JPS63147273A - Method for detecting alignment pattern - Google Patents

Abstract

PURPOSE:To obtain the aligning pattern signal as a diffracted light signal having a high S/N regardless of the thickness of a transparent layer, by irradiating a latent image formed on the transparent layer at a prescribed angle for the detection of said image. CONSTITUTION:A laser beam source 21 supplies the linear polarized light and this luminous flux is turned into a circular polarized light through a polarized beam splitter 22 and a lambda/4 plate 23 to be made incident on a wafer 4 to which a latent image is formed. Here it is assumed that an incident angle theta'' is secured to the normal line of the wafer 4. Thus the luminous flux traces back the original optical path owing to a refractive index distribution and is made incident on the plate 23 to be turned into a linear polarized light. This luminous flux is reflected by the splitter 22 toward a convergent lens 24 and made incident on a photodetector 25. Then the angle theta' is set properly to secure the maximum signal intensity of the latent image. At the same time, an S/N is improved. Thus it is possible to obtain a refracted light signal having a high S/N regardless of the thickness of a resist 7.

Description

[Detailed description of the invention] [Field of invention] The present invention relates to a method for detecting patterns for alignment.

[Prior Art] Conventionally, one method of aligning a reticle to a square in a semiconductor exposure apparatus is to use a latent image.

To align using this method, first, as shown in FIG.
The imaging system 12 is illuminated by the printing light emitted from the illumination system IO.
The image is transferred to a resist, which is a photosensitive layer, on the wafer 13 through the wafer 13. At that time, the printing light is irradiated so that the resist is almost saturated. Then, a latent image as shown in FIG. 4 is formed. Next, this latent image and the mark previously formed on the Unihachi substrate 15 are scanned with a laser beam other than the printing light, and the relative distance between the latent image and the mark is determined by detecting the diffracted light. Perform alignment. This is an alignment method using latent images.

As shown in FIG. 4, the latent image has different refractive index nl+n2 and transmittance TI, T2 between the portion irradiated with the printing light (hatched portion) and the portion not irradiated with the printing light. The distribution is one-dimensional in the horizontal direction of the wafer substrate.

[Problems to be Solved by the Invention] However, when scanning a laser over such a latent image pattern and detecting the diffracted light, when the laser light is perpendicularly incident, the intensity of the diffracted light varies depending on the thickness of the resist. There is a drawback that the angle changes, and in the case of oblique incidence, the diffraction intensity is reduced compared to normal incidence, and a sufficient S/N ratio cannot be obtained.

The present invention eliminates the drawbacks of the conventional example described above and at the same time improves the S/N ratio.
It is an object of the present invention to provide a method for detecting alignment patterns that can obtain a high diffraction light signal.

[Means for Solving the Problems] In order to solve the above problems, the present invention provides an alignment pattern that is periodically repeated in a transparent layer at a predetermined angle with respect to the surface of the transparent layer, as an alignment pattern to be detected. A pattern made of layers having different light transmittances or refractive indexes is used, and each pattern is illuminated and detected at an angle defined by the predetermined angle.

[Operation] A preferable alignment pattern detected in the present invention uses, for example, a coherent light beam divided into two, and at least one of the two beams is illuminated with the alignment pattern on the first object. After imparting the graphic information, these light beams are irradiated onto predetermined portions of the photosensitive layer on the second object at different angles, thereby forming a latent image pattern for positioning on the first object.

In the present invention, the alignment pattern made of the repeating layers is detected by entering illumination light at a predetermined angle corresponding to the angle of the repeating layers in the transparent layer and detecting the diffracted light.

[Example] Hereinafter, the present invention will be explained in detail using the drawings.

FIG. 1 is a diagram showing a latent image detection method according to an embodiment of the present invention. In the figure, 4 is a wafer, 7 is a resist,
8 is a wafer substrate, 21 is a laser light source with a wavelength λ' different from that of the printing light, 22 is a polarizing beam splitter, and 23 is λ
/4 plate, 24 is a condenser lens, and 25 is a photodetector. In the resist 7 of the wafer 4, two coherent parallel light beams Σ1 and Σ2 are transmitted through a pattern for forming a latent image at different angles and then irradiated onto a predetermined portion of the wafer 4 via an imaging system. A distribution of refractive index and transmittance is formed corresponding to a distribution of interference fringes (intensity) formed by the light beams Σi and Σ2. This is the so-called latent image.

FIG. 2 is a cross-sectional view of the wafer 4 on which this latent image is formed. Here, a case will be considered in which the parallel light beam Σ1 is perpendicularly incident on the resist 7. The refractive index of resist 7 before exposure is nl
, the wavelength of the printing light is λ, and the angle between the luminous fluxes Σ1 and Σ2 is 20. According to Snell's rule of seven, the luminous flux Σ! ,
Σ2 is 2 that satisfies sin 2θ = n, 5in2θ'
It forms an angle of 0'. At this time, the fringe of the latent image consists of two light beams Σ1. This occurs at the portion connecting the intersection points of the respective wavefronts W, , W2 of Σ2, that is, the shaded portion of the drawing. As is clear from the figure, the fringe of the latent image is made up of a half angle of the angle formed by both light beams. In other words, the angle between the stripes and the normal to the wafer substrate 8 is θ'
becomes. Also, the fringe pitch is λ/2n1sin (θ
').

In this configuration, when linearly polarized light is supplied from the laser light source 21 (wavelength λ'), the light flux first passes through the polarizing beam splitter 22 . After that, 2 more λ/4 plates
By passing through the sheet metal 3, the light becomes circularly polarized light and enters the wafer 4 on which a latent image is formed.

At this time, θ' (where sin θ
” If the incident angle of =il, sin (re/ 2-θ') is increased, the light beam will reverse the original optical path due to the refractive index distribution and will be incident on the λ/4 plate 2 sheet metal 3, becoming linearly polarized light. This luminous flux is then reflected by the polarizing beam splitter 22 toward the condenser lens 24 and enters the photodetector 25. At this time, the luminous flux that enters each section in the figure from the laser light source 21 is expressed as il+12.13. Then, detector 2
The optical path difference of each light beam up to 5 is λ/sin θ'. Therefore, if θ' is determined so that the following holds true, the signal strength of the latent image will be maximized. Most of the surface reflection of the wafer 4, which causes signal disturbance, goes to the side opposite to the laser light source 21, so the S/N is low.
It also becomes more expensive.

As described above, the latent image formed by illuminating the pattern for forming a latent image with two coherent beams having the same wavelength as the printing light is illuminated with a laser beam having a wavelength different from that of the printing light, regardless of the resist thickness. It can be detected efficiently.

However, since the refractive index of the resist differs depending on the wavelength, it is necessary to adjust the incident angle of the illumination light when detecting a latent image to correspond to the wavelength.

[Modified example of the embodiment] In the above, the incident light and the diffracted light have the same path,
It does not matter if there is a certain angle between the same routes.

[Effects of the Invention] As explained above, according to the present invention, the signal of the alignment pattern can be detected very efficiently as a diffracted light signal with a high S/N, regardless of the thickness of the transparent layer. can.

[Brief explanation of the drawing]

FIG. 1 is a schematic diagram showing a latent image detection method according to an embodiment of the present invention, FIG. 2 is a cross-sectional view of a latent image pattern, FIG. 3 is a schematic diagram of a conventional device, and FIG. , is a cross-sectional view of a conventional latent image pattern. 7: resist, 8: substrate for sea urchins, 21: laser light source for latent image detection, 22: polarizing beam splitter, 23: λ/4 plate, 24: condenser lens, 25; photodetector. Patent Applicant Canon Co., Ltd. Agent Patent Attorney Tatsuo Ito Agent Patent Attorney Tetsuya Ito Figure 3

Claims (1)

[Claims] 1. A pattern consisting of layers having different light transmittances or refractive indexes formed periodically and repeatedly at a predetermined angle with respect to the surface of the transparent layer in the transparent layer, each of which is limited by the predetermined angle. A method for detecting alignment patterns characterized by illuminating and detecting at an angle at which the positioning pattern is detected. 2. The alignment pattern detection method according to claim 1, wherein the angle defined by the predetermined angle is the same for the illumination light and the detection light. 3. The transparent layer is a photosensitive layer, and the pattern is a first
The alignment pattern on the object is illuminated with the first coherent light and transferred to a predetermined portion of the photosensitive layer on the second object, and in parallel, the second coherent light is applied to the predetermined portion of the photosensitive layer on the second object. A latent image pattern for alignment of the first object is three-dimensionally formed on the photosensitive layer on the second object by irradiating the coherent light at a different angle with the first coherent light and causing interference with the first coherent light. A method for detecting alignment patterns according to claim 1.