JPS62177681A - Pattern inspecting method - Google Patents

Abstract

PURPOSE:To detect a defect to some extent of one picture element even when a positioning error is present by comparing a gradient vector obtained from the density distribution of the image of a pattern to be inspected and the density distribution of the image of a reference pattern. CONSTITUTION:When respective density distributions of an image 11 of a multi- value pattern to be inspected and an image 12 of the multi-value reference pattern are f(xi,yj) and g(xi,yj), respective gradient vectors A and B are calculated from these density distributions. At such a time, considering the positioning error of both patterns, a gradient vector A(xi,yj), B(xi,yj) and an adjoining gradient vector B(xk,yl) are compared, the minimum value at the range of k=i-1, i,i+1,l=j-1, j,j+1 of an absolute value ¦B(xk,yl)-A(xi,yj)¦ of the difference in the vectors is made into C(xi,yj), and when the picture element is present which comes to be C(xi,yj)>=V to a set threshold V, it is decided that the defect is present at the pattern to be inspected.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a pattern inspection method applied to comparative inspection of two-dimensional patterns such as circuit patterns of printed circuit boards or integrated circuits.

[Conventional technology]

Conventionally, in a comparative inspection method in which a pattern to be inspected and a reference pattern are compared and inspected, an image of reflected light or transmitted light of the pattern to be inspected obtained by an imaging device is divided into minute pixels, and the amount of reflected light of each of these pixels is calculated. or amount of transmitted light or brightness (
By binarizing fl and O (hereinafter referred to as density), a binarized pattern consisting of a set of pixels with a value of 1.0 is obtained, while a reference pattern is also created in the same way or from design data. A binarization method is often used in which a binarized pattern is obtained and the thus obtained binarized inspected pattern is compared and inspected with a binarized reference pattern.

[The intercostal point that the invention seeks to solve]

However, in the conventional binarization method described above, that is, as shown in FIG. 8, the image 1 of the inspected pattern or reference pattern is divided into minute pixels 2...
In the method of obtaining binarized pattern 3 by binarizing the dark Jff, it is impossible to avoid the occurrence of unevenness of about 1 pixel at the pattern boundary during binarization, and therefore, in principle, the unevenness of several pixels It was not possible to detect defects of a certain degree or less.

Also, suppose that when converting the density of each pixel into a binary value, it is possible to achieve perfect and accurate binarization.

When there is an alignment error between the pattern to be inspected and the reference pattern, in order to prevent this misalignment from being determined as a defect, it is necessary to set the range of pixels of the reference pattern to be compared with the pixels of the pattern to be inspected. For example, if the alignment error is 1
Assuming that there are pixels, in addition to the corresponding pixel, it is also necessary to compare the pixels of one pixel @c, and if this matches the reference pattern, it must be determined that there is no defect. However, in the case of the binarized reference pattern 4VC shown in FIG. 9 (at) and the binarized virtual structure pattern 5 shown in FIG. 6 has occurred, but the alignment error is minus 1 in the X direction.
Since only the number of pixels is generated, the pixel (
X4. Since the adjacent pixel (X4, Y4) in the X direction in FIG. 9 (bl) compared with FIG.
The defect 6 corresponding to two pixels at l will not be detected.

5If the pixel comparison range is expanded by taking into account the alignment error, there is a possibility that defects up to twice the size of the alignment error will be overlooked. The present invention has been made under the above circumstances, and it is possible to reduce the alignment error between the pattern to be inspected and the reference pattern. The object of this invention is to provide a pattern inspection method that can detect defects up to the size of one pixel even if there is a defect.

[Means for solving problems]

In order to solve the above-mentioned problems, the present invention compares each gradient vector obtained from the density distribution of the image of the pattern to be inspected and the density distribution of the image of the reference pattern.

[Effect]

That is, in the present invention, since the gradient vector represents the shape of the concentration distribution by the above means, by comparing the gradient vectors, the partial shapes of the two concentration distributions of the test food pattern and the reference pattern can be compared, As a result, if a shape that does not exist in the reference pattern appears in the pattern to be inspected, it becomes possible to detect the difference even if the alignment error is considerably large.

〔Example〕

The present invention will be described below with reference to an embodiment shown in FIGS. 1 to 5.

Fig. 1 shows a part of the image 1 formed by the transmitted light of the temporary inspection pattern using a multi-value method. ,y
3), the density distribution and amount of transmitted light are respectively expressed. In this case, the image II of the multivalued pattern to be examined
Gradient vector A corresponding to grad f determined by
is calculated using the following formula; its map is shown in Figure 2.

FIG. 3 shows a part of the image 12 created by the transmitted light of the reference pattern corresponding to the pattern to be inspected using a multi-value method from the design data of the pattern to be inspected.
, y is the spatial position coordinate, g(x;, yj) is the pixel (xl
, g3), respectively. In this case, the gradient vector B corresponding to grad g obtained from the image of the multivalued reference pattern is calculated by the following equation; )-g
(xl-1, ylr, >).

Figure 4 shows the map underwater.

[7] a of image 1) of the above-mentioned multivalued pattern to be inspected
Image 12 of the ho distribution f(xl, y3> and the multilevel standard pattern
Each gradient vector A with the concentration distribution g(xl, yj)
and gradient vector B. At this time, gradient vector A(xl, yj) and gradient vector B(
x<,yJ> and its adjacent gradient vector B(Xk,
Y7) and the absolute value of their vector difference; (B(xk,y,)-1xl,yj)1 = i-Z
, i, i-+l, A=j-J+j, j+2 is the minimum value in the range C(xl, y3), and the minimum value of this gradient vector difference C(xi, yj)? Figure 2 and Figure 4.
As shown in the figure.

Next, the minimum value C(Xi, yj
), set a certain threshold value V, and set this threshold value V
If there is a pixel for which C(x i, y j)≧V, the pattern to be inspected has a defect.
For example, if V=40, the fifth
The shaded areas shown in the figure are determined to be defects.

Therefore, as shown in FIG.
If there is a defect with an area equivalent to 11I element, ±1
Conventional methods that only compare densities that take into account pixel alignment errors cannot detect such defects;
The present invention allows complete detection.

Furthermore, FIG. 6 shows another embodiment of the present invention, in which when there is an upright defect with an area equivalent to about one pixel in one image of a multi-valued pattern to be inspected, the corresponding standard The image of the pattern has an overall depression of O, and if the minimum value C(xi, yj) of the gradient vector difference is found in the same manner as above, the value shown in FIG. 7 is obtained. At this time, the threshold value V is 40 (V=40), the pixels in the shaded area shown in FIG. 7 are detected as defects.

That is, in FIGS. 1 and 3, an image 11 of a multivalued inspected pattern and a reference pattern (D image 12
Distributions f(xt, yj) and g(xr.

For example, when comparing only pixel (x5.y,) of the pattern to be inspected shown in FIG. 1, the density of the pixel (x5. + y4) + (X4゜Y5) H(X
4 + Ya) + (x513'4) + (x51
ysL(X5+Y6) l (xe 1Y4)l (x
61y5)+(x,,y,), and in this case, since the density of each pixel is compared as a separate stand-alone pixel, the pixel (xg
The pixel with almost the same density as +yi is the base-stern pixel (x4+Y4)+(X4+Y5>+(x417a)
Therefore, the pixel (Xa+)'s) of the pattern to be inspected is determined to be "not defective", and the surrounding pixels are similarly determined to be "not defective". However,
In the present invention, a gradient vector represents the magnitude and direction of the gradient of a surface, and is a vector that represents the relationship between the density of a point and the density of surrounding pixels, in other words, the shape of the density distribution. For example, the pixel ``5 ty,'' in the pattern to be inspected shown in Figure 1 is determined to be ``protruding'', but this is because the shape is determined by comparing it with the surrounding pixels. On the other hand, there is no so-called "protrusion" in the pixels of the reference pattern shown in Figure 3, so it can be determined that the two are different, and such a determination indicates that the alignment error is quite large. However, since the gradient vector is a graph representing the shape of the concentration distribution, comparing the gradient vectors means comparing the partial shapes of the concentration distributions of two patterns. From this, if a shape that does not exist in the reference pattern appears in the pattern to be inspected, the difference can be found even if the alignment error is quite large.

In the above embodiment, for example, the density distribution f(xr, yj) of the multivalued image of the pattern to be inspected shown in FIG.
The formula for calculating the gradient vector Af obtained from is 2
For example, the gradient vector B may be obtained from the following equation using the Sobel method;

It is also possible to compare only the B direction.

Furthermore, when the pattern inspection method according to the present invention has gradual density changes, even large defects may not be detected, so if used in conjunction with direct comparison inspection using conventional methods, more complete inspection is possible. becomes.

In addition, it goes without saying that the present invention can be modified in various ways without departing from the gist of the present invention.

〔Effect of the invention〕

As is clear from the above description, according to the present invention, the gradient vector ? As a result of this comparison, it is possible to provide an excellent pattern inspection method that can detect defects up to about one pixel even when there is an alignment error.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram showing an example of a multi-level inspected pattern in the pattern inspection method according to the present invention, and FIG. 2 is an explanatory diagram showing a map of gradient vectors similarly obtained from the multi-level inspected pattern. "Figure 3 is an explanatory diagram showing a multi-value reference pattern corresponding to the multi-value test pattern, Figure 4 is an explanatory diagram showing a map of gradient vectors similarly obtained from the multi-value reference pattern, and Figure 5 is an explanatory diagram showing the gradient vector. An explanatory diagram showing a map of the minimum value of the vector difference, FIG. 6 is an explanatory diagram of another multivalued inspected pattern according to the present invention, and FIG. 7 is an explanatory diagram showing the map of the minimum value of the gradient vector difference. Fig. 8 is an explanatory diagram showing a state in which a pattern image obtained by an imaging device is divided into minute pixels and binarized into the density of each pixel, and Fig. 9 (al is also an explanation of the binarized reference pattern). Figure, No. 9
Figure (bl is also an explanatory diagram of the binarized inspected pattern. a degree distribution, 1
2... Image of multivalue reference pattern, "xi + J)...
・Concentration distribution. Applicant's representative Patent attorney Takeshi Suzue 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 (a) Figure 9 (b)

Claims (2)

[Claims] (1) In a pattern inspection method in which a pattern to be inspected and a reference pattern are comparatively inspected, gradient vectors obtained from the density distribution of the image of the pattern to be inspected and the density distribution of the image of the reference pattern are compared with each other. Characteristic pattern inspection method. (2) The pattern inspection method according to claim 1, wherein the image of the reference pattern is created from design data of the pattern to be inspected.