JP5417804B2 - Bubble rate calculation method and bubble rate calculation device - Google Patents

Description

  The present invention relates to a bubble rate calculation method and a bubble rate calculation device for calculating a bubble rate in a bonding region of a semiconductor device formed by bonding a heat dissipation substrate and a semiconductor element.

For example, a semiconductor element used as an electronic component is bonded to a heat dissipation substrate made of copper or aluminum having high thermal conductivity in order to secure a heat dissipation path. At this time, for example, one surface of the rectangular semiconductor element and one surface of the heat dissipation substrate are bonded via a bonding material such as solder or an adhesive, so that bubbles may be mixed in the bonding material.
When a large amount of air bubbles (voids) are interposed between the semiconductor element and the heat dissipation substrate, the thermal conductivity from the semiconductor element to the heat dissipation substrate is reduced and heat dissipation is hindered. As a result, the semiconductor element may be exposed to an excessively high temperature due to self-heating, and the semiconductor element may be deteriorated or destroyed.

  Therefore, when manufacturing a semiconductor device including a semiconductor element and a heat dissipation substrate, for example, in the final stage of the manufacturing process, the ratio of the area of the bubble to the area of the junction region between the semiconductor element and the heat dissipation substrate (that is, the bubble ratio) is obtained. It is necessary to eliminate semiconductor devices whose calculated bubble ratio exceeds the allowable value.

  In order to obtain the bubble rate, the conventional bubble rate calculation device irradiates X-rays so as to be orthogonal to the heat dissipation substrate, images the semiconductor device, and detects the presence or absence of bubbles based on the density of the obtained X-ray image. It is configured as follows. In a semiconductor device, a region where bubbles are present has a higher X-ray transmittance than a region where bubbles are not present (that is, a region where bonding material is present). Therefore, a portion of the bonding region that appears light in the X-ray image is a bubble. Can be considered. Therefore, it is possible to easily obtain the bubble rate by binarizing a multi-value X-ray image with one appropriate threshold value and dividing the number of pixels corresponding to the bubble by the number of pixels in the entire joining region. it can.

  However, a heat radiating board whose material, thickness, etc. are not constant, for example, a heat radiating board in which a hole in which the inner surface is plated with a metal such as a through hole and a via hole is formed in a laminate formed by laminating a conductive layer and an insulating layer. Since the X-ray transmittance of each part of the heat dissipation board itself changes according to the material and thickness of each part, the threshold value when binarizing a multi-value X-ray image is set. It is difficult to set.

In the conventional board inspection method, there are areas where the analysis of the X-ray transmission image is difficult and the inspection results are adversely affected, such as areas where electronic parts are mounted on both sides of the printed circuit board and areas where through holes are opened. Masking is performed as an inspection exclusion region (see Patent Document 1).
Further, in the conventional bonding inspection method, the X-ray fluoroscopic image is corrected to remove the influence of the thickness based on the calibration information indicating the relationship between the thickness of the inspection object and the density of the X-ray fluoroscopic image (Patent Document). 2).
JP 2004-226127 A JP 2000-352559 A

However, in the substrate inspection method described in Patent Document 1, no consideration is given to the inspection exclusion region even if the presence or absence of bubbles is accurately detected in regions other than the inspection exclusion region. On the other hand, in the bonding inspection method described in Patent Document 2, only the thickness of the heat dissipation substrate is considered, and X is based on a logarithmic curve that approximately indicates the relationship between the thickness of the inspection object and the density of the X-ray fluoroscopic image. Since the fluoroscopic image is corrected, there is a possibility that accurate correction is not performed.
As a result, the conventional method has a problem that an inaccurate bubble rate is calculated by dividing the number of pixels of the detected bubble by the number of pixels of the entire bonding region.

  The present invention has been made in view of such circumstances, and the main object of the present invention is to accurately detect the presence or absence of bubbles present in the junction region between the semiconductor element and the heat dissipation substrate, and to accurately detect the bubble rate of the entire junction region. Is to provide a bubble rate calculation method and a bubble rate calculation device.

  According to a first aspect of the present invention, there is provided a method for calculating a bubble ratio, in which an X-ray is irradiated with a bonding region of a semiconductor device formed by bonding a semiconductor element and one surface of a heat dissipation substrate formed with a hole plated with an inner surface facing the one surface. In the bubble rate calculation method for calculating a ratio of bubbles existing in the bonding region to the bonding region using a multi-valued image obtained by irradiating and photographing, the plated portion of the hole included in the multi-valued image The masked multi-valued image is binarized to generate a binary image, and based on the generated binary image, the presence or absence of a region bubble located in the joining region excluding the plated portion is detected and detected. Detecting the presence or absence of contact bubbles in contact with the plated portion in the area bubbles, and based on the detected shape of the contact bubbles, obtain a virtual line indicating the characteristics of the shape, and are surrounded by the calculated virtual line, And on the plated part The presence / absence of a region to be identified as a bubble is detected, and the ratio of the number of pixels in the region to be identified as a detected bubble and the number of pixels in the detected region bubble to the number of pixels in the junction region is determined. And calculating the bubble ratio of the entire joining region.

According to a second aspect of the present invention, there is provided an air bubble ratio calculation device, wherein X-rays are directed toward a surface of a semiconductor device formed by bonding a semiconductor element and one surface of a heat radiating substrate having a hole plated with an inner surface. In the bubble rate calculation device that calculates the ratio of bubbles existing in the bonding area to the bonding area using a multi-valued image obtained by irradiating and photographing, the plated portion of the hole included in the multi-valued image Based on the masking means for masking, the image binarizing means for binarizing the multi-valued image masked by the masking means, and the binary image generated by the image binarizing means, An area bubble detecting means for detecting presence / absence of an area bubble located in the joining area excluding the plated portion, and detecting presence / absence of a contact bubble in contact with the plated portion among the area bubbles detected by the area bubble detecting means. Based on the shape of the contact bubble detected by the contact bubble detection means, a virtual line calculation means for obtaining a virtual line indicating the feature of the shape, and a virtual line obtained by the virtual line calculation means A partial bubble detection means for detecting the presence or absence of a region to be identified as a bubble, which is surrounded and located in the plated portion, and the number of pixels in the region to be recognized as a bubble detected by the partial bubble detection unit and And a total bubble ratio calculating means for calculating a ratio of the total number of pixels of the area bubbles detected by the area bubble detection means to the number of pixels of the bonded area as a bubble ratio of the entire bonded area.

In the bubble rate calculation device according to the third aspect of the invention, when the contact bubble detected by the contact bubble detection means is in contact with one of the outer side and the inner side of the hole at the same position in the circumferential direction of the hole of the plated portion, The imaginary line calculation means is configured to calculate a normal line of the plated portion passing through the intersection of the bubble peripheral portion of the contact bubble and the plated portion, and a tangent of the bubble peripheral portion passing through the intersection, The partial bubble detection means has a minimum number of pixels that are surrounded by two straight lines where the intersection points are located at different positions from the normal line and the tangent line calculated by the virtual line calculation means, and are located in the plated portion. The presence or absence of a region is detected.

In the bubble rate calculation device according to the fourth invention, when the contact bubbles detected by the contact bubble detection means are in contact with both the outside and the inside of the hole at the same position in the circumferential direction of the hole of the plated portion, The virtual line calculation means includes an outer intersection of a bubble peripheral portion of the contact bubble that contacts the plated portion from the outside of the hole and the plated portion, a bubble peripheral portion of the contact bubble that contacts the plated portion from the inside of the hole, and the A straight line passing through the inner intersection with the plated portion is calculated, and the partial bubble detecting means intersects outside the area surrounded by the outer intersection and the inner intersection among the straight lines calculated by the virtual line calculating means. The presence or absence of a region surrounded by two straight lines and located in the plated portion is detected.

In the case of the bubble rate calculation method according to the first invention and the bubble rate calculation device according to the second invention, the heat dissipation substrate can be divided into the inside and outside of the hole plated on the inner surface and the plated portion. At this time, the presence / absence of bubbles located inside and outside the hole (that is, the joining region excluding the plated portion) can be easily and accurately detected by binarization. In addition, the presence or absence of bubbles located in the plated portion can be easily and accurately detected based on the shape of the bubbles in contact with the plated portion. That is, it is possible to accurately detect the presence / absence of bubbles present in the entire bonding region between the semiconductor element and the heat dissipation substrate, and to obtain an accurate bubble ratio in the entire bonding region.

In the case of the bubble rate calculating device according to the third aspect of the invention, the presence / absence of a bubble located in the plated portion is determined by using the tangent of the bubble contacting the plated portion only from the outside or the inside of the hole and the normal of the plated portion. It can be detected easily and accurately.

In the case of the bubble rate calculating device according to the fourth aspect of the invention, the presence / absence of a bubble located in the plating portion connects the bubble in contact with the plating portion from the outside of the hole and the bubble in contact with the bubble from the inside. It can be detected easily and accurately using a straight line.

  Hereinafter, the present invention will be described in detail with reference to the drawings illustrating embodiments thereof.

Embodiment 1.
FIG. 1 is a perspective view showing an external appearance of a bubble rate calculating apparatus according to Embodiment 1 of the present invention.
In the figure, reference numeral 3 denotes a bubble rate calculation device. The bubble rate calculation device 3 is controlled by the calculation device 1 using a personal computer, starts imaging, and the captured X-ray image is calculated by the calculation device 1. And an imaging device 2 for outputting to the camera.
FIG. 2 is a block diagram illustrating a configuration of a main part of the bubble rate calculation device 3. FIG. 3 is a vertical cross-sectional view showing a schematic configuration of a semiconductor device in which the bubble rate calculation device 3 should detect the presence or absence of bubbles.
In the figure, reference numeral 4 denotes a semiconductor device. The semiconductor device 4 is formed by bonding a semiconductor element 41 and a heat dissipation substrate 42.

First, the configuration of the semiconductor device 4 will be described.
The semiconductor element 41 has a rectangular plate shape, and is incorporated in an electric power steering system (not shown), for example, and has a function of controlling driving of the motor.
The heat dissipation substrate 42 dissipates heat generated by the operating semiconductor element 41. Therefore, the heat dissipation substrate 42 is a laminate having a metal substrate 424 functioning as a heat sink, a function of transferring heat from the semiconductor element 41 to the metal substrate 424, and a function of a circuit board on which the semiconductor element 41 is mounted. 423.

The stacked body 423 has a rectangular plate shape, and is formed by alternately stacking conductive layers 421 and insulating layers 422 by thermocompression bonding. The conductive layer 421 is a copper circuit pattern. The insulating layer 422 is made of a composite of glass fiber and epoxy resin (so-called prepreg) and has a high thermal conductivity of 1 W / m · K or more.
The stacked body 423 has a conductive layer 421 on one side (upper surface in the drawing) and an insulating layer 422 on the other surface (lower surface in the drawing), and a nickel plating layer and a gold plating layer (not shown) are provided on the upper surface of the stacked body 423. Thus, one surface (the lower surface in the figure) of the semiconductor element 41 is soldered. In the figure, reference numeral 43 denotes a solder layer that joins the semiconductor element 41 and one surface of the stacked body 423 (that is, one surface of the heat dissipation substrate 42). Bubbles may be mixed in the solder layer 43 during soldering. is there. The bubble rate calculation method of this embodiment detects the presence or absence of bubbles mixed in the solder layer 43 and calculates the bubble rate of the detected bubbles.

A rectangular metal substrate 424 made of aluminum is thermocompression bonded to the lower surface of the stacked body 423.
Further, via holes (holes) 44, 44,... Pass through the stacked body 423 in the stacking direction (vertical direction in the drawing) from the lowermost conductive layer 421 to the uppermost insulating layer 422 in the drawing. Is formed. Each via hole 44 is formed with a plated portion 45 and a resin portion 46 by filling the prepreg with copper plating on the inner surface. The plated portion 45 improves the thermal conductivity from the uppermost insulating layer 422 to the lowermost conductive layer 421. In addition, the plated portion 45 electrically connects the conductive layers 421, 421,. In addition, the plating part 45 which does not electrically connect the conductive layers 421, 421, ... may be provided.

  Since the via holes 44, 44,... Are formed in the stacked body 423, the material in the stacking direction and the thickness of the portion having each material are constant at each position in the layer direction (left and right direction in the figure) of the stacked body 423. is not. Specifically, regarding the thickness of the copper portion (illustrated by an arrow in the vertical direction; hereinafter referred to as copper thickness), the central portion 42a of the via hole 44 penetrates the stacked body 423 in the stacking direction and the resin portion 46 is formed. Therefore, the copper thickness is very thin. However, the inner peripheral portion 42b of the via hole 44 has a very thick copper thickness because the plated portion 45 is formed through the stacked body 423 in the stacking direction. In addition, the outer 42c of the via holes 44, 44,... Is formed by stacking conductive layers 421, 421,... With insulating layers 422, 422,. It is about the middle of the thickness according to each part 42b.

  As a result, when X-rays are irradiated in a direction orthogonal to one surface of the heat dissipation substrate 41, that is, in the stacking direction of the stacked body 423, the X-ray absorption rate in each of the central portion 42a, the inner peripheral portion 42b, and the outer 42c of the via hole 44. (Or X-ray transmittance) is different.

Next, the configuration of the bubble rate calculation device 3 will be described.
As shown in FIGS. 1 and 2, the computing device 1 includes a CPU 10, a ROM 11, a RAM 12, a display unit 13, an operation unit 14, an HDD (hard disk) 15, an external storage unit 16, and an I / F unit 17. Each part of the apparatus is appropriately connected via a bus, a signal line, and the like.
The I / F unit 17 is an interface for electrically connecting the arithmetic device 1 and the photographing device 2.
The external storage unit 16 uses, for example, a CD-ROM drive, and reads a computer program and / or data from a portable recording medium (for example, the CD-ROM 30 on which the computer program of the present embodiment is recorded).

The HDD 15 is an auxiliary storage unit of the bubble rate calculation device 3, and various computer programs and data are read from and written to the HDD 15. More specifically, the HDD 15 stores a computer program that realizes the bubble rate calculation method of the present embodiment in software using hardware elements of a computer. The HDD 15 stores a mask image, which will be described later, as data necessary for executing the bubble rate calculation method of the present embodiment.
The computer program and data read / written to / from the HDD 15 may be read from a recording medium using the external storage unit 16 or may be received via a LAN and generated by the arithmetic device 1. It may be what was done.

The CPU 10 is a control center of the bubble rate calculation device 3, uses the RAM 12 as a main storage unit as a work area, controls each unit according to the computer program and data stored in the ROM 11 and the HDD 15, and executes various processes. For example, the CPU 10 outputs a control signal for controlling the operation of the photographing apparatus 2 to the photographing apparatus 2 via the I / F unit 17.
The display unit 13 uses, for example, a liquid crystal display, and is controlled by the CPU 10 to display a message indicating the operating state of the bubble rate calculation device 3, messages indicating various instructions to the user, and the like. The operation unit 14 includes, for example, a keyboard and a mouse.

The imaging device 2 has an opening 2a on the side surface of the housing for taking in and out the semiconductor device 4 with respect to the inside of the device.
The imaging apparatus 2 includes an X-ray source 21, an X-ray controller 22, a camera 23, and an image processing unit 24. The operation of the X-ray source 21 is controlled by the X-ray controller 22 to which a control signal is input from the arithmetic device 1, and the heat radiation substrate 42 from the semiconductor element 41 side with respect to the semiconductor device 4 placed inside the imaging device 2. X-rays 21a are irradiated perpendicularly to the surface.

  The camera 23 and the image processing unit 24 also operate when a control signal is input from the arithmetic device 1. The camera 23 receives the X-ray 21 a that has passed through the semiconductor device 4 and outputs an analog signal that is a light reception result to the image processing unit 24. The image processing unit 24 performs image processing such as A / D conversion and distortion correction on the analog signal input from the camera 23, and outputs a multi-value X-ray image to the computing device 1 as a result of the image processing.

  The X-ray image input from the imaging device 2 to the arithmetic device 1 is an image (hereinafter, referred to as a region image) showing a state of a bonding region where the semiconductor element 41 of the semiconductor device 4 and one surface of the heat dissipation substrate 42 are bonded. When the bubbles are mixed in the solder layer 43, the region image includes an image showing the bubbles present in the bonding region.

  The user of the bubble rate calculation device 3 places the semiconductor device 4 inside the imaging device 2 through the opening 2a, and starts X-ray imaging by operating the operation unit 14 while viewing the display unit 13. The X-ray image that is the imaging result is displayed on the display unit 13, and the bubble rate that is the bubble calculation result is also displayed on the display unit 13. The user refers to the bubble rate displayed on the display unit 13 to determine whether or not the semiconductor device 4 is acceptable.

  The X-ray image may be configured so that an image taken by an X-ray imaging device separate from the bubble rate calculation device 3 is input to the calculation device 1. In this case, the bubble rate calculation device 3 does not need to include the photographing device 2. Moreover, the structure which the imaging device 2 has the operation | movement control function of the imaging device 2 which the calculating device 1 has, the bubble rate calculating function which detects the presence or absence of a bubble, and calculates a bubble rate may be sufficient. In this case, the bubble rate calculation device 3 does not need to include the calculation device 1.

4 and 5 are flowcharts showing the procedure of the bubble calculation process executed by the bubble rate calculation device 3. The CPU 10 starts executing bubble calculation processing in accordance with a user operation using the operation unit 14.
First, the CPU 10 outputs a control signal for starting X-ray imaging to the imaging apparatus 2 to perform X-ray imaging of the semiconductor device 4 (S11).
Next, the CPU 10 stores the X-ray image input from the imaging apparatus 2 in the HDD 15 (S12).
In the following, it is assumed that the main scanning direction of the X-ray image is the forward direction of the x axis and the sub scanning direction is the forward direction of the y axis.

FIG. 6 is a schematic diagram illustrating an example of an X-ray image captured by the bubble rate calculation device 3.
In the figure, reference numeral 60 denotes an area image, and the area image 60 is included in the X-ray image 600 input from the imaging apparatus 2 to the arithmetic apparatus 1. Each pixel of the X-ray image 600 has a multi-value pixel value of “0.00” or more and “1.00” or less. In the drawing, each of the squares of 9 squares and 12 squares (108 squares) indicates each pixel included in the area image 60. Here, the outer shape of the region image 60 (that is, the shape formed by the peripheral pixels of the region image 60) indicates the outer shape of the heat dissipation substrate 42 (that is, the shape of the peripheral portion of the heat dissipation substrate 42).

  Each of the pixel groups 601 and 602 indicated by fine hatching that rises to the right has, for example, a pixel value “0.20”, and each of the pixel groups 603 and 604 that are indicated by rough hatching that rises to the right is a pixel value “0.60”. The pixel group 605 shown by rough hatching with the lower right has a pixel value “0.70”, and the pixel groups 606, 607, and 608 shown by white have a pixel value “0.90”. However, the pixel group 609 indicated by fine hatching with a lower right side has a pixel value “0.80”. Further, pixels other than the region image 60 of the X-ray image 600 have a pixel value “1.00”.

That is, for example, by comparing each pixel value with a predetermined threshold value (for example, “0.95”) and setting a pixel having a pixel value less than the threshold value as a pixel of the region image 60, the region image 60 and the region image 60 other than As a result, the outer shape of the heat dissipation substrate 42 can be obtained. Note that, for example, a configuration in which an area image in an X-ray image is specified using pattern diagram data described below may be used.
However, for the region image 60, in order to simplify the following description, an example in which the pixel value and the arrangement of the pixel value are extreme is shown.

  After the process of S12 shown in FIG. 4 is completed, the CPU 10 reads a copy of the X-ray image stored in the HDD 15 and develops it in the RAM 12, and specifies a region image in the X-ray image (S13). At this time, it is desirable to mask other than the region image in the X-ray image.

After the process of S13 is completed, the CPU 10 binarizes the area image specified in S13 (S14).
The binarization in S14 simply binarizes the entire area image specified in S13 with a predetermined threshold (for example, “0.30”). 1 ”, and a pixel value“ 0 ”is a pixel having a pixel value less than the threshold value. The presence or absence of bubbles is not detected directly using the region image binarized in the process of S14 (hereinafter referred to as region binary image).

FIG. 7 is a schematic diagram illustrating an example of a region binary image obtained by the bubble rate calculation device 3.
In the figure, reference numeral 61 denotes a region binary image. The region binary image 61 is obtained by binarizing the region image 60 shown in FIG. 6 with a threshold value “0.30”. In the figure, each of squares of 9 squares and 12 squares (108 squares) is an individual pixel included in the region image 60, and each pixel has a pixel value “0” or a pixel value “1”. .

  Each pixel included in each of the pixel groups 611 and 612 indicated by hatching has a pixel value “0”, and each pixel included in a pixel group other than the pixel groups 611 and 612 has a pixel value “1”. That is, the pixel groups 611 and 612 correspond to the pixel groups 601 and 602 that have pixel values less than the threshold “0.30” in the region image 60. Pixel groups other than the pixel groups 611 and 612 correspond to the pixel groups 603 to 609 having a pixel value equal to or greater than the threshold “0.30” in the region image 60.

Next, the CPU 10 reads a mask image from the HDD 15 (S15) and develops it in the RAM 12. The mask image is a binary image, and via design drawing data having respective values such as the radius of each via hole 44 and the thickness of the plating portion 45 (hereinafter referred to as plating thickness) and the size of the heat dissipation substrate 42, and the heat dissipation substrate 42. Are generated using pattern diagram data having the coordinate values of the peripheral portion of the heat dissipation substrate 42 and the center coordinate values of the via holes 44 when the xy coordinate system is associated with the xy coordinate system.
In the present embodiment, the plating thickness is several hundred μm.

FIG. 8 is a schematic diagram showing an example of a mask image used in the bubble rate calculation device 3.
In the figure, 62 is a mask image, and the mask image 62 is stored in advance in the HDD 15 of the arithmetic device 1.
In the figure, each of the squares of 9 squares and 12 squares (108 squares) is an individual pixel included in the mask image 62, and each pixel has a pixel value “0” or a pixel value “1”. .
Each of the pixel groups 621 and 622 shown by hatching has a pixel value “0”, and the pixels other than the pixel groups 621 and 622 have a pixel value “1”.

  Each of the pixel groups 621 and 622 is eight pixels having a pixel value “0”, and is arranged in a square shape surrounding one pixel having a pixel value “1”. Each of the pixel groups 621 and 622 reflects the design radius, plating thickness, and center coordinates of the two via holes 44 and 44.

For example, the coordinates (3, 5) of the pixel located at the center of the region surrounded by the pixel group 621 indicate the design center coordinates of the via hole 44. Of the pixels shown in FIG. 8 and FIGS. 9 to 12, 15, 16, 19, 24, and 27 described later, the pixels that contain “.” Inside are the via holes 44 and 44, respectively. The pixel of the coordinate (henceforth a center pixel) corresponding to the center coordinate of is shown.
Further, as shown in FIG. 8, the number of pixels from the center pixel to any one of the pixels included in the pixel group 621 indicates the design radius of the via hole 44 (hereinafter referred to as via radius). Specifically, the number of pixels is “1”. Further, the number of pixels in the via radial direction of the pixel group 621 indicates the designed plating thickness, and specifically, the number of pixels is “1”.

Furthermore, a pixel having a pixel value “1” inside each of the regions surrounded by the pixel groups 621 and 622 indicates the central portion 42a of each of the via holes 44 and 44 (that is, the portion corresponding to the stack including the resin portion 46). ing. The pixel groups 621 and 622 indicate inner peripheral portions 42 b of the via holes 44 and 44 (that is, portions corresponding to the plated portions 45). A pixel having a pixel value “1” outside each of the regions surrounded by the pixel groups 621 and 622 indicates the outside 42 c of each of the via holes 44 and 44.
The outer shape of the mask image 62 indicates the outer shape of the heat dissipation substrate 42.

The mask image 62 shows an ideal bonding area in design. When the heat dissipation board 42 is manufactured in a precise and incomparable manner based on the via design drawing and the pattern diagram, a mask is obtained by binarizing a region image formed by X-ray imaging of the heat dissipation board 42 to which the semiconductor element 41 is not bonded. When the image 62 is subtracted, the pixel values of each pixel included in the subtraction result image are all “0”. In other words, the two binary images are completely overlapped.
Therefore, in the region image 60 shown in FIG. 6, the plating portions 45, 45,... Included in the region image 60 are masked by masking pixels corresponding to the pixel groups 621 and 622 in the mask image 62 shown in FIG. Can be masked.

  However, in the actual heat dissipation substrate 42, distortion of the laminated body 423, displacement of the opening positions of the via holes 44, 44,. Therefore, it is necessary to correct the mask image 62 according to the state of the heat dissipation substrate 42 actually provided in the semiconductor device 4 imaged by the imaging device 2.

  After the process of S15 shown in FIG. 4 is completed, the CPU 10 adjusts the relative position between the region image and the mask image, and then subtracts the pixel value of the corresponding pixel of the mask image from the pixel value of each pixel of the region image. (S16), a center coordinate correction value to be described later is obtained, and the mask image is corrected (S17). Hereinafter, the corrected mask image is referred to as a mask correction image.

  In the case of the region binary image 61 shown in FIG. 7 and the mask image 62 shown in FIG. 8, in order to adjust the relative position between the region binary image 61 and the mask image 62, the CPU 10 firstly selects the region binary image 61. And the origin of the xy coordinate system set in the region binary image 61 and the origin of the xy coordinate system set in the mask image 62 with reference to the outer shape of the heat dissipation substrate 42 included in each of the mask images 62. To match.

Next, for example, the via hole 44 corresponding to the pixel group 621 of the mask image 62 is a target via hole, and the relative position of the mask image 62 with respect to the region binary image 61 is set to “x _n ” in the x-axis direction and “y” in the y-axis direction. _n ”The mask image 62 is subtracted from the region binary image 61 in a state of being shifted.
However, if the number of pixels corresponding to the via radius of the via hole of interest is “N”, the positional deviation amount (x _n , y _n ) is an integer of −N ≦ x _n , y _n ≦ N, and x _n ≠ y _In some cases, x _n = y _n . In the case of the positional deviation amount (0, 0), the relative position of the mask image 62 with respect to the region binary image 61 is not changed, and the origins of the region binary image 61 and the mask image 62 remain in the same state.

  Here, the reason why the maximum value of the positional deviation amount is limited to the via radius of the via hole of interest is that the design opening position of the via hole of interest and the actual opening position cannot be considered to be significantly different. If the maximum value of the positional deviation amount is set to be unnecessarily large, there arises a problem that the calculation time becomes long.

Each pixel of the subtraction image obtained by subtracting the mask image 62 of the positional deviation amount (x _n , y _n ) from the region binary image 61 has a pixel value “0”, a pixel value “1”, or a pixel value “−1”. Have The CPU 10 has a range of vertical {4N + 1} pixels and horizontal {4N + 1} pixels centered on the central pixel of the target via hole in the region binary image 61 (that is, a range from the central pixel of the target via hole to a pixel corresponding to the diameter of the target via hole). ), The total number of pixels having a pixel value “1” and the number of pixels having a pixel value “−1” is obtained, and the positional deviation amount (x _n , y _n ) having the smallest obtained total number is noted. The via hole center coordinate correction value.

FIG. 9 is a schematic diagram illustrating an example of a subtraction image used in the bubble rate calculation device 3.
In the figure, reference numeral 63 denotes a subtraction image, and the subtraction image 63 is obtained by subtracting a mask image 62 (see FIG. 8) with a positional deviation amount (0, 0) from the region binary image 61 shown in FIG. In the figure, the pixel indicated by rough right-down hatching has a pixel value “0”, the pixel indicated by white outline has a pixel value “1”, and the pixel indicated by fine right-upward hatching is a pixel value. It has “−1”.

In the case of the positional shift amount (0, 0), when a via hole 44 corresponding to the pixel group 621 of the mask image 62 (hereinafter referred to as the upper left via hole 44) is the target via hole, 5 centered on the coordinates (3, 5). For a pixel × 5 pixel range (hereinafter referred to as 25 pixels), the total number of subtraction results is “11”.
On the other hand, when the via hole 44 corresponding to the pixel group 622 of the mask image 62 (hereinafter referred to as the lower right via hole 44) is the target via hole, the total number of subtraction results for 25 pixels centered on the coordinates (8, 3). Is “5”.
Similarly, in the case of the positional deviation amount (−1, 1), when the upper left via hole 44 is the target via hole, the total number of subtraction results for 25 pixels centered on the coordinates (3, 5) is “1”. is there.

On the other hand, when the lower right via hole 44 is the target via hole, the total number of subtraction results is “17” for 25 pixels centered on the coordinates (8, 3).
In the case of the positional deviation amount (0, 1), the total number of subtraction results relating to the upper left via hole 44 is “9”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “11”.
In the case of the positional shift amount (1, 1), the total number of subtraction results relating to the upper left via hole 44 is “11”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “17”.

In the case of the positional shift amount (−1, 0), the total number of subtraction results related to the upper left via hole 44 is “7”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “13”.
In the case of the positional shift amount (1, 0), the total number of subtraction results relating to the upper left via hole 44 is “11”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “11”.
In the case of the positional shift amount (−1, −1), the total number of subtraction results related to the upper left via hole 44 is “9”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “13”.

In the case of the positional shift amount (0, −1), the total number of subtraction results relating to the upper left via hole 44 is “11”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “7”.
In the case of the positional deviation amount (1, −1), the total number of subtraction results relating to the upper left via hole 44 is “13”. On the other hand, the total number of subtraction results for the lower right via hole 44 is “7”.
As a result, for the upper left via hole 44, the CPU 10 sets the positional shift amount (−1, 1) whose total number is “1” as the center coordinate correction value (x _r , y _r ), and the lower right via hole 44. The positional deviation amount (0, 0) with the total number being “5” is set as the center coordinate correction value (x _r , y _r ).

  When the minimum total number is the same for a plurality of positional deviation amounts, the CPU 10 may randomly select a central coordinate correction value from among the multiple positional deviation amounts. For example, the central coordinate correction value of another via hole 44 may be selected. An appropriate center coordinate correction value may be selected by referring to it.

FIG. 10 is a schematic diagram illustrating an example of a mask correction image obtained by the bubble rate calculation device 3.
In the figure, reference numeral 64 denotes a mask correction image, and the mask correction image 64 is obtained by correcting the mask image 62 shown in FIG. 8 based on the center coordinate correction values (x _r , y _r ) of the via holes 44.
In the figure, each of the squares of vertical 9 squares and horizontal 12 squares (108 squares) is an individual pixel included in the mask correction image 64, and each pixel has a value of “0” or “1”. Has a pixel value.
Each of the pixel groups 641 and 642 indicated by hatching has a pixel value “0”, and the pixels other than the pixel groups 641 and 642 have a pixel value “1”.

The pixel group 641 is obtained by correcting the pixel group 621 corresponding to the upper left via hole 44 in the mask image 62 based on the center coordinate correction value (−1, 1). 3 and 5), the center pixel of the pixel group 641 in the mask correction image 64 is the coordinate (2, 6). In other words, the center pixel is moved from the coordinates (3, 5) to the coordinates (2, 6) as indicated by white arrows in the figure.
The pixel group 642 is obtained by correcting the pixel group 622 corresponding to the lower right via hole 44 in the mask image 62 based on the center coordinate correction value (0, 0). 8 and 3), the center pixel of the pixel group 642 in the mask correction image 64 is also at the coordinates (8, 3).

  As described above, the CPU 10 corrects the mask image using the region image. By the way, in the bubble rate calculating device 3 in the present embodiment, a region image formed by X-ray imaging of the heat dissipation substrate 42 to which the semiconductor element 41 is bonded is used. A region image formed by X-ray imaging of the heat dissipation substrate 42 that has not been used may be used. In this case, since the semiconductor device 41, the solder layer 43, and bubbles mixed in the solder layer 43 are not affected, the mask image can be corrected more accurately.

  After the process of S17 shown in FIG. 4 is completed, the CPU 10 masks the pixels corresponding to the pixel group having the pixel value “0” of the mask correction image in the region image, thereby including the plated portion 45 included in the region image. , 45,... Are masked (S18). The CPU 10 in S18 functions as a masking unit. Hereinafter, the area image masked based on the mask correction image is referred to as a masking image.

FIG. 11 is a schematic diagram illustrating an example of a masking image obtained by the bubble rate calculation device 3.
In the figure, 65 is a masking image, and the masking image 65 is formed by masking the region image 60 shown in FIG. 6 based on the mask correction image 64 shown in FIG.
In the figure, each of the pixel groups 651 and 652 indicated by crossing hatching has a masked pixel, and the pixel group 650 indicated by a fine hatching with a right upward has a pixel value “0.20” and is increased to the right. Each of the pixel groups 653 and 654 shown by rough hatching has a pixel value “0.60”, and the pixel group 655 shown by rough right-down hatching has a pixel value “0.70” and is white. Each of the illustrated pixel groups 656, 657, and 658 has a pixel value “0.90”, and a pixel group 659 that is indicated by fine hatching with a lower right has a pixel value “0.80”.

  The masked pixel group 651 corresponds to the plated portion 45 of the upper left via hole 44, and the pixel group 652 corresponds to the plated portion 45 of the lower right via hole 44. In this way, each masked pixel (hereinafter referred to as a mask pixel) corresponds to the plated portion 45 of any one via hole 44. Therefore, a region surrounded by the pixel group 651 corresponds to a region inside the upper left via hole 44, a region surrounded by the pixel group 652 corresponds to a region inside the lower right via hole 44, and the pixel group 651, A region that is not surrounded by any of the pixel groups 651 and 652 except 652 corresponds to a region outside the via holes 44 and 44.

After the process of S18 shown in FIG. 4 is completed, the CPU 10 binarizes the masking image to generate a binary image (S19). The CPU 10 in S19 functions as an image binarization unit.
The binarization executed in the process of S19 is, for example, binarization of the masking image 65 shown in FIG. 11 with a predetermined threshold value except for the masked pixel groups 651 and 652, which is equal to or higher than the threshold value. A pixel having a pixel value is set to a pixel value “1”, and a pixel having a pixel value less than the threshold is set to a pixel value “0”.

  However, a threshold value I for binarizing pixels surrounded by the pixel groups 651 and 652 (specifically, pixels having coordinates (2, 6) and (8, 3)), and pixels other than this pixel are 2 A different threshold value O is used. That is, different threshold values are used for the inner region and the outer region of the via hole 44. This is because, as shown in FIG. 3, the copper thickness is thin at the central portion 42a of the via hole 44, and the copper thickness is thick at the outside 42c. For this reason, the threshold value I applied to the region inside the via hole 44 (ie, the region with high transmittance) is larger than the threshold value O applied to the region outside the via hole 44 (ie, the region with low transmittance). Is set. That is, threshold value O <threshold value I.

The threshold values I and O may be the same value. However, in order to binarize with one type of threshold value, it is necessary to binarize after correcting the difference in X-ray transmittance inside and outside the via hole 44 on the masking image 65.
Further, the threshold used in the process of S14 and the threshold used in the process of S19 may be the same value or different.

FIG. 12 is a schematic diagram showing an example of a binary image obtained by binarizing the masking image 65 shown in FIG.
In the figure, 68 is a binary image, and the binary image 68 is based on the masking image 65 shown in FIG. 11 and the threshold value I = “0.85” for the pixels surrounded by the pixel groups 651 and 652. Is applied and a threshold value O = “0.75” is applied to pixels other than this pixel to binarize.
In FIG. 12, each of the pixel groups 681 and 682 indicated by crossing hatching indicates a masked pixel, and each pixel included in the pixel group 683 indicated by right-upward hatching has a pixel value “0”. Each pixel included in each of the pixel groups 686, 687, and 688 shown in white has a pixel value “1”.

  By the way, in the masking image 65 shown in FIG. 11, the pixel at the coordinates (2, 6), which is the region inside the via hole 44, is not a bubble pixel. However, if the masking image 65 is binarized only with the threshold value O, the pixel value of the coordinate (2, 6) is “1”. Therefore, in the processing of S20 described later, the pixel is a bubble pixel. It will be judged. This is because the central portion 42 a of the via hole 44 has a high X-ray transmittance even if no bubbles are mixed in the solder layer 43. On the other hand, when the masking image 65 is binarized only by the threshold value I, the outside 42c of the via hole 44 has a low transmittance even if bubbles are mixed in the solder layer 43. Is determined not to be a bubble pixel.

After the process of S19 is completed, the CPU 10 performs area bubble detection for detecting the presence or absence of bubbles (hereinafter referred to as area bubbles) located in the bonding area excluding the plated portions 45, 45,... (S20). The CPU 10 in S20 functions as a region bubble detection unit.
Specifically, the CPU 10 determines whether the pixel value of each pixel is “1” or “0” except for the masked pixel group, and determines that the pixel value is “1”. The coordinate value of the obtained pixel is stored in the RAM 12, and the number of pixels of this pixel is counted. This number of pixels is the number of pixels in the region bubble. This is because a pixel having a pixel value “1” is a pixel in a region bubble (hereinafter referred to as a bubble pixel). If there are zero bubble pixels, there is no region bubble, and there are one or more bubble pixels. If so, there is a region bubble.

In the case of the binary image 68 shown in FIG. 12, each pixel included in each of the pixel groups 686, 687, and 688 corresponds to a region bubble, and the number of pixels of the region bubble is “7”.
If the bubble rate is calculated at this time, the number of pixels “7” of the detected region bubble is divided by the total number of pixels “108” (= 12 × 9) of the binary image 68. About 6.48%. However, since the number of pixels of bubbles (hereinafter referred to as “partial bubbles”) located in the plated portions 45 and 45 is not considered at this time, an accurate bubble ratio cannot be obtained.

  After the process of S20 shown in FIG. 4 is completed, the CPU 10 determines whether or not a region bubble exists as shown in FIG. 5 (S21), and if it exists (YES in S21), executes the processes after S22. To do.

13 and 14 are schematic diagrams illustrating an example of bubbles to be detected by the bubble rate calculation device 3. Each of FIGS. 13 and 14 shows the first quadrant of the xy coordinate system with the center coordinate of the via hole 44 as the origin.
Air bubbles 71 shown in FIG. 13 have an elliptical shape extending outside the via hole 44 and the plated portion 45, and do not reach the inside of the via hole 44. On the other hand, the bubble 72 has a circular shape extending over the inner side of the via hole 44 and the plated portion 45, and does not reach the outer side of the via hole 44. Also, the bubbles 73 shown in FIG. 14 are in the form of teardrops that extend outside the via hole 44, the plated portion 45, and the inside of the via hole 44.

Since the plated portion 45 is masked, a portion that does not overlap the plated portion 45 among the bubbles 71 to 73 shown in FIGS. 13 and 14 is detected as a region bubble by executing the process of S20. However, the portion overlapping the plated portion 45 is not detected as a region bubble even when the processing of S20 is executed, and thus needs to be detected separately.
By the way, the bubbles mixed in the solder layer 43 due to the surface tension of the solder often have a round shape such as a circular shape, an elliptical shape, or a teardrop shape. The bubble rate calculation method of the present embodiment detects the presence or absence of partial bubbles located in the plated portion 45 by utilizing such properties.

The bubbles 71 shown in FIG. 13 are in contact only with the outside of the via hole 44 at the same position in the circumferential direction of the via hole 44 of the plated portion 45.
For such bubbles 71, normals 713 and 714 of the plated portion 45 passing through the intersections 711 and 712 of the bubble peripheral portion of the bubble 71 and the plated portion 45 are obtained. The normal line 713 (normal line 714) is a straight line connecting the center point of the via hole 44 and the intersection point 711 (intersection point 712).
Further, tangent lines 715 and 716 of the bubble peripheral edge of the bubble 71 passing through the intersection points 711 and 712 are obtained.

  Of the normal lines 713, 714 and the tangent lines 715, 716, the intersection of the bubble peripheral edge of the bubble 71 and the plated portion 45 is not the same (that is, the intersection exists at a different position). Is selected. In this case, a combination of normal lines 713 and 714, a combination of normal lines 713 and tangent lines 716, a combination of normal lines 714 and tangent lines 715, and a combination of tangent lines 715 and 716 are selected, and normal lines intersecting at intersection 711 are selected. The combination of 713 and tangent 715 and the combination of normal 714 and tangent 716 that intersect at the intersection 712 are excluded.

  Among the regions surrounded by the two selected straight lines and located in the plated portion 45, the region having the smallest area is a region related to a combination of the normal line 713 and the tangent line 716 (a region indicated by vertical hatching). ). Therefore, this region is recognized as a partial bubble. This is because the shape of this region is most similar to the shape of the bubbles 71 located in the plated portion 45.

On the other hand, the bubble 72 is in contact only with the inside of the via hole 44 at the same position in the circumferential direction of the via hole 44 of the plated portion 45.
For such bubbles 72, normals 723 and 724 of the plated portion 45 passing through the intersections 721 and 722 of the bubble peripheral portion of the bubble 72 and the plated portion 45 are obtained. The normal line 723 (normal line 724) is a straight line connecting the center point of the via hole 44 and the intersection point 721 (intersection point 722).
Further, tangents 725 and 726 of the bubble peripheral edge of the bubble 72 passing through the intersections 721 and 722 are obtained.

  Of the normal lines 723 and 724 and the tangent lines 725 and 726, a combination of two straight lines that do not have the same intersection point between the bubble peripheral edge of the bubble 72 and the plated portion 45 is selected. In this case, a combination of normal lines 723 and 724, a combination of normal lines 723 and tangent lines 726, a combination of normal lines 724 and tangent lines 725, and a combination of tangent lines 725 and 726 are selected, and normal lines intersecting at intersection point 721 are selected. The combination of 723 and tangent 725 and the combination of normal 724 and tangent 726 that intersect at the intersection 722 are excluded.

  Among the regions surrounded by the two selected straight lines and located in the plated portion 45, the region having the smallest area is a region related to the combination of the normal lines 723 and 724 (region indicated by horizontal hatching). It is. Therefore, this region is recognized as a partial bubble. This is because the shape of this region is most similar to the shape of the bubble 72 located in the plated portion 45.

The bubbles 73 shown in FIG. 14 are in contact with both the outside and the inside of the via hole 44 at the same position in the circumferential direction of the via hole 44 of the plated portion 45.
For such bubbles 73, the outer intersections 731 and 732 of the bubble peripheral portion of the bubbles 73 contacting the plated portion 45 from the outside of the via hole 44 and the plated portion 45 are obtained, and the plated portion 45 is found from the inside of the via hole 44. Inner intersections 733 and 734 between the peripheral edge of the bubble 73 in contact with the plated portion 45 are obtained.

  Next, outside of a region surrounded by the outer intersection points 731 and 732 and the inner intersection points 733 and 734 among straight lines passing through either the outer intersection points 731 or 732 and any of the inner intersection points 733 and 734 (hereinafter referred to as an intersection surrounding region). Two straight lines 735 and 736 intersecting with each other are obtained. Here, the straight line 735 passes through the outer intersection 731 and the inner intersection 733, and the straight line 736 passes through the outer intersection 732 and the inner intersection 734.

Here, the combination of the straight line 737 passing through the outer intersection 731 and the inner intersection 734 and the straight line 738 passing through the outer intersection 732 and the inner intersection 733 intersects within the intersection surrounding region, and the combination of the straight line 735 and the straight line 737 is Crossing at the outer intersection 731 and the combination of the straight line 736 and the straight line 738 are excluded because they intersect at the outer intersection 732. Similarly, the combination of the straight line 735 and the straight line 738 intersects at the inner intersection point 733, and the combination of the straight line 736 and the straight line 737 intersects at the inner intersection point 734, and thus is excluded.
Incidentally, the straight lines passing through the outer intersections 731 and 732 and the straight lines passing through the inner intersections 733 and 734 are naturally excluded.

  Finally, a region surrounded by the calculated two straight lines 735 and 736 and located in the plated portion 45 (region indicated by hatching) is recognized as a partial bubble. This is because the shape of this region is most similar to the shape of the bubble 73 located in the plated portion 45.

In FIGS. 13 and 14, the case where the shape of each of the bubbles 71 to 73 is known in advance has been described as an example. However, in the actual bubble calculation process, the shape of each of the bubbles 71 to 73 located in the plated portion 45 shown in FIGS. 13 and 14 is not known when it is determined YES in the process of S21.
Here, it is possible to easily determine that the region where the bubble pixels are continuous vertically and horizontally is a lump of bubbles.

  A lump of bubbles not in contact with the via hole 44 can be easily determined to be a single bubble, but the bubble in contact with the via hole 44 from the outside (or the inside) is a single bubble or is not in the via hole 44. It is necessary to determine whether it is a single bubble including other bubbles coming in contact from the inside (or outside).

For this reason, the plating part surrounded by the normal line of the plating part passing through the intersection of the bubble peripheral part of the bubble and the plating part (hereinafter referred to as the normal surrounding part) overlaps the normal surrounding part related to other bubbles. It is determined whether or not a pixel included in one normal line portion is included in another normal line portion.
In the case of FIG. 13, the normal surrounding area surrounded by the normal lines 713 and 714 of the plating portion 45 passing through the intersection points 711 and 712 and the normal surrounding area surrounded by the normal lines 723 and 724 of the plating portion 45 passing through the intersection points 721 and 722. The part does not overlap with other normal surrounding parts. Accordingly, it is determined that the bubbles contacting the via hole 44 from the outside and the bubbles contacting from the inside are single bubbles (that is, the bubbles 71 and 72).

In the case of FIG. 14, a normal surrounding portion surrounded by normal lines 73 a and 73 b of the plating portion 45 passing through the outer intersection points 731 and 732 and a method surrounded by normal lines 73 c and 73 d of the plating portion 45 passing through the inner intersection points 733 and 734. It overlaps with the line go part. Therefore, the bubbles contacting the via hole 44 from the outside and the bubbles contacting from the inside are both a single bubble (that is, the bubble 73).
When the procedure for obtaining partial bubbles as described above is applied to image processing using digital data, for example, the procedure shown in FIGS. 15, 16, and 19 to 24 described later is used.

FIG. 15 is a schematic diagram illustrating an example of a binary image generated by the bubble rate calculation device 3.
In the figure, reference numeral 66 denotes a binary image, and the binary image 66 is obtained by binarizing the masking image in the process of S19. In the figure, 18 pixels 661, 661,..., Respectively indicated by right-up hatching, are mask pixels, and 25 pixels 662, 662,. However, the pixels 663, 663,... Indicated by the right-down hatching have pixel values “0”. Therefore, the pixels 662, 662,... Are bubble pixels.

  The mask pixels 661, 661,... Correspond to 1/4 of the plated portion 45 of the via hole 44 corresponding to the mask pixels 661, 661,..., And in FIG. The first quadrant of the xy coordinate system is shown. Here, the center pixel is a pixel 663 having coordinates (0, 0). Hereinafter, this pixel 663 is referred to as a central pixel 660.

FIG. 16 is a schematic diagram illustrating another example of a binary image generated by the bubble rate calculation device 3.
In the figure, reference numeral 67 denotes a binary image, and the binary image 67 is formed by binarizing the masking image in the process of S19. The 18 mask pixels 671, 671,... Are shown in a right-upward hatching, and 25 bubble pixels 672, 672, and 8 bubble pixels 678, 678,. , And the center pixel 670 and the pixels 673, 673,... Having the pixel value “0” are indicated by right-down hatching.
The mask pixels 671, 671,... Correspond to a quarter of the plated portion 45 of the via hole 44 corresponding to the mask pixels 671, 671,..., And in FIG. The first quadrant of the xy coordinate system is shown.

In the case of YES in S21 shown in FIG. 5, the CPU 10 detects the presence or absence of bubbles (hereinafter referred to as contact bubbles) in contact with each plating portion 45 among the region bubbles detected in the process of S20 (S22). The CPU 10 in S22 functions as a contact bubble detection unit.
In S20, the coordinate value of one or more bubble pixels is stored in the RAM 12, or the coordinate value is not stored. Therefore, if the coordinate value of the bubble pixel is not stored in the RAM 12, there is no region bubble, so there is no contact bubble, and if the coordinate value of one or more bubble pixels is stored in the RAM 12, There may be bubbles.

After completion of the process of S22, the CPU 10 determines whether or not a contact bubble exists (S23), and if it exists (YES in S23), obtains a range of the contact bubble (S24).
In S23, the CPU 10 determines whether or not at least one of the surrounding eight pixels having each bubble pixel stored in the RAM 12 as a target pixel is a mask pixel, and at least one of the eight surrounding pixels is a mask pixel. The pixel of interest determined to be stored in the RAM 12 as a bubble pixel in contact with the mask pixel (hereinafter referred to as a contact bubble pixel).

As described above, the contact bubble pixel is a pixel included in the contact bubble and is in contact with the mask pixel at at least one point. In other words, the contact bubble pixel shares at least one of the four vertices with the mask pixel. Note that it may be configured to determine whether at least one of the four pixels around the target pixel is a mask pixel. In this case, the contact bubble pixel is in contact with the mask pixel on at least one side.
Hereinafter, a bubble pixel that is not a contact bubble pixel (a bubble pixel that is not in contact with the mask pixel) is referred to as a non-contact bubble pixel, and is simply referred to as a bubble pixel when the contact bubble pixel and the non-contact bubble pixel are not distinguished.

For example, the bubble pixel 662 at coordinates (4, 6) shown in FIG. 15 and the bubble pixel 672 at coordinates (4, 6) and the bubble pixel 678 at coordinates (1, 3) shown in FIG. 16 are contact bubble pixels, respectively. is there. In S24, one contact bubble including these contact bubble pixels is obtained.
Specifically, the CPU 10 in S24 determines, for each contact bubble pixel, whether the contact bubble pixel is located outside or inside the via hole 44 corresponding to the mask pixel in contact with the contact bubble pixel. To do.

  Next, for each contact bubble pixel located outside each via hole 44, the CPU 10 in S24 determines whether or not a bubble pixel is included in the surrounding eight pixels having the contact bubble pixel as a target pixel. When it is determined that the bubble pixel is included, the determination is made as to whether or not the bubble pixel is included in the surrounding eight pixels with the bubble pixel as the target pixel. For example, when the contact bubble pixel 662 at the coordinates (4, 6) shown in FIG. 15 is the target pixel, the coordinates (4, 5), the coordinates (4, 7), the coordinates (3, 5) to the coordinates (3, 7 ), And coordinates (5, 5) to coordinates (5, 7) are determined as to whether or not each pixel is a bubble pixel, and coordinates (4, 5), coordinates (5, 5), and coordinates (5, 6) are determined. ) Since each pixel is a bubble pixel, next, for example, the non-contact bubble pixel 662 at coordinates (5, 6) becomes the target pixel. Note that it may be configured to determine whether or not bubble pixels are included in the four pixels around the target pixel.

In this way, the CPU 10 obtains a region where bubble pixels are continuous in the x-axis direction and the y-axis direction starting from the contact bubble pixel, and recognizes the obtained region as a contact bubble located outside the via hole 44. To do. Here, when another contact bubble pixel is included in the obtained contact bubble, it is not necessary to newly obtain a bubble pixel including the contact bubble pixel.
As described above, the bubble pixels 662, 662,... Shown in FIG. 15 and the bubble pixels 672, 672, etc. shown in FIG. 16 are each identified as contact bubbles located outside the via hole 44.

  For each of the contact bubble pixels located inside each via hole 44, the CPU 10 determines that the area obtained in the same manner is a contact bubble located inside the via hole 44. For this reason, the bubble pixels 678, 678,... Shown in FIG. 16 are recognized as contact bubbles located inside the via hole 44.

  After the process of S24 shown in FIG. 5 is completed, the CPU 10 calculates, for each contact bubble obtained in S24, an intersection pixel between the bubble peripheral portion of each contact bubble and the plated portion 45 in contact with the contact bubble ( S25) Based on the normal of the plated portion 45 passing through the calculated intersection pixel, the normal surrounding portion related to the contact bubble is calculated (S26). Furthermore, the CPU 10 determines that each contact bubble obtained in S24 is a contact bubble that is in contact only with the outside of the via hole 44 or a contact bubble that is in contact with only the inside of the via hole 44 based on the normal surrounding area calculated in S26. Or a contact bubble that is in contact with both the outside and the inside of the via hole 44 (S27).

  First, for example, in the case of the binary image 66 shown in FIG. 15, the CPU 10 in S25 selects the pixel having the smallest coordinate value of the x coordinate and the largest coordinate value of the y coordinate among the contact bubble pixels 662, 662,. Is the first intersection pixel, and the pixel having the largest x-coordinate value and the smallest y-coordinate value is the second intersection pixel. In this case, the first intersection pixel is a contact bubble pixel 662 at coordinates (4, 6), and the second intersection pixel is a contact bubble pixel 662 at coordinates (7, 2).

  Similarly, in the case of the binary image 67 shown in FIG. 16, the mask pixels 671, 671,... (That is, plating) from the outside of the region (that is, the via hole 44) that includes the center pixel 670 and is surrounded by the mask pixels 671, 671,. For the contact bubble in contact with the portion 45), the first intersection pixel is the contact bubble pixel 672 at coordinates (4, 6), and the second intersection pixel is the contact bubble pixel 672 at coordinates (7, 2). It is. Further, for the contact bubbles that are in contact with the mask pixels 671, 671,... From the inside of the via hole 44, the first intersection pixel is the contact bubble pixel 672 at coordinates (1, 3), and the second intersection pixel is This is a contact bubble pixel 672 at coordinates (3, 1).

  Next, for example, in the case of the binary image 66 shown in FIG. 15, the CPU 10 in S26 passes the intersection of the coordinates (4, 6) and the intersection of the coordinates (7, 2) calculated in S25 and the center pixel 660. A straight line (that is, a normal line of the plating portion 45 passing through the intersection calculated in S25) is obtained, and mask pixels 661, 661,... (That is, a normal surrounding portion) surrounded by the obtained normal lines (broken line in FIG. 15). Calculate.

  On the other hand, for example, in the case of the binary image 67 shown in FIG. 16, the CPU 10 in S26 uses the coordinates (4, 6) calculated in S25 for the contact bubbles contacting the mask pixels 671, 671,. A straight line (that is, a normal line) passing through the intersection point and the intersection point of coordinates (7, 2) and the center pixel 670 is obtained, and mask pixels 671, 671,... Surrounded by the obtained normal line (broken line in FIG. 16) are obtained. Calculate. Similarly, for the contact bubbles contacting the mask pixels 671, 671,... From the inside of the via hole 44, the CPU 10 intersects the intersection of the coordinates (1, 3) and the intersection of the coordinates (3, 1) calculated in S25, respectively. A straight line (that is, a normal line) passing through the center pixel 670 is obtained, and mask pixels 671, 671,... Surrounded by the obtained normal line (two-dot chain line in FIG. 16) are calculated.

  Further, for example, in the case of the binary image 66 shown in FIG. 15, the mask pixels 661, 661,... Included in the normal surrounding portion calculated in S26 are not included in the normal surrounding portions related to other contact bubbles. The CPU 10 in S27 classifies the contact bubbles to which the bubble pixels 662, 662,... Belong as contact bubbles that are in contact only with the outside of the via hole 44 at the same position in the circumferential direction of the via hole 44 of the plated portion 45.

On the other hand, in the case of the binary image 67 shown in FIG. 16, some mask pixels 671, 671,... Included in the normal surrounding area calculated in S <b> 26 related to the contact bubbles outside the via hole 44 are inside the via hole 44. Since the bubble 10 is included in the normal surrounding portion related to the contact bubble, the CPU 10 contacts the contact bubble to which the bubble pixel 672, 672,... And the bubble pixel 678, 678,. Classify as a bubble.
Based on the contact bubble classification as described above, the CPU 10 in S27 generates an appropriate bubble classification table and stores it in the RAM 12.

  After the process of S27 shown in FIG. 5 is completed, the CPU 10 resets the variable K to “0” (S28), selects one of the via holes 44, 44,... (S29), and plating the selected via hole 44. Based on the shape of the contact bubble in contact with the portion 45, a virtual line indicating the characteristics of the shape of the contact bubble is obtained, and the region surrounded by the obtained virtual line and located in the plated portion 45 is a bubble. Partial bubble detection is performed to detect the presence or absence of a region to be recognized as (S30). In the process of S30, the CPU 10 calls and executes a subroutine (see FIGS. 17 and 18 described later) for performing the partial bubble detection process.

FIGS. 17 and 18 are flowcharts showing a sub-routine of the partial bubble detection processing procedure executed by the bubble rate calculating device 3.
First, the CPU 10 extracts a mask pixel corresponding to the plated portion 45 of the selected via hole based on the center coordinates of the via hole 44 selected in S29 (hereinafter referred to as a selected via hole) in the binary image generated in the process of S19. (S41). In the process of S41, for example, in the case of the binary image 66 shown in FIG. 15, mask pixels 661, 661,... Are extracted, and in the case of the binary image 67 shown in FIG. .

  Next, the CPU 10 refers to the bubble classification table generated in S27, determines whether there is a contact bubble in contact with the mask pixel extracted in S41 (S42), and if it does not exist (NO in S42), End the partial bubble detection process and return to the original routine.

If there is a contact bubble in contact with the mask pixel extracted in S41 (YES in S42), the CPU 10 selects one of the contact bubbles in contact with the mask pixels 661, 661,... Extracted in S41 (S43). It is determined whether or not the selected contact bubble is in contact with both the outside and the inside of the selected via hole (S44).
When the bubble selected in S43 is a contact bubble that is in contact with only the outside or the inside of the selected via hole (NO in S44), the CPU 10 detects the presence / absence of a partial bubble continuous with the contact bubble in S46 and subsequent steps. Execute the process.

  In the following, a case where a contact bubble in contact with only the outside of the selected via hole exists in the first quadrant of the xy coordinate system with the center pixel of the selected via hole as an origin (see FIG. 15) will be described. When a contact bubble is present in any of the second to fourth quadrants, the quadrant in which the contact bubble is present is rotated around the origin, or the x-axis or the y-axis is inverted to change the coordinate axis to the first By making it correspond to one quadrant, it is possible to detect the presence or absence of partial bubbles by the same process. Also, even when one contact bubble exists over a plurality of quadrants, a portion where the contact bubble and the coordinate axis overlap is regarded as a bubble peripheral portion, and the same processing is performed on each of the plurality of quadrants. Thus, it is possible to detect the presence or absence of partial bubbles. In this case, one normal line of the mask area and one tangent line of the contact bubble coincide with the coordinate axis regarded as the bubble peripheral portion.

FIG. 19 is a schematic diagram showing a region surrounded by two normal lines in the binary image 66 shown in FIG. 20 is a schematic diagram illustrating a region surrounded by one normal line and a tangent in the binary image 66, and FIG. 21 is a schematic diagram illustrating a region surrounded by another normal line and a tangent in the binary image 66. is there. FIG. 22 is a schematic diagram showing a region surrounded by two tangent lines in the binary image 66.
In the case of NO in S44 shown in FIG. 17, the CPU 10 calculates the intersection pixel calculated in S25 between the bubble peripheral portion of the contact bubble selected in S43 (hereinafter referred to as the selected bubble) and the plated portion 45 of the selected via hole (FIG. 19 to FIG. 19). 22, the first intersection pixel is a contact bubble pixel 662 at coordinates (4, 6), and the second intersection pixel is a contact bubble pixel 662 at coordinates (7, 2). Execute.

  The CPU 10 determines whether only one intersection pixel is obtained in S25 (S46). When there is only one intersection pixel (YES in S46), there is no partial bubble continuous with the selected bubble (that is, the number of pixels in the region to be recognized as a partial bubble continuous with the selected bubble is “0”). Therefore, the process proceeds to S52 described later.

When there are two or more intersection pixels determined in S25 (NO in S46), the CPU 10 calculates a normal passing through each of the two intersection pixels determined in S25 out of the normal lines of the plated portion 45 of the selected via hole. (S47).
In the case of the binary image 66 shown in FIG. 15, the CPU 10 in S47 calculates the center pixel (0, 0) of the selected via hole, the intersection pixel of coordinates (4, 6), and the intersection pixel of coordinates (7, 2). Calculate the connecting straight line. As a result, normal lines 664 and 665 shown in FIGS. 19 to 21 are obtained.

After the process of S47 shown in FIG. 17, the CPU 10 calculates a tangent passing through each of the two intersection pixels determined in S25 among the tangents of the peripheral edge of the selected bubble (S48).
In the case of the binary image 66 shown in FIG. 15, the CPU 10 in S48 has the smallest coordinate value of the x coordinate and the y coordinate of the contact bubble pixels 662, 662,. The pixel with the largest coordinate value is defined as the first additional pixel, and the pixel with the largest coordinate value for the x coordinate and the smallest coordinate value for the y coordinate is selected from the contact bubble pixels 662, 662,. Let it be a second additional pixel.

  19 to 22, the first additional pixel is a contact bubble pixel 662 at coordinates (4, 5), and the second additional pixel is a contact bubble pixel 662 at coordinates (6, 2). Further, the CPU 10 calculates a straight line connecting the first intersection pixel obtained in S25 and the first additional pixel, and a straight line connecting the second intersection pixel obtained in S25 and the second additional pixel. To do. As a result, tangent lines 666 and 667 shown in FIGS. 20 to 22 are obtained.

Incidentally, even if the selected bubble is a contact bubble that is in contact with only the inside of the selected via hole, the CPU 10 in S48 determines the coordinate value of the x coordinate among the contact bubble pixels 662, 662,... That are in contact with the first intersection pixel. Is the first additional pixel, and the contact bubble pixels 662, 662,... That are in contact with the second intersection pixel are the largest, the coordinate value of the x coordinate and the coordinate of the y coordinate. The pixel having the smallest value is set as the second additional pixel.
The CPU 10 in S47 and S48 as described above functions as a virtual line calculation means.

After the processing of S48 shown in FIG. 17, the CPU 10 selects a combination of two straight lines that do not have the same intersection point between the peripheral edge of the selected bubble and the plated portion 45, out of the two normals and tangents. (S49).
In the case of the binary image 66 shown in FIGS. 15 to 22, a combination including intersection pixels having the same coordinate value is excluded from pixels passing through the normal lines 664 and 665 and the tangent lines 666 and 667. As a result, the combination of the normals 664 and 665, the combination of the normal 664 and the tangent 667, the combination of the normal 665 and the tangent 666, and the combination of the tangents 666 and 667 are selected, and the coordinates (4, 6) are mutually selected. The combination of the normal line 664 and the tangent line 666 passing through the intersection pixel and the combination of the normal line 665 and the tangent line 667 passing through the intersection pixel of the coordinates (7, 2) are excluded.

  By the way, when only one intersection pixel is obtained in S25, the tangent and normal passing through this intersection pixel pass through bubble pixels 662 (that is, intersection pixels) having the same coordinate value. For this reason, even if the processing of S47 and S48 is executed when only one intersection pixel is obtained in S25, the obtained combination of tangent and normal is excluded in S49, and will be described later. There is no point in executing the processing after S50.

Then, the CPU 10 counts the number of pixels of the mask pixels 661, 661,... Surrounded by the two selected straight lines for each combination of the two straight lines selected in S49 (S50), and the smallest counting result C Is added to the variable K (S51). That is, the variable K is the total number of pixels in a region that should be recognized as a partial bubble that is continuous with the selected bubble.
The CPU 10 in S25, S46, and S49 to S51 functions as a partial bubble detection unit.

19 are mask pixels 661, 661,... Surrounded by normal lines 664, 665, and the number of pixels is “7”. Here, the straight line indicating the normal line 664 is y = {3/2} x, and the straight line indicating the normal line 665 is y = {2/7} x. The mask pixels 661, 661,... Surrounded by the normal lines 664, 665 satisfy y ≦ {3/2} x and y ≧ {2/7} x, respectively. Accordingly, for example, the mask pixel 661 at the coordinates (3, 5) passes through the normal line 664 but is not counted as the mask pixel 661 surrounded by the two selected straight lines.
.., Mask pixels 661, 661,... Indicated by fine hatching at the upper right in FIG. 20 are mask pixels 661, 661,... Surrounded by normal line 664 and tangent line 667, and the number of pixels is “7”. .

.., Mask pixels 661, 661,..., Which are shown by fine hatching in the upper right, are mask pixels 661, 661,... Surrounded by the normal line 665 and the tangent line 666, and the number of pixels is “5”.
.., Mask pixels 661, 661,..., Which are indicated by fine hatching in the upper right, are mask pixels 661, 661,... Surrounded by tangent lines 666, 667, and the number of pixels is “5”.
As a result, the minimum number of pixels is “5” among the number of mask pixels 661, 661,... Surrounded by the two straight lines selected in S49.

  After the process of S51 shown in FIG. 17 is completed, the CPU 10 determines whether or not all the contact bubbles contacting the mask pixel extracted in S41 have been selected in S43 (S52), and there is a contact bubble that has not been selected yet. If so (NO in S52), the process returns to S43. If all selections have been made (YES in S52), the partial bubble detection process is terminated and the process returns to the original routine.

When the contact bubble selected in S43 shown in FIG. 17 is a contact bubble that is in contact with both the outside and the inside of the selected via hole (YES in S44), the CPU 10 detects the presence or absence of a partial bubble that continues to this contact bubble. Therefore, the processing from S53 onward shown in FIG. 18 is executed.
In the following, a case will be exemplified in which contact bubbles in contact with both the outside and the inside of the selected via hole are present in the first quadrant of the xy coordinate system with the center pixel of the selected via hole as the origin (see FIG. 16).

  If YES in S44 shown in FIG. 17, the CPU 10 calculates the outer intersection pixel between the bubble peripheral edge of the selected bubble that is in contact with the outside of the selected via hole and the plated portion 45 of the selected via hole, as shown in FIG. Further, an inner intersection pixel between the bubble peripheral portion of the selected bubble in contact with the inside of the selected via hole and the plated portion 45 of the selected via hole is calculated (S53).

  FIG. 23 is a schematic diagram showing four straight lines in the binary image 67 shown in FIG. FIG. 24 is a schematic diagram showing a region surrounded by two straight lines in the binary image 67.

  As shown in FIG. 23, in the case of the binary image 67, the CPU 10 in S53 determines that the coordinate value of the x coordinate is the smallest and y among the contact bubble pixels 672, 672,... Of the selected bubble in contact with the outside of the selected via hole. The pixel with the largest coordinate value is the first outer intersection pixel, and the pixel with the largest x coordinate value and the smallest y coordinate value is the second outer intersection pixel. Among the contact bubble pixels 678, 678,... Of the selected bubble in contact with the inside of the selected via hole, the pixel having the smallest coordinate value of the x coordinate and the largest coordinate value of the y coordinate is set as the first inner intersection pixel. The pixel having the maximum coordinate value of the x coordinate and the minimum coordinate value of the y coordinate is defined as a second inner intersection pixel.

  As a result, in the case of the binary image 67, the first outer intersection pixel is the contact bubble pixel 672 at coordinates (4, 6), and the second outer intersection pixel is the contact bubble pixel at coordinates (7, 2). 672, the first inner intersection pixel is a contact bubble pixel 672 at coordinates (1, 3), and the second inner intersection pixel is a contact bubble pixel 672 at coordinates (3, 1).

  After the process of S53 shown in FIG. 18, the CPU 10 determines whether or not the total of the outer intersection pixel and the inner intersection pixel obtained in S53 is four or more (S54). If the total of the various intersection pixels obtained in S53 is only 2 or 3 (NO in S54), there is no partial bubble continuous with the selected bubble (that is, the region to be recognized as a partial bubble continuous with the selected bubble). Since the number of pixels can be regarded as “0”), the process proceeds to S52.

When the total of the various intersection pixels obtained in S53 is four or more (YES in S54), the CPU 10 draws a straight line passing through the intersection of any of the two outer intersection pixels and any of the two inner intersection pixels. Calculate (S55). A straight line connecting the outer intersection pixels and a straight line connecting the inner intersection pixels are not calculated. The CPU 10 in S53 and S55 functions as virtual line calculation means.
As a result, as shown in FIG. 23, in the case of the binary image 67, straight lines 674 to 677 are obtained.

  After the processing of S55 shown in FIG. 18 is finished, the CPU 10 includes two lines intersecting outside the area surrounded by the two outer intersection pixels and the inner intersection pixels (that is, the intersection surrounding area) 679 among the four straight lines. A combination of straight lines is selected (S56).

  As shown in FIG. 23, in the case of the binary image 67, a combination passing through the same outer intersection pixel or the same inner intersection pixel and a combination passing through the same pixel inside the intersection surrounding area 679 are excluded. As a result, a combination of straight lines 674 and 675 is selected, and a combination of straight lines 674 and 676 that pass through the outer intersection pixels at coordinates (4, 6) and a straight line 675 that passes through the outer intersection pixels at coordinates (7, 2). 677, a combination of straight lines 676, 677 passing through mask pixel 671 at coordinates (3, 3), a combination of straight lines 674, 677 passing through inner intersection pixels at coordinates (1, 3), and a coordinate ( 3,1) is excluded from the combination of straight lines 675,676 passing through the intersection pixel.

  By the way, when there are only two intersection pixels obtained in S53, since there is only one straight line passing through these intersection pixels, the process of S56 cannot be executed. When only three intersection pixels are obtained in S53, the two straight lines passing through these intersection pixels pass through the same intersection pixel, and are therefore excluded in the process of S56. Therefore, there is no point in executing the processing after S57 described later.

Next, the CPU 10 counts the number of pixels of the mask image surrounded by the two straight lines selected in S56 (S57), moves the process to S51, and adds the count result C to the variable K.
As shown in FIG. 24, in the case of the binary image 67, the CPU 10 in S57 determines the mask pixels 671, 671,... Surrounded by straight lines 674, 675, that is, mask pixels 671, 671 shown by fine hatching rising to the right. ,... Are counted.
The CPU 10 in S54, S56, and S57 functions as a partial bubble detection unit.

5 is repeated, all of the via holes 44, 44,... Are selected, and the process of S30 is repeated, so that the number of pixels of the partial bubbles located in the plated portions 45, 45,. Counted.
After completion of the process of S30, the CPU 10 determines whether or not all of the via holes 44, 44,... Have been selected (S31), and if there is a bubble that has not been selected (NO in S31), the process proceeds to S29. Return to.

When all of the via holes 44, 44,... Have been selected (YES in S31), the CPU 10 calculates a bubble rate [%] that is the sum of the partial bubbles and the region bubbles (S32). In the process of S32, the CPU 10 calculates a percentage obtained by dividing the sum of the number of pixels K of all the partial bubbles detected in S30 and the number of pixels of all the area bubbles detected in S20 by the number of pixels of the entire joining area. To do. The CPU 10 in S32 functions as a total bubble rate calculation unit.
As shown in FIG. 24, in the case of the binary image 67, as a result of executing the process of S32, the bubble rate 38.89 (= {25 + 8 + 9} / {12 × 9} × 100) [%] is calculated. If the presence of partial bubbles is ignored and the bubble rate is calculated using only region bubbles, the bubble rate of 30.56 [%] is calculated. That is, a more accurate bubble rate can be obtained by considering the presence of partial bubbles.

  After the processing of S32, the CPU 10 displays the bubble rate calculated in S32 and the X-ray image stored in the HDD 15 on the display unit 13 as the bubble calculation result (S33), and ends the bubble calculation processing. .

  When the area bubble does not exist (NO in S21), the CPU 10 moves the process to S33 and displays the bubble rate 0% and the X-ray image stored in the HDD 15 on the display unit 13 as the bubble calculation result. Then, the bubble calculation process is terminated. That is, in the present embodiment, when there is no region bubble, no contact bubble exists, and therefore, the bubble rate is calculated assuming that there is no partial bubble.

  When the contact bubble does not exist (NO in S23), the CPU 10 calculates the bubble rate [%] by calculating the percentage obtained by dividing the number of pixels of the region bubble detected in S20 by the number of pixels of the entire bonding region. (S34), the process proceeds to S33, the bubble rate [%] calculated in S34 and the X-ray image stored in the HDD 15 are displayed on the display unit 13 as the bubble calculation result, and the bubble calculation process is performed. finish. That is, in the present embodiment, when there is no contact bubble, the bubble rate is calculated assuming that there is no partial bubble.

The above-described bubble rate calculation device 3 includes the presence / absence of partial bubbles located in the plated portions 45, 45,... Of the heat dissipation substrate 42, and the presence / absence of region bubbles located in the joining region excluding the plated portions 45, 45,. Can be detected easily and accurately, and the accurate bubble ratio of the entire joining region can be obtained.
In addition, since the bubble having a complicated curved shape is approximated by a straight line such as a normal line or a tangent line, the calculation time can be shortened.

In the process of S51, the CPU 10 may add the largest count result to the variable K. This is equivalent to certifying the maximum number of pixels as the number of partial bubbles among the number of mask pixels surrounded by the two straight lines selected in S49. In this case, the number of pixels “7” is selected in the binary image 66 shown in FIGS.
In general, the bubbles mixed in the solder layer 43 have a shape with high roundness, such as a circular shape, an elliptical shape, and a teardrop shape. Therefore, as shown in FIG. 13, the region related to the combination of the normal line 713 and the tangent line 716 and the region related to the combination of the normal lines 723 and 724 are recognized as partial bubbles.

However, for example, immediately after the start of mass production of the semiconductor device 4, the capability of the process of soldering the semiconductor element 41 and the heat dissipation substrate 42 is not stable. Therefore, the shape of the bubbles mixed in the solder layer 43 becomes unstable and the roundness tends to be lowered. For this reason, there is a possibility that the partial bubble detected based on the minimum number of pixels as in the present embodiment is estimated to have a smaller area than actual. As a result, since the bubble rate is calculated to be smaller than the actual value, there is a possibility that a defective product is judged as a non-defective product and flows out.
Therefore, by detecting the partial bubbles based on the maximum number of pixels, it is possible to suppress the bubble rate from being calculated smaller than the actual rate, and as a result, it is possible to suppress the outflow of defective products.

  In order to automatically determine whether to detect a partial bubble using the minimum pixel value or to detect a partial bubble using the maximum pixel value, for example, the CPU 10 determines the bubble rate or flatness rate of the region bubble. And whether to use the minimum pixel value or the maximum pixel value is determined based on the calculation result. When using the area bubble flatness, the CPU 10 obtains the maximum diameter and the minimum diameter from the central pixel of the contact bubble selected in S43, for example, and the value obtained by dividing the minimum diameter by the maximum diameter (ie, the flatness ratio) is a predetermined value. If it is within the range, the minimum pixel value is used in the processing of S51.

Embodiment 2. FIG.
FIG. 25 is a flowchart showing the procedure of the bubble calculation process executed by the bubble rate calculation device 3 according to Embodiment 2 of the present invention.
The hardware configuration of the bubble rate calculation device 3 of the present embodiment is the same as the hardware configuration of the bubble rate calculation device 3 of the first embodiment, and the bubble rate calculation device 3 of the present embodiment determines whether or not there are bubbles. The semiconductor device 4 to be detected is the same as the configuration of the semiconductor device 4 of the first embodiment. For this reason, parts corresponding to those of the first embodiment are denoted by the same reference numerals, and description thereof is omitted.
25 are the same as the processes of S11 to S20 shown in FIG. 4 and S21 shown in FIG.

When the area bubble is present (YES in S21), the CPU 10 executes the processes after S61.
The CPU 10 counts the number of pixels in the bonding area excluding the plated portions 45, 45,... (S61).
In the case of the binary image 68 shown in FIG. 12, the pixel group other than the pixel groups 681, 682 corresponds to a junction area excluding the plated portions 45, 45,..., And the number of pixels in this area is “92”. In addition, the pixel groups 686, 687, and 688 correspond to area bubbles, and the number of pixels of the area bubbles is “7”.

After the processing of S61 is completed, the CPU 10 calculates a percentage by dividing the number of pixels of the region bubble detected in S20 by the number of pixels of the joining region excluding the plated portions 45, 45,... Counted in S61. [%] Is calculated (S62). In the case of the binary image 68 shown in FIG. 12, the bubble rate is 7.61%.
Finally, the CPU 10 displays the bubble rate calculated in S62 and the X-ray image stored in the HDD 15 on the display unit 13 as a bubble calculation result (S63), and ends the bubble calculation process.

  When the area bubble does not exist (NO in S21), the CPU 10 moves the process to S63 and displays the bubble rate 0% and the X-ray image stored in the HDD 15 on the display unit 13 as the bubble calculation result. Then, the bubble calculation process is terminated. That is, in the present embodiment, when there is no region bubble, the bubble rate is calculated assuming that there is no partial bubble.

In the above-described bubble ratio calculation device 3, the heat dissipation substrate 42 is placed on the plated portions 45, 45,... And on the inside and outside of the via holes 44, 44,. Can be considered separately. At this time, the presence / absence of bubbles (that is, region bubbles) located in the joining region excluding the plated portions 45, 45,... Can be easily and accurately detected by binarization. Further, by applying the bubble ratio in the bonding area excluding the plated portions 45, 45,... As the bubble ratio of the plated portions 45, 45,. The ratio can be certified as the accurate bubble rate of the entire joining area.
Further, since the bubble ratio in the plated portions 45, 45,... Is compensated with the estimated value, the calculation time can be shortened.

Embodiment 3. FIG.
FIG. 26 is a flowchart showing the procedure of the bubble calculation process executed by the bubble rate calculation device 3 according to Embodiment 3 of the present invention.
The hardware configuration of the bubble rate calculation device 3 according to the present embodiment is the same as the hardware configuration of the bubble rate calculation device 3 according to the first and second embodiments. The semiconductor device 4 whose presence / absence should be detected is the same as the configuration of the semiconductor device 4 of the first and second embodiments. For this reason, parts corresponding to those in the first and second embodiments are denoted by the same reference numerals, and description thereof is omitted.
Moreover, since the process of S11-S17 shown in FIG. 26 is the same as the process of S11-S17 shown in FIG.4 and FIG.25, description is abbreviate | omitted.

After the processing of S17 is completed, the CPU 10 refers to the mask correction image and identifies the pixels of the plated portions 45, 45,... Included in the region image (S81).
In the case of the region image 60 shown in FIG. 6, the CPU 10 in S81 refers to the mask correction image 64 shown in FIG. 10, and sets the pixel group corresponding to the pixel group having the pixel value “0” of the mask correction image 64 to the plated portion 45. , 45 pixel groups.

After the process of S81 shown in FIG. 26 is completed, the CPU 10 binarizes the region image with a plurality of types of threshold values (S82). The CPU 10 in S83 functions as a different threshold binarization unit.
More specifically, the CPU 10 in S82 binarizes the pixel values of the pixels included in the plated portions 45, 45,... Specified in S81 with a threshold value P, and within the joint region excluding the plated portions 45, 45,. The pixel values of the pixels included inside the via holes 44, 44,... Are binarized with a threshold I different from the threshold P, and the pixel values of the pixels included outside the via holes 44, 44,. Binarization is performed with a threshold value O different from any of the above.

  As in the first embodiment, the threshold value I is set in advance according to the X-ray transmittance of the central portion 42a of the via hole 44, and the threshold value O is set in advance according to the X-ray transmittance of the outside 42c of the via hole 44. Is set to a smaller value. On the other hand, the threshold value P is set in advance to a value smaller than the threshold value O according to the X-ray transmittance of the inner peripheral portion 42 b of the via hole 44. That is, threshold value P <threshold value O <threshold value I.

  The threshold values I and O may be the same value. However, in order to set the threshold value O = the threshold value I, it is necessary to correct the difference in X-ray transmittance between the inside and outside of the via hole 44 on the area image before binarization.

FIG. 27 is a schematic diagram illustrating an example of a binary image obtained by the bubble rate calculation device 3.
In the figure, reference numeral 69 denotes a binary image, and the binary image 69 is obtained by binarizing the region image 60 shown in FIG. More specifically, the region image 60 is included in the first pixels included in the pixel groups corresponding to the pixel groups 641 and 642 shown in FIG. 10 (ie, in the plated portions 45 and 45 specified in S81). Pixel value) is binarized with a threshold value P = “0.30”. Further, the pixel value of the second pixel (that is, each pixel of coordinates (2, 6), (8, 3)) surrounded by the pixel groups corresponding to the pixel groups 641 and 642 shown in FIG. = “0.85” and binarized. Further, the pixel values of the pixels other than the first and second pixels are binarized with a threshold value O = “0.75”.

  In FIG. 27, each of the pixel groups 695, 696, 697, and 698 shown in white has a pixel value “1”, and the pixel group shown in fine hatching at the right has a pixel value “0”.

  When the binary image 68 shown in FIG. 12 is compared with the binary image 69 shown in FIG. 27, the binary image 69 corresponds to the pixel groups 681 and 682, which are masked pixel groups of the binary image 68, respectively. It can be seen that the pixel group includes pixels with pixel value “1” (coordinates (3, 7), (8, 4)) indicating bubbles. As described above, in the present embodiment, not only the joint region excluding the plated portions 45 and 45 but also the bubbles present in the plated portions 45 and 45 are detected.

  In the region image 60 shown in FIG. 6, the pixel at the coordinates (3, 7) located in the plated portion 45 corresponds to a bubble, but if the plated portion 45 is binarized with the threshold value O or the threshold value I, this pixel Is determined not to be a bubble. This is because the plated portion 45 has low transmittance even if air bubbles are present.

After the process of S83 is completed, the CPU 10 performs the presence / absence detection of a bubble for detecting the presence / absence of a bubble located in the bonding region (hereinafter referred to as a bonding bubble) (S83). The CPU 10 in S83 functions as a bonded bubble detection unit.
Specifically, the CPU 10 determines whether the pixel value of each pixel is “1” or “0” in the binary image 69 illustrated in FIG. 27 and determines that the pixel value is “1”. The number of pixels J is counted. When the number of pixels J is 0, it is detected that no bonding bubbles are present in the bonding region, and when the number of pixels J is one or more, it is detected that bonding bubbles are present in the bonding region.

After the process of S83 is completed, the CPU 10 calculates the bubble rate [%] of the bonded bubbles (S84). In the process of S84, the CPU 10 calculates a percentage obtained by dividing the number of pixels J of the bonding bubbles detected in S83 by the number of pixels of the entire bonding area. The CPU 10 in S84 functions as a bonding bubble rate calculation unit.
In the case of the binary image 69 shown in FIG. 27, the number of pixels J is nine in the pixel groups 695, 696, 697, and 698, and the total number of pixels in the junction region is 108 (= 12 × 9). Therefore, the bubble rate is 8.33%.
After the process of S84, the CPU 10 displays the bubble rate calculated in S84 and the X-ray image stored in the HDD 15 on the display unit 13 as the bubble calculation result (S85), and ends the bubble calculation process. .

  The above-described bubble rate calculation device 3 includes the presence / absence of partial bubbles located in the plated portions 45, 45,... Of the heat dissipation substrate 42, and the presence / absence of region bubbles located in the joining region excluding the plated portions 45, 45,. Can be detected simultaneously and easily and accurately, and the accurate bubble ratio of the entire joining region can be obtained.

When the threshold values I, O, and P are all set to the same value, the difference in X-ray transmittance between the inside and outside of the via hole 44 and the plated portion 45 is corrected on the region image before binarization. There is a need (see Embodiment 4).
Further, when the threshold value I and the threshold value P are the same value and only the threshold value O is different from these (or the threshold value O and the threshold value P are the same value and only the threshold value I is different from these), The difference in X-ray transmittance between the inside of the via hole 44 and the plated portion 45 needs to be corrected on the area image. Similarly, when the threshold value O and the threshold value P are the same, and only the threshold value I is different from these, the difference in the X-ray transmittance between the outside of the via hole 44 and the plated portion 45 is determined as a region image. It is necessary to correct above.

Embodiment 4 FIG.
FIG. 28 is a flowchart showing the procedure of the bubble detection process executed by the bubble rate calculation device 3 according to Embodiment 4 of the present invention.
The hardware configuration of the bubble rate calculating device 3 of the present embodiment is the same as the hardware configuration of the bubble rate calculating device 3 of the first to third embodiments, and the bubble rate calculating device 3 of the present embodiment is The semiconductor device 4 whose presence or absence is to be detected has the same configuration as that of the semiconductor device 4 of the first to third embodiments. For this reason, the part corresponding to Embodiment 1-3 is attached | subjected with the same code | symbol, and those description is abbreviate | omitted.
Also, the processing of S11 and S12 shown in FIG. 28 is the same as the processing of S11 and S12 shown in FIG.

FIG. 29 is a schematic diagram illustrating an example of an X-ray image captured by the bubble rate calculation device 3.
In the figure, reference numeral 810 denotes an X-ray image. The X-ray image 810 is input from the imaging apparatus 2 and stored in the HDD 15 in the process of S12. Each pixel of the X-ray image 810 has a multi-value pixel value of “0.00” or more and “1.00” or less. In the figure, each square of 20 squares and 25 squares (500 squares) is an individual pixel included in the X-ray image 810.

Each of the pixels 811, 811,..., Shown by rough hatching in the lower right in FIG. 29 has a pixel value P ₁ (= 0.20). In addition, each of the pixels 812, 812,... Shown in white has a pixel value P ₂ (= 0.80). Further, each of the pixels 813, 813,..., Indicated by the rough rising hatching, has a pixel value P ₃ (= 0.40). Each of the pixels 814, 814,... Indicated by the intersecting rough hatching has a pixel value P ₄ (= 0.70). Furthermore, each of the pixels 815, 815,... Shown in white has a pixel value P ₅ (= 1.00). Each of the pixels 815, 815,... Having the pixel value P ₅ clearly corresponds to a portion where the heat dissipation substrate 42 (ie, the semiconductor device 4) does not exist.

Each of the pixels 816, 816,... Indicated by fine hatching with a right lowering has a pixel value P ₆ (= 0.32). In addition, each of the pixels 817, 817,..., Indicated by fine hatching that rises to the right, has a pixel value P ₇ (= 0.16). Each of the pixels 818, 818,... Shown by the fine hatching that intersects has a pixel value P ₈ (= 0.08).
Each of the pixels 811 to 813 is an example of a pixel indicating bubbles mixed in the solder layer 43, and each of the pixels 816 to 818 is an example of a pixel indicating the solder layer 43.

After the process of S12 shown in FIG. 28 is completed, the CPU 10 reads a coefficient image, which will be described later, from the HDD 15 (S71) and develops it in the RAM 12.
A coefficient image is stored in the HDD 15 shown in FIG. 2 as data necessary for executing the bubble calculation method of the present embodiment.

The coefficient image is a multi-value image, and the via design diagram data and pattern diagram data of the first embodiment, and the X-ray transmission table showing the X-ray transmittance through which the X-rays are transmitted for the material of each part of the heat dissipation substrate 42, The heat dissipation substrate 42 to which the semiconductor element 41 is not bonded is generated using each pixel value of a multivalued X-ray image obtained by imaging with the imaging apparatus 2 in the same manner as when the semiconductor device 4 is X-rayed. It is a thing. Note that X-ray absorption may be used instead of X-ray transmittance.
Each pixel value of the coefficient image is a correction coefficient for removing the influence of the material and thickness of each part constituting the heat dissipation substrate 42 from the X-ray image.

The bubble detection method of the present embodiment is a region that indicates a bonding region between the semiconductor element 41 and the heat dissipation substrate 42 based on the material and thickness of each part constituting the heat dissipation substrate 42 and the X-ray image of the heat dissipation substrate 42 itself. Correct the image. As a result, the correction is performed more accurately than the conventional bubble detection method that corrects the region image using an approximate curve that takes into account only the thickness of the heat dissipation substrate 42.
In the following, in order to simplify the description, a case where the bubble transmittance is 100% in the bonding region and the transmittance of the solder layer 43 is 40% is illustrated.

FIG. 30 is a schematic diagram illustrating an example of a coefficient image used in the bubble rate calculation device 3.
In the figure, reference numeral 820 denotes a coefficient image, and the coefficient image 820 is stored in advance in the HDD 15 of the arithmetic device 1.
In the figure, each square of 20 squares and 25 squares (500 squares) is an individual pixel included in the coefficient image 820, and each pixel is not less than “0.00” and not more than “1.00”. It has multi-value pixel values. The pixel value of each pixel indicates the correction coefficient of the corresponding pixel of the X-ray image input from the imaging device 2 to the arithmetic device 1.
Hereinafter, a description will be given with reference to FIGS.

Each of the pixel groups 821, 821,... Shown by hatching in FIG. 30 is 12 pixels each having a correction coefficient R ₁ (= 0.20). Each pixel group 821 is arranged in a square shape surrounding the pixel group 822 shown in white. Each pixel group 822 includes four pixels, each having a correction coefficient R ₂ (= 0.80).
Each pixel group 821 corresponds to the inner peripheral portion 42 b of the via hole 44, and the pixel group 822 corresponds to the central portion 42 a of the via hole 44. The pixel group 821, 821,... Reflects the design radius, plating thickness, and center coordinates of the six via holes 44, 44,.

A pixel group 823 indicated by right-down hatching is 322 pixels having a correction coefficient R ₃ (= 0.40), and corresponds to the outside 42c of the via holes 44, 44,. Therefore, the outer shape of the pixel group 823 indicates the outer shape of the junction region (in other words, the solder layer 43) between the semiconductor element 41 and the heat dissipation substrate 42.
A pixel group 824 indicated by cross hatching is 27 pixels having a correction coefficient R ₄ (= 0.70), and corresponds to, for example, the metal substrate 424 that is pressure-bonded to the stacked body 423. Therefore, the outer shape of the pixel groups 823 and 824 indicates the outer shape of the heat dissipation substrate 42.

Furthermore, a pixel group 825 shown in white is 55 pixels having the correction coefficient R ₅ (= 1.00), and corresponds to a portion where the heat dissipation substrate 42 (ie, the semiconductor device 4) does not exist.
Therefore, the region coefficient image 82 corresponding to the bonding region between the semiconductor element 41 and the heat dissipation substrate 42 is a portion excluding the pixel groups 824 and 825 of the coefficient image 820.

  After the processing of S71 shown in FIG. 28, the CPU 10 reads a copy of the X-ray image stored in the HDD 15 and develops it in the RAM 12, and based on the coefficient image read in S71, an area indicating the joining area in the X-ray image An image is specified (S72).

When the X-ray image 810 and the coefficient image 820 are used, the CPU 10 first compares each pixel value of the X-ray image 810 with a predetermined threshold value (for example, “0.90”), and has a pixel value equal to or greater than the threshold value. Is obtained, the outer shape of the heat dissipation substrate 42 in the X-ray image 810 is obtained. Further, the CPU 10 based on the outer shape of the pixel group 824 indicating an area other than the joint area in the coefficient image 820 and the correction coefficient R ₄ , the pixels 814 and 814 having a pixel value equal to the correction coefficient R ₄ in the X-ray image 810. ,... Are masked to identify the region image 81.
Note that, for example, a configuration in which an area image in an X-ray image is specified using pattern diagram data may be used.

After the process of S72 shown in FIG. 28 is completed, the CPU 10 uses the outer shape of the joining area included in each of the X-ray image and the coefficient image as a reference to the origin of the xy coordinate system set in the area image and the area coefficient image. The origin of the set xy coordinate system is made coincident (S73), and each pixel value of the region image is corrected with the corresponding pixel value of the region coefficient image to generate a corrected region image (S74). The CPU 10 in S74 functions as an image correction unit.
When using the X-ray image 810 and the coefficient image 820, the CPU 10 first sets the xy coordinates set in the region image 81 with reference to the outer shape of the joining region included in the X-ray image 810 and the coefficient image 820, respectively. The origin of the system and the origin of the xy coordinate system set in the area coefficient image 82 are matched.

Further, the CPU 10 generates a corrected region image 83 having a pixel value obtained by dividing each pixel value of the region image 81 by a corresponding pixel value of the region coefficient image 82 (see FIG. 31 described later).
FIG. 31 is a schematic diagram illustrating an example of a corrected region image generated by the bubble rate calculation device 3.
In the figure, reference numeral 83 denotes a corrected area image, and the corrected area image 83 is included in the corrected X-ray image 830 obtained by correcting the area image 81 of the X-ray image 810 in S74. In the figure, each square of 20 squares and 25 squares (500 squares) is an individual pixel included in the corrected X-ray image 830.

Each pixel of the post-correction region image 83 has a multivalued pixel value of “0.00” or more and “1.00” or less. Each of the 97 pixels 831, 831,... Shown in white has a pixel value “1.00”, and each of the 321 pixels 832, 832,. 0.40 ".
On the other hand, the 82 pixels 833, 833,... Shown by the rough rising hatching are masked respectively.

In the corrected area image 83, pixels 831 corresponding to the pixel 811 in the area image 81 has a pixel value B ₁ corrected. Since the pixel 811 of the area image 81 corresponds to the pixel included in the pixel group 821 of the area coefficient image 82, the pixel value B ₁ is calculated as follows based on the pixel value P ₁ and the correction coefficient R _1. The
B ₁ = P ₁ / R ₁ = 0.20 / 0.20 = 1.00 (1)
Similarly, in the post-correction region image 83, pixel 831 corresponding to the pixel 812 in the area image 81 has a pixel value B ₂ after correction. Since the pixel 812 of the area image 81 corresponds to the pixel included in the pixel group 822 of the area coefficient image 82, the pixel value B ₂ is calculated as follows based on the pixel value P ₂ and the correction coefficient R _2. The
B ₂ = P ₂ / R ₂ = 0.80 / 0.80 = 1.00 (2)

In the corrected area image 83, pixels 831 corresponding to the pixel 813 in the area image 81 has a pixel value B ₃ after correction. Since the pixels 813 of the region image 81 correspond to the pixels included in the pixel group 823 of the region coefficient image 82, the pixel value B ₃ is calculated as follows based on the pixel value P ₃ and the correction coefficient R _3. The
B ₃ = P ₃ / R ₃ = 0.40 / 0.40 = 1.00 (3)
In the corrected area image 83, pixels 832 corresponding to the pixel 816 in the area image 81 has a pixel value B ₆ after correction. Since the pixels 816 of the region image 81 correspond to the pixels included in the pixel group 822 of the region coefficient image 82, the pixel value B ₆ is calculated as follows based on the pixel value P ₆ and the correction coefficient R _2. The
B ₆ = P ₆ / R ₂ = 0.32 / 0.80 = 0.40 (4)

In the corrected area image 83, pixels 832 corresponding to the pixel 817 in the area image 81 has a pixel value B ₇ after correction. Since the pixels 817 of the region image 81 correspond to the pixels included in the pixel group 823 of the region coefficient image 82, the pixel value B ₇ is calculated as follows based on the pixel value P ₇ and the correction coefficient R _3. The
_{B 7 = P 7 / R 3} = 0.16 / 0.40 = 0.40 (5)
In the corrected area image 83, pixels 832 corresponding to the pixel 818 in the area image 81 has a pixel value B ₈ after correction. Since the pixel 818 of the area image 81 pixels included in the pixel group 821 of domain coefficient image 82 corresponding pixel value B _8, based on the pixel values P ₁ and the correction coefficient R _1, is calculated as follows The
_{B 8 = P 8 / R 1} = 0.08 / 0.20 = 0.40 (6)

  Since the influence of the material and thickness of each part constituting the heat dissipation substrate 42 is removed from the corrected region image 83 obtained as described above, the pixel 831 having the pixel value “1.00” Regardless of whether it is located in the plating part 45, the bubble located in a joining area | region is shown.

FIG. 32 is a longitudinal sectional view showing a schematic configuration of a semiconductor device in which the bubble rate calculation device 3 should detect the presence or absence of bubbles, and shows an enlarged view of the vicinity of the solder layer 43.
In the figure, P _{1 to} P ₉ are pixel values at corresponding positions on the heat dissipation board 42, and R _{1 to} R _{3 in the} figure are correction coefficients at corresponding positions on the heat dissipation board 42.
In the boundary between the bubble A and the solder layer 43, since it is often the thickness of the solder layer 43 changes gradually, the result obtained by dividing the pixel value P ₉ by the correction factor R ₂ is "0.40" - "1. It becomes an intermediate value of 00 ”.
Accordingly, the presence / absence of bubbles is not directly determined based on the corrected region image 83. First, the corrected region image 83 is binarized, and the thickness of the solder layer 43 at the boundary between the bubble A and the solder layer 43 is first determined. The effects of change need to be removed.

After the process of S74 shown in FIG. 28, the CPU 10 binarizes the corrected region image generated in S74 to generate a binary image (S75). The CPU 10 in S75 functions as a correction binarization unit.
In the case of the post-correction area image 83, the binarization executed in the process of S75 is to binarize the post-correction area image 83 with a predetermined threshold value except for the masked pixels 833, 833,. A pixel having a pixel value equal to or greater than the threshold is set to a pixel value “1”, and a pixel having a pixel value less than the threshold is set to a pixel value “0”. For example, when the threshold value is “0.50”, the pixel values of the pixels 831, 831,... Are “1”, and the pixel values of the pixels 832, 832,.

After the process of S75 is completed, the CPU 10 performs bubble presence / absence detection for detecting the presence / absence of bubbles in the entire joining region (S76). The CPU 10 in S76 functions as bubble presence / absence detection means.
Specifically, the CPU 10 determines whether the pixel value of each pixel is “1” or “0” except for the masked pixels 833, 833,..., And the pixel value is “1”. The number J of pixels determined to be is counted. When the number of pixels J is 0, it is detected that there are no bubbles in the junction area. When the number of pixels J is 1 or more, it is detected that there are bubbles in the junction area.

After the process of S76 is completed, the CPU 10 calculates the bubble rate [%] of the entire bonding area (S77). In the process of S77, the CPU 10 calculates a percentage obtained by dividing the number of pixels J of the bubble detected in S76 by the number of pixels in the entire bonding area. The CPU 10 in S77 functions as an overall bubble rate calculation unit.
In the case of the corrected region image 83, the number of pixels J is 97, that is, the number of pixels 831, 831,..., And the number of pixels in the entire joining region is 418 (= 19 × 22). %.
After the processing of S77 is completed, the CPU 10 displays the bubble rate calculated in S77 and the X-ray image stored in the HDD 15 on the display unit 13 as a bubble calculation result (S78), and ends the bubble calculation processing. .

The bubble rate calculating device 3 as described above does not distinguish between the plated portions 45, 45,... Of the heat dissipation substrate 42 and the bonded regions excluding the plated portions 45, 45,. Presence / absence can be detected easily and accurately, and an accurate bubble ratio in the entire joining region can be obtained.
Further, since the coefficient image uses an X-ray image of the heat dissipation substrate 42, an X-ray image formed by X-ray imaging of the pixels corresponding to the plated portions 45, 45,. Is less likely to be misaligned with the pixels corresponding to the plated portions 45, 45,... Included in the image, so that the processing such as the mask image correction executed in S17 of the first embodiment is applied to the coefficient image. There is no need to execute. As a result, the calculation time can be shortened.

Embodiment 5 FIG.
FIG. 33 is a longitudinal sectional view showing a schematic configuration of a semiconductor device in which the bubble rate calculating device 3 according to Embodiment 5 of the present invention should detect the presence or absence of bubbles.
The hardware configuration of the bubble rate calculation device 3 of the present embodiment is the same as the hardware configuration of the bubble rate calculation device 3 of the first to fourth embodiments.
On the other hand, unlike the semiconductor device 4 of the first to fourth embodiments, the semiconductor device 4 to which the bubble rate calculation device 3 of the present embodiment should detect the presence or absence of bubbles is between the semiconductor element 41 and the heat dissipation substrate 42. The heat propagation substrate 47 is bonded.
In addition, the same code | symbol is attached | subjected to the part corresponding to Embodiment 1-4, and those description is abbreviate | omitted.

First, the configuration of the semiconductor device 4 will be described.
The heat propagation substrate 47 has a rectangular shape made of copper, and one surface (the lower surface in the drawing) of the heat propagation substrate 47 is soldered to one surface of the laminate 423 through a nickel plating layer and a gold plating layer (not shown). Has been. Further, one surface (lower surface in the drawing) of the semiconductor element 41 is soldered to the other surface (upper surface in the drawing) of the heat propagation substrate 47.
For this reason, the heat generated by the operating semiconductor element 41 propagates to the heat dissipation substrate 42 via the heat propagation substrate 47.

In the figure, reference numeral 431 denotes a solder layer that joins the semiconductor element 41 and the heat propagation substrate 47, and bubbles may be mixed into the solder layer 431 during soldering. In the figure, reference numeral 432 denotes a solder layer that joins the heat propagation substrate 47 and one surface of the laminate 423 (that is, one surface of the heat dissipation substrate 42). Bubbles are mixed in the solder layer 432 during soldering. There are things to do. The solder layer 431 is formed at a higher temperature than the solder layer 432.
The semiconductor device 4 is manufactured by joining the semiconductor element 41 and the heat propagation substrate 47 with solder, and then joining the heat propagation substrate 47 to which the semiconductor element 41 is joined and the heat dissipation substrate 42 with solder.

What is necessary is just to obtain | require the bubble rate of the bubble mixed in the solder layer 431 with the conventional bubble rate calculation method.
That is, the user of the bubble rate calculation device 3 places a joined body formed by joining the semiconductor element 41 and the heat propagation substrate 47 with solder inside the imaging device 2. At this time, the heat dissipation substrate 42 is not yet joined to the joined body.
The imaging apparatus 2 captures a multi-value X-ray image by irradiating the X-ray 21 a perpendicularly to the heat propagation substrate 47 from the semiconductor element 41 side, and outputs it to the arithmetic apparatus 1.

  In the arithmetic device 1, the CPU 10 specifies a region image in the input multi-value X-ray image, generates a binary image by binarizing the specified region image with a predetermined threshold value, and generates a pixel value “1”. The percentage obtained by dividing the number of pixels by the pixel value of the entire area image (that is, the bubble ratio of bubbles mixed in the solder layer 431) is calculated. Hereinafter, this binary image is referred to as a bonded binary image.

When obtaining the bubble rate of bubbles mixed in the solder layer 432, the user of the bubble rate calculation device 3 places the semiconductor device 4 in which the heat dissipation substrate 42 is bonded to the bonded body inside the imaging device 2.
The imaging device 2 shoots a multi-value X-ray image by irradiating the X-ray 21 a perpendicularly to the heat dissipation substrate 42 from the semiconductor element 41 side, and outputs it to the computing device 1.

In the arithmetic device 1, the CPU 10 specifies a region image in the input multi-valued X-ray image, and masks the specified region image based on the bonded binary image so as to be mixed into the solder layer 431. Mask the pixels corresponding to the air bubbles. More specifically, the CPU 10 masks the pixels of the area image corresponding to the pixel having the pixel value “1” in the bonded binary image.
After that, by using the masked region image and applying the bubble rate calculation processing of Embodiments 1 to 4 to the unmasked pixels, the bubble rate of bubbles mixed in the solder layer 432 is reduced. Desired.

Specifically, when the bubble rate calculation processing according to the first to third embodiments is applied, the region image specified in S13 is displayed between S13 and S14 illustrated in FIGS. 4, 25, and 26. A process of masking based on the binary image is added. When the bubble rate calculation process according to the fourth embodiment is applied, a process for masking the region image specified in S72 based on the bonded binary image is added between S72 and S73 shown in FIG.
The bubble calculating device 3 as described above has the same effects as those of the first to fourth embodiments.

Embodiment 6 FIG.
The hardware configuration of the bubble rate calculation device 3 according to the present embodiment is the same as the hardware configuration of the bubble rate calculation device 3 according to the first to fifth embodiments. The semiconductor device 4 whose presence or absence is to be detected has the same configuration as that of the semiconductor device 4 of the fifth embodiment.
In addition, the same code | symbol is attached | subjected to the part corresponding to Embodiment 1-5, and those description is abbreviate | omitted.

The bubble ratio of bubbles mixed in the solder layer 431 of the semiconductor device 4 may be obtained by a conventional bubble ratio calculation method as in the fifth embodiment.
That is, the user of the bubble rate calculation device 3 places a joined body formed by joining the semiconductor element 41 and the heat propagation substrate 47 with solder inside the imaging device 2. At this time, the heat dissipation substrate 42 is not yet joined to the joined body.
The imaging apparatus 2 captures a multi-value X-ray image by irradiating the X-ray 21 a perpendicularly to the heat propagation substrate 47 from the semiconductor element 41 side, and outputs it to the arithmetic apparatus 1. Hereinafter, this X-ray image is referred to as a joined body X-ray image.

  In the arithmetic device 1, the CPU 10 specifies a region image in the input multi-valued joined body X-ray image, generates a joined body binary image by binarizing the identified region image with a predetermined threshold, A percentage obtained by dividing the number of pixels having the pixel value “1” by the pixel value of the entire region image (that is, the bubble ratio of bubbles mixed in the solder layer 431) is calculated.

When obtaining the bubble rate of bubbles mixed in the solder layer 432, the user of the bubble rate calculation device 3 places the semiconductor device 4 in which the heat dissipation substrate 42 is bonded to the bonded body inside the imaging device 2.
The imaging device 2 shoots a multi-value X-ray image by irradiating the X-ray 21 a perpendicularly to the heat dissipation substrate 42 from the semiconductor element 41 side, and outputs it to the computing device 1.

The arithmetic device 1 generates a bubble correction image based on the joined body X-ray image. The bubble correction image is a multi-valued image, and each pixel value of the bubble correction image is a correction coefficient for removing the influence of the presence or absence of bubbles in the solder layer 431 from the X-ray image.
The CPU 10 specifies a region image based on the coefficient image in the input multi-value X-ray image, and corrects the specified region image based on the coefficient image and the bubble correction coefficient.
Thereafter, the bubble rate is obtained by applying the bubble rate calculation process of the fourth embodiment using the corrected region image.

That is, before S11 shown in FIG. 28, a process of generating a bubble correction image based on the joined body X-ray image is added, and the process of S74 converts each pixel value of the area image specified in S72 to the coefficient image and The process is changed to a process of generating a corrected area image by correcting with the corresponding pixel value of each of the bubble correction images.
The bubble calculation device 3 as described above has the same effect as that of the fourth embodiment.

The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is not intended to include the above-described meanings, but is intended to include meanings equivalent to the claims and all modifications within the scope of the claims.
In addition, as long as the effect of the present invention is obtained, the bubble rate calculation device 3 may include components that are not disclosed in the first to sixth embodiments.

It is a perspective view which shows the external appearance of the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a block diagram which shows the principal part structure of the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a longitudinal cross-sectional view which shows the typical structure of the semiconductor device which the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention should detect the presence or absence of a bubble. It is a flowchart which shows the procedure of the bubble calculation process performed with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows the procedure of the bubble calculation process performed with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the X-ray image image | photographed with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the area | region binary image calculated | required with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the mask image used with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the subtraction image calculated | required with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the mask correction image calculated | required with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the masking image calculated | required with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows an example of the binary image calculated | required by binarizing the masking image shown in FIG. It is a schematic diagram which shows an example of the bubble ranging from the outer side or inner side of a via hole to the plating part among the bubbles which the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention should detect. It is a schematic diagram which shows an example of the bubble over the inside from the outer side of a via hole among the bubbles which the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention should detect. It is a schematic diagram which shows an example of the binary image produced | generated with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows another example of the binary image produced | generated with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows the subroutine of the partial bubble detection process procedure performed with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows the subroutine of the partial bubble detection process procedure performed with the bubble rate calculating apparatus which concerns on Embodiment 1 of this invention. It is a schematic diagram which shows the area | region enclosed with two normal lines in the binary image shown in FIG. It is a schematic diagram which shows the area | region enclosed with the one normal line and tangent in the binary image shown in FIG. It is a schematic diagram which shows the area | region enclosed with the other normal and tangent in the binary image shown in FIG. It is a schematic diagram which shows the area | region enclosed with two tangents in the binary image shown in FIG. It is a schematic diagram which shows four straight lines in the binary image shown in FIG. It is a schematic diagram which shows the area | region enclosed with two straight lines in the binary image shown in FIG. It is a flowchart which shows the procedure of the bubble calculation process performed with the bubble rate calculating apparatus which concerns on Embodiment 2 of this invention. It is a flowchart which shows the procedure of the bubble calculation process performed with the bubble rate calculating apparatus which concerns on Embodiment 3 of this invention. It is a schematic diagram which shows an example of the binary image calculated | required with the bubble rate calculating apparatus which concerns on Embodiment 3 of this invention. It is a flowchart which shows the procedure of the bubble calculation process performed with the bubble rate calculating apparatus which concerns on Embodiment 4 of this invention. It is a schematic diagram which shows an example of the X-ray image image | photographed with the bubble rate calculating apparatus which concerns on Embodiment 4 of this invention. It is a schematic diagram which shows an example of the coefficient image used with the bubble rate calculating apparatus which concerns on Embodiment 4 of this invention. It is a schematic diagram which shows an example of the area | region image after correction | amendment produced | generated with the bubble rate calculating apparatus which concerns on Embodiment 4 of this invention. It is a longitudinal cross-sectional view which shows the typical structure of the semiconductor device which the bubble rate calculating apparatus which concerns on Embodiment 4 of this invention should detect the presence or absence of a bubble. It is a longitudinal cross-sectional view which shows the typical structure of the semiconductor device which the bubble rate calculating apparatus which concerns on Embodiment 5 of this invention should detect the presence or absence of a bubble.

Explanation of symbols

  10 CPU, 21a X-ray, 3 bubble rate calculation device, 4 semiconductor device, 41 semiconductor element, 42 heat dissipation substrate, 44 via hole (hole)

Claims (4)

A multi-valued image obtained by photographing a bonding region of a semiconductor device formed by bonding a semiconductor element and one surface of a heat dissipation substrate formed with a hole plated with an inner surface by irradiating the one surface with X-rays. In the bubble rate calculation method for calculating the ratio of the bubbles present in the bonding region to the bonding region,
Masking the plated portion of the hole contained in the multi-valued image;
Generate a binary image by binarizing the masked multi-valued image,
Based on the generated binary image, the presence or absence of a region bubble located in the joining region excluding the plated portion is detected,
Detecting the presence or absence of contact bubbles in contact with the plating part among the detected region bubbles,
Based on the detected shape of the contact bubble, a virtual line indicating the feature of the shape is obtained,
Detect the presence or absence of a region that is surrounded by the obtained virtual line and is located in the plated portion and should be certified as a bubble,
A method for calculating a bubble rate, wherein a ratio of the total number of pixels of a region to be recognized as a detected bubble and the number of pixels of the detected region bubble to the number of pixels in the bonded region is calculated as a bubble rate of the entire bonded region . A multi-valued image obtained by photographing a bonding region of a semiconductor device formed by bonding a semiconductor element and one surface of a heat dissipation substrate formed with a hole plated with an inner surface by irradiating the one surface with X-rays. In the bubble rate calculation device that calculates the ratio of the bubbles present in the bonding region to the bonding region,
Masking means for masking the plated portion of the hole included in the multi-valued image;
Image binarization means for binarizing the multi-valued image masked by the masking means to generate a binary image;
Based on the binary image generated by the image binarization means, area bubble detection means for detecting the presence or absence of area bubbles located in the joining area excluding the plated portion;
A contact bubble detection means for detecting the presence or absence of a contact bubble in contact with the plated portion among the region bubbles detected by the region bubble detection means;
Based on the shape of the contact bubble detected by the contact bubble detection means, a virtual line calculation means for obtaining a virtual line indicating the feature of the shape;
Partial bubble detection means for detecting the presence or absence of a region that is surrounded by a virtual line obtained by the virtual line calculation means and is located in the plated portion and should be recognized as a bubble;
The ratio of the total number of pixels of the region to be identified as the bubbles detected by the partial bubble detection unit and the number of pixels of the region bubble detected by the region bubble detection unit to the number of pixels of the junction region A bubble rate calculating device comprising: a total bubble rate calculating means for calculating as follows. When the contact bubble detected by the contact bubble detection means is in contact with one of the outer side and the inner side of the hole at the same position in the circumferential direction of the hole of the plated portion,
The virtual line calculation means includes
A normal line of the plated portion passing through the intersection between the bubble peripheral portion of the contact bubble and the plated portion, and a tangent line of the bubble peripheral portion passing through the intersection are calculated.
The partial bubble detection means includes
The normal line and the tangent line calculated by the imaginary line calculation means detect the presence or absence of an area having the minimum number of pixels located at the plated portion, the intersection being surrounded by two straight lines existing at different positions. The bubble rate calculating device according to claim 2 , wherein: When the contact bubble detected by the contact bubble detection means is in contact with both the outside and the inside of the hole at the same position in the circumferential direction of the hole of the plated portion,
The virtual line calculation means includes
An outer intersection of the bubble periphery of the contact bubble that contacts the plating portion from the outside of the hole and the plating portion, and an inner intersection of the bubble periphery of the contact bubble that contacts the plating portion from the inside of the hole and the plating portion The straight line that passes through is calculated,
The partial bubble detection means includes
The presence / absence of a region surrounded by two straight lines intersecting outside the region surrounded by the outer intersection and the inner intersection among the straight lines calculated by the virtual line calculation means and located in the plating portion is detected. The bubble rate calculation device according to claim 2 or 3 , wherein

Images