Image processing system, inspection system, and image processing method

Abstract

An image processing system according to the present invention processes at least one of image information on a semiconductor device and image information on a probe card. This image processing system comprises: a calculation unit that calculates information related to a matching rate between a comparison image that is the acquired image information and a registered image held in advance; a comparison unit that compares the calculated information related to the matching rate with a threshold value held in advance; and a registration unit that, when the information related to the matching rate is less than the threshold value, adds the comparison image to the registered image and updates the registered image to a new registered image, and when the information related to the matching rate is equal to or greater than the threshold value, does not add the comparison image to the registered image.

Description

Image processing system, inspection system, and image processing method

[0001] The present disclosure relates to an image processing system, an inspection system, and an image recognition method.

[0002] Patent Document 1 discloses an inspection system (inspection device system) that includes multiple inspection devices and a data processing device capable of communicating with each of the inspection devices. This inspection system acquires setup data (device parameters) and index data from at least one of the multiple inspection devices and adjusts the setup data to have a high contribution to the index data. This increases the success rate of the alignment process between the substrate and the probes of the probe card.

[0003] Patent No. 7398930

[0004] The present disclosure provides a technique that can appropriately set a recipe for substrate processing by automatically improving setup data.

[0005] According to one aspect of the present disclosure, an image processing system is provided that processes at least one of image information of a semiconductor device and image information of a probe card, the image processing system comprising: a calculation unit that calculates information related to the match rate between a comparison image, which is the acquired image information, and a pre-stored registered image; a comparison unit that compares the calculated information related to the match rate with a pre-stored threshold; and a registration unit that performs processing to add the comparison image to the registered image and update it to a new registered image if the information related to the match rate is less than the threshold, but not add the comparison image to the registered image if the information related to the match rate is greater than or equal to the threshold.

[0006] According to one aspect, the setup data is automatically improved to appropriately set a recipe for substrate processing.

[0007] 10 is a diagram illustrating the overall configuration of an inspection system including an image processing system according to an embodiment. FIG. 11 is a diagram illustrating an example of an inspection device applied to the inspection system. FIG. 12 is a plan view showing an alignment bridge and a substrate. FIG. 13 is a diagram illustrating an example of image information captured by a camera. FIG. 14 is a block diagram showing functional units formed inside the image processing system. FIG. 15 is a diagram illustrating an example of a score set in a registered image. FIG. 16 is a diagram illustrating an example of a comparison image having a high matching rate with the registered image. FIG. 17 is a diagram illustrating an example of a comparison image having a low matching rate with the registered image. FIG. 18 is a diagram illustrating a first comparative example of a registered image and a comparison image. FIG. 19 is a diagram illustrating a second comparative example of a registered image and a comparison image. FIG. 19 is a diagram illustrating a third comparative example of a registered image and a comparison image. FIG. 10 is a diagram illustrating a procedure for adding a registered image. FIG. 11 is a diagram illustrating a procedure for adding a registered image. FIG. 12 is a diagram illustrating a procedure for adding a registered image. FIG. 13 is a diagram illustrating a procedure for adding a registered image. FIG. 14 is a diagram illustrating a procedure for adding a registered image. FIG. 15 is a diagram illustrating a procedure for adding a registered image. FIG. 16 is a diagram illustrating a procedure for adding a registered image. FIG. 17 is a diagram illustrating a procedure for adding a registered image. FIG. 18 is a diagram illustrating a procedure for adding a registered image. FIG. 19 is a diagram illustrating a process flow for registering a registered image in an image processing method. FIG. 19 is a diagram illustrating a process flow for determining whether to add a comparison image in an image processing method.

[0008] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In the drawings, the same components are denoted by the same reference numerals, and redundant explanations may be omitted.

[0009] An image processing system according to an embodiment acquires and processes image information of a substrate, which is a semiconductor device, and creates setup data for setting a substrate processing recipe based on the image information. The substrate processing recipe includes, for example, alignment processing for adjusting the target position (horizontal position) and target attitude (rotational position) of the substrate during substrate inspection. The following describes a representative image processing system that is applied when performing this substrate alignment processing.

[0010] Note that substrate processing is not limited to substrate inspection, but may also include film formation, etching, modification, cleaning, bonding, stripping, ashing, and other processes performed by a substrate processing apparatus. Even in substrate processing apparatuses that perform such substrate processing, substrate alignment may be performed before the substrate is placed. Alternatively, the alignment process may be performed in a dedicated aligner device before the substrate is loaded into the substrate processing apparatus, and the technology disclosed herein may be applied to this aligner device. Furthermore, the substrate processing is not limited to alignment processing. Therefore, the substrate processing recipe to be set may include various elements depending on the content of the substrate processing. For example, the substrate processing recipe to be set may include the timing of operation of various components, the duration of the substrate processing, the temperature of the substrate, the amount of power and processing gas supplied to the substrate, the pressure within the processing chamber, the amount of plasma generated, and the like.

[0011] 1 , an image processing system 2 applied to alignment processing for substrate inspection is applied to, for example, an inspection system 100 having a plurality of inspection devices 1 that inspect substrates. The inspection system 100 includes an image processing system 2 communicatively connected to each of the inspection devices 1, and a recipe server 3 communicatively connected to each of the inspection devices 1 and the image processing system 2.

[0012] To facilitate understanding of the present disclosure, an example of an inspection device 1 applied to the inspection system 100 will be described below with reference to FIG.

[0013] Each inspection apparatus 1 (inspection apparatuses 1A to 1N) inspects the electrical characteristics of a substrate W. For example, a plurality of devices under test (DUTs) are formed on the surface of the substrate W. The type of substrate W inspected by the inspection apparatus is not particularly limited, and may be a carrier on which a device under test is arranged, a glass substrate, a single chip, an electronic circuit board, or the like.

[0014] The inspection device 1 includes an inspection unit 10 that actually performs the inspection, a loader 13 installed adjacent to the inspection unit 10, and a tester 20 installed above the inspection unit 10. Furthermore, the inspection device 1 has a controller 90 that controls the operations of the inspection unit 10, the loader 13, and the tester 20.

[0015] The inspection unit 10 includes a rectangular parallelepiped housing 11, and an inspection chamber 12 is disposed inside the housing 11. The inspection chamber 12 accommodates a stage 30 on which a substrate W is placed and which transports the substrate W to a desired three-dimensional coordinate position.

[0016] A container such as a FOUP (Front-Opening Unified Pod) for storing a plurality of substrates W is set in the loader 13. The loader 13 is equipped with a transport device (not shown), which removes the substrates W from the container and transfers them to the stage 30 in the inspection chamber 12. The loader 13 also receives the inspected substrates W from the stage 30 by the transport device and stores them in the container.

[0017] The inspection unit 10 includes a probe card 21 above the inspection chamber 12, which is connected to the tester 20 via an interface 23. The probe card 21 has a plurality of probes 22 at a position facing the substrate W. When the substrate W is moved by the stage 30, each probe 22 comes into contact with an electrode pad, solder bump, or the like of each semiconductor device on the substrate W. This allows the tester 20 to output power and various signals to each semiconductor device via the probe card 21 and the interface 23, and also receive signals transmitted from each semiconductor device via the probe card 21 and the interface 23.

[0018] The tester 20 includes a motherboard (not shown) connected to the interface 23. The motherboard has a plurality of test boards (not shown) mounted thereon and is connected to the controller 90. The tester determines the quality of each semiconductor device based on signals transmitted from each semiconductor device on the substrate W on the motherboard and each test board. By appropriately changing the test board, the tester 20 can perform a plurality of types of tests.

[0019] The stage 30 installed in the inspection chamber 12 includes a mounting table 31 having a flat mounting surface 30s capable of supporting the substrate W, a moving unit 32 for moving the mounting table 31, a stage control unit, a motor driver unit, etc. The stage 30 may also include a stage-side camera 19 for capturing an image of the attachment state of the probe card 21 or a contact state between each probe 22 and the substrate W.

[0020] The mounting table 31 is configured by stacking structures such as a bottom plate supported by the moving part 32 and a chuck top having a mounting surface 30s. Although not shown, the mounting table 31 may also include an appropriate holding means (vacuum suction mechanism, mechanical chuck, etc.) for holding the substrate W, a temperature control module for adjusting the temperature of the mounting surface 30s, a temperature sensor for detecting the temperature of the mounting surface 30s, etc.

[0021] The moving unit 32 transports the substrate W placed on the placement surface 30s to an appropriate three-dimensional coordinate position (X-axis, Y-axis, Z-axis) in the inspection chamber 12 and around the vertical axis (θ-axis). For example, the moving unit 32 adjusts the horizontal position of the substrate W by moving horizontally (X-axis-Y-axis directions) between a position near or inside the loader 13 and a position facing the probe card 21. The moving unit 32 also adjusts the circumferential position of the substrate W by rotating the placement table 31 around the vertical axis. Furthermore, the moving unit 32 adjusts the elevation position of the substrate W by moving up and down in the vertical direction (Z-axis) at the position facing the probe card 21 and the substrate W.

[0022] The stage control unit of the stage 30 is connected to a controller 90 (see FIG. 1) of the inspection device 1, and controls the motor driver unit based on commands from the controller 90 to operate the stage 30. The stage control unit includes, for example, a main control unit that controls the operation of the entire stage 30, a PLC that controls the operation of the moving unit 32, a temperature controller that controls the temperature adjustment module, an illumination control unit, a power supply unit, and the like.

[0023] The inspection device 1 is provided with an alignment bridge 40 between the tester 20 and the mounting table 31 of the stage 30 for aligning the substrate W placed on the mounting table 31. The alignment bridge 40 has a camera 41, a moving device 42 that moves the camera 41 back and forth, and a pair of guide rails 43 that guide the movement of the moving device 42.

[0024] The camera 41 of the alignment bridge 40 moves between a standby position set in the Y-axis direction and a center position by a moving device 42. The standby position is, for example, a position adjacent to the side wall of the housing 11 and not overlapping with the probe card 21 above in the vertical direction. The center position is, for example, a position vertically above the mounting surface 30s of the stage 30 placed at the reference position, below the probe card 21, and coincident with the center of the probe 22 of the probe card 21.

[0025] The camera 41 is attached to the underside of the moving device 42 and is connected to the controller 90. The camera 41 captures an image of the substrate W facing it below the center position and recognizes the position of the substrate W based on the image information. For example, a CCD (Charge Coupled Devices) sensor or a CMOS (Complementary Metal Oxide Semiconductor) sensor can be used as the camera 41. The controller 90 recognizes the position, attitude, thickness, etc. of the substrate W and / or the position of the mounting table 31, etc., based on the image information from the camera 41.

[0026] 3A, the moving device 42 is provided so as to bridge the pair of guide rails 43, and includes a plurality of rolling elements that roll on each guide rail 43, a motor that rotates each rolling element, etc. The moving device 42 is connected to a controller 90, and by driving the motor based on commands from the controller 90, the moving device 42 moves along the pair of guide rails 43 and positions the camera 41 at a center position or a standby position.

[0027] The pair of guide rails 43 are arranged with a width wider than the diameter of the mounting table 31. The guide rails 43 extend parallel to each other along the Y-axis direction in FIG. 2 and guide the movement of the moving device 42 in the Y-axis direction.

[0028] The controller 90 of the inspection device 1 is a computer having a processor, memory, input / output interface, communication interface, etc. (not shown). The processor is one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a circuit made up of a plurality of discrete semiconductors, etc., and executes and processes programs stored in memory. The memory includes a main storage device made up of a semiconductor memory, etc., and an auxiliary storage device made up of a disk, a drive, a semiconductor memory (flash memory), etc.

[0029] For example, the controller 90 controls the alignment bridge 40 to perform alignment processing for the substrate W. In the alignment processing, the controller 90 first moves the moving device 42 to position the camera 41 at the center position. Furthermore, the controller 90 causes the camera 41 to capture an image of the substrate W below. For example, as shown in FIG. 3B , image information PD captured by the camera 41 includes the positions and orientations of a plurality of arranged devices under test (DUTs), scribe lines SL, etc. Note that FIG. 3B illustrates an enlarged image of a specific device under test among a plurality of devices under test provided on the substrate W, i.e., an enlarged pattern of the device under test.

[0030] The controller 90 recognizes the positional deviation (deviation direction, deviation amount) of the substrate W based on the image information PD captured by the camera 41 and the target position and target orientation of the recipe. Then, the controller 90 moves or rotates the stage 30 by θ based on the recognized deviation of the substrate W, thereby appropriately adjusting the position and orientation of the substrate W. At this time, the inspection apparatus 1 may perform orthogonal axis alignment of the substrate W based on the image information PD. Orthogonal axis alignment is performed, for example, by calculating a line connecting pads in a semiconductor device on the substrate W in the image information PD and rotating the stage 30 in the θ direction to align this line with the Y-axis direction of the stage 30.

[0031] Furthermore, in order to inspect the substrate W, the controller 90 operates the moving unit 32 to bring the aligned substrate W into contact with the probes 22 of the probe card 21. In this contact state, the controller 90 performs an electrical inspection of the substrate W using the tester 20. Furthermore, after the inspection by the tester 20 is completed, the controller 90 causes the moving unit 32 to lower and horizontally move the substrate W, and returns the substrate W to the loader 13.

[0032] As described above, in the inspection system 100, image information PD of the substrate W is captured by the cameras 41 in multiple inspection devices 1, and alignment processing is performed based on this image information PD and the stored recipe (target position and target posture of the substrate W).

[0033] Returning to Figure 1, the image processing system 2 of the inspection system 100 receives and stores image information PD captured by each inspection device 1 and processes the image information PD to create setup data for the target position and target posture, which is a recipe for alignment processing. The image processing system 2 transmits the created setup data to the recipe server 3. As a result, the recipe server 3 updates (corrects) and stores the recipe for alignment processing based on the setup data of the image processing system 2. Then, the inspection system 100 transmits the corrected recipe stored in the recipe server 3 based on a recipe request from the inspection device 1, thereby enabling the inspection device 1 to perform alignment processing using the updated recipe.

[0034] Well-known computers can be used for the image processing system 2 and the recipe server 3. Although the image processing system 2 and the recipe server 3 are configured as separate devices in FIG. 1 , the image processing system 2 and the recipe server 3 may be the same device. For example, the image processing system 2 and the recipe server 3 may be configured as a management server that manages the operating status, errors, etc. of multiple inspection devices 1.

[0035] The image processing system 2 includes a processor 2p, a memory 2m, an input / output interface, a communication interface, etc. Like the processor of the controller 90 of the inspection device 1, the processor 2p is a combination of one or more of a CPU, a GPU, an ASIC, an FPGA, a circuit consisting of multiple discrete semiconductors, etc. The memory 2m includes a main memory device and an auxiliary memory device having a capacity sufficient to store multiple pieces of image information PD for each of the above-mentioned inspection devices 1. In other words, in this disclosure, the image processing system 2 is an electronic circuit including a CPU, a GPU, an ASIC, an FPGA, etc., and performs various control operations described in this specification by executing instruction codes stored in the memory 2m or by circuit design for a specific application. Meanwhile, the recipe server 3 is a storage device that stores recipes and includes a memory 3m, a communication interface, etc. The memory 3m stores recipes created by the image processing system 2.

[0036] In detail, the processor 2p of the image processing system 2 executes a program stored in the memory 2m to form an internal functional block that processes a plurality of pieces of image information PD and creates setup data, as shown in Fig. 4. The image processing system 2 includes an image acquisition unit 201, a score calculation unit 202, an image comparison unit 203, a registration unit 204, etc.

[0037] The image acquisition unit 201 acquires image information PD captured by each inspection device 1 and stores it in memory 2m. For example, the image acquisition unit 201 stores image information of the same product A in memory 2m in a manner that links the image information with each other, even if the image was captured by different inspection devices 1. This enables the image processing system 2 to easily collect image information of the same product A, and to accurately and quickly create setup data (registered images described below) corresponding to product A.

[0038] The score calculation unit 202 processes the image information PD stored in the memory 2m to calculate a score according to the substrate processing (alignment processing in this embodiment) of the substrate W. This score is a numerical representation of the degree of match between the previously registered image information PD (hereinafter also referred to as a registered image) and the image information PD to be compared (hereinafter also referred to as a comparison image).

[0039] 5A to 5C, for example, when the comparison pattern CP of the semiconductor device in the comparison image completely matches the registered pattern EP of the semiconductor device in the registered image, the score is set to 1. In other words, a score of 1 means that the square-shaped comparison pattern CP completely overlaps the square-shaped registered pattern EP in FIG. 5A. Conversely, when the comparison pattern CP of the semiconductor device in the comparison image completely mismatches the registered pattern EP of the semiconductor device in the registered image (they are separated by a predetermined distance or more), the score is set to 0. Then, the image processing system 2 calculates the score of the comparison image by evenly allocating the matching rate of the comparison image to the registered image using a decimal value between 0 and 1.

[0040] For example, if the substrate processing is an alignment processing, the score is the degree of positional deviation (including horizontal position and circumferential position) of the comparison pattern CP of the comparison image relative to the registered pattern EP of the registered image. In the example of Figure 5B, where the comparison pattern CP of the comparison image is significantly deviated from the registered pattern EP of the registered image, the score is 0.2. On the other hand, in the example of Figure 5C, where the comparison pattern CP of the comparison image is slightly deviated from the registered pattern EP of the registered image, the score is 0.8.

[0041] 5A to 5C illustrate an example of calculating a score related to the misalignment between the registered pattern EP of the registered image and the comparison pattern CP of the comparison image in response to the alignment process. In the case of substrate processes other than the alignment process, the score is not limited to the misalignment between the registered pattern EP of the registered image and the comparison pattern CP of the comparison image. For example, the score may be the matching rate of brightness, color, shape, defects, etc. between the registered image and the comparison image, or a single score may be calculated by combining a plurality of these parameters.

[0042] Conventional inspection systems use only one registered reference image to monitor the positional deviation of the comparison image of the substrate W relative to this registered reference image. However, when inspection is performed using multiple inspection devices, there are differences in the characteristics, model, and machine differences between the devices. Furthermore, when substrate processing is performed on the substrate W using substrate processing devices, there are differences in the content of the substrate processing performed by the substrate processing devices. For this reason, the appearance of the substrate W after substrate processing, including inspection, may change, and there is an inconvenience in that the appearance of the substrate W cannot be adequately addressed using only one registered reference image.

[0043] Therefore, the inspection system 100 according to the embodiment is configured to expand the range of reference registered images, i.e., setup data, by performing processing such as adding the comparison image to the registered image based on the score calculated for the comparison image. In other words, the inspection system 100 expands the range of substrate processing recipes by expanding the range to include setup data with different characteristics.

[0044] Therefore, the image comparison unit 203 (see FIG. 4) of the image processing system 2 determines whether or not to register a comparison image as a new registered image based on the calculated score of the comparison image. For a comparison image with a low score, there is a possibility that the substrate W has been misaligned in the original alignment process. Therefore, for a comparison image with a low score, the user can manually teach the position of the substrate W to add it to the registered image.

[0045] An example of the process for registering a comparison image will be described in detail below with reference to FIGS. 6A to 6C. The image comparison unit 203 holds a first threshold and a second threshold corresponding to the scores described above. In this case, the first threshold is a threshold for determining whether the comparison image sufficiently matches the registered image. The second threshold is a threshold for determining whether the comparison image matches or does not match the registered image. In other words, if the score of the comparison image is less than the second threshold, the comparison image is determined to not match the registered image. If the score of the comparison image is equal to or greater than the second threshold, the score of the comparison image is further compared with the first threshold. If the score is equal to or greater than the first threshold, the comparison image is determined to match sufficiently. On the other hand, if the score of the comparison image is less than the first threshold but equal to or greater than the second threshold, the comparison image is determined to be close to the registered image but not sufficiently matched. In the illustrated example, a score of 0.6 is set as the first threshold, and a score of 0.5 is set as the second threshold.

[0046] 6A, when the image comparison unit 203 compares the registered pattern EP of the registered image with the comparison pattern CP of the comparison image and determines that the score of the comparison image is 0.6 or higher, the image comparison unit 203 considers the comparison image to be a sufficient match with the registered image. In this case, the comparison image can be said to be an image that is sufficiently close to the registered image, so the image comparison unit 203 determines not to add a comparison image but to continue using the original registered image.

[0047] On the other hand, as shown in FIG. 6B , when the image comparison unit 203 compares the registered pattern EP of the registered image with the comparison pattern CP of the comparison image and determines that the score of the comparison image is less than 0.6 but greater than or equal to 0.5, the image comparison unit 203 determines that the comparison image is close to the registered image, as described above, but does not sufficiently match. In this case, the image comparison unit 203 determines to add the comparison image to the registered images. This allows the image processing system 2 to expand the range of registered images.

[0048] 6C , when the image comparison unit 203 compares the registered pattern EP of the registered image with the comparison pattern CP of the comparison image and determines that the score of the comparison image is less than 0.5, it regards the comparison image as mismatched with the registered image. In this case, since there is a possibility that the substrate W in the comparison image is misaligned as described above, the image comparison unit 203 makes an operator call and determines that the position of the substrate W should be taught by manual operation by the user to add the registered image.

[0049] 4, the registration unit 204 of the image processing system 2 manages the registered images and performs operations such as adding a new comparison image to the registered images. For example, if there are no registered images, the registration unit 204 registers the first registered image. Also, for example, if the image comparison unit 203 determines that an image should be added, the registration unit 204 adds that image to the registered images.

[0050] Hereinafter, an example of a change in a registered image due to the addition of a comparison image in the image processing system 2 will be described with reference to FIGS. 7A to 7D.

[0051] As shown in FIG. 7A , the score calculation unit 202 calculates the score of comparison image 1 based on the registered pattern EP1 of the first registered image, which has been registered in advance, and the comparison pattern CP of comparison image 1. For example, if the score of comparison image 1 is 0.55, the image comparison unit 203 selects the process of adding the comparison image to the registered images as described above. As a result, the registration unit 204 updates the registered image to a new (second) image by adding comparison image 1 to the first registered image, as shown in the right diagram of FIG. 7A . The second (two) registered images correspond to registered patterns EP1 and EP2, and their first and second thresholds overlap each other.

[0052] 7B , the score calculation unit 202 calculates the score of comparison image 2 based on the registered patterns EP1 and EP2 of the updated second registered image and the comparison pattern CP of the next comparison image 2. For example, if the score of comparison image 2 is 0.4 relative to all the scores of the second registered images, the image comparison unit 203 determines that the comparison images do not match, as described above. In this case, the image comparison unit 203 issues an operator call to have the user manually teach the position of the substrate W (see the dotted line in FIG. 7B ). As a result, the registration unit 204 updates the second registered image to a new (third) registered image by adding comparison image 2 to the second registered image, as shown in the right diagram in FIG. 7B . The third registered image corresponds to registered patterns EP1 to EP3, and one of the first and second thresholds overlaps with another of the first and second thresholds.

[0053] 7C, the score calculation unit 202 calculates the score of comparison image 3 based on the registered patterns EP1 to EP3 of the updated third registered image and the comparison pattern CP of the next comparison image 3. For example, if the score of comparison image 3 is 0.6 with respect to any of all registered images, the image comparison unit 203 determines that comparison image 3 matches the registered image. As a result, the image processing system performs processing to not register comparison image 3.

[0054] Furthermore, as shown in FIG. 7D , the score calculation unit 202 calculates the score of comparison image 4 based on the registered patterns EP1 to EP3 of the updated third registered image and the comparison pattern CP of the next comparison image 4. For example, if the score of comparison image 4 is 0.3 relative to all registered images, the image comparison unit 203 determines that the comparison images do not match, as in FIG. 7B . In this case, the image comparison unit 203 issues an operator call to have the user manually teach the position of the substrate W. In response to the user's manual operation, for example, as shown in the right diagram of FIG. 7D , the third registered image may be updated to a new (fourth) registered image in which an outlying image is added. That is, the fourth registered image may be in a state where one of the first and second thresholds overlaps with another first and second threshold, or in a state where the first and second thresholds are spaced apart from another first and second threshold, as in registered pattern EP4.

[0055] In this way, the image processing system 2 can appropriately expand the setup data required for alignment processing by appropriately adding registered images based on the scores (matching rates) of the comparison images. In other words, the inspection system 100 can automatically improve the setup data and repeatedly operate it.

[0056] The inspection system 100 and the image processing system 2 according to the embodiment are basically configured as described above, and their operation (image processing method) will be described below.

[0057] First, the processing flow for registering a registered image in the image processing method will be described with reference to the flowchart in Fig. 8. The image processing system 2 first registers reference setup data (registered image) for the alignment processing before performing the alignment processing in each inspection device 1. At this time, the image processing system 2 sequentially executes the processing flow of steps S101 to S106 shown in Fig. 8.

[0058] The image acquisition unit 201 of the image processing system 2 first acquires image information PD for registration from the inspection device 1 (step S101). In acquiring the image information PD for registration, for example, a substrate W is placed on the mounting table 31 of the inspection device 1 that serves as a reference, and the stage 30 is manually positioned by a user, and the camera 41, which has been moved to the center position, captures an image of the substrate W below. This allows the image processing system 2 to receive the image information PD for registration from the inspection device 1.

[0059] Next, the image processing system 2 determines whether or not to newly register the acquired image information PD (step S102). "New registration" refers to, for example, setting a new registered image without relying on a previously registered image, even if a previously registered image has been registered. One example of this is when, in order to perform alignment processing on a new product A, the previously used registered image of product B becomes unnecessary, and a new registered image corresponding to product A is registered. For example, when a user inputs a command for new registration, the image processing system 2 determines that the acquired image information PD is to be newly registered (step S102: YES) and proceeds to step S103.

[0060] In step S103, the image processing system 2 erases all previously stored registered images in order to register new image information PD. Note that if there is a possibility that previously registered images will be used later, the image processing system 2 may leave the previously registered images without executing step S103.

[0061] Thereafter, the image processing system 2 registers the acquired image information PD as a registered image (step S104). As a result, a registered image for alignment processing is stored in the memory 2m, and the image processing system 2 transmits this registered image as setup data to the recipe server 3. The recipe server 3 can transmit an alignment processing recipe (target position, target posture) based on this registered image, or the registered image itself, to each inspection device 1.

[0062] On the other hand, if it cannot be determined in step S102 that a new registration is required because, for example, a command for new registration has not been input by the user (step S102: NO), the process proceeds to step S105. In step S105, the image processing system 2 searches the image information storage area in memory 2m to determine whether or not there is a previously registered image. In other words, if there is no previously registered image, the acquired image information PD will be the data to be used as the first registered image. Therefore, if there is no previously registered image (step S105: NO), the process proceeds to the above-mentioned step S104, where the image information PD is registered as the registered image in step S104. On the other hand, if there is a previously registered image (step S105: YES), the process proceeds to step S106.

[0063] In step S106, the image processing system 2 executes a reset subroutine to compare the previously registered image in the memory 2m with the image information for the current registration and determine whether or not to add the image information for the current registration. This reset subroutine will be described in detail later.

[0064] By performing the above process of registering the registered image, the inspection system 100 can quickly start the alignment process of each inspection device 1 based on this registered image.

[0065] Next, the processing flow of the image processing method for comparing the registered image held by the above registration with a newly acquired comparison image and adding the comparison image to the registered image as needed will be described with reference to the flowchart in Figure 9. The image processing system 2 acquires image information PD when alignment processing is performed in each inspection device 1 and determines whether or not a registered image should be added. At this time, the image processing system 2 sequentially executes the processing flow of steps S111 to S123 shown in Figure 9.

[0066] The image processing system 2 first acquires a comparison image, which is image information PD for comparison (step S111). For example, every time each inspection device 1 performs alignment processing, the image processing system 2 receives image information PD captured by the camera 41 and stores it in the memory 2m.

[0067] When the comparison between the registered image and the comparison image is started, the image processing system 2 reads out the first registered image (i.e., n=1) from the registered images stored in the memory 2m (step S112). If there are multiple registered images, the other registered images are read out in the subsequent step S116. That is, the image processing system 2 compares each of the multiple registered images with one comparison image by adding registered images during operation of the inspection system 100. This allows the image processing system 2 to accurately determine whether or not a new comparison image needs to be registered.

[0068] The image processing system 2 determines whether there is a registered image to compare with the comparison image (step S113). For example, when comparing the first registered image with the comparison image, it is natural that there is a registered image. If there is such a registered image (step S113: YES), the process proceeds to step S114. On the other hand, there may be no registered images beyond the second registered image. If there is such a registered image (step S113: NO), the process proceeds to step S117.

[0069] In step S114, the score calculation unit 202 of the image processing system 2 calculates a score, which is the matching rate of the comparison image with the registered image, using the registered pattern EP of the current registered image and the comparison pattern CP of the comparison image. As described above, the score of the comparison image is calculated as a numerical value between 0 and 1, including decimal points.

[0070] If there is a previously compared registered image, the score calculation unit 202 takes into account the scores of the comparison image calculated from each registered image and the current score, and extracts the highest score from among the scores (step S115). This allows the image processing method to perform the subsequent processing flow using the highest score.

[0071] The image processing system 2 then proceeds to step S116, increments the number of registered images in step S116 (n = n + 1), and prepares to read out the next registered image. For example, if the first registered image has been compared with the comparison image, the second registered image is read out. The image processing system 2 then returns to step S113 to determine whether or not the incremented registered image exists. If a registered image exists, the image processing system 2 proceeds to step S114 again, and the same processing flow is repeated. On the other hand, if a registered image does not exist (step S113: NO), the image processing system 2 proceeds to step S117.

[0072] In step S117, the image processing system 2 determines whether the maximum match score for the comparison image that was not added is less than the second threshold (0.5). That is, by going through step S105, the score of the comparison image becomes the maximum match score, which is the highest score when compared with multiple registered images. Therefore, the image processing system 2 compares the maximum match score with the second threshold. Then, as described above, if the maximum match score is 0.5 or greater, it means that the comparison pattern CP of the comparison image matches the registered pattern EP of the registered image. Therefore, if the maximum match score is 0.5 or greater (step S117: NO), the process proceeds to step S118.

[0073] The image comparison unit 203 of the image processing system 2 compares the calculated maximum match score of the comparison image with a first threshold value (0.6) stored in advance and determines whether the maximum match score is less than 0.6 (step S118). If the maximum match score is less than 0.6 (step S118: YES), the image processing system 2 determines whether to register the comparison image as described above, and proceeds to step S119.

[0074] In step S119, the registration unit 204 of the image processing system 2 performs a process of adding the comparison images having the maximum matching scores of 0.5 or more and less than 0.6 to the registered images. For example, the registration unit 204 associates the comparison images determined to be added with the current registered images and stores them, so that the comparison images can be handled as registered images.

[0075] On the other hand, if the maximum match score is 0.6 or more (step S118: NO), the comparison image sufficiently matches the registered image and does not need to be added. Therefore, image processing system 2 ends the process without adding the current comparison image to the registered images. As a result, comparison images with a score of 0.6 or more will not be added as registered images.

[0076] On the other hand, returning to step S117, if the maximum matching score is less than 0.5, the comparison pattern CP of the comparison image is significantly deviated from the registered pattern EP of the registered image, and manual teaching by the user is required. Therefore, if the maximum matching score is less than 0.5 (step S117: YES), proceed to step S120.

[0077] In step S120, the image processing system 2 makes an operator call to have the user perform a manual operation. This operator call includes information that a large positional deviation has occurred in the comparison image and that manual operation is necessary, and also includes an instruction for the inspection device 1 that is the target of the manual operation.

[0078] The user of the inspection system 100 manually moves the stage 30 on which the substrate W is placed for the specified inspection device 1 to specify (teach) the position to a position facing the camera 41 (step S121). As a result, the target inspection device 1 captures an image of the substrate W with the camera 41 and acquires image information again, and then transmits the image information to the image processing system 2.

[0079] Then, when the image processing system 2 receives image information from the inspection device 1 that has been manually operated, it replaces the image information with a new comparison image (step S122).

[0080] The image processing system 2 then adds the replaced comparison image to the registered image (step S123). As described above, the comparison image taught by the user may have a score close to that of the registered image (see FIG. 7B), or may have a score far from that of the registered image (see FIG. 7D). After completing step S122, the image processing system 2 ends the processing flow for the current comparison image.

[0081] Next, the reset subroutine (step S106 in FIG. 8) performed when registering the above-mentioned registered image will be described with reference to FIG. 10. The reset subroutine is a subroutine that, if there is a previously registered image, compares the previously registered image with the current registered image and determines how to handle the current registered image.

[0082] In this reset subroutine, the image processing system 2 sequentially executes steps S131 to S139 shown in Fig. 10. However, the reset subroutine basically performs substantially the same procedures as the processing flow shown in Fig. 9, which compares a registered image with a comparison image and processes the comparison image. Therefore, in the following description of the reset subroutine, explanations of parts that overlap with Fig. 9 may be omitted.

[0083] Specifically, the image processing system 2 first sets the image information for registration acquired in the process flow for registering the registration image in FIG. 8 as the current registration image to be processed in the resetup subroutine (step S131).

[0084] Thereafter, the image processing system 2 performs steps S131 to S136 that are substantially the same as steps S112 to S116 shown in Fig. 9. Then, in step S133, when the comparison of all previously registered images with the image information PD for current registration is completed (step S133: NO), the process proceeds to step S137.

[0085] Furthermore, the image processing system 2 performs steps S137 and S138 in substantially the same manner as steps S118 to S119 shown in FIG. 9 . However, in step S137, it is determined whether the maximum matching score of the current registered image is equal to or less than 0.6 (first threshold value). If the maximum matching score is equal to or less than 0.6 (step S137: YES), it means that there is no image information similar to the current registered image among the previously registered images. Therefore, in step S138, the image processing system 2 performs a process of adding the current registered image to the previously registered images as is (without an operator call or manual operation by the user). On the other hand, if the maximum matching score exceeds 0.6 (step S137: NO), it means that there is image information similar to the current registered image among the previously registered images. Therefore, the image processing system 2 does not perform step S138 and does not add the current registered image.

[0086] By performing the reset subroutine in this way, the image processing system 2 can appropriately determine whether to add or not add the current registered image, and can smoothly set the registered image to be used in the alignment process. In other words, the image processing system 2 can expand the range of registered images in the alignment process even before the alignment process is performed.

[0087] 8 to 10, the image processing system 2 can expand the range of the registered image. In this way, the repeated automatic growth and stopping of the range of the registered image can be said to be an efficient system that prevents frequent operational shutdowns in the inspection system 100. In other words, the inspection system 100 can automatically improve and operate the registered image (setup data) using the image processing system 2, thereby increasing the success rate of the alignment process.

[0088] The image processing system 2, inspection system 100, and image processing method disclosed herein are not limited to the above-described embodiment, and various modifications are possible. For example, the above-described embodiment has been described using an inspection system 100 having multiple inspection devices 1 as an example. However, the same image processing system 2 and image processing method can be adopted even in an inspection system 100 having only one inspection device 1. In this case, it goes without saying that the controller 90 of the inspection device 1 can also serve as the image processing system 2 and the recipe server 3.

[0089] In the above embodiment, image processing of image information in alignment processing of a substrate W (semiconductor device) has been described in the inspection apparatus 1. However, the image processing system 2, inspection system 100, and image processing method disclosed herein are not limited to this, and may be applied to, for example, alignment processing of the probe card 21 when attaching the probe card 21 to the interface 23. In this way, the image processing system 2, inspection system 100, and image processing method can also increase the success rate of alignment processing of the probe card 21.

[0090] Furthermore, the image processing system 2 may arbitrarily design the thresholds (first threshold, second threshold) for determining the score of the comparison image without being limited to 0.6, 0.5, etc. The first threshold and second threshold may be appropriately set depending on the object (substrate W or probe card 21), the content of the substrate processing, etc. For example, the first threshold and second threshold may be variable values ​​that change depending on switching between product A and product B. For example, when performing alignment processing on fine product A, the image processing system 2 may narrow the range of the first threshold and the second threshold to make it more difficult to register the comparison image. Furthermore, the image processing system 2 may use only one threshold (e.g., the first threshold) for determining the score of the comparison image or the registered image.

[0091] The technical ideas and effects of the present disclosure explained in the above embodiments will be described below.

[0092] A first aspect of the present disclosure is an image processing system 2 that processes at least one of image information PD of a semiconductor device (substrate W) and image information PD of a probe card 21, and includes a calculation unit (score calculation unit 202) that calculates information related to the match rate between a comparison image, which is the acquired image information PD, and a registered image that is stored in advance, a comparison unit (image comparison unit 203) that compares the calculated information related to the match rate with a threshold value that is stored in advance, and a registration unit 204 that adds the comparison image to the registered image and updates it to a new registered image when the information related to the match rate is less than the threshold, but does not add the comparison image to the registered image when the information related to the match rate is equal to or greater than the threshold.

[0093] As described above, the image processing system 2 can add comparison images to the registered images stored in advance as needed. This allows the image processing system 2 to automatically improve the setup data (registered images) and appropriately set a recipe for substrate processing, including inspection, using this setup data. For example, when performing alignment processing during inspection, a registered image with a wide range can increase the success rate of the alignment processing.

[0094] The comparison unit (image comparison unit 203) has a first threshold (score of 0.6) as a threshold value, and a second threshold (score of 0.5) that is lower than the first threshold value, and determines to add the comparison image to the registered images as is when the information related to the matching rate is less than the first threshold value and equal to or greater than the second threshold value. As a result, the image processing system 2 can easily expand the range of registered images by adding the comparison image to the registered images when the matching rate with the registered images is somewhat different.

[0095] Furthermore, the first and second thresholds of the pre-stored registered images and the first and second thresholds of the newly added registered images overlap each other, so that the registered images are registered in a state where they are related to each other even if they have different first and second thresholds.

[0096] Furthermore, if the match rate of the comparison image is less than the second threshold, the comparison unit (image comparison unit 203) reacquires the comparison image based on a manual operation by the user and determines whether to add the reacquired comparison image to the registered image. As a result, if the match rate of the comparison image with the registered image is low, an appropriate comparison image can be obtained by manual operation by the user and added to the registered image.

[0097] In addition, the first and second thresholds of the registered image stored in advance and the first and second thresholds of the registered image newly added by the user's manual operation may overlap or be separated from each other, which allows the multiple registered images to be separated from each other, making it possible to reliably set the comparison image by the user's manual operation.

[0098] Furthermore, when there are multiple registered images, the calculation unit (score calculation unit 202) and the comparison unit (image comparison unit 203) sequentially calculate and compare information relating to the match rate between one comparison image and the multiple registered images, and determine whether to add the comparison image to the registered images. This allows the image processing system 2 to smoothly compare the comparison image with multiple registered images and determine whether to add a new comparison image.

[0099] Furthermore, the image information PD of the semiconductor device (substrate W) or the image information PD of the probe card 21 is captured when performing alignment processing for the semiconductor device or the probe card 21. This allows the image processing system 2 to effectively improve the setup data when performing alignment processing for the semiconductor device or the probe card 21, thereby increasing the success rate of the alignment processing.

[0100] A second aspect of the present disclosure is an inspection system 100 including an inspection apparatus 1 that inspects a semiconductor device (substrate W) and an image processing system 2 that processes at least one of image information PD of the semiconductor device and image information of a probe card 21 having a plurality of probes 22 that contact the semiconductor device during inspection, wherein the image processing system 2 includes a calculation unit (score calculation unit 202) that calculates information related to a match rate between a comparison image, which is the acquired image information PD, and a pre-stored registered image, a comparison unit (image comparison unit 203) that compares the calculated information related to the match rate with a pre-stored threshold, and a registration unit 204 that adds the comparison image to the registered image and updates it to a new registered image if the information related to the match rate is less than the threshold, but does not add the comparison image to the registered image if the information related to the match rate is equal to or greater than the threshold. Even in this case, the inspection system 100 can appropriately set the inspection recipe by automatically improving the setup data of the inspection apparatus 1.

[0101] A third aspect of the present disclosure is an image processing method for processing at least one of image information PD of a semiconductor device (substrate W) and image information PD of a probe card 21, the image processing method including: (A) calculating information relating to a match rate between a comparison image, which is the acquired image information PD, and a pre-stored registered image; (B) comparing the calculated information relating to the match rate with a pre-stored threshold; and (C) adding the comparison image to the registered image to update it to a new registered image when the information relating to the match rate is less than the threshold, while not adding the comparison image to the registered image when the information relating to the match rate is equal to or greater than the threshold. Even in this case, the image processing method can appropriately set a substrate processing recipe by automatically improving the setup data.

[0102] The image processing system 2, the inspection system 100, and the image processing method according to the presently disclosed embodiments are illustrative in all respects and are not limiting. The embodiments can be modified and improved in various ways without departing from the spirit and scope of the appended claims. The matters described in the above-described embodiments can be configured in other ways as long as they are not inconsistent, and can be combined as long as they are not inconsistent.

[0103] This application claims priority to Japanese Patent Application No. 2024-110533, a basic application filed on July 9, 2024 with the Japan Patent Office, and Japanese Patent Application No. 2024-179327, a domestic priority application filed on October 11, 2024, the entire contents of which are incorporated herein by reference.

[0104] 2 Image processing system 21 Probe card 100 Inspection system 202 Score calculation unit 203 Image comparison unit 204 Registration unit W Substrate

Claims

1. An image processing system that processes at least one of image information of a semiconductor device and image information of a probe card, comprising: a calculation unit that calculates information relating to the match rate between a comparison image, which is the acquired image information, and a registered image that is stored in advance; a comparison unit that compares the calculated information relating to the match rate with a threshold value that is stored in advance; and a registration unit that adds the comparison image to the registered image to update it to a new registered image when the information relating to the match rate is less than the threshold, but does not add the comparison image to the registered image when the information relating to the match rate is equal to or greater than the threshold.

2. The image processing system of claim 1, wherein the comparison unit has a first threshold as the threshold and a second threshold that is lower than the first threshold, and when the information related to the matching rate is lower than the first threshold and equal to or greater than the second threshold, makes a decision to add the comparison image to the registered image as is.

3. The image processing system according to claim 2, wherein the first threshold and the second threshold of the registered image stored in advance and the first threshold and the second threshold of the newly added registered image have overlapping portions.

4. The image processing system according to claim 2, wherein the comparison unit, when the difference is less than the second threshold, reacquires the comparison image based on a manual operation by a user and determines whether to add the reacquired comparison image to the registered image.

5. The image processing system of claim 4, wherein the first threshold and the second threshold of the registered image stored in advance and the first threshold and the second threshold of the registered image newly added by manual operation of the user have overlapping portions or are spaced apart from each other.

6. An image processing system according to any one of claims 1 to 5, wherein, when there are multiple registered images, the calculation unit and the comparison unit sequentially calculate and compare information relating to the matching rate between one comparison image and multiple registered images, and determine whether or not to add the comparison image to the registered images.

7. The image processing system according to any one of claims 1 to 5, wherein the image information of the semiconductor device or the image information of the probe card is captured when performing alignment processing of the semiconductor device or the probe card.

8. An inspection system comprising: an inspection apparatus for inspecting semiconductor devices; and an image processing system for processing at least one of image information of the semiconductor device and image information of a probe card having a plurality of probes that come into contact with the semiconductor device during inspection, wherein the image processing system comprises: a calculation unit that calculates information relating to the match rate between a comparison image, which is the acquired image information, and a pre-stored registered image; a comparison unit that compares the calculated information relating to the match rate with a pre-stored threshold; and a registration unit that adds the comparison image to the registered image to update it to a new registered image when the information relating to the match rate is less than the threshold, but does not add the comparison image to the registered image when the information relating to the match rate is equal to or greater than the threshold.

9. An image processing method for processing at least one of image information of a semiconductor device and image information of a probe card, comprising: (A) a step of calculating information relating to the degree of match between a comparison image, which is the acquired image information, and a registered image stored in advance; (B) a step of comparing the calculated information relating to the degree of match with a threshold value stored in advance; and (C) a step of adding the comparison image to the registered image and updating it to a new registered image when the information relating to the degree of match is less than the threshold, and not adding the comparison image to the registered image when the information relating to the degree of match is equal to or greater than the threshold.

Images