JP3042403B2 - LSI image displacement calculation method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for calculating a displacement of an LSI image in alignment of an observed image of an LSI shape.

[0002]

2. Description of the Related Art Conventionally, an LSI image position shift calculating method is as follows.
For example, it is required for automatic operation in an LSI analysis device such as an electron beam tester, and is used for positioning an LSI-shaped observation image. This alignment requires a short processing time and a low error.

In general, in the LSI image misalignment calculation method, the observed two-dimensional image of the LSI is obtained by adding image quality deterioration such as noise, distortion, and blur. For this reason, it is difficult to completely match the shape of a reference image such as an LSI mask diagram even if the displacement is corrected. Further, a method of directly deriving the entire amount of positional deviation from only the local match / mismatch of the shapes cannot be adopted. A technique for determining the quantity is needed.

[0004] As a rudimentary method suitable for this, for example, there is a method of evaluating the overall coincidence while shifting the two images little by little, and obtaining the amount of deviation that maximizes the coincidence. However, this method has a disadvantage that a very long processing time is required when the displacement range is wide because the displacement is performed in two dimensions.

[0005] In connection with such a calculation of the displacement, an LSI
A technique for high-speed calculation of the displacement between the electron beam tester image and the LSI mask diagram has been proposed in a document (Okubo et al., "Study of EB Automatic Positioning Method for LSI Wiring in EB Tester", Academic Journal Promotion Committee 132 Committee 1
17th meeting for the study group, pp.153-158).

In the above method, contour lines such as wiring in an electron beam tester image are checked for each direction, and the distribution amount of contour lines in a direction perpendicular to each of the X axis and the Y axis is projected. In this method, the amount of displacement is calculated by comparing the amounts of projection of two images for each axis using the projected contour lines.

In this method, the distribution of contour lines such as wiring is represented by X,
The positional deviation is calculated in one dimension by disassembling each axis of Y in each direction. For this reason, by examining the degree of coincidence while shifting slightly in a one-dimensional space, it is possible to perform higher-speed processing than the above-described general method of calculating a positional shift. In this method, the amount of projection of a contour of a wiring or the like on each axis for each direction is used.

As another conventional example similar to the present invention in the technical field, there is an "object image recognition processing method" disclosed in Japanese Patent Laid-Open No. 5-225344 in which a template is checked against a plurality of positioning conditions.

A feature point for matching and a map are associated with each other, and a feature point extracted from an image to be matched is converted into a map coordinate system by a given conversion formula to examine a neighborhood positional relationship. There is an "image alignment device" disclosed in Japanese Patent No. 181949.

[0010]

However, in the above-mentioned conventional method, such information may not be suitable for position determination. That is, depending on the type of the integrated circuit element or the image observation position therein, the wiring in the X-axis or Y-axis direction may be distributed at a substantially uniform density or may be distributed almost periodically. In such a case, an erroneous displacement amount may be calculated.

FIG. 6 shows an example of a target image which leads to an erroneous displacement in the conventional method. However, actually LSI
There is image quality degradation in the image obtained by observing, but this is not represented in this example. FIG. 6 shows that there is a possibility that the displacement in the X-axis direction is erroneous. In FIG. 6, two wirings are present on the right and left sides of the center. Correspondingly, there are four Y-axis contours on the right and left sides, respectively.

FIG. 7 shows the projection amount of the contour in the Y-axis direction on the X-axis in the target image of FIG. Here, since the distributions corresponding to the four contour lines in the Y-axis direction on the right and left sides in FIG. 6 are both similar distributions, errors may be introduced between them.

An object of the present invention is to provide a method for calculating an LSI image position shift with a high speed and a low error rate.

[0014]

In order to achieve the above object, an LSI image position shift calculating method according to the present invention provides a target image obtained by observing a two-dimensional shape of an LSI and a two-dimensional model LSI corresponding to the target image. 2 with the reference image referencing the shape
Feature extraction operation for extracting each feature element distributed in one image, orthogonal projection operation for projecting the distribution of the two extracted feature elements in two axial directions perpendicular to each other, and feature for the two projected images A comparison operation of independently comparing one-dimensionally the projection amounts of the elements for each axis to calculate a difference amount, and the two difference amounts calculated by the comparison operation
For each of the at least two types of feature elements
Calculated by performing a shadow, and integrating at least two types of the difference amounts
Then, it is characterized by the amount of displacement of the LSI with respect to the model LSI.

Further, two kinds of feature elements even without least the above, polygonal under constant diameter or less, closed curve under constant diameter or less, corner contour, diagonal contours, may include any of.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, an embodiment of an LSI image position shift calculating method according to the present invention will be described in detail with reference to the accompanying drawings. FIGS. 1 to 5 show an embodiment of an LSI image position shift calculating method according to the present invention.

FIG. 1 is a block diagram showing a processing procedure for explaining a method for calculating a displacement of an LSI according to the present embodiment. In FIG. 1, a target image 11 and a reference image 12 are input to a feature element extraction unit 15 in parallel. Target image feature element distribution 2 extracted by feature element extraction unit 15
1 and the reference image element distribution 22 are the projection unit 2 on the X and Y axes.
5 is input. From the projected images, the target image feature element projection amount (X-axis) 31 and the target image feature element projection amount (Y
Axis) 32 and reference image feature element projection amount (X axis) 3
3. The reference image feature element projection amount (Y axis) 34 is extracted. Each of these extracted feature element projection amounts 31,
32, 33, and 34, the one-dimensional displacement calculating unit 35
An axis shift amount 41 and a Y-axis shift amount 42 are calculated.

In each of the above-mentioned block components, the target image 11 is a two-dimensional LSI image obtained by an analyzer such as an electron beam tester. The reference image 12 to be compared with the target image 11 is an image representing a two-dimensional shape serving as a template for the target image 11. This reference image 1
Reference numeral 2 denotes an image serving as a reference observed by an analyzer for a different sample having the same design as the LSI of the target image 11 or an LSI mask diagram or the like.

The feature element extraction unit 15 searches for and extracts a portion having a shape corresponding to a feature element in the images of the target image 11 and the reference image 12, and obtains its distribution. As the characteristic element, a partial shape that is included in the LSI and is effective for position determination is used. Examples of characteristic elements include the following.

(A) Polygons or closed curves having a certain diameter or less (corresponding to contacts) (b) Bent angles of contour lines (c) Contour lines in oblique directions (d) Contour lines in X-axis and Y-axis directions The image examples of the elements are shown in FIGS.
(C) and (d) show. Extraction of characteristic elements such as contour lines
This can be realized by using various image processing methods.

In the X-axis and Y-axis projection units, the sum of the extracted X- and Y-coordinate values may be obtained based on the distribution of the extracted characteristic elements. As an example of feature element extraction and projection, FIG. 3 shows an example in which the feature element (a) is extracted from the target image in FIG. 6 and projected on the X axis.

In the calculation of the one-dimensional displacement, a one-dimensional displacement is calculated by comparing the projection amounts of the characteristic elements on the X and Y axes with respect to the two images for each axis. In order to calculate the displacement, a correlation is obtained between the projection amount on the target image 11 side and the projection amount on the reference image 12 side while shifting the projection amounts for the two images by a fixed width. What is necessary is just to find the difference amount of the positional deviation that maximizes the degree.

For example, when the projection amounts for the target image 11 and the reference image 12 are obtained as shown in FIGS. 4 (a) and 4 (b), the degree of correlation when they are shifted by a fixed width is shown in FIG. become that way. From FIG. 5, it is possible to derive the difference amount of the shift that gives the maximum degree of correlation.

Further, in order to further reduce the possibility of leading to an erroneous displacement amount, the above-described operation can be performed on a plurality of types of characteristic elements, and the displacement can be calculated by integrating them. As an example of a method of integrating the results of a plurality of types of characteristic elements, there is a method of obtaining the relationship between the positional deviation and the degree of correlation for each characteristic element, and obtaining the best positional deviation from the sum of these.

According to the above embodiment, two perpendicular to each other
It is divided into two axes (X-axis, Y-axis) and calculates the displacement in one dimension.
Since the amount of displacement is obtained using not only the contour lines in the X and Y axis directions but also other characteristic elements having good position discrimination, the possibility of error occurrence can be further reduced. Therefore, it is possible to calculate the amount of positional deviation between the target image 11 and the reference image 12 that has a high speed and a lower error rate.

In the present embodiment, as information for projecting the distribution on the X and Y axes to calculate a one-dimensional shift, it is not necessary to use only edges (in different directions) representing contours of wiring and the like. , The distribution of other characteristic elements unique to the LSI is effectively used. Examples of the characteristic element include a small hole corresponding to a contact hole, a bending angle of a wiring (by direction), a wiring contour in an oblique direction, and the like. These are extracted from the target image and the reference image, projected onto two axes that are perpendicular to each other, and the amount of displacement is calculated using the amount of projection.

In other words, the method of calculating the displacement according to the present embodiment is characterized in that a characteristic element distributed in each image is provided for each of the target image obtained by observing the two-dimensional shape of the LSI and the reference image of the LSI corresponding to the target image. , The distribution of the feature elements is projected on two axes perpendicular to each other, and the projection amounts for the two images are independently compared one-dimensionally for each axis to calculate the displacement. It is characterized in that the amount of displacement between the image and the reference image is calculated.

The above embodiment is an example of a preferred embodiment of the present invention, but the present invention is not limited to this, and various modifications can be made without departing from the gist of the present invention.

[0029]

As is apparent from the above description, the method of calculating the displacement of an LSI image according to the present invention employs a target image obtained by observing the two-dimensional shape of the LSI and a model LS corresponding to the target image.
Each feature element distributed in two images with a reference image referencing the two-dimensional shape of I is extracted. 2 extracted
The distribution of one feature element is projected in two axis directions perpendicular to each other, and the amount of projection of the feature element with respect to the two projected images is independently and one-dimensionally compared for each axis to calculate a difference amount. The difference between the two calculated by this comparison operation is small.
At least two types of feature elements are projected
The calculated and at least two types of difference amounts are combined into a positional shift amount of the LSI with respect to the model LSI. Therefore, the error rate can be further reduced in order to obtain a position shift amount with high speed and high discrimination.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an embodiment of an LSI image position shift calculation method according to the present invention.

FIG. 2 is a diagram illustrating an example of a feature element.

FIG. 3 is a diagram showing an example in which a polygon / closed curve having a certain diameter or less is extracted from the conventional target image shown in FIG. 6 and projected on the X axis.

FIG. 4 is a diagram illustrating an example of a projection amount for a target image and a reference image.

5 is a diagram illustrating a correlation degree when the projection amounts for the target image and the reference image illustrated in FIG. 4 are shifted by a fixed width.

FIG. 6 is an example of a conventional image that leads to an incorrect displacement.

7 is a diagram showing a projection amount of a contour in a Y-axis direction in an object image onto an X-axis in the conventional example of FIG. 6;

[Description of Signs] 11 Target image 12 Reference image 15 Feature element extraction unit 21 Target image feature element distribution 22 Reference image feature element distribution 25 Projection unit to X and Y axes 31 Target image feature element projection amount (X axis) 32 Target Image feature element projection amount (Y-axis) 33 Reference image feature element projection amount (X-axis) 34 Reference image feature element projection amount (Y-axis) 35 One-dimensional displacement calculator 41 X-axis displacement amount 42 Y-axis displacement amount

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields investigated (Int. Cl. ⁷ , DB name) G06T 1/00 G06T ⁷ /00-7/60 G01B 11/00 G01N 21/88

Claims (2)

(57) [Claims] 1. Extraction of respective characteristic elements distributed in two images: a target image obtained by observing a two-dimensional shape of an LSI and a reference image referred to by a two-dimensional shape of a template LSI corresponding to the target image A feature extraction operation, an orthogonal projection operation of projecting the distribution of the extracted two feature elements in two axis directions perpendicular to each other, and a projection amount of the feature element on the two projected images for each axis. And a comparison operation for calculating the difference amount by independently performing one-dimensional comparison. The two difference amounts calculated by the comparison operation are at least
It is calculated by projecting each of the two types of feature elements.
An LSI image position shift calculating method, wherein at least two types of difference amounts are combined to obtain a position shift amount of the LSI with respect to the template LSI. 2. The method according to claim 1, wherein the at least two types of characteristic elements include any one of a polygon having a certain diameter or less, a closed curve having a certain diameter or less, a turning angle of a contour, and an oblique contour. 2. The method for calculating an LSI image position shift according to item 1.