JPS6336543A - Method and apparatus for automatic inspection of semiconductor device - Google Patents

Abstract

PURPOSE:To make it possible to change visual inspection by an operator into automatic inspection and to shorten the time of inspection, by converting images, which are taken by using a low magnification camera and a high magnification camera, into a binary coded picture, and detecting deviation in bonding the wire of an electrode pad part, the diameter of a ball, the shape of the ball, yield of separation and the like. CONSTITUTION:Images, which are taken by using a low magnification camera 1 and a high magnification camera 5, are converted into a binary coded picture by a binary coded picture device. The deviation of the mounting position of an element pellet is obtained by a pattern matching method. Thereafter, deviation in bonding the wire of the electrode pad part of the element pellet, the diameter of a ball, the shape of the ball, yield of separation and the like are detected based on the position of the center of the ball of the wire, which is bonded to the electrode pad of the element pellet, the diameter of the ball and the roundness of the ball. Furthermore, e.g., the bonding state of a mounting solder material at the rear surface of the element pellet is judged based on an image, which is obtained through an ultrasonic wave microscope including an acoustic lens 3. Images, which are obtained by using a prism 6, an optical fiber 21 and an image comera 20, are converted into a binary coded picture. The height of a wire loop is detected, and whether the result is within a specified range or not is judged.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to the appearance inspection after mounting and bonding in the assembly process of semiconductor devices, and in particular to automatic inspection of semiconductor devices that automatically performs measurement 2 judgment using image processing. This invention relates to an inspection method and an inspection device.

[Conventional technology]

Conventionally, one of the assembly processes for semiconductor devices is a mounting process in which the semiconductor element pellet is fixed to the base, and a bonding process in which a metal wire is connected to the mounted semiconductor element pellet. Afterwards, it is necessary to carry out a visual inspection to check whether the quality satisfies the predetermined standards.Normally, there are about 15 inspection items in these process inspections. It is necessary to inspect each time.

In order to perform such inspections, conventional methods have been adopted in which an inspector visually observes the semiconductor device using a stereomicroscope or metallurgical microscope, and then judges whether the quality is good or bad based on the observation results. There is.

The inspection results obtained in this way are used when readjusting various setting conditions of the mounting and bonding process of the semiconductor device, thereby improving the quality of the semiconductor devices processed thereafter.

[Problem that the invention seeks to solve]

In the conventional inspection method described above, approximately 15 items are inspected visually by the inspector, which requires individual skill on the part of the inspector and increases eye fatigue for the inspector.Moreover, the parts to be observed are semiconductors. As devices become more highly integrated and the number of pins increases, there is a problem in that the testing time required for one semiconductor device becomes longer and testing errors are more likely to occur. Furthermore, since the visual inspection by the inspector depends on his/her senses, it is difficult to perform a quantitative inspection.

Furthermore, as mentioned above, the processing conditions are readjusted based on the test results, but since it takes a long time for the tester to perform the test, the time required for this test is A large number of semiconductor devices that do not meet the specifications are manufactured over time, resulting in a decrease in manufacturing yield.

[Means for solving problems]

The automatic inspection method for semiconductor devices of the present invention converts a screen imaged using a low-magnification camera and a high-magnification camera into a binarized image using a binarization image device, and performs a pattern matching method to detect mounting positional deviations of element pellets. After determining the ball center position of the wire bonded to the electrode pad of the element pellet, the ball diameter and the roundness of the pole, the bonding deviation of the wire at the electrode pad of the element pellet and the ball diameter are determined.

This method detects the shape of the ball and the occurrence of peeling.

The automatic inspection device for semiconductor devices of the present invention also includes a low-magnification camera that images the entire mounted device pellet, and a high-magnification camera movable in the x and y directions that images the electrode pads of the bonded device pellet. , a binarized image part that displays only the element pellet or only the wire bonded to the electrode pad of the element pellet as white or black information from the screen of a low-magnification camera or a high-magnification camera, and binarization from the binarized image part. An image processing unit that detects misalignment of the mounting position of the element pellet, misalignment of the wire bonded to the electrode pad of the element pellet, ball diameter, ball shape, peeling, etc. from the image, and a lead frame or lead frame to which the element pellet is mounted or bonded. It has a conveyance section that moves the pan cage.

〔Example〕

Next, the present invention will be explained with reference to the drawings.

FIG. 1 is a diagram showing the overall configuration of an apparatus according to an embodiment of the present invention. The low magnification camera 1 is arranged so as to image the entire semiconductor element pellet P mounted on a lead frame (or package) L that is transported by a transport section (not shown). The potting section 2 is installed inside the device so as to apply liquid to the back surface of the lead frame. An acoustic lens 3, which is a part of the ultrasonic microscope (not shown), is provided inside the device with an X
, Y, and Z stage 11. Furthermore, a drying section 4 for removing the applied liquid is also installed inside the device.

In addition, the high magnification camera 5 is mounted on the X and Y stages 12 so as to image the electrode pads and lead portions of each element pellet.
It is installed on.

Then, the prism 6 is moved in X and Y directions so as to image the loop shape of the wire of the electrode pad of the element pellet.

X, Y stage 13 and Z, θ moved in Z, θ directions
Mounted on the stage 14, an image camera 20 receives the image obtained from the prism 6 through an optical fiber 21, and electrically creates an image from it. Monitors 7 to 1O are the low magnification cameras and high magnification cameras.

Display images obtained from the acoustic lens and prism. 2
2 is a supply magazine elevator, and 23 is a storage magazine elevator.

This inspection device having the above-described general configuration is composed of the following inspection item sections. That is, (1) mount position device displacement inspection section, (2) mount wettability inspection section, (3) element pellet back wettability inspection section, (4) first bonding inspection section,
(5) a second bonding inspection section and (6) a bonding loop inspection section.

Each inspection item will be explained below.

(1) Mount misalignment inspection unit The mount misalignment inspection unit is composed of the low-magnification camera 1, a circuit that electrically holds a screen on which the element pellet is imaged, and an image processing unit that calculates inspection results from the held screen.

The mount misalignment inspection method uses a screen captured by a low-magnification camera 1 of the element pellet P, and converts this into a binary image whose level is set in advance so that the element pellet is displayed as white information. , the element pellet is projected with white information. Furthermore, this is compared with the element pellet image at the optimal mounting position, which is stored in advance, and the image processing unit performs pattern matching by calculating the amount of deviation of the damaged image and confirming that the amount of deviation of the mount position is within a specified range. Determine whether or not.

As an image processing method other than the pattern matching method, as shown in FIG. A method may be used in which the amount of displacement of the mounting position of the element pellet is calculated from the number of pixels of white information indicating the pellet portion and their positions.

(2) Mount wettability test section The mount wettability test section is composed of the low magnification camera 1, a circuit that holds the screen thereof, and an image processing section that calculates the test results.

The image processing method for performing the mount wettability test is to first align the position of the element pellet P on the screen obtained from the low-magnification camera 1 using a pattern matching method or image processing as shown in FIG. put out Then, a region S2 for inspecting the wettability of the brazing material is selected along the detected element pellet P with an interval of one pixel or more on the outer periphery of the element pellet P, as shown in FIG. Then, the level is set in advance so that the solder metal of the mount is displayed as white information, and the image is converted into a binary image, and the solder metal of the mount is displayed as white information.

As shown in Figure 3, if the region S2 where the wettability of the brazing filler metal placed along the element pellet is inspected is white information, the wettability is considered to be good, and conversely, if the black information is above a certain percentage, the wettability is good. judge it to be bad.

(3) Element pellet back surface wettability inspection unit ultrasonic microscope device, acoustic lens 3 of the ultrasonic microscope that improves ultrasonic conductivity, lead frame L3 [potting unit 2 that applies liquid such as pure water to the surface , an XYX stage 11 that moves the acoustic lens 3 in the X, Y, and Z directions; a drying section 4 that removes the applied liquid after inspection; and a It is composed of an image processing unit that determines the bonding state and determines whether mountability is acceptable or not.

The method for inspecting the back surface of a device pellet will first be explained using the case of potting on the surface of the device pellet 2 shown in FIG. 4 and the case of potting on the back surface of a lead frame or the like shown in FIG.

First, a liquid such as pure water is applied from the potting section 2.

Then, the acoustic lens 3 is brought into contact with the droplet-shaped liquid, and ultrasonic vibrations are generated by a piezoelectric element (not shown) and transmitted to the element pellet. The acoustic lens is moved in the Z direction by the stage 11 to a position where it can collect the reflected wave reflected from the back surface of the element pellet, and brought close to the element pellet P by a certain distance, and then driven in the X and Y directions by the stage 11. Scan parallel to the plane of the element pellet to take an image of the entire back surface of the element pellet. Ultrasonic microscopes take advantage of the fact that the properties of reflected waves of ultrasonic waves differ due to differences in the elastic properties of materials, converting the state of the reflected waves into electrical signal characteristics and displaying them on a screen. From this ultrasonic microscope image, it is easy to display only the mounting brazing material on the back side of the element pellet as white information like a binary image by processing the electrical signals obtained from the reflected waves of the ultrasonic waves. .

In advance, the position of the element pellet is detected by the pattern matching method or other method as described in the mount position deviation inspection section, and a portion corresponding to the back side of the element pellet is detected on the screen obtained by the ultrasonic microscope. The wettability of the back surface of the element pellet is determined based on the detected position of the element pellet, and if that part is white information corresponding to the brazing material of the mount or is larger than a certain area, the wettability of the back surface of the element pellet is passed.

After performing the above inspection, the potting area is removed by blowing warm dry gas as shown in the drying area 4 in Figure 4, or by pressing a sponge-like material that absorbs liquid (not shown). Remove the liquid applied in step 2.

(4) The high-magnification camera 5 is large enough to photograph the entire electrode pad of the first bonding inspection part element pellet, and the high-magnification camera is X-shaped so as to image each electrode pad to be inspected.

An XY stage 12 that moves in the Y direction and a magnification camera 5 project a specific part of the element pellet to accurately detect the position of the element pellet (P) and calculate inspection results from the screen showing each electrode pad. It consists of an image processing section and an image processing section.

Here, the positional deviation, ball diameter, and ball shape when the wire is bonded to the electrode pad of the element pellet using the ball bonding method are inspected.

First, the high magnification camera 5 is moved to image a specific portion of the element pellet P, and the exact position of the element pellet is calculated by a technique such as a pattern matching method. The high magnification camera 5 is moved from the calculated position of the element pellet to the electrode pad position to which the wire W is bonded, and the sixth
You will get a screen like the one shown. The level is set in advance so that the wire is displayed as white information, and the screen obtained from the high magnification camera is converted into a binary image, and the wire W portion bonded to the electrode pad is displayed as white information.

Eight arbitrary points of the wire ball WB portion displayed as white information are scanned and determined. From point S, scanning is performed, for example, in the X direction, and outer edge points A and B of the wire ball on the boundary between white information and black information are detected. Scanning is performed from point B in the vertical direction of line segment AB to find point C on the outer edge of the ball, which is on the boundary between white information and black information. A crushed wire ball can be considered circular.
Therefore, since the ball diameter is geometrically the same as the distance between ACs, it can be obtained by calculating the distance between ACs from the number of pixels between ACs and the resolution of the screen. Since the center position of the ball can be regarded as the midpoint between AC, the position of the midpoint between line segments AC on the screen and the X and Y stages 12 for moving the high magnification camera are determined.
Calculated from the amount of movement. Since the position of the element pellet P has been detected in advance, the center position of the electrode pad is calculated.

The amount of bonding position deviation is calculated by comparing the ball center position obtained in this way with the pad center position. Regarding the ball shape, the ball diameter is calculated at several locations by changing the scanning direction from point S using a binary image in which the wire portion is shown with white information. Then, the roundness is calculated from the ball diameters obtained from several locations (for example, A'C'/AC from the distance of line segment A'C' as shown in Fig. 7 and the distance of line segment AC in Fig. 6). Calculate the value of , and if it is within the specified value, it will be passed.

If the ball diameter, ball center position deviation amount, and roundness obtained above are within the specified ranges, it is determined to be passed.

However, if the electrode pad that is supposed to be wire-bonded is captured by a high-magnification camera and converted into a binary image, and there is no white information indicating the wire portion, it is determined that the wire has separated from the electrode pad. Using this method, all electrode pads are inspected for misalignment, ball diameter, ball shape, and peeling.

(5) Second bonding inspection section The high magnification camera 5 is capable of showing the entire wire-bonded lead, and the XY camera is moved in the X and Y directions so as to image each lead to be inspected. It consists of a stage 12 and an image processing section that images a specific part of the lead frame or package with a high-magnification camera, accurately detects the position of the lead frame L, and calculates the inspection result from the screen on which each lead is imaged.

Here, the second bonding position shift, crushed shape, and peeling state in the bonding state of the leads and wires of the lead frame or package are inspected.

First, the high magnification camera 5 is moved to image a specific portion of the lead frame L or package, and the exact position of the lead frame or package L is detected by a technique such as a pattern matching method. Then, when bonding is performed with a bongoo, use the recorded position data of the second bonding or the bonding position data of the lead at the designed value to adjust the height so that the center of the screen of the high-magnification camera is located at the center of the capillary trace. Magnification camera X, Y stage 1
2 to move it. Observation areas SSI~8 are arranged on the obtained screen as shown in FIG.

This observation area SSI~8 is arranged and the size of the area is changed from before so that if the bonding state passes the standard, observation area SSI~5 is located in the lead part, and observation areas S36~8 are located in the wire part. I'll decide. First, a screen obtained by a high-magnification camera is converted into a binary image whose level is set in advance so that the wire portion is displayed as white information, and the wire is displayed as white information. In observation areas SSI~8, remember whether each area contains white information or black information (if white information and black information are mixed in the area, if the white information exceeds a certain ratio, the area is determined to be white information).

Next, set the level of the screen of the same high-magnification camera in advance so that the lead part is displayed as white information.
Convert it into a valued image and display the lead as white information. It is stored whether each area in the observation areas SSI to SSI8 is white information or black information.

If the bonding information passes the standard, the observation area SS
I to 5 are white information in a binary image in which the leads are shown as white information, and observation areas SS6 to SS8 are black information because they correspond to wire portions. However, observation area S
If there are scratches on the lead side at positions S6 to S8, the observation area S
Since S6 to S8 are black information, a binarized image in which the wire is shown as white information is used for the purpose of accurate inspection.

'In the binarized image in which the wire is shown as white information, the observation areas SSI-5 are black information, and the observation areas S86-9 are white information if the wire is bonded. However, when the width of the wire collapse is small, the observation area SS6 or SS8 becomes black information on the binarized screen where the wire becomes white information, as shown in FIG.

At this time, if the observation area SS7 is also black information, it means that there is no wire portion in the screen, and it is determined that the wire has peeled off from the lead. If the observation area SS4 or SS2 is white information in the same binarized image, this indicates that the width of the wire collapse is large, as shown in FIG. The collapsed shape of the second bonding and the occurrence of peeling are determined based on whether observation areas SS2.4, 6, and 7.8 are white information or black information in the binarized image in which the wire is displayed as white information. .

Next, referring to FIG. 11, a case will be described in which a bonding position shift occurs and bonding is performed at a location other than the wire lead portion. In the binarized image in which the lead is displayed as white information, one or more of the observation areas SS 1° 2.4.5 is not on the lead, so it should be black information. Therefore, if the observation areas SS1.2 and SS4.5 are not white information in the binarized image displayed with the lead as white information, it is determined that a bonding position shift has occurred. Repeat the same inspection method for all leads.

(6) Bonding loop inspection section The prism 6 having a special shape, the optical fiber 21 bonded to this prism 6, the image receiving section that creates a screen from the optical fiber 21, and the prism in x, y, z, and θ directions. an XY stage 13 and a Zθ drive unit 14 that move to
It is composed of an image processing section that calculates inspection results from the image obtained from the prism 6.

Here, the bonding loop height is inspected to determine whether the wire bonded to the electrode pad of the element pellet is within a specified height range from the surface of the element pellet.

The prism used is prism 6 as shown in Figure 12.
The tip of the fiber is shaved at a required angle, and the shape is such that the light incident from the side is totally reflected and proceeds to the optical fiber joint surface 6a, and the part 6b where the light enters to obtain the required screen on the side. The rest is coated with a light-shielding coating 24.

First, the prism 6 is moved between the XY stage 13 and the Zθ drive unit 14.
The element pellet is moved in the X, Y, and θ directions, and further moved in the Z-axis direction to a certain height toward the element pellet surface, and a specific portion of the element pellet is captured as an image from the prism side surface 6b. The position of the element pellet I-P is calculated using a method such as a pattern matching method from image data stored in advance.

Next, the center position of the wire-bonded electrode pad of the element pellet P is calculated, and the cross line CLx in the image from the side of the prism is calculated.

The prism 6 is moved in the X, Y, and θ directions so that the center of CLY coincides with the center of the previous electrode pad.

Then, the screen obtained from the prism is converted into a binarized image whose level is set in advance so that the wire portion is displayed as white information, and the screen obtained from the prism is converted into a binarized image in which the wire portion is displayed as white information. Create a screen like the one shown. Then, cross line CLx of the screen in the Y direction on the screen
, CLy starts scanning from the center point p, and detects point g on the boundary between white information and black information. Then, the distance between the two points pg on the screen is determined from the number of pixels and the resolution of the screen.

The actual height of the wire loop from the pellet surface is determined by an arithmetic expression determined from the distance between pg and the geometrical relationship. The bonding loop height is inspected to see if the loop height of the wire thus obtained is within a specified range. As an inspection method using other image processing methods, in a binary image where the wire is shown as white information, if the wire loop height is within a specified height range, the area where the highest point of the wire loop should be shown is The observation area 5SIO is set at the front side, and the above-mentioned inspection point g is the observation area 5SI.
There is a method of inspecting the bonding loop height depending on whether the bonding loop is located within O.

Once all the electrode pads of the wire-bonded element pellet P have been inspected using the image processing method described above, the prism is moved away from the element pellet surface in the Z-axis direction, and the element pellet is conveyed. This is to prevent the bonded wire from interfering with the prism and being cut when it moves through the system.

Next, in the automatic inspection method for each inspection item described above,
A system for performing inspection processing will be described with reference to FIG. 15, since the same screen can be used as image data for multiple inspection items.

Movement of each camera and image input, control of potting section,
An image capturing/transporting control unit 31 moves the acoustic lens of the ultrasonic microscope, inputs images from the ultrasonic microscope, moves the special prism described above, inputs images 9, and controls the transport unit (not shown). The data input control unit 32 instructs that the screen data obtained from the image capture/transport control unit 31 be numbered to each element being inspected so that each screen data corresponds to each element. The image data is stored in a specific part of the image memory unit 33. Then, the inspection board control unit 35 selects from among the image data stored and held in the image memory unit 33 by the data output control unit 34.
Outputs the necessary image data in response to requests from. In order to shorten the inspection processing time, the inspection board control section 35 outputs image data to each inspection board of the image processing section 36, which is composed of a plurality of inspection boards for each inspection item, and performs processing in parallel. The test results are inputted from the combined test board, the test results are totaled for each element, and the data is output to the central control section 37. At the same time, the inspection board control section requests the data output control section 34 for the next image data for the inspection board that has output the inspection results, outputs the image data to the inspection board again, and causes the inspection board to perform image processing one after another. The central control unit 37 stores the test results of each element in the product-by-product test result data memory unit 38. Then, parameter correction data 39 is output to the mounter and bonder, and the inspection results are fed back so that the mounting and bonding conditions are optimized. In addition, the central control section 37 includes the image capturing/conveyance control section 3'1. Data person/output control unit 32
．． 34 and the inspection board control section 35, the timing of data exchange and the like is adjusted.

Here, the present invention can also be configured so that mounting and bonding are performed in one device, as shown in FIG. 16. A mounting recognition camera 41 determines the pellet adsorption position, and after mounting with a mount head 42, only a mount inspection is performed using a low magnification camera 43. A mount inspection unit (not shown) sends data to the bonder side regarding elements determined to be defective in mounting based on the mountability inspection results. Then, only for elements with good mounting, i'p f! for bonder. The camera 44 and the bonding head 45 operate to perform wire bonding, and for elements with mounting defects, wire bonding is not performed and the elements are fed by a transport system (not shown) to shorten the process time. In the figure, 46 is a supply magazine, and 47 is a storage magazine.

〔Effect of the invention〕

As explained above, the present invention provides a low magnification camera.

Mount bonding inspection can be performed using the image processing method described above on images obtained from high-magnification cameras, ultrasonic microscopes, prisms, etc., and the conventional visual inspection by operators can be shifted to automation using equipment. is possible. As a result, the mount/bonding inspection apparatus of the present invention has the effect of being able to inspect one element pellet in about 1/30th of the time required by an operator through image processing that performs high-speed arithmetic processing. Furthermore, the automatic inspection device of the present invention can quantitatively calculate the difference between the optimum mounting condition and bonding condition of the element pellet under inspection, and automatically correct the parameters of the mounter and bonder. There are effects that can be done.

[Brief explanation of drawings]

Fig. 1 is an external perspective view of the automatic mount/bonding inspection device of the present invention, Fig. 2 shows an image for detecting mount positional deviation, Fig. 3 shows an image for inspecting mount wettability, Fig. 4, and Figure 5 shows how a potting device and drying device are used to obtain an image of the back side of a device pellet using an ultrasonic microscope, and Figures 6 and 7 show the bonding positions of the wires on the electrode pads of the device. Misaligned ball diameter. FIGS. 8 to 11 are diagrams showing screens for detecting wire bonding position deviation, crushed shape, peeling, etc. of the lead part of the lead frame or package 7 cage, and FIGS. Figures 12 to 14 are diagrams showing a method of inspecting wire loop height using a prism, Figure 15 is a system configuration diagram of an embodiment of an automatic mount/bonding inspection device, and Figure 16 is only the mount inspection device. FIG. 2 is an external view of the device attached to a mount/bonding device. 1...Low magnification camera, 2...Potting section, 3...
...Acoustic lens, 4...Drying section, 5...High magnification camera, 6...Prism, 7-10...Monitor, 11,
12.13... XY stage, 14... Zθ drive section, 20... Image camera, 21... Optical fiber, 31
...Image shooting/transport control section, 32...Data input control section, 33...Memory unit, 34...Data output control section, 35...Inspection board control section, 36...Image processing Part, 37... Central control unit, 38... Data memory unit, 39... Data for parameter correction, 4
1... Mount recognition camera, 42... Mount head, 43... Low magnification camera, 44... Bonder recognition 3
Camera for 5, 45...bonding head. -' Agent Patent Attorney Akio Suzuki ゛・,)' Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 1O Figure 1I

Claims (10)

[Claims] (1) Convert the screen imaged using a low-magnification camera and a high-magnification camera into a binarized image using a binarized image unit, and determine the mounting positional deviation of the element pellet by a pattern matching method. Automatic inspection method for semiconductor devices that detects bonding deviation, ball diameter, ball shape, occurrence of peeling, etc. of wire in element pellet electrode pad part from ball center position, ball diameter, and roundness of wire bonded to pad . (2) Multiple mask-like observation areas are set up on the screen of a high-magnification camera, and the binarized image in which only the wire or lead is displayed as white information or black information can be used to detect bonding deviation, crushed shape, peeling, etc. of the wire in the lead part. Claim 1 detecting bonding appearance inspection items on the lead side
The method for automatically inspecting a semiconductor device as described in Section 1. (3) Set an area corresponding to the outer periphery of the element pellet in the screen of a low-magnification camera, and compare that area with white or black information from the binarized image in which only the brazing material of the mount is shown as white or black information. 2. The automatic testing method for a semiconductor device according to claim 1, wherein the mount wettability is tested based on whether or not the mount wettability is present. (4) Only the brazing material of the mount on the back side of the element pellet obtained using an ultrasonic microscope is projected as white or black information, and the area corresponding to the element pellet in the screen has white or black information of a certain ratio or more. 2. The method for automatically testing a semiconductor device according to claim 1, wherein the mount wettability of the back surface of a device pellet is tested based on whether or not the mount wettability of the back surface of the element pellet is tested. (5) Only the wire portion appears as white or black information in the images obtained from the prism, optical fiber, and image camera2
Scans from the center of the wire-bonded electrode pad in the value image, detects the highest point of the wire loop, detects the wire loop height from the geometric relationship, and determines whether it is within the specified range. An automatic testing method for semiconductor devices according to claim 1. (6) A low-magnification camera that images the entire mounted device pellet, a high-magnification camera that can move in the X and Y directions that images the electrode pads of the bonded device pellet, and the screens of these low-magnification and high-magnification cameras. A binarized image section that displays only the element pellet or only the wire bonded to the electrode pad of the element pellet as white or black information, and a binarized image from the binarized image device that detects the mounting positional deviation of the element pellet and the element. Misalignment of the wire bonded to the electrode pad of the pellet, ball diameter,
An automatic inspection device for semiconductor devices that has an image processing section that detects ball shape and peeling, and a transportation section that moves a lead frame or package on which element pellets are mounted or bonded. (7) A high-magnification camera movable in the X and Y directions that displays the bonded leads of the lead frame or package, and a binary image unit that displays only the wires or only the leads as white or black information from the high-magnification camera screen. 7. The semiconductor device inspection apparatus according to claim 6, further comprising an image processing section that detects the occurrence of positional deviation, crushed shape, and peeling of the wire bonded to the lead portion from the binarized image. (8) A low-magnification camera that images the entire mounted element pellet and its surroundings, a binarized image section that displays only the brazing material protruding from the screen of the low-magnification camera to the outer periphery of the mount as white or black information, and this binarization. 7. The semiconductor device inspection apparatus according to claim 6, further comprising an image processing section that inspects the wettability of the element pellet from an image. (9) an ultrasonic microscope having an acoustic lens movable in the Z, Y, and Z directions, and a potting section for applying a liquid that improves the transmission of ultrasonic waves between the acoustic lens and the back surface of the lead frame or element pellet; A drying section removes the liquid, and an image processing section uses reflected waves from the back side of the element pellet to project only the brazing material of the mount on the back side of the element pellet as white or black information on the screen and detects the bonding property of the mount on the back side of the element pellet. An automatic testing device for semiconductor devices according to claim 6. (10) a prism movable in the X, Y, Z, and θ directions;
An optical fiber connected to the prism, an image camera that displays the light sent from the prism through the optical fiber as an image, and a binarized image unit that displays only the wire portion as white or black information from the obtained image. 7. The automatic inspection apparatus for semiconductor devices according to claim 6, further comprising an image processing section that calculates the bonding leave height of the wire from the binarized image.