JP3435578B2 - Pattern inspection 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 automatically inspecting a pattern formed on a film carrier or the like by etching.

[0002]

2. Description of the Related Art Conventionally, a film carrier used for mounting ICs and LSIs has a polyimide film having a thickness of about 75 to 125 μm, a copper foil attached with an adhesive, a photoresist applied on both sides, and a mask. Exposure, development and etching are performed to form a lead pattern.

After the pattern is formed in this manner, the photoresist is removed, and the surface of the lead is plated with Sn, Au and solder, and the film carrier process is completed.
After the completion of this process, the pattern is visually inspected by a human using a microscope.

[0004]

Problems to be Solved by the Invention As described above, there are problems in that visual inspection of a fine pattern requires skill and results in heavy use of the eyes. on the other hand,
As an alternative to the visual inspection, it is conceivable to use a pattern matching method in which the pattern is imaged by a TV camera and the inspection is performed based on the matching rate with the measured pattern as the reference pattern. However, the coincidence rate is measured on a pixel-by-pixel basis as represented by the following formula, and is not suitable for inspecting a fine pattern such as a film carrier. Matching rate = {(entire pixel-non-matching pixel) / (entire pixel)} × 100% In the case of a pattern in which leads of the same shape are continuous, the lead is There is a case in which the positions are shifted by one, and there is a problem in the case of repeating the pattern of the same shape.

[0005]

[0006]

SUMMARY OF THE INVENTION The present invention is directed to a measured pattern obtained by converting the edge data of a non-defective measured pattern into a set of straight lines, and both of these straight lines which are located opposite to each other. The width W between the straight lines and the center line L _M are registered as master pattern information, the measured pattern is stored in the image memory, the master pattern and the measured pattern are aligned, and the inspection range is set. Then, the periphery of this inspection range is inspected in a fixed direction to extract the edge data of the measured pattern, and a straight line that divides the angle between the straight lines adjacent to each other in the master pattern into two equal parts is calculated, By this straight line, the master pattern is divided into areas corresponding to the respective straight lines, and the edge data of the measured pattern existing in each of the areas is made to correspond to the straight lines of the master pattern, respectively. Measured parameter - down of Ejjide - data and Masutapata - good to detect an error between the straight down, but which is adapted to determine defective.

[0007]

The grayscale image of the pattern to be measured, which is determined to be non-defective, is stored in the image memory, and the grayscale histogram of the grayscale image stored in this image memory is binarized to obtain the above-mentioned grayscale image. The edge data of the measurement pattern is obtained, the straight line is registered from this edge data by the least squares method, and the center line L _M and the width W of both straight lines facing each other are obtained, and this is used as the master pattern. It is registered as information and this is used as a reference pattern. The edge data of the measured pattern is associated with each straight line of the master pattern, the two are compared and collated, the error is inspected, and the non-defective product or the defective product is determined.

[0008]

Embodiments of the present invention will be described in detail with reference to FIGS. FIG. 1 shows an embodiment of the present invention, a processing flow of a master pattern M, FIG. 2 is a processing flow for associating a measured pattern m with a master pattern M, and FIG. FIG. 4 is a basic operation diagram of the pattern inspection apparatus of the present invention, FIG. 4 is a system configuration diagram of the present invention, and FIGS. 5 to 18 are explanatory diagrams for explaining a pattern inspection method according to the present invention.

First, a specific pattern inspection apparatus will be described with reference to the basic operation diagram shown in FIG. 3 and the system configuration diagram shown in FIG. In this embodiment, a pattern of TAB tape is taken as an example of the pattern to be measured m, and the case of inspecting the pattern will be described.

Referring to FIG. 3, the pattern inspection apparatus according to the present invention comprises an unwinding section 1 for feeding a film of a TAB tape (not shown) wound around a reel, a fixed-size feeding section, and a positioning section 2. A detection unit 3 for determining a non-defective product and a defective product, a constant-size feeding / positioning unit 4 for sending the tape for which the determination process has been completed to a cutting process, a punching unit 5 for cutting a defective product tape,
It is composed of a winding portion 6 for winding the tape again on the reel.

FIG. 4 shows a system configuration diagram. The system control section 7 controls the entire system, and a menu is displayed.
Color CRT8 that displays the results and results, product type setting, registration,
A keyboard 9, a sequencer unit 10, a printer 11, an image memory 12, a CPU (measuring unit) 13, a signal selector 14 and the like for selecting an operation menu are controlled.

Under the control of the system control section 7, the sequencer section 10 controls the operation switch / indicator lamp 15, the tape planner section 16, the tape puncher section 17, the XY table 18 and the like. is doing. 19 is a floppy disk.

The motor 20 is driven by the driver 20 in the X direction.
Through the sequencer unit 10, a table controller (3
Axis) 22 and Y-S-P driver 23 to control the motor 24 in the Y direction and the motor 25 for the span axis, and
The -Y table 18 is drive-controlled in the X direction, the Y direction, and the span axis direction, respectively. Therefore, the TAB tape is picked up by the cameras 26 and 27 on the XY table 18, and the image is digitally converted by the A / D converter 28 and stored in the image memory 12 on a pixel-by-pixel basis. Detection unit 3
Is the camera 26, 27 and its moving mechanism part, XY target.
It is composed of a bull 18, a CPU 13, an A / D converter 28, and an image memory 12.

Next, the operation will be described. The TAB tape wound around the reel is the sequencer section 10
Under the control of (1), the tape planner unit 16 sends the frames one by one to the predetermined positions of the cameras 26 and 27 on the XY table 18 and positions them. XY table 18 is Y
And the drive motors 24 and 21 in the X-axis direction and the table controller 22 controlling the three-axis directions.

One frame TAB positioned at a predetermined position
The pattern of the tape is picked up by the two cameras 26 and 27, and the grayscale image thereof is converted into a digital signal by the A / D converter 28 and stored in the image memory 12 in pixel units. On the other hand, in the image memory 12, the master pattern information of the TAB tape determined to be non-defective at the start of the inspection is created and stored. This master pattern M and TA
The B pattern (measured pattern m) is compared in the detecting section 3 by a method to be described later, and a non-defective product / defective product is determined.

In this way, the TAB tape, which has been subjected to the processing of determining whether each frame is a good product or a defective product, is sequentially sent to the tape puncher unit 17 for each frame, and after the defective product is punched out, In the winding section 6, the reel is wound again and the inspection is completed.

Next, a method of inspecting the TAB pattern for determining whether the TAB pattern is a good product or a defective product in the detection unit 3 will be described with reference to FIGS. 1 and 2. Prior to the inspection, it is necessary to prepare a reference master pattern M from the TAB tape determined to be non-defective and register it in the image memory 12 as master pattern information.

A method of creating the master pattern M will be described below with reference to FIG. Black and white cameras 26, 27
The grayscale image of the pattern of the TAB tape which is determined as a non-defective product by imaging in FIG. 5 shows a part of the pattern, and FIG. 6 and FIG. 7 show enlarged views thereof. Then, it is once registered in the image memory 12 (step 40). 5 and 6
As shown in, the edge coordinates (edge data) are extracted on the assumption that the grayscale image intersects with the threshold value SL at the midpoint where the gray image changes from white to black and from black to white. (Step 41).

[0019] Masutapata - the emission M Ejjide - the data extraction is binarized from the density histogram of the entire image, crosses the threshold S _L white on both sides of the threshold SL, the black pixels ratably X or Y coordinate value (hereinafter, edge data
N ₁ (4.5, ₂ ), N ₂ (5, 2.5),
N ₃ (5.5, ₃ ) ... Is obtained (step 4)
1).

Next, as shown in FIG. 8, each edge data N ₁ , N ₂ , N ₃ ... Of the master pattern M, which is the set of points thus obtained, is calculated by the least squares method. .. are straightened as straight lines A ₁ , A ₂ , A ₃ ... (Step 4
2). In this case, each edge data N ₁ , N ₂ , N
₃ ... beat perpendicular to a straight line A _1, A _2, A _3, ... from each of Ejjide - from data N _1, N _2, N ₃ ... up straight A _1, A ₂ · · Distance ΔN ₁ , ΔN ₂ , ΔN
_{If 3} ... Is 0.3 pixels or less, it is included in the straight lines A ₁ , A _2, ... As straight edge data. In this way, Masutapata are extracted as a set of points - the emission M Ejjide - data _{N 1, N 2, N 3} ···
Is linearized, and finally the master pattern M is straight line A
It is converted into a set of ₁ , A ₂ , A ₃ ... (Step 4)
2).

Next, the width W of the master pattern M is obtained. As shown in FIG. 9, the straight line A ₁ and the straight line A _n, a straight line A ₂ and the line A _n which face each other - find the center line L _{M 1} ..., the center line de - data is Masutapata - down information Is registered as (step 43). This information is used to check the directionality of the lead when inspecting the measured pattern m.

When a straight line H _M perpendicular to the obtained center line L _M is drawn, this straight line H _M is the straight lines A ₁ and A _n and A ₂ and A _{n -1.}
Width of master pattern M is the width W up to the point where
Is equal to That is, the center line L _M becomes the locus of the center of the inscribed circle whose diameter is the distance between the straight lines A ₁ and A _n, and the diameter is the width W of the master pattern M (step 4).
4). The master pattern M sequentially obtained in this way
Straight lines A ₁ , A ₂ , A _3, ..., Center line L _M and width W
Is registered as master pattern information, and a master pattern M serving as a reference is created (step 45).

Next, the measured pattern of the TAB tape is actually measured.
M is inspected by the procedure shown in FIG. The TAB tape is wound on a reel, fed out one frame at a time from the unwinding unit 1, and the cameras 26 and 27 capture an image of the measured pattern m. At this time, the captured image becomes a grayscale image as shown in FIG. 5, which is digitized by the A / D converter 28 and temporarily stored in the image memory 12. The master pattern M and the measured pattern m stored in the image memory 12 are read by the CPU 13 of the detection unit 3,
After alignment, the patterns to be measured are compared and collated to inspect the measured pattern m (step 46).

Here, the measured pattern m is set as a master pattern.
Fig. 1 shows the prerequisites for comparing and collating
A description will be given based on 0. (1) Masutapata - emission M and the measured pattern - Ejjide of emissions m - data _{n 1, n 2, n 3} ··· , for example, as one method for setting the inspection range, is masked by the window 30, the window Only the patterns in 30 are subject to inspection (step 47). (2) The measured pattern m, which is partially cut out by the window 30, is always the window 30 as shown in FIG.
The pattern portion having the shape shown by the pattern 31 which is in contact with any one of the four sides is not recognized as edge data. (3) The inspection always proceeds from the starting point S around the four sides of the window 30 in a fixed direction, and finally returns to the starting point S to complete the screen inspection.

Under such preconditions, the pattern m to be measured is inspected by the following procedure. FIG. 11 shows the state at the time of inspection, and shows the edge data of the master pattern M.
Is linearized by the method of least squares, and straight lines A ₁ , A ₂ ,
A ₃ are indicated by.., Extracted the measured pattern - the emissions m Ejjide - data _{n 1, n 2, n 3} ··· are indicated by a black circle (48).

Therefore, first, correspondence must be made to which straight line of the master pattern M the edge data n ₁ , n ₂ , n ₃ ... Of the measured pattern m correspond to. . In order to make this correspondence, as shown in FIGS. 11 and 12, straight lines A ₁ of the master pattern M adjacent to each other are provided.
And the straight line A ₂ , the straight line A ₂ and the straight line A ₃ have an angle a,
Straight lines A ₂ ′, A ₃ ′, etc. that divide b ... into two equal parts are calculated.

Therefore, the master pattern M is the straight line A
₂ ', A _3' is divided in., The straight line _{A 1, A 2, A 3} ·
Areas corresponding to (hereinafter, areas A ₁ , A ₂ , A ₃ ...
It means being divided into a referred), the measured pattern - the emissions m Ejjide - data _{n 1, n 2, n 3} ··· are all areas A _1, A _2, A ₃ correspond to ... Attached (49).

Next, the error amount (deviation) Δd between the master pattern M and the measured pattern m must be measured. For this purpose, as shown in FIG.
Straight line A from the edge data n ₁ , n ₂ , n ₃ ...
₁ and A ₂ ... Perpendicular lines h ₁ , h ₂ , h ₃ ...
The down distance [Delta] d ([Delta] d ₁ this perpendicular line _{h 1, h 2, h 3} ··· is, crossing the straight line A _1, A ₂ · · · respectively,
When determining the [Delta] d ₂ · · ·), the amount of this distance [Delta] d is an error, i.e., Masutapata - emission M and the measured pattern - a shift amount between the emission m (50).

Next, the case where various inspections are actually carried out will be described (51). (1). Inspection of disconnection As shown in FIG. 11, the edge data of the master pattern M is converted into straight lines A ₁ , A ₂ ...
It is divided into areas A ₁ , A ₂ . On the other hand, the measured pattern - data n _1, n - Ejjide showing the boundary coordinate of emissions m
₂ and n ₃ ... Are labeled and extracted as a set of continuous edge data n ₁ , n ₂ , n ₃ .

The edge data n ₁ , n ₂ , n ₃ ... (Indicated by black circles) of the extracted measured pattern m correspond to which straight line of the master pattern M, respectively. Are divided into areas A ₁ , A _2, ...

Therefore, the determination as to whether or not there is a disconnection is made by referring to FIG.
As shown in FIG. 4, the edge data n of the measured pattern m
₁ , n ₂ ... N ₉ are the areas A ₁ of the master pattern M,
Corresponds to A ₂ ... A ₇ . In this case, the edge data n ₁ , n ₂ , and n of the measured pattern m, which are associated with the areas A ₁ , A _2, ... A ₇ of the master pattern M, respectively.
n ₃ ... Is as follows.

Area A ₁ ... Edge data n ₁ , n ₂ , n ₃ Area A ₂ ... Edge data n ₄ , n ₅ Area A ₃ ... Edge data 0 Area A ₄ ... Edge data 0 area A ₅ ... Edge data 0 area A ₆ ... Edge data n ₆ and n ₇ Area A ₇ ... Edge data n ₈ and n ₉

As is clear from this result, since there is no edge data of the measured pattern m corresponding to the areas A ₃ , A ₄ and A ₅ of the master pattern M, the disconnection occurs in such a case. Is determined.

(2). Short Circuit Inspection FIG. 15 shows a state during a short circuit inspection. As shown in the figure, the master pattern M lead and the measured pattern m lead are If the positions in contact with the windows 30 are the lead position and the lead positions ',', the edge data of the measured pattern m is different from the master pattern M at the other positions. Must be judged to be a short circuit.

Therefore, in order to determine a short circuit as described above, first, edge data is extracted for each lead of the pattern m to be measured, and as shown in FIG.
Labeled as a, b, c, d ... S, t, u.

When the measured pattern m has a short circuit,
The edge data of the labeled read pattern 'is a, b, c, j, k, l, m, n, o, p, q,
r, s, t, u, d, e, f, g, h, i, a,
Edge data j, k, l, m of lead pattern '
n, o, p, q, r, s, t, u are extracted. However, such edge data j, k, l, m, n, o,
Since p, q, r, s, t, and u are not registered as the straight line data of the read pattern of the master pattern M, in this case, the patterns of the measured pattern m are read. -It is judged that the short pattern and the red pattern are short-circuited.

(3). Inspection of Pattern Width FIG. 17 shows a state at the time of inspecting the thinness. As shown in FIG. 13, edge data n ₁ , n of the measured pattern m are measured.
₂ , n _3, ... From the straight lines A ₁ , A of the master pattern M
_A perpendicular line is drawn on ₂ ..., and the length Δd (error amount) up to this perpendicular line (Δd ₁ , Δd ₂ ...)
When is small, it is determined as a good product as it is.

For edge data having a large error amount .DELTA.d, a perpendicular is drawn to the center line L.sub.M of the master pattern _M , the distance is determined, and the edge data of the measured pattern m facing each other is obtained. The width w (w ₁ , w ₂ ...) Of the data is obtained.
In the case of this embodiment, when the width W of the master pattern M is set, is the width w of the measured pattern within the range of (2/3) W <w <(4/3) W? Looking at whether or not it is within this range, it is determined to be a non-defective product, and if it is out of the range, it is determined to be a defective product as thin or thick. However, in general, when the determination coefficients are A and B (however, A <B), when the width w and the width W of the master pattern are AW <w <BW, it is determined as a good product, When the width w is out of this range, the pattern width is determined to be unsuitable and the product is determined to be defective.

(4). Inspection of misalignment (bending) FIG. 18 shows a state at the time of inspecting misalignment, in which the measured pattern m corresponding to the straight lines A ₁ , A ₂ ... Of the master pattern M, respectively. The edge data of the measured pattern m is changed from the edge data of the measured pattern m to the straight lines A ₁ , A _2, ...
・ Put a perpendicular line on and the length (error amount) to this perpendicular line Δd
₁ , Δd ₂ , Δd ₃ , ... Δd _n are calculated.

When the lead width of the master pattern M is W, only when this error amount Δd _n is Δd _n > (1/2) W, it is judged that the position is misaligned and it is judged as a defective product. in the case of, i.e., the error amount [Delta] d _n Galli - when de width W / 2 less than is judged to be good.

The pattern m to be measured is the same shape.
If it is formed as a repeated pattern,
There is a case where the master pattern M and the measured pattern m are aligned with one lead pattern shifted, but in this case, there is no determination result as the position shift, but
At the final stage of the inspection, the edge data of the master pattern M corresponding to the final read pattern of the measured pattern m does not exist, and it is judged as a defective product. Can check.

[0042]

[0043]

According to the present invention, the measured pattern obtained by converting the edge data of the non-defective measured pattern into a set of straight lines,
Width W between these straight lines located opposite each other
And the center line L _M are registered as master pattern information,
The measured pattern is stored in the image memory, the master pattern and the measured pattern are aligned with each other, the inspection range is set, and the periphery of the inspection range is inspected in a certain direction to measure the measured pattern. Edge data and extract the master pattern.
A straight line that divides the angle between the straight lines adjacent to each other into two equal parts is divided, and the master pattern is divided into regions corresponding to the respective straight lines by this straight line, and the measured patterns existing in the respective regions are divided. The edge data is associated with each straight line of the master pattern, and the error between the edge data of the measured pattern and the straight line of the master pattern that are associated with each other is detected to judge whether the product is a good product or a defective product. I did so,
Any pattern can be applied to the measured pattern.
Further, it is possible to automatically inspect the measured pattern, which has conventionally been visually inspected by a human. Therefore, the labor cost is greatly saved. In addition, since it is possible to eliminate inaccuracies due to individual differences in the person inspecting, accurate inspection results can be obtained.

[Brief description of drawings]

FIG. 1 shows an embodiment of the present invention, in which a master pattern
Processing flow.

FIG. 2 shows an embodiment of the present invention, in which a master pattern
Processing flow for processing the correspondence between the pattern and the measured pattern.
Is.

FIG. 3 is a basic operation diagram of a pattern inspection device showing an embodiment of the present invention.

FIG. 4 is a system configuration diagram showing an embodiment of the present invention.

FIG. 5 shows an embodiment of the present invention, in which the measured pattern
Is part of the

FIG. 6 shows an embodiment of the present invention, in which a master pattern
FIG. 6 is an explanatory diagram for obtaining edge data of the image, and is an enlarged view of FIG. 5.

FIG. 7 shows an embodiment of the present invention and is an enlarged view of FIG.

FIG. 8 shows an embodiment of the present invention, which is a master pattern.
It is an explanatory view for making a straight line.

FIG. 9 shows an embodiment of the present invention, which is a master pattern.
It is an explanatory view for obtaining width W of the input.

FIG. 10 shows an embodiment of the present invention and is a measured pattern displayed on a window.

FIG. 11 shows an embodiment of the present invention and is a diagram showing the relationship between the straight line of the master pattern and the edge data of the measured pattern.

FIG. 12 shows an embodiment of the present invention and is an explanatory diagram for dividing an area of a master pattern.

FIG. 13 illustrates an embodiment of the present invention and is an explanatory diagram for obtaining a deviation (error amount).

FIG. 14 shows an embodiment of the present invention and is an explanatory diagram of an inspection for disconnection.

FIG. 15 shows an embodiment of the present invention and is an explanatory diagram of a short circuit inspection.

FIG. 16 shows an embodiment of the present invention and is an explanatory diagram of a short circuit inspection.

FIG. 17 is a view for explaining the pattern width inspection according to the embodiment of the present invention.

FIG. 18 shows an embodiment of the present invention and is an explanatory diagram of inspection for positional deviation (bending).

[Explanation of symbols]

M master pattern A ₁ , A ₂ ... master pattern M straight line N ₁ , N ₂ ... master pattern edge data H _M master pattern straight line L _M master pattern center line W master pattern Width Δd error amount (deviation) m measured pattern n ₁ , n ₂ ... Edge data of measured pattern 12 image memory 30 window

Claims (6)

(57) [Claims] 1. The measured pattern obtained by converting the edge data of a non-defective measured pattern into a set of straight lines, and a width W and a center line between both straight lines of these straight lines which are located opposite to each other. Masutapata and L _M - registered as down information to be measured pattern - storing in to the image memory, the Masutapata - ting the measured pattern - as well as the down align, set the inspection range, the inspection range Of the measured pattern is extracted by extracting the edge data of the measured pattern, and a straight line that bisects the angle between the adjacent straight lines of the master pattern is calculated. Is divided into areas corresponding to the respective straight lines, and the edge data of the measured pattern present in the respective areas are respectively associated with the straight lines of the master pattern, and the respective associated Measurements pattern - emissions of Ejjide - motor and the Masutapata - good to detect an error between the straight down, pattern characterized by the determination of the defective - down method of inspection. 2. A perpendicular line is drawn from the edge data of the measured pattern to the straight line of the associated master pattern, and the length Δd to this straight line is taken as the error amount. The pattern inspection method according to Item 1. 3. When the edge data of the measured pattern to be associated with the straight line of the master pattern does not exist, it is determined as a defective product as a disconnection of the measured pattern. The pattern inspection method according to claim 1. 4. When the edge data of the labeled pattern to be measured is made to correspond to the straight line of the master pattern and a surplus occurs in the edge data of this pattern to be measured, the edge data of this pattern is measured. The method for inspecting a pattern according to claim 1, wherein the measurement pattern is judged as a short circuit due to a short circuit. 5. When the error amount Δd is large, a perpendicular line is drawn from the edge data of the master pattern, which is opposed to each other, to the center line L _M of the master pattern, and the perpendicular line is drawn from the edge data of the opposed patterns. By obtaining the distance to the center line L _M , the width w of the edge data
When the determination coefficients are A and B (however, A <B), when this width w and the width W of the master pattern are AW <w <BW, it is determined as a non-defective product and the width 3. The pattern inspection method according to claim 2, wherein when w is out of the range, the pattern width is determined to be unsuitable and a defective product is determined. 6. When the error amount Δd is 1 / N or more (where N = 1, 2, 3 ...) Of the width W of the master pattern (where N = 1, 2, 3 ...), a defective or defective product is obtained. The method for inspecting a pattern according to claim 2, wherein