Efem including a wafer substrate backside defect detecting device

Abstract

The present embodiment may provide an EFEM comprising: a Front-Opening Unified Pod (FOUP) on which a substrate is loaded; a main body having a first space formed therein that communicates with the FOUP through a door and allows the substrate being transported to stay there; a transport device disposed inside the main body and which mounts the substrate on a transport arm to remove the substrate from the FOUP; a lighting device including a light source that irradiates light in the direction of the lower surface of the substrate; a vision camera that captures images of the lower surface of the substrate mounted on the transport arm in real time inside the main body while the transport device removes the substrate; and a defect detection device that analyzes images captured by the vision camera to detect defects occurring on the lower surface of the substrate.

Description

Technology Field

[0001] This embodiment relates to an EFEM including a device for detecting defects on the lower surface of a wafer substrate. Background Technology

[0002] With the recent acceleration of the transition to an information society, the demand for highly integrated electronic devices is surging.

[0003] Highly integrated electronic devices, such as high-resolution display devices and high-density, high-performance semiconductor devices, are manufactured by integrating multiple electronic structures onto a single area through a high-precision surface treatment process.

[0004] Semiconductor manufacturing processes used in the manufacture of highly integrated electronic devices are known to include thin film deposition, polishing, photolithography, etching, ion implantation, cleaning, and heat treatment, and highly integrated electronic devices are manufactured by a composite process in which these different processes are applied one or more times.

[0005] The process system may have process processing devices such as an Equipment Front End Module (EFEM), a transfer device, and a process chamber to apply multiple processes to a highly integrated electronic device.

[0006] A substrate to be processed—for example, a semiconductor wafer—waits in the EFEM among these process processing devices, and then is moved to a suitable process chamber by a transfer device and undergoes the necessary process.

[0007] The EFEM may have a Load Port Module (LPM), a Front-Opening Unified Pod (FOUP), an EFEM chamber, etc.

[0008] The load port module is a device to which a semiconductor wafer storage device called a pod is combined.

[0009] A number of semiconductor wafers can be loaded in the pod, and a transfer device including a transfer device sequentially transfers the semiconductor wafer substrates (W) loaded in the pod to the process chamber.

[0010] Semiconductor wafer processing takes place within a high-cleanliness cleanroom, but wafers can be loaded into a POW to provide an even higher level of cleanliness. Meanwhile, during the semiconductor manufacturing process, the back side of the wafer is prone to defects such as contamination, scratches, and cracks.

[0011] These defects can cause oxidation, etching, or foreign matter formation in the device, thereby reducing yield. Since the substrate to be processed—e.g., a semiconductor wafer—resides in the EFEM for a significant amount of time, the detection of defects on the back side of the wafer within the EFEM is recognized in the industry as an important factor for improving yield.

[0012] Therefore, currently, a method is used to inspect the back of the wafer on a sample basis using a separate stand-alone instrument.

[0013] However, while the transfer device moves the wafer substrate (W) from the pod to the process device, it cannot detect defects on the back side of the wafer in real time, which poses a risk of mass defects, and there is a problem that the turnaround time (TAT) increases due to stand-alone inspection.

[0014] To overcome the aforementioned problems, real-time defect detection technology is being attempted by installing a vision camera and lighting device inside the EFEM, but there is a need for technology that can overcome practical problems such as space constraints, maintaining turnaround time (TAT), and high-precision detection accuracy. The problem to be solved

[0015] Against this backdrop, the objective of the present embodiment is, in one aspect, to provide a technology that can detect defects on the bottom surface of a wafer in real time while a transfer device moves a wafer substrate from a FOUP to a process device within an EFEM, thereby preventing mass defects, solving the problem of increased turnaround time (TAT) caused by the use of existing standalone measuring instruments, and overcoming the technical challenges of space constraints and high-precision detection. means of solving the problem

[0016] To achieve the aforementioned objective, one embodiment may provide an EFEM comprising: a Front-Opening Unified Pod (FOUP) on which a substrate is loaded; a main body having a first space formed therein that communicates with the FOUP through a door and allows the substrate being transported to stay there; a transport device disposed inside the main body and which places the substrate on a transport arm to remove the substrate from the FOUP; a lighting device including a light source that irradiates light in the direction of the lower surface of the substrate; a vision camera that captures images of the lower surface of the substrate placed on the transport arm in real time inside the main body while the transport device removes the substrate; and a defect detection device that analyzes images captured by the vision camera to detect defects occurring on the lower surface of the substrate.

[0017] Another embodiment may provide an EFEM in which the lighting device comprises a plurality of light sources spaced apart from the lower surface of the transfer arm.

[0018] Another embodiment may provide an EFEM in which the light is short-wavelength infrared.

[0019] Another embodiment may provide an EFEM in which the light source includes a ring structure formed around the lens of the vision camera.

[0020] Another embodiment may provide an EFEM in which the lighting device includes a diffuser plate disposed on the path where the light is irradiated.

[0021] Another embodiment may provide an EFEM in which the vision camera is fixedly positioned on the lower surface of the transfer arm.

[0022] Another embodiment may provide an EFEM in which the number of vision cameras is multiple.

[0023] In another embodiment, the defect detection device may provide an EFEM including a deep learning-based image processing algorithm.

[0024] In another embodiment, the defect detection device may provide an EFEM that generates mapping data of the distribution of defects present on the lower surface of the substrate based on an image captured by the vision camera.

[0025] Another embodiment may provide an EFEM comprising a defect alarm device that transmits an alarm signal to a user when the number of defects detected by the defect detection device exceeds a preset threshold.

[0026] Another embodiment may provide an EFEM comprising: a Front-Opening Unified Pod (FOUP) on which a substrate is loaded; a main body having a first space formed inside that is in communication with the FOUP through a door and in which the substrate being transported can stay; a transport device disposed inside the main body and which takes the substrate out of the FOUP by placing the substrate on a transport arm; a side storage for temporarily storing the substrate in a side storage chamber; a lighting device including a light source disposed above or below the side storage chamber and irradiating light in the direction of the substrate; a vision camera disposed above or below the side storage chamber and capturing images of the substrate in real time; and a defect detection device that analyzes images captured by the vision camera to detect defects occurring in the substrate. Effects of the invention

[0027] As described above, according to the present embodiment, a technology is provided that can detect defects on the lower surface of a wafer in real time through a vision camera and a lighting device while the transfer device moves the wafer substrate from the FOUP to the process device within the EFEM, thereby preventing mass defects, solving the problem of increased turnaround time (TAT) caused by the use of existing standalone measuring instruments, and significantly improving process yield by overcoming the technical challenges of space constraints and high-precision detection. Brief explanation of the drawing

[0028] FIG. 1 is a side view showing the configuration of a process system according to one embodiment. FIG. 2 is a top view showing the configuration of a process system according to one embodiment. FIG. 3 is a diagram showing a transfer device moving a wafer substrate from a FOUP of an EFEM chamber according to one embodiment. FIG. 4 is a drawing showing a wafer bottom surface defect detection device and a transfer device inside an EFEM according to one embodiment. FIG. 5 is a top view of a wafer bottom defect detection device and a transfer device inside an EFEM according to one embodiment. FIG. 6 is a top view of a wafer bottom defect detection device and a transfer device inside an EFEM according to another embodiment. FIG. 7 is a drawing showing a wafer bottom surface defect detection device and a transfer device inside an EFEM according to another embodiment. FIG. 8 is a top view of a wafer bottom defect detection device and a transfer device inside an EFEM according to another embodiment. FIG. 9 is a drawing showing a camera mounted on a side storage (SSB) according to another embodiment. FIG. 10 is a drawing showing a substrate being transferred to a side storage (SSB) according to another embodiment. Specific details for implementing the invention

[0029] Some embodiments are described in detail below with reference to exemplary drawings. It should be noted that in assigning reference numerals to the components of each drawing, the same components are given the same reference numeral whenever possible, even if they are shown in different drawings. Furthermore, in describing the present invention, if it is determined that a detailed description of related known components or functions could obscure the essence of the invention, such detailed description is omitted.

[0030] Additionally, terms such as first, second, A, B, (a), (b), etc., may be used when describing components. These terms are intended only to distinguish the component from other components and do not limit the nature, order, or sequence of the component. Where it is stated that a component is "connected," "combined," or "joined" to another component, it should be understood that the component may be directly connected or joined to the other component, but that another component may also be "connected," "combined," or "joined" between each component.

[0031] FIG. 1 is a side view showing the configuration of a process system according to one embodiment, and FIG. 2 is a top view showing the configuration of a process system according to one embodiment.

[0032] Referring to FIGS. 1 and 2, the process system (100) may include an EFEM (110), a load lock device (120), a return chamber (130), a process chamber (140), etc.

[0033] The EFEM (110) is an interface module for supplying a substrate (W), such as a wafer, to a process chamber (140), and the substrate (W) can be introduced into and removed from the process system (100) through the EFEM (110).

[0034] The substrate (W) to be processed can stay in the EFEM before being transferred to the process chamber (140) and then be transferred to the process chamber (140) via the return chamber (130) when necessary.

[0035] The substrates (W) staying in the EFEM can be sequentially transferred to the load lock device (120). Then, the substrates (W) transferred to the load lock device (120) are transferred to the return chamber (130), and the transfer device placed in the return chamber (130) transfers the substrates (W) to the process chamber (140) so that the substrates (W) can be processed.

[0036] The process system (100) may include a plurality of process chambers (140a, 140b, 140c), and each of the process chambers (140a, 140b, 140c) may apply different processes to the substrate (W). A transfer device placed in the return chamber (130) may introduce the substrate (W) into the first process chamber (140a) to allow the first process to be applied to the substrate (W), and may introduce the substrate (W) removed from the first process chamber (140a) into the second process chamber (140b) to allow the second process to be applied to the substrate (W). Additionally, the transfer device may introduce the substrate (W) into the third process chamber (140c) to allow the third process to be applied to the substrate.

[0037] Doors (D1, D2, D3, D4) are arranged between each device to minimize the mixing of gases between each device. For example, a second door (D2) may be arranged between the EFEM (110) and the load lock device (120), a third door (D3) may be arranged between the load lock device (120) and the return chamber (130), and a fourth door (D4) may be arranged between the return chamber (130) and the process chamber (140). Additionally, each door (D1, D2, D3, D4) may be opened only when the substrate (W) is being moved and may remain closed at other times.

[0038] The EFEM (110) may include an EFEM chamber (112), a pod (114), a load port module (116), etc.

[0039] A number of substrates (W) can be loaded into the pod (114). The number of substrates (W) can be transferred sequentially to the EFEM chamber (112).

[0040] The pod (114) can be connected to the EFEM chamber (112) through the first door (D1). A transfer device may be placed inside the EFEM chamber (112), and when the first door (D1) is opened, the transfer device can take out the substrate (W) from the pod (114) and transfer it to the load lock device (120).

[0041] The EFEM (110) may include a plurality of pods (114a, 114b, 114c). Each pod (114a, 114b, 114c) may be connected to the EFEM chamber (112) at different locations. A transfer device placed in the EFEM chamber (112) may sequentially open first doors (D1a, D1b, D1c) placed in each pod (114a, 114b, 114c) and sequentially remove substrates (W) from each pod (114a, 114b, 114c).

[0042] The internal space of the pod (114) may be relatively narrow compared to the EFEM chamber (112). Because the pod (114) has a relatively narrow space, the gas atmosphere surrounding the substrate can be well controlled. For example, when a substrate such as a semiconductor wafer is exposed to moisture, oxygen, etc., an oxide film may form on its surface, but this problem can be minimized by purging the internal space of the pod (114) with nitrogen gas (nitrogen).

[0043] A load port module (116) supporting the pod (114) can supply nitrogen gas into the internal space of the pod (114). The load port module (116) may include a nitrogen gas supply device, a nitrogen gas pipe, a Mass Flow Controller (MFC), a filter, etc. Nitrogen gas supplied from the nitrogen gas supply device can be delivered into the internal space of the pod (114) through the nitrogen gas pipe, at which time the MFC can control the flow of the nitrogen gas fluid and a filter in the nitrogen gas distribution path can remove foreign substances.

[0044] According to this structure in which the load port module (116) purges nitrogen gas into the internal space of the pod (114), the time required to lower the internal humidity of the pod (114) can be shortened, the effect of suppressing contamination of a substrate (W), such as a semiconductor wafer, can be increased, static electricity generation during transport of the substrate (W) can be suppressed, the diffusion of particles can be prevented, and the possibility of corrosion of the substrate (W) by particles can be reduced.

[0045] Meanwhile, since moisture can react with fumes generated from process processing devices to oxidize or etch the device and can react with fine reaction particles to form foreign matter, it is important to minimize moisture within the EFEM (110). Since the pod (114) has a relatively narrow space, it may be easy to reduce humidity, but since the EFEM chamber (112) has a relatively wide space, it may not be easy to reduce humidity.

[0046] Various technologies have been attempted to lower humidity in these EFEM chambers, but most had problems such as being expensive or having low effectiveness.

[0047] FIG. 3 is a diagram showing a transfer device moving a wafer substrate from a FOUP of an EFEM chamber according to one embodiment.

[0048] Referring to FIG. 3, in one embodiment, a transfer device (320) can take out and move a wafer substrate (W) loaded in the FOUP (310) of the EFEM chamber (300).

[0049] FOUP (310) is a storage device loaded with wafers (W), and FOUP (310) is mounted in the load port of the EFEM chamber (300), and may have a number of wafer substrates loaded inside.

[0050] The transfer device (320) may refer to a device that takes a wafer substrate (W) out of the FOUP (310) and moves it to the process equipment.

[0051] A transfer device (320) according to one embodiment may include a driving unit, and the driving unit may mean a driving device that moves a moving part (not shown) connected to a transfer arm (321) in the forward and backward directions and supplies kinetic energy to rotate a support part in the forward and backward directions.

[0052] The transfer arm (321) may refer to a device that securely holds and moves the wafer substrate (W).

[0053] The support member can facilitate the movement of the transfer device (320) and provide stability, and mainly supports devices connected to moving parts such as the transfer arm (321) or acts as a joint connecting each support member, and can ensure precision and flexibility in the wafer transfer operation.

[0054] The support member can physically support moving parts such as the transfer arm (321) to allow the transfer device to operate stably without shaking, and can absorb external vibrations or load changes that occur during transport.

[0055] The support members (430, 431, 432) can support multi-axis movements such as rotation, up and down, and forward and backward movements of the transfer arm (321). This allows the transfer device (320) to perform various operations during the process of taking out and moving the wafer substrate (W) from the FOUP (310).

[0056] The support member may include a multi-axis joint structure, and each axis may be responsible for specific directional movement (e.g., rotation, up and down, forward and backward) of the joints including the transfer arm.

[0057] According to one embodiment, the support member may have a sensor built in to detect position, speed, and torque, and thereby it may be possible to control the operation of the transfer device (320) in real time.

[0058] The interior of the support may include gears or bearings that control the power transmitted from the drive unit, making it possible to provide smooth and precise movement.

[0059] According to one embodiment, when a driving unit generates power, this power can be transmitted to a support unit. The support unit can control the movement of the robot arm by converting the transmitted power into rotational, linear, or compound motion.

[0060] Based on sensor data, the support member can form a movement linked with the transfer arm (321), and it may be possible to control the transfer arm (321) to move accurately to a desired position and angle. Through this, the operational stability and precision of the transfer device (320) are improved, thereby reducing contamination and damage to the wafer substrate (W), and the transfer speed can be improved through efficient power transmission between the drive member and the transfer arm (321).

[0061] The transfer arm (321) can take out the wafer substrate (W) from the FOUP and move it to the process equipment.

[0062] At this time, depending on the forward and backward movement of the transfer arm (321), the substrate (W) placed on the top of the transfer arm (321) can move into or out of the buffer (450).

[0063] A transfer arm (321) according to one embodiment may include a plurality of rod structures, thereby enabling stable movement of the substrate (W), and at the same time, since only a portion of the lower surface of the substrate supports the substrate, it may be possible to take a picture by a vision camera of the portion of the lower surface of the substrate that is not supported.

[0064] FIG. 4 is a drawing showing a wafer bottom surface defect detection device and a transfer device inside an EFEM according to one embodiment.

[0065] Referring to FIG. 4, an EFEM according to one embodiment may include a transfer device (420), a lighting device (440), a vision camera (430), etc.

[0066] The transfer device (420) can perform the role of taking the substrate (W) out of the FOUP and transferring it to the process chamber.

[0067] The transfer device (420) may include a transfer arm (421), and the transfer arm (421) may have a mounting surface on which a substrate (W) is placed.

[0068] The transfer arm (421) can be formed with a plurality of rod structures, thereby stably supporting the substrate (W) while exposing the lower surface of the substrate (W) to enable real-time shooting by the vision camera (430). With this structure, the transfer device (420) can maintain the turnaround time (TAT) without interfering with the defect detection process even while the substrate (W) is moving within the EFEM, and can significantly improve process efficiency by minimizing the time delay caused by the use of existing standalone measuring instruments.

[0069] The lighting device (440) can perform the function of irradiating light in the direction of the lower surface of the substrate (W) and may include a light source (441).

[0070] The light source (441) is positioned at the bottom of the transfer arm (421) to uniformly illuminate the entire lower surface of the substrate (W).

[0071] The light source (441) can be spaced apart in multiple numbers and formed into a ring structure and arranged around the lens (431) of the vision camera (430).

[0072] With this arrangement, the lighting device (440) can provide contrast that highlights micro-defects (e.g., scratches, contamination, cracks) of the substrate (W), and can detect internal defects by irradiating light that passes through the silicon substrate (W) using short-wavelength infrared (SWIR).

[0073] Additionally, the lighting device (440) may include a diffuser plate in the light path, and the diffuser plate can optimize light scattering to minimize shadow effects or hot spots. Through these functions, the lighting device (440) can overcome the limitations of the existing uniform lighting method to obtain high-resolution images, and can maintain high defect detection accuracy while exhibiting a contamination prevention effect in the cleanroom environment of the EFEM.

[0074] The vision camera (430) is fixedly positioned at the lower part of the transfer arm (421) and can perform the role of capturing images of the lower surface of the substrate (W) in real time, and may include a lens (431).

[0075] The vision camera (430) can perform imaging while the transfer device (420) is transporting the substrate (W), and can be configured in multiple units to simultaneously image the lower surface of the substrate (W) from multiple angles. With this configuration, the vision camera (430) can overcome the field of view limitations of a single camera to comprehensively detect complex defect shapes, and can enable full inspection while maintaining TAT through a high-speed imaging function.

[0076] In addition, the vision camera (430) may have a dustproof structure suitable for a cleanroom environment, and accordingly, the detection of false detections caused by fine dust can be minimized, thereby significantly improving inspection reliability.

[0077] The transfer device (420), lighting device (440), and vision camera (430) can be organically combined with the defect detection device.

[0078] Specifically, the transfer device (420) can provide an operating space for the lighting device (440) and the vision camera (430) during the process of moving the substrate (W), and the lighting device (440) illuminates the lower surface of the substrate (W) through a light source (441) so that the vision camera (430) can obtain a clear image.

[0079] The image obtained from the vision camera (430) can be transmitted to a defect detection device and analyzed by a deep learning-based image processing algorithm, thereby generating a distribution of defects as mapping data, and if the number of defects exceeds a threshold, an alarm signal can be transmitted through a defect alarm device.

[0080] The defect alarm device can perform the role of transmitting an alarm signal to the user when the number of defects detected by the defect detection device exceeds a preset threshold.

[0081] The fault alarm device can generate visual (e.g., LED indicator or screen notification), auditory (e.g., buzzer sound), or network-based (e.g., email, SMS, or system log) alarms through a user interface (UI) or external device, and can be integrated with the EFEM's control console or upper-level monitoring system.

[0082] The fault alarm device may include an alarm module (e.g., buzzer, LED array, or wireless transmitter) in terms of hardware, and may provide a threshold setting interface in terms of software, allowing the user to customize the threshold (e.g., more than 5 faults or more than 1 fault / cm²).

[0083] The defect alarm device can have a dustproof and contamination-resistant structure suitable for cleanroom environments and includes a battery backup function to maintain the alarm even in the event of a power failure. With this configuration, the defect alarm device can reduce the real-time response time to within one second and automatically induce process stoppage or defective wafer classification, thereby minimizing production losses.

[0084] The defect alarm device can be organically combined with the defect detection device and can determine alarm conditions by receiving data analyzed from the defect detection device.

[0085] Specifically, the defect detection device can calculate the number and type of defects using a deep learning-based algorithm and then transmit a signal to the defect alarm device, and the defect alarm device can classify the alarm intensity (e.g., warning, error, emergency) in multiple stages based on this.

[0086] The defect alarm device can reflect real-time feedback of captured data through integration with a vision camera and lighting device, and can include the location of defects in alarm messages by referring to mapping data. Through this organic integration, the defect alarm device can enhance the monitoring capabilities of the entire EFEM system and significantly improve process stability by preventing the risk of mass defects in advance.

[0087] The defect alarm device can overcome the limitations of manual monitoring in existing systems to provide automated responses, thereby reducing production costs and improving yield.

[0088] The defect alarm device may additionally include AI-based prediction capabilities and can warn of potential defects in advance by learning past defect patterns.

[0089] In addition, the defect alarm device can support remote alarms through cloud integration and improve integrated management capabilities on the production line.

[0090] According to one embodiment, real-time full inspection can be implemented within EFEM through the above-described configurations, thereby preventing the occurrence of large-scale defects in advance and significantly improving process yield.

[0091] In addition, one embodiment can be integrated into a compact design that takes into account the space constraints of the EFEM, allowing for installation without modifying the existing system and improving the productivity of the entire system.

[0092] In another embodiment, the lighting device (440) may optionally include a dark field lighting mode, which can highlight low-contrast defects by irradiating light at an oblique angle to the lower surface of the substrate (W).

[0093] In addition, the defect detection device may additionally include a machine learning-based repeatability evaluation function, which can ensure the long-term stability of the system by verifying the consistency of inspection results under the same conditions.

[0094] The vision camera is fixedly positioned at the bottom of the transfer arm and can perform the role of capturing images of the lower surface of the substrate in real time, and may include a lens.

[0095] The vision camera may include a high-resolution image sensor and can provide micron-level resolution to detect minute defects on a substrate.

[0096] Vision cameras can maintain uniform brightness by linking with a special lighting system and can have high-speed performance capable of capturing and processing tens of images per second. Vision cameras can be integrated with software to apply edge detection or pattern recognition algorithms, and can detect defects with high accuracy by utilizing deep learning-based models.

[0097] A vision camera according to one embodiment may have a dustproof structure suitable for a cleanroom environment, and accordingly, false detection caused by fine dust can be minimized.

[0098] In addition, the vision camera is composed of multiple units to capture images of the underside of the substrate from various angles, and this configuration overcomes the field of view limitations of a single camera, enabling comprehensive detection of complex defect shapes.

[0099] In some cases, a vision camera can enable 100% inspection while maintaining the turnaround time (TAT) through high-speed shooting capabilities. Due to these capabilities, the vision camera can significantly improve process efficiency by minimizing the time delay of conventional sample inspection methods, and can support real-time image analysis by being organically combined with a defect detection device.

[0100] A vision camera can acquire clear images or videos based on light provided from a lighting device, and transmit the video to a defect detection device to generate mapping data of the defect distribution through deep learning-based analysis.

[0101] Due to this organic combination, the vision camera can maintain stable operation during the movement of the transfer device, and can achieve the effects of preventing mass defects and improving process yield.

[0102] A lighting device according to one embodiment can perform the role of irradiating light in the direction of the lower surface of a substrate and may include a light source.

[0103] The lighting device may include a bright field lighting mode and can provide uniform brightness by directly irradiating light onto the substrate surface.

[0104] The lighting device may optionally include a dark field lighting mode and can emphasize scattered light at defective areas by illuminating light at an oblique angle.

[0105] In addition, the lighting device may include a short-wavelength infrared (SWIR) light source and can detect internal defects, such as internal cracks or air pockets, by irradiating light that passes through a silicon substrate.

[0106] The lighting device can project a structured light pattern, and a digital light projector can be used to analyze the three-dimensional topography of the substrate surface.

[0107] In another embodiment, the lighting device may include an LED array, have high intensity and long lifespan, and provide wide-angle coverage tailored to the substrate size.

[0108] The lighting device can be compactly installed in the limited space of the EFEM and can support high-speed scanning by synchronizing with the vision camera. Through these capabilities, the lighting device can overcome the limitations of conventional visible light-based lighting to expand the defect detection range and significantly improve inspection accuracy by maintaining high uniformity.

[0109] The lighting device can be organically combined with a vision camera to provide optimized light to the underside of the substrate, and can be linked with a defect detection device to support real-time analysis.

[0110] FIG. 5 is a top view of a wafer bottom defect detection device and a transfer device inside an EFEM according to one embodiment.

[0111] Referring to FIG. 5, an EFEM according to one embodiment may include a transfer device (520), a lighting device (540), a vision camera (530), etc.

[0112] The transfer device (520) may include a transfer arm (521), and the transfer arm (521) may be arranged in a plurality of rod structures extending in a horizontal direction to support the substrate (W). Due to this upper surface arrangement, the transfer arm (521) can expose the lower surface of the substrate (W) over a wide area and provide a planarly stable support structure to maintain the alignment of the substrate (W) during movement.

[0113] In addition, the multiple rods of the transfer arm (521) are spaced apart to secure the field of view of the vision camera (530), thereby enabling high-speed transfer and defect detection to be performed simultaneously within the limited planar space of the EFEM.

[0114] The vision camera (530) may include a lens (531) and is positioned at the lower center of the transfer arm (521) to comprehensively photograph the lower surface of the substrate (W) from an upper perspective. Due to this planar arrangement, the vision camera (530) can cover the entire circular surface of the substrate (W) as a single field of view, and when multiple cameras are configured, the radial distribution of defects can be accurately captured through multi-angle shooting.

[0115] The transfer device (520), the lighting device (540), and the vision camera (530) can be organically combined with the defect detection device. Specifically, the transfer device (520) can maintain the alignment of the lighting device (540) and the vision camera (530) during the planar movement process, and the lighting device (540) can optimize image acquisition quality by concentrating light around the lens (531) of the vision camera (530) through a ring structure.

[0116] The image acquired from the vision camera (530) can be transmitted to a defect detection device for analysis, and through this planar coupling relationship, even edge defects of the substrate (W) can be detected in real time within the EFEM, and the compactness of the entire system can be enhanced by a design that takes space efficiency into account.

[0117] FIG. 6 is a top view of a wafer bottom defect detection device and a transfer device inside an EFEM according to another embodiment.

[0118] Referring to FIG. 6, an EFEM according to another embodiment may include a transfer device (620), a lighting device (640), a vision camera (630), etc.

[0119] The transfer device (620) may include a transfer arm (621), and the transfer arm (621) is formed as a rod structure extending in a horizontal direction so as to be able to place and move the substrate (W). According to the present embodiment, the transfer arm (621) can increase the stability of the substrate (W) by providing a wider support surface on the upper surface, and in particular, minimize alignment deviations when the substrate (W) rotates. Due to this upper surface arrangement, the transfer device (620) can improve space efficiency by optimizing the movement path of the substrate (W) within the EFEM.

[0120] The lighting device (640) may include a light source (641), and the light source (641) may be spaced apart from each other at the bottom of the transfer arm (621) to illuminate the lower surface of the substrate (W).

[0121] According to one embodiment, the light source (641) is arranged in a square array on the upper surface to uniformly cover the entire area of ​​the substrate (W), and the lighting intensity can be easily controlled through a linear distribution instead of a ring structure. With this arrangement, the lighting device (640) can provide consistent lighting in the center and edge regions of the substrate (W), and can accommodate various substrate shapes by compensating for the limitations of the existing ring structure.

[0122] The vision camera (630) may include a lens (631) and may be fixedly positioned at the lower part of the transfer arm (621) by a vision camera fixing device (632). The vision camera fixing device (632) may include an adjustable structure, which can maintain an optimal shooting position according to the movement of the substrate (W) by adjusting the angle and height of the lens (631) in real time.

[0123] The vision camera fixing device (632) can stably fix the vision camera (630) to the lower part of the transfer arm (621) and optimize the shooting position and angle of the lens (631).

[0124] The vision camera fixing device (632) may include an adjustable bracket structure, which enables fine adjustment in horizontal and vertical directions, thereby allowing the focus and field of view of the lens (631) to be adjusted in real time according to the movement path of the substrate (W).

[0125] The vision camera fixing device (632) can be made of a durable metal material, such as stainless steel or aluminum alloy, to prevent corrosion or dust accumulation in a cleanroom environment, and can include a vibration-absorbing pad to minimize image distortion caused by external vibrations.

[0126] Additionally, the vision camera fixing device (632) may include an automatic adjustment sensor, which can detect changes in the position of the substrate (W) and automatically correct the angle of the lens (631).

[0127] The vision camera fixing device (632) can accommodate various substrate sizes and shapes and can maintain high precision in defect detection by providing a stable shooting environment even during high-speed transport.

[0128] The vision camera fixing device (632) is organically combined with the lighting device (640) and the defect detection device, and can support real-time analysis of the image acquired through the lens (631), and can increase equipment integration with a compact design that takes into account the space constraints of the EFEM.

[0129] The light source (641) is a core component of the lighting device (640) and is spaced apart from each other to illuminate the lower surface of the substrate (W).

[0130] The light source (641) can be composed of a high-efficiency LED module and can detect surface and internal defects of the substrate (W) by emitting white light or short-wavelength infrared (SWIR) light.

[0131] The light source (641) can be arranged in a square or grid arrangement on the upper surface and can be designed to uniformly cover the entire area of ​​the substrate (W).

[0132] The light source (641) may include a variable dimming function and can dynamically adjust the light intensity according to the reflectance or color change of the substrate (W).

[0133] The light source (641) may include a thermal management system, for example, a thermoelectric element (TEC) or a fanless cooling structure, which can prevent contamination by heat in the cleanroom environment of the EFEM.

[0134] The light source (641) can optimize light scattering by combining with a diffuser plate, which can minimize shadow effects and provide high illumination uniformity. Due to these functions, the light source (641) can overcome the limitations of conventional linear lighting arrays to accommodate various substrate shapes and support high-resolution image acquisition, thereby significantly improving the reliability of defect detection.

[0135] The light source (641) can be synchronized with the vision camera (630) to enable high-speed scanning, and can enhance the real-time monitoring environment through integration with the defect detection device.

[0136] The vision camera (630) can precisely capture the planar defect distribution of the substrate (W) and can accommodate various substrate sizes due to the flexibility of the fixture (632).

[0137] In addition, when multiple vision cameras (630) are configured, cross-photography is possible on the upper surface, which can increase the comprehensiveness of defect detection.

[0138] The transfer device (620), the lighting device (640), and the vision camera (630) can be organically combined with the defect detection device. Specifically, the transfer device (620) supports the alignment of the lighting device (640) and the vision camera (630) in the upper surface movement path, and the lighting device (640) provides uniform light to the lower surface of the substrate (W) through a light source (641), enabling the vision camera (630) to obtain a clear image.

[0139] The image acquired from the vision camera (630) can be stably transmitted to the defect detection device through the vision camera fixing device (632), thereby enabling the generation of defect distribution mapping data and the transmission of alarm signals. This organic combination relationship can implement real-time full inspection by efficiently utilizing the upper surface space of the EFEM, and can improve process yield by overcoming space constraints.

[0140] The lighting device (640) may additionally include a variable dimming function in the light source (641), which can increase detection accuracy by dynamically adjusting the light intensity according to the reflectance of the substrate (W).

[0141] Additionally, the vision camera fixing device (632) may include an automatic adjustment sensor, which can optimize the focus of the lens (631) by detecting changes in the position of the substrate (W) in real time.

[0142] The lighting device (540) may include a light source (541), and the light source (541) may be arranged in a ring structure centered on the lens (531) of the vision camera (530) to uniformly illuminate the lower surface of the substrate (W) in a circular manner. Due to this upper surface ring arrangement, the lighting device (540) can evenly distribute light to the entire edge area of ​​the substrate (W), and overcome the limitations of the conventional linear arrangement method, thereby maintaining consistent lighting even when the substrate (W) is rotated or its position changes. In addition, the multiple spaced arrangements of the light source (541) can prevent hot spots, and by combining with a diffuser plate in a planar manner to enhance the light scattering effect, the contrast of fine defects can be increased.

[0143] FIG. 7 is a drawing showing a wafer bottom surface defect detection device and a transfer device inside an EFEM according to another embodiment.

[0144] Referring to FIG. 7, an EFEM according to another embodiment may include a transfer device (720), a lighting device (740), a vision camera (730), etc.

[0145] The transfer device (720) may include a transfer arm (721), and the transfer arm (721) may perform the function of placing and moving the substrate (W).

[0146] The lighting device (740) may include a light source (741) and a diffuser plate (742) and may perform the function of irradiating light onto the lower surface of the substrate (W).

[0147] The light source (741) can be formed in a ring structure and arranged in a circular pattern around the lens (731) of the vision camera (730) to uniformly illuminate the entire lower surface of the substrate (W).

[0148] The ring structure can be configured by connecting multiple LED modules in a circular shape, and each LED can emit high-intensity white light or SWIR light.

[0149] The light source (741) can be designed so that the inner and outer diameters of the ring match the diameter of the substrate (W), and the inner diameter can be adjusted to an optimized numerical range so as not to obstruct the field of view of the vision camera (730).

[0150] The ring structure can optimize light reflection and concentration and can be positioned coaxially with the vision camera (730) in a coaxial array to minimize glare on the highly reflective surface of the substrate (W).

[0151] The light source (741) may include a variable voltage control function, which can adjust the light intensity according to the material or process step of the substrate (W).

[0152] Additionally, the light source (741) may include a built-in heat sink for heat dissipation, which can prevent thermal contamination in the EFEM's cleanroom environment and ensure a long lifespan. Due to this ring structure, the light source (741) can overcome the limitations of conventional linear arrays and provide uniform illumination up to the circular edge region of the substrate (W), and can increase detection accuracy by enhancing the contrast of fine defects.

[0153] The light source (741) can be combined with a diffuser plate (742) to amplify the light scattering effect, and can maintain a stable light source during high-speed shooting by synchronizing with the vision camera (730).

[0154] The diffuser plate (742) can be placed on the light path of the light source (741) and can perform the role of minimizing shadow effects by optimizing the scattering of light.

[0155] The diffuser plate (742) can be made in the form of a thin plate using a high-quality optical material, for example, acrylic (polycarbonate, PC) or polymethyl methacrylate (PMMA), and may contain fine scattering particles (e.g., silica or polymer beads) inside so that light rays can be scattered in multiple directions as they pass through.

[0156] These fine particles can reduce the straightness of light and mitigate the point light source characteristics of the light source (741), thereby providing a uniform light distribution on the surface of the substrate (W).

[0157] The surface of the diffuser plate (742) can be processed into a fine texture, which can control the reflection of light to prevent hotspots or uneven brightness areas.

[0158] The diffuser plate (742) can be placed directly in front of or above the LED light source (741) in the light path and can be customized into a circular, square, or ring shape to fit the size of the substrate (W).

[0159] The thickness of the diffuser plate (742) can be adjusted to a range of 1 to 5 mm, which can enable a compact design that takes into account the space constraints of the EFEM.

[0160] The diffuser plate (742) can optimize light scattering to minimize shadow effects, specifically reducing shadows or uneven lighting patterns that may occur due to the spacing between LED light sources (741).

[0161] This can prevent defect detection errors that may be particularly prominent in the edge region of the substrate (W).

[0162] The diffuser plate (742) can achieve high uniformity by causing the directionality of light to be lost, thereby increasing the contrast and clarity of the image acquired by the vision camera (730).

[0163] Additionally, the diffuser plate (742) can maintain stable illumination even during high-speed transport and can compensate for illumination imbalances caused by positional fluctuations that occur as the substrate (W) moves. The diffuser plate (742) may include a dustproof coating suitable for a cleanroom environment and has heat resistance, thereby preventing deformation caused by LED heat generation. Due to these functions, the diffuser plate (742) can overcome the limitations of the conventional direct irradiation method to increase the accuracy of detecting micro-defects and significantly improve process yield in a mass production environment.

[0164] The vision camera (730) may include a lens (731) and can capture the lower surface of the substrate (W) in real time. In this embodiment, the vision camera (730) is positioned on the side below the light source (741) and the diffuser plate (742) to capture an image of the lower surface of the substrate (W).

[0165] The transfer device (720), lighting device (740), and vision camera (730) can be organically combined with the defect detection device.

[0166] Specifically, the transfer device (720) can provide an operating space for the lighting device (740) and the vision camera (730) during the process of moving the substrate (W), and the lighting device (740) can illuminate the lower surface of the substrate (W) through a light source (741) and a diffuser plate (742) so that the vision camera (730) can obtain a clear image.

[0167] The light source (741) may include a multi-wavelength LED within a ring structure, which can simultaneously detect surface and internal defects by irradiating visible light and SWIR.

[0168] Additionally, the diffuser plate (742) may further include an adaptive haze control function, which can increase inspection precision by automatically adjusting the degree of scattering according to changes in the thickness of the substrate (W).

[0169] FIG. 8 is a drawing showing a wafer bottom defect detection device and a transfer device inside an EFEM according to another embodiment.

[0170] Referring to FIG. 8, an EFEM according to another embodiment may include a transfer device (820), a lighting device (840), a vision camera (830), etc.

[0171] The transfer device (820) may include a transfer arm (821), and the transfer arm (821) may perform the function of placing and moving the substrate (W).

[0172] The transfer arm (821) may include a reinforced support structure and can reduce shaking during movement through an asymmetric arrangement that takes into account the center of gravity of the substrate (W), thereby increasing stability during high-speed transfer of the substrate (W) and ensuring the accuracy of the defect detection process.

[0173] The lighting device (840) may include a light source (841), and the light source (841) may be formed in a ring structure to illuminate the lower surface of the substrate (W).

[0174] The light source (841) can be positioned around the lens (831) of the vision camera (830), and the outer diameter of the ring can be extended to the edge of the substrate (W) to uniformly cover the entire area.

[0175] The light source (841) can be composed of high-efficiency LEDs and can include independent power control functions for each segment within the ring, allowing for adjustment of lighting intensity in specific areas.

[0176] The ring structure of the light source can optimize light distribution by reflecting the circular characteristics of the substrate (W) and can improve edge defect detection capability compared to the existing linear array.

[0177] The light source (841) may include a heat-resistant coating, which can prevent contamination by heat in the EFEM's cleanroom environment.

[0178] The vision camera (830) may include a lens (831) and can photograph the lower surface of the substrate (W) in real time.

[0179] The transfer device (820) can align the field of view of the lighting device (840) and the vision camera (830) along the movement path of the substrate (W), and the lighting device (840) can provide uniform light to the lower surface of the substrate (W) through a light source (841) so that the vision camera (830) can obtain a clear image.

[0180] The image acquired from the vision camera (830) can be transmitted to a defect detection device for analysis, and this combined surface area relationship can maximize space utilization within the EFEM to efficiently perform real-time inspection.

[0181] The vision camera can perform the role of capturing images of the lower surface of the substrate in real time within the EFEM and can be composed of multiple units.

[0182] Multiple vision cameras can be fixedly positioned at the bottom of the transfer arm, and each vision camera includes a high-resolution image sensor capable of detecting minute defects on the substrate in micron units.

[0183] Multiple vision cameras can simultaneously capture the lower surface of a substrate from multiple angles, and, for example, can overcome the field of view limitations of a single camera by dividing the coverage into side, center, and edge areas. Each of the multiple vision cameras may include a lens, and each lens may be designed in various ways, such as wide-angle or telephoto types, to comprehensively cover the entire surface of the substrate.

[0184] Multiple vision cameras can have a dustproof structure suitable for cleanroom environments, thereby minimizing false detections caused by fine dust. With this configuration, the multiple vision cameras can enable 100% inspection while maintaining the turnaround time (TAT) through high-speed shooting capabilities, and can significantly improve inspection reliability by comprehensively detecting complex defect shapes.

[0185] Multiple vision cameras can be organically combined with a lighting device, and can acquire clear images based on light provided by the lighting device.

[0186] Multiple vision cameras can be linked with a defect detection device and analyzed by a deep learning-based image processing algorithm, thereby generating mapping data of the defect distribution and transmitting an alarm signal when the number of defects exceeds a threshold.

[0187] Multiple vision cameras may additionally include an independent focus adjustment mechanism for each camera, which can automatically correct for changes in the position of the substrate to improve detection accuracy.

[0188] In addition, multiple vision cameras may include stereo vision capabilities, which can overcome the limitations of conventional planar imaging by enabling three-dimensional defect analysis.

[0189] Multiple vision cameras can be linked with a defect detection device and analyzed by a deep learning-based image processing algorithm, thereby generating mapping data of the defect distribution and transmitting an alarm signal when the number of defects exceeds a threshold.

[0190] Deep learning-based image processing algorithms can analyze images captured by a semiconductor wafer defect detection system to automatically identify and classify defects.

[0191] The algorithm is based on a deep network structure such as a convolutional neural network (CNN) and can perform image feature extraction, pattern recognition, and anomaly detection, and can be suitable for high-speed processing environments by applying a real-time object detection model such as YOLO (You Only Look Once).

[0192] In addition, using models that integrate the Inception module or skip connections allows for the effective extraction of multi-scale features, which enables the accurate classification of various defect types such as micro-scratches, particles, cracks, and contaminants.

[0193] The algorithm can be trained using a large amount of training data and can implement automated defect classification by taking Scanning Electron Microscope (SEM) images or optical images as input.

[0194] For example, a CNN model that combines edge detection and segmentation can precisely analyze the location and size of defects and overcome the limitations of existing manual inspection or simple image processing methods, thereby improving detection speed to tens of frames per second.

[0195] The algorithm can include self-learning capabilities and maintain long-term accuracy by updating the model in real-time when new defect patterns occur.

[0196] Mapping data can be generated as 2D or 3D data that visualizes the location, type, and distribution of defects in semiconductor wafer inspection.

[0197] Mapping data can be generated by mapping defect information collected from inspection tools to the wafer's coordinate system, and can be expressed in the form of heatmaps or vector graphics to intuitively analyze defect density or patterns.

[0198] For example, mapping data can be automatically stored in a manufacturing execution system (MES), which can track the causes of defects and generate lot-level statistics. Mapping data can sequentially identify critical areas through a feedback loop and, when combined with deep learning models, enable automatic labeling.

[0199] Mapping data can detect defects such as cracks, scratches, and chipping during wafer production, and process efficiency can be improved through data visualization tools.

[0200] In one embodiment, a vision camera is installed below the wafer transfer module to photograph the back side of the wafer and detect defects. The vision camera is fixedly positioned at the bottom of the transfer arm and can photograph the back side of the wafer in an exposed state in real time while the wafer is being removed from the FOUP. It includes a high-resolution image sensor and can identify micro-defects (e.g., scratches, contamination, cracks) in micron units.

[0201] The vision camera can have a dustproof structure suitable for cleanroom environments, thereby minimizing false detections caused by fine dust. The vision camera can be integrated without altering the structure of the existing EFEM and can naturally perform inspections during wafer transfer, allowing for the immediate detection of defects without a separate process.

[0202] The vision camera is composed of multiple units that can capture images of the back of the wafer from various angles, overcoming the field of view limitations of a single camera to comprehensively detect complex defect shapes. The vision camera enables real-time inspection while maintaining the TAT of the existing EFEM, and can prevent yield reduction caused by contamination or damage during the process.

[0203] One embodiment can prevent the occurrence of mass defects and improve process yield by inspecting the back side of the wafer in real time.

[0204] 100% inspection can perform real-time imaging and analysis on all wafers during transport, overcoming the limitations of sample inspection methods to comprehensively monitor defects in the entire lot. The inspection method includes high-speed image processing capabilities to analyze tens of images per second, allowing it to proceed without delay within the EFEM.

[0205] 100% inspection can prevent the accumulation of defects during the process by detecting minute defects early, and can improve production yield. 100% inspection can enhance economic efficiency in mass production environments and can support automatic classification by linking with defect detection devices.

[0206] One embodiment allows for the design of a vision camera installation space and structure while maintaining the Turn Around Time (TAT) of the existing equipment (EFEM). The vision camera can be compactly placed in a limited space below the transport module and can overcome the structural limitations of the existing EFEM by including a special bracket structure.

[0207] Inspection can be performed without increasing TAT by synchronizing with the wafer transfer speed, and the overall size of the EFEM can be changed by optimizing the installation space.

[0208] TAT maintenance design can minimize the cost of modifying existing equipment and significantly improve production line efficiency.

[0209] One embodiment can develop vision specifications, software (S / W), and hardware (H / W) to detect defects on the back side of a wafer.

[0210] The vision specifications may include a high-resolution CMOS or CCD sensor and, together with a wide-angle lens, can capture micron-sized defects.

[0211] The vision specifications may include a high-resolution CMOS (Complementary Metal-Oxide-Semiconductor) or CCD (Charge-Coupled Device) sensor and, together with a wide-angle lens, can capture micron-sized defects.

[0212] Vision specification high-resolution CMOS (Complementary Metal-Oxide-Semiconductor) sensors are characterized by fast read speeds and low power consumption, making them suitable for real-time wafer defect detection.

[0213] CMOS sensors can provide a resolution of 1 to 2 μm at the pixel level, which allows for the precise capture of defects such as fine scratches or particles on the wafer surface.

[0214] On the other hand, high-resolution CCD (Charge-Coupled Device) sensors have high sensitivity and low noise characteristics and can achieve a resolution of 0.5 to 1 μm, which can be advantageous for detecting internal defects (e.g., cracks, air pockets) in silicon wafers.

[0215] CMOS and CCD sensors can be designed in a hybrid configuration that combines their respective advantages, which can maintain stable image quality under various lighting conditions (e.g., short wavelength infrared (SWIR)).

[0216] The vision specifications, combined with a wide-angle lens, can capture micron-sized defects. The wide-angle lens can extend the field of view (FOV) to a range of 60 to 120 degrees, enabling the entire 300mm wafer to be covered in a single shot.

[0217] The wide-angle lens may include an aspherical design, which can minimize distortion and maintain resolution in the edge area. The focal length of the lens can be adjusted from 10 to 50 mm, thereby enabling depth of field clarity capable of capturing defects in the micron range (e.g., 0.5 to 5 μm).

[0218] Wide-angle lenses can be equipped with multiple coatings (e.g., anti-reflective coatings) to reduce light reflection and prevent image distortion caused by dust or contamination in cleanroom environments. With this configuration, the wide-angle lens maximizes the field of view of the vision camera, enabling the analysis of the entire area of ​​the wafer in a single scan and the detection of minute defects without omission, even during high-speed transport.

[0219] The CCD sensor can convert light input through a wide-angle lens into a digital signal, and the high-resolution data can be transmitted to software in real time and analyzed by a defect detection device.

[0220] It can be implemented with a compact design that takes into account the space constraints of the EFEM, and can support video processing of tens of frames per second by linking with high-speed processing hardware (H / W).

[0221] Vision specifications can overcome the limitations of existing sample inspection methods to enable 100% wafer inspection and significantly improve process yield.

[0222] The software (S / W) can perform edge detection, pattern recognition, and anomaly detection by integrating deep learning-based algorithms, and can classify defect types with high accuracy.

[0223] The hardware (H / W) may include a high-speed processing board linked to an industrial PC and may have a dustproof and heat-resistant structure suitable for cleanroom environments. The vision system can precisely detect micro-contamination or damage on the back of the wafer and can enhance process reliability by overcoming the detection limitations of existing technologies.

[0224] One embodiment may link vision image data with equipment communication to prevent mass defects through real-time inspection. Vision image data can be transmitted to a defect detection device in real time for analysis, and the analysis results can be transmitted to an EFEM control system or a higher-level monitoring system to trigger an alarm or process stoppage signal. This communication linkage can be implemented via Ethernet or wireless protocols, and data delay can be maintained within one second. Through this function, the communication linkage enables early detection of mass defects, reduces production costs, and significantly improves process stability.

[0225] One embodiment can be verified through repeatability and continuity evaluations to enhance vision detection capability. Repeatability evaluation can confirm an agreement rate of 99% or higher in more than 100 inspections under the same conditions, and continuity evaluation can ensure system reliability by evaluating stability during long-term operation (e.g., more than 1,000 hours). This evaluation process may include automated test scripts, and software (S / W) parameters can be optimized based on the verification results.

[0226] Multiple vision cameras can transmit images acquired from each camera to a defect detection device in real time, and the defect detection device, composed of an industrial PC or dedicated hardware, can process multiple input streams. The multiple vision cameras can automatically analyze captured images in conjunction with a deep learning-based image processing algorithm, and the deep learning-based image processing algorithm may include a Convolutional Neural Network (CNN) or a Recurrent Neural Network (RNN). The deep learning-based image processing algorithm is trained using training data to automatically classify defect types such as scratches, particles, contaminants, and cracks on the underside of the substrate, and can maintain a detection accuracy of over 95%. The deep learning-based image processing algorithm can integrate edge detection, pattern recognition, and anomaly detection functions, and can correct image distortion occurring during the real-time movement of the substrate. Through this analysis, the deep learning-based image processing algorithm can overcome the limitations of conventional manual analysis and improve the processing speed of large amounts of data to tens of frames per second.

[0227] The distribution of defects can be generated as mapping data, which can be configured to visualize the location, size, and type of defects in a 2D or 3D coordinate system on the underside of the substrate. The mapping data can be automatically stored in a Manufacturing Execution System (MES) and linked with a database for process analysis to track the causes of defects. The mapping data can be generated in the form of a heatmap or vector graphics and can support lot-based statistical analysis by including a defect density calculation function. Through these capabilities, the mapping data can provide immediate feedback in the event of mass defects and improve process yield.

[0228] If the number of defects exceeds a threshold, an alarm signal can be transmitted, and the alarm signal can be sent via a user interface or an external alarm device in the form of a visual, auditory, or network message.

[0229] The threshold can be set by the user and, for example, defined as a number of defects of 5 or more or a density of specific type defects of 1 or more / cm².

[0230] Alarm signals can trigger process stop commands or the classification of defective substrates, which can be linked with a higher-level monitoring system to trigger an automated response.

[0231] Alarm signals can prevent the risk of mass defects in advance and significantly increase the stability of the production line.

[0232] Multiple vision cameras, deep learning-based image processing algorithms, mapping data generation, and alarm signal transmission can be organically combined.

[0233] Multiple vision cameras can supply images to a defect detection device, and a deep learning-based image processing algorithm can generate mapping data based on analysis results and issue an alarm signal if the statistics of the mapping data exceed a threshold.

[0234] The entire system can implement real-time full inspection within EFEM, and can significantly improve process efficiency by overcoming the limitations of existing sample inspection methods.

[0235] FIG. 9 is a drawing showing a vision camera mounted on a side storage buffer (SSB) according to another embodiment.

[0236] Referring to FIG. 9, the side storage (900) is a space for temporarily storing or buffering a substrate (W) and may include a support structure that supports the substrate (W) through its outer edge. Through this support structure, the entire rear surface of the substrate (W), excluding its edge, may be exposed, and in this state, inspection or thickness measurement using an upper vision camera (920) and a lower vision camera (930) may be possible.

[0237] The side storage (900) may be composed of a side storage chamber (911) and a side storage body (912).

[0238] The upper vision camera (920) is positioned on the upper side of the side storage chamber (911) to capture images of the upper surface of the substrate (W) in real time, and the upper lighting device (921) can uniformly illuminate the upper surface to increase the accuracy of defect detection.

[0239] The lower vision camera (930) is positioned at the bottom of the side storage body (912) to photograph the back side of the substrate (W), and the lower lighting device (931) illuminates the back side to detect fine defects.

[0240] The upper vision camera (920) and the lower vision camera (930) may each include a high-resolution image sensor and may have the capability to identify defects in the micron range.

[0241] In addition, it includes a dustproof structure suitable for cleanroom environments, which can minimize false detections caused by fine dust.

[0242] The upper lighting device (921) and the lower lighting device (931) may include a short-wavelength infrared light source, thereby providing a function to detect internal defects of the substrate (W). With this configuration, the side storage (900) can be linked with the transfer device of the EFEM to perform additional inspection after the substrate (W) is removed from the FOUP, thereby preventing mass defects and significantly improving the process yield.

[0243] FIG. 10 is a drawing showing a substrate being transferred to a side storage (SSB) according to another embodiment.

[0244] Referring to FIG. 10, the side storage (1000) is a space for temporarily storing or buffering a substrate (W), and the substrate (W) can be moved to the side storage (1000) by being removed from inside a FOUP or EFEM by a transfer device (1020).

[0245] The transfer device (1020) may include a transfer arm (1021), and the transfer arm (1021) may perform the function of stably moving the substrate (W) by placing it on the transfer device. During this transfer process, a structure supported through the outer edge of the substrate (W) is applied so that the entire rear surface excluding the edge can be exposed, and in this state, vision inspection or thickness measurement can be performed in real time.

[0246] The side storage (1000) may be composed of a side storage chamber (1011) and a side storage body (1012), and a substrate (W) may be loaded into the side storage chamber (1011) by a transfer arm (1021).

[0247] The upper vision camera (1020) is positioned on the upper side of the side storage chamber (1011) to capture images of the upper surface of the substrate (W) in real time, and the upper lighting device (1021) can uniformly illuminate the upper surface to increase the accuracy of defect detection.

[0248] The lower vision camera (1030) is positioned at the bottom of the side storage body (1012) to photograph the back side of the substrate (W), and the lower lighting device (1031) illuminates the back side to detect fine defects (e.g., scratches, contamination, cracks).

[0249] The upper vision camera (1020) and the lower vision camera (1030) may each include a high-resolution image sensor and have the capability to identify defects in the micron range. Additionally, they may include a dustproof structure suitable for a cleanroom environment to minimize false detections caused by fine dust. The upper lighting device (1021) and the lower lighting device (1031) may include a short-wavelength infrared light source, providing the ability to detect even internal defects of the substrate (W). With this configuration, the side storage (1000) is organically linked with the transfer device (1020) of the EFEM, allowing for additional inspection to be performed during the process of moving the substrate (W) from the FOUP to the side storage (1000), thereby preventing mass defects and significantly improving the process yield.

[0250] A vision camera according to one embodiment can capture the state of the substrate at the start of the process in real time when the vision camera is positioned on the FOUP (Front-Opening Unified Pod) side.

[0251] Specifically, at the initial stage when the substrate (W) is removed from the pod, the upper and lower surfaces of the substrate can be simultaneously captured using an upper or lower vision camera, thereby immediately detecting defects such as contamination, scratches, and cracks prior to the process, which can be used as reference data for comparative analysis of subsequent processes.

[0252] In the case of this embodiment, defect inspection is initiated from the moment the substrate enters the EFEM to prevent initial defects in advance, and the stability of the entire production line can be enhanced by generating mapping data of the type and location of micro-defects in conjunction with a deep learning-based algorithm.

[0253] Meanwhile, according to another example, if the vision camera is positioned on the side storage (SSB), substrate defects during the process finishing stage can be inspected in real time.

[0254] For example, as a substrate that has been processed in the process chamber is transferred to side storage, the upper vision camera and the lower vision camera capture images of the front and back surfaces of the substrate from multiple angles, and can detect additional defects that occur during the process (e.g., deformation due to heat treatment or etching, attachment of foreign substances).

[0255] Inspection of the substrate at the side storage (SSB) location can analyze accumulated defects by reflecting the state immediately after the end of the process, and can increase inspection reliability by enabling the detection of internal defects using short-wave infrared (SWIR) through organic linkage with the lighting device.

[0256] In addition, if the number of defects exceeds a preset threshold, an immediate notification is sent via a defect alarm device, allowing for the automation of process stoppage or defect classification.

[0257] One embodiment allows for the real-time detection of defects in the substrate throughout the entire process flow by strategically positioning the vision camera, and enables full inspection (comprehensive monitoring of all substrates) by combining initial inspection on the pod side and final inspection on the SSB side.

[0258] Accordingly, the limitations of existing sample-based inspection methods can be overcome to minimize the risk of mass defects and significantly improve process yield while maintaining turnaround time (TAT), and the compact design that takes into account the space constraints of the EFEM system allows for easy integration into existing facilities.

[0259] Terms such as "include," "compose," or "have" as described above, unless specifically stated otherwise, mean that the relevant component may be inherent; therefore, they should be interpreted as allowing for the inclusion of additional components rather than excluding them. All terms, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains, unless otherwise defined. Commonly used terms, such as those defined in advance, should be interpreted in accordance with their meaning in the context of the relevant technology and should not be interpreted in an ideal or overly formal sense unless explicitly defined in the present invention.

[0260] The foregoing description is merely an illustrative explanation of the technical concept of the present invention, and those skilled in the art to which the present invention pertains will be able to make various modifications and variations within the scope of the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are intended to explain, not limit, the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by these embodiments. The scope of protection of the present invention shall be interpreted by the claims below, and all technical concepts within an equivalent scope shall be interpreted as being included within the scope of rights of the present invention.

Claims

Claim 1 EFEM comprising: a Front-Opening Unified Pod (FOUP) on which a substrate is loaded; a main body communicating with the FOUP through a door and having a first space formed inside for the substrate being transported to stay therein; a transport device disposed inside the main body and capable of placing the substrate on a transport arm to remove the substrate from the FOUP; a lighting device including a light source that irradiates light toward the lower surface of the substrate; a vision camera that captures images of the lower surface of the substrate placed on the transport arm in real time inside the main body while the transport device removes the substrate; and a defect detection device that analyzes images captured by the vision camera to detect defects occurring on the lower surface of the substrate. Claim 2 In claim 1, the lighting device comprises a plurality of light sources spaced apart from the lower surface of the transfer arm, EFEM. Claim 3 In paragraph 1, the light is short-wavelength infrared, EFEM. Claim 4 EFEM according to claim 1, wherein the light source comprises a ring structure formed around the lens of the vision camera. Claim 5 In claim 1, the lighting device comprises a diffuser plate disposed on the path where the light is irradiated, EFEM. Claim 6 In claim 1, the vision camera is fixedly positioned on the lower surface of the transfer arm, EFEM. Claim 7 In claim 1, the number of the vision cameras is a plurality, EFEM. Claim 8 In claim 1, the defect detection device comprises an EFEM including a deep learning-based image processing algorithm. Claim 9 In claim 1, the defect detection device generates mapping data for the distribution of defects present on the lower surface of the substrate based on an image captured by the vision camera, EFEM. Claim 10 EFEM according to claim 1, comprising a defect alarm device that transmits an alarm signal to a user when the number of defects detected by the defect detection device exceeds a preset threshold. Claim 11 EFEM comprising: a Front-Opening Unified Pod (FOUP) on which a substrate is loaded; a main body communicating with the FOUP through a door and having a first space formed inside for the substrate being transported to stay; a transport device disposed inside the main body and capable of placing the substrate on a transport arm to remove the substrate from the FOUP; a side storage for temporarily storing the substrate in a side storage chamber; a lighting device including a light source disposed above or below the side storage chamber to irradiate light toward the substrate; a vision camera disposed above or below the side storage chamber to capture images of the substrate in real time; and a defect detection device that analyzes images captured by the vision camera to detect defects occurring in the substrate.

Images