JPH06214376A - Copying correcting method for polygonal pattern - Google Patents

Abstract

PURPOSE:To improve copying correcting accuracy by automatically performing alignment by using a common pattern between a reference pattern and the pattern of a copying correcting spot and correcting copying by using the reference pattern where line width is corrected by the plotting method of a focusing ion beam device. CONSTITUTION:The normal reference pattern corresponding to the pattern to be corrected is searched from the patterns formed on a mask 8, and the contour information of the reference pattern is obtained by an automatic pattern acquisition method, and registered in a computer 13. The picture display of a CRT 14 is restored to the copying correcting position of the mask 8, and the normal acquired reference pattern is superposed on the correcting position of the mask 8. Next, the contour of the normal pattern and the pattern of the mask 8 are accurately aligned according to an automatic correcting alignment method. Then, an area where the line width is corrected is set by an automatic copying correcting area setting means, and the bit map of the area is formed and plotted as a plotting area.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for copying and correcting a polygonal pattern of a photomask, reticle or the like having a focused ion beam processing device.

[0002]

2. Description of the Related Art An image of a reference pattern obtained from a focused ion beam processing apparatus or the like is subjected to preprocessing such as noise removal, and then binarized to obtain a bitmap of the reference pattern. The reference pattern is obtained by scanning and irradiating the same normal patterned focused ion beam in the vicinity of the pattern to be corrected on the mask, and detecting the secondary charged particles.

This reference pattern is manually designated as a copy position, and a drawing area is set at that position. Here, in a drawing method such as a focused ion beam processing apparatus, since a pattern is formed to be wider than the drawing area by a certain amount, a reference area bitmap is subjected to a certain number of pixel reduction processing as the line width correction, and is used as the drawing area. Next, the drawing area is drawn by a drawing device such as a focused ion beam processing device to copy and correct the reference pattern.

[0004]

However, the above method has a problem in accuracy because it can only set the copy correction area in units of one pixel because of the bitmap processing. Further, since the reference pattern was manually moved to the copy position, there was a problem in the accuracy of the copy position.

[0005]

Means for Solving the Problems After preprocessing such as noise removal is performed on an image of a reference pattern, binarization processing is performed to obtain a bitmap of the reference pattern. The contour of the pattern on this bitmap is acquired by the tracing method. Next, polygonal approximation is performed by intercept linear approximation using the least square method to obtain contour information. After the reference pattern is manually designated as the copy position, a common pattern of the reference pattern and the copy location is designated, and accurate alignment is automatically performed.
Next, as a line width correction, a pattern is obtained in which each side of the reference pattern is translated by a specified width. This pattern is used for drawing and copying and correction.

[0006]

[Operation] A reference pattern in which the common pattern of the reference pattern and the pattern of the correction portion of the mask is used, their positions are automatically aligned, and the line width is corrected by the drawing method of the processing region of the focused ion beam processing apparatus Can be used to correct the mask pattern with high accuracy.

[0007]

Embodiments of the present invention will now be described in detail with reference to the drawings. FIG. 1 shows an overall configuration diagram of a focused ion beam apparatus according to the present invention. An ion source 1 emits an ion beam 2. A scanning electrode 3 is composed of X and Y electrodes, and scans an irradiation spot of the ion beam 2 on the mask 8 which is a sample over a predetermined range in the XY plane. An objective lens 4 forms an image of the spot of the ion beam 2 on the surface of the mask 8 which is an irradiation target. A gas gun 5 sprays the organic compound vapor 6 on the white defect portion of the mask 8 and simultaneously irradiates it while scanning the ion beam to attach a light shielding film to correct the white defect.

Further, in the repair of the black defect portion, the unnecessary adhered portion is irradiated with the ion beam 2 in a limited manner, and is removed by the sputtering to correct it. Reference numeral 9 denotes an XY stage on which the mask 8 is placed and which is moved in the X or Y direction. A detector 10 detects the intensity of the secondary ions 7 knocked out from the surface of the mask 8 by the irradiation of the ion beam 2.
The planar distribution of the secondary ion intensity corresponds to the pattern formed on the mask 8. Reference numeral 11 is an A / D converter, which converts an analog amount called secondary ion intensity into digital data. This digital data is taken into the computer 13, the pattern image of the mask 8 is enlarged and reproduced, and displayed on the CRT 14. A scanning circuit 12 receives the ion beam irradiation range from the computer 13 and controls the scanning electrodes 3.

Here, the shapes of the mask pattern and the reference pattern of the basic mask pattern already stored in the computer 13 are compared, and the portion where these patterns are deviated becomes the pattern correction shape. In other words, it is necessary to find the pattern correction shape, which is called pattern copying.

An actual pattern copying method is shown in FIG.
It will be described according to (4). Acquisition of Reference Pattern (Refer to FIG. 1 (1)) A normal reference pattern 21 corresponding to the pattern to be corrected is searched for from the pattern formed on the mask 8, and the reference pattern 21 is subjected to contour information acquisition by an automatic reference pattern acquisition method. Obtained and registered in the computer 13. The mask 8
The shape of the pattern formed in is obtained by irradiating a focused ion beam in a certain range and detecting secondary charged particles (secondary electrons or secondary ions) obtained by the irradiation with the detector 10. To be

Manual correction alignment (see FIG. 2 (2)) The image display of the CRT 14 is returned to the copy correction position (the position where the pattern should be corrected) of the actual mask 8 and the contour of the normal reference pattern 21 obtained in the previous stage. Is moved by using the mouse and overlapped with the corrected position of the mask 8. The correction pattern of the mask 8 is represented by numeral 21a in FIG.

Automatic Correction Positioning (Refer to FIGS. 2 (3) and 2 (4)) Input an edge portion to be matched for position adjustment between the correction pattern 21a and the reference pattern 21. Here, the contour of the normal pattern 21 and the pattern of the mask 8 are accurately aligned according to the automatic correction alignment method.

Copy correction area setting (see FIG. 2 (4)) An area for which line width correction has been performed by the automatic copy correction area setting method is set based on a rectangular input by the operator. Create a bitmap of this area and draw it as a drawing area.

Next, the method used in the above procedure will be described in detail. [A] Automatic reference pattern acquisition method The image of the reference pattern 21 is converted into a bitmap by binarization processing. The contour of the reference pattern 21 is traced from this bitmap to obtain a point sequence 21b as shown in FIG.
At this time, the point sequence 21b is ordered so that the inside of the pattern is seen on the right hand side. Next, polygon approximation is performed by the RAMER method. The initial value is a polygon having two vertices at the upper left and lower right of the edge point sequence as in AE in FIG. Next, the point B farthest from the section AE and its distance, and the section EA
The point F most distant from the point F and its distance are obtained. If these distances are larger than the specified distance, it is registered as a vertex. This is repeated until registration of the vertex is no longer performed.

Next, as shown in FIG. 4, the whole point sequence 2 between the vertices
A regression line in a two-dimensional space is obtained from 1b and is taken as the side of the polygon. Here, a general shape such as a mask pattern is made of straight lines, but since there is a small R only at the corners,
The point (vertex) of the section break is not used in the calculation of the regression line, and is less affected by the R portion.

In order to further improve the approximation accuracy, the intercept linear approximation by the Pavlidis division fusion method shown in FIG. 5 is used. Initial value Almost no influence of the initial value on the final result, but the closer to the optimal solution, the faster the focusing. The output of the RAMER method sufficiently satisfies this condition.

Division processing (see FIG. 5 (i)) Since the error can be estimated in advance due to the nature of the image, the average unbiased variance of the error is set in advance, and if the average unbiased variance of each section is larger than that value, the average is calculated. Divide until the amount of unbiased variance is below a certain value.

Fusion processing (see FIG. 5 (ii)) The error can be reduced by performing division, but instead, the error is incorporated in the contour information. Therefore, the sections are fused in a range in which the average unbiased variance amount of each section is equal to or less than a certain value.

Local optimization processing (see FIG. 5 (iv)) The division / fusion processing optimizes the number of sections. On the other hand, in the local optimization processing, the position of the break (vertex) of the section is locally optimized. The vertex is moved back and forth to find the sum of the unbiased variances in all sections, and the point at which the value gives the minimum value is taken as the new vertex.

If the vertex does not move in the optimization process, the process ends. Otherwise, continue the process from the division process. Finally, the polygon is obtained from this section information. As shown in Fig. 6, a polygon that passes through the apex as the break of the trace point and the average position of the two adjacent points on both sides and draws a straight line with the average of the slopes of the regression lines on both sides to find the intersection of this line and the regression line Register as the apex of.

[B] Automatic Positioning As shown in FIGS. 7A and 7B, the contour is moved so that the edges of the contour crossing the input rectangle coincide with the edges of the pattern. When there are multiple sides, the required movement amounts are combined.

[C] Automatic copy correction area setting method The copy correction area is recognized from each side of the input rectangle and the contour polygon. Next, each side of the copy correction area is moved by the designated distance to set the copy correction area. (A) Each side of the contour polygon is regarded as a straight line and moved by the specified distance. Next, the new vertex is the intersection of the straight lines corresponding to the adjacent sides of the original polygon. At this time, a temporary piece is inserted at the apex portion as shown in FIG.

(A) Check whether the direction is the same as the direction of the side of the original polygon. If they are different, the sides intersect in the middle, so the intersection is obtained and the part including the abnormal direction edge is deleted. (C) Create a bitmap from the contour information to set the correction area. Therefore, the usual method of painting the inside of the closed curve from the painting start point cannot be used. Therefore, in FIG.
Paint with the polygon scan conversion method. However, as shown in FIG. 10, a polygon that has disappeared in the reduction process may remain. In this way, it is necessary to recognize the beginning and end of the pattern in the direction and order of the sides, rather than simply determining whether or not to paint with the outline as a boundary.

(D) If the obtained quadrilateral is simply bit-mapped, a step occurs as shown in FIG. 11 (a), which causes a problem in accuracy. This is because a ± 1/2 pixel digital error (rounding error) occurs before and after the step. Therefore, Figure 1
Instead of moving the ½ pixel quadrilateral downward as in 1 (b), the step disappears if the entire bitmap is moved up ½ pixel. Here, in the case of FIG. 11A, a maximum ± 1/2 pixel digital error occurs,
It can be seen that it is less than that in the case of FIG. It is the regression line of the edge that affects the correction accuracy. Therefore, as shown in FIG. 11B, the offset amount is calculated for each contour side so that the digital error at the center of the side on the regression line of the edge becomes 0, and the automatic copy correction area setting method movement in FIG. 11C is performed. An offset amount is a vector synthesized according to the amount synthesis.

[0025]

The copy correction accuracy of this apparatus is ± 1 pixel or less, and a marked improvement in accuracy can be achieved as compared with the conventional method.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment.

FIG. 2 is a plan view showing a procedure of a pattern copying method.

FIG. 3 is a plan view showing polygonal approximation by the RAMER method.

FIG. 4 is a plan view showing polygonal approximation by a regression line.

FIG. 5 is a plan view showing a split fusion method of PAVLIDIS.

FIG. 6 is a plan view showing vertices obtained by polygonal approximation.

FIG. 7 is a plan view showing automatic alignment.

FIG. 8 is a plan view showing a virtual side to be inserted for line width correction.

FIG. 9 is a plan view showing painting by a scan conversion method.

FIG. 10 is a plan view showing a polygon disappeared in the reduction processing.

FIG. 11 is a plan view showing a step removal process.

[Explanation of symbols]

 1 Ion Source 2 Ion Beam 3 Scanning Electrode 4 Objective Lens 5 Gas Gun 6 Organic Compound Vapor 7 Secondary Ion 8 Mask 9 XY Stage 10 Detector 11 A / D Converter 12 Scanning Circuit 13 Computer 14 CRT

Claims (1)

[Claims] 1. A contour point sequence of a reference pattern is acquired by a tracing method from an image obtained from a focused ion beam processing apparatus, and the contour point sequence is contour information by polygonal approximation by intercept linear approximation using the least square method. A method of automatically performing alignment using a common pattern of a reference pattern and a copy correction location, and highly accurately performing copy correction using a reference pattern whose line width has been corrected by the drawing method of a focused ion beam apparatus.