JPH0228887A - Pattern defect inspecting device - Google Patents

Abstract

PURPOSE:To precisely inspect the thick defect and thin defect of a pattern by dividing a memory part into plural blocks, calculating places to shown an abnormality for each block, comparing them with a threshold value and judging the distribution condition of the blocks to show the abnormality. CONSTITUTION:First, the length in the width direction of the pattern is measured in a pattern width measuring part 1. Next, its size is compared with that of a prescribed value (standard pattern width) in a comparing part 2. Pattern width measuring data are developed to two dimensions and stores in a memory part 4 of a defect detecting part 3. Next, the memory part 4 where the data are stored is divided into the plural blocks, the data of each block are sent to a number-of-pieces comparing part 5. The number-of-pieces comparing part 5 calculates the places where the data of the pattern comparing part 2 shows the abnormality and decides whether or not their values exceed a large value. When the blocks which exceed the threshold value continue vertically, horizontally, or obliquely, it is judged that there is a pattern defect.

Description

[Detailed Description of the Invention] [Summary] The present invention relates to a pattern width defect inspection device that detects pattern width abnormalities by continuously detecting the pattern width. The purpose of the present invention is to provide a pattern inspection device that can accurately inspect defects, and includes a pattern width measurement section that measures the length in the width direction of a predetermined pattern, and a size difference between the output of the pattern width measurement section and a predetermined value. In a pattern defect inspection device comprising a comparison section that performs comparison and a detection section that detects defects in pattern width based on the output of the comparison section, the detection section expands the binary output from the comparison section into two dimensions. A memory section for storing data, a number comparison section that divides the memory section into multiple blocks, counts the number of abnormalities in each block, and compares it with a threshold value, and a distribution of blocks whose output shows an abnormality. It is characterized by comprising a defect determination section that determines the state.

[Industrial Field of Application] The present invention relates to a pattern defect inspection device that detects abnormalities in pattern width by continuously detecting pattern width.

When a machine performs the visual inspection of conventional printed board patterns that was done by humans, the machine determines that there is a defect even if it is something that humans can flexibly judge. For this reason, the machine's judgment unit is adjusted to determine that there is no defect unless it is a major defect, which has the disadvantage of overlooking defects that should be detected.

In particular, there has been a demand for the development of an apparatus that can effectively inspect patterns that are known as thickening defects and thinning defects.

[Prior art] In the process of visually inspecting patterns generated on printed circuit boards, the visual inspection by humans was replaced with inspection by automatic inspection equipment, with the intention of quickly performing large-volume inspections. It became so.

The width of the pattern is inspected at each predetermined point, and it is determined whether the width is within a normal value range. Recently, it has become necessary to judge patterns that are "thick" or "thin" as defects. That is, when looking at a top view in the longitudinal direction of the pattern as shown in FIG. 6, A in FIG. 6 is a protrusion defect, B is a defect, C is a thickening defect, and D is a thinning defect. To explain with reference to FIG. 7, a thickening defect refers to a case where the pattern width continues to be slightly less than the maximum specified value (a) and exceeding the thickening limit fbl. In addition, a thinning defect refers to a case where the pattern width is continuously at a value slightly exceeding the minimum specified value (dl) and below the thinning limit value (C) in Fig. 7.Such thickening defects and thinning defects Even if the width at each measurement point is within the minimum specified value fdl or maximum specified value fa1, if the same condition continues in the direction perpendicular to the width direction, the electrical characteristics of the lead pattern may become unsatisfactory. Therefore, it is necessary to detect it as a defect.

Furthermore, in order to inspect patterns for defects, it is appropriate to judge the condition over a wider range than the size of the pattern and then make a final judgment. Therefore, the normal value of the pattern is stored in advance in a storage device, and the normal value of the pattern is compared with the pattern to be inspected, and if there is a difference, it is determined as a defect.

When this means is executed, the comparison is made bit by bit, so most of the patterns are determined to be defective due to quantization errors that occur when reading the pattern to be inspected. Therefore, it is better to compare the area, side length, center of gravity, etc. of the patterns, and judge the quality by looking over the range of patterns. This means is versatile and effective even if the shape of the pattern is diversified.

[Problems to be Solved by the Invention] Conventional mechanical pattern width inspection devices do not detect protrusion defects or defects even if the device judges them as defects when they clearly deviate from the maximum or minimum specified values. Since the thickening defect or the thinning defect is determined to be normal at each measurement point, it is determined that the thickening defect or thinning defect is not a defect. In other words, it is necessary to judge whether or not there is a defect based on the results of macroscopic observation, and the structure has not been designed to satisfy this point. Furthermore, in the configuration in which the data is compared with a stored normal pattern, it takes a long time to create dictionary data and perform inspection processing, so it is not possible to inspect many types of patterns in real time.

To perform a large number of tests, it is necessary to prepare a large number of identical processing circuits,
This problem can be solved by parallel processing, but the problem is that the equipment becomes extremely expensive.

Furthermore, it was not possible to accurately inspect thickening defects and thinning defects.

SUMMARY OF THE INVENTION An object of the present invention is to improve the above-mentioned drawbacks and to provide a pattern defect inspection apparatus that can accurately inspect thick and thin defects in a pattern with a relatively simple configuration.

[Means for Solving the Problems] FIG. 1 is a diagram showing the principle structure of the present invention. In FIG. 1, 1 is a pattern width measurement section, 2 is a comparison section, 3 is a defect detection section, 4 is a memory section, 5 is a number comparison section, and 6 is a defect determination section.

A pattern width measurement section 1 that measures the widthwise length of a predetermined pattern, a comparison section 2 that compares the output of the pattern width measurement section 1 with a predetermined value, and the comparison section 2
The present invention has the following configuration in a pattern defect inspection apparatus comprising a detecting section 3 that detects defects with respect to the pattern width based on the output of the detecting section 3. That is, the detection section 3 includes a memory section 4 that expands and stores the binary output from the comparison section 2 in two dimensions, and a memory section 4 that divides the memory section 4 into a plurality of blocks and detects a location indicating an abnormality in each block. It consists of a number comparison section 5 that counts and compares it with a threshold value, and a defect determination section 6 that determines the distribution state of blocks whose output from the number comparison section indicates an abnormality.

[Operation] The pattern width measuring section 1 first measures the length of a predetermined pattern in the width direction as in the conventional method. Next, the comparing section 2 compares the width with a predetermined value (standard pattern width), that is, determines whether it is smaller than the maximum standard value and larger than the fat limit value, or larger than the minimum specified value and smaller than the thin limit value. The determination output is applied to the defect detection section 3 to detect the presence or absence of a defect. Therefore, the pattern width measurement data is developed two-dimensionally and stored in the memory section 4 of the defect detection section 3. Next, the memory section 4 in which the data is stored is divided into a plurality of blocks, and the data for each block is sent to the number comparison section 5. The number comparator 5 counts the number of locations where the data from the pattern comparator 2 indicates an abnormality, and determines whether the value exceeds a threshold value. When blocks exceeding the threshold are consecutive vertically, horizontally, or diagonally, it is determined that there is a pattern defect.

[Example] FIG. 2 is a diagram illustrating a specific example of the pattern width measuring section 1, in which a shift register is used to measure the pattern width. In FIG. 2, 7 generally indicates a shift register. The dots on the register that look like red stamps are the parts of the serial data obtained by scanning the pattern that are related to length measurement. FIG. 3 exemplifies the names of bit strings for length measurement of a pattern, and the bit strings in the left and right direction from the point of interest are named a, a', and so on. The logic of each bit of the shift register is determined by a determination circuit within the pattern width measuring section 8. The bit string for length measurement is extended in eight directions as shown in FIG. 3, centering on the point of interest, and the number of bits in the pattern is measured in each direction. For example, when the difference in length between opposing bit strings such as a and a' is smaller than a preset value, the direction indicates that the point of interest is located at the center of the pattern. If the center signal is "1" due to an abnormality in two directions out of a pair of four directions, the point of interest is assumed to be at the center of the pattern. Next, when this condition is met, add the lengths of the bit string patterns in two opposing directions of the eight direction bit strings,
If there is a direction in which overflow occurs, that direction is defined as the pattern direction, and the direction orthogonal to that direction is defined as the pattern width.

Next, a pattern width calculation section 8 calculates the pattern width. The pattern width signal is applied to the pattern width calculation section 8 into which the maximum specified value and minimum specified value are input, and a size judgment is made. For example, if it is smaller than the maximum specified value and thicker than the limit value, or conversely, if it is larger than the minimum specified value and thinner and smaller than the limit value, it will be set to "1", and otherwise it will be set to "0", then a value of "1" will be obtained. The pattern width is found to be abnormal.

Next, this judgment value is obtained in real time and applied to the memory section 4 of the detection section 3. FIG. 4 shows an example of the memory section 4. For example, the memory section 4 is configured using a shift register,
0" or "°1" is stored. Therefore, the memory section 4 is divided into blocks and the number of abnormal values "0" in each block is counted. Counting and comparison shown in Fig. 4 This is carried out in section 9. It is determined whether the block indicates an abnormality by comparing it with a preset threshold value indicated as a count reference value. In each block, the abnormal value "
If the number of "0" is greater than a predetermined value, the block is set as "1", otherwise it is set as "0". A plurality of counting and comparing sections 9 are provided, and this is an example of the number comparing section 5 shown in FIG. .

Note that the range of block nxn must be greater than or equal to the minimum continuous value when determining thickening or thinning defects.

For example, if the inspection resolution is 10μm, the width abnormality is 400μm.
When a thickening (thinning) defect is defined as a case where the number of defects continues for m or more, the size of one block is 100 x 100 μm, and n is 4 or more. Next, the arrangement of the blocks indicating "al" is examined to confirm whether there is a thick or thin wiring defect.

The defect determination dictionary 10 is used as an example of the defect determination section 6. The dictionary 10 is connected as shown in FIG.
An example of storing 0 is shown in FIG. In the case of a thick defect in a simple lead, the combinations of "1" blocks lined up vertically, horizontally, and diagonally as shown in Figure 5 A, B-C are registered as code information in the defect dictionary, and the bent part in the 45° direction is If it is included, register the dictionary code as shown in the fifth diagram. The defect determining unit compares the distribution state of abnormalities that are "1" in the range of nxn blocks with a dictionary to determine whether or not the block is defective.

Even if a plurality of patterns are included inside the memory nXn, it is possible to handle them using a dictionary.

[Effects of the Invention] In this way, according to the present invention, since the state of each point to be measured on a pattern is summarized as a block, only the continuity of the abnormal part of the pattern is extracted, so that minute details of the pattern can be detected. It is effective because it provides a sufficient inspection margin against fluctuations, rotations, etc.

To create a dictionary, you only need to register the general characteristics of the pattern to be inspected, and there is no need to input data for each pattern.If you know the angular component of the pattern, it is possible to automatically generate a dictionary from that. The time required for inspection will be even faster.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the principle configuration of the present invention. FIG. 2 is a diagram showing an example of the pattern width measuring section of FIG. 1 as an embodiment of the present invention. FIG. 3 is a diagram showing the names of bit strings for Bakun length measurement. Figure, 4th
The figure shows an example of the memory section 4 and defect determination section 6 in FIG. 1, FIG. 5 shows the dictionary storage pattern of FIG. 4, FIG. 6 shows the types of defects, and FIG. FIG. 3 is a diagram showing the relationship between defects and specified values. 1-Pattern width measurement section 2--Comparison section 3-Detection section 4-Memory section 5-Number comparison section 6-Defect determination section Patent applicant Fujitsu Ltd. Representative Patent attorney Eisuke Suzuki Misfire JJIP in August, Phys. J Su Seiko Diagram 1 fP Code Dictionary @Hiragi Pattern Diagram 5

Claims (1)

[Claims] I. a pattern width measuring section (1) that measures the length of a predetermined pattern in the width direction;
A comparison unit (1) that compares the output with a predetermined value
2) and a detection section (3) that detects a defect in the pattern width based on the output of the comparison section (2), wherein the detection section (3) detects defects from the comparison section (2). A memory section (4) that expands and stores the binary output of A pattern defect inspection apparatus comprising: a number comparison section (5) for determining the number of defects; and a defect determination section (6) for determining the distribution state of blocks whose output from the number comparison section indicates an abnormality. II. The pattern defect inspection apparatus according to claim 1, wherein the defect determination section is provided with a dictionary in which the defect status of the pattern to be inspected is recorded, and the defect determination unit compares and determines the contents of the dictionary.