Method and device for determining quality of semiconductor chips
The semiconductor chip quality determination device efficiently and accurately inspects the end faces of semiconductor chips by moving diagonally above the chips, extracting data on evaluation lines, and using an epi-illuminated microscope to detect defects, ensuring high inspection efficiency and reliability.
Patent Information
- Application Number
- JP2023539480
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-08-05
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2041-08-05
AI Technical Summary
Existing methods are inefficient and unreliable in determining the quality of the end faces of semiconductor chips, particularly for those arranged in a matrix, due to their small size and the inability to accurately inspect for scratches or dust.
A semiconductor chip quality determination device that moves diagonally above the chips to acquire images, extracts data on continuous evaluation lines, identifies optimal images based on a standard, and determines quality using an epi-illuminated microscope with a specific inclination angle and imaging pitch to ensure focus and accuracy.
Enables efficient and highly accurate quality inspection of semiconductor chip end faces by capturing focused images and reliably detecting defects such as scratches or dust.
Smart Images

Figure 0007588439000001 
Figure 0007588439000002 
Figure 0007588439000003
Abstract
Description
[Technical field]
[0001] The present invention relates to a method and an apparatus for quickly and reliably determining the quality of chip end faces in the manufacturing process of semiconductor chips such as semiconductor lasers LDs (Laser Diodes). [Background technology]
[0002] An LD chip generally has a double heterostructure in which an active layer that emits light at a laser emission wavelength is sandwiched between a P-type cladding layer and an N-type cladding layer.
[0003] When a forward voltage is applied between the P-type and N-type layers, both end faces of the active layer function as reflectors, causing the laser light to go back and forth within the active layer while being amplified, resulting in stimulated emission. Edge-emitting lasers can be broadly categorized based on their reflection structure into Fabry-Perot lasers, which use semiconductor cleavage planes as reflectors, distributed feedback lasers (DFB lasers) that form a diffraction grating in the waveguide, and distributed Bragg reflector lasers (DBR lasers) that form diffraction gratings before and after the active region.
[0004] For a semiconductor chip having such a configuration, for example, in the final manufacturing stage, it becomes necessary to determine whether or not the stimulated emission aperture formed on the end face of the chip has any scratches or dust. [Prior art documents] [Patent documents]
[0005] [Patent Document 1] JP 2017-207356 A [Patent Document 2] JP 2002-296203 A [Patent Document 3] Japanese Patent Application Publication No. 10-040387 [Patent Document 4] Japanese Patent Application Publication No. 10-002725
[0006] Regarding quality determination devices, various inventions including Patent Documents 1 to 4 are known, but no method is known for efficiently determining the quality of the end faces of many semiconductor chips arranged vertically and horizontally. That is, the above inventions mainly target the chip plane, and are not configured to reliably determine the quality of the end faces of today's increasingly miniaturized semiconductor chips. Summary of the Invention [Problem to be solved by the invention]
[0007] SUMMARY OF THE PRESENT DISCLOSURE The present invention has been made in view of the above problems, and has as its object to provide a method and apparatus for quickly and reliably determining the pass / fail state of an end face of a semiconductor chip. [Means for solving the problem]
[0008] In order to achieve the above-mentioned object, the semiconductor chip quality determination device of the present invention is configured to include an imaging device that moves back and forth from diagonally above a semiconductor chip placed on a light-reflective mounting surface to obtain images of the end faces of the semiconductor chip, and a determination device that determines the quality of the semiconductor chip based on the images obtained by the imaging device, and the determination device is configured to include a first means for extracting data on an evaluation line that continues inside and outside the semiconductor chip from a plurality of images obtained during the movement of the imaging device, a second means for evaluating the manner in which the data on the evaluation line changes for the plurality of images and identifying an optimal image from the plurality of images, and a third means for determining the optimal image identified by the second means based on a standard image obtained in advance, and determining the quality of the semiconductor chip corresponding to the image.
[0009] In addition, the method for determining whether a semiconductor chip is good or bad according to the present invention is composed of an imaging device that moves back and forth from diagonally above a semiconductor chip placed on a light-reflective mounting surface to acquire images of the end faces of the semiconductor chip, and a judgment device that judges whether the semiconductor chip is good or bad based on the images acquired by the imaging device, and is composed of a first step of extracting data on an evaluation line that continues inside and outside the semiconductor chip from a plurality of images acquired during the movement of the imaging device, a second step of evaluating the manner in which the data on the evaluation line changes for the plurality of images and identifying an optimal image from the plurality of images, and a third step of judging the optimal image identified in the second step based on a standard image acquired in advance and judging the quality of the semiconductor chip corresponding to the image.
[0010] According to the present invention, since it is not necessary to hold the end surface 10a of the thin semiconductor chip 10 on a horizontal plane, efficient and highly accurate quality inspection can be realized. Note that, for the sake of inspection efficiency, it is preferable to arrange the semiconductor chips 10 vertically and horizontally, but this arrangement is not necessarily limited to the aligned arrangement shown in FIG. 1(b).
[0011] In addition, in the present invention, the imaging device EQU for capturing an enlarged image of the semiconductor chip 10 moves back and forth from diagonally above the semiconductor chip 10 placed on the light-reflective mounting surface, so that the end face of the thin plate-like chip can be captured widely. In order to capture a wide end face, it is advantageous to have a small inclination angle θ of the imaging device relative to the mounting surface during its reciprocating movement, but if it is too small, there is a drawback that optical components such as the objective lens come into contact with the mounting base or the semiconductor chip 10.
[0012] Considering this point, the staggered arrangement shown in FIG. 1(c) is more effective than the aligned arrangement shown in FIG. 1(b) because it ensures a wider gap (W1 / W2) between the semiconductor chips 10 in the direction of the illumination light 20.
[0013] However, a transmission microscope is not suitable for determining the quality of the end face 10a of each of a large number of semiconductor chips 10 arranged vertically and horizontally. Therefore, in the present invention, a reflected light microscope shown in Fig. 2(a) is used, and the imaging device of the present invention is configured to include the reflected light microscope and an external illumination 4.
[0014] 2(a), a reflected light microscope generally includes an objective lens 1 facing a sample 6, an imaging lens 2 that forms an image of imaging light 30 on an image sensor 7, an illumination lens 3 that receives illumination light 20, a splitter 5 that directs the illumination light 20 that has passed through the illumination lens 3 to the sample 6, and a light source 4 that generates the illumination light 20. As shown in the figure, imaging light 30, which is a reflected wave from the chip end face 10a (sample 6), passes through the splitter 5 and is directed to the image sensor 7.
[0015] Here, if the inclination angle θ is too small, the adjacent chip 10 in front as seen from the camera side will block the illumination light, causing unnecessary spot noise in the direct reflected wave from the end face 10a of the chip. On the other hand, if the inclination angle θ is too large, the end face 10a cannot be captured widely. Therefore, taking into consideration the arrangement pitch (W1 / W2) of the semiconductor chips 10 in the illumination direction and the height difference of the inspection range on the chip end face 10a, the inclination angle θ is preferably determined in the range of 20° to 40°.
[0016] Furthermore, the determination device according to the present invention extracts data on an evaluation line that continues from the inside to the outside of the semiconductor chip 10 for a plurality of images acquired while the imaging device EQU is moving (first means). The extraction of data on the evaluation line may be a real-time process in accordance with the imaging process of the imaging device, or may be a batch process for a plurality of images.
[0017] In the present invention, the reason why a plurality of images are acquired during the movement of the imaging device EQU is to accurately acquire a focused image of the semiconductor chip 10. That is, in the present invention, a plurality of images are acquired during the process of moving the imaging device EQU forward or backward from a state where the imaging device EQU has been temporarily positioned to a limit position, and the best focused image is selected from among the images.
[0018] 2(c) is a block diagram illustrating the configuration of an imaging device EQU that can move back and forth in the X and Z directions. As shown in the figure, the Z direction of the present invention forms an inclination angle θ with respect to the X direction. Note that a mounting table 40 that holds a semiconductor chip 10 is configured to be arbitrarily movable in two orthogonal axial directions on a horizontal plane, a vertical direction, and an inclined direction based on a camera that views the semiconductor chip 10 in a plan view.
[0019] As shown in FIG. 2(c), the imaging device EQU is configured to include a stepping motor Mx that realizes reciprocating movement in the X direction, a horizontal slide base 45 that can move reciprocatingly in the X direction based on the rotation of the stepping motor Mx and a first ball screw mechanism, a slide unit 44 that incorporates a stepping motor Mz and a second ball screw mechanism, a slide base 43 that can move reciprocatingly in an inclined direction along the slide unit 44, and an imaging main body 42.
[0020] Illumination light 20 is introduced into the imaging main body 42, and passes through the objective lens 1 to illuminate the end face 10a of the semiconductor chip 10. Then, a reflected wave from the chip end face 10a is imaged on the image sensor 7 of the imaging camera 41 as imaging light.
[0021] Incidentally, among the multiple acquired images, it is necessary to include an image that is accurately in focus in the inspection range including the original judgment target (judgment area). Therefore, the timing of acquiring the image is determined based on the depth of field δ of the imaging device EQU, the depth of the inspection range, etc.
[0022] As shown in FIG. 2(b), the depth of field δ is the allowable range of the distance between the sample surface 6 and the lens. Therefore, the depth of field δ means the range in which the image is in focus. This depth of field depends on the focal length and aperture of the lens, but the image of an object that exists within the depth of field exists within the imaging range (depth of focus) of an image sensor 7 such as a CCD (Charge Coupled Device). Generally, the shorter the focal length of the lens, the shallower the depth of field, and the narrower the aperture, the deeper the depth of field.
[0023] To evaluate the quality of the end face 10a of the semiconductor chip 10, a total magnification of 10 to 20 times is preferable, and the corresponding depth of field δ is about δ=10 μm to 2 μm. Also, it is preferable to intermittently capture multiple images every time the imaging device EQU moves a specified distance (imaging pitch) Pi, and the imaging pitch Pi is determined in consideration of the height dimension H of the inspection range that includes the judgment region (judgment target).
[0024] The photographing pitch Pi will be explained below, where it is assumed that the top surface of the semiconductor chip 10 is the origin line ORG (see FIG. 1(a)) as shown in FIG. 3(a), and the range H downward from the origin line ORG is the inspection range H. Note that, as shown in FIG. 3(b) and FIG. 3(c), the imaging device EQU moves back and forth along a moving line LN with an inclination angle θ, and the depth of field is represented as δ.
[0025] In such a case, for example, on the forward path on the moving line LN, the bottom line of the inspection range H can be captured in the focused field of view for the first time in the state shown in Fig. 3(b). That is, in the state shown in Fig. 3(b), the focal position F1 is located X1 in front of the origin line ORG, and the position +δ / 2 from the focal position F1 coincides with the bottom end of the inspection range (the lower line of H).
[0026] In this case, H*SIN(θ)+X1=δ / 2 (Equation 1) holds, and rearranging this equation gives X1=δ / 2-H*SIN(θ). Note that X1 does not necessarily have to be a positive value, and may be a negative value.
[0027] Next, when the imaging device EQU moves further forward on the forward path on the moving line, the state shown in Fig. 3(c) is reached, where the position of -δ / 2=-X2 from the focal position F2 coincides with the upper end of the inspection range H (i.e., the origin line ORG). In this state shown in Fig. 3(c), the relationship X2=δ / 2 (Equation 2) is established.
[0028] Incidentally, in the state of Fig. 3(c), the imaging device EQU needs to capture the lowest end of the inspection range (the +H line) as well, and the relationship H*SIN(θ)≦δ (formula 3) is required. (Formula 3) specifies the tilt angle θ, and once the inspection range H and depth of field δ are determined, the tilt angle θ needs to satisfy the relationship SIN(θ)≦δ / H (formula 3). This relationship also applies to the case of Fig. 3(b).
[0029] Then, based on (Equation 1) and (Equation 2), the relational equation X1+X2=δ-H*SIN(θ) (Equation 4) is calculated, and if an image is captured at a distance range where Pi=X1+X2 or less, an image focused on the inspection range H can be obtained without fail. Therefore, the shooting pitch Pi on the movement line must be Pi≦δ-H*SIN(θ) (Equation 4).
[0030] Even when the photographing pitch Pi is the maximum pitch Pi=δ-H*SIN(θ), if photographing is repeated at the same pitch Pi from any photographing start timing, one photographed image can always be obtained somewhere between the focal position F1 and the focal position F2. Therefore, in the present invention, the maximum value of the photographing pitch Pi is determined based on the inclination angle θ of the moving line, the vertical dimension H of the inspection range, and the depth of field δ.
[0031] By adopting such a configuration, the image captured at the optimal timing always includes the inspection range of +H below the origin line ORG within the depth of field δ, and an image of the inspection range in focus is obtained. Note that, for example, when the depth of field δ=14 μm, the inspection range H=10 μm, and the tilt angle θ=40°, the shooting pitch Pi is Pi≦7.5 μm, but by narrowing the shooting pitch Pi below the maximum value of 7.5 μm, multiple images in focus can be obtained.
[0032] On the other hand, regarding the condition of the inclination angle θ, for example, when the inspection range H=10 μm, the depth of field δ=10 μm, and the shooting pitch Pi is the maximum value δ-H*SIN(θ), according to (Formula 3), SIN(θ)≦10 / 10, and the inclination angle θ can be any value. Therefore, in this case, the inclination angle θ can be set as small as possible by adopting a staggered arrangement as shown in Figure 1(c).
[0033] Furthermore, when the shooting pitch Pi is the maximum value δ-H*SIN(θ), for example, when H=5 μm and the depth of field δ=2 μm, from SIN(θ)≦δ / H··· (Equation 3), SIN(θ)≦0.4 and therefore θ≦23.5°. However, when the shooting pitch Pi is shorter than the maximum value δ-H*SIN(θ), the tilt angle θ does not necessarily need to satisfy the relationship SIN(θ)≦δ / H··· (Equation 3).
[0034] In any case, in the present invention, a plurality of images are captured at a predetermined capture pitch Pi. Figures 4(a) to 4(e) show five images of the same semiconductor chip 10 captured at the capture pitch Pi, including images in which the inspection range is not in focus and images in which the focus is poor. Each image captures the end face 10a of the semiconductor chip 10, but since the mounting surface 40 is light reflective, a virtual image of the semiconductor chip 10 is also captured.
[0035] In the present invention, it is necessary to extract the most focused image from multiple images such as those shown in Fig. 4. Therefore, in the present invention, an evaluation line EVL that continues from the inside to the outside of the semiconductor chip 10 is defined, and image data along this evaluation line EVL is extracted (first means).
[0036] The evaluation line EVL is not particularly limited, but preferably, as shown by a white line in Figures 4(a) to 4(e), a straight line that intersects with the origin line ORG (Figure 3(a)) and extends from the end surface 10a of the semiconductor chip 10 to the inside is used as the evaluation line EVL. Note that the intersecting line does not necessarily have to be perpendicular to the origin line ORG.
[0037] When the image captured by the imaging device EQU is an RGB color image, the poured data poured along the evaluation line EVL is preferably one-dimensional data, and for example, brightness data (V value) obtained by converting RGB data to HSV is preferably used as the poured data. In the HSV conversion, the V value (Value of Brightness) is specified as the maximum value among the R value (Red), G value (Green), and B value (Blue) in the RGB space.
[0038] However, if any of the single colors of RGB has a large change in brightness, it is preferable to use light quantity data that has passed through an appropriate single-color filter. Note that the image is not limited to a color image, and a monochrome image may also be used.
[0039] In addition, the quality of focus is evaluated based on the sharpness of the edges detected from the pouring data, and the image with the sharpest edge is selected as the image with the best focus. In the example of Figure 4, image 4 or image 5 is selected as the best image.
[0040] The above operations are realized as computer processing, and the brightness data (V value) obtained by HSV conversion is smoothed, for example, by averaging processing, and the smoothed data is then subjected to, for example, differentiation processing to extract positive and negative edges.
[0041] Then, the positive or negative edge is binarized based on a predetermined threshold value TH to generate a pulse wave, the pulse width is evaluated, and the image with the narrowest pulse width is selected as the optimum image (second means).Whether the positive edge or the negative edge is the issue depends on the positive or negative level difference in brightness (V value) between the inside and outside of the semiconductor chip 10.
[0042] Fig. 5(a) shows HSV brightness data, and Fig. 5(b) shows the waveform after averaging the brightness data. Fig. 5(c) shows the waveform after the averaged waveform has been subjected to differentiation, and Fig. 5(d) shows the pulse wave after binarization. As shown, the brightness data (Fig. 5(a)) fluctuates slightly in level, but by applying averaging, it is possible to remove minute level components.
[0043] The averaging process is realized, for example, by a convolution operation that averages three consecutive data sets including the two preceding and following data sets (see FIG. 5(b)), and the differentiation process is realized, for example, by a difference operation (convolution) between the previous data set and the next data set (see FIG. 5(c)). However, the above processes are not limiting, and any suitable digital filter process that actually realizes the integration process or differentiation process can be used.
[0044] In any case, the luminance data on the evaluation line EVL of all captured images is subjected to averaging, differentiation, and binarization to obtain pulse waves, and the image with the smallest pulse width is selected as the optimum image. In the case of an out-of-focus image, as shown in Fig. 5(a), the data after differentiation changes gradually, so the pulse width of the binarized signal becomes wider. Hereinafter, the process of obtaining a pulse wave by averaging, differentiation, and binarization may be referred to as edge detection.
[0045] Once the optimum image has been selected in the above manner, the optimum image extracted by the second means is then judged based on a standard image acquired in advance, and the quality of the semiconductor chip 10 corresponding to the image is judged (third means).
[0046] The specific method of the third means may be any method, but for example, a boundary portion showing a significant brightness difference is identified based on an edge detection process, and the background of the captured image is first removed. At this time, it is also preferable to remove the virtual image reflected on the mounting table 40 based on the edge detection process or the like.
[0047] Next, the captured image is evaluated from the viewpoint of whether the semiconductor chip 10 related to the captured image has scratches or dust. The specific method is appropriate, but for example, the similarity between the captured image and the feature points or feature amounts previously extracted for the standard image is calculated, and the quality of the semiconductor chip 10 corresponding to the image is judged based on the evaluation value (similarity). In this case, a deep learning method may be used. If it is simply necessary to judge the presence or absence of scratches or dust, an appropriate evaluation line may be set, and the quality of the semiconductor chip 10 may be judged based on the transition of the brightness data on the evaluation line. The transition of the brightness data is judged, for example, by the number of pulses or the pulse width for the pulse wave after the edge detection process. Effect of the Invention
[0048] As described above, according to the present invention, the quality of the end face of a semiconductor chip can be determined quickly and reliably. [Brief description of the drawings]
[0049] [Figure 1] 1 is a diagram illustrating an arrangement state of semiconductor chips and an operation of an imaging device. [Diagram 2] 1 is a diagram illustrating a configuration of an imaging device and a depth of field. [Diagram 3] 11 is a diagram for explaining a maximum value of a shooting pitch. [Figure 4] 1 is a diagram showing a sampling image. [Diagram 5] 1 is a diagram illustrating a data processing procedure. [Figure 6] 1 is a diagram for explaining the operation of the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0050] The present invention will be described in detail below based on an embodiment, but this is merely an example and the specific description does not limit the present invention in any way. Fig. 6 is a flow chart for explaining the operation of the imaging device EQU and the determination device according to the embodiment. The determination device controls the operation of the imaging device EQU, which has a built-in incident-light microscope whose basic configuration is shown in Fig. 2(a).
[0051] 2(c), and includes a stepping motor Mx that realizes reciprocating motion in the X direction and a stepping motor Mz that realizes reciprocating motion in the Z direction. In this embodiment, the movement line LN of the imaging device EQU is inclined by θ=30° with respect to the mounting surface 40 of the semiconductor chip 10.
[0052] Based on the above conditions, the photographing pitch Pi must be Pi≦δ-H*SIN(30)=δ-H / 2, where δ is the depth of field and H is the height of the inspection range. Considering the depth of field δ=10 μm and the height of the inspection range H=10 μm, the photographing pitch Pi in this embodiment is set to Pi=4 μm, which is slightly narrower than the maximum value of 5 μm. Therefore, when the pulse speed of the stepping motor Mz is, for example, 1400 pps (pulse per second), the side surface 10a of the semiconductor chip 10 is repeatedly photographed every 40 pulses. Note that under these operating conditions, the photographing cycle is 40 / 1400=28.6 mS.
[0053] 5, first, feature parameters are identified (ST1) for a reference image of a reference chip that is a typical non-defective product of the semiconductor chip 10. In addition, for a reference range that is a pass / fail judgment region of the reference chip, a horizontal range H1 / HMAX to H2 / HMAX from the left end of the reference chip and a vertical range V1 / VMAX to V2 / VMAX from the top end are identified.
[0054] The pass / fail judgment area does not necessarily have to be one, but may be multiple, for one semiconductor chip 10. For example, when three pass / fail judgment areas AR1 to AR3 are provided, the first area AR1 closest to the objective lens 1 is set as the reference area, and optimal images of the other judgment areas AR2 and AR3 are specified based on the relative positional relationship with the reference area AR1.
[0055] Specifically, when capturing images while moving the imaging device EQU closer, based on the relative positional relationship with the reference area (first area AR1), image i+n, which is shifted by n pixels from the optimal image i capturing the first area AR1, and image i+m, which is shifted by m pixels, are identified as the optimal images for the second area AR2 and the third area AR3, respectively.
[0056] On the other hand, in the case of the present embodiment where images are captured while moving the imaging device EQU away, the third area AR3 which is the farthest from the objective lens 1 is set as the reference area, and image i+n which is shifted by n pixels from the optimal image i capturing the reference area (third area AR3), and image i+m which is shifted by m pixels are identified as the optimal images of the second area AR2 and the first area AR1, respectively.
[0057] Next, the imaging device EQU is set to the imaging start position. The procedure for setting the imaging start position is as shown in FIG. 5(b). First, the mounting table 40 is moved to an optimal position in the horizontal plane XY direction corresponding to the arrangement position of the semiconductor chip 10 that is the imaging target (ST20). If necessary, it may be moved in the vertical Z direction or in an inclined direction, and by moving in an inclined direction, the inclination angle θ=30° between the mounting surface 40 and the movement line LN of the imaging device EQU can be appropriately changed. In any case, as a result of these moving operations, the semiconductor chip 10 that is the imaging target is initially positioned close to the imaging device EQU.
[0058] Next, move the imaging device EQU forward or backward in the X direction in Fig. 2(c) so that the chip end face 10 is largely displayed at the center of the imaging screen until it reaches the limit state. Since the movement in the X direction is executed with an accuracy of, for example, 1 μm, an accurate positioning operation is realized. Thus, in this embodiment, in addition to the movement operation (ST20) of the mounting table 40 by the camera that captures the plan view of the semiconductor chip 10, the imaging device EQU is moved to the optimal position based on the microscope image (ST21), enabling precise positioning. However, at this timing, a focused imaging image has not yet been obtained.
[0059] Therefore, next, in order to obtain a focused imaging image, in this embodiment, the imaging device EQU is made to start the return movement (ST2). Note that the forward movement may be used instead of the return movement, but in this embodiment, the return movement is adopted corresponding to the positioning operation in step ST2.
[0060] When the shooting operation is started, it is determined whether the shooting timing (in this embodiment, after +40 pulses have elapsed) has been reached (ST3). If the image acquisition timing is reached, the end face 10a of the semiconductor chip 10 is photographed and stored (ST4). Next, the region of the semiconductor chip 10 is determined, and an evaluation line extending from the upper part of the semiconductor chip 10 to the inside of the semiconductor chip 10 is specified (ST5).
[0061] Subsequently, for the brightness data on the evaluation line, averaging processing such as calculating the moving average value is executed to remove noise (ST6), and the data after the noise processing is differentiated to detect positive and negative edges (ST7).
[0062] Next, in this embodiment, focusing on the positive edge, based on a predetermined threshold TH, the positive edge is binarized (ST7), and the pulse width PL of the binarized pulse wave is calculated (ST8). Then, this pulse width PL is compared with the minimum value MIN up to that point. If PL < MIN, the minimum value MIN is rewritten with the current pulse width PL, and the image number i corresponding to the pulse width PL is stored (ST10).
[0063] The above-described processes ST4 to ST10 are repeatedly executed at each imaging timing. Then, when the imaging device has moved as far away from the semiconductor chip 10 as possible, the imaging device EQU is returned to the imaging start position, and the optimal image is specified based on the image number i corresponding to the minimum value MIN stored at that time (ST12).
[0064] After the above process is completed, the background is removed from the identified optimum image i, and necessary feature parameters are calculated. Then, the similarity is calculated based on a comparison with the feature parameters of the reference image, and the quality of the semiconductor chip 10 corresponding to the optimum image i is judged (ST13).
[0065] Then, for the semiconductor chip 10 that is evaluated as a defective product, the arrangement position on the mounting surface, etc. are stored, and the semiconductor chip 10 is subjected to the necessary subsequent processing (ST14). Since the pass / fail judgment of the specific semiconductor chip 10 is completed by the above processing, the processing of ST3 to ST25 is repeated for the adjacent semiconductor chip 10 thereafter.
[0066] Although the embodiment has been specifically described above, the specific configuration can be appropriately changed without departing from the spirit of the present invention. For example, in the embodiment, the chip end face is photographed in a stopped state, but since the moving speed is low, it may be photographed while moving. In addition, the method of detecting the minimum pulse width PL is not limited to the sequential operation shown in FIG. 6. For example, after acquiring and saving all the required number of images, a batch process in which edge extraction is performed on each image and the optimal image is selected is also suitable. [Explanation of symbols]
[0067] ST5 1st method ST12 Second means ST13 Third means
Claims
1. The semiconductor chip is placed on a light-reflective mounting surface, and the imaging device moves back and forth from an obliquely upward direction to capture an image of the end face of the semiconductor chip. The semiconductor chip is then subjected to a pass / fail test based on the captured image. The determination device includes: a first means for extracting data on an evaluation line extending from the inside to the outside of a semiconductor chip from a plurality of images captured while the imaging device is moving; a second means for evaluating a change in data on the evaluation line for the plurality of images and identifying an optimal image from the plurality of images; and a third means for judging the optimal image identified by the second means based on a standard image acquired in advance, and judging the quality of the semiconductor chip corresponding to the image.
2. 2. The quality determining device according to claim 1, wherein the imaging device reciprocates at an inclination angle of 20 degrees to 40 degrees with respect to the placement surface.
3. The quality determination device according to claim 1 , wherein the imaging device captures an enlarged image at a predetermined magnification.
4. The plurality of images are intermittently acquired while the imaging device is stationary or moving, each time the imaging device moves a specified distance; 4. The quality determination device according to claim 1, wherein the specified distance is determined based on an inclination angle θ when the imaging device is moved, a vertical dimension H of an inspection range required for quality determination, and a depth of field δ.
5. 5. The quality determining device according to claim 1, wherein the data on the evaluation line is brightness data.
6. 6. The quality determining device according to claim 1, wherein the second means evaluates a transition pattern of data on the evaluation line, and extracts an image showing the sharpest transition pattern as an optimum image.
7. 7. The quality determining device according to claim 6, wherein the second means evaluates a transition state based on a differential processed value of data on the evaluation line.
8. 8. The quality determining device according to claim 1, wherein the imaging device is configured to be capable of reciprocating on a horizontal plane.
9. 9. The quality determining device according to claim 1, wherein the semiconductor chips are arranged in a staggered pattern in a plan view, thereby ensuring a wide gap in the direction of the illumination light.
10. A plurality of pass / fail judgment areas including a reference judgment area are set on the semiconductor chip, The pass / fail judgment device according to any one of claims 1 to 9, which is configured to select optimal images from the plurality of captured images for other pass / fail judgment areas in correspondence with an optimal image for the reference judgment area selected from the plurality of captured images.
11. The semiconductor chip is placed on a light-reflective mounting surface, and the imaging device moves back and forth from an obliquely upward direction to capture an image of the end face of the semiconductor chip. The semiconductor chip is then subjected to a pass / fail test based on the captured image. a first step of extracting data on an evaluation line extending from the inside to the outside of a semiconductor chip from a plurality of images acquired while the imaging device is moving; a second step of evaluating a change in data on the evaluation line for the plurality of images to identify an optimal image from the plurality of images; A method for determining whether a semiconductor chip is good or bad, comprising: a third step of determining whether the optimal image identified in the second step is good or bad based on a standard image acquired in advance, and determining whether the semiconductor chip corresponding to the image is good or bad.
Citation Information
Patent Citations
Visual inspection device for tape carrier package
JP1998002725A
Device for judging substrate quality
JP1998040387A
Method of inspecting floating of chip component and its device
JP2002296203A
Bump forming device, bump forming method, inspecting device and inspecting method
JP2004214481A
Device and method for substrate inspection
JP2007013112A
