levels A ₁ , A ₂ , and A ₃ of the signals I _m1 , I _m2 , and I _m3 can be stored in the circuit unit 58 of FIG. The pupil surface EP in the projection optical system PL shown in FIG. 15 is the same as the Fourier transform surface EP shown in FIG. The optical axis of the objective lens provided on each of the alignment systems RA, TTRA, and TTLA for detecting an object (mark of wafer or mark of reference mark plate FG) on the wafer stage WST through the projection optical system PL is substantially the optical axis AX on the wafer stage WST side. It is set to be parallel to. When the reticle side as well as the wafer side of the projection optical system PL are telecentric (in the case of FIG. 15), the optical axis of the objective lens of each alignment system is parallel to the optical axis AX of the projection optical system PL even on the reticle side. The extension of the optical axis of the objective lens passes through the center of the pupil surface EP of the projection optical system PL (the portion through which the optical axis AX passes).

The effective radius of the pupil surface EP corresponds to the numerical aperture (N, A) which determines the resolution (minimum resolution line width) of the projection lens PL, and currently a projection lens of about N. A. = 0.5 to 0.7 is developed.

FIG. 28 shows an example of the main part of the alignment system TTLA among the alignments shown in FIG. 15, and shows two incident light beams LF (LF in FIG. 24) for detecting the grating mark MG or the reference mark plate FG on the wafer. + LF and beam -LF) correspond to correction optical system CG, beam splitter 20 (corresponding to half mirror 20 in FIG. 24), objective lens OBJ (corresponding to objective lens 22 in FIG. 24), 2 Incident on the projection lens PL through two mirrors MR. At this time, the surface FC which is conjugate with the surface of the wafer W is formed between two mirror MRs, and the two beams LF intersect in the rear surface FC. The beam ± LF is relayed by the projection lens PL, and irradiates the lattice marks MG on the wafer as well.

The interference beam BM from the grating mark MG passes through the center of the pupil EP of the projection lens PL, and through the mirror MR, the objective lens OBJ, and the beam splitter 20, the dichroic mirror DCM (the dichroic mirror in FIG. 24). (Equivalent to (32)), and the wavelength is divided here. For example, the incident light beam ± LF two wavelengths λ _1, λ _2, that if, dichroic mirror, DCM interference beam having a wavelength λ ₁ interference beam B _m1 photoelectric element induced in (36A), and the wavelength λ ₂ of a dichroic B _m To the photoelectric element 36B.

In such an alignment system TTLA, when the incident light beam ± LF includes a plurality of wavelength components (around 30 to 40 nm apart), the influence of chromatic aberration (axis image and magnification) of the projection lens PL or the chromatic aberration of the objective lens OBJ As a result, the crossing area of the beam LF irradiated onto the wafer is slightly shifted in the Z direction or the XY direction for each wavelength component. A correction optical system CG for correcting an error occurring in accordance with chromatic aberration is provided in the optical path of the incident light beam ± LF as in the 28th degree. In the correction optical system CG, the lens is composed of a convex lens, a combination lens thereof, a parallel flat glass, or the like, and the adjustment optical systems 14, 16, and 18 shown in FIG. 24 may be used.

In the case of the alignment system TTRA in FIG. 15, the mark DDM for D / D alignment on the reticle R is a diffraction grating, and the relative position offset of the mark DDM and the lattice mark MG on the wafer W corresponding to the mark DDM is equal to 24th degree. When detecting by the heterodyne system, as shown in FIGS. 17A and 17B, a parallel plate-shaped correction plate PGP is provided on the pupil surface EP of the projection lens PL, and the incident light beam (± LF) on the correction plate PGP is provided. Alternatively, the phase diffraction gratings PG ₁ , PG ₂ , and PG ₃ can be arranged only at the position where the interference beam BM passes, and the influence of chromatic aberration on the axial and chromatic aberration on the magnification can be reduced.

Next, a twelfth embodiment of the present invention will be described. In the present embodiment, the configuration shown in Fig. 24 is based on, and as described with reference to Figs. 4, 5A, and 5B, the lattice mark is used in addition to the interference beam of the ± 1st order diffracted light from the lattice mark. The structure which detects the interference beam of 0th order light and 2nd order diffraction light of was also added. In the case of photoelectric conversion of an interference beam between zero-order light and second-order diffracted light into a single photoelectric device, and using the photoelectric signal to detect the position offset of the grating mark, the incident light beam for grating mark illumination is multiplied. After receiving the interference beams (multiple wavelengths) of the 0th order light and the 2nd order light with a single photoelectric device, it is difficult to detect a good position offset as it is. The reason for this is that the photoelectric signal IR02 ₁ obtained by photoelectric detection for each of the three wavelength components λ ₁ , λ ₂ , and λ ₃ , for example, an interference beam of zero-order light and secondary light as shown in FIGS. 29A to 29D. Observing the waveforms of IK02 ₂ and IK02 ₃ makes it easy to understand. That is, as shown in FIGS. 29A to 29D, the photoelectric signal I _mn when the phase difference of the three photoelectric signals IK02 _n (n = 1, 2, 3) is an interference beam of ± 1st order diffracted light (FIG. 7) This is because the phase difference is large compared to the phase difference of the reference). For this reason, when a light intensity change of each wavelength having originally a large phase difference is received by a single photoelectric element, the amplitude of the photoelectric signal (the amplitude of the alternating current) becomes extremely small due to the cancellation effect of the intensity of each wavelength. The interference beams of the zeroth order light and the secondary light are generated at symmetrical angles on both sides of the interference beam BM of the first-order diffraction light D _ln .

However, the Fig. 29A, Fig. 29B claim, the 29C turning the fourth diagram showing a zero-order-secondary beam of light interference, for example, the primary interference of light beams of three wavelengths interference beam that appears on the left side of the BM to the λ _1, The photoelectric signal I _ms which shows the wavelength in the heterodyne system of each photoelectric signal IK02 ₁ , IK02 ₂ , and IK02 ₃ when photoelectric detection is individually performed for each of λ ₂ and λ ₃ , and the 29D or 26D is the same reference signal as that of the 26D. Shows the waveform.

On the other hand, the Fig. 30A, Fig. 30B claim, the fourth Figure 30C of the zero-order-secondary interference of light beams shown in degrees, an interference beam that appears on the right side of the ± 1-order light beam interference BM 3 wavelengths λ _1, λ Waveforms of the heterodynes of the photoelectric signals IK20 ₁ , IK20 ₂ , and IK20 ₃ when photoelectric detection is individually performed every ₂ and λ ₃ are _shown , and FIG. 30D shows the waveform of the same photoelectric signal I _ms as the 29D degree. . As shown in FIGS. 29A, 29B, 29C, 30A, 30B, and 30C, the phase offsets Δβ of the signals IK02 _n and IK20 _n (n = 1, 2, 3) _01, has a _{△ β 02, △ β 03,} △ β 21, △ β 22, △ β 23 are simultaneously produced, the gap is strong dependence of the wavelength, the larger, the reverse trend in signal IK02 _n and IK20 _n for the same wavelength .

Thus, the configuration of this embodiment will be described with reference to FIG. FIG. 31 is a part of the configuration of FIG. 24, specifically, a photoelectric detection system of various interference beams from the grating mark MG. Therefore, the same reference numerals are given to members having the same function as those in FIG. Incident light system 100 in FIG. 31 includes light sources LS ₁ , SL ₂ , SL ₃ , mirror MR, dichroic mirror DCM ₄ , DCM ₅ , radial grating RRG as a frequency shifter, and lens 10 shown in FIG. 24. And the spatial filter 12, the adjusting optical systems 14, 16, 18, and the like, and emit a pair of incident light beams + LF, -LF. The incident light beam ± LF including the components of the wavelengths λ ₁ , λ ₂ , and λ ₃ is partially reflected by the half mirror 20 to enter the objective lens 22, and a part of the incident light beam is incident on the reference light receiving system 110. . The reference light receiving system 110 is composed of the waveform selection filter 24, the lens 26, the reference grid SG, and the spatial filter 38 in FIG. 24, and guides the reference light B _ms to the photoelectric element 40.

Then, when the grating MG on the wafer W is irradiated by the incident light beam ± LF through the objective lens 22, the interference beam BM of the ± first-order diffracted light is generated vertically and at the same time in the zero direction in the direction of travel of each incident light beam. Various interference beams of secondary light are generated. The interference beams of the 0th-secondary light are directed to the dichroic mirrors 32 and 34 through the objective lens 22 and the half mirror 20, where they are separated for each wavelength component. First, the dichroic mirror 32 reflects two interference beams of zero-order secondary light of wavelength lambda ₁ , and the interference beams are received at 36Al and 36A2, respectively. Of course, the interference beam B _m1 of ± 1 order light of wavelength λ ₁ is reflected by the dichroic mirror 32 and received by the photoelectric element 36A.

In addition, the interference beams of 0-second order light having wavelengths λ ₂ and λ ₃ transmitted through the dichroic mirror 32 and the interference beams B _m2 and B _m3 of ± 1 order light are each wavelength component of the dichroic mirror 34. Separated, _two interference beams of 0-secondary light of wavelength λ ₂ are received by the photoelectric elements 36B ₁ and 36B ₂ , respectively, and the interference beams B _m2 of ± 1st-order light are received by the photoelectric element 36B. do. Next, two interference beams of 0-second order light having a wavelength λ ₃ transmitted through the dichroic mirror 34 are received by the photoelectric elements 36C ₁ and 36C ₂ , and the interference beam B _{m3 of} ± 1 order light. Is received by the photoelectric element 36C.

As is apparent from the above configuration, in this embodiment, the photoelectric signal I _ms from the photoelectric element 40 is used as a reference signal, and each photoelectric element 36A, 36A ₁ , 36A ₂ , 36B, 36B ₁ , 36B ₂ , 36C, 36C _A signal processing circuit for obtaining the phase difference of the photoelectric signal from ₁ , 36C ₂ ) is required. An example of the simplest circuit configuration for that is shown in FIG.

FIG. 32 is an improvement of a part of the processing circuit shown in FIG. 25, and in hardware, each photoelectric signal other than the reference signal I _ms input to the A / D converter circuit 50 in FIG. The addition of the analog multiplexer 120 is different. The analog multiplexer 120 includes three input 1 output switching switches SS ₁ , SS ₂ , and SS ₃ , and each switch SS ₁ , SS ₂ , SS ₃ is switched in response to an external switching signal SN.

The switch SS ₁ selects _one of three photoelectric signals I _m1 , IK02 ₁ , and IK20 ₁ obtained by receiving an interference beam having a wavelength λ ₁ , and the switch SS ₂ receives three photoelectric rays obtained by receiving an interference beam having a wavelength λ ₂ . One of the signals I _m2 , IK02 ₂ , IK20 ₂ is selected, and the switch SS ₃ is connected to select one of three photoelectric signals I _m3 , IK02 ₃ , IK20 ₃ obtained by receiving an interference beam of wavelength λ ₃ . . In this embodiment, however, the switches SS ₁ to SS ₃ are linked to each other, so that three measurement signals (photoelectric signals) input to the A / D converter circuit 50A at the same time are signals detected under the same diffraction state. That is, when the switches SS ₁ to SS ₃ are switched to the intermediate position, they become exactly the same as in FIG. 25, and the signals I _m1 to I _m3 which photoelectrically detect the interference beam BM of ± 1st order light for each wavelength component are A / D. When supplied to the converter circuit 50 and the three switches SS ₁ to SS ₃ are switched to the positions shown in FIG. 32, the wavelengths of the 0th-secondary light generated on the left side of the interference beam BM of ± 1st light are each wavelength. The photoelectrically detected signals IK02 ₁ , IK02 ₂ , and IK02 ₃ are supplied to the A / D converter circuit 50 every time. Of course, when the three switches SS ₁ to SS ₃ are switched to the rightmost positions, the photoelectric signals IK20 ₁ , IK20 ₂ , and IK20 ₃ are supplied to the A / D converter circuit 50.

In addition, in the amplitude detection and amplitude ratio detection circuit 58 shown in FIG. 25, the ratio data C _n1 , C _n2 , C _n3 grouped for each of the interference beams having different diffraction states are shown in FIG. 1, 2, 3) is changed to output. Of these data, C _n1 (n = 1, 2, 3) is the same as the ratio C ₁ , C ₂ , C _{3 in} FIG. 25, and C _n2 (n = 1, 2, 3) is the photoelectric signal IK02 _n. is the ratio for each wavelength obtained in (n = 1, 2, 3), and C _n3 (n = 1, 2, 3) is the ratio for each wavelength obtained in the photoelectric signal IK20 _n (n = 1, 2, 3). to be.

The phase difference and position offset detection circuit 26 shown in FIG. 25 shows offset amounts ΔXn ₁ , ΔXn ₂ , ΔXn ₃ (n = 1, 2) grouped for each interference beam having different diffraction states in FIG. 32. , 3) to output. Of these offsets, ΔX _n1 (n = 1, 2, 3) is the same as the offset amounts ΔXn ₁ , ΔXn ₂ , ΔXn ₃ of FIG. 6, and ΔXn ₂ (n = 1, 2, 3) is photoelectric signal _{IK02 n (n = 1, 2} , 3) and the offset amount for each wavelength component obtained _{from, △ Xn 3 (n = 1} , 2, 3) is a photoelectric signal _{IK20 n (n = 1, 2} , 3) Offset amount for each wavelength component obtained from In addition, the detection circuit 56 intermediates the values according to the phase differences _{Δβ 0n} and _{Δβ 2n} (n = 1, 2, 3) as described in FIGS. 29A to 29D and 30A to 30D. Calculate as

Next, the weighted averaging circuit 60 in FIG. 25 is changed to a selective weighted averaging circuit in FIG. 32, and calculates the final position offset amount ΔX based only on the photoelectric detection result of the interference beam BM of ± 1st order light. Based on the operation mode of FIG. 1, which is the same as that of FIG. 25, the second operation mode that calculates the final offset amount ΔX based only on the photoelectric detection result of the interference beam of the 0th-second order light, and the interference beam photoelectric detection result of all And a third calculation mode for calculating the final offset amount [Delta] X. These three operation modes can be appropriately selected by the operator, but when the third operation mode is designated, two to three calculation algorithms can also be selected. Such mode designation and algorithm selection will be described later in detail.

Also in the case of this embodiment, the stage WST is firstly positioned so that the lattice mark of the reference mark plate FG on the wafer stage WST is irradiated with the incident light beam ± LF from the objective lens 22. Then, the switching signal SN is given to the analog multiplexer 120, and the switches SS ₁ to SS ₃ are set at, for example, the position shown in FIG. 32, and the interference of the 0th-order secondary light generated from the lattice mark of the reference mark plate FG. Signal IK02 _n (n = 1, 2, 3) in the photoelectric signal obtained by photoelectric detection of the beam is digitally sampled by the A / D converter circuit 50, and each waveform of the signal IK02 _n is temporarily stored in the memory circuit 54. Remember.

The waveform data in the memory circuit 54 is analyzed by the amplitude detection circuit 58, and the amplitude value (peak to peak) of each signal IK02 _n is calculated and stored as J02 _n (n = 1, 2, 3). .

Next switch SS ₁ to SS ₃ to claim 23, also of the switch to the right, and the reference mark plate FG of the grating mark resulting from the 0th-order -2-order light of the photoelectric detection of the interference beam to IK20 signal of the photoelectric signal obtained by _n (n = 1, 2, and 3 are digitally sampled by the A / D converter circuit 50, and each waveform of the signal IK20 _n is temporarily stored in the memory circuit 54. At this time, when the storage capacity of the memory circuit 54 is not large enough, each waveform data of the signal IK02 _n stored in advance is erased and the waveform data of the signal IK20 _n is overwritten. Thereafter, the amplitude detection circuit 58 analyzes the waveform data in the memory circuit 54 and sets the amplitude value (peak to peak) of each signal IK20 _n (n = 1, 2, 3) to J20 _n (n = 1). , 2, 3) to be stored.

Finally, each of the signals I _mn (n = 1, 2, 3) obtained by switching the switches SS ₁ to SS ₃ to the intermediate position and photoelectrically detecting the interference beam of ± 1st order light generated from the lattice mark of the reference mark plate FG equally stored in the memory 54, the waveform data, and storing the calculated amplitude values _{J11 n (n = 1, 2} , 3) of the signals I _mn by the amplitude detection circuit 58.

Since the preliminary operation is completed by the above, the wafer W to be actually positioned and aligned next is placed on the stage WST, and the lattice mark MG on the wafer W is irradiated with the incident light beam ± LF from the objective lens 22. Position the stage WST if possible.

Then, in the same manner as in the detection of the lattice mark of the reference mark plate FG, switching the switches SS ₁ to SS ₃ of the multiplexer 120 in sequence puts each signal waveform data into the memory circuit 54 and onto the wafer. each amplitude value of the signal of each interference beam generated from the grating mark MG which is obtained by photoelectrically detecting _{I mn, IK02 n, IK20 n} (n = 1, 2, 3) respectively by the detection circuit (58) E _n (No. 26A Fig. To 26D)), E02 _n , and E20 _n (see FIGS. 29A to 29D and 30A to 30D).

In addition, when the switches SS ₁ to SS ₃ are switched to store one set of the signals I _mn , IK02 _n , and IK20 _n in the memory circuit 54, the waveform of the reference signal I _ms is also under the same time axis. Remember to. And switch before switching the SS ₁ to SS _3, the phase difference, analysis of waveform data of the signal stored in the signal I _mn, IK02 _n, IK20 _n by the offset detection circuit 56, and phase △ Ψ _n, △ β _n , One corresponding to among Δβ _{2n and} one corresponding to position offset amounts ΔX _n1 , ΔX _n2 , ΔX _n3 (n = 1, 2, 3) are sequentially calculated.

In this way, when the amplitude value and the position offset amount for each wavelength are obtained with the detection light (interference beam) having different diffraction states, the amplitude ratio detection circuit 58 performs the following calculation.

Next, the averaging circuit 60 calculates the offset amount (displacement) ΔX most likely to be most accurate, but is the same as in the case of FIG. 25 in the first calculation mode using only the interference beam BM of ± 1st order light.

ΔX = (C ₁₁ · ΔX ₁₁ + C ₂₁ · ΔX ₂₁ + C ₃₁ · ΔX ₃₁ ) / (C ₁₁ + C ₂₁ + C ₃₁ ).

On the other hand, in the second operation mode using only the interference beams of the 0th-secondary light, the phase difference _{Δβ 0n} and the interference beam obtained by the detection of the interference beam of the 0th-secondary light occurring on the left side of the interference beam BM of the ± 1st-order light The algorithm which calculates the position offset amount for each wavelength is employ | adopted from the average phase difference with phase difference (DELTA) (beta) _2n obtained by detection of the interference beam of the 0th-order secondary light which generate | occur | produces on the right side of BM. The averaging of the phase difference is an averaging which should be carried out in principle in the case of position detection using an interference beam of 0th order light and ± 2nd order light, unlike the averaging which reduces the random component to improve the accuracy.

Thus, in this embodiment, based on the algorithm, the averaging circuit 60 first calculates each position offset amount obtained from the signal IK02 _n from each position offset amount ΔX _n2 (n = 1, 2, 3) and the signal IK20 _n . The average value ΔXA _n (n = 1, 2, 3) for each wavelength with _{ΔX n3} (n = 1, 2, 3) is calculated as follows.

ΔXA ₁ = (ΔX ₁₂ + ΔX ₁₃ ) / 2

ΔXA ₂ = (ΔX ₂₂ + ΔX ₂₃ ) / 2

ΔXA ₃ = (ΔX ₃₂ + ΔX ₃₃ ) / 2

In addition, the averaging circuit 60 measures the average value CA _n (n = 1, 2, 3) for each wavelength component of the amplitude ratios C _n2 and C _n3 of the interference beam of the 0th-order secondary light obtained by the amplitude ratio detecting circuit 58. Calculate as

CA ₁ = (C ₁₂ + C ₁₃ ) / 2

CA ₂ = (C ₂₂ + C ₂₃ ) / 2

CA ₃ = (C ₃₂ + C ₃₃ ) / 2

Thereafter, the averaging circuit 60 sets the average ratio CA _n for each wavelength component as an intermediate coefficient, weights averages the average position offset amount ΔXA _n for each wavelength component as follows, and is most likely The offset amount ΔX is calculated.

ΔX = (CA ₁ , ΔXA ₁ + CA ₂ , ΔXA ₂ + CA ₃ , ΔXA ₃ ) / (CA ₁ + CA ₂ + CA ₃ )

By the above, the position or position offset detection of the grid mark MG by a 2nd operation mode is achieved.

In the third operation mode, the first algorithm simply averages the position offset amount calculated in the first operation mode and the position offset amount calculated in the second operation mode, and the second algorithm weighted averages of the two position offset amounts. Either can be set in advance by an operator. Therefore, the position offset amount finally calculated in the first calculation mode (the mode using the detection result of the interference beam of the ± 1st order light) is ΔXM ₁ , and the position offset amount finally calculated in the second calculation mode is ΔXM. _{If it is 2} , the position offset amount determined by the first algorithm is calculated as (ΔXM ₁ + ΔXM ₂ ) / 2.

On the other hand, in the second algorithm, the offset amount ΔXM ₁ calculated in the first calculation mode and ΔXM ₂ calculated in the second calculation mode are weighted average using predetermined weighting factors Q ₁ and Q ₂ . As an example, the weighting factor Q ₁ is the amplitude values E ₁ , E ₂ , E ₃ of the signals I _mn (n = 1, 2, 3) obtained by photoelectric detection of the interference beam BM of ± 1st order light (Fig. 26A). Corresponding to the sum of Fig. 26D), the weighting factor Q ₂ is obtained for each wavelength of the signals IK02 _n , IK20 _n (n = 1, 2, 3) obtained by photoelectric detection of an interference beam of zero-order secondary light. Corresponds to the sum of the average amplitude values (E02 ₁ + E20 ₁ ) / 2, (E02 ₂ + E20 ₂ ) / 2, and (E02 ₃ + E20 ₃ ) / 2. Therefore, in the second algorithm, the offset amount? X of the grid mark MG is determined by the following calculation.

ΔX = (Q ₁ , △ XM _I + Q ₂ , △ XM ₂ ) / (Q ₁ + Q ₂ )

Further, in principle, since high order diffracted light and its light intensity are small, the interference of 0th and -2nd order diffracted light is compared with the light intensity amplitude (corresponding to E _n ) of the interference beam BM of the ± 1st order diffracted light. The light intensity amplitude (corresponding to E02 _n , E20 _n ) of the beam is significantly smaller. Therefore, when the weighting factors Q ₁ and Q ₂ are determined simply by the sum of the amplitudes of the signals I _mn , IK02 _n , and IK20 _n , in most cases, the weighting factor Q ₁ becomes larger than the coefficient Q ₂ . The coefficient Q ₂ is preferably corrected to increase the calculated value by, for example, a predetermined ratio (10 to 30% as an example).

Next, a thirteenth embodiment of the present invention will be described with reference to FIG. In this embodiment, the structure of the reference mark plate FGN on the wafer stage WST shown in FIG. 24 is changed to a transmission lattice (a lattice that is not asymmetric to the amplitude transmittance), and photoelectric detection of the interference beam generated through the lattice by was to obtain the denominator (reference value) to be used at the time of output from each photoelectric signal I _mn, detecting the amplitude ratio _n of the IK02, IK20 _n circuit 58.

33 shows a partial surface of the wafer stage WST, and when the incident light beam ± LF (here, two wavelengths of wavelengths λ ₁ and λ ₂ ) is irradiated with a grating on the reference mark plate FG, zero from the grating toward the inside of the stage is shown. Light shielding, ± 1st order light, ± 2nd order light is generated. These diffracted light are bent at right angles in the mirror MR to enter the lens system G ₅ having a Fourier transform function, and either component of the offset of wavelengths λ ₁ and λ ₂ is selected in the wavelength selective filter 24 having the automatic conversion function. , Respectively, the interference beams B _mrn , ± B _1r , and ± B _2r enter the photoelectric element group DTR. The wavelength selective filter 24 is comprised so that the filter which permeate | transmits the wavelength (lambda) ₁ , and cuts off the wavelength (lambda) ₂ , and the filter which has the opposite characteristic may be inserted into and detached from an optical path. Accordingly, when the filter for selecting the wavelength λ ₁ used ± 1-order light beam B _mr1 interference due to zero-order-secondary interference of light beam wavelength λ ₁ and ± B _lr by wavelength λ ₁ reaches the photoelectric device group DTR , The interference beam B _mr2 of ± 1st order light due to the wavelength λ ₂ reaches the photoelectric element group DTR. Therefore, the wavelength when using the λ ₁ selected filter has obtained a photoelectric signal I _mr1, photoelectric signals IR02 _1, IR20 _1, the wavelength λ ₂ when using the selected filter, the photoelectric signal I _mr2, photoelectric signals IR02 _2, IR20 ₂ Is obtained.

In the case of the heterodyne system, these photoelectric signals appear as sinusoidal waveforms of the same frequency as the bit frequency, and the three switches SS ₁ to SS ₃ of the analog multiplexer 120 of the processing circuit shown in FIG. Connected to the A / D converter 50 selectively. Specifically, the three switches SS ₁ to SS ₃ in FIG. 32 are changed to 5 input 1 outputs, and 2 inputs thereof are photoelectric signals 1 _mr1 , IR02 ₁ , IR20 ₁ obtained at the time of detecting the reference mark plate FG. It is used to switch the input between the set of photoelectric signals and the set of photoelectric signals Imr ₂ , IR02 ₂ and IR20 ₂

The amplitude values of these photoelectric signals are obtained from the amplitude detection circuit 58 in FIG. 32 and stored. When calculating the amplitude ratio, the following calculation is performed, for example.

C ₁₁ = I _m1 / I _mr1

C ₂₁ = I _m2 / I _mr2

C ₁₂ = IK02 ₁ / IR02 ₁

C ₂₂ = IK02 ₂ / IR02 ₂

C ₁₃ = IK20 ₁ / IR20 ₁

C ₂₃ = IK20 ₂ / IR20 ₂

In this embodiment, the interference beam of the diffracted light transmitted through the reference mark plate is photoelectrically detected by the photoelectric element group DTR, and the phase information of each photoelectric signal obtained from the element group DTR and the phase information of the photoelectric signal I _ms as reference signals. In comparison with the above, the position offset of the reference mark plate FG, or the measurement of the position, that is, the operation of a part of the baseline measurement can be used in combination.

Next, a fourteenth embodiment of the present invention will be described with reference to FIG. In the present embodiment, the polarization directions of a pair of incident light beams + LF and -LF complementary to the lattice mark MG for measurement (alignment) on the wafer W (or reference mark plate FG) through the objective lens 22 are complementary. Let's do it. That is, in the case of linearly polarized light, the polarization directions of the incident light beams + LF and -LF are orthogonal to each other, and in the case of circularly polarized light, the beams + LF and -LF are set to be reversely polarized light. For this reason, the two incident light beams ± LF do not interfere with each other, and the ± first-order diffraction light BM of wavelengths λ ₁ , λ ₂ , and λ ₃ vertically generated from the grating mark MG does not interfere with each other.

Therefore, when the photoelectric detecting the ± 1-order diffracted light BM through the objective lens 22, a small mirror MR _2, uses a polarizing beam splitter PBS as the analyzer (analyzer). In this way, the ± 1st order light BM transmitted through the polarization beam splitter PBS interferes with each other to become the first interference beam BP ₁ , and the ± 1st order light BM reflected from the polarization beam splitter PBS interferes with each other to the second interference BPS ₂ . do. The polarities of these interference beams B _P1 and B _P2 are complementary to each other, but each interference beam is intensity-modulated into a sinusoidal shape in accordance with the bit frequency if it is a heterodyne system. In addition, the phases of the intensity modulation of the interference beams B _P1 and B _P2 differ by about 180 degrees.

In addition, in the case where the half wave plate HW shown in the figure is a case where the linearly polarized light directions of the incident light beams ± LF and the first-order diffraction light BM orthogonal to each other are different (rotating) from the polarization separation direction of the polarization beam splitter BPS, It is provided for the purpose of correcting the linear polarization direction between +/- 1st order diffracted light BM. For this reason, when the linear polarization directions orthogonal to each other between ± 1st-order diffraction light BM coincide with the polarization separation direction of the polarization beam splitter PBS from the beginning, or when the incident light beams + LF and -LF are reversely rotated to become circularly polarized light, It is not necessary to use the half wave plate HW.

In this embodiment, the interference beam BP1 is discriminated for each wavelength through the dichroic mirrors 332 and 34, the component of the wavelength λ ₁ of the interference beam BP1 is received by the photoelectric element 36A ₁ , and the component of the wavelength λ ₂ . Is received by the photoelectric element 36B ₁ , and the component of wavelength λ ₃ is received by the photoelectric element 36C ₁ . Similarly, the dichroic mirrors 32 and 34 also discriminate the wavelength of the interference beam BP ₂ , and receive the light by the photoelectric elements 36A ₂ , 36B ₂ , and 36C ₂ for each component of the wavelengths λ ₁ , λ ₂ , and λ ₃ . .

In addition, both output signals of the photoelectric elements 36A ₁ and 36A ₂ are subtracted by the differential amplifier to be the photoelectric signal I _m1 , and both output signals of the photoelectric elements 36B ₁ and 36B ₂ are subtracted by the differential amplifier to provide the photoelectric signal I _m2 , and both output signals of the photoelectric signals 36C ₁ and 36C ₂ are subtracted by the differential amplifier to become the photoelectric signal I _m3 .

The use of the differential amplifier in this manner is, for example, the output signal of the photoelectric element 36A ₁ and the output signal of the photoelectric element 36A ₂ are in phase out of each other (180 ° difference), and are included in both outputs. The phase noise component is subtracted and canceled, and the substantial S / N ratio of the signal I _m1 is improved. The objective lens 22 shown in FIGS. 24, 31, and 34 is corrected to some extent on at least axial chromatic aberration among various chromatic aberrations occurring in the wavelength ranges λ ₁ to λ _{3 used} . It is preferable. For example, when the band of wavelengths λ ₁ to λ ₃ to be used is 100 nm or less, such axial chromatic aberration selects a base material of a plurality of lens elements constituting the objective lens 22 and has a lens having a different refractive index dispersion ratio. By combining the elements, correction can be made to some extent. Of course, such chromatic aberration does not need to be corrected completely in the objective lens 22, but can be corrected by the adjustment optical systems 14, 16, and 18 shown in FIG.

As mentioned above, although each Example of this invention was described, when grating mark MG on the wafer W or the reference mark plate FG is detected by the homodyne system, the level change of each photoelectric signal is prescanned by prescanning the grating mark MG in the pitch direction. You need to sample. In that case, the simplest method is to measure the clock signal C _ps for the signal waveform sampling shown in FIG. 25 or 32 degrees from the measurement pulse (for example, every 0.02 µm) from the laser interferometer 44 for position measurement of the stage WST. 1 pulse). In this way, waveform data of each photoelectric signal generated during prescan of the grating mark MG over several pitches is stored in the memory circuit 54 corresponding to the grating position of the grating mark MG.

In the method of irradiating two incident light beams ± LF to the grating mark MG, it is preferable that the two incident light beams ± LF have an incident angle symmetrical with respect to at least the pitch direction of the grating mark MG. In the method of projecting one incident light beam into a beam, the incidence angle is preferably set to zero (vertical incidence) with respect to the pitch direction of the grating mark MG.

By the way, when projecting the multi-wavelength illumination beam to the grating mark MG (or reference mark) for measurement, a plurality of laser beams are not co-synthesized once for each wavelength as shown in FIGS. 21, 23, and 24 degrees. It may be configured so as to separate and enter the non-measurement direction orthogonal to the measurement direction (pitch direction) of the mark position on the Fourier transform surface of the lattice mark MG. That is, the angle of incidence to the grating mark MG for each wavelength of the plurality of illumination beams may be different from the non-measurement direction.

Claim 35 of the two wavelengths incident on the rear group lens system G ₂ or the objective lens 22 of the projection lens beam turning ± LFλ _1, showing the incident light of the shape of ± LFλ _2, and their beam ± LFλ _1, ± LFλ ₂ is lattice It consists of two beams of optical axes on the Fourier transform plane (the pupil plane) EP for the mark MG. And then the pitch direction of the grating mark MG also ground and a direction perpendicular to a wavelength λ beam ± LFλ ₁ and wavelength λ ₂ beam ± LFλ ₂ the left-right direction in the non-measuring direction (surface in the same figure on the deflection plane Fourier EP of the _first )to be.

As a result, the interference beams B _m1 and B _{m2 of} the ± first-order diffracted light generated from the lattice mark MG and returned to the Fourier transform plane EP also pass through positions separated in the non-measurement direction on the Fourier transform plane EP for each wavelength. The interference beam B _m1 is generated from the mark MG by irradiation of the incident light beam ± LFλ ₁ , and the interference beam B _m2 is generated from the mark MG by irradiation of the incident light beam ± LFλ ₂ , and the incident light beam and the interference light beam are Fourier transform planes. On EP, it is distributed like 36th degree, for example.

37. The method of claim 36, also, when setting the orthogonal axis (the measurement axis and the non-measuring axis) to the center of the deflection plane Fourier EP as the origin, the incident light beam of the second set ± LFλ _1, ± LFλ ₂ non-measuring axes offset in the direction of The amount Dh corresponds to the offset amount in the non-measuring direction of the interference beams B _m1 and B _m2 by the first-order diffracted light. In this way, when the beam irradiating the lattice mark MG is inclined in the non-measurement direction for each wavelength component, the interference beams B _m1 and B _m2 are also distributed separately in the Fourier transform surface EP, so that the light receiving surface of each photoelectric detector is It is arrange | positioned on the Fourier conversion surface EP or on the surface conjugated with the surface EP, and similarly, photoelectric detection is attained.

That is, if a plurality of interference beams (interference of ± 1st order diffraction light, 0-2nd order diffraction light) to be photoelectrically detected are separated on the Fourier transform plane EP for each wavelength, they are the same dichroic mirror as in each of the above embodiments. Photoelectric detection is possible individually without using. Therefore, it is not necessary to use a wavelength selection element such as a dichroic mirror or a band pass filter as a method of separating the detection interference beam for each wavelength.

In addition, the multi-wavelength of the incident light beam is not limited to the laser light source, and can be realized even when using light from a halogen lamp and light from a high-brightness LED. When using the light from a halogen lamp, the wavelength selection filter which has a predetermined | prescribed bandwidth is provided, and the light (broadband light) of about 20-100 nm wavelength width selected by this filter is guided and used, for example with an optical fiber. In this case, since the incident light beam irradiating the lattice mark MG on the wafer has a continuous intensity distribution within the selected wavelength bandwidth, an interference filter (bandwidth of 3 to 10 nm) extracting only a specific wavelength component in front of each photoelectric device in the light receiving system is used. It may be arranged fixedly or interchangeably.

As described above, according to the eighth to eighth embodiments of the second invention of the present application, the diffracted light generated from the lattice mark for position detection on the substrate is multiplied or widened by the illumination light for position detection. Photoelectric detection is independently performed for each wavelength component, and the mark position information is detected for each photoelectric signal obtained thereby and averaged in the calculation. Therefore, high-precision position detection that reduces the influence of the asymmetry of the mark and the thickness of the resist layer is possible. Become. In addition, when photoelectric detection of the diffracted light from the mark is performed to obtain an independent photoelectric signal for each wavelength component, even if the intensity is different for each wavelength component of the illumination light, it does not impair the averaging effect by the multi-wavelength as in the prior art. There is also an advantage.

Next, according to each of the eighth to fourteenth embodiments, even when the diffracted light to be photoelectrically detected is made of higher order components, the higher order diffracted light (0th order and 2nd order) multiplied into a single photoelectric element as in the prior art There is no cancellation phenomenon that occurs when receiving an interference beam of light, etc. at the same time, and the position detection and alignment with high precision can be made remarkably compared with the prior art.

Furthermore, in the second invention of the present application, the attenuation ratio (amplitude ratio) of the intensity level of the diffracted light is obtained for each wavelength component detected photoelectrically, and the diffraction light whose signal amplitude with a small attenuation ratio is relatively large is positioned by a large weighted averaging operation. Since detection is performed, the effect of a high precision of position detection is also acquired remarkably compared with simple averaging.

Next, in the embodiment of the present invention, the incident light beams having multiple wavelengths are simultaneously irradiated with the lattice mark MG. However, a high-speed shutter is provided after each of the light sources LS ₁ , SL ₂ , SL ₃ in FIGS. 6 and 24. Any one of the incident light beams of λ ₁ , λ ₂ , and λ ₃ may be switched in time series for injection. In this case, several photoelectric elements of a light receiving system can be made common without separating and preparing for every wavelength previously. In this way, when the incident light beam is emitted in time series for each wavelength component, the high-speed shutter function must be set on the device, but the number of photoelectric elements and the number of parts in the signal processing circuit (particularly the number of A / D converters and memory chips) Not only can be significantly reduced, but the wavelength selection filter 24 in the stage WST shown in FIG. 33 can also be omitted.

Thus, an example of the case where the incident light beam for each wavelength component is converted in time series will be described as the fifteenth embodiment with reference to FIG. 37 is based on the structure of FIG. 6 described above, and therefore the same reference numerals are attached to the same members or similar members as those in FIG.

In FIG. 37, the three laser light sources LS ₁ , LS ₂ , LS ₃ emit laser beams LB ₁ , LB ₂ , LB ₃ having different wavelengths λ ₁ , λ ₂ , and λ ₃ , respectively. ₁ is set to a semiconductor laser light source having a lambda ₁ = 0.635 µm, a light source LS ₂ is a semiconductor laser light source having a lambda ₂ = 0.690 µm, and a light source LS ₃ is a semiconductor laser light source having a lambda ₃ = 0.760 µm.

Each of these laser light sources LS ₁ , LS ₂ , LS ₃ includes a driving circuit for supplying a stabilized driving current, a compensation circuit for compensating the influence of the temperature change of the laser element, or a variation in the oscillation center wavelength to monitor the wavelength. A laser power supply unit including a feedback control circuit for feedback control of the drive current so as to be stable is provided. These laser power supply units control the radiation of the laser beam from each laser light source CS ₁ , CS ₂ , CS ₃ to be turned off in response to the sequential signals CS ₁ , CS ₂ , CS ₃ from the switching control circuit TSC, respectively. do.

In this embodiment, in order to change the radiation of the laser beam from the laser light sources LS ₁ , LS ₂ , LS _{3 in order} by a predetermined irradiation time, the command signal CQ from the controller 62 in FIG. 10 will be described in detail later. In response, the switching control circuit TSC can be programmed. Therefore, a light source for oscillating a laser beam in any arbitrary time is limited to one of the three laser beams _{LS 1, LS 2, LS 3} . The switching control circuit TSC can change the beam oscillation timing of each light source LS ₁ , LS ₂ , LS ₃ depending on the content of the command signal CQ.

These three beams LB ₁ , LB ₂ , LB ₃ are aligned to pass through the coaxial optical path through the appropriate mirror MRa, dichroic mirrors DCM ₄ , DCM ₅ , and any one of the three beams is mirror MRb as beam LB0. Reflected at and incident perpendicularly to the rotating radial grating RRG. This rotary grating plate RRG acts as a frequency shifter for increasing or decreasing the frequency of diffracted light of each order to be diffracted in accordance with the angular velocity as shown in FIG.

Then, the grating RRG from the grating RG is a wavelength λ ₁ of the beam first-order diffracted beam upon irradiation of the LB ₁ ± D _11, the wavelength λ ₂ of the beam 1-order diffracted beams produced at the time of irradiation of LB ₂ ± D _12, or In the irradiation of the beam LB ₃ of the wavelength λ ₃ , one set of the first-order diffraction beams ± D ₁₃ occurs in response to the sequential driving of the switching control circuit TSC. The diffraction angle θ _n of the first diffraction beam ± D _1n for each wavelength is expressed as follows.

Sinθn = λ _1n / Prg

Where n represents the number of wavelengths and Prg represents the pitch of the grating RG.

Next, if the first diffraction beam ± D _1n receives a constant frequency shift Δf depending on the wavelength, and the speed at which the grating RG of the grating plate RRG crosses the beam LB ₀ is V, Δf = V / Prg. And the + 1st order diffraction beam + D _1n becomes higher by Δf for the frequency of the 0th order light D ₀ , and the -1st order diffraction beam -D _1n is lowered by Δf for the frequency of the 0th order light D ₀ . For this reason, the rotating grating RRG acts as a frequency shifter.

Incident light beams ± LF and zero-order light D ₀ , which are composed of any one of three sets of first-order diffraction beams ± D _1n (n = 1, 2, 3) of the three wavelength components, are collimator lenses as shown in FIG. (10) is converted so that the chief rays are parallel to each other and reaches the beam selection member 12 as a spatial filter, where the 0th order light D ₀ is blocked, and the incident light beam ± LF by the first order diffracted light ± D _1n To pass. Then, the incident light beam ± LF reaches the beam splitter (half mirror) 20 through the adjusting optical systems 14, 16, 18 made of parallel flat glass of which the tilt amount is variable. The adjustment optical system 14 has a function of eccentric with respect to the optical axis of the lens 10 without changing the distance in the Fourier space of the incident light beam + LF and the incident light beam -LF, and the adjustment optical systems 16 and 18 have the incident light beam + It has the function of adjusting the position with respect to each optical axis of LF and incident light beam -LF separately.

The incident light beam ± LF is divided into two in the beam splitter 20, one is incident on the objective lens 32, and the other is a condenser lens (Fourier) through adjustment optical systems 24A and 24B composed of parallel flat glass. Incident on the conversion lens;

On the other hand, the incident light beams ± LF incident on the objective lens 22 become parallel beams, respectively, and simultaneously irradiate the grating MG on the wafer W at symmetrical incident angles. Thus, on the grating MG, an interference fringe made by the interference of the incident light beam ± D ₁₁ of wavelength λ _1, an interference fringe made by the interference of the incident light beam ± D ₁₂ of wavelength λ ₂ , and an incident light beam ± D ₁₃ of wavelength λ ₃ . One of the interference fringes appears. The interference fringes for each of these wavelengths appear when the three beams LB ₁ , LB ₂ , and LB ₃ irradiate the rotating grating plate RRG at the same time, and overlap the same pitch and the same phase.

Further, because of the frequency difference 2 · Δf between the incident light beams + LF and -LF, the interference fringe is observed as if moving on the grating MG at a constant speed in one direction. The moving speed is proportional to the speed V of the grating RG of the rotating radial grating RRG. In relation to the frequency difference 2 占, the lighting time of each of the three light sources LS ₁ , LS ₂ , LS ₃ by the switching control circuit TSC is sufficiently longer than the period of the frequency difference 2Δf (bit frequency, for example, For example, if the frequency difference 2Δf is 10 Hz (cycle 0.1 ms), the lighting time of each of the three light sources LS ₁ , LS ₂ , and LS ₃ is preferably 10 ms or more. .

Further, as is apparent from FIG. 37, the wafer W surface (lattice MG) and the radial lattice plate RRG are disposed so as to be conjugated to each other by the composite optical system of the collimator lens 10 and the objective lens 22. Therefore, the diffraction image by the ± 1st-order diffraction light of the grating RG of the radial grating plate RRG is formed on the grating MG of the wafer W, but the 0th-order light D ₀ is blocked so that the diffraction image of 1/2 of the pitch of the grating RG (interfering fringe strength) Distribution). The pitch Pif on the wafer W of the interference fringe is set to 1/2 of the pitch Pmg of the grating MG in the same manner as in the previous embodiment.

When the above relationship is satisfied, the first-order diffracted light from the grating MG is generated vertically by irradiation of the incident light beam ± LF. That is, the interference beam BM (interference beam S _m1 of wavelength λ ₁ , wavelength) in which the first-order diffraction light generated vertically by the irradiation of the incident light beam + LF and the first-order diffraction light generated vertically by the irradiation of the incident light beam -LF are interfered with. one of the interference beam B _m2 of lambda ₂ and the interference beam B _m3 of wavelength lambda ₃ ). The interference beam BM is a bit light intensity modulated at the frequency 2Δf.

In this way, in order to generate the ± first order diffracted light (interference beam BM) in the same direction, on the Fourier transform plane of the incident light beam ± LF for each wavelength with the focal length of the objective lens 22 as F0. The distance D _Ln from the optical axis may be set to D _L n = F ₀ · sinQn = ± F ₀ · λ _n / Pmg (n = 1, 2, 3). The setting of the interval D _Ln for each wavelength can be adjusted by appropriately determining the pitch of the grating RG of the rotational radial grating plate RRG and the focal length of the collimator lens 10.

In addition, since the interference fringe formed on the wafer W is formed as a diffraction image of the grating RG of the radial grating plate RRG, in principle, the pitch of the interference fringe by one wavelength component among three wavelengths λ ₁ , λ ₂ , and λ ₃ If the pitch of the lattice marks MG of the wafer W is an integer multiple, the pitch of the interference fringes by other wavelength components is also such a relationship, and the interference fringes for each wavelength component are also beams LB ₁ , LB ₂ , LB from the light source. _{If 3} is irradiated at the same time, they will be completely coincident and will not cause phase offset or position offset with each other.

However, in practice, interference fringes for each wavelength component generate position offsets, phase offsets, and pitch offsets according to the degree of chromatic aberration of the optical system such as the objective lens 22 and the collimator lens 10. Therefore, in order to correct such an offset, the adjusting optical systems 14, 16, and 18 are provided in the same manner as in FIGS.

The interference beam BM generated vertically from the grating MG by the irradiation of the interference fringe as described above passes through the objective lens 22 and the beam splitter 20 to reach the spatial filter 28. The spatial filter 28 is disposed at or near the Fourier transform plane with respect to the objective lens 22, and has an opening for transmitting only the interference beam BM (± first order diffracted light) in this embodiment. The interference beam BM transmitted through the spatial filter 28 is converted into a parallel beam by the lens system (Fourier transform lens) 30 and then reflected by the mirror 32 and received by the photoelectric element DT ₀ .

The photoelectric element DT ₀ has the same function as the photoelectric element 36A in FIG. 6 except that the light-receiving interference beams B _m1 , B _m2 , and B _m3 are each intensity-modulated at the bit frequency 2 · Δf. . Therefore, the photoelectric signal I _mn output from the photoelectric element 36A is all a bit frequency while the interference beam BM from the lattice mark MG is present, that is, while any of the three LS ₁ , LS ₂ , and LS ₃ irradiates the beam. It becomes a waveform which changes in level into a sine wave shape at the same frequency as 2 (D).

On the other hand, the incident light beam ± LF (primary beams ± D ₁₁ , ± D ₁₂ ,) incident on the condensing lens 26 through the adjusting optical systems 24A and 24B, which are made of parallel flat glass or the like, is transmitted through the beam splitter 20. 1 set of ± D ₁₃ ) is irradiated superimposed on the transmission-type reference grid SG. Here again, the reference grating SG is disposed in conjugate with the rotational radial grating plate RRG with respect to the composite optical system of the collimator lens 10 and the condenser lens 26. For this reason, a one-dimensional interference fringe is formed on the reference grating SG by the beam interference of the incident light beam ± LF, and it moves at a speed corresponding to the bit frequency 2Δf.

In addition, the adjustment optical systems 24A and 24B compensate for the interference fringes generated on the reference grating SG for each wavelength component so as not to generate a position offset or a pitch offset from each other because of chromatic aberration of the condensing lens 26.

If the pitch of the reference grid SG and the pitch of the interference fringe are appropriately set, the ± first-order diffracted light generated from the reference grating SG proceeds in the same direction as the interference beam B _ms , which passes through the spatial filter 38 to allow photoelectricity. The light is received by the element 40. The photoelectric signal I _ms of the light receiving element 40 is a waveform which level changes to a sine wave shape at the same frequency as the bit frequency 2? Δf, and the signal I _ms is a heterodyne reference signal.

In the above configuration, the reference lattice SG is manufactured by depositing a chromium layer on a glass plate and etching the chromium layer so that transparent lines and shading lines are alternately formed, so that at least an asymmetrical, resist layer such as lattice mark MG on the wafer W is formed. An almost ideal grating with no problem, i.e., amplitude transmittance, is made as a symmetric grating.

For this reason, even if the pair of incident light beams irradiated to the reference grating SG is only the incident light beam corresponding to one of the three wavelengths λ ₁ , λ ₂ , and λ ₃ , sufficient accuracy is obtained. In this way, if an interference fringe is formed on each reference wavelength SG sequentially and the interference beam Bms generated by the reference grating SG is photo-divided for each wavelength, the reference signal according to the wavelength λ ₁ and the wavelength λ ₂ are formed. Since the reference signal and the reference signal according to the wavelength? 3 are obtained separately, the position measurement of the grating mark MG can be performed for each wavelength. Moreover, even if the interference fringes generate a fixed position offset (phase offset) from each of the three wavelength components formed on the wafer W, it can be measured as an offset amount in advance. It will be described later in detail.

By the way, the wafer W shown in FIG. 37 is mounted on the wafer stage which moves two-dimensionally in the surface (XY plane) perpendicular | vertical to the optical axis of the objective lens 22. FIG. The two-dimensional movement on the stage WST is performed by the drive source 42 including the drive motor, and may be a system in which the feed screw is rotated by the motor or a system in which the stage main body is directly interlocked by the electromagnetic force of the linear motor. The coordinate position of the stage WST is subsequently measured by the laser interferometer 44. The measured value of this laser interferometer 44 is used for feedback control of the drive source 42.

Next, the reference mark plate FG is prepared in a part of wafer stage WST. On the reference mark plate FG, a reflective intensity grating (pitch is the same as the grating MG on the wafer) formed by patterning a line and space with a chromium layer on the surface of the quartz glass.

For this reason, the intensity grating is characterized in that, unlike a phase grating such as grating mark MG formed with irregularities on the wafer W, the diffraction efficiency without asymmetry does not depend on the wavelength of illumination light (or detection light), that is, asymmetry in amplitude reflectance. It has no features. The reflectance of the chromium layer also does not change in the wavelength band of the illumination light for position detection (generally 0.5 to 0.8 mu m). Therefore, the reference mark plate by using the intensity grating on the FG, may accurately obtain a photoelectric signal I _m1, I _m2, ratios of amplitude variation and trade of I _m3 of photoelectric elements (36A) is obtained for each wavelength.

In the configuration of FIG. 37, a semiconductor laser is used as the light source, but in this case, a stereoscopic optical system for removing astigmatism between the semiconductor lasers LS ₁ , LS ₂ , LS ₃ and the dichroic mirrors DCM ₄ , DCM ₅ . installing the parallel flat glass plates of the plurality of pictures, and so on), and the beam component of each wavelength of the beam LB ₀ incident to the rotating radial grating RRG is preferably the same diameter. Also in other cases, it is preferable to provide a beam shaping optical system for each wavelength component of the diameter of the coaxial beam LB ₀ .

In FIG. 37, for the sake of simplicity, the rotational radial grid RRG is used as the frequency shifter, but in addition, two acoustic optical modulators (AOM) are used as in (E) Japanese Patent Application Laid-Open No. 6-882 / 5, or the center wavelength is used. Only the first agent laser light source oscillating at λ ₁ and the second agent laser light source oscillating at the center wavelength λ ₂ may be used as the light source. However, in the case of the Zeeman laser, since the two laser beams having the complementary polarization directions are generally oscillated and give a frequency difference of several hundreds of kHz to several MHz between the beams, the bit frequency of the interference beam for photoelectric detection is also high, The elements 36A, 40 and the like use a highly reliable PIN diode or photomal.

The various dichroic mirrors shown in FIG. 37 may be replaced with a dispersing element such as a prism. One prism in this case has the same function as a set of two dichroic mirrors DCM ₄ , DCM ₅ .

Next, an example of a position detection and position control circuit suitable for the apparatus of FIG. 37 will be described with reference to FIG.

FIG. 38 is based on the configuration of the control circuit of FIG. 25 described above, and like reference numerals denote circuit blocks having the same functions as the circuit blocks in FIG.

In the heterodyne system of FIG. 37, the signals I _m1 , I _m2 , I _m3 for each wavelength from the photoelectric element 36A are generated while the interference beam BM is generated from the lattice MG on the wafer W or the reference mark plate FG. I _ms from the photoelectric element 40 becomes an sine wave alternating current waveform as shown in FIGS. 39A to 39D.

However, in FIGS. 39A to 39D, the signals I _m1 , I _m2 , and I _m3 are laser light sources LS ₁ , LS ₂ , LS in response to signals CS ₁ , CS ₂ , CS ₃ from the switching control circuit TSC in FIG. 37. This is a signal when each of ₃ lights up. The reference signal I _{ms is} also output separately for each wavelength (in time division) in response to switching of lighting of the laser light sources LS ₁ , LS ₂ , LS ₃ , but is represented by one signal waveform in FIG. 39D.

39D shows a temporal change in intensity of the signal I _ms as a reference signal, and FIGS. 39A, 39B, and 39C show time-divisionally receiving, for each wavelength, an interference beam BM from the lattice mark MG on the wafer W, respectively. An example of the temporal intensity change of the signals I _m1 , I _m2 , and I _m3 as shown in FIG. Here, if, based on the phase of the signal I _ms, the signal phase of the I _m1 is for the signal I _ms - shifted △ by Ψ _1, the phase of the signal I _m2 is for a signal I _ms - eogeutnamyeo △ by Ψ _2, and the signals I _m3 is shifted by + DELTA Ψ ₃ with respect to the signal I _ms . In addition, the amplitude of the I signal _m1 (peak to peak of the AC component), is the amplitude of E _1, E ₂ is the amplitude of the signal I _{m2, m3} signal I are to E _3.

And, in the circuit block shown in FIG. 38, the photoelectric signal I _mn from the photoelectric element 36A and the photoelectric signal I _ms from the photoelectric element 40 are analog-to-digital conversion (A / D converter) circuit units ( 50, where the instantaneous intensity level of each signal is converted to a digital value in response to a clock signal (pulse) C _ps from the sampling clock generation circuit 52.

The clock generation circuit 52 is output from the position controller 62 to be described later, and controls the timing of sending the clock signal C _ps in response to the command signal CQ transmitted to the switching control circuit TSC in FIG. The timing is always the clock signal C _ps while the switching control circuit TSC outputs the signals C _sn (n = 1, 2, 3) to light the three light sources LSn (n = 1, 2, 3). Is set to output

The frequency of the clock signal C _ps is determined to be sufficiently higher than the signal I _mn (n = 1, 2, 3) and the bit frequency (2 · Δf) of I _ms , and the clock signal C _PS is the waveform memory circuit unit 54. It is also sent to and used to update the memory address when storing digital values (data) from the A / D converter 50.

At this time, in response to the command signal CQ, the waveform memory circuit unit 54 switches the storage area (address area) of the respective digital waveform data of the photoelectric signals I _mn and I _ms to correspond to each of the light sources that are turned on. For example, six RAM regions M1A, M1B, M2A, M2B, M3A, and M3B in 8K byte units are reserved as the waveform data storage space of the memory circuit unit 54. Then, the light source digital waveform data of the signal I _mn (I _m1) outputted from the A / D converter 50 while the LS ₁ is lit, the λ ₁ is the clock signal in response to the C _PS RAM region and sequentially stored in the M1A At the same time, the digital waveform data of the signal I _ms output from the A / D converter 50 is sequentially stored in the RAM area M1B in response to the clock signal C _ps .

When the light source LS ₂ is turned on in response to the command signal CQ, the digital waveform data of the signal I _mn (I _m2 ) is sequentially stored in the RAM area M2A, and at the same time, the digital waveform data of the signal I _ms is stored in the RAM area M2B. When the light source LS ₃ is turned on, the digital waveform data of the signal I _mn (I _m3 ) is sequentially stored in the RAM area M3A, and at the same time, the digital waveform data of the signal I _ms is sequentially stored in the RAM area M3A. Switch to

Accordingly, each of the three RAM regions MnA (n = 1, 2, 3) in the waveform memory circuit unit 54 has the same signal I _mn (n = 1, 2, 3) as in the 39A, 39B, and 39C degrees. ) Is digitally sampled over a predetermined period (e.g., 10 cycles or more), and each of the three RAM areas MnB (n = 1, 2, 3) in the memory circuit unit 54 is provided. The waveform data of the signal I _ms like the 39D diagram is digitally sampled over the same period as the signal I _mn .

At this time, each waveform data of the three measurement signals I _{mn in} the waveform memory unit 54 has different timings on the time axis, but each of the three measurement signals I _mn and the reference signal I _ms are the common clock signal C _ps. since at the same time sampling by, ask the reference signal phase difference △ Ψ _1, △ Ψ of the three measurement signal I _mn each waveform data on the basis of waveform data of I _{ms 2,} △ Ψ _3, the respective wavelengths λ _1, λ ₂ , the position offset amount of the grid mark MG for each λ ₃ can be known accurately.

In addition, when the rotational radial lattice board RRG is used, since the bit frequency has an upper limit of several Hz, the clock signal C _ps may also be several tens of Hz. In addition, when using a frequency shifter in which two acousto-optic modulators (AOMs) are arranged in tandem as in (E) JP-A 6-82215, the bit frequency is a frequency of a high frequency drive signal (several tens of MHz or more) added to each AOM. Since it is determined to be twice the difference of, it can be determined relatively freely.

And each waveform data in the memory circuit unit 54 shown in FIG. 38 is a detection circuit unit of phase difference (DELTA) _n (n = 1, 2, 3, position offset (DELTA) _Xn (n = 1, 2, 3). The phase difference ΔΨ ₁ , ΔΨ ₂ , and ΔΨ ₃ as shown in Figs. 39A, 39B, and 39C are calculated by digital calculation (Fourier integral method). As assumed, if the pitch Pmg of the lattice mark MG of the wafer W and the pitch Pif of the interference fringe irradiated thereon are set to Pmg = 2Pif, one period of each waveform of FIGS. 39A to 39C corresponds to Pmg / 2. do.

In addition, since the phase difference measurement is generally performed in the range of ± 80 degrees, the detection circuit 56 sets the calculated phase differences ΔΨ ₁ , ΔΨ ₂ , ΔΨ ₃ within ± Pmg / 4 according to the above equation (7). Convert to position offset amounts ΔX ₁ , ΔX ₂ , and ΔX ₃ . This offset amount ΔX _n represents an offset within ± Pmg / 4 of the grating mark MG with respect to the reference grating SG.

If the resolution of the phase difference measurement is about 0.2, the resolution of the offset amount is almost (0.2 / 180) Pmg / 4, and if the pitch Pmg is 4 µm, about 0.002 µm (2 nm) is obtained as a practical range.

On the other hand, the signal amplitude and amplitude ratio detection circuit unit 58 carries out respective waveform data such as the 39th to 39th degrees stored in the three RAM regions MnA (n = 1, 2, 3) in the waveform memory circuit unit 54. It is read out, and detected by the amplitude value of the signals _{I mn E 1, E 2,} E 3 to a digital operation.

In addition, the detection circuit unit 58 has amplitude values A _l of the photoelectric signals I _m1 , I _m2 , and I _m3 obtained when the interference beam BM generated from the grating of the reference mark plate FG is received by each photoelectric element 36A in advance. , A ₂ and A ₃ are remembered.

That is, before measuring the lattice mark MG on the wafer W, the lattice mark of the reference mark plate FG is moved below the objective lens 22, and signals like the 39th, 39th, and 39th degrees from the photoelectric element 36A are transmitted. After the waveform is generated and stored in the waveform memory circuit unit 54, the amplitude detection circuit 50 detects and stores the amplitude values A ₁ , A ₂ , and A ₃ .

At this time, the stop position of the stage WST where the mark plate FG is detected is read out from the laser interferometer 44 and stored, and at the offset amount detecting circuit unit 56, the position offset amounts ΔX _b1 , ΔX _b2 , If ΔX _{b3 is} also obtained, it can be used as data at the time of baseline determination.

In addition, the base line referred to here refers to the mutual minute error when the position offset amounts ΔX _b1 , ΔX _b2 , and ΔX _b3 of the lattice marks on the mark plate FG measured for each wavelength differ only in extremely small amounts. it means. Originally, in the incident light system shown in FIG. 37, the position of the mark plate FG if the interference fringes of the respective wavelengths generated on the reference mark plate GF by the respective beams of the wavelengths λ ₁ , λ ₂ , and λ ₃ exactly match. The respective values of the offset amounts DELTA X _b1 , DELTA X _b2 , and DELTA X _b3 may completely coincide.

However, as a practical matter, when the resolution is about 2 nm, it is difficult to adjust the incident light system and the detection system so that the position offset amounts ΔX _b1 , ΔX _b2 , and ΔX _b3 coincide with this resolution. For this reason, it becomes residual as a mark plate the position offset amount measured by the FG _b1 X △, △ X _{b2, b3} X △ mutual difference of the offset of the alignment system unique illustrated in Figure 37 (base line error).

The base line error is the position offset amount ΔX _b1 where the position offset amounts ΔX ₁ , ΔX ₂ , and ΔX ₃ for each wavelength obtained by detecting the lattice mark MG on the wafer W are determined by the detection circuit 56. , ΔX _b2 , ΔX _b3 are compensated for by the correction calculation. As an example, in the apparatus of FIG. 37, since the interference beam B _ms obtained from the reference grating SG is also changed to any one of wavelengths λ ₁ , λ ₂ , and λ ₃ , the reference mark plate measured under any one wavelength, for example, wavelength λ ₁ . Based on the position offset amount ΔX _b1 of the FG, the respective values of ΔX _b2 ΔX _b1 = ΔXb ₂₁ and ΔX _b3 ΔXb ₁ = ΔXb ₃₁ are calculated and stored.

And △ X ₂ with respect to the wafer W grating marks MG at a position offset amount _{△ X 1, △ X 2,} △ X 3 measurements for on-calculating △ X ₁ = so that the △ X ₂₁ corrects the value of △ X _2, The value of DELTA X ₃ may be corrected and calculated so that DELTA X ₂ -DELTA X ₁ = DELTA X _b31 .

Alternatively, as a more concise method, the position offset amounts ΔXb ₁ , ΔXb ₂ , and ΔXb ₃ of the reference mark plate FG determined according to the switching for each wavelength of the interference beam B _ms are stored, and the lattice mark MG on the measured wafer is stored. the amount of position offset _{△ X 1, △ X 2,} △ X 3 a _{△ X 1 - △ X b1,} △ X 2 - △ X b2, △ X 3 - may be calculated as the correction △ X _b3.

In addition, the amplitude ratio detection circuit unit 58 measures the ratio C of the amplitude values E ₁ , E ₂ , and E ₃ obtained when the amplitude values A ₁ , A ₂ , A ₃ , which are stored in advance, and the lattice mark MG on the wafer W are detected. ₁ , C ₂ , C ₃ are calculated as C ₁ = E ₁ / A ₁ , C ₂ = E ₂ / A ₂ , C ₃ = E ₃ / A ₃ . This ratio C ₁ , C ₂ C ₃ corresponds to the weighting coefficient set in the example of FIG. 25.

The data of the position offset amounts ΔX ₁ , ΔX ₂ , ΔX ₃ and the ratios C ₁ , C ₂ , C ₃ obtained as described above are sent to the weighted averaging calculation circuit unit 60, where the weighted grid mark MG The offset amount ΔX of is calculated. This operation is performed according to the following equation.

ΔX = (C ₁ · ΔX ₁ + C ₂ · ΔX ₂ + C ₃ · ΔX ₃ ) / (C ₁ + C ₂ + C ₃ )

The offset amount ΔX thus obtained is an offset in the pitch direction of the grating mark MG with respect to the reference grating SG. When this data is sent to the position control and the indicator 62, the wafer W is aligned in real time (positioning). Is also sent to the servo control circuit unit 64.

This servo control circuit unit 64 has two functions, one of which is a function (feedback servo mode) for controlling the feedback of the drive source 42 until the offset amount ΔX becomes a predetermined value. In the case of this function, the operations of the A / D converter circuit 50, the memory circuit unit 54, the offset amount detection circuit unit 56, and the averaging circuit unit 60 are connected to the command signal CQ from the position controller 62. In response, it is repeated one after another, and the value of the offset amount [Delta] X is calculated every extremely short time (for example, several msec).

In addition, the amplitude detection circuit unit 58 ratio C _1, C _2, C ₃ of the calculation by the good manyirado one of the first, may be carried out every time for each calculation of the offset amount △ X. Ratio C _1, C _2, that is, if every time of performing the calculation of the C ₃ every calculation of the offset amount △ X by the weighted averaging circuit unit 60 is non-C _1, the value of C _2, C ₃ if some change Of course. In the case where the calculation of the ratios C ₁ , C ₂ , and C ₃ is made only once or plural times, the same ratio value is used during the detection of the same lattice mark MG thereafter.

On the other hand, another function of the servo control circuit unit 64 is a function (interferometer servo mode) for moving the wafer stage WST based on the measured value of the laser interferometer 44. This function is, for example, positioning a lattice of the reference mark plate FG on the stage WST or a lattice mark MG on the wafer W directly below the objective lens 22, or any position on the wafer W based on the position of the detected lattice mark MG. Is used when positioning a point just below the objective lens 22.

The position control and indicator 62 also has a function of displaying the coordinate position of the grid mark MG and the obtained offset amount ΔX in addition to the above-described command to switch the output servo mode of the command signal CQ. In some cases, the values of the ratios C ₁ , C ₂ , and C ₃ serving as weighting coefficients when the grid mark MG is detected are also stored and stored. In this case, the same lattice mark MG is formed at a plurality of positions on the wafer W, and when the positions of these marks MG are sequentially stored, the marks of any part on the wafer W are also stored if the ratios C ₁ , C ₂ , and C ₃ are stored sequentially. It is possible to verify whether MG has a problem due to asymmetry or nonuniformity of the resist layer.

The portion where the weighting factors (ratios C ₁ , C ₂ , C ₃ ) are greatly changed may be graphically displayed on the wafer W. FIG. At this time, if the wafer before applying the resist layer is mounted in the apparatus of FIG. 8 through a thermochemical process such as a diffusion process or an etching process to obtain a change in weighting coefficient, the influence on the wafer surface caused by the thermochemical process is indirectly investigated. You may. In addition, by applying a resist layer to the wafer to obtain a change in the weighting coefficient, the effect of the resist layer can be indirectly investigated as compared with the change in the weighting coefficient before application.

In the above fifteenth embodiment, the reference mark plate FG is provided on the stage WST, and the signal amplitude change rate for each wavelength, i.e., the ratio C ₁ , C ₂ , C ₃ , is obtained using this. It is not necessary to provide a photoelectric element for directly detecting the light intensity of the beams B _r1 and B _r2 of a part of the incident light beams LB ₁ , LB ₂ , and LB ₃ .

However, when the reference mark plate FG having a chrome surface of a known reflectance is fixed on the wafer stage WST, as mentioned above, the mark plate FG can be used for measurement of various baseline quantities, measurement of focus state or measurement of focus state. Can be. Therefore, the position detection apparatus according to the fifteenth embodiment shown in FIG. 37 can be applied to the projection exposure apparatus that requires measurement of the baseline amount in the same manner as shown in FIGS. 15 and 16 above.

By the way, when the heterodyne position detection device of the 37th degree is mounted on the alignment system TTLA shown in FIG. 16, the incident light beam ± LF is time-divisionally divided into a plurality of wavelength components (offset about 30 to 40 nm from each other). When switched, the intersecting area of the beam ± LF irradiated on the wafer due to the influence of chromatic aberration (axial and magnification) of the projection lens PL or the chromatic aberration of the objective lens OBJ subtly in the Z direction or the XY direction for each wavelength component There may be a shift.

Therefore, in the 17A and 17B, the correction optical system CG may be provided to correct an error occurring in accordance with the chromatic aberration in the optical path of the incident light beam ± LF.

40 shows the configuration of the position detection apparatus according to the sixteenth embodiment of the present invention for switching the illumination beam to time division, the basic configuration of which is similar to the twenty-first degree, and accordingly the same reference numeral to the twenty-first degree and the member of the same function. Attach the. Here, the case where the relative position offset amount of the pitch direction (it is X direction) between two diffraction gratings RG and MG is measured by the homodyne system is illustrated. The beams LB ₁ and LB ₂ as illumination beams are respectively emitted from different laser light sources at different wavelengths λ ₁ and λ ₂ , are adjusted to be coaxial, and then converted to time division to be beam-splitter BS, mirror MR ₁ to the grating RG. Irradiated vertically.

The beam splitter BS amplitude-divides the beams B _r1 , B _r2 of a portion (a few%) of the beams LB ₁ , LB ₂ to guide the photoelectric element DT ₁ . When the photoelectric element DT ₁ receives beam B _r1 having a wavelength λ ₁ by irradiation with the cast name beams LB ₁ and LB ₂ , the photoelectric element DT ₁ outputs a photoelectric signal I _r1 indicating the intensity value and is irradiated with the cast name beam LB ₂ . When the beam B _r2 of wavelength lambda ₂ is received by the light, the photoelectric signal I _r2 indicating the intensity value is output.

From the grating RG, a plurality of diffracted light beams are generated similarly to the twenty-first degree by irradiation of the beams LB ₁ and LB ₂ (parallel beams).

In FIG. 40, these diffraction beams are first-order diffraction beams + D ₁₁ , -D ₁₁ generated from LB ₁ of wavelength λ ₁ , and first-order diffraction beams + D ₁₂ , -D ₁₂ generated from beam LB ₂ of wavelength λ ₂ . And 0th order beam D ₀ . Of course, even higher order diffracted light is generated for each of the beams LB ₁ and LB ₂ of each wavelength, but only the first diffraction beam is shown here for the sake of simplicity.

Each diffraction beam enters the imaging optical system divided into the front-group lens system G ₁ and the rear-group lens system G ₂ in the same manner as in the twenty-first degree.

Then, the beam D ₀ from the grating RG is shielded by the small mirror MR ₂ disposed at the center of the Fourier transform surface (the pupil surface) EP, and it is supported to enter the rear lens group G ₂ . In addition, each primary diffraction beam is a parallel beam similarly to the beams LB ₁ and LB ₂ when exiting from the grating RG, but converges as a beam waste at the position of the Fourier transformed EP in the operation of the entire lens system G ₁ .

Here, when the pitch of the grating RG is set to Prg, the first-order diffraction beam ± D ₁₁ generated by the beam LB ₁ of the wavelength λ ₁ and the first-order diffraction beam ± D ₁₂ generated by the beam LB ₂ of the wavelength λ ₂ are the 21st. Similarly, parallel beams are superimposed on the reflective grating MG for measurement, which is formed on the object side in a concave-convex shape through the rear lens group G ₂ , respectively. At this time, the pitch direction of the grating MG also coincides with the X direction, and on the grating MG, the one-dimensional interference fringe of the wavelength λ ₁ due to the light beam interference of the first-order diffraction beam ± D ₁₁ (the pitch direction is the X-direction), or the first-order One of the one-dimensional interference fringes (the pitch direction is the X direction) of the wavelength λ ₂ due to the light beam interference of the diffraction beam ± D ₁₂ is generated.

At this time, since the light having the wavelength λ _{1 and} the light having the wavelength λ ₂ are different wavelengths, for example, even if _two beams of name beams LB ₁ and LB ₂ are irradiated at the same time, there is no interference between the first diffraction beams ± D ₁₁ and D _12. Does not occur. And importantly, the interference fringes of wavelength λ ₁ generated by the first diffraction beam ± D ₁₁ and the interference fringes of wavelength λ ₂ generated by the first diffraction beam ± D ₁₂ are exactly the same in pitch so that they are a single interference fringe. Appear as

The pitch Pif of the intensity distribution of this interference fringe is determined by the pitch Prg of the grating RG and the magnification M of the imaging optical systems G ₁ and G ₂ , and is represented by Pif = M · Prg / 2. For example, if the pitch Prg is 4 µm and the magnification M is 1/4 (the pattern size of the grid RG is reduced to 1/4 at the lattice MG side), the pitch Pif of the interference fringe is 0.5 µm. Here, when the pitch Pmg of the grating MG for measurement is determined by the relationship of Pmg = 2Pif, that is, the relationship of Pmg = M · Prg, the refraction light is generated from the grating MG as the incident light beam as the first diffraction beam ± D ₁₁ . .

For example, one re-diffracted light generated from the grating MG using the first diffraction beam + D ₁₁ as the incident light beam is the first diffracted light (wavelength λ ₁ ) traveling perpendicularly from the grating MG, and the first diffraction beam -D One re-diffracted light generated from grating MG as ₁₁ incident light beams is + 1st order diffracted light (wavelength λ ₁ ) running perpendicularly from grating MG. These first-order diffracted light of the wavelength lambda ₁ traveling vertically have interference intensities depending on the phase state of each other, and become the interference beam BM (B _m1 ) to reach the mirror MR ₂ .

In addition, although re-diffracted light is generated from the grating MG using the first diffraction beam ± D ₁₂ as the incident light beam, -first-order diffraction light generated from the grating MG by irradiation of the first diffraction beam + D ₁₂ (wavelength λ ₂ ) Travels perpendicularly from the grating MG, and the + 1st order diffracted light (wavelength λ ₂ ) generated from the grating MG by irradiation of the first diffraction beam -D ₁₂ also proceeds vertically from the grating MG. These first-order diffraction light beams of wavelength lambda ₂ running vertically also have interference intensities depending on phase states of each other, and reach the mirror MR ₂ as the interference beam BM (B _m2 ). I.e. interference beam BM is the main illumination beam LB _1, B _m1 interference beam having a wavelength λ ₁ in response to the switching of LB ₂ is any one of an interference beam having a wavelength λ ₂ B _m2. This interference beam BM is reflected by the mirror MR ₂ and reaches the photoelectric element DT ₀ through the lens system G ₃ constituting the photoelectric detection system.

The photoelectric element DT ₀ outputs the photoelectric signal output while the casting name beam LB ₁ is being irradiated to the circuit units CU ₁ and CU ₃ as the photoelectric signal I _ml at a level corresponding to the intensity of the interference beam B _m1 , and casting name beam LB. the ₂ outputs the photoelectric signal output while the signal is radiated as a photoelectric I _m2 of the level corresponding to the intensity of the interference beam B _m2 to the circuit unit CU CU ₂ and _4.

Incidentally, in FIG. 40, circuit units CU ₁ and CU ₂ are separately shown and circuit units CU ₃ and CU ₄ are separately shown in order to easily explain the function of the signal processing circuit. Since LB ₁ and LB ₂ are switched time-divisionally, either one of the circuit units CU ₁ and CU ₂ and any one of the circuit units CU ₃ and CU ₄ may be used.

The circuit unit CU ₁₁ obtains the ratio C ₁ of the signal I _t1 from the photoelectric element DT ₁ to the amplitude value of the photoelectric signal I _m1 by the calculation of I _m1 / I _r1 , and the circuit unit CU ₂ is obtained from the photoelectric element DT ₁ . The ratio C ₂ between the signal I _r2 and the amplitude value of the photoelectric signal I _m2 is obtained by calculation of I _m2 / I _r2 . The data of these ratios C ₁ and C ₂ are output to the circuit unit CU ₅ which calculates a weighted average similarly to the twenty-first degree.

In this embodiment, since the homodyne method is employed, the intensity of the interference beams B _m1 and B _m2 changes according to the change in the relative position of the gratings RG and MG in the X direction, for example, when the gratings RG and MG stop in a certain state. If present, the levels of signals I _m1 and I _m2 continue to take these constant values, respectively. Therefore, the interference fringes on the grid MG generated by the grid RG and the grid MG are relatively scanned in the X direction by a certain amount (more than the pitch Pif minutes of the interference fringes), and the level change of the electrostatic waveforms of the signals I _m1 and I _m2 generated during that time is _measured . The peak value and the bottom value in the sample are sampled, and the difference value is used as the amplitude value for the calculation of the circuit units CU ₁ and CU ₂ , respectively.

Here, since the level change of the signal I _m1 (same as I _m2 ) according to the change in the positional relationship between the interference fringe and the grating MG is the same as that described in FIGS. 22A to 22D, the description thereof is omitted.

The circuit units CU ₃ and CU ₄ are based on the amplitude values of I _m1 and I _m2 and the X-direction of the interference fringe and the grating MG based on a preset function or a conversion equation F (I _m2 ) or F (I _m2 ). The position offset amounts ΔX ₁ and ΔX ₂ are calculated. The position offset amounts ΔX ₁ , ΔX ₂ are ± Pmg / 4 from this, for example, based on one specific peak or bottom point of each signal I _m1 , I _m2 in FIG. 22D. It is calculated | required as a value within a range.

In this embodiment, since the cast name beams LB ₁ and LB ₂ are switched time-divisionally and irradiated, X of the grid mark MG (wafer) when the cast name beam LB ₁ is irradiated to detect the level change of the photoelectric signal I _m1 . The microscopic start position of the lattice mark MG (wafer) at the time of detecting the level change of the photoelectric signal I _m2 by irradiating the fine starting position of the direction and the casting name beam LB ₂ shall be made to match.

In addition, on the actual device configuration, the coordinate position of the movable stage for moving the grid mark MG is measured in order by a measuring device (laser interferometer, etc.) having a sufficiently high resolution than the grid pitch Pmg, and based on the measured value of the measuring device. Servo control of the movable stage is made to realize the fine starting position. Further, as a more preferable configuration, A / D conversion of the level change of each photoelectric signal I _m1 , I _m2 is performed in response to each of position pulses (for example, one pulse for every 0.02 μm movement) for position measurement from such a measuring device. In addition, waveform storage means in which the digital value of the converted level is sequentially stored, such as a waveform memory circuit (RAM) in which an address and a moving position are associated, is provided.

In this way, by simply reading the stored waveform data, the coordinate positions of the bottom and peak serving as reference values on the signal waveforms and amplitude values of the signals I _m1 and I _m2 are immediately obtained. However, the above-described functions or conversion equations F (I _m1 ) and F (I _m2 ) use sine or cosine functions because each signal I _m1 and I _m2 is sinusoidal. As an example, if the peak level of the signal I _m1 described above is E _p1 , the bottom level is E _b1 , and the level of the signal I _m1 for the detected position is e ₁ ,

(E _p1 + E _b1 ) / 2 + {(E _p1 -ES) sinΨ ₁ } / 2 = e ₁

By calculating the radial Ψ ₁ satisfying the following and substituting it into the following exchange equation using the value of the pitch Pmg, the offset amount ΔX ₁ from the reference point can be known.

ΔX ₁ = Pmg ・ Ψ ₁ / 4π

Similarly, the offset amount ΔX ₂ from the reference point using I _m2 is also calculated, and the data of these offset amounts ΔX ₁ , ΔX ₂ is transferred to the circuit unit CU ₅ which performs the weighted average operation, and the ratio C obtained above. _The following calculation is performed with ₁ and C ₂ as weighting coefficients.

ΔX = (C ₁ · ΔX ₁ + C ₂ · ΔX ₂ ) / (C ₁ + C ₂ )

The offset amount ΔX obtained in this calculation is the position offset amount with respect to the grid RG of the grid MG to be finally obtained.

As can be seen from this calculation formula, the offset amount ΔX is determined to give more weight to the side of the measurement result of the position offset amount using the interference beam of the wavelength component of the higher intensity in the interference beam BM. As described above, in the present embodiment, the beams LB ₁ and LB ₂ of two different wavelength components are used to irradiate the gratings Rg and MG, and the interference beams BM to be received are photoelectrically detected for each wavelength and the interference beams B for each wavelength. _Since the results of position offset detection using _m1 and B _m2 are individually weighted averaged according to the amplitude of the received light for each wavelength, a more reliable position detection result is obtained.

Further, in the optical exposure arrangement shown in FIG. 40, when the grating RG is a grating mark on the mask, the grating MG is a mark on the wafer, and the imaging systems G ₁ and G ₂ are transparent lenses on the wafer of the mask pattern, the projection exposure apparatus Alignment apparatus can be realized.

41 shows a schematic configuration according to the seventeenth embodiment, and since the basic configuration is the same as that of the twenty-third embodiment, the same reference numerals are assigned to the same members. 17 embodiment, the illumination two beams LB _1, through the LB ₂ lens system G ₄ and incident on the mirror MR ₂ disposed at the center of the pupil plane of the imaging optical system (G _1, G _2), down from the mirror MR ₂ The beams LB ₁ and LB ₂ which are bent by are time-divisionally switched and irradiated perpendicularly to the grating MG as parallel beams through the rear lens group G ₂ . Then, the first diffraction beam ± D ₁₁ of wavelength λ ₁ diffracted by grating MG or the first diffraction beam ± D ₁₂ of wavelength λ ₂ is crossed (imaged) on the grating RG through lens systems G ₁ and G ₂ . Since the grating RG is a transmission type, ± first-order diffraction light of the re-diffracted light generated from the grating RG by irradiation of the first-order diffraction beam ± D ₁₁ or ± D ₁₂ proceeds to the imaging optical system and the reflection side in a direction perpendicular to the grating RG. Through the mirror MR ₃ , the alignment objective lens G ₅ and the spatial filter 28, it becomes an interference beam BM ( _one of the interference beam B _m1 of the wavelength λ _{1 and} the interference beam Bm of the wavelength λ ₂ ) and is received by the photoelectric element DT ₀ . . Other configurations (photoelectric elements DT ₁ , circuit units CU ₁ , CU ₂ , CU ₃ , CU ₄ , CU ₅ ) are the same as those in the 40th degree.

In this way photoelectric device DT _0, and outputs either the photoelectric signal I _m1, I _m according to the time-division switching of the primary illumination beam LB _1, LB _2.

The present embodiment has a configuration in which the relationship between the incident light and the received light of the beam is opposite to that in FIG. 40, but this configuration forms the grating MG on the wafer, the grating RG on the reticle (mask), and the lens system G ₁ , G. ₂ is applicable to the apparatus of (F) JP-A-3-3224, which is a reduced projection lens for projection exposure of a reticle pattern. However, in the apparatus disclosed in the publication (F), a small lens that refracts the first diffraction beams ± D ₁₁ and ± D ₁₂ by a small amount is installed on the pupil surface E _p of the projection lens, and the chromatic aberration generated in the projection lens is corrected. However, when the embodiment of FIG. 41 is applied, a small lens (e.g., a chromatic dispersion is performed for optimal correction for each of two sets of first diffraction beams ± D ₁₁ and ± D ₁₂ having different wavelengths). It is necessary to install a large-print printing base material and an aspherical lens).

As described above, in the embodiment of FIG. 17, since the cast name beams LB ₁ and LB ₂ are configured to directly enter the lattice marks MG on the wafer, for example, the first-order diffraction beams ± D ₁₁ and ± D ₁₂ generated from the lattice marks MG. Each intensity can be higher than the intensity of the diffraction beam (interference beam BM) generated from the grating MG in FIG.

Next, an eighteenth embodiment of the present invention will be described below. In the present embodiment, as described above based on the configuration shown in FIG. 37, in addition to the interference beam of ± 1st order diffracted light from the grating mark, the interference beam (order difference) of the 0th order light and the second diffracted light from the grating mark. The structure which also detects the interference of two diffraction light) was added. Although a method of photoelectric conversion of an interference beam of zero-order light and second-order diffracted light into a single photoelectric device and detecting the position offset of the grating mark using the photoelectric signal has been attempted, the incident light beam for grating mark illumination is multiplied. When the interference beam (which is multi-wavelength) of the 0th order light and the 2nd order light is received in a single phase with a single photoelectric device, it is difficult to detect a good position offset as it is.

The reason for this great reason is that the phase difference between the three photoelectric signals IK02 _n (n = 1, 2, 3) is ± 1st order diffracted light as described with reference to FIGS. 29A to 29D and 30A to 30D. This is because the phase difference is large compared with the phase difference of the photoelectric signal I _mn (FIGS. 39A to 39C) in the case of an interference beam.

For this reason, when the light intensity change for each wavelength which originally has a large phase difference is received by a single photoelectric element, the amplitude of the photoelectric signal (the amplitude of an alternating current) will become extremely small by the effect of canceling the intensity of each wavelength.

Thus, the configuration of this embodiment will be described with reference to FIG. FIG. 42 is a part of the configuration of FIG. 37, specifically, a light spot detection system of various interference beams from the grating mark MG. Therefore, the same reference numerals are given to members having the same function as those in FIG. Incident light system 100 in FIG. 42 includes light sources LS ₁ , LS ₂ , LS ₃ , mirror MR, mirror DCM ₄ , DCM ₅ , radial grating RRG as a frequency shifter, lens 10, and spatial filter shown in FIG. 37. (12) and adjustment optical systems 14, 16, 18, etc., and emits a pair of incident light beams + LF, -LF.

The light transmission beam ± LF which is sequentially switched to any one of the wavelengths λ ₁ , λ ₂ , and λ ₃ is partially reflected by the half mirror 20 and is incident on the objective lens 22, and a part is incident on the reference light receiving system 110. do. The reference light receiving system 110 is composed of the adjustment optical system 24A. 24B, the lens 26, the reference grating SG, and the spatial filter 38 in FIG. 37, and guide light B _ms is directed to the photoelectric element 40. FIG. . Then, when the lattice mark MG on the wafer W is irradiated by the optical transmission beam ± LF through the objective lens 22, the interference beam BM of the ± first-order diffraction light is generated vertically and is zero in the direction of travel of each optical transmission beam. An interference beam BM of secondary-secondary light, and BM ₂₀ occurs. The interference beam BM of the ± 1st order diffracted light and the interference beams BM ₀₂ , BM ₂₀ of the 0th order and 2nd order light are reflected from the mirror 32 through the objective lens 22 and the half mirror 20, and the interference beam BM is photoelectric. The element DT ₀ is received and the interference beams BM ₀₂ and BM ₂₀ are received by the photoelectric elements DT _2a and DT _2b , respectively.

In response to the conversion of the incident light beam ± LF to each wavelength component, the interference beam M is _one of the interference beam B _m1 of the wavelength λ _1, the interference beam B _m2 of the wavelength λ _{2, and} the interference beam B _m3 of the wavelength λ ₃ . do. Similarly, the interference beams BM ₀₂ and BM _{20 of the} 0th-order secondary light also have any one of three wavelengths λ ₁ , λ ₂ , and λ ₃ in response to the wavelength switching of the incident light beam ± LF.

Each opto-electronic device also DT _0, DT _{2a, 2b} DT is when placed in the Fourier transform plane, or near the surface of the objective lens 22, the 0th-order-secondary light interference beam BM _02, BM ₂₀ is photoelectric each wavelength The light is offset horizontally on the elements DT _2a and DT _2b . Therefore, the light-receiving surface of each photoelectric element _2a DT, DT _2b is sized so that the horizontal offset estimated. In addition, when a spatial filter is provided immediately before each photoelectric element DT ₀ , DT _2a and DT _2b , it is also necessary to determine the size of the opening for beam selection in consideration of this horizontal offset. Alternatively, a prism (or flat-parallel glass) made of a material having a large color dispersion may be provided immediately before each of the photoelectric elements DT _2a and DT _2b to reduce the horizontal offset due to each wavelength on the light receiving surface. .

As can be seen from the above configuration, in the present embodiment, the photoelectric signal I _ms from the photoelectric element 40 is used as a reference signal, and the photoelectric signals I _mn , IK02 _n , IK20 from the respective photoelectric elements DT ₀ , DT _2a , DT _2b . _The signal processing circuit which calculates each phase difference of _n (n = 1, 2, 3) is needed. An example of the simplest circuit configuration for this purpose is shown in FIG.

43 is an improvement of a part of the processing circuit shown in FIG. 38. On the hardware, the converter circuit unit 50 of FIG. 38 is divided into four channels for the A / D conversion IC circuits ADCa, ADCb, ADCc, and ADCd. in configuration and, with reference to each channel signal Ims, measurement signal I _mn, IK02 _n, by applying a respective IK20 _n, the pulse signals from the four almost the signal at the same time generating a sampling clock of claim 38, a circuit (52) C _The difference is that digital sampling is possible in response to the _PS .

In addition, the waveform memory circuit unit 54 is a four-channel memory back for simultaneously storing signal waveform data from each of the A / D conversion IC circuits ADCa, ADCb, ADCc, and ADCd as shown in FIG. 44 in this embodiment. MM _an , MM _bn , MM _cn , MM _dn , and memory areas a ₁ to a ₃ , b ₁ to b ₃ , c ₁ to c ₃ , for the number of wavelengths (three here) for each bag; d ₁ to d ₃ are prepared.

In addition, the waveform memory circuit unit 54 is capable of recording in each memory back MM _an , MM _bn , MM _cn , and MM _dn so as to cooperate with the wavelength switching of the incident light beam ± LF in response to the command signal CQ from the position controller 62. The memory areas to be switched are sequentially switched. In addition, the address counters of the twelve memory areas a ₁ to a ₃ , b ₁ to b ₃ , c ₁ to c ₃ , and d ₁ to d ₃ are commonly updated in response to the pulse signal C _ps , but are simultaneously written. This is done only in four memory areas in response to the command signal CQ.

That is, between the incident light beam ± LF and the wavelength λ ₁ , the digital waveform data of the signal I _mn is in the memory region a ₁ , the digital waveform data of the signal IK02 _n is in the memory region b ₁ , and the digital waveform data of the signal IK20 _n is the memory region. the c _1, and the digital waveform data of the reference signal I _ms (λ ₁₎ is stored in the memory area ₁ d.

Similarly, between the incident light beam ± LF and the wavelength λ ₂ , the digital waveform data of the signal I _mn is in the memory region a ₂ , and the digital waveform data of the signal IK02 _n is in the memory region b ₂ , and the digital waveform data of the signal IK20 _n is the memory region c. ₂ and the digital waveform data of the reference signal I _ms (λ ₂ ) are stored in the memory region d ₂ , while the incident light beam ± LF is between the wavelength λ ₃ , the digital waveform data of the signal 1 _mn is stored in the memory region a ₃ , The digital waveform data of the signal IK02 _n is stored in the memory area b ₃ , the digital waveform data of the signal IK20 _n is stored in the memory area c ₃ , and the digital waveform data of the reference signal I _ms (λ ₃ ) is stored in the memory area d ₃ .

In addition, the amplitude detection and amplitude ratio detection circuit 58 shown in FIG. 38 has data of ratios C _n1 , C _n2 , C _n3 grouped for each interference beam having different diffraction states in FIG. 43 (n is n corresponding to a wavelength). = 1, 2, 3). Of the data of this ratio, Cn ₁ (n = 1, 2, 3) is the same as the ratio C ₁ , C ₂ , C _{3 in} FIG. 38, and C _n2 (n = 1, 2, 3) is the photoelectric signal IK02 _n. The ratio for each wavelength obtained from (n = 1, 2, 3), and C _n3 (n = 1, 2, 3) is the ratio for each wavelength obtained from the photoelectric signal IKL20 _n (n = 1, 2, 3). to be.

In addition, the phase difference and position offset detection circuit 56 shown in FIG. 38 shows offset amounts ΔX _n1 , ΔX _n2 , ΔX _n3 (n = 1, 2, 3) to output. _{ΔX n1} (n = 1, 2, 3) in the offset amount is the same as the offset amounts ΔX ₁ , ΔX ₂ , ΔX _{3 in} FIG. 38, and _{ΔX n1} (n = 1, 2, 3) Is the offset amount for each wavelength component obtained from the photoelectric signal IK02 _n (n = 1, 2, 3), and ΔX _n3 (n = 1, 2, 3) is the photoelectric signal IK02 _n (n = 1, 2, 3). The offset amount for each wavelength component obtained from the In addition, the detection circuit 56 is configured according to the phase differences _{Δβ 0n} , _{Δβ 1n} , _{Δβ 2n} (n = 1, 2, 3) as described in FIGS. 29A to 29D and 30A to 30D. The value is calculated in the middle.

In addition, the weighted averaging circuit 60 in FIG. 38 is changed to a selective weighted averaging circuit in FIG. 43, and the final position offset amount is set to? X based only on the photoelectric detection result of the interference beam BM of ± 1st order light. The first operation mode, which is the same as the 38th degree to calculate, the second operation mode which calculates the final offset amount ΔX based only on the photoelectric detection progress of the interference beam of the 0th-second order light, and the final based on the result of all the interference beam photoelectric detection And a third operation mode for calculating the typical offset amount ΔX. These three operation modes can be appropriately selected by the operator, but when the third operation mode is designated, two to three calculation algorithms can be selected. Such mode designation, algorithm, and selection will be described later in detail.

Also in this embodiment, first, positioning of the stage WST is performed in accordance with the interferometer 44 so that the lattice mark of the mark plate FG on the wafer stage WST is first irradiated with the incident light beam ± LF from the objective lens 22.

Thereafter, in response to the command signal CQ from the position controller 62 (FIG. 38), the three light sources LS ₁ , LS ₂ , LS ₃ have a predetermined time (for example, about 100 times the period of the bit scan number 2Δf). The light is sequentially turned on every time, and the lattice mark of the reference mark plate FG is irradiated by the incident light beam ± LF which is sequentially wavelength-converted. Thus, for example, if the three light sources LS ₁ , LS ₂ , LS ₃ are switched on in order of shorter wavelengths, they are _first output from the photoelectric element DT ₀ while the light source LS ₁ (wavelength λ ₁ ) is turned on. The digital waveform data of the signal _Imn (n = 1) is stored in the memory area a ₁ in FIG. 44 in response to the sampling pulse signal C _PS .

At the same time, the signal digital waveform data of the IK02 _{n (n} = 1) from the photoelectric device DT ₂ in response to the pulse signal C _PS is stored in the memory area b _1, signal IK20 _{n (n} = 1) from the photoelectric device DT _2b The digital waveform data of is stored in the memory area c ₁ in response to the pulse signal C _PS , and the digital waveform data of the signal I _ms (wavelength λ ₁ ) from the photoelectric element 40 is stored in the memory area in response to the pulse signal C _PS . It is stored in d ₁ .

In the same manner, the memory regions a _n corresponding to the signals 1 _mn , IK02 _n , IK20 _n , and I _ms (n = 2, 3) from the respective photoelectric elements while each of the light sources LS ₂ and LS ₃ are lit. , b _n , c _n , and d _n (n = 2, 3), respectively.

Next, each waveform data in the memory back MM _bn of the memory circuit unit 54 is analyzed by the amplitude detection circuit 58, and the amplitude value (peak to peak) of the signal IK02 _n for each wavelength is determined by J02 _n ( It calculates as n = 1, 2, 3), and stores it. Similarly, the amplitude detection circuit 58 analyzes each waveform data in the memory back MM _cn and sets the amplitude value (peak to peak) of the signal IK20 _n (n = 1, 2, 3) for each wavelength to J20 _n. It calculates and stores as (n = 1, 2, 3), analyzes each waveform data in the memory back MM _an , obtains and stores the amplitude value J11 _n (n = 1, 2, 3) of the signal I _mn for each wavelength. do.

Since the preliminary operation is completed by the above, the next stage WST to be actually positioned is placed on the stage AST, and the stage WST so that the lattice mark MG on the wafer W is irradiated by the incident light beam ± LF from the objective lens 22. Position it.

The three light sources LS ₁ , LS ₂ , and LS ₃ are sequentially switched on and lighted in the same manner as in the detection of the lattice mark of the reference mark plate FG, and the respective photoelectric signals I _mn , IK02 _n , and IK20 _n (n = 1, 2). 3), each waveform data of I _ms is simultaneously put into the memory circuit unit 54. After that, the amplitudes of the respective signals of the signals I _mn , IK02 _n , IK20 _n , (n = 1, 2, 3) stored in the respective memory regions a _n , b _n , c _n of the memory circuit unit 54. The values are calculated by the detection circuit 58 as E _n (see Figs. 39A to 39D), E02 _n and E20 _n (Figs. 29A to 29D and 30A to 30D), respectively.

On the other hand, the phase difference and phase offset detection circuit unit 56 is stored in each of the memory regions a _n , b _n , and c _n of the memory circuit unit 54, and the signals I _mn , IK02 _n , and IK20 _n (n = 1, 2, 3) for each wavelength are read for each signal I _mn, IK02 _n, IK20 _n reference signal phase for the _{I ms △ Ψ n, △ β} 0n, △ β 2n (n = 1, 2, 3) and a position offset Calculate the amounts ΔX _n1 , ΔX _n2 , and ΔX _n3 (n = 1, 2, 3) sequentially.

In this way, when the amplitude value and the position offset amount for each wavelength are obtained for each detection light (interference beam) having a different diffraction state, the amplitude ratio detection circuit 58 performs the following calculation.

(A1) C ₁₁ = E ₁ / J11 ₁

(A2) C ₂₁ = E ₂ / J11 ₂

(A3) C ₃₁ = E ₃ / J11 ₃

(B1) C ₁₂ = E02 ₁ / J02 ₁

(B2) C ₂₂ = E02 ₂ / J02 ₂

(B3) C ₃₂ = E02 ₃ / J02 ₃

(C1) C ₁₃ = E20 ₁ / J20 ₁

(C2) C ₂₃ = E20 ₂ / J20 ₂

(C3) C ₃₃ = E20 ₃ / J20 ₃

Next, the most likely offset amount [Delta] X is calculated by the averaging circuit unit 60, but in the first calculation mode using only the interference beam BM of ± 1st order light, it is the same as in the case of FIG.

ΔX = (C ₁₁ · ΔX ₁₁ + C ₂₁ · ΔX ₂₁ + C ₃₁ · ΔX ₃₁ ) / (C ₁₁ + C ₂₁ + C ₃₁ ).

On the other hand, in the second operation mode using only the interference beams of the 0th-secondary light, the phase difference _{Δβ 0n} obtained by the detection of the interference beam of the 0th-secondary light generated on the left side of the interference beam BM of the ± 1st light and the interference beam An algorithm for calculating the position offset amount for each wavelength is adopted from the average phase difference with the phase difference Δβ _2n obtained in accordance with the detection of the interference beam of the 0th-order secondary light generated on the right side of the BM. The averaging of the phase difference is different from the averaging for the purpose of reducing the so-called random components to improve the accuracy, and is an averaging that should be performed in principle when performing position detection using an interference beam of 0th order light and ± 2nd order light. .

So in this embodiment, the algorithm as a base, and averaging circuit unit 60 first, the signal for each position offset amount _{△ X n2 (n = 1,} 2, 3) and the signal of each location offset obtained from IK20 _n obtained from IK02 _n The average value ΔXA _n (n = 1, 2, 3) for each wavelength with the amount ΔX _n3 (n = 1, 2, 3) is calculated as follows.

ΔXA ₁ = (ΔX ₁₂ + ΔX ₁₃ ) / 2

ΔXA ₂ = (ΔX ₂₂ + ΔX ₂₃ ) / 2

ΔXA ₃ = (ΔX ₃₂ + ΔX ₃₃ ) / 2

In addition, the averaging circuit unit 60 has an average value CA _n (n = 1, 2, 3) for each wavelength component of the amplitude ratios C _n2 , C _n3 , of the interference beam of the 0th-order secondary light obtained by the amplitude ratio detection circuit unit 58. ) Is calculated as follows.

CA ₁ = (C ₁₂ + C ₁₃ ) / 2

CA ₂ = (C ₂₂ + C ₂₃ ) / 2

CA ₃ = (C ₃₂ + C ₃₃ ) / 2

Then, the averaging circuit unit 60 is the average ratio as a weighting factor CAn average position offset amount △ XA _n weighted mean most accurate offset amount △ X to as the follows for each of the wavelength components of each wavelength component Calculate

ΔX = (CA ₁ , ΔXA ₁ + CA ₂ , ΔXA ₂ + CA ₃ , ΔXA ₃ ) / (CA ₁ + CA ₂ + CA ₃ )

By the above, the position or position offset detection of the grid mark MG by a 2nd operation mode is achieved.

In the third operation mode, either one of the first algorithm that simply averages the position offset amount calculated in the first operation mode and the second operation mode position offset amount, and the second algorithm that weights the two position offset amounts may be selected. It is possible to set in advance by the operator. Therefore, the position offset amount finally calculated in the first operation mode (the mode using the result of the detection of the interference beam of the ± 1st order light) is ΔXM ₁ , and the position offset amount finally calculated in the second operation mode is ΔXM. _{If it is 2} , the position offset amount determined by the second algorithm is calculated as (ΔXM ₁ + ΔXM ₂ ) / 2.

On the other hand, in the second algorithm, the offset amount ΔXM ₁ calculated in the first calculation mode and ΔXM ₂ calculated in the second calculation mode are weighted average using predetermined weighting factors Q ₁ and Q ₂ . As an example, the weighting factor Q1 is the amplitude values E ₁ , E ₂ , E ₃ of the signals I _mn (n = 1, 2, 3) obtained by photoelectric detection of the interference beam BM of ± 1st order light (Figs. 39A to Corresponding to the sum of Fig. 39D), the weighting factor Q ₂ is the average amplitude value for each wavelength of the signals IK02 _n and IK20 _n (n = 1, 2, 3) obtained by photoelectric detection of an interference beam of 0-second order light. Corresponds to the logarithm of (E02 ₁ + E20 ₁ ) / 2, (E02 ₂ + E20 ₂ ) / 2, and (E02 ₃ + E20 ₃ ) / 2. Therefore, in the second algorithm, the offset amount? X of the grid mark MG is determined by the following calculation.

ΔX = (Q ₁ · △ XM ₁ + Q ₂ · △ XM ₂ /) / (Q ₁ + Q ₂ )

In principle, the higher the diffracted light, the smaller the light intensity, so that the light intensity amplitude of the interference beam of 0-second order light (corresponding to E _n ) is higher than that of the interference beam BM of ± 1st order light. Corresponding to E02 _n and E20 _n ) becomes considerably small. Thus, simply discard signal I _{mn, n} IK02, it increased if I to determine the weighting factors Q _1, Q ₂ in the sum of the amplitudes of the IK20 _n only, the coefficient Q ₂ side than the most part the weighting factors Q _1. Here, the coefficient Q ₂ is preferably corrected so as to increase the calculated value by, for example, a predetermined ratio (10 to 30% as an example).

Next, a nineteenth embodiment of the present invention will be described with reference to FIG. In this embodiment, basically the same as the thirty-third degree, the structure of the reference mark plate FG on the wafer stage WST shown in FIG. 37 is changed to a transmissive lattice mark (lattice without asymmetry in amplitude transmittance), The denominator (reference value) used when calculating the amplitude ratios of the photoelectric signals I _mn , IK02 _n , and IK20 _n to the detection circuit unit 58 by ray detection of the interference beam transmitted through the lattice mark was determined. Therefore, the same code | symbol is attached | subjected to the same thing as 33rd degree in each member in FIG.

45 shows a partial cross-sectional view of the wafer stage WST, and when the incident light beam LF (here, the wavelength of the wavelengths λ ₁ and λ ₂ ) is irradiated by sequentially switching the lattice marks on the reference mark plate FG for each wavelength component From this grating mark, 0th order light, ± 1st order light, ± 2nd order light generate toward the stage interior.

These diffracted light is bent at right angles in the mirror MR and incident on the lens system G ₅ having a Fourier transform function, and the interference beam Bmrn (n = 1, 2) by ± first-order diffracted light generated perpendicularly from the reference mark plate FG And the interference beams ± B1 _r (wavelength λ ₁ ) and ± B2 _r (wavelength λ ₂ ) by the 0th-order second-order diffracted light and enter the photoelectric element group DTR.

The photoelectric element group DTR receives the interference beam B _mrn and outputs an AC photoelectric signal I _mrn according to the bit frequency, and the interference beams + B1 _r (wavelength λ ₁ ) and + B _2r (wavelength λ ₂ ). A light receiving unit that receives the light in common and outputs the photoelectric signal IR20 _n of the alternating current according to the bit frequency, and receives the interference beams -B _1r (wavelength λ ₁ ) and -B _2r (wavelength λ ₂ ) in common, It consists of a light-receiving unit that outputs a photoelectric signal IR02 _n.

Therefore, the incident light beam with a wavelength λ ₁ ± LF for the wavelength λ ₁ 0 car ± 1-order diffracted light beam B _mr1 interference due to an interference beam and ± B _1r wavelength λ ₁ of the light-shielding -2 photoelectric device according to the DTR group is reached, and the incident light beam with a wavelength λ ₂ of ± LF is for a wavelength of the interference beam of the ± 1 light shielding by the zero-order beam and interference of the light shielding -2 ± B _2r and wavelength λ ₂ λ ₂ due to the photoelectric element group B _mr2 Reach DTR. For this reason, when the wavelength λ ₁ becomes the photoelectric signal I _mr1, photoelectric signals IR02 _1, IR20 ₁ obtained from each light-receiving unit, the wavelength λ ₂ days when is a photoelectric signal I _mr2, photoelectric signals IR02 _2, IR20 ₂ obtained from the respective light-receiving units .

In the case of the heterodyne system, these photoelectric signals are represented by sinusoidal waveforms of the same frequency as the bit frequency, and the photoelectric signals I _mrn , IR02 _n , and IR20 _n represent the A / D converter circuit unit 50 shown in FIG. 43. ) Are input to the input signals I _mn , IK02 _n , and IK20 _n respectively.

Specifically, a switch for switching the signal I _mrn and the signal I _mn in the circuit of FIG. 43 and inputting it to the A / D converter circuit 50, and switching the signal IR02 _n and the signal IK20 _n to input the A / D converter circuit 50. And a switch for switching the signal IR02 _n and the signal IK20 _n to be input to the A / D converter circuit 50, and interlocking these three switches with the command signal from the position controller 62 (Fig. 38). It may be configured to switch in response to the response.

The photoelectric signals from these photoelectric element groups DTR are temporarily stored in the waveform memory circuit unit 54, and then each amplitude value of these photoelectric signals is obtained and stored by the amplitude detection circuit unit 58 in FIG. . When the amplitude ratio is obtained, the following calculation is performed, for example.

C ₁₁ = I _m1 / I _mr1

C ₂₁ = I _m2 / I _mr2

C ₁₂ = IK02 ₁ / IR02 ₁

C ₂₂ = IK02 ₂ / IR02 ₂

C ₁₃ = IK20 ₁ / IR20 ₁

C ₂₃ = IK20 ₂ / IR20 ₂

In this embodiment, since the photoelectric detection of the diffracted light transmitted through the reference mark plate FG is performed by the photoelectric element group DTR, the phase information of each photoelectric signal obtained from the element group DTR and the phase information of the photoelectric signal I _ms as a reference signal are obtained. By comparing with the above, the position offset of the reference mark plate FG or the measurement of the position, that is, the operation of a part of the baseline measurement can be combined.

Next, a twentieth embodiment of the present invention will be described with reference to FIG. This embodiment is basically the same as the eighteenth embodiment, and a pair of incident light beams + LF for irradiating the lattice mark MG for measurement (alignment) on the wafer W (or reference mark plate FG) through the objective lens 22. And the polarization direction of -LF are complementary relations. That is, in the case of linearly polarized light, the polarization directions of the incident light beams + LF and -LF are orthogonal to each other, and in the case of circularly polarized light, the incident light beams + LF and -LF are set to reverse rotation polarizations. For this reason, the two incident light beams ± LF do not interfere with each other, and the ± first-order diffraction light BM of the wavelengths λ ₁ , λ ₂ , and λ ₃ vertically generated from the grating mark MG does not interfere with each other.

Therefore, when the photoelectric detecting the ± 1-order diffracted light BM through the objective lens 22, a small mirror MR _2, uses a polarizing beam splitter PBS as the analyzer (analyzer). In this way, the ± 1st order light BM transmitted through the polarization beam splitter PBS interferes with each other to become the first interference beam BP ₁ , and the ± 1st order light BM reflected from the polarization beam splitter PBS interferes with each other to interfere with the second interference beam BP _2. It becomes

These interference beams BP ₁ and BP ₂ are complementary to each other, but each interference beam is intensity modulated into a sinusoidal shape in accordance with the bit frequency if it is a heterodyne system. In addition, the phases of the intensity modulation of the interference beams BP ₁ and BP ₂ are exactly different by 180 °.

In addition, the half wave plate (HW) shown in the figure has a different linear polarization direction of the incident light beam ± LF and the first-order diffraction light BM from the polarization separation direction of the polarization beam splitter PBS (rotating). In this case, it is provided for the purpose of correcting the linear polarization direction between the ± first order diffracted light BM. For this reason, when the linear polarization directions orthogonal to each other between ± 1st-order diffraction light BM coincide with the polarization separation direction of the polarization beam splitter PBS from the beginning, or when the incident light beams + LF and -LF become circularly polarized light of reverse rotation, It is not necessary to use / 2 wave plate HW.

Thus, in the present embodiment, the interference beam BP _{1 is configured} to receive the photoelectric element DT0 _a : 36A _{1 in} FIG. 18, and the interference beam BP _{2 is configured} to receive the photoelectric element DTO _{b in} 36A ₂ of FIG. 18. It was in a configuration to obtain a photoelectric signal I _mn by subtracting the amount of the photoelectric signal of the DTO, and _a DTO _b to the differential amplifier. In the case of using such a differential amplifier, since the photoelectric signal of the photoelectric element DTO _{a and} the photoelectric signal of the photoelectric element DTO _{b are out} of phase with each other (180 ° difference), the common-phase noise components included in both signals are subtracted. This is because it is canceled and the actual S / N ratio of the signal I _mn is improved.

By the way, the first 37, second 42 degree, 46 degree by the objective lens 22 are wavelength used Station some extent with respect to at least the axial chromatic aberration in the various types of chromatic aberration generated in the (λ ₁ to λ ₃₎ shown in It is desirable that it is corrected. For example, when the band of wavelengths λ ₁ to λ ₃ to be used is 100 nm or less, such axial chromatic aberration is selected by selecting materials of a plurality of lens elements constituting the objective lens 22 and combining lens elements of different refractive indices and dispersion ratios. Some correction is possible. Of course, the chromatic aberration does not need to be corrected completely by the objective lens 22, but can be corrected by the adjustment optical systems 14, 16 and 18 shown in FIG.

In the above, each Example of 15th-20th of this invention was described. When grating mark MG on the wafer W or reference mark plate FG is detected by the homodyne system, it is necessary to prescan the grating mark MG in the pitch direction to sample the level change of each photoelectric signal. In this case, the simplest method is to measure the clock signal C _PS for the signal waveform sampling shown in FIG. 38 or 43 to the laser interferometer 44 for the position measurement of the stage WST (for example, 0.02). 1 pulse per micrometer).

In this way, waveform data of each photoelectric signal generated while prescanning the grating mark MHG over several pitches is stored in the memory circuit 54 corresponding to the grating position of the grating mark MG. However, the prescan of the stage WST requires the number of times of switching and responding to each wavelength of the incident light beam ± LF.

In the method of irradiating the two incident light beams ± LF to the grating mark MG, it is preferable that the two incident light beams ± LF have a symmetrical incident angle with respect to at least the pitch direction of the grating mark MG. In the method of projecting one incident light beam on the MG, the incident angle is preferably set to zero (vertical incidence) with respect to the pitch direction of the grating mark MG. That is, it means that the incident light beam may be inclined in the direction (non-measurement direction) orthogonal to the pitch direction of the grating mark MG.

By the way, when projecting the illumination beam switched for each wavelength to the grating mark MG (or reference mark) for measurement, a plurality of laser beams for each wavelength, such as 40 degrees, 41 degrees, and 37 degrees, are coaxial once. The incident structure separated in the non-measurement direction orthogonal to the measurement direction (pitch direction) of a mark position may be sufficient in the Fourier transformation surface of the grating mark MG without making it so suitably. That is, the angle of incidence to the grating mark MG can be varied in the non-measurement direction for each wavelength of the plurality of illumination beams. The configuration therefor can be made exactly the same as those of the 35th and 36th degrees.

The generation of the incident light beam is not limited to the laser light source, but it is also possible to use light from a halogen lamp and light from a high brightness LED. When using light from a halogen lamp, a plurality of wavelength selective filters (or interference filters) having narrow bandwidths in different wavelength portions are provided so as to be interchangeable, and the filters are time-divisionally switched to select light having a wavelength width selected, for example. For example, light guides may be used for wafers or the like. In this case, since the incident light beam irradiating the lattice mark MG on the wafer has a continuous spectral intensity distribution even within a narrow wavelength bandwidth of selection, an interference filter (bandwidth of 3 to 10 nm) extracting only a specific wavelength component in front of each photoelectric element in the light receiving system. ) May be arranged fixedly or interchangeably.

According to each of the fifteenth to twentieth embodiments of the present invention, the illumination light for position detection is switched for each of a plurality of wavelength components, and the diffracted light generated from the lattice marks for position detection on the substrate is individually for each wavelength component. Photoelectric detection is performed, and the mark position information is detected for each photoelectric signal thus obtained and averaged by calculation. Therefore, highly accurate position detection with reduced influence of the asymmetry of the mark and the thickness of the resist layer is possible. In addition, since diffraction light from the mark is obtained by obtaining an independent photoelectric signal for each wavelength component when photoelectric detection is performed, there is an advantage that the averaging effect due to multiple wavelengths is not impaired even if the intensity of each wavelength component of the illumination light is different.

Further, according to each of the fifteenth to twentieth embodiments, even when the diffracted light to be photoelectrically detected is made of higher order components, higher order diffracted light (multi-wavelength, interference beam of secondary light, etc.) multiplied by a single photoelectric element The canceling phenomenon that occurs when the light is simultaneously received is eliminated, and position detection and alignment with high precision are possible.

Furthermore, the attenuation ratio (amplitude ratio) of the intensity level of the diffracted light for each photoelectrically detected wavelength component was obtained, and the position detection was performed by a large weighted averaging operation on diffracted light having a small attenuation ratio and a relatively large signal amplitude. Compared with the simple averaging, the effect of the positional detection accuracy is also significantly higher.

1 is a graph showing a result of simulating wavelength dependence of position detection error when a conventional coherent mark position detection apparatus using a single wavelength is used.

2 is a partial cross-sectional view showing the cross-sectional shape of the grid mark used as a model of the simulation.

3 is a graph showing the results of simulating the change in reflectivity of a substrate with a resist under the conditions of multi-wavelength determined by the present invention.

4 is a diagram showing diffraction light shapes of respective orders generated from lattice marks by light irradiation of multiple wavelengths.

5A and 5B simulate the detection error when detecting a mark having the same structure as the second degree using the first-order diffracted light and the detection error when detecting a mark using the zero-order -second order diffracted light. graph.

6 is a diagram showing the configuration of the position detection device according to the first embodiment of the present invention.

FIG. 7 is a perspective view showing the detailed structure of the rotational radial grating plate and the appearance of generation of diffracted light in FIG. 6; FIG.

FIG. 8 is a graph showing the results of simulating the change in reflectance of a substrate with a resist when the wavelength values set in the case of FIG. 3 are separately combined and satisfied while satisfying the conditions of the multi-wavelength determined by the present invention.

9 is a block diagram showing the configuration of a signal processing circuit applied to the apparatus shown in FIG.

10A and 10B are waveform diagrams showing examples of waveforms of two photoelectric signals processed in the signal processing circuit of FIG.

11 is a partial cross-sectional view schematically showing the cross-sectional structure of a small step difference grating mark formed on a semiconductor wafer and a resist layer covering the surface thereof.

12 is a diagram showing a configuration of a position detection device according to a second embodiment of the present invention.

13A to 13D are diagrams schematically showing the position detection shape of the grating mark and the change of the photoelectric signal by the apparatus of FIG.

14 is a diagram showing the configuration of a position detection device according to a third embodiment of the present invention.

15 is a view showing the configuration of a projection exposure apparatus to which the position detection device according to the present invention is applied as a fourth embodiment of the present invention.

FIG. 16 is a diagram showing a partial configuration of a TTL alignment system in FIG. 15. FIG.

17A and 17B show a configuration of a projection exposure apparatus to which the position detection apparatus according to the present invention is applied as the fifth embodiment of the present invention.

18 is a diagram showing a part of a configuration of a position detection device according to a sixth embodiment of the present invention.

19 is a diagram showing the configuration of a position detection device according to a seventh embodiment of the present invention, and showing a configuration in the case where a spectroscopic detector for receiving an interference beam is provided for each wavelength component separately.

20 is a block diagram showing a configuration of a signal processing circuit applied to the apparatus of FIG. 19. FIG.

21 is a diagram showing the configuration of a position detection device according to an eighth embodiment of the present invention.

22A to 22D show changes in relative positional relationship between the interference fringes and the grating and changes in the level of each detection signal.

23 is a diagram showing the configuration of a position detection device according to a ninth embodiment of the present invention.

24 is a diagram showing the configuration of a position detection device according to a tenth embodiment of the present invention.

25 is a block diagram showing a signal processing circuit applied to the apparatus according to the tenth embodiment.

26A to 26D show examples of signal waveforms to be received in the memory of the processing circuit of FIG. 25;

27 is a block diagram showing a modification of the signal processing circuit applied to the apparatus shown in FIG. 24 as an eleventh embodiment.

FIG. 28 is a partial enlarged view of the TTL alignment system of the apparatus shown in FIG. 15. FIG.

29A to 29D show examples of waveforms of photoelectric signals for respective wavelengths obtained by the respective interferences of the 0th order light and the 2nd order light from the diffraction grating.

30A to 30D show an example of a waveform of a photoelectric signal for each wavelength obtained by respective interference of zero order light and secondary light from a diffraction grating.

31 is a diagram showing the configuration of a position detecting device according to a twelfth embodiment of the present invention.

32 is a block diagram showing the configuration of a signal processing circuit applied to the apparatus of FIG. 31;

33 is a sectional view showing a partial configuration of a device according to a thirteenth embodiment of the present invention.

34 is a diagram showing the configuration of a position detection device according to a fourteenth embodiment of the present invention.

35 shows a modification of the projection scheme of the illumination beam disclosed in each embodiment of the present invention.

36 shows an arrangement example of each beam on a Fourier transform plane in the projection method of the illumination beam of FIG. 35;

37 is a diagram showing the configuration of a position detection device according to a fifteenth embodiment of the present invention.

FIG. 38 is a block diagram showing a signal processing circuit applied to the apparatus shown in FIG. 37. FIG.

39A to 39D show examples of complex signal waveforms in the memory of the processing circuit of FIG. 38;

40 is a diagram showing the configuration of a position detection device according to a sixteenth embodiment of the present invention.

41 is a diagram showing the configuration of a position detection device according to a seventeenth embodiment of the present invention.

42 is a diagram showing the configuration of a position detection device according to an eighteenth embodiment of the present invention.

43 is a block diagram showing a configuration of a signal processing circuit applied to the apparatus of FIG. 42. FIG.

FIG. 44 is a diagram explaining a memory bank arrangement in the waveform memory circuit unit in FIG. 43; FIG.

45 is a sectional view showing a partial configuration of a device according to a nineteenth embodiment of the present invention.

46 is a diagram showing the configuration of a position detection device according to a twentieth embodiment of the present invention.

Explanation of symbols on the main parts of the drawings

10: collimator lens 22: objective lens

38: space filter 40: photoelectric element

50: A / D converter circuit

Claims (17)

A method for detecting the position of a position detection grating mark formed in a small step structure on the surface of a flat object: (a) irradiating an illumination light beam onto the grating mark at a predetermined angle of incidence, wherein the illumination light beam has n (n ≧ 3) different wavelengths λ ₁ , λ ₂ , λ ₃ , ..., λ _n) a includes a plurality of incoherent beams (coherent beams) having the n number of wavelengths are within the range of about _{± 10% (1 / λ 1} - 1 / λ 2) = (1 / λ 2 - 1 / λ ₃ ) = ... = (1 / λ _n-1-1 / λ _n ) is set to approximately satisfy the relationship, where the wavelengths are λ ₁ <λ ₂ <λ ₃ ... <λ irradiating said illumination light beam with a condition of _n ; (b) performing photoelectric detection of changes in the amount of light of the diffracted light component generated from the grating mark in a specific direction upon irradiation of the illumination light beam having the n wavelength components; And (c) determining the position of the grating mark based on the signal obtained in the photoelectric detection. A method of measuring the relative positional relationship between a first diffraction grating and a second diffraction grating in a periodic direction: (a) irradiating an illumination beam comprising different first and second wavelengths onto the first diffraction grating, thereby providing a plurality of first diffraction beams having the first wavelength and a plurality of second having the second wavelength. Generating two diffracted beams; (b) irradiating the plurality of first diffraction beams and the plurality of second diffraction beams on the second diffraction grating to generate a plurality of re diffraction beams from the second diffraction grating; (c) photoelectric detection of a specific rediffracted beam of the plurality of rediffracted beams and included in the first photoelectric signal and the specific rediffracted beam corresponding to the beam component of the first wavelength included in the specific rediffracted beam Photoelectric detection to output a second photoelectric signal corresponding to the beam component of said second wavelength; (d) calculating a first position displacement amount representing the positional relationship based on the first photoelectric signal, and calculating a second position displacement amount indicating the positional relationship based on the second photoelectric signal; And (e) calculating a weighted average of the first and second positional displacements with weighting factors corresponding to the intensities of the first and second photoelectric signals, whereby the relative between the first and second diffraction gratings Confirming the positional relationship. Applying a coherent illumination beam to a diffraction grating pattern formed at a predetermined pitch in the pitch direction on the substrate to be detected, and changing the intensity of the specific diffraction light beam among the plurality of diffraction light beams generated from the grating pattern; In the method for detecting the position of the grating pattern in the pitch direction: (a) receiving by the first photovoltaic device first interfering light by ± n-order diffracted light beams generated from the grating pattern on the substrate, and generating a first photoelectric signal; (b) receiving, by a second photovoltaic device, second interference light by ± m (m ≠ n) order diffracted light beams generated from the grating pattern on the substrate, and generating a second photoelectric signal; (c) detecting a position of the grating pattern on the substrate with respect to the pitch direction based on the first photoelectric signal, and a second of the grating pattern on the substrate with respect to the pitch direction based on the second photoelectric signal Detecting a location; And (d) calculating the constant position of the grating pattern on the substrate by averaging the detected first position and the second position by assigning weights according to amplitude information of the first and second photoelectric signals; , The method of detecting the position of the grid pattern. A method of measuring the relative positional relationship between a first diffraction grating and a second diffraction grating in a periodic direction: (a) irradiating an illumination beam comprising first and second wavelengths different from each other onto the first diffraction grating, thereby providing a plurality of first diffraction beams having the first wavelength and a plurality of second having the second wavelength. Generating two diffracted beams; (b) irradiating the plurality of first diffraction beams and the plurality of second diffraction beams on the second diffraction grating to generate a plurality of re-diffracted beams from the second diffraction grating; (c) photoelectrically detecting the re-diffracted beams to obtain first and second photoelectric signals; And (d) determining the positional relationship using the first and second photoelectric signals weighted according to their respective amplitudes. In the method for measuring position information of an object having a diffraction grating pattern: Irradiating a first illumination beam having a first wavelength and a second illumination beam having a second wavelength different from the first wavelength on the diffraction grating pattern vertically; Detecting a first interference beam between diffraction beams generated from the diffraction grating pattern irradiated with the first illumination beam; And And detecting a second interference beam between diffraction beams generated from the diffraction grating pattern irradiated with the second illumination beam. The method of claim 5, wherein And the first interference beam and the second interference beam are detected by respective detectors. The method of claim 5, wherein And the first wavelength beam and the second wavelength beam irradiate the diffraction grating at substantially the same time. The method of claim 5, wherein And the first wavelength beam and the second wavelength beam irradiate the diffraction grating at different timings. In a method of exposing a substrate: Aligning the substrate using a method for measuring position information of an object having a diffraction grating pattern according to claim 60; And Forming a pattern on the aligned substrate. A device for measuring positional information of an object having a diffraction grating pattern: A first illumination beam having a first wavelength and a second illumination beam having a second wavelength different from the first wavelength, disposed on a predetermined side with respect to the object, irradiated vertically on the diffraction grating pattern Illuminated beam system; And Disposed on the same side as the predetermined side with respect to the object, detecting a first interference beam between diffraction beams generated from the diffraction grating pattern irradiated with the first illumination beam and irradiated with the second illumination beam And a detector for detecting second interfering beams between diffracted beams generated from the diffraction grating pattern. In an apparatus for exposing a substrate with a pattern formed on a mask: A projection optical system for projecting the pattern formed on the mask onto the substrate; And A position detection system for detecting a mark formed on the substrate without passing through the projection optical system, The position detection system, A beam illumination system arranged on a predetermined side with respect to the substrate and irradiating an illumination beam vertically on the mark with respect to the periodic direction of the mark; A first detection system arranged on the same side as the predetermined side with respect to the substrate, wherein ± n-order (n is a natural number) diffraction beams generated from the mark irradiated with the illumination beam are predetermined with respect to a first reference grating; A first photoelectric element for detecting a beam from said first reference grating irradiated with said &lt; RTI ID = 0.0 &gt; n &lt; / RTI &gt; And A second detection system arranged on the same side as the predetermined side with respect to the substrate, wherein the ± m difference (m is a natural number other than n) diffracted beams generated from the mark irradiated with the illumination beam are transferred to a second reference grating; And a second optical element for detecting a beam from the second reference grating irradiated with the ± m-diffraction beams, each of the second optical element being inclined at a predetermined angle relative to the second reference grating. Substrate exposure apparatus. In the medium for detecting the position information of the mark formed on the substrate: A beam illumination system arranged on a predetermined side with respect to the substrate and irradiating an illumination beam vertically on the mark with respect to the periodic direction of the mark; A first detection system arranged on the same side as the predetermined side with respect to the substrate, wherein ± n-order (n is a natural number) diffraction beams generated from the mark irradiated with the illumination beam are predetermined with respect to a first reference grating; A first photoelectric element for detecting a beam from said first reference grating irradiated with said &lt; RTI ID = 0.0 &gt; n &lt; / RTI &gt; A second detection system arranged on the same side as the predetermined side with respect to the substrate, wherein the ± m difference (m is a natural number other than n) diffracted beams generated from the mark irradiated with the illumination beam are transferred to a second reference grating; A second optical element comprising a second optical element for respectively inclining at a predetermined angle relative to the second photoelectric element and a second photoelectric element for detecting a beam from the second reference grating irradiated with the ± m diffraction beams; And And an operator electrically connected to the first and second photoelectric elements, for calculating the position information of the mark based on detection by the photoelectric elements. In the method for detecting positional information of a mark formed on a substrate: Irradiating an illumination beam vertically on the mark with respect to the periodic direction of the mark; The first ± n order (n is a natural number) diffraction beams generated from the mark irradiated with the illumination beam so as to be inclined at a predetermined angle with respect to a first reference grating, respectively, and the first irradiated with the ± n order diffraction beams Detecting a beam by the first photovoltaic device from the reference grating; And ± m-difference (m is a natural number other than n) diffraction beams generated from the mark irradiated with the adjustment beam are respectively inclined at a predetermined angle with respect to a second reference grating, and irradiated with the ± m-difference diffraction beams And detecting a beam from the second reference grating by a second photoelectric element different from the first photoelectric element. The method of claim 13, Calculating the position information of the mark based on detection results by the photoelectric elements. In the projection exposure method: Aligning the substrate based on the position information detected using the mark position information detection method according to claim 14; And Projecting exposure to the aligned substrate in a pattern formed on a mask. In a device for detecting positional information of a mark formed on a substrate: A first illumination beam having a first wavelength and a second illumination beam having a second wavelength different from the first wavelength, arranged on a predetermined side with respect to the substrate, on the mark with respect to the periodic direction of the mark; Beam lighting system irradiated with; A detection system arranged on the same side as the predetermined side with respect to the substrate, wherein diffraction beams generated from the mark irradiated with the first illumination beam or the second illumination beam are each inclined at a predetermined angle with respect to a reference grating; And a photoelectric device for detecting a beam from the reference grating irradiated with the diffraction beams; And And a computing device electrically connected to the photoelectric element, for calculating the position information of the mark based on detection by the photoelectric element. In the method for detecting positional information of a mark formed on a substrate: Irradiating a first illumination beam having a first wavelength and a second illumination beam having a second wavelength different from the first wavelength on the mark perpendicular to the periodic direction of the mark; Causing diffraction beams generated from the mark irradiated with the first illumination beam or the second illumination beam to be inclined at a predetermined angle with respect to a reference grating respectively; And Detecting a beam by a photoelectric element from said reference grating irradiated with said diffraction beams, and calculating said positional information of said mark based on detection by said photoelectric element.