Mounting device and mounting method
The mounting apparatus and method address reproducibility issues in semiconductor chip bonding by using a sigmoid function to fit a brightness profile, optimizing the ROI size, and calculating center positions, resulting in enhanced precision and accuracy for alignment and mounting.
Patent Information
- Application Number
- JP2023086949
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-26
- Publication Date
- 2024-12-06
AI Technical Summary
Existing methods for high-precision alignment in semiconductor chip bonding, such as flip chips and through silicon vias, suffer from reduced reproducibility due to variations in edge position detection caused by camera noise, leading to inaccuracies in alignment and mounting.
A mounting apparatus and method that uses a sigmoid function to fit a brightness profile of an alignment mark, focusing on specific regions to minimize noise impact, and calculates center positions based on detected edge positions, ensuring accurate alignment and mounting by optimizing the size of the region of interest (ROI) to reduce variations.
The solution enhances reproducibility and accuracy in detecting edge positions, allowing for precise alignment and mounting of semiconductor chips, even in the presence of noise, thereby improving the overall precision of the bonding process.
Smart Images

Figure 2024170031000001_ABST
Abstract
Description
[Technical field]
[0001] The present disclosure relates to a mounting apparatus and a mounting method. [Background technology]
[0002] In order to achieve low power consumption and high driving speed, semiconductor devices are becoming more multi-layered. The process of stacking semiconductor chips such as CoC (Chip on Chip) and CoW (Chip on Wafer) or the chip bonding process, which is a process of mounting semiconductor packages, is changing from the conventional contact connection method via wire bonding to a connection method using flip chips or through silicon vias (TSVs). The existing contact connection method via wire bonding was sufficient with a bonding accuracy of several tens of microns. However, flip chips, which make direct contact between bumps and pads, require an accuracy of several microns. In particular, chip bonding using through silicon vias requires sub-micron accuracy.
[0003] In the bonding process, there is a method of recognizing an alignment mark by using an upper and lower two-view optical system (for example, Patent Document 3). Specifically, the upper and lower two-view optical system is inserted between a lower object to be bonded held on a bonding stage and an upper object to be bonded held on a bonding head. The upper and lower two-view optical system recognizes alignment marks on the bonding surface of the lower object to be bonded and on the bonding surface of the upper object to be bonded. The upper and lower two-view optical system is a combination of a camera for recognizing the upper side and a camera for recognizing the lower side. The upper and lower two-view optical system has a drive shaft at least in a horizontal plane so that it can enter the gap between the lower object to be bonded and the upper object to be bonded from the side just before bonding. Then, based on the recognition result, the objects to be bonded are aligned and bonded.
[0004] In order to detect the position of an alignment mark, the edge position of the alignment mark is detected. As a method for detecting the edge position, for example, an edge position detection device disclosed in Patent Document 1 obtains a luminance profile of an area including an edge of a pattern, and detects the edge position by applying a high-order approximation equation to an inclined portion indicating the edge.
[0005] Moreover, the edge position detection device of Patent Document 2 acquires a luminance profile of an inspection image showing a group of pattern elements on a substrate.The edge position detection device then detects the edge position by applying a bilaterally symmetric model function, which is a combination of four bell-shaped functions corresponding to the four concave portions and three bell-shaped functions corresponding to the three convex portions, to the luminance profile having four concave portions and three convex portions alternately arranged. [Prior art documents] [Patent documents]
[0006] [Patent Document 1] Patent No. 5563942 [Patent Document 2] Patent No. 6355487 [Patent Document 3] Patent No. 5876000 Summary of the Invention [Problem to be solved by the invention]
[0007] Incidentally, the brightness of an image contains camera noise such as dark shot noise, read noise, and photon shot noise. For this reason, for example, when an edge position is obtained for a group of images of the same object taken at different times using Patent Document 1 or Patent Document 2, the result of function fitting may vary between images due to the influence of brightness variations caused by camera noise. In this case, different edge positions are detected in the group of images even for the same object. Therefore, in situations where highly accurate position detection is required, this is undesirable as it leads to a decrease in reproducibility. This makes it difficult to achieve highly accurate mounting.
[0008] The present disclosure has been made to solve such problems, and improves the reproducibility of detection of edge positions of alignment marks using a fitting function in order to perform highly accurate alignment between objects to be bonded in a bonding process, thereby providing a mounting device and a mounting method capable of achieving highly accurate mounting. [Means for solving the problem]
[0009] A mounting apparatus according to one embodiment includes a profile acquisition unit that acquires, from an image of a region including an alignment mark formed on an object to be bonded, a luminance profile along one direction of a region of interest including a first region, a second region, and a third region that are aligned along the one direction in the image; a fitting unit that fits a fitting function including a sigmoid function having an inflection point and a curvature to the luminance profile and detects an edge position of the alignment mark from the inflection point; and a bonding unit that bonds another object to the object to be bonded using the detected edge position, wherein the second region corresponds to the alignment mark, and the first region and the third region have a difference in luminance from the second region, The fitting function is TIFF2024170031000002.tif17105, μ 1 and μ 2 indicates the inflection point, and a 1 and a 2denotes the curvature, and b and c denote constants.
[0010] The mounting apparatus may further include a position calculation unit that calculates a center position of the alignment mark based on the detected edge position.
[0011] In the above mounting device, the second region is a rectangular shape having sides along the one direction and another direction that intersects with the one direction, and the position calculation unit may calculate a first center position of the alignment mark in the one direction based on the edge position detected in the one direction, and calculate a second center position of the alignment mark in the other direction based on the edge position detected in the other direction.
[0012] In the above mounting device, the profile acquisition unit may acquire the luminance profile of the maximum range of the region of interest that satisfies at least one of the following: an error between the luminance profile and the fitting function is equal to or less than a predetermined value; and a calculation time for detecting the edge position is equal to or less than a predetermined time.
[0013] In the above mounting device, the image has a fourth region around the second region, the fourth region being brighter or lower than the first region and the third region, and corresponding to another mark, and when the region of interest includes the fourth region, the error between the brightness profile and the fitting function may be greater than a predetermined value.
[0014] In the above mounting device, the area of interest may include a first edge portion having an edge of the alignment mark between the first area and one of the second areas, and a second edge portion having the edge between the first area and the other of the second areas, and the fitting function may include a first sigmoid function having the inflection point and the curvature of the first edge portion, and a second sigmoid function having the inflection point and the curvature of the second edge portion.
[0015] The above mounting device further includes a bonding head that holds the object to be bonded, a bonding stage that holds another object to be bonded to the object to be bonded, and a top-bottom two-field optical system that is inserted between the object to be bonded and the other object to be bonded and captures images of the object to be bonded and the other object to be bonded, and when the alignment mark of the object to be bonded is a first alignment mark and the alignment mark of the other object to be bonded is a second alignment mark, the top-bottom two-field optical system may have a single image sensor that acquires images of the first alignment mark and the second alignment mark.
[0016] A mounting method according to one embodiment includes the steps of: acquiring, from an image of a region including an alignment mark formed on an object to be bonded, a luminance profile along one direction of a region of interest including a first region, a second region, and a third region aligned along the one direction in the image; fitting a fitting function including a sigmoid function having an inflection point and a curvature to the luminance profile, and detecting an edge position of the alignment mark from the inflection point; and bonding another object to the object to be bonded using the detected edge position, wherein the second region corresponds to the alignment mark, and the first region and the third region have a difference in luminance from the second region, The fitting function is TIFF2024170031000003.tif17105, μ 1 and μ 2 indicates the inflection point, and a 1 and a 2 denotes the curvature, and b and c denote constants.
[0017] The mounting method may further include a step of calculating a center position of the alignment mark based on the detected edge position.
[0018] In the above implementation method, the second region is a rectangular shape having sides along the one direction and another direction that intersects with the one direction, and in the step of calculating the center position of the alignment mark, a first center position of the alignment mark in the one direction may be calculated based on the edge position detected in the one direction, and a second center position of the alignment mark in the other direction may be calculated based on the edge position detected in the other direction.
[0019] In the above implementation method, in the step of acquiring the luminance profile, the luminance profile of the maximum range of the region of interest that satisfies at least one of the following: an error between the luminance profile and the fitting function is equal to or less than a predetermined value; and a calculation time for detecting the edge position is equal to or less than a predetermined time.
[0020] In the above implementation method, the image has a fourth region around the second region, the fourth region being brighter or lower than the first region and the third region, and corresponding to another mark, and when the region of interest includes the fourth region, the error between the brightness profile and the fitting function may be greater than a predetermined value.
[0021] In the above implementation method, the area of interest may include a first edge portion having an edge of the alignment mark between the first area and one of the second areas, and a second edge portion having the edge between the first area and the other of the second areas, and the fitting function may include a first sigmoid function having the inflection point and the curvature of the first edge portion, and a second sigmoid function having the inflection point and the curvature of the second edge portion.
[0022] The above mounting method may further include the steps of holding the object to be bonded by a bonding head, holding another object to be bonded to the object to be bonded by a bonding stage, inserting an upper and lower two-field optical system between the object to be bonded and the other object to be bonded, and, when the alignment mark of the object to be bonded is a first alignment mark and the alignment mark of the other object to be bonded is a second alignment mark, acquiring an image of the first alignment mark and the second alignment mark by one image sensor in the upper and lower two-field optical system. Effect of the Invention
[0023] The present disclosure makes it possible to provide a mounting apparatus and a mounting method that can achieve highly accurate mounting. [Brief description of the drawings]
[0024] [Figure 1] 11 is a diagram illustrating an example of an ROI of an image including an alignment mark in the method for detecting the edge position of an alignment mark according to Reference Example 1. FIG. [Diagram 2] 11 is a diagram illustrating an example of an ROI of an image including an alignment mark, a portion with a high intensity of a luminance gradient, and the intensity of the luminance gradient in a method for detecting an edge position of an alignment mark according to Reference Example 1; FIG. [Diagram 3] 1 is a graph illustrating the intensity of the brightness gradient of an ROI in a method for detecting the edge position of an alignment mark in accordance with Reference Example 1, where the horizontal axis indicates the position on the ROI in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient. [Figure 4] 1 is a graph illustrating the intensity of the brightness gradient of an ROI in a method for detecting the edge position of an alignment mark in accordance with Reference Example 1, where the horizontal axis indicates the position on the ROI in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient. [Diagram 5] 1 is a graph illustrating the intensity of the brightness gradient of an ROI in a method for detecting the edge position of an alignment mark in accordance with Reference Example 1, where the horizontal axis indicates the position on the ROI in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient. [Figure 6] 1 is a graph illustrating the intensity of the brightness gradient of an ROI in a method for detecting the edge position of an alignment mark in accordance with Reference Example 1, where the horizontal axis indicates the position on the ROI in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient. [Figure 7] 13 is a diagram illustrating an example of an ROI of an image including an alignment mark in a method for detecting the edge position of an alignment mark according to Reference Example 2. FIG. [Figure 8] 11 is a graph illustrating a luminance profile in the method for detecting the edge position of an alignment mark according to Reference Example 2, where the horizontal axis indicates the position on the ROI in the x-axis direction and the vertical axis indicates the intensity of luminance. [Figure 9] 13 is a diagram illustrating an example of a deviation of the ROI of an alignment mark according to Reference Example 2. FIG. [Figure 10] 13 is a diagram illustrating an alignment mark and the intensity of a brightness gradient in a method for detecting the edge position of an alignment mark according to Reference Example 3. FIG. [Figure 11] 11 is a graph illustrating the intensity of the brightness gradient of an ROI in the method for detecting the edge position of an alignment mark in accordance with Reference Example 3, where the horizontal axis indicates the position of the ROI in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient. [Figure 12] 13 is a graph illustrating a brightness profile of an ROI in a method for detecting an edge position of an alignment mark according to Reference Example 4, where the horizontal axis indicates the position on the ROI in the x-axis direction and the vertical axis indicates brightness intensity. [Figure 13] 1 is a configuration diagram illustrating a mounting device according to a first embodiment. [Figure 14] 3 is a configuration diagram illustrating an optical section of an upper and lower two-view optical system in the mounting device according to the first embodiment. FIG. [Figure 15] 4 is a diagram illustrating an example of an alignment mark captured by an image sensor in the mounting device according to the first embodiment. FIG. [Figure 16] 2 is a block diagram illustrating a processing device in the mounting device according to the first embodiment. FIG. [Figure 17]4 is a diagram illustrating an image captured of a region including an alignment mark formed on an object to be bonded in the mounting device according to the first embodiment. FIG. [Figure 18] 1 is a graph illustrating a luminance profile and a fitting function in the mounting device according to the first embodiment, where the horizontal axis indicates the position on the ROI in the x-axis direction, and the vertical axis indicates the intensity of luminance. [Figure 19] 1 is a graph illustrating a fitting function in the mounting device according to the first embodiment, in which the horizontal axis indicates the position on the ROI in the x-axis direction, and the vertical axis indicates the intensity of luminance. [Figure 20] 4 is a diagram illustrating an example of a curvature depending on a coefficient of a fitting function in the mounting device according to the first embodiment. FIG. [Figure 21] 4 is a diagram illustrating an example of an edge position detected in the mounting apparatus 1 according to the first embodiment. FIG. [Figure 22] 4 is a diagram illustrating an example of a center position calculated by a position calculation unit in the mounting device 1 according to the first embodiment. FIG. [Figure 23] FIG. 4 is a flow chart illustrating a mounting method using the mounting device according to the first embodiment. [Figure 24] 4 is a flow chart illustrating a method for detecting the center position of an alignment mark according to the first embodiment. [Diagram 25] 4 is a diagram illustrating an example of an ROI extracted by a profile acquisition unit in the implementation method according to the first embodiment. FIG. [Figure 26] 11 is a diagram illustrating data obtained by averaging luminance of an ROI in the implementation method according to the first embodiment. FIG. [Figure 27] 1 is a graph illustrating the reproducibility when the size of the ROI is changed in the mounting method of embodiment 1, where the horizontal axis indicates the ratio of the length of the ROI to the length of the alignment mark in the x-axis direction, and the vertical axis indicates the reproducibility. [Figure 28] This is a graph illustrating a case where the accuracy decreases as the size of the ROI increases in the mounting method of embodiment 1, where the horizontal axis indicates the ratio of the length of the ROI to the length of the alignment mark in the x-axis direction, and the vertical axis indicates the error. [Figure 29] 1 is a graph illustrating the relationship between ROI size and calculation time in the mounting method of embodiment 1, where the horizontal axis indicates the ratio of the ROI length to the alignment mark length in the x-axis direction, and the vertical axis indicates the calculation time. [Diagram 30] 4 is a diagram illustrating a GUI in the mounting device according to the first embodiment. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0025] For clarity of explanation, the following description and drawings have been omitted or simplified as appropriate. In addition, in each drawing, the same elements are given the same reference numerals, and duplicate explanations are omitted as necessary.
[0026] (Overview of the embodiment) As an example, the mounting device of this embodiment detects the edge position of an alignment mark by fitting using a fitting function in order to align objects to be bonded with high accuracy in a bonding process. Specifically, the mounting device obtains a luminance profile of a region of interest (hereinafter referred to as ROI) from an image including the alignment mark and its vicinity. Then, the mounting device fits a predetermined fitting function to the luminance profile. In this way, the mounting device detects the edge position of the alignment mark. Then, the mounting device bonds the objects to be bonded based on the detected edge position.
[0027] There are several methods for finding edge positions by fitting a fitting function to a brightness profile in an ROI. However, since a captured image contains various noises, each brightness in the brightness profile has a variation. Due to this variation, different edge positions are detected between images captured at different times, even if the images are of the same object. This reduces reproducibility. In this embodiment, in order to suppress such variation, the size of the ROI to be fitted is increased according to the idea of the law of large numbers.
[0028] In order to clarify the edge position detection method used by the mounting device of this embodiment, first, the edge position detection method according to the reference example will be described. Then, the mounting device and the edge position detection method of this embodiment will be described in comparison with the reference example. Note that the edge position detection method according to the reference example is also included in the technical concept of this embodiment. Moreover, the mounting device of this embodiment does not exclude mounting the workpiece using the edge position detection method according to the reference example.
[0029] (Reference Example 1) A method for detecting the edge position of an alignment mark according to Reference Example 1 will be described. FIG. 1 is a diagram illustrating an example of an ROI 82 of an image including an alignment mark 80 in a method for detecting an edge position 81 of an alignment mark 80 according to Reference Example 1. As shown in FIG. 1, the method for detecting the edge position 81 of an alignment mark 80 according to Reference Example 1 uses an image of the alignment mark 80. The alignment mark 80 is formed on a predetermined surface of the object to be bonded Ma. The predetermined surface may be, for example, a bonding surface of the object to be bonded Ma or an interface in a multilayer structure, as long as it is a surface on which the alignment mark 80 is formed.
[0030] The portion of the image corresponding to the alignment mark 80 is a high-luminance region 86H. On the other hand, the portion of the bonding object Ma in the image that is not the alignment mark 80 is a low-luminance region 86L. The low-luminance region 86L has a lower luminance than the high-luminance region 86H. The high-luminance region 86H has a higher luminance than the low-luminance region 86L. The difference in luminance between the alignment mark 80 and its surroundings is produced by the difference in reflectance. Therefore, depending on the wavelength of the illumination light that illuminates the alignment mark 80, the portion corresponding to the alignment mark 80 may become a low-luminance region and its surroundings may become a high-luminance region. Also, when observing a transmission image by irradiating transmitted light from the bonding head or bonding stage side, the portion corresponding to the alignment mark 80 becomes a low-luminance region and its surroundings become a high-luminance region. Therefore, the regions having a difference in luminance that are aligned along one direction in the image are called the first region, the second region, and the third region. The portion corresponding to the alignment mark 80 is the first region. The non-alignment mark 80 portions around the alignment mark 80 are the first region and the third region. The first region and the third region are located on both sides of the second region in one direction. This distinguishes the first region, the second region, and the third region. In the following description, the second region is a high-luminance region, and the first and third regions are low-luminance regions. Of course, as described above, it is not excluded to set the second region as a low-luminance region and the first and third regions as high-luminance regions.
[0031] The ROI 82 partially includes the alignment mark 80 formed on the object to be bonded Ma. Specifically, the ROI 82 includes one side 83 of the alignment mark 80 and a peripheral region 84 adjacent to the side 83. The peripheral region 84 is a portion of the object to be bonded Ma where the alignment mark 80 is not arranged.
[0032] Here, for convenience in explaining the method for detecting the edge position 81 of the alignment mark 80, an xyz Cartesian coordinate system is introduced. The predetermined surface on which the alignment mark 80 is formed is defined as the xy plane. The direction perpendicular to the predetermined surface on which the alignment mark 80 is formed is defined as the z-axis direction. For example, the alignment mark 80 may be rectangular having sides along the x-axis and y-axis directions.
[0033] FIG. 2 is a diagram illustrating an ROI 82 of an image including the alignment mark 80, a portion 85 with a high intensity of the luminance gradient, and the intensity of the luminance gradient in a method for detecting an edge position 81 of an alignment mark 80 according to Reference Example 1. As shown in FIG. 2, the method for detecting an edge position 81 of an alignment mark 80 according to Reference Example 1 calculates the intensity of the luminance gradient in the ROI 82 from a portion of the ROI 82 of the image. For example, the intensity of the luminance gradient may be calculated by a Sobel filter. The portion 85 with a high intensity of the luminance gradient is located between a peripheral region 84 and a side portion 83. That is, the portion 85 with a high intensity of the luminance gradient includes the edge position 81 of the alignment mark 80.
[0034] Next, a luminance profile along the x-axis direction in the image is obtained. The luminance profile shows, for example, intensities I1, I2, and I3 along the x-axis direction. Intensity I2 is the intensity of a pixel showing a peak in a portion 85 where the intensity of the luminance gradient is high. Intensities I1 and I3 are the intensities of two adjacent pixels on either side of the peak.
[0035] Next, a quadratic function is fitted to the luminance profile including the intensities I1, I2, and I3. That is, a quadratic function is applied to the luminance profile including the pixel with the peak intensity of the luminance gradient and the intensities of the pixels around it. The apex of the fitted quadratic function is detected as the edge position 81.
[0036] In the method of Reference Example 1, when the image sharpness is low and the profile of the intensity of the brightness gradient is gentle, the edge position 81 is likely to vary due to the influence of noise. Figures 3 to 6 are graphs illustrating the intensity of the brightness gradient of ROI 82 in the method of detecting edge position 81 of alignment mark 80 according to Reference Example 1, where the horizontal axis indicates the position on ROI 82 in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient.
[0037] For example, in a profile with a steep luminance gradient as shown in Fig. 3, when the luminance gradient intensity at position x=4 on ROI 82 increases by 5 points as shown in Fig. 4, the edge position 81 of alignment mark 80 shifts by 0.01 from x=4.92 to x=4.91. However, in a profile with a gentle luminance gradient as shown in Fig. 5, when the luminance gradient intensity at position x=4 on ROI 82 increases by 5 points as shown in Fig. 6, the edge position 81 of alignment mark 80 shifts by 0.25 from x=4.90 to x=4.65. In this way, in the method of Reference Example 1, the change in edge position 81 due to the change in the luminance gradient intensity at the pixel is large.
[0038] (Reference Example 2) Next, a method for detecting the edge position 81 of the alignment mark 80 according to Reference Example 2 will be described. FIG. 7 is a diagram illustrating an example of an ROI 82 of an image including the alignment mark 80 in the method for detecting the edge position 81 of the alignment mark 80 according to Reference Example 2. FIG. 8 is a graph illustrating an example of a luminance profile in the method for detecting the edge position 81 of the alignment mark 80 according to Reference Example 2, where the horizontal axis indicates the position on the ROI 82 in the x-axis direction and the vertical axis indicates the intensity of luminance. FIG. 9 is a diagram illustrating an example of a deviation of the ROI 82 of the alignment mark 80 according to Reference Example 2.
[0039] As shown in Fig. 7, the method of Reference Example 2 calculates the center of gravity using pixels with brightness equal to or higher than a threshold value in ROI 82. Specifically, the center of gravity position is calculated by weighting brightness using the following formula (1). As a result, the method of Reference Example 2 can improve repeatability by increasing the number of target pixels to be calculated.
[0040]
number
[0041] As shown in FIG. 8, in the method of Reference Example 2, when the center of gravity position is calculated using only bright areas including pixels equal to or greater than the threshold, only pixels equal to or greater than the threshold are used for calculation. Also, when no threshold is set and all pixels in the ROI 82 are used, as shown in FIG. 9, even a shift in the position of the ROI 82 by one pixel results in a different target pixel being calculated. This causes the results of the edge position 81 to vary greatly. Also, the position of the ROI 82 may be set by roughly recognizing the alignment mark 80. Furthermore, the edge position 81 of the image is usually blurred. For this reason, it is difficult to obtain results that are not shifted by even one pixel in template matching.
[0042] (Reference Example 3) Next, a method for detecting edge position 81 of alignment mark 80 according to Reference Example 3 will be described. Fig. 10 is a diagram illustrating alignment mark 80 and the intensity of the brightness gradient in the method for detecting edge position 81 of alignment mark 80 according to Reference Example 3. Fig. 11 is a graph illustrating the intensity of the brightness gradient of ROI 82 in the method for detecting edge position 81 of alignment mark 80 according to Reference Example 3, where the horizontal axis indicates the position of ROI 82 in the x-axis direction and the vertical axis indicates the intensity of the brightness gradient.
[0043] As shown in FIG. 10, the method of Reference Example 3 fits a predetermined function such as a Gaussian function to the profile of the intensity of the luminance gradient in an ROI 82 including a side portion 83 and a peripheral region 84. In this way, the method of Reference Example 3 detects an edge position 81 of an alignment mark 80. However, as shown in FIG. 11, in the method of Reference Example 3, the intensity of the luminance gradient in the dark portion is slightly different from the intensity of the luminance gradient in the bright portion. That is, there is a difference in the amount of noise between a high-luminance region 86H and a low-luminance region 86L. For this reason, the accuracy decreases when a symmetrical function is fitted. The difference in the amount of noise between the high-luminance region 86H and the low-luminance region 86L is considered to be due to photon shot noise in the fitting error expressed by the following formula (2).
[0044]
number
[0045] The first term in the root is the dark shot noise. The second term is the read noise. The third term is the photon shot noise, which is expressed by the following formula (3). Here, QE is the quantum efficiency (constant within the image), N is the photon flux density (dark < light), and t is the exposure time (constant within the image).
[0046]
number
[0047] In this way, in the method of Reference Example 3, there is a difference in height between the flat portions in the high luminance region 86H and the low luminance region 86L, and the fitting error of the Gaussian function increases.
[0048] (Reference Example 4) Next, a method for detecting the edge position 81 of the alignment mark 80 according to Reference Example 4 will be described. FIG. 12 is a graph illustrating an example of a luminance profile of the ROI 82 in the method for detecting the edge position 81 of the alignment mark 80 according to Reference Example 4, where the horizontal axis indicates the position on the ROI 82 in the x-axis direction and the vertical axis indicates the intensity of luminance. In FIG. 12, the position on the ROI 82 in the x-axis direction may indicate the number of a pixel along the x-axis. For example, the 100th position on the x-axis indicates the edge position 81. The position on the -x-axis side of the 100th indicates the position of the peripheral region 84, and the position on the +x-axis side of the 100th indicates the position of the side portion 83 of the alignment mark 80.
[0049] As shown in FIG. 12, the method of Reference Example 4 finds an edge position 81 by fitting a sigmoid function as shown in the following equation (4) to a luminance profile in an ROI 82 including a side portion 83 and a surrounding region 84.
[0050]
number
[0051] However, due to the influence of noise contained in the luminance profile, the luminance is not flat even in the flat portion of the side portion 83. Therefore, variations may occur when fitting a fitting function to the flat portion of the pattern of the side portion 83. Furthermore, if the alignment mark 80 has a large amount of blur or is small, a luminance profile with a sufficient number of pixels cannot be obtained, resulting in reduced accuracy.
[0052] (Embodiment 1) Next, a description will be given of the mounting apparatus and method for detecting the edge position 81 of the alignment mark 80 according to the first embodiment. In the method of this embodiment, a fitting function including a sigmoid curve as shown in the following equation (5) is fitted to the luminance profile.
[0053]
number
[0054] The method for detecting the edge position 81 of this embodiment improves the reproducibility of detection of the edge position 81 by optimizing the parameters used for fitting. Specifically, the pixels of the non-edge portion located in the high brightness region 86H and the low brightness region 86L including the flat portion of the sigmoid curve are expanded within a range in which the fitting function and the shape of the brightness profile do not deviate. At this time, the maximum range of the ROI 82 is included within an allowable range for the calculation time required to detect the edge position 81. This makes it possible to suppress the variation of individual pixels when detecting the edge position 81 of the alignment mark 80.
[0055] First, a mounting device according to the first embodiment will be described. The mounting device of this embodiment aligns the upper and lower objects to be bonded using an upper and lower two-view optical system. Then, the mounting device mounts the upper object to be bonded on the lower object to be bonded. Specifically, for example, the mounting device detects the edge position 81 of the alignment mark 80, thereby detecting the center position of the alignment mark 80. Then, the mounting device aligns and bonds the object to be bonded Ma based on the detection results of the edge position 81 and the center position of the alignment mark 80.
[0056] Fig. 13 is a configuration diagram illustrating the mounting apparatus 1 according to the first embodiment. As shown in Fig. 13, the mounting apparatus 1 includes a bonding head 10, a bonding stage 20, a two-field optical system 30 (upper and lower fields of view), and a processing device 90. Each component of the mounting apparatus 1 other than the processing device 90 may be disposed on a base frame 70. The processing device 90 may also be disposed on the base frame 70.
[0057] The base frame 70 is a reference structure of the mounting device 1. The base frame 70 is, for example, a rectangular parallelepiped having a base 71, an upper frame 72, and a side frame 73. The side frame 73 supports the upper frame 72 on the base 71. Note that the base frame 70 may have other shapes as long as each component of the mounting device 1 can be arranged thereon.
[0058] Here, for convenience of explanation of the mounting apparatus 1, an XYZ orthogonal coordinate axis system is introduced. For example, the direction perpendicular to the upper surface of the base 71 is the Z-axis direction, and two perpendicular directions in a plane parallel to the upper surface of the base 71 are the X-axis direction and the Y-axis direction. The +Z-axis direction is the upward direction, and the -Z-axis direction is the downward direction. Note that the upward and downward directions are for convenience of explanation of the mounting apparatus 1, and do not limit the direction in which the mounting apparatus 1 is arranged when it is actually used. The xyz orthogonal coordinate axis system in which the predetermined surface of the object Ma on which the above-mentioned alignment mark 80 is arranged is the xy plane coincides with the case in which the predetermined surface of the object Ma is parallel to the upper surface of the base 71. Each configuration of the mounting apparatus 1 will be described below.
[0059] <Bonding head> The bonding head 10 holds an object to be bonded Mb. The object to be bonded Mb is a member to be bonded onto the object to be bonded Ma. The object to be bonded Mb is, for example, a member such as a die. Note that the object to be bonded Mb is not limited to a die, and may be a member such as a wafer, a chip, an interposer, etc. The bonding head 10 has a head 11 and a driving mechanism 12.
[0060] The head 11 holds the object to be bonded Mb. For example, the head 11 may suction-hold the object to be bonded Mb. The driving mechanism 12 is fixed to the upper frame 72. The driving mechanism 12 translates the head 11 in the X-axis direction, the Y-axis direction, and the Z-axis direction. At the same time, the driving mechanism 12 may rotate the head 11 about a rotation axis that rotates around the X-axis, the Y-axis, and the Z-axis. In this manner, the bonding head 10 may function as a bonding tool.
[0061] Specifically, the driving mechanism 12 may have, as a function of a bonding tool, translation axes for translating the head 11 in the directions of the X-axis, Y-axis, and Z-axis, as well as rotation axes Tx, Ty, and Tz for rotating the head 11 around each axis. This allows the bonding head 10 to adjust the relative position and parallelism of the upper object to be bonded Mb and the lower object to be bonded Ma. Also, the bonding head 10 can bond the object to be bonded Ma and the object to be bonded Mb.
[0062] <Bonding stage> The bonding stage 20 holds the object to be bonded Ma. The object to be bonded Ma is a member to be bonded to the object to be bonded Mb. The object to be bonded Ma is, for example, a member such as a wafer. Note that the object to be bonded Ma is not limited to a wafer, and may be a member such as a chip, a die, or an interposer. The object to be bonded Ma may be a member that becomes the bottom layer when constituting a laminate. The bonding stage 20 has a stage 21 and a driving mechanism 22.
[0063] The stage 21 holds the object to be bonded Ma. For example, the stage 21 may adsorb the object to be bonded Ma. The driving mechanism 22 is fixed to the base 71. The driving mechanism 22 translates the stage 21 in the X-axis direction and the Y-axis direction. This allows the bonding stage 20 to move the object to be bonded Ma in the X-axis direction and the Y-axis direction. The driving mechanism 22 may translate the stage 21 in the Z-axis direction, or may rotate the stage 21 around a rotation axis that rotates around the X-axis, Y-axis, and Z-axis.
[0064] The bonding stage 20 may have, as a bonding tool, translation axes for translating the stage 21 in the directions of the X-axis, Y-axis, and Z-axis, as well as rotation axes Tx, Ty, and Tz for rotating the stage 21 around each axis, instead of or in addition to the bonding head 10. This allows the bonding stage 20 to adjust the relative position and parallelism of the upper object to be bonded Mb and the lower object to be bonded Ma. Also, the bonding stage 20 can bond the object to be bonded Ma and the object to be bonded Mb.
[0065] Therefore, at least one of the bonding head 10 and the bonding stage 20 functions as a bonding unit that bonds the object to be bonded Mb to the object to be bonded Ma. That is, the mounting apparatus 1 includes a bonding unit. The bonding unit includes at least one of the bonding head 10 and the bonding stage 20.
[0066] <Upper and lower 2-field optical system> The upper and lower two-field optical system 30 is inserted between the bonded object Ma and the bonded object Mb, and photographs the bonded object Ma and the bonded object Mb. For example, the upper and lower two-field optical system 30 may photograph the bonded object Ma and the bonded object Mb simultaneously. The upper and lower two-field optical system 30 is not limited to being inserted between the upper bonded object Mb and the lower bonded object Ma, and photographing the upper bonded object Mb and the lower bonded object Ma, but may be inserted between the left bonded object Ma and the right bonded object Mb, and photographing the left bonded object Ma and the right bonded object Mb, for example. In other words, the upper and lower two-field optical system 30 may photograph not only up and down, but also two opposite directions such as left and right. The upper and lower two-field optical system 30 may also photograph two opposite directions such as left and right simultaneously. In this sense, the upper and lower two-field optical system 30 may simply be called a two-field optical system.
[0067] The upper and lower two-view optical system 30 includes an optical unit 31 and a driving mechanism 32. The driving mechanism 32 is fixed to a base frame 70. The driving mechanism 32 is fixed to, for example, an upper frame 72. The driving mechanism 32 translates the optical unit 31 in the directions of the X-axis, the Y-axis, and the Z-axis. The driving mechanism 32 may rotate the optical unit 31 about a rotation axis that rotates around the X-axis, the Y-axis, and the Z-axis. The driving mechanism 32 moves the optical unit 31 between a plurality of alignment marks 80. The driving mechanism 32 moves the optical unit 31 in the Z-axis direction to adjust the focus of the optical unit 31. The driving mechanism 32 may adjust the inclination of the upper and lower two-view optical system 30.
[0068] <Optical section with two upper and lower field of view optical system> FIG. 14 is a configuration diagram illustrating the optical section 31 of the upper and lower two-view optical system 30 in the mounting device 1 according to the first embodiment. As shown in FIG. 14, the optical section 31 of the upper and lower two-view optical system 30 has a plurality of optical members, lighting, and an image sensor 35 such as a camera. The plurality of optical members include, for example, an objective lens 33a, an objective lens 33b, a tube lens 34a, and a tube lens 34b. An image of the upper object Mb to be bonded is formed on the image sensor 35 via the objective lens 33b and the tube lens 34b. An image of the lower object Ma to be bonded is formed on the image sensor 35 via the objective lens 33a and the tube lens 34a.
[0069] For example, the image sensor 35 may acquire an image of the alignment mark 80 formed on the object to be bonded Ma and the alignment mark 80a formed on the object to be bonded Mb simultaneously. The multiple optical members form an image of each alignment mark on the image sensor 35. Therefore, when the alignment mark 80 of the object to be bonded Ma is the first alignment mark and the alignment mark 80b of the other object to be bonded Mb is the second alignment mark, the upper and lower two-field optical system 30 has one image sensor that acquires an image of the first alignment mark and the second alignment mark. As another configuration, the configuration for forming an image of the upper object to be bonded Mb on the image sensor 35 and the configuration for forming an image of the lower object to be bonded Ma on the image sensor 35 may be configured by a common objective lens and a common imaging lens.
[0070] In Fig. 14, the objects Ma and Mb to be bonded are both shown on the left side, but in reality, as shown in Fig. 13, the objects Ma and Mb to be bonded are arranged to face each other with the optical unit 31 in between. Therefore, some optical members are omitted in Fig. 14. For example, in Fig. 14, the bending of the optical path is minimized in order to explain the upper and lower two-view optical system 30, but in reality, the optical path may be bent appropriately by a plane mirror in order to incorporate it into a compact housing. The optical unit 31 may further include a housing for fixing a plurality of optical members, and lighting for illuminating the alignment marks 80 and 80a.
[0071] <Alignment mark> 15 is a diagram illustrating an alignment mark 80 captured by the image sensor 35 in the mounting device 1 according to the first embodiment. As shown in FIG. 15, for example, the image sensor 35 acquires an image of the alignment mark 80 formed on the object to be bonded Ma and the alignment mark 80a formed on the object to be bonded Mb. The multiple optical members form images of the alignment marks 80 and 80a on the image sensor 35. As another configuration, as described above, the configuration for forming an image of the upper object to be bonded Mb on the image sensor 35 and the configuration for forming an image of the lower object to be bonded Ma on the image sensor 35 may be configured by a common objective lens and a common imaging lens.
[0072] The objects to be bonded Ma and Mb are formed with upper and lower paired alignment marks 80 and 80a. When the objects to be bonded Ma and Mb are large relative to the field of view, the objects to be bonded Ma and Mb are moved to a plurality of locations, and the paired alignment marks 80 and 80a are recognized at each location. At least one of the bonding head 10 and the bonding stage 20 functioning as a bonding tool adjusts at least one of the relative position and the parallelism of the upper object to be bonded Mb and the lower object to be bonded Ma based on the image acquired by the image sensor 35 so as to minimize the positional deviation of the alignment marks 80 and 80a at one or more pairs of positions, and bonds the objects to be bonded Ma and Mb.
[0073] The captured image of the alignment marks 80 and 80a may include a portion corresponding to another mark 87. The portion corresponding to another mark 87 includes, for example, other alignment marks, circuit patterns, etc. formed on the objects to be bonded Ma and Mb.
[0074] <Processing equipment> Fig. 16 is a block diagram illustrating a processing device 90 in the mounting device 1 according to the first embodiment. As shown in Fig. 16, the processing device 90 includes a profile acquisition unit 91, a fitting unit 92, and a position calculation unit 93. The processing device 90 may be an information processing device such as a microcomputer, a personal computer, or a server. The processing device 90 may further include a processor PRC, a memory MMR, a storage device STR, and a user interface UI.
[0075] The storage device STR may store the processes executed by each component of the processing device 90 as a program. The processor PRC may also load the program from the storage device STR into the memory MMR and execute the program. In this way, the processor PRC realizes the functions of each component of the processing device 90, such as the profile acquisition unit 91, the fitting unit 92, and the position calculation unit 93. The user interface UI may include input devices such as a keyboard, a mouse, and an imaging device, and output devices such as a display, a printer, and a speaker.
[0076] Each of the components of the processing device 90 may be realized by dedicated hardware. Also, a part or all of each component may be realized by a general-purpose or dedicated circuit, a processor PRC, or a combination of these. These may be configured by a single chip, or may be configured by multiple chips connected via a bus. A part or all of each component may be realized by a combination of the above-mentioned circuits, etc. and a program. Also, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an FPGA (Field-programmable Gate Array), a quantum processor (quantum computer control chip), etc. may be used as the processor PRC.
[0077] Furthermore, when some or all of the components of the processing device 90 are realized by a plurality of information processing devices, circuits, etc., the plurality of information processing devices, circuits, etc. may be centrally arranged or distributed. For example, the information processing devices, circuits, etc. may be realized in a form in which each is connected via a communication network by a client-server system, a cloud computing system, etc. Furthermore, the functions of the processing device 90 may be provided in a SaaS (Software as a Service) format.
[0078] FIG. 17 is a diagram illustrating an image of an area including an alignment mark 80 formed on the object to be bonded Ma in the mounting device 1 according to the first embodiment. As shown in FIG. 17, the profile acquisition unit 91 acquires an ROI 82 including a high brightness area 86H and a plurality of low brightness areas 86L from an image of an area including an alignment mark 80 formed on the object to be bonded Ma. The high brightness area 86H is a portion corresponding to the alignment mark 80 in the image. The plurality of low brightness areas 86L are portions located on both sides of the alignment mark 80 in the x-axis direction. The plurality of low brightness areas 86L are, for example, portions of the object to be bonded Ma that are not alignment marks. The low brightness area 86L is lower in brightness than the high brightness area 86H. The plurality of low brightness areas 86L sandwich the high brightness area 86H from both sides in the x-axis direction. The x-axis direction may be replaced with the y-axis direction. In the following description, the x-axis direction may also be replaced with the y-axis direction. The edge position 81L is located between the high brightness region 86H and one low brightness region 86L, and the edge position 81R is located between the high brightness region 86H and the other low brightness region 86L. In other words, the profile acquisition unit 91 acquires a brightness profile along one direction of the ROI 82 including the first region, the second region, and the third region aligned along one direction in the image from an image of a region including the alignment mark 80 formed on the bonding object Ma. The second region corresponds to the alignment mark 80. The first region, the third region, and the second region have a difference in brightness.
[0079] 18 is a graph illustrating a luminance profile and a fitting function in the mounting apparatus 1 according to the embodiment 1, where the horizontal axis indicates the position on the ROI 82 in the x-axis direction and the vertical axis indicates the intensity of luminance. As shown in FIG 18, the profile acquiring unit 91 acquires a luminance profile along the x-axis direction of the ROI 82 from an image captured of a region including the alignment mark 80 formed on the object to be bonded Ma.
[0080] The fitting unit 92 fits the luminance profile with a fitting function including a sigmoid function having an inflection point and a curvature. Then, the fitting unit 92 sets the inflection points as the edge positions 81L and 81R of the alignment mark 80. Here, the fitting function is the above-mentioned formula (5).
[0081] Fig. 19 is a graph illustrating a fitting function in the mounting device 1 according to the first embodiment, where the horizontal axis indicates the position on the ROI 82 in the x-axis direction and the vertical axis indicates the intensity of luminance. Fig. 20 is a diagram illustrating a curvature depending on the coefficient of the fitting function in the mounting device 1 according to the first embodiment. As shown in Figs. 19 and 20, μ 1 and μ 2 indicates the inflection point, and a 1 and a 2 indicates the curvature and positive or negative, and b and c indicate constants. Note that the character for the curvature in the denominator of exp() in the above formula is written as a in the specification due to font restrictions. 1 and a 2 This indicates that.
[0082] a 1 and a 2 If is positive, the fitting function increases. 1 and a 2 If is negative, the fitting function decreases. 1 and a 2 The larger the absolute value of a, the more gradual the step between the low-luminance region 86L and the high-luminance region 86H becomes. 1 and a 2 The smaller the absolute value of is, the steeper the step between the low luminance region 86L and the high luminance region 86H becomes.
[0083] The ROI 82 includes a first edge portion having an edge of the alignment mark 80 between the high brightness region 86H and one low brightness region 86L, and a second edge portion having an edge between the high brightness region 86H and the other low brightness region 86L. The fitting function is 1 and curvature a 1 and the inflection point μ of the second edge portion. 2 and curvature a 2 and a second sigmoid function having the following function:
[0084] The fitting unit 92 calculates μ so that the error between the intensity of each point of the luminance profile and the value of the fitting function is minimized. 1 , μ 2 , a 1 , a 2 , b and c are estimated. The initial values for these are as follows: 1 The initial value of μ is the position where the output value obtained by applying a Sobel filter to the luminance profile is maximum. 2 The initial value is the position where the output value obtained by applying a Sobel filter to the luminance profile is the minimum. 1 , a 2 , b, and c use the results of images captured in advance as initial values. The estimation method uses the downhill simplex method, but there are no limitations.
[0085] 21 is a diagram illustrating an example of an edge position 81 detected in the mounting apparatus 1 according to the first embodiment. As shown in FIG. 21, an edge position 81L of the alignment mark 80 is a function of the x-axis direction. 1 x The edge position 81R is μ in the x-axis direction. 2 x The same process is performed on the Y-axis direction. That is, the two edge positions 81 of the alignment mark 80 in the y-axis direction are determined by the distance μ 1 y and μ 2 y It is calculated from:
[0086] FIG. 22 is a diagram illustrating a center position calculated by the position calculation unit 93 in the mounting device 1 according to the first embodiment. As shown in FIG. 22, the position calculation unit 93 calculates the center position of the alignment mark 80 based on the detected edge position 81L and edge position 81R. For example, the position calculation unit 93 calculates the center position of the alignment mark 80 in the x-axis direction from the following formula (6). The position calculation unit 93 also calculates the center position of the alignment mark 80 in the y-axis direction from the following formula (7). Specifically, as an example, the high brightness area 86H corresponding to the alignment mark 80 may be a quadrangle having sides along one direction and another direction intersecting the one direction. The position calculation unit 93 calculates a first center position of the alignment mark 80 in one direction based on the edge position 81 detected in one direction, and calculates a second center position of the alignment mark 80 in the other direction based on the edge position 81 detected in the other direction.
[0087]
number
number
[0088] The bonding unit bonds the object Mb to the object Ma by using the calculated center position of the alignment mark 80. The bonding unit may bond the object Mb to the object Ma by using at least one of the edge position 81L and the edge position 81R of the alignment mark 80. The bonding unit includes at least one of the bonding head 10 and the bonding stage 20.
[0089] <Implementation method> Next, a description will be given of a mounting method using the mounting apparatus 1. Fig. 23 is a flow chart illustrating the mounting method using the mounting apparatus 1 according to the first embodiment.
[0090] As shown in step S11 of FIG. 23, the lower object to be bonded Ma is held by the bonding stage 20.
[0091] Next, as shown in step S12, the upper object to be bonded Mb is held by the bonding head 10. Specifically, for example, a member supply unit such as a die lifter is disposed on the bonding stage 20. Therefore, by moving the bonding stage 20, the die lifter is positioned below the bonding head 10. This allows the head 11 of the bonding head 10 to grip the upper object to be bonded Ma from the die lifter.
[0092] Next, as shown in step S13, the mounting section is moved. Specifically, the mounting section for the lower object Ma is moved below the bonding head 10 by moving the bonding stage 20.
[0093] Next, as shown in step S14, the upper and lower two-field optical system 30 is inserted between the upper object to be bonded Mb and the lower object to be bonded Ma. Next, as shown in step S15, the objects to be bonded Ma and Mb are photographed by the image sensor 35 of the upper and lower two-field optical system 30. The objects to be bonded Ma and Mb may be photographed simultaneously. Specifically, the alignment marks 80 and 80a are captured within the field of the upper and lower two-field optical system 30 inserted between the objects to be bonded Ma and Mb. As a result, the image sensor 35 captures an image including the alignment marks 80 and 80a. In this way, an image of the alignment mark 80 formed on the object to be bonded Ma and the alignment mark 80a formed on the object to be bonded Mb photographed by one image sensor 35 in the upper and lower two-field optical system 30 is acquired.
[0094] Next, as shown in step S16, the positional deviation between the upper alignment mark 80 and the lower alignment mark 80a is detected. Specifically, the center positions of the alignment marks 80 and 80a shown in the acquired image are detected by a detection flow described later.
[0095] Next, as shown in step S17, the positional deviation is adjusted. Specifically, using the center positions detected based on the images of the alignment marks 80 and 80a acquired by the image sensor 35, at least one of the bonding head 10 and the bonding stage 20 is caused to adjust the relative positions of the objects Ma and Mb.
[0096] Next, as shown in step S18, the upper object to be bonded Mb and the lower object to be bonded Ma are bonded. Specifically, for example, a bonding head 10 as a bonding tool is lowered, and the upper object to be bonded Mb is pressure-welded to the lower object to be bonded Ma. In this way, the object to be bonded Mb is bonded to the object to be bonded Ma using the detected edge position 81.
[0097] Next, a method for detecting the positional deviation of the alignment mark 80 in step S16 described above will be described. The positional deviation of the alignment mark 80 is detected by detecting the center position of the alignment mark 80. FIG. 24 is a flow chart illustrating a method for detecting the center position of the alignment mark 80 according to the first embodiment. FIG. 25 is a diagram illustrating an example of an ROI 82 cut out by the profile acquisition unit 91 in the mounting method according to the first embodiment. FIG. 26 is a diagram illustrating data obtained by averaging the brightness of the ROI 82 in the mounting method according to the first embodiment.
[0098] 24 and 25, the profile acquisition unit 91 extracts an ROI 82 from an image including the alignment mark 80. The ROI 82 includes a high-luminance region 86H corresponding to the alignment mark 80, and a plurality of low-luminance regions 86L sandwiching the high-luminance region 86H from both sides in the x-axis direction. The plurality of low-luminance regions 86L correspond to portions of the object to be bonded Ma where the alignment mark 80 is not present.
[0099] Next, as shown in step S22, the profile acquisition unit 91 may perform noise reduction. For example, the profile acquisition unit 91 may perform noise reduction by using a Gaussian filter.
[0100] Next, as shown in step S23 and FIG. 26, the profile acquisition unit 91 performs averaging in the vertical direction. For example, the profile acquisition unit 91 averages the luminance in the y-axis direction to form one-dimensional data along the x-axis direction. The profile acquisition unit 91 may also average the luminance in the x-axis direction to form one-dimensional data along the y-axis direction. In this manner, from an image of a region including the alignment mark 80 formed on the bonding object Ma, a luminance profile along the x-axis direction of the ROI 82 including the high luminance region 86H and a plurality of low luminance regions 86L located on both sides of the alignment mark 80 in the x-axis direction is acquired. In other words, from an image of a region including the alignment mark 80 formed on the bonding object Ma, a luminance profile along one direction of the region of interest including the first region, the second region, and the third region aligned along one direction in the image is acquired. The second region corresponds to the alignment mark 80. The first region, the third region, and the second region have a difference in luminance.
[0101] Next, as shown in step S24 and in FIG. 18, the fitting unit 92 fits the luminance profile with a fitting function including a sigmoid function having an inflection point and a curvature. The fitting unit 92 fits the luminance profile with a fitting function including a sigmoid function having an inflection point and a curvature so as to minimize the error between the intensity of each point of the luminance profile and the value of the fitting function. 1 , μ 2 , a 1 , a 2 , b and c are estimated. 1 and μ 2 indicate the edge positions 81L and 81R of the alignment mark 80, respectively. That is, the inflection point μ 1 and μ 2 The edge positions 81L and 81R of the alignment mark 80 are detected from the edge positions 81L and 81R.
[0102] 22, the position calculation unit 93 calculates the center position of the alignment mark 80 based on the calculated edge position 81. When calculating the center position of the alignment mark 80, the position calculation unit 93 may calculate the center position in the x-axis direction and the center position in the y-axis direction. In this manner, the center position of the alignment mark 80 is detected, thereby detecting the positional deviation of the alignment mark 80.
[0103] Next, a method for determining the size of the ROI 82 will be described. The size of the ROI 82 is the size of the luminance profile when fitting with the fitting function. Fig. 27 is a graph illustrating an example of repeatability when the size of the ROI 82 is changed in the mounting method according to the first embodiment, where the horizontal axis indicates the ratio of the length of the ROI 82 to the length of the alignment mark 80 in the x-axis direction, and the vertical axis indicates repeatability. The "Repeatability (%)" index on the vertical axis indicates that the repeatability is better as it becomes smaller.
[0104] As shown in FIG. 27, the larger the size of ROI 82 in the x-axis direction, the better the reproducibility of the calculated edge position 81 of alignment mark 80. For example, the more the length of ROI 82 relative to the length of alignment mark 80 increases to 200%, 300%, the better the reproducibility of edge position 81. This is based on a concept conforming to the law of large numbers. In other words, the luminance values of each pixel vary. Even with such variation, the μ 1 and μ 2 can be converged to a predetermined average value. Therefore, from the viewpoint of accuracy, it is preferable that the size of ROI 82 is large.
[0105] On the other hand, there are two drawbacks to a large size of the ROI 82: an increase in error in function fitting due to the inclusion of another mark 87, and an increase in the calculation time required to detect the edge position 81. Therefore, in order to perform highly accurate alignment, it is necessary to determine the maximum range of the size of the ROI 82 taking into account at least the first drawback.
[0106] Regarding the first point, we will explain the increase in error in function fitting due to the inclusion of another mark 87. Fig. 28 is a graph illustrating a case in which the size of ROI 82 increases and the accuracy decreases in the mounting method according to the first embodiment, where the horizontal axis indicates the ratio of the length of ROI 82 to the length of alignment mark 80 in the x-axis direction, and the vertical axis indicates the error.
[0107] 28, consider the case where another mark 87 is present near the alignment mark 80, which is the detection target of the edge position 81. Even if the size of the ROI 82 increases so as to approach the other mark 87, the intensity profile maintains the shape of equation (5) until it includes the other mark 87. Therefore, the fitting error is constant.
[0108] However, when the ROI 82 includes another mark 87, a new high-luminance region is added to the luminance profile. Then, the shape of the assumed luminance profile and the shape of the fitting function differ. For this reason, the fitting error between the luminance profile and the fitting function increases. Therefore, while changing the size of the ROI 82, edge position detection processing including noise reduction, vertical averaging, and fitting is performed on luminance profiles of various sizes. The maximum range in which the fitting error is equal to or smaller than the threshold value is set as the size of the ROI 82. The threshold value is set as the fitting error that can achieve the desired accuracy while changing the size of the ROI 82. Alternatively, the user may set the threshold value.
[0109] In this way, the bonding object Ma around the alignment mark 80 may have another mark 87 with higher brightness than the low brightness region 86L. When the ROI 82 includes the mark 87, the error between the brightness profile and the fitting function becomes larger than a predetermined value. Therefore, the profile acquisition unit 91 acquires the brightness profile of the ROI 82 in the maximum range that satisfies that the error between the brightness profile and the fitting function is equal to or smaller than the predetermined value. Note that the above-mentioned another mark 87 is a mark when the second region is the high brightness region 86H and the first and third regions are the low brightness regions 86L. When the second region is the low brightness region 86L and the first and third regions are high brightness regions, the another mark 87 is lower in brightness than the first and third regions. Therefore, the image may have a fourth region around the second region corresponding to the alignment mark 80 that is higher or lower in brightness than the first and third regions and corresponds to another mark.
[0110] Regarding the second point, an explanation will be given of an increase in the calculation time for detecting the edge position 81. Fig. 29 is a graph illustrating an example of the relationship between the size of ROI 82 and the calculation time in the mounting method according to the first embodiment, where the horizontal axis indicates the ratio of the length of ROI 82 to the length of alignment mark 80 in the x-axis direction, and the vertical axis indicates the calculation time.
[0111] As shown in FIG. 29, the larger the size of the ROI 82, the longer the calculation time. This is because the number of times data is accessed increases. Therefore, the maximum size that requires a calculation time equal to or less than a threshold is set as the size of the ROI 82. The threshold may be predefined as a specification of the mounting device 1, or may be set by the user. In this manner, the profile acquisition unit 91 acquires a luminance profile of the maximum range of the ROI 82 that satisfies that the calculation time for detecting the edge position 81 of the alignment mark 80 is equal to or less than a predetermined time.
[0112] As described above, the size of the ROI 82 is determined based on at least one of the range in which a brightness profile that does not deviate from the shape of the expected fitting function is obtained and the range in which the calculation time is acceptable. This allows for edge detection with high reproducibility.
[0113] As another method for determining the size of the ROI 82, the processing device 90 of the mounting device 1 may further include a GUI (Graphical User Interface). The user may determine the size of the ROI 82 using the GUI.
[0114] FIG. 30 is a diagram illustrating a GUI in the mounting device 1 according to the first embodiment. As shown in FIG. 30, when a user sets an ROI 82, the ROI 82 may be displayed on the GUI. The user may change the size of the ROI 82 on the GUI by operating a mouse or the like. The GUI may display the ROI 82 determined by the above-mentioned method for determining the size of the ROI 82 as a separate display such as a dotted line and a solid line, or may display an indicator of the policy by which the user determines the size of the ROI 82. In addition, the quality of each ROI 82 may be displayed on the GUI with a symbol such as good (◎), good (◯), standard (△), or bad (×).
[0115] When determining the size of these ROIs 82, the size may be determined once and continue to be used as long as the pattern of the alignment mark 80 of the object to be bonded Ma is the same. Therefore, the size may be determined in advance other than the processing flow for bonding the object to be bonded Ma. However, the timing for determining the size of the ROI 82 is not limited to this. For example, when the pattern of the alignment mark 80 of the object to be bonded Ma changes frequently, the size of the ROI 82 may be determined before cutting out the ROI 82.
[0116] Next, the effects of this embodiment will be described. The mounting device 1 of this embodiment fits a fitting function including a sigmoid curve to the luminance profile of a large number of pixels in an image including the alignment mark 80. This allows the mounting device 1 to reduce the variation when detecting the center position of the alignment mark 80, and improve the reproducibility of the detection of the alignment mark 80.
[0117] In particular, since the number of pixels constituting the luminance profile can be increased, it is possible to make the luminance profile conform to the law of large numbers, thereby improving the reproducibility of detecting the center position of the alignment mark 80.
[0118] Furthermore, since the luminance profile includes the edge positions 81L and 81R at both ends of the alignment mark 80, fitting can be performed using more pixels than when only one edge position 81 is included. Specifically, the high luminance region 86H and the low luminance region 86L can be used for fitting to the maximum extent. This can improve reproducibility, particularly when the alignment mark 80 has a large amount of blur or is small. As a result, the accuracy of alignment can be improved, and high-precision joining can be performed.
[0119] Furthermore, the profile acquisition unit 91 acquires a luminance profile of the maximum range of ROI that satisfies that the error between the luminance profile and the fitting function is equal to or less than a predetermined value. This makes it possible to reduce the fitting error. The profile acquisition unit 91 may also acquire a luminance profile of the maximum range of ROI that satisfies that the calculation time for detecting the edge position 81 is equal to or less than a predetermined time. This makes it possible to shorten the calculation time.
[0120] The mounting device 1 further includes an upper and lower two-view optical system that is inserted between the objects Ma and Mb and captures the objects Ma and Mb simultaneously. Therefore, alignment can be performed from an image that captures the alignment mark 80 formed on the object Ma and the alignment mark 80a formed on the object Mb simultaneously, and mounting can be performed with high accuracy.
[0121] The present disclosure is not limited to the above-described embodiment, and can be appropriately modified without departing from the spirit of the present disclosure. For example, the configurations of the first embodiment can be omitted or combined as appropriate. In addition, the following implementation program for causing a computer to execute the implementation method is also included in the technical concept of the present embodiment.
[0122] (Appendix 1) acquiring, from an image of a region including an alignment mark formed on an object to be bonded, a luminance profile along one direction of a region of interest including a first region, a second region, and a third region aligned along the one direction in the image; fitting a fitting function including a sigmoid function having an inflection point and a curvature to the luminance profile, and detecting an edge position of the alignment mark from the inflection point; a step of joining another workpiece to the workpiece by using the detected edge position; An implementation program for causing a computer to execute the the second region corresponds to the alignment mark; the first region, the third region, and the second region have a difference in luminance; The fitting function is TIFF2024170031000011.tif17105, μ 1 and μ 2 indicates the inflection point, a 1 and a 2 denotes the curvature, b and c are constants. Implementation program. (Appendix 2) The mounting program according to claim 1, further causing a computer to execute a step of calculating a center position of the alignment mark based on the detected edge position. (Appendix 3) the second region has a rectangular shape having sides along the one direction and another direction intersecting the one direction, In the step of calculating the center position of the alignment mark, calculating a first center position of the alignment mark in the one direction based on the edge position detected in the one direction, and calculating a second center position of the alignment mark in the other direction based on the edge position detected in the other direction; An implementation program as described in Appendix 2. (Appendix 4) In the step of acquiring the luminance profile, acquiring the luminance profile of the region of interest of the maximum range that satisfies at least one of the following: an error between the luminance profile and the fitting function is equal to or less than a predetermined value; and a calculation time for detecting the edge position is equal to or less than a predetermined time. 2. An implementation program as described in Appendix 1. (Appendix 5) the image has a fourth region around the second region, the fourth region being brighter or darker than the first region and the third region, the fourth region corresponding to another mark, When the region of interest includes the fourth region, an error between the luminance profile and the fitting function becomes greater than a predetermined value. Implemented program according to appendix 4. (Appendix 6) the region of interest includes a first edge portion having an edge of the alignment mark between the first region and one of the second regions, and a second edge portion having the edge between the first region and the other of the second regions; the fitting function includes a first sigmoid function having the inflection point and the curvature of the first edge portion, and a second sigmoid function having the inflection point and the curvature of the second edge portion. 2. An implementation program as described in Appendix 1. (Appendix 7) holding the workpiece by a bonding head; a step of holding another object to be bonded to the object to be bonded by a bonding stage; A step of inserting an upper and lower two-view optical system between the object to be bonded and the other object to be bonded; a step of acquiring an image of the first alignment mark and the second alignment mark by one image sensor in the upper and lower two field of view optical system, when the alignment mark of the object to be bonded is a first alignment mark and the alignment mark of the other object to be bonded is a second alignment mark; The implementation program according to claim 1, further causing a computer to execute the above steps.
[0123] The program includes a set of instructions (or software code) that, when read into the processor 90, causes the computer to perform one or more functions described in the embodiments. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, the computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disc (DVD), Blu-ray® disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, the transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals. [Explanation of symbols]
[0124] 1 Mounting equipment 10 Bonding Head 11 Head 12 Drive mechanism 20 Bonding Stage 21 Stage 22 Drive mechanism 30 Upper and lower two-view optical system 31 Optics Department 32 Drive mechanism 33a, 33b Objective lenses 34a, 34b Imaging lenses 35 Image Sensor 70 Base Frame 71 Foundation 72 Upper Frame 73 Side frame 80, 80a alignment marks 81 Edge Position 82 ROI 83 Side 84 Surrounding Areas 85 High Part 86H High brightness area 86L Low brightness area 87 Marks 90 Processing Equipment 91 Profile Acquisition Section 92 Fitting section 93 Position calculation section Ma, Mb welded material MMR Memory PRC Processor STR storage UI User Interface
Claims
1. a profile acquisition unit that acquires, from an image captured of a region including an alignment mark formed on an object to be bonded, a luminance profile along one direction of a region of interest including a first region, a second region, and a third region that are aligned along the one direction in the image; a fitting unit that fits a fitting function including a sigmoid function having an inflection point and a curvature to the luminance profile and detects an edge position of the alignment mark from the inflection point; a joining section that joins another workpiece to the workpiece by using the detected edge position; Equipped with the second region corresponds to the alignment mark; the first region, the third region, and the second region have a difference in luminance; The fitting function is and μ 1 and μ 2 indicates the inflection point, a 1 and a 2 denotes the curvature, b and c are constants. Mounting equipment.
2. The alignment mark detecting device further includes a position calculating unit that calculates a center position of the alignment mark based on the detected edge position. The mounting device according to claim 1 .
3. the second region has a rectangular shape having sides along the one direction and another direction intersecting the one direction, the position calculation unit calculates a first center position of the alignment mark in the one direction based on the edge position detected in the one direction, and calculates a second center position of the alignment mark in the other direction based on the edge position detected in the other direction; The mounting device according to claim 2 .
4. the profile acquisition unit acquires the luminance profile of the maximum range of the region of interest that satisfies at least one of the following: an error between the luminance profile and the fitting function is a predetermined value or less; and a calculation time for detecting the edge position is a predetermined time or less. The mounting device according to claim 1 .
5. the image has a fourth region around the second region, the fourth region being brighter or lower than the first region and the third region, the fourth region corresponding to another mark, When the region of interest includes the fourth region, an error between the luminance profile and the fitting function becomes greater than a predetermined value. The mounting device according to claim 4 .
6. the region of interest includes a first edge portion having an edge of the alignment mark between the first region and one of the second regions, and a second edge portion having the edge between the first region and the other of the second regions; the fitting function includes a first sigmoid function having the inflection point and the curvature of the first edge portion, and a second sigmoid function having the inflection point and the curvature of the second edge portion. The mounting device according to claim 1 .
7. A bonding head that holds the object to be bonded; a bonding stage for holding another object to be bonded to the object to be bonded; an upper and lower two-view optical system that is inserted between the object to be bonded and the other object to be bonded and captures images of the object to be bonded and the other object to be bonded; Further equipped with When the alignment mark of the object to be bonded is a first alignment mark and the alignment mark of the other object to be bonded is a second alignment mark, the upper and lower two-field optical system has one image sensor that captures an image of the first alignment mark and the second alignment mark; The mounting device according to claim 1 .
8. acquiring, from an image of a region including an alignment mark formed on an object to be bonded, a luminance profile along one direction of a region of interest including a first region, a second region, and a third region aligned along the one direction in the image; fitting a fitting function including a sigmoid function having an inflection point and a curvature to the luminance profile, and detecting an edge position of the alignment mark from the inflection point; a step of joining another workpiece to the workpiece by using the detected edge position; Equipped with the second region corresponds to the alignment mark; the first region, the third region, and the second region have a difference in luminance; The fitting function is and μ 1 and μ 2 indicates the inflection point, a 1 and a 2 denotes the curvature, b and c are constants. How to implement it.
9. The method further comprises calculating a center position of the alignment mark based on the detected edge position. The mounting method according to claim 8.
10. the second region has a rectangular shape having sides along the one direction and another direction intersecting the one direction, In the step of calculating the center position of the alignment mark, calculating a first center position of the alignment mark in the one direction based on the edge position detected in the one direction, and calculating a second center position of the alignment mark in the other direction based on the edge position detected in the other direction; The mounting method according to claim 9.
11. In the step of acquiring the luminance profile, acquiring the luminance profile of the region of interest of the maximum range that satisfies at least one of the following: an error between the luminance profile and the fitting function is equal to or less than a predetermined value; and a calculation time for detecting the edge position is equal to or less than a predetermined time. The mounting method according to claim 8.
12. the image has a fourth region around the second region, the fourth region being brighter or lower than the first region and the third region, the fourth region corresponding to another mark, When the region of interest includes the fourth region, an error between the luminance profile and the fitting function becomes greater than a predetermined value. The mounting method according to claim 11.
13. the region of interest includes a first edge portion having an edge of the alignment mark between the first region and one of the second regions, and a second edge portion having the edge between the first region and the other of the second regions; the fitting function includes a first sigmoid function having the inflection point and the curvature of the first edge portion, and a second sigmoid function having the inflection point and the curvature of the second edge portion. The mounting method according to claim 8.
14. holding the workpiece by a bonding head; a step of holding another object to be bonded to the object to be bonded by a bonding stage; A step of inserting an upper and lower two-view optical system between the object to be bonded and the other object to be bonded; a step of acquiring an image of the first alignment mark and the second alignment mark by one image sensor in the upper and lower two-field optical system, in which the alignment mark of the object to be bonded is a first alignment mark and the alignment mark of the other object to be bonded is a second alignment mark; Further equipped with The mounting method according to claim 8.
Citation Information
Patent Citations
Car braking method
JP1980063942A
Coupling member for condenser microphone
JP1983076000A
Support detecting device
JP1988055487A
