KR20240051370A - Spin coater with funcion of inspecting coating quality of semiconductor photoresist apparatus - Google Patents

Abstract

The present invention includes a spin chuck that fixes and rotates the wafer; a supply nozzle installed on the upper side of the spin chuck and supplying photoresist to the upper surface of the wafer on the spin chuck; a main control unit that controls the operation of the spin chuck and the supply nozzle; a defect stopping unit that stops the operation of the spin chuck and the supply nozzle when an operation control signal due to a defective photoresist is received from the outside; A photoresist coating quality inspection system that inspects the quality of the photoresist supplied from the supply nozzle or the photoresist supplied to the upper surface of the wafer, and provides the operation control signal to the defect stopper when the quality of the photoresist is poor. ; After the photoresist is applied to the wafer, a defect rate receiver is configured to receive the defect rate from a wafer yield management system that calculates the defect rate for each die of the wafer through probe test data inspected for each die of the wafer in a subsequent process. In addition, the photoresist coating quality inspection system includes a first DVS (Dynamic Vision Sensor) installed in the spin coater to acquire an image of the photoresist supplied from the supply nozzle of the spin coater, and a top surface of the wafer on the spin chuck from the supply nozzle. A second DVS (Dynamic Vision Sensor) installed on the spin coater to obtain an image of the photoresist supplied to and spreading by centrifugal force, It includes a deep learning determination unit that determines whether there is a defect, and an auxiliary control unit that outputs the operation control signal when a defect occurs in the photoresist according to the determination result of the deep learning determination unit.

Description

Spin coater with funcion of inspecting coating quality of semiconductor photoresist apparatus}

The present invention relates to a spin coater, and more specifically, the status of the nozzle used in photo resist (PR) coating, which is one of the semiconductor manufacturing processes, and the coating area of the photo resist using DVS (Dynamic Vision Sensor) and deep learning. This relates to a spin coater capable of inspecting the coating quality of semiconductor photoresist spraying equipment that enables accurate defect detection using technology.

Generally, the photolithography process for pattern formation during the semiconductor device manufacturing process involves applying photoresist to a wafer, exposing it to light using a photomask, and developing it in a developer solution, so that a certain pattern of the photomask is transferred to the photoresist film. It's fair. At this time, a process of applying photoresist to the wafer is performed, which is mainly performed using a spin coater.

In this spin coater, when an externally supplied wafer is placed on a rotary chuck that fixes and rotates the wafer, the rotary chuck fixes the wafer and starts rotating. And the photoresist supply nozzle moves from the standby position to the spraying position and sprays a certain amount of photoresist on the center of the rotating wafer. This liquid photoresist spreads widely on the rotating wafer by centrifugal force and by its own viscosity. It is distributed in an almost even thickness across the wafer. Afterwards, the liquid photoresist goes through a series of curing processes such as baking to form a solid state and is then inputted into the exposure process.

Quality control of the photoresist coating on the wafer by such a spin coater is not only closely related to the wafer defect rate, but is also becoming more important in realizing fine line widths in semiconductors.

As a technology for controlling the coating quality of conventional photoresists, Korean Patent Publication No. 10-2003-0085692, “Photoresist sensing device and sensing method for spin coater,” has been proposed, which is a rotating device that can be rotated by a driving means. Chuck and; a photoresist supply nozzle equipped with a third sensing unit capable of detecting the amount of photoresist and applying a certain amount of photoresist to the wafer located on the upper part of the rotary chuck; a photoresist container provided with a first sensing unit capable of detecting the amount of photoresist and containing photoresist to be dispersed on the wafer; a pumping unit provided with a second sensing unit capable of detecting the amount of photoresist and pumping the photoresist contained in the photoresist container so that it can be supplied to the photoresist supply nozzle; A controller is provided that controls the driving means to rotate the rotary chuck and controls the pumping unit to pump the photoresist contained in the photoresist container and move it to the nozzle for supplying the photoresist.

However, the prior art had limitations in accurately identifying defects that cause poor photoresist coating quality, and as a result, it had a problem in that it did not significantly contribute to reducing the defect rate of wafers.

Another conventional method for detecting photoresist coating defects is to detect defects that show a clear difference in shape compared to the normal area when using a machine vision method, and detects the difference in brightness of the background compared to the lighting. , and shape detection using contours (edges) and defect detection methods using a discrete Fourier transform algorithm that converts the image into frequency space are used. In this case, the following problems may occur.

The first limiting factor of the machine vision method is that the detection method must be created differently every time the image data background and defect elements change, and each time the object of inspection changes, data with new features are collected and features are extracted from the data. Since it is necessary to extract and apply the algorithm, it is difficult to construct an integrated inspection algorithm for various types of defects, and there is a problem that a large amount of additional time and cost are incurred.

In order to solve the problems of the prior art as described above, the present invention uses a DVS (Dynamic Vision Sensor) to determine the status of the nozzle used in photo resist (PR) coating, which is one of the semiconductor manufacturing processes, and the coating area of the photo resist. and deep learning technology to enable accurate defect detection, and in particular to detect the photoresist supply status of the nozzle for photoresist coating, which is a key process equipment in the semiconductor lithography process, to meet the requirements of the semiconductor industry. The purpose is to satisfy the requirements for high reliability and increased yield of devices.

Other objects of the present invention may be easily understood through the description of the examples below.

In order to achieve the above-mentioned purpose, a spin chuck for fixing and rotating the wafer; a supply nozzle installed on the upper side of the spin chuck and supplying photoresist to the upper surface of the wafer on the spin chuck; a main control unit that controls the operation of the spin chuck and the supply nozzle; a defect stopping unit that stops the operation of the spin chuck and the supply nozzle when an operation control signal due to a defective photoresist is received from the outside; A photoresist coating quality inspection system that inspects the quality of the photoresist supplied from the supply nozzle or the photoresist supplied to the upper surface of the wafer, and provides the operation control signal to the defect stopper when the quality of the photoresist is poor. ; After the photoresist is applied to the wafer, a defect rate receiver is configured to receive the defect rate from a wafer yield management system that calculates the defect rate for each die of the wafer through probe test data inspected for each die of the wafer in a subsequent process. In addition, the photoresist coating quality inspection system includes a first DVS (Dynamic Vision Sensor) installed in the spin coater to acquire an image of the photoresist supplied from the supply nozzle of the spin coater, and a top surface of the wafer on the spin chuck from the supply nozzle. A second DVS (Dynamic Vision Sensor) installed on the spin coater to obtain an image of the photoresist supplied to and spreading by centrifugal force, A deep learning determination unit for determining whether a defect exists, and an auxiliary control unit for outputting the operation control signal when a defect occurs in the photoresist according to the determination result of the deep learning determination unit, and the deep learning determination unit is configured to use the deep learning algorithm. The weight learned by is modified to be proportional to the defect rate and used when determining defects in the photoresist, and the first DVS is composed of a pair and arranged at an angle of 90 degrees around the supply nozzle. A spin coater capable of inspecting the coating quality of semiconductor photoresist spraying equipment is provided.

The auxiliary control unit may further include an alarm generation unit that generates an alarm according to an alarm control signal output when a defect in the photoresist occurs according to the determination result of the deep learning determination unit.

The deep learning determination unit determines a defect in the photoresist according to the spacing amount of the photoresist supplied downward from the bottom of the supply nozzle with respect to a virtual vertical line extending vertically downward from the bottom of the supply nozzle, or Defects in the photoresist can be determined based on the ratio of areas on the wafer on which the photoresist is applied that have a color or shape that is different from that of the center.

According to the spin coater capable of inspecting the coating quality of the semiconductor photoresist spraying equipment according to the present invention, the state of the nozzle used in photo resist (PR) coating, which is one of the semiconductor manufacturing processes, and the coating area of the photo resist are monitored by DVS (DVS). Dynamic Vision Sensor) and deep learning technology enable accurate defect detection, improve the detection and processing ability of the image state of the photoresist application process, monitor the nozzle state in real time, and if an abnormal defect state is detected, By sounding an alarm in the unit, production efficiency can be maximized by preventing process defects and enhancing yield. In particular, it detects the photoresist supply status of the nozzle for photoresist coating, which is a key process equipment in the semiconductor lithography process. , has the effect of satisfying the demands for high reliability and increased yield of semiconductor devices required by the semiconductor industry.

Figure 1 is a front view showing the inside of a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 2 is a configuration diagram showing a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 3 is a plan view showing the arrangement of the first DVS in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 4 is a side view for explaining defect determination criteria of a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 5 is a plan view illustrating another example of a defect determination standard for a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 6 is a block diagram of the CNN hardware accelerator of the deep learning judgment unit in the spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 7 is a block diagram of the Convolution Layer module of the deep learning judgment unit in the spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 8 is a block diagram of the Max Pooling Layer module of the deep learning judgment unit in the spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 9 is a block diagram of the Fully Connected Layer module of the deep learning judgment unit in the spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 10 is a block diagram of the input matrix and filter and pooling structure for each layer of the deep learning judgment unit in the spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 11 is a block diagram showing an example of a deep learning judgment unit in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 12 is a block diagram of the Faster R-CNN-based detection implementation of the deep learning judgment unit in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 13 is a diagram for explaining the case of ground-truth and anchor box (Left dispensing detection) of the deep learning judgment unit in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 14 is a diagram illustrating fine adjustment of a support box as a reference box in the deep learning judgment unit of a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 15 is a diagram for explaining the 1x1xN convolution technique of the deep learning judgment unit in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 16 is a C-language based Quasi-Simulator software network structure of the deep learning judgment unit in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 17 is an image showing a learning data sample of the deep learning judgment unit in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.
Figure 18 is an example of an inspection monitoring UI in a spin coater capable of inspecting photoresist coating quality according to an embodiment of the present invention.

Since the present invention can be subject to various changes and can have various embodiments, specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, but should be understood to include all changes, equivalents, and substitutes included in the technical idea and scope of the present invention, and may be modified into various other forms. The scope of the present invention is not limited to the following examples.

Hereinafter, embodiments according to the present invention will be described in detail with reference to the attached drawings, and identical or corresponding components will be assigned the same reference numerals regardless of the reference numerals, and redundant description thereof will be omitted.

Figure 1 is a front view showing the inside of a spin coater equipped with a spin coater capable of inspecting the quality of photoresist coating according to an embodiment of the present invention, and Figure 2 is a coating of a semiconductor photoresist spraying equipment according to an embodiment of the present invention. This is a configuration diagram showing a spin coater capable of quality inspection.

Referring to Figures 1 and 2, the spin coater 10 capable of inspecting photoresist coating quality according to an embodiment of the present invention includes a spin chuck 220, a supply nozzle 210, a main control unit 240, and a defect stopper. It may include a unit 250 and a photoresist coating quality inspection system 100.

The spin chuck 220 fixes and rotates the wafer to be coated with photoresist.

The supply nozzle 210 is installed on the top of the spin chuck 220 and supplies photoresist supplied from the outside to the upper surface of the wafer on the spin chuck 220.

The main control unit 240 controls the operations of the spin chuck 220 and the supply nozzle 210.

The defect stopping unit 250 stops the operation of the spin chuck 220 and the supply nozzle 210 when an operation control signal due to a photoresist defect is received from the outside.

The photoresist coating quality inspection system 100 may include a first Dynamic Vision Sensor (DVS) 110, a second Dynamic Vision Sensor (DVS) 120, a deep learning judgment unit 130, and an auxiliary control unit 140. , and may further include an alarm generating unit 150.

The first DVS 110 is installed in the spin coater 10 to obtain an image of the photoresist supplied from the supply nozzle 210 of the spin coater 10. Like the second DVS 120, which will be described later, the first DVS 110 is about 20 times faster than a general image sensor and is useful for identifying moving objects by accurately recognizing movement even at a relatively long distance.

The first DVS 110 needs to acquire images from various directions in order to accurately detect whether the photoresist supplied downward from the supply nozzle 210 has a uniform thickness and is properly supplied vertically. In order to accurately detect defects in small numbers, as shown in FIG. 3, they can be formed in pairs and arranged at an angle of 90 degrees around the supply nozzle 210.

Each of the first DVSs 110 may be installed in the casing 230 of the spin coater 10 to face the photoresist supplied downward from the supply nozzle 210.

A second Dynamic Vision Sensor (DVS) 120 is installed on the spin coater 10 to obtain an image of the photoresist supplied from the supply nozzle 210 to the upper surface of the wafer on the spin chuck 220 and spread by centrifugal force. For example, the second DVS 120 may be installed to face downward on the nozzle arm 211 to which the supply nozzle 210 is fixed.

The deep learning determination unit 130 determines whether the photoresist is defective using a deep learning algorithm from images acquired from the first and second DVSs 110 and 120. The deep learning judgment unit 130 is located in the control box 231 provided in the casing 230 of the spin coater 10 along with the auxiliary control unit 140, alarm generation unit 150, and defect rate receiver 160, which will be described later. It can be accepted.

For example, as shown in FIG. 4, the deep learning determination unit 130 supplies downward from the bottom of the supply nozzle 210 with respect to a virtual vertical line (H) extending vertically downward from the bottom of the supply nozzle 210. Defects in the photoresist can be determined according to the amount of separation of the photoresist (P), for example, the maximum separation distance (d). This spacing amount can be calculated based on a predetermined thickness of the photoresist (P) through image processing. In addition, as another example, the deep learning determination unit 130 determines the photoresist according to the ratio of the area (A2) having a color or shape that is different from the center (A1) on the wafer (W) on which the photoresist is applied, as shown in FIG. Defects can be identified by using an area calculation algorithm to calculate the difference in color or shape for each pixel through an image process, and then use the center (A1) as a reference to determine the area (A2) with a different color or shape. It can be configured to calculate the area ratio and use it to determine defects in photoresist.

The deep learning decision unit 130 has, for example, a CNN hardware accelerator as shown in FIG. 6 to implement a deep learning algorithm. This CNN hardware accelerator can be implemented with Verilog-based PL (Programmable Logic) hardware, Data transmission can use the AMBA AXI4 bus, and can be implemented as a hardware structure that improves the speed of CNN by processing convolution operations and other operations in parallel for real-time processing of CNN in an embedded environment. In Figures 7 to 10 As shown, it can be composed of a Convolution module, Max Pooling module, and Fully Connected module. Their operation process stores the weights, biases, and input images required for CNN calculation in Block Memory (weight memory, bias memory, and input image) through the Bus Interface, and operates sequentially starting from the Convolution1 module. Max Pooling is used for all It is performed simultaneously on the feature map, and the calculation results of each module are stored in Block Memory. Here, the hardware structure can be implemented so that a multiplier can be used to minimize the execution time of the convolution layer, the convolution operation can be performed in 1 clk, and calculations can be performed continuously using a double buffer.

In addition, the deep learning decision unit 130 may specifically include the configurations shown in FIG. 11, and can be manufactured as an FPGA board for neural network system development, AXI4 simulation environment and logic design, Channel write input arbiter, address/ Data fifo design, Channel write input arbiter, address/data fifo design, Boundary write split design, AXI I/F write wrapper design, AXI Master wrapper I/F, Slave I/F design, FPGA top module design, parallel processor Simulation environment construction and modeling, object recognition algorithm mapping on a parallel processor, etc. are implemented, learned weights are stored in memory, and camera interface design is also implemented.

In the deep learning decision unit 130, the implementation of the base object detection algorithm uses Faster R-CNN to distribute and drop the photoresist to the left and right at the photoresist supply nozzle 210, as shown in FIG. 12, An algorithm for detecting defective areas such as residuals is implemented, and the region proposal includes an extraction method in CNN, allowing the entire process from input to output to be operated as a single network, shortening processing time and improving learning efficiency. It is implemented to be consistent. The Fast R-CNN configuration can be configured to enable accurate image acquisition even for subjects moving at a speed of 1,000 rpm or more, which is difficult for existing image sensors to measure.

Referring to FIG. 13, the deep learning decision unit 130 extracts a feature map from the input image using CNN and downsampling (max pooling), and selects a foreground (object) area where the object to be detected is located and a background (not object). ) Divide the area, pass the feature map through CNN (Conv.3), pass it again through Conv.1, and connect it to a fully connected layer (FC) to classify the background area and foreground area. . This is the process of calculating the position of the foreground area and a rectangle (bounding-box) circumscribed to the foreground area. The location is the center coordinate (x, y) of the rectangle surrounding the foreground area, and the size of the rectangle is the width (w) and length. Height (h) is expressed as x, y, w, h, and the result is six values of foreground, background, x, y, w, and h, which can be used to express the proposed area. The foreground and background each have real values between 0 and 1. This value is obtained using an anchor box and a ground-truth box, and the anchor box is a window for searching the detection area. The value for distinguishing between foreground and background is Equation 1 below.

Here, A is the area of the support box, B is the area of the reference box, and S is the area where the two boxes overlap, allowing you to set area values for the foreground and background.

Referring to FIG. 14, the detection process consists of a ROI (region of interesting) pooling layer, two fully connected layers, left & right dispensing, drop, and residues regions. It may consist of a detection process. The output of the ROI pooling layer passes through two fully connected layers (FC) and is divided into Left & Right dispensing, drop, residues areas, and background. The process of fine-tuning the position of the rectangle surrounding the object area is to confirm the detected area and is also the process of finding a rectangle circumscribed to the detected object.

Referring to Figures 15 and 16, in the case of detection of the reference box and support box, a simulator for operation verification for RTL can be developed, which is necessary for verification to confirm hardware and accurate operation. This can be implemented with a C-language-based simulator, and network compression such as the insertion of a 1 It can be implemented using techniques. In addition, the amount of computation can be improved by reducing the existing network layer and introducing new network techniques such as route (skip connection) and reorg to minimize recognition performance degradation. The process of implementing the profiling function for verification, the process of creating a log file for the input/output information of each layer, and the process of generating the weight and parameter information of each layer as a log file can be performed.

Referring to FIG. 17, the data to be learned has many difficulties in accessing data collection due to security issues due to the nature of the semiconductor industry, making it very insufficient for learning defect data images. Therefore, in order to secure sufficient data for learning, it is possible to use related maintenance companies to secure supply nozzles and photoresists for photoresist coating used in industrial sites, and then secure them through similar environmental experiments at other locations. It can also be secured by using transformation (parallel movement, angle adjustment, etc.) processing on the original image. Learning can be done through four steps: the first step is learning the RPN using training data, the second step is learning the detector using the proposal region obtained from the learned RPN, and the third step is convolution. After fixing the weights of the (CNN) and Max pooling steps, the RPN can be fine-tuned. The fourth step is the process of fine-tuning the detector using the proposed area obtained from the third step. Here, in the third and fourth steps, the weights of the convolution and Max pooling processes are shared. Additionally, by integrating RPNs that exist separately outside into the network, the object detection time can be greatly shortened by sharing the convolution and Max pooling parts.

When a defect occurs in the photoresist according to the determination result of the deep learning determination unit 130, the auxiliary control unit 140 outputs an operation control signal for stopping the photoresist coating operation to the defect stopping unit 250. At this time, the defect stopping unit 250 receives a control signal from the auxiliary control unit 140 and stops the operation of the spin coater 10. In addition, the auxiliary control unit 140 outputs a control signal for generating an alarm when a photoresist defect occurs according to the determination result of the deep learning determination unit 130.

The alarm generator 150 generates an alarm according to a control signal from the auxiliary control unit 140, and displays an alarm message or sounds an alarm through the display unit 170 provided on the outside of the casing 230. It may be composed of a speaker to output, or it may be composed of a light emitting unit to provide an alarm by emitting light or flashing.

The spin coater 100 capable of inspecting the quality of photoresist coating according to an embodiment of the present invention includes an input unit 180 for inputting data or commands, a power supply unit (not shown) that distributes power required for operation, and various data. A memory unit (not shown) for storing programs, etc. may be provided, and after application of photoresist to the wafer, data from a probe test inspected for each die (chip) of the wafer in the subsequent process may be provided. A defect rate receiver 160 may be provided to receive the defect rate as data from the wafer yield management system 300, which calculates the defect rate for each wafer die, using wired or wireless communication.

At this time, the deep learning determination unit 130 may modify the weight learned by the deep learning algorithm to be proportional to the defect rate and use it when determining defects in the photoresist. Here, the proportionality constants range from 0.1 to 10 and can be obtained experimentally to minimize the defect rate. Here, the wafer yield management system 300 receives probe test data inspected for each wafer die, and uses the input probe test data to calculate the yield for each wafer according to a designated yield calculation method. , the corresponding yield along with identification information for each wafer is provided to the defect rate receiver 160.

The spin coater 10 may be provided with a nozzle opening/closing unit 212 for opening and closing the supply of photoresist through the supply nozzle 210, and may include a nozzle arm 211 on which the supply nozzle 210 is fixed within the casing 230. A nozzle transfer unit 213 may be provided to load the onto the upper side of the spin chuck 220 or unload it from the upper side of the spin chuck 220, and the spin coater 10 may be provided with a nozzle opening and closing unit to perform a predetermined process. A main control unit 240 may be provided to control (212), the nozzle transfer unit 213, and the spin chuck 220.

According to the spin coater capable of inspecting the coating quality of the semiconductor photoresist spraying equipment according to the present invention, type data on the state of the supply nozzle of the spin coater is quantified using deep learning technology, and the photoresist application process is quantified. By improving the detection processing ability of the image status, the supply nozzle status can be monitored in real time, and if an abnormal defect condition is detected, an alarm is sounded in the relevant unit, thereby maximizing production efficiency by preventing process defects and enhancing yield.

In addition, according to the present invention, the deep learning-based technology classifies the image based on the defect criteria of the supply nozzle state of the semiconductor photoresist, and the defect area classification and object detection use the input image using a convolutional neural network (CNN). It is a machine vision method in that it can improve defect image processing capabilities and increase inspection reliability by using an image processing algorithm that creates a feature map and performs defect area classification and object detection in image data. It becomes differentiated.

In this way, damage caused by thickness abnormalities and unevenness of the resist film due to supply defects from the photoresist nozzle of the photo-lithography process, which is the initial stage of the wafer manufacturing process, is prevented, resulting in economical savings through import substitution and cost reduction. effect can be achieved.

Although the present invention has been described with reference to the accompanying drawings, it goes without saying that various modifications and variations can be made without departing from the technical spirit of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be determined by the claims described below as well as equivalents to these claims.

110: 1st DVS 120: 2nd DVS
130: deep learning judgment unit 140: auxiliary control unit
150: Alarm generating unit 160: Defect rate receiving unit
170: display unit 180: input unit
210: Supply nozzle 211: Nozzle arm
212: nozzle opening/closing unit 213: nozzle transfer unit
220: spin chuck 230: casing
231: control box 240: main fisherman
250: Defect termination unit 300: Wafer yield management system

Claims (3)

A spin chuck that fixes and rotates the wafer;
a supply nozzle installed on the upper side of the spin chuck and supplying photoresist to the upper surface of the wafer on the spin chuck;
a main control unit that controls the operation of the spin chuck and the supply nozzle;
a defect stopping unit that stops the operation of the spin chuck and the supply nozzle when an operation control signal due to a defective photoresist is received from the outside;
A photoresist coating quality inspection system that inspects the quality of the photoresist supplied from the supply nozzle or the photoresist supplied to the upper surface of the wafer, and provides the operation control signal to the defect stopper when the quality of the photoresist is poor. ;
A defect rate receiver configured to receive the defect rate from a wafer yield management system that calculates the defect rate for each die of the wafer through data from a probe test inspected for each die of the wafer in a subsequent process after the photoresist is applied to the wafer;
Including,
The photoresist coating quality inspection system is,
A first DVS (Dynamic Vision Sensor) installed in the spin coater to acquire an image of the photoresist supplied from the spin coater's supply nozzle;
A second DVS (Dynamic Vision Sensor) installed in the spin coater to obtain an image of the photoresist supplied from the supply nozzle to the upper surface of the wafer on the spin chuck and spreading by centrifugal force;
a deep learning determination unit that determines whether the photoresist is defective using a deep learning algorithm from the images acquired from the first and second DVS; and
An auxiliary control unit configured to output the operation control signal when a defect occurs in the photoresist according to the determination result of the deep learning determination unit;
Including,
The deep learning judgment unit,
The weight learned by the deep learning algorithm is modified to be proportional to the defect rate and used when determining defects in the photoresist,
The first DVS is,
It consists of a pair and is arranged at an angle of 90 degrees around the supply nozzle,
The second DVS is characterized in that the supply nozzle is installed on the fixed nozzle arm so that it faces downward.
A spin coater capable of inspecting the coating quality of semiconductor photoresist spraying equipment. In claim 1,
A spin coater capable of inspecting the coating quality of a semiconductor photoresist spraying equipment, wherein the auxiliary control unit further includes an alarm generation unit that generates an alarm according to an alarm control signal output when a defect in the photoresist occurs according to the determination result of the deep learning determination unit. In claim 1,
The deep learning judgment unit,
With respect to an imaginary vertical line extending vertically downward from the bottom of the supply nozzle, defects in the photoresist are determined according to the spacing amount of the photoresist supplied downward from the bottom of the supply nozzle, or the photoresist is applied. A spin coater capable of inspecting the coating quality of semiconductor photoresist spraying equipment, which determines defects in the photoresist according to the ratio of areas on the wafer that have a color or shape that is different from the center.

Images