JPH04109647A - Automatic external appearance inspecting device - Google Patents

Abstract

PURPOSE:To improve a detecting ratio and to reduce an erroneously detecting ratio by providing processing means for setting an optimum differentiating formula in response to the dense or pale of an image signal of a pattern, differentiating the image signal and aligning with its differentiated value. CONSTITUTION:Processing means 14, 15 for setting an optimum differentiating formula in response to the dense or pale of an image signal of a pattern, differentiating the image signal based on the formula and aligning with its differentiated value, are provided. That is, since a differentiated signal for differentiating the signal of the pattern by the set optimum formula indicates a position where the density change of the image is vigorous and a position having an edge, relative positional deviation between the patterns can be known by comparing the differentiated signals of the two points of the same pattern. Thus, only a defective part can be subjected to defect check to improve a detecting accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a technique for inspecting the appearance of the surface of an object such as a semiconductor wafer, and in particular, to an effective technique used to ensure alignment of patterns to be inspected. It is about a certain technology.

[Conventional technology]

For example, when mass producing LSIs (Large Scale Integrated Circuits), the most important issue is improving the yield of the wafer processing process for forming semiconductor elements. Most of the causes of this decrease in yield are poor appearance, and reducing this has become an important issue. For this reason, automation of wafer visual inspection is required.

By the way, the present inventor has discovered that semiconductor wafers, substrates, masks,
We investigated the problem of density conversion when using image processing to inspect the appearance of objects to be inspected, such as reticles and liquid crystals.

The following are the techniques studied by the present inventor, and the outline thereof is as follows.

In other words, in image processing for visual inspection, the inspection target is imaged with a television camera, etc., the image information is converted to multiple gradations, and this is compared in two different locations with the same pattern. Defects are compared based on the occurrence of defects (e.g., size, size, etc.). When comparing patterns, each image signal is differentiated to detect lines and edges and extract features.

Regarding the pattern differentiation method, for example, see "Image Recognition Theory" by Makoto Nagao (published by Corona Publishing, February 1982) 4
It is described on pages 5 to 53.

In addition, as an example of defect inspection of patterns of printed wiring boards, etc., there is Utility Model Application Publication No. 1-195350, which converts the reflected light from the pattern part into an electrical signal and compares it with a reference voltage to obtain the binary value of the film. A configuration for performing the conversion and making the determination is shown.

5 Problems to be Solved by the Invention] However, the above-mentioned "Image Recognition Theory" does not disclose a specific means for optimally and automatically setting a differential formula or differential threshold depending on the object; The inventors have discovered that there is a problem in that conditions must be determined empirically and experimentally based on the image signal.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to provide a technique that can automatically determine an optimal differential equation and differential threshold in response to changes in products or processes.

The above and other objects and novel features of the present invention will become apparent from the description of this specification and the accompanying drawings.

[Means to solve the problem]

A brief overview of typical inventions disclosed in this application is as follows.

That is, it is an automatic visual inspection device that aligns the same pattern parts at two different locations on an object to be inspected and determines a defect based on the fact that there is a difference of more than a certain level between the patterns at the two locations, and which detects an image signal of the pattern. A processing means is provided for setting an optimal differential equation depending on the shading of the image, differentiating the image signal based on this differential equation, and performing the positioning based on the differential value.

[Effect]

According to the above-mentioned means, the differential signal obtained by differentiating the image signal of the pattern using the set optimal differential equation indicates a position where the direct density change is large, that is, a position where an edge exists;
By comparing differential signals at two points of the same pattern, the amount of relative positional deviation between the patterns can be determined. Thereby, it becomes possible to use only the missing WIIfN) for defect determination, and the detection accuracy can be improved. Also,
The differential threshold value that is set when the differential signal is binarized is set to an optimal value depending on each object to be inspected.

Therefore, it is possible to always set optimal conditions even if the inspected object (sample) is different.

〔Example〕

FIG. 1 is a block diagram showing an embodiment of an automatic visual inspection apparatus according to the present invention.

As shown in FIG. 1, a sample stage 2 is attached to the upper surface of an X-Y stage l that can freely move in the X and Y directions, and a sample (in this example, a semiconductor wafer) is placed on the sample stage 2.
3 is set. On the other hand, a light fi4 is provided to illuminate the surface of the sample 3, which is the object to be inspected, and a condenser lens 5 is disposed on the optical path of the light fi4.

An objective lens 6 is disposed above the sample 3, and a half mirror 7 is disposed above this and on the output optical path of the condenser lens 5. Furthermore, an imaging means 8 is arranged at the focal position of the objective lens 6.

This imaging means 8 photoelectrically converts the reflected light from the sample 3, and is a -dimensional line sensor or a two-dimensional ITV.
(Industrial TV) Constructed using a camera. A signal processing circuit 9 for amplifying the image signal, correcting distortion, A/D conversion, etc. is connected to the imaging means 8, and this signal processing circuit 9 has an analog signal for converting the analog signal into a digital signal. /digital (A/D) converter 10 is connected, and to this A/D converter 10, a delay memory 11, a differentiation circuit 12a1, and a defect detection section 13 are connected.

A deviation correction section 14 is inserted between the delay memory 11 and the defect detection section 13, and the delay memory 11 further includes a differentiation circuit 12.
b is connected. A deviation amount detection circuit 15 is connected to the differentiating circuits 12a and 12b, and its output is connected to the deviation correction unit 1.
4.

An image memory 16 for storing an output image of one of them is connected to the delay memory 11 and the differentiation circuit 12a, and a main control 817 using a microcomputer is connected to the differentiation circuits 12a, 12b and the image memory 16.

In the above configuration, in order to conduct an external appearance inspection, first, the sample 3 is placed on the sample stage 2, and the light source 4 is turned on. The output light reaches a half mirror 7 through a condenser lens 5, and further reaches a specific pattern portion of one chip on a sample 3 through an objective lens 6. The reflected light from the illuminated portion of the sample 3 passes through the half mirror 7 and forms a pattern image on the imaging means 8 .

The image signal photoelectrically converted by the imaging means 8 is processed by the signal processing circuit 9, and then converted into multiple gradations (256 gradations for 1 byte) by the A/D converter 10, and the signal is It is temporarily stored in the delay memory 11.

Next, the same pattern of another chip on the same wafer is irradiated with the light source 4, and after being received by the imaging means 8, it is processed by the signal processing circuit 9 and the A/D converter 10.
2a and performs differentiation, and generates an output signal when the value exceeds a certain threshold value. Further, image data is read out from the delay memory 11, inputted to the differentiating circuit 12a to be differentiated, and an output signal is generated when the value exceeds a certain threshold value.

Each output of the differentiating circuits 12a and 12b is transmitted to the deviation amount detection circuit 1.
5, how far can the degree of coincidence be maximized by shifting in the X and Y directions (or is the degree of mismatch small?)
Detect.

On the other hand, the image information of the A/D converter 10 of the chip currently being detected and the image information of other chips stored in the delay memory 11 are input to the defect detection section 13 and a comparison process for defect detection is performed. It will be done. At this time, the output of the delay memory 11 is used to correct the relative positional shift between the two images by the shift correction section 14 according to the shift amount given from the shift amount detection circuit 15. This eliminates misalignment in the normal part of the pattern. The defect detection section 13 compares the output of the deviation correction section 14 and the output of the A/D converter 10, and determines a defect having a brightness difference and size larger than -window.

Processing such as data setting, control, and judgment for each circuit is performed by the main control unit 17.
It is used for temporary storage according to the selection of a.

Next, pattern differentiation will be explained.

As described on pages 46 to 54 of the above-mentioned document "Image Recognition Theory", -th order differential or second order differential is generally used to detect edges of a pattern, which is one means of defect detection. The method of differentiation is as described in the literature,
There are many types. For example, if the image signal is expressed as f (x, y), it will be as follows.

-4f (x, y)+f (x, y+1)+f
(x'-, 1, y) + f (x-1, y) + f (x,
y-1) Through this calculation, the X and Y directions are simultaneously differentiated.

Further, differentiation only in the X direction can be performed by the following calculation.

-2f (x,y)+f (x-1,y)+f
(x+1.y) FIG. 2 is a block diagram showing details of the differentiating circuits 12a and 12b.

At the input stage, there is a line shifter 18a for shifting a signal for one line in the operating direction of the line sensor of the imaging means 8.
~18d are connected in series, input point and line shifter 18
A plurality of 1-bit shift registers 19 are connected to the output and input of each of a to 18d. Each output terminal of the shift register 19 is connected to a differential calculation circuit 20 , and the main control section 17 is connected to the differential calculation circuit 20 via a differential mode register 21 . A differential threshold register 22 is connected to the main control a17, and a differential binarization section 23 is connected to this output terminal and the output terminal of the differential calculation circuit 20.

In the configuration of FIG. 2, line shifters 18a to 18d
Then, the shift register 19 cuts out signals at 25 points (5×5). On the other hand, the differential calculation circuit 20 has 25
Select multiple input signals from among the point signals and perform differential operation. At this time, the type of arithmetic expression is set by the differential mode register 21, and the arithmetic expression of the differential arithmetic circuit 20 is changed according to this set value. The differential threshold value set in the differential threshold register 22 is used for the calculation formula of the differential calculation circuit 20, and the differential binarization unit 23 performs binarization processing, and the result is sent to the defect detection unit 13. .

Next, we will explain how to determine the differential equation in Figures 3 (a) and (b).
).

This will be explained with reference to (C).

FIG. 3(a) schematically shows a pattern imaged by the imaging means 8, and the shading of the pattern on the xx' cross section in the same figure is indicated by brightness '!'! J3 et al.]. The signal waveform obtained by differentiating the grayscale signal of FIG. 3(b) is shown in FIG. 3(C).

The signal value of the differential signal differs depending on the difference in shading of the pattern, the size of the pattern, and the like. Furthermore, the signal value differs depending on the differential equation. Therefore, the A/D converter 10
The image signal after the D conversion is taken into the image memory 16, and a calculation is performed in the main control s17 using a plurality of candidate differential expressions to check the peak value of the differential value. In the example of Figure 3(C),
v3 in the figure is the peak value.

Whether or not edges of a pattern can be detected with good contrast depends on whether the peak value of the differential value is high or not, so the differential expression with the highest peak value of the differential value is determined to be the optimal differential expression. Note that the average value of the absolute values of the differential values may be calculated, and the arithmetic expression for which this average value is the highest value may be set as the optimal differential expression.

The optimum differential expression thus obtained is set in the mode register 21.

Next, a method for setting the differential threshold value will be explained.

Third! I! In J(C), as the threshold value is increased, the number of edges that can be binarized at VI, Vz, and Vs in the figure changes. FIG. 4 shows the relationship between the number of edges detected and the threshold value.

As shown in FIG. 4, as the threshold value increases, the number of detected edges decreases. Since there is a large variation in edge detection between the thresholds V,, V,, V, setting a differential threshold in this region will increase the variation in the number of edges detected, making it impossible to perform stable detection. . Therefore, the differential threshold value may be set between 0 and V+ in the example shown in FIG. Therefore, the threshold value vth is set to a value slightly lower than V in FIG. If the tendency of change in the number of detected edges does not differ much depending on the object to be inspected, vth may be set as the peak value V of the differential value multiplied by a certain ratio.

FIG. 5 is a flowchart showing a method for determining the optimal differential equation.

First, in step 51), an image signal captured by the imaging means 8 is manually calculated by differential calculation CB(N) using the differential equation (N).
, i)) is performed (step 52).

Here, 1 means the first threshold. Step 5
2 is executed N times (step 53).

Next, an evaluation function H(N) is found (step 54), and N that gives the maximum value is found out of this H(N), and this is set as the optimal differential equation (step 55).

FIG. 16 is a flowchart showing a method for determining the optimal differential threshold.

First, find the initial differential threshold Th (step 61
). Next, the number of edges equal to or greater than the threshold value Th (i) is calculated [→E(i>)L (step 62), and this is executed M times (step 63). That is, by adding a certain ratio to Th, Th(1) is obtained (step 64).
. Through steps 62 and 63, it is possible to graph the relationship between the threshold value and the number of edges as shown in FIG. Next, M
When it is determined that the calculation of the number of edges is completed (step 63
), the optimum threshold value Vrh (a point with little variation) is determined (step 65).

Although the invention made by the present invention has been specifically explained based on Examples above, it goes without saying that the present invention is not limited to the above-mentioned Examples and can be modified in various ways without departing from the gist thereof. .

For example, in the above embodiment, the differential equation is calculated using an A/D
This is performed by the main control unit 17 based on the converted image signal, and candidate differential expressions are sequentially set in the mode register 21, and the output of the differential calculation circuit 20 for each differential expression is stored in the image memory 16. The present invention can be similarly achieved by making a determination using the main control B17 based on the peak value of the differential value or the average value of the absolute value.

In addition, in FIG. 2, one differential threshold register 22 and one differential binarization section 23 are provided, but
When performing differentiation individually in the Y direction, two may be provided for each.

More than! ! Hu mainly explained the case in which the invention made by the present inventor is applied to the visual inspection of semiconductor wafers 1, which is the field of use thereof, but the invention is not limited to this and can be applied to pattern inspection of masks, reticles, printed circuit boards, etc. is possible.

〔Effect of the invention〕

Among the inventions disclosed in this application, the effects obtained by typical inventions can be briefly explained as described by Geromi.

That is, automatic visual inspection i aligns the same pattern parts at two different locations on the object to be inspected, and determines a defect if there is a difference of more than a certain level between the patterns at the two locations.
t, a processing means is provided for setting an optimal differential equation according to the density of the image signal of the pattern, differentiating the image signal based on this differential equation, and performing the positioning based on the differential value. This makes it possible to perform accurate positioning, improve the detection rate, and reduce the false detection rate. In particular, when used for visual inspection of semiconductor wafers, it is possible to improve the yield of semiconductor devices.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of an automatic visual inspection apparatus according to the present invention, FIG. 2 is a block diagram showing details of a differential circuit, and FIG.
), (b), and (C) are explanatory diagrams explaining the method for determining the differential formula, Figure 4 is a characteristic diagram showing the relationship between the number of detected edges and the threshold value, and Figure 5 is a diagram showing the method for determining the optimal differential formula. Flowchart FIG. 6 is a flowchart showing a method for determining the optimal differential threshold, and FIG. 7 is a correlation diagram showing the relationship between the threshold determined in FIG. 6 and the number of edges. DESCRIPTION OF SYMBOLS 1... X-Y stage, 2... Sample stand, 3... Sample, 4... Light source, 5... Condensing lens, 6... Objective lens, 7... Half mirror 8 ...imaging means, 9.
...Signal processing circuit, 10...A/D converter, 11...
・Delay memory, 12a, 12b... Differential circuit, 13.
. . . Defect detection section, 14 . Line shifter, 19... Shift register, 20... Differential operation circuit, 21... Differential mode register, 22...
- Differential threshold register, 23...differential binarization section. Agent Patent Attorney Daiwa Tsutsui Diagram-〉Distance diagram Diagram 4 Diagram yh Threshold value-〉 Diagram

Claims (1)

[Scope of Claims] 1. An automatic visual inspection device that aligns the same pattern parts at two different locations on an object to be inspected and determines a defect based on the fact that the patterns at the two locations have a difference of more than a certain level. , characterized in that processing means is provided for setting an optimal differential equation according to the density of the image signal of the pattern, differentiating the image signal based on this differential equation, and performing the positioning based on the differential value. Automatic appearance inspection equipment. 2. The automatic visual inspection apparatus according to claim 1, wherein the differential equation is set as the optimal differential equation when the peak value of the signal resulting from the differentiation is maximized. 3. The automatic visual inspection apparatus according to claim 1, further comprising means for binarizing the differential result using a differential threshold, and the differential threshold is set based on the number of detected edges. 4. The differential threshold value is set relative to the maximum value of the differential value,
4. The automatic appearance inspection apparatus according to claim 3, wherein the value is set to a constant ratio. 5. The automatic method according to claim 3, characterized in that the differential threshold value is made variable, a change in the number of edge detections at that time is determined, and a differential threshold value for a portion where the change is small is determined as an optimum threshold value. Appearance inspection equipment.