JPS6239813B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an apparatus for detecting circuit pattern defects in LSI photomasks and the like with higher accuracy.

FIG. 1 shows a pattern diagram in a conventional pattern defect detection method, in which two patterns 1 and 2 of the same shape shown in a and b of the figure are compared with each other as shown in c of the figure, If there is a mismatched portion a, b between patterns 1 and 2, this portion can be regarded as a mismatched portion in shape between patterns 1 and 2, that is, a defect. Ideally, it would be possible to detect small mismatched parts by overlapping them as shown in c in the same figure, but in reality it is difficult to perfectly align the two patterns 1 and 2, so for example, it is possible to detect alignment errors. Assuming D, it is difficult to detect the small defect a on the corner as shown in Figure 2.
Only large defects b can be detected.

SUMMARY OF THE INVENTION An object of the present invention is to provide a pattern defect detection device which eliminates the above-mentioned drawbacks of the prior art and is capable of detecting even small defects at corners.

That is, in order to achieve the above object, the present invention provides first and second imaging devices that synchronously scan and image two aligned patterns that should originally be the same; first and second binarization circuits that convert video signals obtained from each of the imaging devices into binary pixelized signals;
Each of the binary pixelated signals obtained from the digitization circuit is synchronously and corresponded to each other while being isolated and sequentially stored in an L-shaped storage factor group formed by linearly arranging a plurality of storage factors. first and second memories,
2 stored in the outer L-shaped memory factor group (P ₁ to Pn) stored in each of the first and second memories.
Value pixelized signals (P ₁ ) to (Pn) are (P ₁ )=…=(Pn)
, and the binary pixelated signals (q ₁ ) to (qm) stored in the inner L-shaped memory factor group (q ₁ to qm) are (q ₁ )=(q ₂ )=...=(qm ), and (P ₁ )≠
(q ₁ ), and X, Y from one of the first and second specific pattern detection means. A comparison means is provided for determining that one pattern has a defect when a corner is detected in the coordinates and no corner is detected within a range of (X±D, Y±D) from the other. This is a pattern defect detection device.

The present invention will be explained in detail below with reference to embodiments shown in FIGS. 3 to 6.

FIG. 3 shows a specific example of the hardware configuration of the pattern defect detection apparatus according to the present invention. Patterns 1 and 2, which are positioned in the figure and are identical in nature, are detected by pattern imagers 4, 4' via lenses 3, 3' and video signals 5, 5' obtained from imagers 4, 4'. are converted into binary picture elements by the binarization circuits 6 and 6', respectively, and are
As described in Japanese Patent No. 131469, etc., the signals are first output and input into the memories 7 and 7', respectively, by a well-known technique. Here, the image is spatially quantized (two-dimensionally cut out) and stored in memory. That is, the image is divided into a certain minimum unit of squares (memory factor) and stored in memory. One square (memory factor) is called a picture element. In the case of a binarized image, the content of a picture element is "1" or "0". That is, as shown in FIG. 7, each of the memories 7 and 7' has shift register groups 7a and 7a' that store one raster, and a binary pixelated signal output from the shift register group, for parallel input and parallel output. It is composed of memories 7b and 7b' having a plurality of memory elements vertically and horizontally so as to cut out and store them.

Next, the shape of the local pattern is detected from each memorized image by each of the specific pattern detection means 8, 8'. The specific pattern in each of the specific pattern detection means 8, 8' is as follows:
For example, as shown in FIG. 4, L-shaped picture element groups 10 and 10' in which a plurality of memory elements are linearly arranged are provided separated by a predetermined distance. That is, memories 7, 7'
As shown in FIG. ₄ , among _all picture elements stored in each of
The binary pixelated signals (P ₁ ) to (Pn), (q ₁ ) to (qm) stored in the section 10 (q ₁ , q ₂ ...qm) are When the following relationship is satisfied, it is determined that there is a corner 11 in each of the specific pattern detection means 8, 8'. Each of the specific pattern detection means 8, 8' is configured based on the above equation (1), for example, a theoretical circuit (AND circuits 8a, 8a', NOR circuits 8b, 8b',
OR circuits 8c, 8c') are constructed as shown in FIG. As an example of the L portion, in addition to the corner portion in the direction of 12 in FIG. 4, that is, FIG.
3, 14, 15, these corners can be formed, for example, into two sets of L-shaped windows separated in time by setting the read pattern from each of the memories 7, 7' accordingly. By checking the data content within, it is possible to detect the presence of corners in the pattern.

Next, the comparison means 9 compares the corners of the patterns determined by each of the specific pattern detection means 8 and 8'. If a corner exists in both patterns, the pattern is determined to be normal; if it exists only on one side, the pattern is determined to be defective. That is, when the specific pattern detection means 8 (or 8') detects the presence of a corner at the coordinates X and Y of pattern 1 (or 2), the other specific pattern detection means 8' (or 8) Check whether a corner in the same direction is detected within the coordinate range (X±D, Y±D) of pattern 2 (or 1) (D is a value that allows positional deviation). . If no corners in the same direction are found, the comparison means 9 outputs a defect signal.

The shape of the detected corner defect is determined by the arrangement shown in FIG. For example, the solid line 16 in Figure 6
Since the corner shown in does not satisfy formula (1), it is not determined to be a corner. Therefore, if the other comparison pattern shape has a corner 17 as shown by the broken line, the pattern 16
is determined to be defective.

Although the above embodiment has been described using a corner as an example of a specific pattern, it is of course applicable to detecting defects such as minute irregularities and sharp bends.

According to the detection method according to the present invention, even a slight deformation of a corner of a pattern as shown in FIG. 6, for example, can be determined to be a defect, and it is now possible to discover defects with minute deformation that could not be detected using conventional inspection methods. , fix this discovered defect (discard it if it cannot be fixed)
By doing this, we can make better quality masks.
It can be used for IC and LSI manufacturing. the result
The yield of IC and LSI manufacturing has improved, making manufacturing more efficient, and its industrial value is enormous.

[Brief explanation of the drawing]

FIG. 1 is a pattern diagram of a general pattern defect detection method, and FIG. 2 is a pattern diagram when an alignment error occurs in the method of FIG. 1. 3 to 7 relate to the present invention, and FIG. 3 is a block diagram showing an embodiment of a pattern defect detection device;
Figure 4 is an example of a pattern for detecting the presence of a corner, Figure 5 is a diagram showing the type (direction) of a corner,
FIG. 6 is a diagram showing an example of detecting a corner and an example of not detecting a corner, and FIG. 7 is a diagram showing a more specific example of an embodiment of the apparatus shown in FIG. 3. 1, 2...Pattern, 7,7'...Memory, 8,
8'... Specific pattern detection means, 9... Comparison means, 1
0...Inner L part, 10'...Outer L part, 11-15...
L part.

Claims (1)

[Claims] 1. A first and second imaging device that synchronously scans and images two aligned patterns that should be the same, and a video signal obtained from each of the first and second imaging devices. The first and second binarization circuits that convert into value pixelization signals and the binary pixelation signals obtained from the first and second binarization circuits are synchronized and made to correspond to each other. L-shaped 1 formed by linearly arranging multiple memory elements separated by
first and second memories sequentially stored in a set of memory factor groups, and outer L-shaped memory factor groups (P ₁ to Pn) stored in each of the first and second memories. The binary pixelated signals (P ₁ ) to (Pn) are (P ₁ )=
(P ₂ )=...=(Pn), and the binary pixelated signals (q ₁ ) to (qm) stored in the inner L-shaped memory factor group (q ₁ to qm) are (q ₁ ) = (q ₂ ) =...= (qm) and (P ₁ )≠(q ₁ ) to detect whether or not a corner has been cut out; first and second specific patterns; A corner is detected in the X and Y coordinates from one of the detection means and the first and second specific pattern detection means, and no corner is detected within a range of (X±D, Y±D) from the other; 1. A pattern defect detection device comprising: a comparison means for determining that one of the patterns has a defect when the pattern is defective.