JPH0827174B2 - Pattern detection method - Google Patents

Description

The present invention relates to a pattern detection method, for example, a highly integrated LS.
The present invention relates to a method of detecting a positioning pattern in a mask aligner (exposure device) used for manufacturing.

[Prior Art] Conventionally, for example, in a mask aligner or the like, a method of photoelectrically detecting a linear pattern or the like for alignment on a wafer is known.

However, since a photoresist is coated on the wafer, a large number of interference fringes appear when it is illuminated with single wavelength light. FIG. 6 (A) shows a number of interference fringes a that appear when a linear mark, which is a pattern for alignment, is illuminated with a single wavelength light.

When this is scanned in the direction of the arrow in the figure and converted into an electric signal, a video signal b as shown in FIG. 6 (B) is obtained.

Conventionally, in order to obtain the center from this signal b, an appropriate threshold value c is set in advance and the signal b
It was converted into the binarized signal d shown in FIG. 6C, and the center e of the binarized signal d, that is, the center position of the pattern on the wafer was obtained from this.

[Problems to be solved by the invention]

However, this method cannot be detected when the video signal 1b does not reach the threshold value c as shown in FIG. 6 (D) or when the video signal 2b is not completely symmetrical as shown in FIG. 6 (E). Alternatively, the binarized signal 1d of FIG. 6 (F) was obtained, and 1e of FIG. 6 was detected as the center position, and the correct center position could not be detected.

Further, as shown in FIG. 7 (A), the threshold value c of the video signal 3b is
When noise f is present in the vicinity or when the video signal 4d is slightly asymmetrical due to some defects in the linear pattern itself as shown in FIG. 8 (A), FIG. ) 2
Since the binarized signal 2d and the binarized signal 3d in FIG. 8 (B) were obtained, the correct center position could not be detected.

In order to eliminate these drawbacks of the conventional method, Japanese Patent Publication No. Sho 56-2284 discloses a method in which a video signal is folded from a certain point to obtain the degree of matching between both signals, and the point where the matching degree is the best is obtained as the center position. Is proposed. However, the image signals actually obtained are various as shown in FIGS. 9A to 9G due to uneven coating of resist, optical shading, difference in wafer material, etc. For example, it is the same as the video signal 3 shown in FIG. 4 (A), and the original signal is not required to have symmetry.

Further, when the linear pattern is viewed at right angles to the length direction, the contrast and the width of the linear pattern appear different as in the pattern 2 shown in FIG. 2 due to the uneven coating of the resist. Therefore, the center position to be obtained differs greatly depending on which position the arrow indicating the scanning direction of the pattern is brought to obtain the video signal in FIG. 6 (A).

Further, in the above publication, although the position and width of the measurement window are actually important, there is no description of a method for determining the position and width. It will be detected more often.

An object of the present invention is to eliminate the above-mentioned drawbacks of the conventional technique, to influence noise and asymmetry included in a video signal detected by an image pickup device, variations and distortions of pattern video on the same linear pattern, and sampling. An object of the present invention is to provide a method for accurately determining the center position of a pattern by minimizing the influence of the quantization error due to.

[Means and Actions for Solving Problems] In order to achieve the above object, the present invention provides a two-dimensional window of a predetermined size for a video signal obtained by imaging a pattern by the imaging means, and The video signal is integrated with respect to one direction of the two-dimensional coordinates in the dimension window, the first pattern detection window is set for the integration signal, and the first pattern detection window is scanned with the pattern detection window. An evaluation value is sequentially obtained using the integrated signal in the one-pattern detection window, a rough position of the pattern is detected based on the first evaluation value signal, and the first pattern detection is performed with the rough position as a center. A second pattern detection window having a width smaller than that of the window is set, and the integrated signal in the first pattern detection window is set to the second pattern detection window.
While sequentially scanning the pattern detection window, the evaluation value is sequentially obtained using the integrated signal in the second pattern detection window, and the center position of the pattern is obtained based on the second evaluation value signal. And

In a preferred embodiment of the present invention, a plurality of two-dimensional windows are set for a linear pattern in the same field of view, and random noise or partially generated noise is eliminated by integrating video signals in the two-dimensional windows. , The contrast and skewness of the integrated signal are obtained by moving the first window of a predetermined width (width smaller than the two-dimensional window), and the integrated signal that does not meet the conditions is removed. The skewness is calculated again while moving the second window of the specified width (however, smaller than the first window) around the obtained centered position, and the skewness value is weighted to the center of gravity before and after the skewness value crosses zero. By averaging, the center position in one window is obtained. Furthermore, the minimum / maximum values are removed from the plurality of center positions obtained for each of the plurality of two-dimensional windows and the averaged values are averaged to obtain a statistically more accurate center position.

Further, in the above-mentioned publication, a method for actually determining the position and width of an important measurement window is not described, but in the present invention, considering this point, the prealignment accuracy is about 8 to 10 μm. The window has been set and the pattern must be inside the window, and the width of the measurement window is gradually reduced to improve the accuracy of detecting the center position.

Description of Embodiments Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a flow chart showing an outline of the processing flow of a method for detecting a mask pattern on a wafer according to an embodiment of the present invention. FIG. 2 is a view in which a window for projection is applied to the alignment mask pattern on the wafer.

In FIG. 2, 1 is a mask pattern screen on the wafer where the image signal of the portion is obtained by the image pickup means, 2 is an optical image of an alignment pattern, and 3 is scanning when scanning the optical image. The direction is shown, and the video signal at that time is shown at 3 in FIG. Reference numeral 4 in FIG. 2 is a two-dimensional window for integrating the video signals, and an example of the integrated signal is shown in 10 of FIG. 4 (B).

FIG. 3 shows an actual application example, 1 is a video signal screen,
Reference numeral 2 is a light image of the alignment pattern, and 4 is a two-dimensional window for obtaining an integrated signal, and nine windows are arranged here. This number can be any number as long as it can be treated statistically.

FIG. 4 shows the relationship between the video signal, the integrated signal, and the three measurement windows. Reference numeral 3 in the same figure (A) is a video signal, 10 in the same figure (B) is an integrated signal, 4 is a two-dimensional window for integration, and 5 is a window for obtaining the contrast and skewness of the integrated signal while shifting the window 4. The width 8 is smaller than the width 7 of the window 4. Reference numeral 6 denotes a window having a width 9 which is slightly larger than the width of the alignment pattern and which is the center of the point having the minimum skewness obtained in the window 5, and is a window for obtaining the final center position.

FIG. 5 shows the calculation result of a certain pattern.
Is the integrated signal, 11 is the contrast of the integrated signal 1, 12 is the skewness value, and 13 is the zero-cross point of the skewness.

Hereinafter, each process will be described in detail with reference to the flowchart of FIG.

First, the video signals in the range of number 1 in FIG. 2 obtained by the image pickup means in step S1 are integrated in the y direction in the first two-dimensional window 4 of N, and FIG. Numbering
An integrated signal P (i) as shown in 10 is obtained.

here, Is. In this equation (1), the integrated signal y _j (i) is y
The video signal is as shown by 3 in FIG. 4 (A) when one coordinate j in the direction is determined, and 1 indicates the width of integration in the y direction. As a result, random noise included in the video signal is removed, and spike noise is diluted to obtain a clean integrated signal P (i). On this occasion,
The integrated width l in the equation (1) needs to be wide enough to remove noise, but if it is made too large, it is affected by the distortion of the optical image of the alignment pattern caused by uneven coating of the resist, resulting in a large measurement error. The width may be appropriate, so the width should be appropriate.

Next, in step S2, the contrast C is obtained from the integrated signal P (i) by the following formula, and signals whose contrast C is below a predetermined threshold value in step S3 are excluded from measurement targets and excluded from N in step S4. To do.

here, It is defined as the standard deviation, the mean, and the value called cartosis, respectively. Further, n is the range of number 7 in FIG. 4 (B), and 1 to n is the width of i in the x direction.

By this exclusion processing, it is possible to remove the place where the contrast partially deteriorates due to the influence of shading or the like from the measurement target and improve the center detection accuracy.

Next, a measurement window shown by number 5 in FIG. 4 (B) is set with a width 8 smaller than the integration window 4, and the measurement window is shifted in the integration signal P (i) in step S5. , Contrast C shown in equation (2)
And skewness S (j).

Skewness S (j) is Is. However, w is the width of the measurement window, P _min is the minimum value of P (i) shown in FIG. 4 (B).
(I, j) reduces the influence of the offset when obtaining the skewness, and has a value normalized by the area and has a probabilistic meaning. Also, M (j) is And the center of gravity in the window 5 is represented by the first moment. Also, Σ (j) is Is the standard deviation within window 5.

Next, in step S6, the position at which the contrast C obtained as described above exceeds a predetermined threshold and the absolute value of the skewness S becomes the minimum is set as P _{1 and is set} as a temporary center position. On this occasion,
If there is no point P _{1 in} step S7, that is, if there is no point that satisfies the condition even if the whole range (range 7 in FIG. 4) is examined, in step S8, it is excluded from measurement targets from N exclude.

Next, the final measurement window shown by 6 in FIG. 4 (B) is set with the width slightly larger than the width of the actual alignment pattern with the point P ₁ as the center. This width is, when the width of the actual alignment pattern with x _0, 1.1x ₀ ~2.0x about ₀ is preferred.

Then, in step S9, the skewness S is calculated again within the range 8 in FIG. 4 while shifting the measurement window 6 above.
In step S10, a point (position 13 in FIG. 5) at which the skewness S value crosses zero is obtained. If there is no zero-cross point, the process branches to step S12. If a zero-crossing point is found, the center-of-gravity positions before and after the zero-crossing point are weighted by the skewness value and averaged to remove the quantization error in step S11, and the final center position M ₀ (k)
Ask for.

However, S ₁ , S ₂ , M ₁ , and M ₂ are the skewness value and the center of gravity position before and after zero crossing, respectively.

At this time, if the value of skewness S does not cross zero, the value of P ₁ obtained previously is compared with another M ₀ (k), and if it is within a certain threshold, the value of P ₁ is adopted as the center position. To do.

Next, in step S12, it is determined whether or not the processing has been completed for all N two-dimensional windows. If not, the process returns to step S1 and the processing for the next window is performed. When N times have been completed, the obtained M ₀ (k) are averaged to obtain a pattern center close to true. Needless to say, various statistical treatments of M ₀ (k) can be considered depending on the pattern and the arrangement of the integration windows, but in the case of FIG. 3, the excluded central position is predicted from the calculated central position and included. Alternatively, the pattern center position close to the true position may be obtained from the center of gravity of all N center positions.

In the above embodiment, the central position was measured by directly using the integrated signal P (i) (signal 10 in FIG. 4 (B)).
Absolute value of difference signal of integrated signal d (i) = | P (i + 1) -P (i-1) |
The above algorithm may be applied using (5).

In this case, it can be seen that the influence of the background and shading is reduced, and that using d (i) directly without obtaining P (i) to simplify the calculation has little effect on the result. There is. Further, it is possible to prevent misidentification due to skewness by obtaining the approximate pattern position using the difference value.

Further, in the above-described embodiment, the case of detecting the linear pattern in the y direction has been described, but the linear pattern in the x direction can be detected in the same manner by exchanging the x and y directions.

Further, the alignment pattern can be applied to a straight line, a cross, a quadrangle, a circle, or the like, and may be any pattern as long as it can find a plurality of pattern center positions in the x and y directions by the same method.

A digitized signal can be used as the video signal.

EFFECTS OF THE INVENTION As described above, according to the present invention, the first pattern detection window is set for the integrated signal obtained in the two-dimensional window, and the integrated signal in the detection window is used. The rough position of the pattern is detected based on the first evaluation value obtained as described above, and a second pattern detection window having a width smaller than the first pattern detection window is set around the rough position as a center. A second evaluation value is obtained by using the integrated signal in the second pattern detection window while scanning the integrated signal in the pattern detection window with the second pattern detection window, and the second evaluation value signal Since the center position of the pattern is calculated based on the pattern, high-precision pattern that is not affected by noise influence, shading, difference in background brightness, difference in appearance of interference fringes, resist unevenness, etc. Turn position can be detected. Further, by applying it to a mask aligner or the like, it becomes possible to perform precise alignment such as alignment pattern on a wafer.

[Brief description of drawings]

FIG. 1 is a flow chart showing a processing flow of a method for detecting an alignment pattern on a wafer according to an embodiment of the present invention, and FIGS. 2 and 3 apply a projection window to an alignment pattern. FIG. 4 (A) is a diagram showing a video signal waveform, FIG. 4 (B) is a diagram showing a noise-free waveform obtained by integrating the video signal waveforms shown in FIG. 4 (A), FIG. 5 is a diagram showing the integrated waveform, contrast and skewness in an overlapping manner, FIG. 6 (A) is a diagram showing a positioning pattern coated with photoresist, and FIG. 6 (B) is an image. 6 (C) is a diagram showing a signal waveform, FIG. 6 (C) is a diagram showing a binarized signal waveform binarized by a predetermined threshold which is a conventional method, and FIG. 6 (D) is an image whose level is smaller than the threshold. The signal waveform is shown in FIG. 6 (E), which is an asymmetric video signal waveform. FIG. 6 (F) shows a binarized signal waveform obtained by binarizing the video signal shown in FIG. 6 (E) by a threshold value, and FIG. 7 (A) shows noise. FIG. 7 (B) is a diagram showing a video signal waveform, FIG. 7 (B) is a diagram showing a binarized signal waveform obtained by binarizing the video signal shown in FIG. 7 (A) by a threshold value, and FIG. 8 (A) is The figure which shows the asymmetrical video signal waveform detected by the error of an alignment pattern, FIG.8 (B) shows the binarized signal waveform which binarized the video signal shown in FIG.8 (A) by the threshold value. FIGS. 9A to 9G are diagrams showing various video signal waveforms.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 7/00 7/60 H01L 21/027 21/68 F H01L 21/30 502 M (72) Invention Person Teruhiro Hiroi 53, Imaiue-machi, Nakahara-ku, Kawasaki-shi, Kanagawa Canon Co., Ltd. Kosugi Plant (56) Reference JP-A 63-105603 (JP, A) JP-A 54-103373 (JP, A) JP-A Sho 53-69063 (JP, A)

Claims (2)

[Claims] 1. A two-dimensional window of a predetermined size is provided for a video signal obtained by imaging a pattern by an imaging means,
Video signals are integrated in one direction of the two-dimensional coordinates in the two-dimensional window, a first pattern detection window is set for the integrated signal, and the integrated signal is scanned by the pattern detection window. An evaluation value is sequentially obtained by using the integrated signal in the first pattern detection window, a rough position of the pattern is detected based on the first evaluation value signal, and the first pattern is centered on the rough position. A second pattern detection window having a width smaller than that of the detection window is set, and the second pattern detection window is scanned while the integrated signal in the first pattern detection window is being scanned by the second pattern detection window. A pattern characterized in that evaluation values are sequentially obtained using the integrated signal in the window, and the center position of the pattern is obtained based on the second evaluation value signal. Way out. 2. The pattern according to claim 1, wherein there are a plurality of the two-dimensional windows, and the final pattern center position is detected by averaging the center positions of the patterns obtained for each of them. Detection method.