JP2550976B2 - Alignment method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial field of application) The present invention is an alignment for transferring a circuit pattern or the like formed on a mask (or reticle) to a substrate sensitive to ultraviolet rays, X-rays or the like by the step-and-repeat method. Regarding the method.

(Prior Art) Conventionally, when a circuit pattern is exposed on a wafer by a step-and-repeat method, a pattern portion to be exposed of a mask or reticle (hereinafter collectively referred to as a mask) and one shot area on the wafer are previously formed. Must be accurately aligned. In particular, in the case of the step-and-repeat method, since a plurality of shot areas to be exposed are formed in a matrix on the wafer, when the wafer is stepped, it is accurately aligned with the next shot area. Is required. Currently, a reduction projection type exposure apparatus is mainly used as an exposure apparatus used for manufacturing a semiconductor device, and a pattern of a mask (reticle) is superposed as an image on one shot area on a wafer through a projection optical system. Then, as an alignment method of this type of apparatus, the alignment mark on the mask and the alignment mark not associated with the shot area on the wafer are simultaneously observed by an alignment optical system provided above the mask, and the positional deviation amount of both marks is determined. A method is known in which a wafer or a mask is finely moved so as to be zero. In addition, the alignment mark on the wafer is detected only through the projection optical system on the assumption that the mask is positioned at a predetermined position with high accuracy with respect to the optical axis of the projection optical system, and the alignment mark is projected. A method of moving and stopping the wafer so as to be arranged at a predetermined position in the image plane is also known. Particularly, according to the former alignment method, since the mask and the wafer are directly observed by the exposure apparatus, there is an advantage that an offset hardly occurs.

(Problems to be Solved by the Invention) However, in general, the projection optical system is satisfactorily corrected for chromatic aberration only for illumination light for exposure (eg, g-line of wavelength 436 nm, i-line of wavelength 365 nm). In the method of simultaneously observing the mask and the wafer through the projection optical system, the wavelength of the illumination light for alignment is the same as the wavelength of the exposure light,
Alternatively, there is a problem that the wavelength is limited to the vicinity. In other words, when the illumination light for alignment becomes coherent light or quasi-monochromatic light, unexpected noise components (speckles, interference fringes, etc.) are generated during mark detection on the wafer surface with a complicated layer structure (resist layer, etc.) There arises a problem that the optical information from the light source cannot be correctly detected. Therefore, a proposal to make the illumination light for alignment a plurality of wavelengths,
There are proposals to continuously change the incident angle of the alignment illumination light (monochromatic light) on the wafer surface, but in either case, it is undeniable that there will be a problem in the device configuration such that the configuration of the alignment system becomes complicated. .

(Means for Solving Problems) Therefore, the present invention provides a method for enabling highly accurate alignment when the marks are illuminated for alignment.

In the present invention, the marks associated with the shot areas on the sensitive substrate (wafer) are composed of n kinds of mark patterns having different shapes (for example, two kinds having the same planar shape and concave and convex cross-sectional shapes). Then, the optical system for detecting the mark on the wafer selects one of m different detection modes (for example, two types of bright field detection and dark field detection) to detect the alignment state. Is configured as follows.
Further, the plurality of shot areas on the wafer are divided into n × m (for example, 2 × 2) groups, and the mark detection is executed by changing the combination of the n kinds of mark patterns and the m kinds of detection modes for each group. Then, an optimal combination (for example, dark field detection with a concave mark pattern) was selected and applied to the alignment of the wafer to be processed next.

(Operation) According to the present invention, since the same alignment optical system is used to switch to different detection modes and different mark patterns to detect the alignment marks on the wafer (or mask), the accuracy of mark position detection by noise components can be improved. The less degraded detection modes and mark patterns can be selected for application to subsequent wafer processing.

(Embodiment) FIG. 2 is a perspective view showing the configuration of a projection exposure apparatus suitable for applying the method according to the first embodiment of the present invention. In FIG. 2, the exposure light from the exposure illumination system 1 uniformly illuminates the reticle R via the main condenser lens 2, and the pattern on the reticle R is projected onto the wafer W via the projection lens 3. The wafer W is mounted on the stage 4 which moves two-dimensionally by the step-and-repeat method. The stage 4 is provided with a reference mark FM having the same pattern shape as the alignment mark formed on the wafer W.
Three wafer alignment microscopes 5Y, 5θ, 5X are provided around the projection lens 3 for global alignment of the wafer W.
Is arranged. Also, oblique incident light type focus detection systems 6a and 6b which obliquely irradiate the surface of the wafer W with an imaging light flux (non-photosensitive wavelength) and photoelectrically detect the position of the reflected light to detect the height position of the wafer surface are also provided. It is provided. Further, each of the wafer alignment microscopes (hereinafter referred to as WAM) 5Y, 5θ, and 5X enters the surface of the wafer W by injecting light from the illumination system 7 that generates non-photosensitive illumination light to the resist on the wafer. Illuminate. The image sensor 8 observes the alignment marks detected by WAM5Y, 5θ, and 5X, respectively. WAM5Y and 5θ detect alignment marks for the y direction on the wafer, and WAM5X shows x.
Detect alignment mark for direction. Now, between the reticle R and the main condenser lens 2, reticle alignment microscopes R _xy and Rθ for positioning the reticle R with respect to the apparatus are provided. Further, between the reticle R and the main condenser lens 2, a step alignment of a TTR (through the reticle) method for simultaneously observing the mark on the reticle R and the back projection image of the mark on the wafer W by the projection lens 3 is simultaneously observed. Microscopes (hereinafter referred to as SAMs) 10a and 10b are provided. The SAM 10a detects a shift in the y direction between the reticle R and one shot area on the wafer W, and the SAM 10b detects a shift in the x direction between the reticle R and the shot area. The images of both marks on the reticle R and the wafer W are observable by the two projection elements 11 and 12, and the illumination light (the same wavelength as the exposure light) to both marks is transmitted from the illumination system 13 through the SAMs 10a and 10b. Sent. The image sensor 11 is for observing the bright field images of both marks, and the image sensor 12 is for observing the dark field images of both marks. The pupil of the projection lens 3 is in the optical path from the SAMs 10a, 10b to the image sensor 12. A spatial filter 14 that blocks specular reflection light is provided at a position conjugate with. Therefore, the illumination light from the illumination system 13 is determined so as to pass through the pupil center of the projection lens 3 with a size smaller than the pupil diameter. The illumination system 13 may be a He-Cd laser light source when the exposure light is g-line.

FIG. 3 (A) shows the shape of marks provided in association with each shot area of the wafer W shown in FIG. 2 and comprises a pair of mark patterns WMa, WMb. Mark pattern WMa
And WMb are separated from each other by a fixed distance (for example, 30 μm) and have the same shape in plan view, but the cross-sectional shape is defined as a complementary shape of convex and concave.

Further, FIG. 3B shows a planar shape of a mark RM provided around the reticle R, and here, it is composed of two parallel bar patterns (light-shielding parts), and a transparent part inside the bar pattern is formed. With the mark pattern WMa or WMb of the wafer W positioned, the amount of deviation between the reticle and the wafer is detected.

Next, the operation of this embodiment will be described with reference to the flow chart shown in FIG. This flowchart is executed after the global alignment of the wafer W is completed by the WAMs 5Y, 5θ, and 5X shown in FIG. 2. Here, a part of the periphery of the rotation pattern region of the reticle R is on the wafer W. A vernier pattern for verifying the overlay with the shot area is formed, and when the reticle R is illuminated by the illumination system 1, this vernier pattern has a predetermined accuracy with a vernier pattern provided in advance around the shot area on the wafer W. They are arranged so that they are superimposed and transferred.

Now, as shown in FIG. 4, a plurality of shot areas SA are arranged in a matrix on the wafer W. Here, the shot arrangement on the wafer W is four as shown by the imaginary line in FIG. Blocks (groups) A, B, C and D are divided. Then, for the shot area SA included in the A block, the TTR alignment system is aligned (mark detection) in the dark field detection mode using the image sensor 12, and for the shot area SA included in the B block,
Initialization is performed so that marks are detected in a bright field detection mode using the image sensor 11 (step 100). Of course, the C block and the D block are similarly initialized. Symbol BL in each shot area SA in FIG. 4 means bright field (bright field) detection, and symbol DK means dark field (dark field) detection.

Next, at step 101, which of the mark patterns WMa and WMb is to be used among the marks WM associated with each shot area is designated for each of the four blocks A, B, C and D. As a result, alignment by all combinations (4 ways) of two detection modes and two mark patterns WMa, WMb is possible on one wafer W. In FIG. 4, the symbol projections in each shot area SA are mark patterns.
It means WMa, and the symbol concave means the mark pattern WMb.

Next, in step 102, alignment with the reticle R is performed in the alignment mode designated for each shot area, and the shot area is exposed. When exposure for all shots on one wafer is completed in this way, the wafer W is taken out and developed in step 103.

When the development is completed, a resist image of a vernier pattern is formed in association with each shot area of the wafer W according to the overlay accuracy at the time of exposure. The overlay accuracy of this vernier pattern is detected by another measuring device or an alignment sensor (for example, WAM5Y, 5X, etc.) included in the exposure apparatus itself (step 104). Then, in step 105, the absolute value of the overlay deviation amount and the average deviation value σ are calculated for each of the four blocks A, B, C, and D. As a result, for example, in each block, the sum (± 2σ or ± 3σ) of the absolute value and the deviation σ is calculated as shown in Table 1.

As a result, it can be seen that the alignment mode applied to the D block, that is, the method of detecting the dark field of the mark pattern WMb (concave pattern) is the most accurate (step 106). Therefore, when the same wafer is processed thereafter, the alignment mode with high accuracy is automatically set.

If the vernier pattern (resist image) on the wafer can be sequentially detected by the alignment sensor in the device,
Since steps 104, 105 and 106 can also be automatically executed by the apparatus, except for step 103, the apparatus itself determines the optimum alignment mode.

By the way, in the present embodiment, the checking operation as shown in FIG. 5 is also executed so as to verify the optimum value for the focus state at the time of alignment. The wafer W at this time is the same as the previously used wafer. The reticle R may be attached as it is. The flow chart of FIG. 5 is also executed after the global alignment of the wafer, but since the optimum mark shape and mark detection method (alignment mode) are determined here, each shot area on the wafer W is selected only in that alignment mode. On the other hand, alignment and exposure are performed. First, in step 110, it is designated to change the focus offset value for each row of the shot array as shown in FIG. That is, an offset of −2 μm for the first row, an offset of −1 μm for the second row, an offset amount of 0 for the third row, an offset of +1 μm for the fourth row,
An offset of +2 μm is applied to the fifth row. Here, the focus offset refers to the height position of the wafer W at which the surface of the wafer W is detected as the focal point by the oblique incident light type focus detection systems 6a and 6b as a reference, and the direction approaching the projection lens 3 from there is negative. That is, the wafer W is finely moved with the direction of moving away set to be positive. This fine movement is performed by the Z stage provided on the stage 4.

Next, under the focus offset value designated in step 111, TTR alignment is performed for each shot area, and exposure is performed by the step-and-repeat method. Next, in step 112, the wafer is developed, and in step 113, the vernier pattern (resist image) for each shot area is detected and the overlay error amount is read. Then, in step 114, the sum of the absolute value of the deviation amount and the average deviation value σ (± 2σ or ± 3 for each row shown in FIG. 6 is performed.
σ) is calculated. As a result, the overlay accuracy for each row is calculated as shown in Table 2.

As a result, the focus state has an offset value of 0 μm.
Alternatively, it can be seen that the best alignment accuracy can be obtained by setting +1 μm. This offset value is automatically set as an alignment condition when processing similar wafers thereafter (step 115).

Incorporating this focus state as an alignment condition minimizes the influence of the telecentric error and aberration of the projection lens 1 and the alignment optical system of the TTR system, and the alignment error due to the quality of the observed image of the mark WM on the wafer W. It means to minimize the occurrence.

As described above, in the embodiment of the present invention, as the TTR type alignment optical system, the type in which the image of each mark is picked up by using the illumination light having the same wavelength as the exposure light is shown, but the present invention is not limited to this. For example, even in a method (so-called dark-field detection) in which a spot light of a laser beam is scanned on a reticle and a wafer and scattered light or diffracted light generated from each mark is photoelectrically detected on a pupil conjugate plane of an optical system. Two detection modes can be selected by providing a photoelectric detector for detecting a visual field (detection of specular reflection light). Further, the present invention can be similarly applied to an alignment system in which a laser beam is transmitted to the wafer not through the reticle R but through the projection lens 3 to form spot light on the wafer. Also in this case, specular reflection light or scattered light (diffracted light) from the mark on the wafer irradiated with the spot light may be selectively detected. At this time, since the reticle and the wafer do not have to be conjugated with each other under the laser beam for illumination, a wavelength different from the exposure light can be used. When using illumination light with a wavelength different from the exposure light for TTR type alignment, an optical system such as a small lens for chromatic aberration correction, a mirror, a prism, etc. may be inserted in the alignment optical path between the reticle and the projection lens. The present invention can be utilized as well. Besides, the present invention can be similarly applied to an off-axis type alignment system (WAM5Y, 5θ, 5X, etc.) that does not use a projection lens. In this case, it is necessary to increase the stroke of the stage 4 so that the marks WM in all shot areas on the wafer can be detected by the WAMs 5Y and 5X.

On the other hand, with respect to the pattern shape of the mark WM on the wafer, the same effect can be obtained even if the planar size is different or the planar shape itself is different. For example, a pair of large and small mark patterns having similar planar shapes and different sizes, a pair having different pattern widths, or a bar pattern having one linear edge and the other having a diffraction grating pattern can be used. Needless to say, the mark patterns may have two or more different shapes. Furthermore, the mark detection mode by the alignment optical system is bright field,
Other than the dark field, a method of selecting whether image detection is pupil detection may be used. Image detection is to detect an image of a mark and to arrange the image sensor and the scanning slit shown in FIG. 2 on the image forming plane. Pupil detection is to be a spatial filter or photoelectric detector on the pupil conjugate plane of the optical system. Is arranged to detect the distribution of light information from the mark on the pupil.

The present invention can be similarly applied to a step-and-repeat type (or step-and-scan type) exposure apparatus other than the projection type exposure apparatus, such as a proximity X-ray stepper.

As described above, according to the present invention, since the alignment condition is changed for each group on one sensitive (photosensitive) substrate, it is easy to compare the alignment accuracy obtained by each condition (combination). Therefore, the optimum alignment condition can be immediately obtained. Further, an error that varies from wafer to wafer (for example, wafer rotation) also appears as an offset in each alignment, but the offset value is a σ value (average deviation) because it is a measurement of a deviation amount within one substrate. However, it can be obtained with accurate accuracy.

In addition, if the method of deciding the shot area for grouping is changed appropriately, the run
There is also an advantage that the dependence on out (stretching) can be reduced. Furthermore, since the exposure apparatus can be self-checked, there is an effect that the specified alignment accuracy can be maintained for a long time.

[Brief description of drawings]

FIG. 1 is a flow chart for explaining the operation according to the embodiment of the present invention, FIG. 2 is a perspective view showing the configuration of a projection exposure apparatus suitable for the embodiment of the present invention, and FIG. 3 (A) is a mark on a wafer. FIG. 3B is a diagram showing the shape of the pattern, FIG. 3B is a diagram showing the shape of the marks on the reticle, and FIG. 4 is a plan view showing the arrangement of shot areas on the wafer that determine the alignment conditions.
FIG. 5 is a flow chart for explaining the operation for determining the focus state, and FIG. 6 is a plan view showing the arrangement of shot areas for determining the alignment condition including the focus state.

Claims (3)

(57) [Claims] 1. Prior to sequentially superposing and exposing a plurality of shot areas on a sensitive substrate on a pattern on a mask, an alignment mark formed accompanying the shot areas is optically detected to detect the alignment marks. In the method of aligning a pattern of a mask and a shot area of the sensitive substrate, the marks are composed of n kinds of pattern marks (n is an integer of 2 or more) having different shapes, and optical systems for detecting the marks are different from each other. m types (m is an integer of 2 or more)
One of the detection modes is selected to detect the alignment state, the plurality of shot areas on the sensitive substrate are divided into n × m groups, and each of the n types is divided into n groups. The alignment method is characterized in that the mark detection is performed by changing the combination of the mark pattern and the m types of detection modes, the optimum combination is selected and applied to the alignment of the sensitive substrate to be subsequently processed. 2. The alignment accuracy when the focus state of the optical system with respect to the mark is changed after the optimum combination is selected, and the focus state with good accuracy is set as an alignment condition. The method according to claim 1. 3. The m types of detection modes include at least two types, a mode of detecting light information from the mark in a bright field and a mode of detecting light information in a dark field. The method according to claim 1.