JPH0210202A - Position detecting apparatus by double linear fresnel zone plate through illumination with multi wavelengths - Google Patents

Abstract

PURPOSE:To achieve high throughput by providing a cylindrical lens between an objective lens and a linear sensor, processing signals obtained from said linear sensor and detecting the position of an alignment mark. CONSTITUTION:A wafer 15 and a mask 14 are overlapped with each other by a gap 18 of several tens mum. In the right and left upwards in a slantwise direction from the wafer 15 and mask 14, there are provided an illuminating optical device 5 and a detecting optical device 10 confronting each other. An illuminating angle 19 of incident light from an optical axis of the device 5 for a normal line of the mask 14 or wafer 15 is selected to be equal to a slantwise detecting angle 20 of an optical axis of the device 10. The illuminating light and detecting light are thus inclined for the mask 14 and wafer 15. Therefore, alignment marks 16, 17 within an X-ray exposing region 13 can be detected without moving the devices 5, 10. Furthermore, because the device 10 is placed outside an optical path of the region 13, the marks 16, 17 can be detected even during the exposing time, and it is not necessary to move the device 10 at every alignment; thereby, high throughput can be achieved.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention is applicable to ultra-precision position measurement (0.01 μm order).
The equipment, especially masks and wafers, are 0. This invention relates to a position detecting device using a double linear Fresnel zone plate using multiple wavelengths, which is a position detecting device on the order of O1 pm.

[Prior Art] Mask and wafer alignment methods that are currently in practical use can be roughly divided into the pattern meter J11 method, single diffraction light method, and a hybrid method that partially combines both of these methods. , these will be explained individually below.

First, I will list the names of the documents that introduce these topics.

Literature l-Precision Machinery 5115/1985゜PP, 156~
162 Sentence f42- Tectu+1cal Proceedi
ngsScmicon/West 1987. May
19-21 San McCro, CA, PP, 24
2-247 Document 3 - 1986 Japan Society for Precision Engineering Autumn Conference Academic Lecture Proceedings PP, 347-348 Document 4 - JP-A-6
1-236117 Publication Text +lL 5-Special Issue 11i'
(No. 61-1933, 448 + U Document 6 - "Optical Technology Application System" published by Seikodo Co., Ltd. (March 20, 1981) PP, 165-167 Document 7 - 1984 Precision Society Autumn Conference Academic Collection of conference papers PP, 443-444 Reference 8 - Collected papers of the 1962 Precision SL Society Autumn Conference Academic Conference PP, 353-354 Reference 9 - Collected papers of the 1961 Precision Engineering Society Autumn Conference Academic Conference I11. 703-704 Reference 10 - 1960 Precision Machinery Society Spring Conference Academic d^ Conference Paper East PP, 301-302 Text 11i 1 1 - Semicon News
198 [i, 7. P, 20 Reference 12-Preliminary Lectures of the 48th Academic Conference on Applied Physics (1988) 11,448 Reference 13 J, Vac, Sci Tccbnol, l
'+(4) INov,/Dec, 1981. PP
, 1224-1228 sentences dQ l 4 -5oli
d 5tate Technology, 21.5
(+985 Oki, 175 Reference 15-J, Vac, Sci Teclu+ol, 1
li(6) Nov, / Dec, 197'l, PP,
1954-1'158 Reference 16-J, Vac, Sci
Technol, 19(4).

Nov,/Dec, 1981. PP, 1219-1
22'1 sentence + i & 17- I E E E Tra
nsactions onElecLron Dev
ices, Vol, ED-26, NO, 4PP, 72:
l~728 Document 1 B-1987 (Showa 62) Autumn, f548
Annual Conference of Japan Society of Applied Physics, 2nd Minute III) P 427, 18a-F-8
Sentence 119-times, 18a-F-9 Sentence $Q 2 0 - Sem1conductor
1llorld 1 9 8 5.

5, PP, 105-111 (1) Pattern measurement method (detection power method using a bifocal lens using O birefringence (development organization) Bell Labs and NTT (4F1 omitted) Detecting alignment marks using a bifocal lens using birefringence In addition, it uses a step-and-repeat method in which alignment is performed for each exposure area in minutes.(For NTT's example, see Reference 1:1) (Disadvantages) (1) There is a limit to the reduction in lens diameter, and since the mark position is located at a scribe line quite far (about 10 mm) from the exposure area, alignment errors due to expansion and contraction of the mask are added, reducing alignment accuracy.

(II) Due to (I), a mark cannot be provided within the exposure area.

(m) Depends on the symmetry of marks due to binarization processing.

(■) Since detection is performed using a two-dimensional camera, high-speed detection is not possible (30Hz or less).

■Detection method using a 22 bifocal lens (it is unknown whether it is due to birefringence).

(Development organization) Kart 2u (Overview) Marks are detected using an optical microscope with dual focus. The mask and wafer are moved under this microscope for alignment, and after completion, the mask and wafer are locked.
It is moved to the exposure position and X-ray irradiation is performed. (Refer to Document 2) (Disadvantages) The exposure concept is different from other X-ray aligners, and alignment and exposure are performed at different locations, so they cannot be compared generally, but at least servo control and high-speed detection during exposure are not an integral part. It is.

■ Oblique detection (Development organization) [1 at the manufacturing facility (outline) By tilting an objective lens with a small NA outside the X-ray exposure area and detecting marks on the disk and wafer obliquely, ':A The device is capable of detecting marks provided within the exposure area even under light.

(See Sentence 1! I & 3, 4.5).

(Disadvantages) (I) Detection is performed using a two-dimensional camera (Reference 3.
4), high-speed detection is not possible.

(II) A process called symmetric pattern matching (see Reference 5) is performed, and this is a calculation method based on the symmetry of the marks, and the detection accuracy greatly depends on the symmetry of the marks.

(m) Due to oblique imaging, the imaging range is extremely narrow, making it difficult to improve the S/N ratio through raster compression, etc.
Highly process dependent in terms of mark shape retention.

(2) Diffraction light method σ) Equal pitch double diffraction grating method (development organization) MIT (Overview) Diffraction gratings with the same pitch are provided on the mask and wafer, and the superimposed signal is generated by the intensity difference of the first-order transmitted diffraction light. get. The positional displacement is expressed by the relationship Δφm=(double m/d)ΔX, Δφm=phase displacement ΔXdouble phase displacement m: diffraction order, O, I#Lm with a grating pinch of 10ILm.

200 people with 1.2 μm grid pinch.

The overlay accuracy can be obtained. (Refer to Document 6.7) (Disadvantages) (I) The influence of positional deviation signals due to gap fluctuations is extremely high. FIG. 2 of Reference 7 shows how the positional deviation signal is greatly affected by the gap variation (±0.04 pm).

(11) Due to monochromatic light alignment, the diffracted light intensity is greatly affected by light interference phenomena such as standing wave effects in thin films.

(m) Depends on the symmetry of the mark (detection accuracy decreases with an asymmetrically shaped grating).

■Double pitch double diffraction grating method (development organization) NTT Tsuken (General WS) Double pitch diffraction grating eliminates the influence of gap fluctuations, which was a drawback of the conventional double diffraction grating method, and is used for gear love detection. , instead of the conventional) orcus signal.
This method uses a diffracted light intensity signal from a diffraction grating. (Refer to Documents 8, 9 and 10) Document 8 provides an overview of the x6 aligner equipped with a PS, which is a detection device using the double-pitch/edge double diffraction grating method, and Document 9 provides an overview of the detection device. There is.

(Disadvantages) (I) Gap dependence is considerably reduced compared to the equal-pitch 2-ton diffraction grating method, but gap control on the order of 0.1×m is still required.

(II) Since it is a monochromatic light alignment, there is a problem of light interference, and the influence of the process (step, step width, tXII
I! 2 deposited layers) or unavoidable 0 process effects are shown in Ref. 10.

(III) Depends on the symmetry of the grating marks.

■Method of using a bare diffraction grating (Development organization) Electric Research Institute O11 (omitted) As shown in the drawing in Reference 11, by placing a pair of diffraction gratings on the wafer side, the gap between the mask and the wafer is The principle is that the positional deviation detection signal () is not affected by fluctuations. As shown in Reference 12, the two characteristics are (a) the two diffraction gratings are not directly opposed, and (b) the grating period is set to a non-integer multiple. (Refer to Documents 11 and 12) (Disadvantages) (I) Since it is a monochromatic light alignment, the influence of the process due to light interference problems cannot be avoided.

(TI) Depends on the symmetry of the grating marks.

(m) As shown in sentence fi12, the gap dependence is based on simulation and has not been demonstrated.

02 Double Fresnel Zone Targeting Method (Development Institution) Be1.1 Lab (Outline) Circular Fresnel Zone Targeter 1~ (FZT) with focal length f formed on the mask and focal length l1l (ft gap) formed on the wafer 1- The FZT is irradiated with a parallel laser beam, and the position and gear amplifier are detected by detecting the position change of the focused spot that occurs at that time. (Refer to Document 13) (Disadvantages) (I) As shown in Table 1 and Figure 2 of Sentence 1113, since monochromatic light alignment is used, the influence of the process due to light interference cannot be avoided, and sufficient S /N ratio cannot be obtained, and film thickness control becomes difficult due to standing wave effects in thin films.

(II) High gap dependence Although the gap dependence is low compared to the diffraction grating method, the light intensity of the spot focused by Fresnel diffraction varies due to gap variation.

(m) The 0 circular FZT that depends on the symmetry of the mark can be thought of as a kind of convex lens, so if the symmetry of the mark is impaired, a phenomenon equivalent to astigmatism or coma aberration occurs in lenses. Detection accuracy decreases.

(rV) The mark shape is complex and pattern changes are likely to occur due to process changes.

(V) Asymmetry of the image point occurs due to spurious images generated near the focus focused by Fresnel diffraction, and position 2 (
It makes a difference.

■Double circular Fresnel zone target method (development organization) M
icronix (outline) The alignment principle is the diffraction light method ■Ba1l
The same thing as the lab one is applied.

Micronix sells it by incorporating it into aligners under the trade name Mx-15. (Refer to Document 14) (Disadvantages) (I) The disadvantages mentioned in ① above are still present.

(m)' A servo control system is not set in the light beam.

(■) High-speed detection is impossible because it is detected using a two-dimensional camera.

■Method of combining a linear Fresnel zone plate and a diffraction grating (development organization) Th, omson-C3F.

Position detection is performed using a transmissive linear Fresnel zone plate (LFZP) as a mask mark and a reflective linear diffraction grating as a wafer mark. Since the focal length of the LFZP corresponds to the amount of gap between the mask and the wafer, vertical alignment is also possible. (Refer to Documents 15 and 16) (Disadvantages) (I) Since tn color light is used as alignment light, there is a problem of light interference.

(II) Check the symmetry of the diffraction grating of wafer 42
2.1 degrees depends.

(m) High gap dependence.

(IV) In order to improve the detection accuracy, it is necessary to increase the S/N ratio, so it is necessary to arrange the LFZP and the diffraction grating in a matrix, and the exclusive area of the mask is 1000.
An extremely large area of approximately gm square is required.

(V) High overlay accuracy is required between those marks.

The alignment method using diffracted light has been explained individually from CD to ■ above, but regarding the common point that "it depends on the symmetry of the diffraction alignment mark", see Reference 17. The alignment accuracy of diffraction gratings with various profiles as shown in FIG. 2 is being investigated. It has become clear that in an asymmetric diffraction grating like the profile in (e) in Figure 2, the detection accuracy decreases because ψ1 in w, which is equation (3) in this document, is no longer zero. There is.

Furthermore, Reference 118 discusses the correction of an asymmetric diffraction grating, and it is shown that the asymmetry is corrected by the rotation part from the response signal Jn obtained from the following equation from Δ which defines the asymmetry. ing.

Jn=In [l+cos2k (x-Δ)・n input/L
i] Further, 1 Reference 19 reports on making the diffraction grating asymmetric due to the standing wave effect of the resist film thickness.

(3) Hybrid method (pattern measurement method 10-diffracted light method) Nikon's optical stepper: N5R150 is a method that combines two methods using a diffraction grating and a pattern meter i1-.
There is an LSA (laser stamp alignment) detection device installed in the 5G3A.

(J11 omitted) A diffraction grating mark is provided on the wafer, and as the stage moves, a beam passes over the alignment mark and diffracted light is generated, and a signal waveform is obtained based on the light intensity. After this is received by a detector, it is A/D converted, stored in a waveform memory as a digital signal, and its center position is measured by a waveform processor. The light intensity of the diffraction grating is used as the detection signal, and the process of determining the position from the waveform calculates the center position of the waveform data by pattern measurement. (Hybridization) (See Reference 20) (Disadvantages) (I) Monochromatic light alignment.

(II) Depends on the symmetry of the grating mark.

(m) The detection speed depends on the signal processing system (digital system).

Table 1 summarizes the main problems of the conventional pattern measurement method and the diffracted light method.

Table 1: Problems with conventional methods (Note) X is a problem.

Reason for △: In the case of pattern measurement of N013, it is possible by oblique detection (■) in (1), or by using a bifocal lens using double Jrit folding (■) in (1), or (1)
According to method (■), it is an integral part.

In the case of N013 Fresnel zone plate.

According to (2) (2), Thomson-C3F, it is possible, although it has a drawback that the mark is large (1 mm x 1 mm). However, (2) (Φ Micronix
According to (Mx-15), a mark is provided on the scrub line, so it is an unacceptable part.

In the case of No. 4 Fresnel zone plate, if a circular Fresnel zone plate is used like Micron ix (Mx-5) in (2) ■ or Ba1l lab in (2) ■, 2 Since a dimensional camera is used as a light receiving element, high-speed detection is impossible due to its responsiveness (30 Hz).

[Problems to be solved by the present invention] The present invention aims to provide a novel alignment device that solves the six problems of the prior art listed in Table 1 above. In other words, as the ultra-high integration of semiconductors progresses at a rapid pace, 16M, 64M, and 25M will be mass-produced in the future.
In an X-ray aligner that uses synchrotron radiation (SOR) as a radiation source, which is used in the production of memory devices such as 6M, we are trying to provide an alignment device that can detect the position of masks and wafers on the order of 0.01 gm. Currently used optical Stellar Per and X
Alignment methods for line aligners are broadly divided into two types: (0-pattern measurement method (contrast method, edge detection method), and method using diffracted light.Each method has its own advantages and disadvantages. Although it is difficult to say which is better or worse, the purpose of this IJII is to provide a position detection device that uses a pipe-lit alignment method that combines the advantages of (1) and (2).

[Means for Solving the Problems] A position detection device of the present invention detects the relative position of a first object and a second object in a direction orthogonal to the optical axis, which are separated by a small distance in the optical axis direction of xVj exposure light. A position detection device,
Linear Fresnel Zone Plate (LFZP) on each object
The illumination device is configured to simultaneously illuminate these LFZPs with light of multiple wavelengths from the same direction, and as described above, the focus of that wavelength is aligned with the diffraction focus located at a different position for each wavelength of the FZP. The above-mentioned LFZP is provided with an objective lens having a chromatic aberration such that
A linear sensor is provided on the imaging plane for one-dimensionally scanning the linear diffraction focal image in a direction perpendicular to its longitudinal direction and converting it into an electrical signal, and so that it is compressed in the direction and imaged onto the above linear sensor.

Double linear Fresnel using multi-wavelength illumination having means for detecting the position of each of the alignment marks by arranging a cylindrical lens between the objective lens and the linear sensor and processing signals obtained from the linear sensor. This is a position detection device using a zone plate.

A single LFZ as the alignment mark of the first object
It is convenient to use a pair of LFZPs as alignment marks for the second object, with a single LFZP of the first object positioned between the pair of LFZPs of the second object.

The first object is a mask, the second object is a wafer, and it is practical to provide a transparent window area in the mask portion through which the diffracted light and incident light of the alignment mark on the wafer pass.

The illumination device and the objective lens are arranged at opposite positions above the left and right shoulders of the first object alignment mark and the second object alignment mark, and the incident angle of the illumination device is such that the incidence angle of the illumination device is such that the objective lens The position detection device of the present invention can be placed outside the X-ray exposure region by arranging it so that the detection angle is equal to the detection angle of the detection optical system.

If the means for processing the signals obtained from the linear sensor and detecting the positions of the alignment marks is means for performing similarity pattern matching processing, highly accurate position detection is performed.

[Function] When the LFZP provided on each object is simultaneously illuminated with light of a plurality of wavelengths, diffraction focal points of the LFZP are formed at positions with different focal lengths depending on the wavelengths. The focal length of each LFZP is chosen so that the diffraction focal points of the LFZPs of both objects are located on the same focal plane for each wavelength of light. An objective lens with chromatic aberration is arranged so that the focal point of each wavelength coincides with the diffraction focal plane of each wavelength, and the diffraction focal images of each wavelength are superimposed and focused on the same imaging plane by this objective lens. imaged. The linear diffraction focal images of each LFZP, which are superimposed and formed on the same imaging plane, are one-dimensionally transformed in a direction perpendicular to the long plane direction by a linear sensor placed on the imaging plane. It is scanned and converted into an electrical signal. At this time, a cylindrical lens placed between the objective lens and the linear sensor allows
The linear diffraction focal image is compressed in its long plane direction and formed on the linear sensor. The signals obtained from the linear sensor are processed to detect the position of each alignment mark, and the two are aligned.

[Example] Hereinafter, with reference to FIGS. 1 to 17, the configuration and operation of a position detection device using a tuple linear Fresnel zone plate using dual wavelength illumination according to the present invention will be described. FIG. 1 is a schematic side view when the position detection device of the present invention is applied to a mask and a wafer, and FIG. 2 is a schematic side view thereof. A light beam 1 is placed diagonally above the left and right sides of the wafer 15 and the mask 14, which are stacked with a gap 18 of several tens of pm.
51 optical device 5 and detection optical device 10 are arranged at opposing positions. The incident illumination angle 19 of the optical axis of the illuminating IJJ optical device 5 with respect to the normal line of the mask 14 or the wafer and the oblique detection angle 20 of the optical axis of the detection optical device 211O are equal, and are usually within the range of 10° to 45''. (25° in the case of the figure) 6 In the present invention, by tilting the illumination light and the detection light with respect to the mask and the wafer in this way,
This allows the alignment marks 16 (mask) and 17 (wafer) within the X-ray exposure area 13 to be detected without moving the illumination optical device 5 and the detection optical device 10. This is especially true when referring to the fjS2 diagram.
It's obvious. Since the detection optical device 10 is provided outside the optical path of the exposure area 13, it can detect the alignment marks 16 (17) even during exposure, and there is no need to move the detection optical device 10 for each alignment. High throughput is possible.

Although the illumination optical device 5 uses laser light in FIG. 1, any other type of illumination light may be used as long as it can provide an emission line with a narrow half-width. In the case of Figure 1,
IJj from laser l! , long input, and laser beam with wavelength input 2 from laser 3 (input, 〉input 2) are combined by a heme splitter 2 and reflected obliquely downward by a mirror 4 as parallel light. I'm doing the lighting. The detection optical device 10 includes an objective lens 9 having axial chromatic aberration, and since this objective lens 9 has different focal lengths for wavelength input and input light, the wavelength input at different object positions By using the fact that the object point of the light of , and the object point of the light of wavelength input 2 are superimposed and formed at the same image point position (described later), a mask 14 made of a linear Fresnel zone plate (LFZP) is formed. The focal points (linear focal points) at different positions depending on the wavelengths 1 and 2 of the alignment mark 16 of the alignment mark 16 of the wafer 15 and the alignment mark 17 of the wafer 15 are the focus of the light of the objective lens 9 and the focus of the light of the wafer 15. Match.

The objective lens 9 and the relay lens 8 form an image superimposed on a linear sensor 6 having a long and narrow detection area in the Y direction in the figure. In addition, relay lens 8
is not necessarily required, but a linear mark 16, 17 is located between the relay lens 8 and the linear sensor 6, and is located within the plane of Figure i1, and is perpendicular to the optical axis of the detection optical system 10. A cylindrical lens 7 (described later) is arranged to compress the image and form an image on the linear sensor 6, thereby enabling high-speed detection. The detection signal from the linear sensor 6 is subjected to signal processing, which will be described later, to align the position of the alignment mark 16 in the Y direction and the alignment mark 17.
The respective positions in the Y direction are detected. Mask 1 below
4. Alignment mark 16.17 on wafer 15 side
, objective lens 9. Processing of signals from the cylindrical lens 7 and the linear sensor 6 will be explained in detail.

First, the alignment mark (mask mark) 16 on the mask 14 and the alignment mark 17 on the wafer 15 will be explained with reference to FIG. Both the mask mark 16 and the wafer mark 17 are linear Fresnel zone plates (LFZ) having straight stripes (zones) in the X direction in FIG.
It consists of P).

As shown in Fig. 4, the LFZP 23 generally has a linear focal point 24 (Fresnel focal point) parallel to the stripe direction (X direction in the figure), and the focal length varies depending on the wavelength. In the circular Fresnel zone plate, the parallel incident light beam is condensed into a conical point focal point, but in the LFZP, the beam is focused in a cross section perpendicular to the fringes (y-z plane).
The focal point 24 is a straight line because the light is focused on the focal point only within the stripe, and is reflected (transmitted) as parallel light without undergoing diffraction within the cross section (X-Z plane) parallel to the stripes. It can be understood that the circular Fresnel zone plate functions similarly to a convex lens made of a spherical surface, while the LFZP functions similarly to a convex cylindrical lens.

Now, returning to FIG. 3, a pair of wafer marks 17 made of LFZP are provided on the wafer 15 at a predetermined pressure 611 in the Y direction, as shown in the figure. mask 14
1-, a mask mark 16 made of LFZP is provided between two wafer marks 17. mask 1
4J-, a transparent mask window 22 is provided at a corresponding position on the wafer LFZP 17 so that the diffracted light from the wafer LFZP 17 can pass therethrough. Note that the wafer mark 17 is formed on the wafer mark 15 using resist or other thin film.

As shown in FIGS. 5 and 6, the mask 14 and the wafer 15 are stacked 1. with a gap 18 (pm) apart. As shown in Figure 1, the same wavelength of light and λ are used! Ki, each of the mask mark 16 and wafer mark 17 has an incident wavelength λ8. Fresnel diffraction occurs due to the incident light, and two focal planes are formed by the light of each wavelength. Focal length f of mask LFZP l 6
m and the focal length #fw of the wafer LFZP l 7 are selected to differ by a gap of 18 minutes (6 gm) (δ
=fw-fm), the focal length of LFZP differs depending on the wavelength, so as shown in Figure 1, the wavelength input.

LFZP
16 and 17 focal length f m .

Both fw and fw are selected to be large compared to δ, so it can be considered that the focal planes of the wavelengths of both input and input lights coincide with each other as shown in the figure, from the relationship δ = fw - fm. Then, the focal plane 25.26 in the x-Z plane due to mark 16.17 (the focal point is straight in this plane) is i
In the range where the incident angle and detection angle 0 in diagram f are not very large, it can be considered as a plane perpendicular to the optical axis of the detection optical system as shown in FIG. In addition, the mask mark (LFZP)
16 wavelengths included 1. Fresnel diffraction due to the input light is expressed by the following equation.

(Regarding the light at the wavelength input) (Regarding the light at the wavelength λ2) Similarly, each wavelength input of the wafer mark (LFZP) 17,
, Fresnel diffraction due to input 2 is expressed by the following equation.

(About wavelength input light) (About wavelength input light) In the above formulas (1) to (4).

γ: Distance from the center of the m-th zone edge of the mask LFZP γ: Distance from the center of the m-th zone edge of the wafer LFZP fl: Focal length of light with wavelength λ1 of the mask LFZP f8□: Wavelength input of the mask LFZP Focal length of 2nd light f wl: Focal length of wafer LFZP's wavelength input light, f w*: Focal length of wafer LFZP's wavelength input 2nd light, and when irradiating with wavelength input and input 2 light at the same time ,λ
Since the light of , and the light of input 2 become mutually incoherent light, there is no audible interference between wavelengths, and this property is important in this emission 1.

Next, the two-claw pointless detection system 12 using chromatic aberration shown in FIG. 1 will be explained. The basics of this detection system have already been filed by the applicant (Japanese Patent Application No. 1983).
First, to explain this using Fig. 7, the objective lens Ob has chromatic aberration for light rays of different wavelengths, for example, for g-line (input = 435 nm) and e-line (input; 546 nm). The two types of rays have different focal lengths. Therefore, the image point formed when a mark on the same object point is imaged using g-line light is different from the image point formed when e-line light is used.

For example, the focal length of objective lens Ob with numerical aperture NA=0.4 and magnification n=1o is F for e-line and 2g-line light, respectively.
e = 12.5 mm, Fg = 12°741 mm, and the image point distance S'' that occurs when the object distance S is 13.75 mm is S''g = 137.5 mm, S'' e
= 137,615mm. Using such an objective lens Ob, an object point with a minute interval δ, for example, a mask M
A and wafer WA are respectively on the optical axis J2M as shown in the figure.
.

Since the focal lengths Fg and F's of the lights gIIa and eila are different, the image of the mark on the mask MA by the two light beams is formed at point B, and the wafer W
The images of the marks A, J- are generated at points B and C, respectively. That is, due to the chromatic aberration of the objective lens system, the images of the mark on the mask MA and the mark on the wafer WA are formed at the same position at point B. However, the mark on the mask MA is an image formed by gIIa light, and the mark on the wafer W
The mark on A is an image formed by e-line light.

Now, if you place a detection system such as a television camera at point B and observe it, you will be able to observe the marks on the mask MA and the marks on the wafer WA at the same time, but due to monochromatic aberration, each will be observed with a blurred image. be done. This blur image is filtered using a combination of a filter that blocks G-line light and transmits E-line light, and a filter that transmits G-line light and cuts E-line light (referred to as a pattern barrier filter). By using , it is possible to observe the images of the alignment marks on the mask and wafer at different positions at the same point B by removing blurring due to each chromatic aberration (4 yen open/close 62-196174 (see issue).

The above is the content of the prior invention, and the objective lens used in the present invention is a chromatic aberration objective lens that actively generates chromatic aberration in this way and is well corrected for the aberration. below,
An example of the optical system is shown in Table 2.

(Margin below) 2　　　　　　　　　　 (objective lens)

(including field lens and relay lens) This IJ1
In this case, the objective lens 9 used has the same properties as the objective lens Ob shown in FIG. 7, and the focal plane 25 of the light with the wavelength λ1 in FIG.
When the focal plane 26 of the light is placed at the position M'' in Figure i7,
At the position B in FIG. 7, the wavelength λ of LFZP16, 17,
The focal image of the light at the wavelength and the focal image of the light at the input wavelength are overlapped to form an image. To explain this using FIG. 8, the alignment marks 16, . Diffraction focal plane 2 due to light with 17 wavelengths input
5 and the diffraction focal plane 26 of the light with the wavelength input 2 is the objective lens 9.
It is arranged so that the positional relationship perfectly matches the bifocal plane for light having a wavelength of input, .lambda.2. Therefore, both focal planes 25 and 26 are aligned with the objective lens 9 and the relay lens 8.
The image passes through the linear sensor 6 and is superimposed on the linear sensor 6 to form an image.

This situation is further explained using Fig. 9 (a) in Fig. 0.
shows the strong diffracted light component IH3 at the focal plane 25 of the light at the wavelength input from the mask mark 16 and the wafer mark 17, and the stipple 25 in the beam of the light with the wavelength λ1 is of course the light at the wavelength input. Since the Fresnel focus is due to the wavelength λ
The diffracted light intensity 31 due to the light of wavelength 2 shows a peak.On the other hand, the diffracted light due to the light of wavelength 2 is far away from its focal point (gap of 8 gm), so its light intensity 32 is very weak, as shown in Figure 1. As shown by the dotted line, the light intensity distribution is low and gentle. (b) of the figure shows the diffracted light intensity distribution at the focal plane 26 of the light of wavelength λ2 from the mask mark 16 and the wafer mark 17, and since the focal plane 26 of the incident light is the Fresnel focal point of the incident light, The intensity 31' of the diffracted light due to the incident light is the peak. On the other hand, since the diffracted light due to the wavelength input light is far away from its focal point, its light intensity 32' is very weak, resulting in a low and gentle light intensity distribution as shown by the solid line in the figure. (c) of the figure shows the light intensity distribution at the imaging surface 6 of the diffracted light due to the wavelength input light and the wavelength input light from the mask mark and the wafer mark.
Since the light of each wavelength is incoherent and does not interfere with each other, the light intensity distribution at It will be done. In other words, the light intensity is expressed as the addition of (diffracted light intensity at the imaging plane 6) = Because of this addition, the gain of the peak signal used as the alignment signal obtained by t'J from the linear sensor 6 is doubled. As a result, the S/N ratio is also improved.

Next, the action of the cylindrical lens 7 in tJS8Lilf shown in FIG. 1 will be explained. When the Fresnel diffraction j and 1% point images of the enlarged mask mark 16 and wafer mark 17 are detected by a two-dimensional camera using the two-drying point detection system 12 using chromatic aberration, the detection speed of the camera is 30 Hz, so
It is difficult to increase the detection speed beyond that. Furthermore, since the focused image is a straight line extending in the X direction in the 81'A, and digital raster compression processing is also required, the processing speed is greatly reduced. Therefore, if this focal image is detected with a one-dimensional camera, that is, a linear sensor 6 made of a CCD or the like, instead of a two-dimensional camera, the detection speed can be dramatically increased to around 10 Hz. Since the linear sensor 6 is arranged in the direction perpendicular to the focal line (Y direction), only a part of the focal image with the straight line is detected by the linear sensor 6.
A straight line extending in the direction 4.1. A cylindrical lens 7 that compresses a point image in the X direction is used to enable more effective detection of a focused image.

I will explain this point using the 101st percentile. In this figure, one LFZP is shown as an example, and the objective lens 9 is omitted. By arranging the cylindrical lens 7 having a generatrix in the direction (Y direction) perpendicular to the stripes of LFZP16 or 17,
The light intensity distribution 33 of the linear focal point extending in the X direction of 7 is
When compressed in the direction, the distribution becomes as shown by reference numeral 34, and all the light can be made incident on the linear sensor 6 arranged in the Y direction. If the compression ratio of the cylindrical lens 7 in the It will be done. Furthermore, by using the cylindrical lens 7, raster compression that was previously performed digitally can be performed optically, and the raster compression time becomes virtually zero, and the detection speed required for raster compression and reliability detection is 1 OKHz. It becomes extremely fast forward and backward. Moreover, unnecessary white noise components introduced in the Lasque scan are also reduced, resulting in a high S/N ratio.

Regarding the use of the above-mentioned cylindrical lens, the present applicant has filed an application as "Alignment mark position detection device" (No. 1, No. 11/+63-63921).

Next, a signal processing method for processing the signal obtained by the linear sensor 6 to detect the position will be described.

One method is the similarity pattern matching method (application II/162-24319 filed by the same applicant).
(See No. 4) ;: (1)? Ji<:k, as shown in Fig. 11(a), two marks of the same shape are placed at a predetermined distance apart, and this is scanned in the direction of the mark spacing to generate an input as shown in Fig. 11(b). The image signal V(j) is li;:
0) A power ii! The if'l signal V(j) is differentiated and the same figure (
The edge extraction signal V'(j) as shown in c) is obtained, and the autocorrelation function A (W, P, R) defined by the following equation (5) is obtained.

-P V'(j+w)...(5) Here, P: starting point of the correlation interval R: Correlation interval W: Width of left and right pattern V' (j): Image at pixel j 1 of human power value V(j) Order differential a1 The distance between the marks in FIG. 11(a) is A (W).

P, R) take the maximum value W, and w2 take 27fr[''
exists in between.

In the conventional center position detection method using pattern matching, it is assumed that each mark is symmetrical as shown in FIG. However, the accuracy of center detection decreases. However, when using the above-mentioned method of similarity pattern matching, there is a method of determining the position based on the process characteristic that close marks are simultaneously destroyed while maintaining similarity due to the processing process. This makes it an extremely low-dependence method on semiconductor processing processes.

Now, the signal obtained from the linear sensor 6 is shown in Figure 12 (
In the case of the mask mark 16 as shown in FIG. , since two wafer marks 17 are used, two corresponding signals 41 and 43 appear, and their positional relationship is such that wafer mark signals 41 and 43 exist on both sides of the corresponding signal 42 of the mask mark 16. do. This output signal is digitized by an A/D converter and subjected to first-order differential processing to obtain a signal as shown in FIG. 12(b). In the figure (b), signals 44 and 46 correspond to wafer marks, and 45 corresponds to mask marks, respectively. For example, by applying the above-described similarity pattern matching method to the signal shown in FIG.
The positional relationship in the Y direction (FIGS. 2, 5, etc.) of the wafer 15 and the wafer 15 can be detected, and alignment of both becomes possible.

Here, the mask mark 1 is detected using the position detection device of the present invention.
The principle of aligning wafer mark 6 and wafer mark 17 will be explained below.Linear Fresnel zone plate (LFZP)
As explained above, has the same effect as a convex cylindrical lens having a focal length f. In FIG. 13, the LFZP 23 with stripes perpendicular to the plane of the paper (X direction) has a linear focus 24 in the same direction, so that the LFZP of the mask or wafer is perpendicular to the stripes (Y direction). When moving by 66μm, the focal point also moves 66g in the same direction.
M moves. The amount of movement of this focal point is calculated as the amount of deviation of LFZP (Δδu
m).

Therefore, by detecting the focus shift hl, it is possible to know the amount of shift of the mask or wafer. Therefore, by providing a mark made of LFZP on each of the mask and the wafer, and simultaneously detecting the position of each focus, the mask or wafer shift amount can be determined. Mark and wafer mark alignment can be performed. In this invention, LFZP
multiple (principle) - does not necessarily have to be limited to two colors)
At the same time, the LFZP focal position differs depending on the wavelength, which is corrected by an objective lens with chromatic aberration, and the images are focused on the same focal plane. (N), and the image is formed on a linear sensor scanning in the Y direction, as shown in FIG.

Mask mark 1 on the imaging plane (linear sensor) 6
6 or the movement amount of the focal point of the wafer mark 17 is (ΔδxN
) gm. This movement fit is obtained by simultaneously processing the signals from the linear sensor 6 for the mask 14 and the wafer 15, and the center positions of the marks are made to coincide with each other for both.

The cylindrical lens 7 compresses the focal point of the LFZP 16° 17 extending in the X direction in this direction to increase the processing speed, amplify the signal strength, etc.

Here, I would like to make some additional additions. First, the 6th
In the diagrammatic explanation, the X-
It has been stated that the focal planes 25 and 26 on the Z+ plane can be considered to be a plane perpendicular to the optical axis of the detection optical system in the range where the incident angle and detection angle are not very large, but when 0 becomes large, this is not the case. The focal plane of marks 16 and 17 must be considered as a plane parallel to the mask or wafer surface, as shown at 25' and 26'' in FIG. 6. Therefore, the optical system after objective lens 9 is It is necessary to make some modifications to the arrangement shown in Fig. 1.The easiest way is to change the optical axes of the objective lens 9, relay lens 8, and cylindrical lens 7 to the mask 14.
What is necessary is to rotate it around the Y direction so that it is perpendicular to the plane of the wafer 15.

Next, regarding the number of slits in the LFZP, as reported in Reference 15, the intensity of one diffracted light depends on the number of slits in the LFZP, so it is necessary to determine an appropriate number of slits. In addition, there are two types of LFZP: clear center type and dark center type, so you need to choose one of them.
When determining the type, it is necessary to determine the appropriate number and type through experiments while looking at the effects of the treatment process.

FIG. 16 shows changes in signal strength due to differences in the number and type of LFZP slits.

In addition, in this invention, the detection signal is obtained by receiving the Fresnel diffracted light intensity with a linear sensor and converting it into an A/D signal. An example similar to the converted signal is shown in FIG. 17 (
Reference 13). As described above, the signal after A/D conversion is subjected to first-order differentiation, similarity pattern matching, interpolation calculation, etc.

The JliWjR specifications of the alignment device shown in FIG. 1 using the position detection device of the present invention are as follows.

1. Alignment mark wafer mark = LFZPx
2 Mask mark = LFZPx 1 2, detection signal: wavelength incident on the focusing focal plane due to Fresnel diffraction, and diffracted light intensity 3 due to light of λ2, detection device +j! 1: Utilizing chromatic aberration due to multiple wavelengths 2
Optical raster compression optical system 4 using heavy focal point detection device and cylindrical lens, light receiving element: linear sensor 5, signal processing method: position detection of alignment mark by similarity pattern matching, etc.Next, the position detection device of the present invention is used. The following describes how the problems No. 1 to No. 6 of the various conventional alignment methods shown in Table 1 are solved by the alignment method.

No. 1 Process dependence (No. in Table 1)
It corresponds to (Turn below) ■? Light interference problem due to n-wavelength alignment light.

According to this invention: 1. Since the Fresnel diffracted light with a double-continent wavelength is detected by a bifocal detection system that utilizes monochromatic aberration, it fundamentally solves the light interference problem of the conventional alignment method using a diffraction grating.

An example of the interference problem that occurs in the 5-diffraction grating method is given in Reference 113. In the alignment method using a circular Fresnel zone plate at the Bel Institute, the thickness of the thin film causes an interference effect, which causes the intensity of the Fresnel diffracted light to decrease. It has been reported that the temperature changes (see Table 1 of Reference 13).

Furthermore, it has been reported that the intensity of diffracted light changes sinusoidally for various thin films depending on the illumination wavelength (see FIG. 2 of Document 13).

Furthermore, in Document 19, in an alignment method using single-diffraction light, the change in interference intensity is kept small by making the resist film thickness uniform, thereby improving alignment accuracy.

(It depends on the symmetry of the remark.

According to this invention, this problem is basically solved by combining the following techniques.

(a) Use the LFZP bare, and set the position as the position of the alignment mark.

(b) Similarity pattern matching is applied as arithmetic processing for determining the position.

Next, the dependence of mark symmetry solved by the combination of the above two techniques will be explained in relation to the prior art.

First, we will introduce alignment methods using circular Fresnel zone targets that are indirectly related to IJI, and explain that they depend on the symmetry of marks.

Circular Fresnel Zone proposed by Be1l Lab.
The method using targets is reported in Reference 13,
This method is the best! As shown in FIG. 8(a), two bears of circular Fresnel zone targets 50 are provided on the mask 7 at the center thereof.

This is intended to align the center of the circular Fresnel zone provided on the wafer L and the target 51.

In addition, Mic that follows the method of the B e 1.1 laboratory mentioned above.
An X-ray aligner for RAN I
The objective is to align the center of a circular Fresnel zone target (51) provided on the wafer with the center of the bare wafer.

As is clear from these FIGS. 18(a) and (b), since one circular Fresnel zone target is used as the wafer alignment mark, the center position of the wafer alignment mark changes due to the influence of the process. Alignment accuracy will always be affected.

Taking the case of FIG. 18(b) as an example, how the alignment accuracy depends on the symmetry of the wafer mark in the detection signal from the alignment mark will be described below. 19(a), (b), and (C) show detection signals when the symmetry of the wafer mark changes. As shown in FIG. 19(a), the detection signal 53 from the wafer mark 51 maintains the symmetry. , since the center position 55 of the mask mark and the center position f156 of the wafer mark match,
There is no detection error. On the contrary.

As shown in (b), the symmetry of the wafer mark may be broken and it may shift to the left. This means that, for example, the intensity of the diffracted light from the wafer mark will not lose its symmetry due to the influence of the process.
As shown in (b), the center position 2 is to the left of the original center position.
! 56 is shifted, and (b) an alignment error shown by reference numeral 57 in the figure occurs. (C) is the same as (b), when the symmetry of the wafer mark is broken and it shifts to the right.
For the same reason as in (b), an alignment error 57 occurs in the right direction.

In contrast to these conventional methods, the present invention uses a third method.
As shown in the figure, two bare wafer marks 17 are provided. Taking the case of FIG. 3 as an example, the reason why alignment accuracy is extremely difficult to depend on the symmetry of marks will be explained. Figure 20(a).

(b) and (C) show the case where the symmetry of the detection signal from the wafer mark changes.
b) and (c) are the changes fiP, R in equation (5) of similarity pattern matching explained in IJ1 with reference to FIG.
, W are shown. As shown in (a), when the detection signal 59 of the wafer mark 17 (FIG. 3) maintains symmetry, what is the center position of the wafer mark that is bare with the center position 61 of the mask mark (one)? Match. On the other hand, as shown in (b), when the symmetry of each wafer mark is broken and shifted to the left, the center position 62 of each mark is shifted to the left. The way the symmetry of the left and right marks is broken changes while maintaining the similarity because the marks are close to each other. Therefore, we transformed it while maintaining this similarity,
When applying the similarity pattern mounting method to the left and right marks that have become asymmetrical, the center positions of the left and right bare marks will be as shown in Figure 2 (Fig.
a) It is found that it exists near position ffl (position 61) in the figure and is not easily affected by the asymmetry of the left and right marks. That is, according to the correlation formula expressed by the following equation (5), <j + w) ... (5) Here, P: starting point of the correlation section 1111 R: correlation section W: detection of the left side of the width of the left and right pattern When the signal is moved parallel to the right, it overlaps with the detection signal on the right (mathematically equivalent to a correlation operation), and the position of best match represents the distance of symmetry between the left and right patterns.

High dependence on N002 gap.

Basically, the gap dependence in the present invention is as follows.

It is determined by the depth of focus of the objective lens of the detection device that uses chromatic aberration. In other words, if the numerical aperture of the objective lens is NA, the detection accuracy is not affected by the gear amplitude variation as long as the gap variation is within the Rayleigh focal depth defined by the following equation (6).

However, when input: illumination wavelength For example, input = 546 nm (e-line), NA
=0.4, the depth of focus is ±1.5 pm and is not affected by gap fluctuations within that range.

The 21st [ffl] shows the relationship between the focal plane of Fresnel diffraction and the depth of focus of the objective lens. In this figure, LFZP 1 provided on a mask 14 and a wafer 15 (FIG. 3) J:,
The position of the diffraction focal plane 25, 6 or 17, changes in response to changes in the optical axis direction of the mask and the wafer, respectively. Therefore, the detection surface of the diffracted light is fixed at one point (
- plane), the detection intensity is greatly affected by the gap and constantly fluctuates, and the alignment accuracy becomes extremely dependent on the gap, so the accuracy becomes unstable and decreases. However, the objective lens 9 has the formula (6)
If there is a depth of focus 63 expressed by , and there is an object point within that range, the optical detection and resolution waterfall can be considered constant, so
Even if the diffraction focal plane 25 changes within the focal depth 63, the diffracted light intensity distribution detected on the imaging plane does not change.

As can be seen from the above description, according to the present invention, a detection device with low gap dependence can be realized in the same way as pattern measurement, even though a single diffracted light is used.

No. 3N Servo during exposure by detecting alignment marks provided in the light area 1ljlσ1 As is clear from the device external configuration shown in FIG.
Since the oblique detection method is adopted, an alignment mark is provided within the exposure area, and the mark is aligned with the exposure wavelength (Xl
It is possible to detect even during exposure without interfering with a). Therefore, the wafer position can be detected and corrected without retracting the detection optical system during exposure.
Since the entire wafer can be subjected to slap-and-repeat exposure, alignment accuracy and throughput can be significantly improved.

As an example in which the arrangement of the detection optical device is similar to that of the device of the present invention, there is a device by Hitachi, Ltd. that performs side detection, as shown in 1 Document 3. This has an arrangement as shown in 22nd IA, and uses pattern measurement and a normal epi-illumination microscope. On the other hand, in the present invention, a detection device mainly based on LFZP (diffraction method) and chromatic aberration is used, and in this point it is greatly different from that in Document 1, but the detection optical system 67 in FIG. If the position detection system 12 (FIG. 1) using chromatic aberration is replaced, the alignment mark is replaced with LFZP, and the image detection system 68 is replaced from a two-dimensional camera with a one-dimensional linear sensor, the system shown in FIG. The external shape will almost match that of the device.

N014 High-speed detection As explained in relation to Figure 1O of the technology that enables high-speed detection, by combining a position detection device that uses monochromatic aberration with a cylindrical lens and a near sensor, the detection speed can be increased to several tens of tens of degrees. It is possible to more than double the amount. For details on the method of combining a cylindrical lens and a linear sensor in a position detection device using chromatic aberration, refer to the specification of Japanese Patent Application No. 63-63921 ``Alignment Mark Position Detection Device''.

In the conventional method using Fresnel diffraction, the shape of the NO.5 mark is complex and pattern changes are likely to occur due to process changes, as described in Reference 13.1.
A circular Fresnel zone target (plate) was used as shown in Figure 4. FIG. 23(a) shows a plan view of such a circular Fresnel zone plate. On the other hand, in the present invention, a linear Fresnel zone plate as shown in FIG. 23(b) is shown. , the mark shape is considerably simplified.

Pattern changes based on process changes are less likely to occur.

No. 6 Regarding this problem, the asymmetry of the focal point caused by the spurious image generated near the focal point of Fresnel diffraction causes a position error. Since it is included in the dependent problems, it can be explained in the same way. In other words, the collapse of the mark symmetry due to the spurious image makes the detection signal asymmetric, similar to #'2F due to the process, so basically the following The problem is solved by two techniques.

(a) Use LFZP bare and find the position of the alignment mark from that position.

(b) Similarity pattern matching is applied as arithmetic processing to find the relative position of the alignment mark.

The solutions to the problems of the embodiments of the present invention and the conventional examples have been described above, but as is clear from the above explanation, the present invention is applicable to In the aligner, the mask and wafer are
It can be applied not only to alignment devices that can detect positions on the order of 1 gm or more, but also to aligners that use a light source other than X-rays and perform proximity exposure, and that require precise positioning of the mask and wafer. In addition, it can be applied to a device that precisely aligns two relatively close objects (several 10 meters to several hundreds of grams). This issue 11
is used as a mask and wafer position detection device in an X-ray aligner, enabling high precision and high throughput. Further, it is used as a means for obtaining position signals for servo systems such as an x-y stage and a mask stage.

[Effects of the Invention] As explained in detail in the embodiment, the effect of this IJI is that it can solve the problems of Nos. 001 to 6 of the prior art in the l's1 table. especially.

(1) Possible to perform servo control during exposure to detect alignment marks provided within the exposure area (2) Extremely low process dependence ■ Does not depend on mark symmetry ■ Standing wave effect due to light interference in thin films It is extremely effective in that (3) gap dependence is extremely low and (4) high-speed detection is possible.

Therefore, from more than 1, according to this statement,
Ultra-precise position J-(on the order of 0.01 m), in particular, it is possible to detect the positions of masks and wafers on the order of 0.014 m, and high throughput is possible.

[Brief explanation of the drawing]

FIG. 1 is a schematic side view of a position detection device according to the present invention. Figure 2 is a schematic top view of the machine. FIGS. 3(a) and 3(b) are a top view and a side view of the alignment mark, and FIG. 4 is an explanatory diagram of the action of LFZP. 5 and 6 are explanatory diagrams of Fresnel diffraction of an alignment mark made of LFZP. FIG. 7 is an optical path diagram showing the effect of an objective lens having monochromatic aberration. FIGS. 8(a) and (b) are explanatory diagrams of the relationship between the Fresnel diffraction focus and the bifocal focus detection device. FIGS. 9(a), (b), and (c) are the Fresnel focal plane and the imaging plane. FIG. 10 is an explanatory diagram of optical raster compression using a cylindrical lens. FIGS. 11(a), (b), and (c) are explanatory diagrams of similarity pattern matching. 12(a) and 12(b) are waveform diagrams of the detection signal and its first-order differential signal, FIGS. 13 and 14 are explanatory diagrams of the alignment principle of the present invention, and FIG. 15 is an illustration of other implementations. A drawing similar to FIG. 1 in the example; FIG. 16 is an explanatory diagram showing changes in signal strength due to differences in the number and type of LFZP slits; FIG. 17 is a detection signal waveform diagram after A/D conversion. FIGS. 18(a) and 18(b) are layout diagrams of mask marks and wafer marks in a conventional example. 191″A (a), (b), and (c) are waveform diagrams of detection signals in the case of the conventional example, and Figure 20 (a), (b), and (c) are detection signals in the case of the present invention FIG. 21 is an explanatory diagram of the gap dependence of the present invention. FIG. 22 is a perspective view of a conventional aligner, and FIG.
a) and (b) are one circular Fresnel zone plate and LF
It is a comparison diagram of ZP. l: laser (wavelength λl), 2: beam splitter, 3
: Laser (wavelength input a), 's: Mirror 5: Illumination optical device, 6: Linear sensor, 7: Cylindrical lens,
8: Relay lens, 9: Objective lens, 1O: Detection optical device, 11: High-speed detection system without linear sensor and chromatic lens, 12: Bifocal detection system using chromatic aberration. 13: X-ray exposure area, 14: Mask, 15: Wafer-1
6:7 alignment mark = LFzP. 17: Alignment mark = LFZPx2° 18: Gap, 19: Oblique incident illumination angle (0), 20: Oblique detection angle (0), 21: Opaque area, 22: Mask window (
transparent part)

Claims (5)

[Claims] (1) In a position detection device that detects the relative position in a direction perpendicular to the optical axis of a first object and a second object that are separated by a small distance in the optical axis direction of X-ray exposure light, a linear Fresnel zone is placed on each object. An alignment mark consisting of a plate (LFZP) is provided, and an illumination device is configured so that these LFZPs are illuminated with light of a plurality of wavelengths from the same direction at the same time, and the focal point of the wavelength is set at a diffraction focal point at a different position for each wavelength of the LFZP. An objective lens having a chromatic aberration such that the A cylindrical lens is provided on the image forming surface for one-dimensionally operating a linear sensor to convert the linear sensor into an electrical signal, and a cylindrical lens is provided so that the linear diffraction focal image is compressed in the longitudinal direction and imaged on the linear sensor. is arranged between the objective lens and the linear sensor, and has means for processing the signal obtained from the linear sensor to detect the position of each of the alignment marks. Position detection device with double linear Fresnel zone plate. (2) A single L as an alignment mark for the first object
A pair of LFZPs are used as alignment marks of the second object, and the single LFZP of the first object is arranged so as to fit between the pair of LFZPs of the second object. The position detection device according to claim 1. (3) A claim characterized in that the first object is a mask, the second object is a wafer, and a transparent window area is provided in a portion of the mask through which diffracted light and incident light of an alignment mark on the wafer pass. 2. The position detection device according to 1 or 2. (4) The illumination device and the substitute lens are arranged in opposing positions diagonally above the first object alignment mark and the second object alignment mark, and the incident angle of the illumination device is set so that the angle of incidence of the illumination device is 4. The position detection device according to claim 1, wherein the position detection device is arranged to be equal to a detection angle of a detection optical system including an objective lens. (5) The device according to any one of claims 1 to 4, wherein the means for detecting the position of each alignment mark by processing the signal obtained from the linear sensor is a means for performing a similarity pattern matching process. Position detection device.