JPH08185438A - Cad device for pattern image - Google Patents

Abstract

PURPOSE: To generate raster data to be freely enlarged/reduced on a CAD with a less data amount in comparison with a run length vector. CONSTITUTION: The raster data of an object are converted into the run length vector described by both the coordinates of a start point and an end point for each scan line by raster/vector conversion and when the first run length vector is overlapped with the second run length vector following this first run length vector, a first trapezoidal form with both the vectors as upper and lower bottoms is generated. Concerning the run length vectors after the third run length vector, the new trapezoidal forms are successively generated as well by repeating the similar operation within an allowable range. Then, CAD data are formed by parallelly moving the respective upper and lower bottoms of the final trapezoid to the outside just by the dimension corresponding to the 1/2 picture elements of the raster data.

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION The present invention relates to a pattern image CAD.
The present invention relates to a CAD device for pattern images, which is suitable for application in image measurement of a precision etching product.

[0002]

2. Description of the Related Art ICs to be mounted as etching products
There is a lead frame used to electrically connect to the (integrated circuit) chip.

FIG. 27 shows an example of a lead frame for one chip as viewed from one side. A die pad (island) 10 for mounting a chip (not shown) is located at the center of the lead frame. The die pad 10 is supported by an outer frame 12 via a tab suspension bar 14, an inner lead 16 is arranged around the die pad 10 with its tip close to the die pad 10, and an outer lead continuous with the inner lead 16 is provided. 18 is the above-mentioned outer frame 1 via the dam bar 20 etc.
Supported by 2. Also, the inner lead 16 has
A fixing tape 22 made of plastic is attached to prevent deformation of the leads 16. The broken line in the figure is the mold line.

Taking the lead frame as an example, an outline of the steps from etching product design to product completion is shown in FIG.
It becomes like 8.

CAD of lead frame pattern design
(Computer Aided Design), first, CAD data of the same product (A) as the target product size is created in the product pattern design process of CAD1, and then the actual etching is performed in the etching correction process of CAD2. The correction allowance for the side etching, which is excessively removed from the width of the resist pattern in the process, is defined as (A) product dimension C above.
(B) Machining dimension CAD data that is the prototype of the resist pattern is created by adding it to the AD data, and this machining dimension CAD data is drawn by a laser plotter in the next pattern manufacturing process, and the drawn pattern is printed on a glass plate. (C) A glass original plate pattern is created. This original plate pattern is created for both the front and back surfaces of the lead frame.

Then, using the glass original plate as a mask, the resist coated on the metal material such as the copper plate which is the base material of the lead frame is exposed (baked),
By developing and burning (curing) (D) to form a plate-making pattern (resist pattern), etching for removing the metal material in the exposed portion is performed, and then the adhering resist is peeled off. A lead frame, or (E) product pattern is obtained.

In the manufacturing process of the lead frame,
The (E) product pattern is preferably the same as the (A) product dimension CAD data which is the design pattern. Therefore, (B) processing dimension CAD data (this is (C) glass original plate pattern, which is designed by adding a correction allowance to this (A),
The size difference between (D) the pattern having the same size as the plate-making pattern) and (E) above is large, and usually has a difference of several tens of μm.

Similarly, as another etching product to be microfabricated, there is a shadow mask for a color television. In comparison with this, the lead frame has an irregular shape and the post-process performed after the etching is complicated. It has the feature of being

Further, as a feature of the lead frame, there is a gap (space to be removed by etching) between the island 10 to which the chip is attached and the tip of the inner lead to be wire-bonded, and the etching liquid easily enters there. The etching of the tip portion of the inner lead is likely to proceed and the tip tends to be tapered, but a sufficient tip width is required for wire bonding.

As described above, the CAD data (B) for forming the plate-making pattern (D), which serves as a mask for etching a lead frame which is difficult to process, is side-etched in excess of the resist pattern as described above. It is created by performing a correction for adding the size to be added to the product size CAD data of (A) as a correction margin. Conventionally, the correction allowance used for the above-mentioned etching correction has been set based on experience.

Further, it is necessary to measure the dimension of the actually manufactured product, for example, in order to certify that the dimension of the inner lead tip is within the tolerance (allowable range from the target value). In this case, conventionally, local dimension measurement is often performed for some leads, but there is also a demand for all leads.

However, in the method of determining the etching compensation amount based on experience as described above, the design of the processing size pattern of (B) is performed for a new pattern or a fine pattern having a fine pitch which has not been experienced in the past. If required, the correction allowance cannot be set appropriately, so that a defect occurs in the product dimension, and it is likely that it is necessary to correct again and etch again each time. In particular, a lead frame having a difficult fine pattern is subjected to trial and error many times, resulting in a large delay in delivery of the product.

Further, in the past, in general, only local dimension measurement was performed, so that it was easy to overlook a defective portion, but it took too much time to measure all leads (all pins), and as a result, It takes a lot of time and effort to determine the tolerances and certify the products.

As a method for solving the above-mentioned inconvenience,
It is extremely important that at least the design pattern and the product pattern are closely overlapped and collated. However, conventionally, there is no technique for accurately superimposing a design pattern and a product pattern, and when there is a request for collation between the two, for example, the design pattern of a CAD drawing drawn on a film by a magnifying projector and the actual product. Were overlaid and projected on the screen for verification.

Therefore, an optical microscope is used to input a two-dimensional precision pattern such as a lead frame as a divided image in raster data, which is raster-vector converted into CAD data, which is then converted into another CAD data by a CAD device, for example. It is conceivable to collate with the processing dimension pattern.

As a conventional raster vector conversion,
(1) A core line extraction method of extracting a center line from a raster image and converting it into vector data, (2) An outline (contour) extraction method of extracting white and black edges of the raster image and converting it into vector data, (3 ) A run length vector method is known in which a raster length vector corresponding to the length of a white or black area of a raster image is converted.

[0017]

[Problems to be Solved by the Invention] However, (1)
The core line extraction method in (2) cannot be applied to a wide area pattern because only the center line is extracted. The outline extraction method in (2) is for the inside and outside of the pattern, and in the case of a lead frame, the hole part and the metal part. Area cannot be distinguished and expressed, and it is particularly difficult to extract the contour of the loop when inputting in division, and therefore it is impossible to fill when converting to raster data.

The run length vector method (3) is a method of compressing raster data, but a vector expressed by coordinates of start and end points of run length from run length data expressed by white or black run length. Since it is converted into data, the amount of data becomes enormous, and the amount of data is too large to be used as CAD data.

The present invention has been made to solve the above-mentioned conventional problems, and provides a CAD device for a pattern image that can generate raster data that can be freely scaled up and down with a smaller data amount than a run-length vector. This is an issue.

[0020]

SUMMARY OF THE INVENTION The present invention provides a pattern image CAD apparatus provided with a Lux / Bex conversion unit for converting image-input raster data into vector data.
The raster / vector conversion means is provided with means for selecting an object to be converted into vector data from white or black raster data.
The function to convert the selected target raster data into a run length vector described by both the start point and end point coordinates of each scan line, the first run length vector, and the adjacent second run length vector are overlapped. And the function to generate a first trapezoidal shape with the upper and lower bases of both,
If the second run length vector and the third run length vector adjacent to it overlap, and the start point and end point are within the allowable range with respect to the extension lines on both sides of the trapezoid, the first run length vector and the first run length vector A function to generate a new second trapezoidal shape with the upper and lower bottoms of both run length vectors of 3 respectively,
Even after the fourth run length vector, the same operation is repeated to successively generate new trapezoidal shapes on condition that the extension lines on both sides of the first trapezoid are within the allowable range, If the condition is not satisfied, a function of parallelly translating the upper and lower bases of the final trapezoid generated in the immediately preceding operation outward by a dimension corresponding to ½ pixel of the raster data to form CAD data, In the case where a single run length vector remains, it has a function of translating it in the orthogonal direction outward by an amount corresponding to 1/2 pixel of the raster data to form a rectangle and using it as CAD data. By solving the above, the above problems are solved.

The present invention also provides a CAD for the above pattern image.
In the apparatus, the finally generated CAD data having a trapezoidal shape is filled and displayed, and a function of generating raster data at an arbitrary scale is provided.

[0022]

In the present invention, the white or black raster data input as the target image is designated, and the raster data of the designated image is converted into a run length vector for each scan line to obtain the X-axis or A trapezoid is formed between the first run-length vector and the last run-length vector within the allowable range between adjacent run-length vectors that are parallel to the Y-axis, and CAD data is created using the trapezoidal coordinates. Since it is expressed, it is possible to generate raster data that can be freely scaled up and down with a small amount of data.

As a result, it has become possible to easily perform the difference calculation between figures, which is difficult with the vector expression by the outline extraction method, using the raster data. Therefore, it becomes possible to easily measure the image of the precise pattern shape of the etching product.

Further, according to the present invention, when the finally generated CAD data of the trapezoidal shape is filled and displayed and the raster data is generated at an arbitrary scale, the displayed pattern image is visually recognized. Can be easily recognized.

[0025]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of a CAD system including a CAD device according to an embodiment of the present invention.

The CAD system is a sample (actual product)
A sample mounting device 30 for mounting a sample, an optical microscope 32 for enlarging a sample set in the mounting device 30, a CCD camera 34 for receiving an image observed by the microscope 32 and converting it into a color video signal, and the CCD An image processing device 36 for processing a color video signal from the camera 34,
A TV monitor 38 capable of color-displaying image data processed by the image processing device 36, and a raster / vector conversion function for generating CAD data from image data input from the image processing device 36 in addition to a normal drawing function. Or 2
An engineering workstation (EWS) 40, which constitutes the CAD apparatus of this embodiment, has the functions of superposing the CAD data described above, moving (shifting) their positions relative to each other, and measuring dimensions.

Further, in the above CAD system, the sample mounting apparatus 30 includes a manual rotary stage 42 having a sample mounting portion (not shown), an XY stage 44 for moving the sample in the plane direction, and a sample in the vertical direction. The XY stage 44 and the Z stage 46 are configured to move, and the XY stage 44 and the Z stage 46 operate by receiving a command from the workstation 40 via the interface RS232C.
The Y stage controller 48 and the auto focus controller 50 are driven and controlled respectively. Further, a laser position detector is attached to the XY stage 44, and the position measurement in the XY directions is performed by a laser scale counter 52 which is also operated by a command from the workstation 40, and the measured value is fed back to the workstation 40. Then, the position measurement value of the XY stage 44 is corrected by the XY stage controller 48.

Further, the autofocus controller 5 described above
An image signal used for autofocus is directly input to CCD 0 from the CCD camera 34, and an autofocus monitor (not shown) separately provided with an image captured through the microscope 32. It is possible to directly view the image on the TV monitor 38 from the autofocus controller 50 by using the monochrome (B /
The video signal of W) is input.

FIG. 2 shows the mounting device 30 and the optical microscope 3.
2 is a perspective view showing the appearance of the CCD camera 34 and the CCD camera 34,
The XY stage 44 shown in FIG.
An X drive motor 54 which is composed of an A stage and a Y stage 44B and is connected to the stage controller 48.
The A and Y drive motors 54B are movable in the X and Y directions, and the rotary stage 42 for mounting the sample is mounted on the Y stage 44B so that it can be manually rotated.

In addition, the X stage 44A and the Y stage 44
Scale patterns 56A and 56B each made up of a fine diffraction grating are attached to the side surface of B, and laser light is used to scale the positions of both stages 44A and 44B moved by the X drive motor 54A and Y drive motor 54B. , 56B are provided with an X position detector 58A and a Y position detector 58B for irradiating and detecting the laser scale counter 5B.
Connected to 2.

Further, the Z stage 46 is arranged below the X stage 44A, and the Z stage 46 can be moved back and forth in the vertical direction by a Z drive motor 54C. It is connected to the auto focus controller 50, and the distance between the objective lens 32A of the optical microscope 32 and the sample is increased or decreased based on a control signal from the controller 50 to perform auto focus on the microscope 32. There is.

Below the Z stage 46, a transmissive light source unit 60 which also serves as a support is arranged.
0 has a built-in transmissive light source (not shown) that projects light from below onto the microscope 32, and a transmissive light source switch 60A and a light amount adjusting volume 60B are attached to its side wall.

Further, an epi-illumination light source unit 62 is attached to the microscope 32, an epi-illumination light source (not shown) is built in the unit 62, and an epi-illumination light source switch 62A and a light quantity adjusting volume 62B are provided on a side wall of the unit 62. And are attached.

Therefore, when capturing the microscopic image of the sample with the CCD camera 34, it is possible to use at least one of the light source of transmission and epi-illumination.

Next, referring to FIG. 3, the image processing device 36 will be described.
The features of the configuration and the processing function will be described. The processing device 36 has an image input / processing / binarization processing function, for example, S manufactured by Seiko Denshi Kogyo Co., Ltd.
V-2110 (trade name) can be used.

The image processing device 36 is the CCD camera 3
Each of the R (red), G (green), and B (blue) signals input from 4 can be stored for each screen, and 3 for R image, G image, and B image each surrounded by a rectangle are stored. Monochrome memory for storing one frame memory and Y (luminance) signal
One frame memory for W image and H obtained by processing the R, G, B signals by a 3 × 3 matrix operation unit
(Hue), S (saturation), and V (luminance) image data are stored in the three frame memories for each image of H, S, and V, and the R signal and B signal are processed by the image arithmetic operation unit. A total of eight frame memories of the RB difference image frame memory that stores the difference image data of both obtained as described above are provided.

The reason why such different color signals are used is as shown in the table of FIG. 4 for the actual sample (actual pattern).
Depending on the material used and the type of image required, the type of light source suitable for use and the optimum color signal may differ.

That is, there are two types of original patterns, one for the front side and the other for the back side of the lead frame, both of which are glass dry plates (the pattern is formed of an opaque film on the glass plate), so that a black and white transmission image is obtained. The frame memory of the B / W image is the optimum plane because it can be obtained with good contrast.

Since the plate-making pattern is a resist pattern formed on the front and back surfaces of the lead frame,
It depends on the type of metal material and the type of resist, and it is necessary to use a reflected light source to receive the reflected image.

When casein is used as the resist, since the resist has a reddish color at the stage of heat-curing after development, the 42-alloy material of silver-white is used as an optimum plane for the B image frame. Although a memory can be used, the copper (Cu) material itself has a reddish color, and the difference is not clear in the B image. Therefore, the frame memory of the V image is the optimum plane.

When a blue-based dry film is used as the resist, the R image is optimal for 42 alloy, but the frame memory for the RB difference image is optimal for copper material.

In the case of the product pattern after the etching is completed and the resist film is removed, the penetrating transmission image and the reflection images on both the front and back surfaces can be received, and the transmission image is black and white B as described above. / W when the image is a reflection image, H
(Hue) The image becomes the optimum plane.

Further, among the product patterns, when the fixing tape 22 (denoted as TP in the table) made of polyimide resin is attached to the inner lead as shown in FIG. 27, the tape is red. Since the transparency is high in the system, in order to obtain a completely transparent image where the image is not input on the tape, B
The / W image becomes the optimum plane. However, it is necessary to appropriately set a threshold value for binarization described later.

On the contrary, in order to capture a transmission image including a tape, blue which is opaque to the tape is suitable, so that the B image becomes an optimum plane.

Furthermore, when only the tape portion is desired to be photographed, the H image becomes the optimum plane for the reflection image using the epi-illumination light source.

As described above, when the optimum plane to be used is selected according to the object to be captured as an image, the corresponding image signal is input to the binarization processing unit from the eight frame memories. Binarization processing is performed on the image data input by the binarization processing unit. The threshold value set at that time can be arbitrarily set from the gradation values of 0 to 255, for example.

Morphology processing for removing black spots or white spots on the image caused by the fine roughness of the surface of the actual pattern, etc. on the image data binarized by the binarization processing section. I do. However, in the case of a transmission image, such a spot does not occur, so that it is not necessary to perform it.

Whether the spot to be removed is white or black is set, a predetermined number of morphology is set, and the image is expanded / contracted for that number of times to remove the spot.

Next, the binary image obtained by performing the above processing is the workstation 4 which functions as a CAD device.
0, where the binary image is processed by the raster / vector conversion unit and converted into CAD data. As the workstation 40, normal CAD software and actual matching CAD software (for example, CAD manufactured by Computer Vision Co., Ltd.)
It is possible to use, for example, Sparc Station 10 (trade name) of Sun Microsystems, which is activated by software Medusa (trade name).

The raster / vector conversion section has a method of converting to general outline CAD data and a method of converting image data of a white or black area into CAD data of a trapezoidal area, which will be described in detail later. There is. When the signal conversion processing is performed by this raster / vector conversion unit, the RV vertex thinning coefficient for determining the accuracy of linear approximation is set. The smaller this coefficient is, the less jagged the line is in the case of the outline, and the finer the trapezoid can be extracted in the case of the trapezoid area.

When selecting either of the above two conversion methods and the trapezoidal area conversion method,
It is necessary to select either white or black and determine the target area for trapezoidal area processing.

FIG. 5 conceptually shows the procedure of processing by the trapezoidal area CAD data conversion method.

FIG. 5A shows a pattern (hatched portion) of the raster data input by the camera, and FIG. 5B shows a state in which the above pattern is converted into run length vector data (arrow) for each scan. FIG. 3C shows a state in which the trapezoidal area CAD data conversion according to the present invention is applied to the above run length vector data and a trapezoidal approximation is performed. The diagonal line for each trapezoid shown in FIG. 5C indicates not only the outline of the trapezoid but also graphic information as an area including the inside.

FIG. 5D shows the result of applying the conventional outline extraction method shown for reference to the pattern shown in FIG.

Workstation 40 in this embodiment
The trapezoidal approximation processing function of the raster / vector conversion unit in FIG. 6 will be described with reference to FIGS.

Image data (raster data) input now
A white or black pattern to be converted into vector data from is selected, and the raster data of the pattern is converted into a run length vector described by the coordinates of the start point and the end point of each scan line, and the result is shown in FIG. In FIG. 6, four run length vectors (1) to (4) are shown to simplify the description.

As shown by the broken line in FIG. 6, the first of (1)
When the run length vector and the second run length vector (2) adjacent to the run length vector overlap each other, as shown in FIG. 7, the first trapezoidal shape (1A, 1A, 1A, 1
B, 2A, 2B).

Then, as shown in FIG. 8, the second run length vector (2) and the third run length vector (3) adjacent to the second run length vector (3) are overlapped, and the start point and the end point are both sides of the trapezoid. When the value is within the range of the permissible value (indicated by the double-headed arrow length) with respect to the extension line (broken line) of the first and third run length vectors (1), (3), as shown in FIG. )
The new second trapezoidal shape (1
A, 1B, 3A, 3B).

Then, as shown in FIG. 10, the fourth run-length vector (4) is also operated in the same manner, provided that it exists within the allowable range based on the extension lines of both sides of the first trapezoid. A new trapezoidal shape is repeatedly generated in sequence, but when this condition is not satisfied (this corresponds to FIG. 10),
As shown by the upward and downward arrows in FIG. 11, each of the upper and lower bases of the final trapezoid is outside by a size corresponding to 1/2 pixel of the raster data, that is, the upper base is upward and the lower base is downward. To the CAD data as shown in FIG.

By thus setting the reference for newly generating a trapezoidal shape as an extension line of each side of the first trapezoidal shape, it becomes possible to approximate the pattern to a trapezoidal shape within a certain allowable range.

In addition, the single run length vector (4) isolated and left in the above process is divided by ½ of the raster data in both orthogonal directions, as also indicated by the arrows in FIG.
A rectangle (trapezoid) is parallel-moved to the outside by a dimension corresponding to a pixel, which is used as CAD data. Therefore, the CAD data of the trapezoidal approximation of the above pattern has four vertex coordinates (X1, X2, X3, X4, Y1, Y2) and (X3, X
It is sent as two trapezoidal data (4, X5, X6, Y2, Y3) (each two vertices have a common Y coordinate).

As described above, in the embodiment, the pattern can be formed by a set of trapezoids sharing a straight line parallel to the X axis.

Further, in this embodiment, the CAD data in which the pixels are included inside the trapezoidal shape as described above is used.
The created pattern of the final trapezoid can be filled in with an arbitrary color, displayed on the screen, and converted into raster data at an arbitrary scale.

Next, the operation of this embodiment will be described with reference to the flowchart of FIG.

First, before starting a specific operation, basic setting and adjustment of system functions are performed. In particular, when the lens of the microscope 32 or the camera 34 is replaced, it is necessary to perform orthogonal adjustment of the camera 34 and the XY stage 44. This is done by manually rotating the camera mount. In this orthogonal adjustment, as shown schematically in the monitor screen in FIG. 14, a minute mark (may be a minute dust) indicated by an X mark on the sample mounting portion is used as a reference point, and this range does not deviate from the monitor screen. If the reference point does not deviate from the reference line (horizontal line) on the monitor when it is horizontally moved in the left and right X directions, it can be performed by setting OK.

Further, in order to provide the image measuring function, it is necessary to measure the size per pixel and the screen feed pitch in advance. This is one screen size (512 × 4 in this embodiment).
In measuring the stage feed value corresponding to 80 pixels), specifically, FIG.
As shown in, the reference points indicated by X are moved to four points on the reference line at positions of ¼ and 3/4 in both the X direction and the Y direction on the screen, and the X direction at that time, The stage moving distance in the Y direction is determined by the laser scale counter 5
It is possible to measure with high accuracy by using the count value of 2. In this case, the size per pixel is Xs /
256, Ys / 240, and the screen feed pitches in the X and Y directions are calculated as 2Xa and 2Ys.
For the measurement of the above dimensions and pitch, the feeding pitch by the XY stage drive motor (step motor), for example, 1 μm may be used without using the laser scale counter.

On the premise that the above preparatory operation has been completed, the sample is set in step S1. Specifically, as shown in FIG.
The sample (lead frame) is attached to a predetermined position of the sample, and the operator operates the rotary stage 42 while viewing the image of the sample taken from the camera 34 displayed on the monitor 38 to adjust the sample orthogonally. To do.

If the monitor screen displaying the horizontal edge of the sample input by the CCD camera 34 whose orthogonal adjustment with the XY stage 44 has already been completed is as shown in FIG. The rotary stage 42 is manually rotated to a position such that the edge does not deviate from the horizontal reference line even when the sample is largely moved in the X-axis direction, and orthogonal adjustment between the sample and the XY stage 44 is performed.

Then, in step S2, a light source to be used is selected and its light amount is adjusted. That is, the switch 60A or 6
A transmission light source or an epi-illumination light source is selected by turning on any of 2A. The desired light source is selected, and the light amount is adjusted by the light amount volume of 60B or 62B so that the brightness signal falls within the specified range by looking at the monitor of the autofocus device. In some cases, both light sources can be used at the same time.

In the next steps S3 to S6, the menu is displayed on the menu screen (each can be displayed on the same screen as the window) displayed on the monitor screen as schematically shown in FIGS. 17 to 19, for example. It is executed by selecting.

First, in step S3, the center of an island (die pad) captured as an image is designated.

In this embodiment, the island 10 is shown in an enlarged manner in FIG. 20, and a point P _{T at} the upper end and a point P at the lower end of the island 10 are shown as shown on the monitor screen on the right side.
_By inputting _{B so} as to match the center of the camera input screen in the Y-axis direction, the respective Y coordinate values Y _T , Y _B
Is calculated, and by inputting the point P _L on the left end and the point P _R on the right end at the center of the screen in the X direction, the X coordinate values X _L and X _R are calculated. There is. Therefore, from the coordinate values of the edge positions corresponding to the boundaries of these four black and white areas (the black parts are indicated by diagonal lines),
The coordinates (X, Y) of the center of the island, which is the alignment origin, are calculated by the following equation.

X = (X _R + X _L ) / 2, Y = (Y _T + Y _B ) / 2

If the CAD system has an edge detection function, the same as the above, even if the edge portions of the black and white boundaries on the left, right, top and bottom are not aligned with the centers of the X and Y coordinates on the screen, You can specify the center. By specifying the center of the island of the input image in this way,
It is possible to accurately perform the overlay display in which the center coincides with the center of the island of the design pattern of CAD data.

At the next step S4, a sampled area is designated.

In the case of photographing the lead frame for one chip, for example, shown in FIG. 27, right and left of the lead frame input from the camera 34 and displayed on the monitor while moving the XY stage 44. The shooting area can be designated by sequentially moving the upper and lower end portions into the screen and designating respective points (four endpoints defining the rectangular area) with the cursor, for example. At that time, a plurality of areas including a partial area (for example, an inner lead bonding area) can be designated.

When CAD data of the product design dimension is input, the dimension is read from the CAD data and the coordinates of the four endpoints can be automatically set, for example, by using the dimension value, and the photographing area is set. Can also be specified. In this case, the area can be designated in a short time, and the CAD pattern design pattern to be executed later can be easily aligned with the actual pattern input as the image.

Information about the center of the island and the shooting area on the image designated in steps S3 and S4 is stored in the setting file XY.

Then, in step S5, conditions for autofocus (AF) are set. Here, the mode is selected and the limit value is set. In this mode, one reference point (position) is determined in the Z-axis direction applied to a flat sample, and the 2WAY method in which the Z stage 46 is finely moved vertically to focus on the reference point, and a surface with large unevenness is used. From the state in which the Z stage 46 is lowered to a predetermined lower position beyond the focusing point to be applied, the stage 46 is gradually raised to bring the sample closer to the lens to focus the image 1WA
There are a Y system and selection of a lens to be used from objective lenses (five types in this embodiment).

The limit is an approach limit distance set to prevent the lens from colliding with the sample during autofocus. In addition, here, 1 WAY is used for the product sample with severe unevenness as the mode, and 2 W when it is not.
Select AY, use a 20x objective lens as a lens to capture with a resolution of 1 μm / 1 pixel, and use a focus parameter setting file for that. 2 mm from the origin as the limit value, the maximum value of the focus working distance is 1 limit. Set a normal default value such as "/ 2". The conditions set in this step are stored in the setting file AF.

When the 2WAY method is selected as the AF mode, the edge portion of the sample is brought into the screen and autofocus is activated or the Z stage 46 is manually moved to determine the focus origin (reference point Z coordinate value) is also set. The autofocus is executed by moving the Z stage 46 up and down slightly with reference to this position as described above.

In the next step S6, conditions for image processing are set. The contents of the selection are the selection of the input plane to be used from the eight types of image frame memories shown in FIG. 3 that use different colors, the setting of the binarization threshold value when creating a binary image, white or black. Condition and raster vector (RV) to remove unnecessary points of image from image data
It is a conversion condition. The conditions set in this step are stored in the setting file SV.

After the conditions have been stored in the respective setting files by the above operation, the raster / vector conversion method (outline (contour) mode or trapezoidal area mode) is selected.

The batch processing of step S7 for selecting the contour mode and step S8 for selecting the trapezoidal mode is automatically executed in the workstation 40, and the raster data of the entire image-captured photographing area is converted into vector data. CAD data created by the output file LFX
The data is converted into various CAD system formats in step S9, and various collation processes are performed. This collation (overlapping) process is performed by selecting a menu on the screen of the workstation.

Next, a CAD device (workstation 4
The batch processing of steps S7 and S8 executed inside 0) will be described with reference to the flowchart of FIG.

First, in step S11, the data etc. stored by the above operation are read from the above-mentioned setting files XY, AF, SV, etc., and the cell division of the photographing area is calculated.

The contents of each setting file read here are exemplified below.

File XY-Position of four sides of island (center of island) -Number of input designated areas-Coordinate value of each rectangular area File SV-Input color plane number-Binarization threshold value-Morphology direction and number of times (+: white to black) ,-: Black to white) -RV thinning coefficient file AF (normally fixed) -cell unit execution or fixed focus selection-AF mode (2 modes for each of 5 lens types (1 or 2W)
AY))-Soft limit values (upper and lower limits of Z stage) Lens file (fixed) -Field size by lens-Actual size of 512 x 480 pixels (including non-rectangular distortion)

The morphology of the contents of the file SV
In the direction +, the black point is removed from the white image, and-is the white point removed from the black image. Also, in the contents of the file AF, "execute in cell unit" is applied to a sample having minute unevenness such as a lead frame by executing autofocus every time an image is captured, and fixed focus is applied to a glass original plate. Applies to those with high flatness. The soft limit value is the same as the limit value set in step 5, and is the upper limit value of the Z stage movement set by the operator to prevent the sample from colliding with the lens, or the lower limit value to prevent it from lowering more than necessary. is there.

In the lens file, the field size of each of the above five types of objective lenses, the actual size corresponding to all the pixels of the CCD camera 34 used in this embodiment (the distortion due to the lens is corrected). 4 dimensions) and are stored.

Similarly, the calculation of the cell division, which is executed in step S11, is performed by, for example, four lead frames for one chip.
0 mm x 40 mm, 512 x CCD camera 34
If it is assumed that the visual field size of 480 pixels is 496 μm × 464 μm, it means that the lead frame is divided into 80 × 86 units (cells) as shown in FIG. It is an input unit when sequentially moving in the Y direction to input an image of the entire lead frame, and also a feed pitch (offset amount) when feeding to the next cell. However, in practice, the offset amount is usually set to about 90% of the cell size in order to make the boundary of each captured image clear.

When the calculation of the number of cell divisions in step S11 is completed, in step S12, the XY stage controller 48 moves the XY stage 44 in response to a command from the workstation to bring the field of view of the optical microscope 32 to the first photographing position. Set. At that time, the actually measured position (corresponding to the X and Y coordinate values) actually measured by the laser scale counter 52 is fed back to the workstation 40.

Next, in step S13, autofocus is performed by the autofocus controller 50 at the camera setting position based on a command from the workstation 40, and the controller 50 reads the Z coordinate value of the in-focus position. Workstation 4
Send to 0.

Then, in step S14, the workstation 40 issues an instruction to the image processing device 36 (SV), the image processing device 36 is activated, and an image frame is input from the CCD camera 34 to the processing device 36. , A binarized image is generated by performing binarization on the input raster image and morphological processing for removing unnecessary points from the image.

Next, in step S15, raster / vector conversion is executed. First, the binary image data generated in step S14 is read from the image processing device 36 into the workstation 40, converted into vector data by the RV conversion section, and transmitted to the image processing device 36 again. , While displaying it on the TV monitor 38, converting the vector data into CAD data and storing it in a file, at the same time calculating the offset amount and scaling once sent from the XY stage and moving to the next cell, Returning to step S12, the process up to step S15 is executed for the cell, and this process is repeated.

As shown by the broken line in FIG. 21, when the above step S14 is completed, a command for moving to the next cell is issued to the XY stage controller 48 and the autofocus controller 50 to prepare for the preparation. By doing so, the RV conversion process, the XY stage movement, and the autofocus process can be performed in parallel.

The above image capturing is performed, for example, with reference to FIG.
By executing the above for all areas, the entire lead frame for one chip can be displayed on the monitor 38 as vector data, and the CAD data of the lead frame can be generated.

The CAD data created in this way is converted into the CAD data as shown in step S9 of the flow chart of FIG.
When the data is converted into various CAD system formats, the product design dimension data originally input as CAD data on the screen of the monitor 38 and the machining dimension data in which the correction allowance is added are displayed as a graphic pattern and are also displayed in these data. In the present embodiment, C from the image input data
The product pattern of the lead frame converted into the AD data can be displayed in an overlapping manner.

FIG. 23 shows the data obtained by actually performing the trapezoidal approximation process described above in the step 15 of raster / vector conversion on the screen. In this figure, a large rectangle represents a cell of a unit captured by the camera 34, and a thin pattern A having a rounded tip.
Is the inner lead of the product, the result of trapezoidal approximation of the inner side is shown by the line in the X direction (horizontal direction), the pattern B on the outer side is the resist pattern (plate making pattern), and the outer side of this B The regions are likewise trapezoidal. In this figure, the lines corresponding to the large vertical and horizontal rectangles represent cells in image capturing units. FIG. 24 shows a state in which the patterns of FIG. 23 (set of final trapezoids) obtained by the trapezoidal approximation are filled with different colors.

FIG. 25 shows an example of collation on the screen of one lead frame. The line drawing A on the outer side is a processing dimension pattern, and the inner line B is a product design pattern before being corrected to A. C is the actual product pattern (inner lead) obtained by actual etching. This product pattern is a through image using a frame memory for B / W images. From this FIG. 25, it can be seen that the correction margin setting is almost appropriate.

FIG. 26 also shows a screen display near the tip of the inner lead. The outer line drawing D is a front reflection image and the inner E is a back reflection image. It is an example of collation of an image displayed by being painted out.

From this figure, it is possible to clearly understand the difference in the degree of etching on the front side and the back side, which cannot be recognized from the transmission image. The lead frame is usually etched by spraying an etching solution from both upper and lower sides with the surface on which the chip is mounted facing down. The reason for making the surface side down is that it is necessary to secure a sufficient width for wire bonding on the surface of the inner lead tip portion, but etching is easier to proceed on the upper surface. From this FIG. 26,
It can be clearly understood that there are etching differences between the front and back.

The basic characteristics and performance of the CAD system of this embodiment described in detail above will be briefly summarized as follows.
It looks like this:

There are two methods for converting raster data into vector data, that is, a normal outrun approximation and a novel trapezoidal area approximation.

Since full-color image input processing is possible, it becomes possible to collate the resist pattern and the pattern for the front and back of the product with each other or with the design pattern or the like, so that it is possible to perform dimension comparison and measurement. It is possible to objectively evaluate etching. Since it has its own automatic positioning function, the center of the island can be calculated automatically. It functions as an interface to various commercially available CAD systems such as Medusa (trade name) manufactured by Computer Vision.

As the measurement performance, resolution: 1 μm,
Measurement accuracy: 1μm guaranteed, visual field size: 496 × 464μm
(512 × 480 pixels), measurement size: 200 mm square (expandable), measurement target: product (transmission image, reflection image by front and back),
Plate making (resist reflection image) and glass original plate (transmission image) can be mentioned.

Therefore, the CAD system of the present embodiment is used for automatic calculation of etching compensation, dimensional control for each manufacturing process, quality control such as product dimensional inspection, etching simulator, etc., for providing data to research and development. Available.

As described above in detail, according to the present embodiment, the raster data input as an image is converted into vector data and the CA is converted.
By adopting a trapezoidal area approximation method when creating D data, it is possible to generate raster data that can be scaled up and down with a smaller amount of data compared to the run length vector.
Further, since the pattern created by the trapezoidal area approximation can be enlarged or reduced or painted, the two can be clearly recognized even when different patterns are displayed in an overlapping manner, and thus the difference between the two can be accurately measured.

Further, since the etching compensation amount can be quantified by the objective data, it is possible to reduce the replacement of the compensation amount due to trial and error, and as a result, the delivery time can be shortened.

Further, since the tolerance judgment can be performed collectively and automatically, it is possible to greatly reduce the labor of the authentication, and the oversight can be eliminated, so that the accuracy can be improved. .

Further, since the CAD pattern, the original pattern, the plate-making pattern, and the product pattern can be compared with each other, the accuracy of the production line can be grasped and the quality can be maintained for each process.

The present invention has been specifically described above, but the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the invention.

For example, the sample is not limited to the lead frame, and may be a shadow mask for a color television, for example.

[0115]

According to the first aspect of the present invention, it is possible to generate raster data that can be scaled up and down with a smaller amount of data than the run length vector.

According to the second aspect of the invention, even when different patterns are displayed in an overlapping manner, both can be clearly recognized, so that the difference between the two can be accurately measured.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a CAD system according to an embodiment of the present invention.

FIG. 2 is a CAD system sample mounting device, a microscope,
Oblique view showing a CCD camera

FIG. 3 is a block diagram showing a frame memory and a processing function of the image processing device of the CAD system.

FIG. 4 is a diagram showing an optimum input plane for each sample.

FIG. 5 is an explanatory diagram conceptually showing the CAD data creation function of the raster / vector conversion unit.

FIG. 6 is an explanatory diagram showing a procedure of a trapezoidal approximation method.

FIG. 7 is another explanatory diagram showing a specific procedure of the trapezoidal approximation method.

FIG. 8 is still another explanatory diagram showing a specific procedure of the trapezoidal approximation method.

FIG. 9 is still another explanatory diagram showing the specific procedure of the trapezoidal approximation method.

FIG. 10 is still another explanatory diagram showing the specific procedure of the trapezoidal approximation method.

FIG. 11 is still another explanatory diagram showing a specific procedure of the trapezoidal approximation method.

FIG. 12 is still another explanatory diagram showing a specific procedure of the trapezoidal approximation method.

FIG. 13 is a flowchart showing the operation of the embodiment.

FIG. 14 is an explanatory diagram showing a monitor screen at the time of orthogonal adjustment of the camera and the XY stage.

FIG. 15 is an explanatory diagram showing a monitor screen when a dimension per pixel and a screen feed pitch are calculated.

FIG. 16 is an explanatory diagram showing a monitor screen at the time of orthogonal adjustment of the sample and the XY stage.

FIG. 17 is an explanatory diagram illustrating a CAD system menu screen.

FIG. 18 is another explanatory diagram illustrating the menu screen of the CAD system.

FIG. 19 is still another explanatory diagram illustrating the menu screen of the CAD system.

FIG. 20 is an explanatory diagram showing an example of a method of designating the center of an island.

FIG. 21 is a flowchart showing the procedure of batch processing executed inside the CAD device.

FIG. 22 is an explanatory diagram showing a cell division calculation method.

FIG. 23 is an explanatory diagram showing an example of an actual superimposed display pattern that is trapezoidally approximated.

FIG. 24 is an explanatory diagram showing a state in which the display pattern is filled.

FIG. 25 is an explanatory diagram showing an example of a screen in which a plurality of patterns are displayed in an overlapping manner.

FIG. 26 is an explanatory diagram showing another example of a screen in which a plurality of patterns are displayed in an overlapping manner.

FIG. 27 is a plan view showing an example of a lead frame.

FIG. 28 is an explanatory view conceptually showing the manufacturing process of the lead frame.

[Explanation of symbols]

 30 ... Sample mounting device 32 ... Optical microscope 34 ... CCD camera 36 ... Image processing device 38 ... TV monitor 40 ... Workstation (EWS) 42 ... Rotation stage 44 ... XY stage 44A ... X stage 44B ... Y stage 46 ... Z stage 48 ... XY stage controller 50 ... Auto focus controller 52 ... Laser scale counter 54A ... X drive motor 54B ... Y drive motor 54C ... Z drive motor 56A, 56B ... Scale pattern 58A ... X position detector 58B ... Y position detector 60 ... Transmission Light source unit 60A ... Transmissive light source switch 60B ... Light intensity adjustment volume 62 ... Epi-illumination light source unit 62A ... Epi-illumination light switch 62B ... Light intensity adjustment volume

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Teruaki Iinuma 1-1-1, Ichigaya-Kagacho, Shinjuku-ku, Tokyo Dai Nippon Printing Co., Ltd. (72) Inventor Masataka Yamaji 1-chome, Ichigaya-Kagacho, Shinjuku-ku, Tokyo No. 1 Dai Nippon Printing Co., Ltd.

Claims (2)

[Claims] 1. A pattern image CAD apparatus having a Lux / Bex conversion means for converting image-inputted raster data into vector data, comprising means for selecting an object to be converted into vector data from white or black raster data. A function of the raster vector conversion means for converting the selected target raster data into a run length vector described by both the start point and end point coordinates of each scan line, the first run length vector, and the adjacent run length vector When the second run length vector overlaps, the function of generating the first trapezoidal shape having the upper base and the lower base as the two, and the second run length vector and the adjacent third run length vector overlap each other. If the start point and the end point are within the allowable range for the extension lines on both sides of the trapezoid, the first and third run length vectors are The function to generate a new second trapezoidal shape with the upper and lower bases, respectively, and that it exists within the allowable range even after the fourth run length vector based on the extension lines of both sides of the first trapezoid. A function to generate a new trapezoidal shape sequentially by repeating the same operation under the conditions, and if the above conditions are not satisfied, the upper and lower bases of the final trapezoid generated by the previous operation are set as 1 /
A function to make CAD data by parallel translation to the outside by a dimension equivalent to 2 pixels. A CAD device for a pattern image, which has a function of moving in parallel to form a rectangle and using it as CAD data. 2. A pattern image according to claim 1, which has a function of displaying the finally generated CAD data of the trapezoidal shape in a painted manner and generating raster data at an arbitrary scale. CAD
apparatus.