TW202328813A - A compensation method and system using for an etching process, and deep learning system - Google Patents

Abstract

The present invention provides a compensation method using for an etching process, which includes: photographing a workpiece after an etching process to obtain an etching circuit image; measuring the workpiece according to the etching circuit image to obtain a circuit measurement information of the workpiece; calculating an etching compensation information according to the circuit measurement information; according to the etching compensation information to produce a desired exposure image; and sending the desired exposure image to an exposure device to update the exposure film of the etching process.

Description

Compensation method, compensation system and deep learning system for etching process

The present invention relates to a compensation method, a compensation system and a deep learning system, in particular to a compensation method, a compensation system and a deep learning system for an etching process.

In the modern manufacturing process, with the gradual miniaturization and integration of electronic products, the frequency of circuit signals increases day by day, and the lines of circuit boards and integrated circuits become thinner and thinner. As the lines become thinner and thinner, the consistency of the cross-sectional area of the lines becomes more and more critical to the image of circuit characteristics such as resistance and impedance. Slight changes will cause the final electrical performance not to be as expected.

During the circuit etching process, the original design circuit layout density, circuit shape... and other factors will cause the flow and contact conditions of the etching solution to be different, and also cause different etching rates. Therefore, when making negatives (or LDI programs), The above considerations must be taken into account to make appropriate shape or width adjustments for regional lines with different characteristics.

Traditionally, production personnel make initial settings for line etching based on experience, and adjust parameters by actually measuring etching results. The traditional method is limited by the measurement efficiency and production pressure, resulting in insufficient measurement locations and insufficient parameter verification, resulting in lower-than-expected line production quality.

The main purpose of the present invention is to provide a compensation method for etching process, comprising: photographing a workpiece through an etching process to obtain an etching image; measuring the workpiece according to the etching image to obtain a line of the workpiece measuring information; calculating etching compensation information according to the line measuring information; generating an exposure image to be imaged according to the etching compensation information; and sending the exposure image to be imaged to an exposure device to update the exposure of the etching process Negatives.

Another object of the present invention is to provide a compensation system for an etching process, including a measurement device, a compensation device, and an image generation device. The measurement device photographs a workpiece that has undergone an etching process to obtain an etching image, and analyzes the etching image to generate a line measurement information of the workpiece. The compensation device is coupled to the measurement device, and calculates etching compensation information according to the line measurement information. The image generation device is coupled to the compensation device, generates an exposure image to be imaged according to the etching compensation information, and provides it to an exposure device to update the exposure film of the etching process.

Another object of the present invention is to provide a deep learning system for setting up the above-mentioned compensation system for the etching process, the deep learning system includes a data storage device and a processing device. The data storage device is connected to the image generation device, and the image generation device receives the etching image and the exposure image to establish a sample database. The processing device is connected to the data storage device to access the sample database, wherein the processing device includes a deep neural network, and the deep neural network is trained using the etching image and the exposure image stored in the sample database road.

Therefore, the present invention can carry out full-scale and comprehensive measurement, so as to quickly enter mass production and quickly achieve the best compensation through the functions of fully automatic fast compensation value setting, fully automatic production intermediate monitoring and real-time feedback adjustment. Value setting, monitoring in smart production, and the effect of improving and stabilizing quality.

The detailed description and technical contents of the present invention are described as follows with respect to the accompanying drawings. Furthermore, for the convenience of explanation, the proportions of the drawings in the present invention are not necessarily drawn according to the actual scale, but are exaggerated. These drawings and their proportions are not intended to limit the scope of the present invention.

The present invention can be used in the etching process of the workpiece line. When the line etching is completed, the form feedback compensation value after the line etching is confirmed by automatic optical detection to correct the exposure image, so as to correct the etching process in an instant and comprehensively during the process, ensuring The workpieces produced in the process can be standardized. On the other hand, errors caused by environmental changes in a large number of manufacturing processes can also be avoided. Workpieces in the present invention may include, but are not limited to, printed circuit boards (PCBs), flexible circuit boards (FPCs), ceramic substrates, integrated circuit wafers or integrated circuit wafers, and changes in these embodiments are not part of this invention. The scope of the invention is intended to be limited.

One of the embodiments of the present invention will be described below. Please refer to FIG. 1 , which is a schematic block diagram of the compensation system used in the etching process of the present invention, as shown in the figure.

The present embodiment provides a compensation system 100 for an etching process, which is used to measure various data of lines after etching of a workpiece, such as information including line length, width, shape, and cross-sectional area. The compensation system 100 for the etching process mainly includes a measurement device 10 , a compensation device 20 coupled to the measurement device 10 , and an image generation device 30 coupled to the compensation device 20 .

The measuring device 10 photographs the workpiece W undergoing an etching process to obtain an etching image, and analyzes the etching image to generate circuit measurement information of the workpiece W. In one embodiment, the measurement device 10 includes an image capture device 11 , and an image processing device 12 connected or coupled to the image capture device 11 .

The image capturing device 11 may include, but not limited to, a line scan camera or an area scan camera, for example. In another embodiment, the image capture device 11 of the measurement device 10 may include a dual-lens device (for example, including an up-view lens and a side-view lens), which can take pictures in the up-view direction and obtain the width of the upper frame of the line and the width of the lower frame of the line. Width, the height of the line is obtained by shooting in the side-view direction, and the cross-sectional area of the line is calculated and obtained through the information. In another embodiment, the measurement device 10 may include a three-dimensional image camera, which generates line measurements through 3D scanning techniques (such as Time of Fight (TOF), Triangulation, and stereo vision) Information, such methods of obtaining line measurement information are not within the scope of the present invention.

The image processing device 12 is connected to the image capture device 11 to obtain an etching image of the workpiece W and generate line measurement information of the workpiece W. In one embodiment, the line measurement information includes line width, line pitch, line direction, line thickness, line volume or line type. In one embodiment, the image processing device 12 of the measuring device 10 can measure the entire surface of the workpiece W according to the etched circuit image, so as to obtain circuit measurement information of all circuits on the workpiece W.

The compensation device 20 is coupled to the measurement device 10 and calculates etching compensation information according to the line measurement information. Specifically, the compensation device 20 calculates etching compensation information according to the difference between the line measurement information and a preset information. The default information is, for example, the line information of the motherboard or good product. The etching compensation information is, for example, the compensation value that should be corrected at the line position corresponding to the gap value of each mask on the exposure image, and the conversion of these compensation values (corrected by the gap value) is the compensation value) is converted according to the etching environment by substituting the correction parameters. The correction parameters can be, for example, machine parameters (such as LDI accuracy, incident angle, output power, etc.), mask registration error, material characteristics and uniformity, and environmental parameters ( Such as ambient temperature, humidity, etc.) or any other conditions that may affect the final etching result, all or part of these variable parameters are substituted to generate a compensation value.

The image generating device 30 is coupled to the compensating device 20 , generates an exposure image to be imaged according to the etching compensation information, and provides it to the exposing device 40 to update the exposure film of the etching process. Specifically, the image generation device 30 may pre-store or acquire the original exposure image by an external device. The original exposure image referred to here is the exposure negative film used when performing the etching of the corresponding workpiece, so as to serve as a standard image based on the original exposure image. The original exposure image is compensated according to the corresponding etching compensation information, and an updated exposure image is output. For example, when the etching compensation information is coordinates (x, y) and the compensation value is +2 μm, when the original mask on the coordinates (x, y) is 10 μm, modify the corresponding mask on the exposure image to 12 μm for Compensation; in practice, the compensation value can also be divided into two blocks on the left and right of the corresponding coordinates, so as to perform compensation and correction on the two sides of the mask respectively.

The exposure device 40 receives the exposure image from the image generation device 30, and updates the exposure film in the database with the received exposure image, so as to perform exposure and development according to the updated exposure film. For example, but not limited to, it may include a digital direct imaging exposure machine , Laser Direct Imaging Device (LDI), inner layer exposure machine, outer layer exposure machine, solder mask exposure machine, parallel light exposure machine or non-parallel light exposure machine.

In one embodiment, the compensation system 100 of the present invention can be integrated on the automatic optical inspection equipment, and the exposure film is corrected in real time according to the finished product while the automatic optical inspection is being performed, and the exposure film is sent to the exposure device 40 at the front end of the production line. Correct the etch process. In the above-mentioned embodiment, the compensation system 100 of the present invention may further include an image detection device 50 connected or coupled to the measurement device 10 to detect defects on the workpiece W according to the etching image, so as to obtain defect detection information of the workpiece W. The defect detection can be performed by comparing the etched image with the master image (or good product image) through traditional algorithms (such as image subtraction method) to confirm the location of the defect, or through a trained neural network (such as A deep learning model, or a machine learning model) performs defect identification after inputting the workpiece W, and the changes in these embodiments are not within the scope of the present invention. In another embodiment, the compensation system 100 of the present invention can also be implemented as an independent detection device, such as being installed at the front end or rear end of automatic optical detection, and the changes in these embodiments are not within the scope of the present invention. .

In an embodiment, the image processing device 12 , the compensating device 20 , and the image generating device 30 may be implemented by different computers, servers, PLCs or similar independent devices. In another embodiment, the image processing device 12, the compensation device 20, and the image generating device 30 can be co-constructed into a single computer, server, PLC or similar independent device, and the storage unit is executed after the processor loads the stored programs. The processor can be, for example, a central processing unit (Central Processing Unit, CPU), or other programmable general purpose or special purpose microprocessor (Microprocessor), digital signal processor (Digital Signal Processor, DSP), programmable Controller, application specific integrated circuit (Application Specific Integrated Circuits, ASIC), programmable logic device (Programmable Logic Device, PLD) or other similar devices or a combination of these devices. In one embodiment, the present invention can be implemented in combination with automatic optical inspection equipment (AOI), and the automatic optical inspection equipment compares the measured line measurement information with the preset information to generate etching compensation for the workpiece W to be tested information, and according to the compensation information, adjust the computer aided manufacturing software (Computer Aided Manufacturing, CAM) file of the original exposure image/film to generate an updated exposure image/film, and provide it to the exposure device 40 .

In another embodiment, after the automatic optical inspection equipment measures the workpiece W and obtains the line measurement information, it can provide the line measurement information to an external computer (not shown in the figure) to adjust the original exposure image/film The CAM file enables the external computer to generate updated exposure images/negatives to provide to the exposure device 40; specifically, for example, the system makes a corresponding database (such as a lookup table) after integrating and analyzing the training results and stores it in the external computer. , when the external computer receives the line measurement information, it finds the corresponding data (such as compensation value or target value) from the database according to the detected value of the line measurement information, and generates an exposure image based on the data /negatives.

Regarding the implementation process of the compensation system of the present invention, please refer to "Fig. 2" and "Fig. 3" together, which are schematic diagrams of the cross-section of the circuit and isometric diagram of the circuit in the measurement of the full board surface of the present invention, as shown in the figure. In one embodiment, the image generating device 20 includes but is not limited to generating an exposure image through the following compensation method.

After the image processing device 12 obtains the etching image from the image capture device 11 , it stores the etching image in the image database A1 to prepare for measurement.

After the etching image is scheduled, the measuring device 10 generates measurement information of the workpiece W through the etching image. The measurement information is the information of each line in the etching image, including line width, line spacing, line direction, line thickness, line volume, or line type. When measuring the entire board, the identification of the circuit can be searched by the color or shape of the circuit in the etching image to segment the boundary between the circuit and the substrate on the etching image. Such image segmentation methods are not intended by the present invention limited range. The "full board surface measurement" mentioned in the present invention is to measure the image of the entire object to be measured. Specifically, it can be used to scan the object to be measured with a line scan camera, and then measure the entire surface after splicing all the line segment images. The image of the object to be tested can be obtained by taking multiple images of the object to be tested, or the partial area of the object to be tested can be photographed by the surface scanning camera to form an image of the entire surface of the workpiece, or the image of the entire surface of the object to be measured can be captured at one time.

Specifically, the image processing device 12 obtains the line information in the following manner. The image processing device 12 can first perform image pre-processing procedures (such as image enhancement, noise removal, contrast enhancement, edge enhancement, feature extraction, image compression, Image conversion, etc.), and segment the image after the image pre-processing procedure, or extract the boundary to divide the region of interest (Region of Interest, ROI). The method of extracting the region of interest can be, for example, binarization, or through a machine learning system (Machine Learning), a deep learning system (Deep Learning) and other similar neural networks, which will be etched into the image after training by the system The circuit area and the substrate area are separated. It should be noted that the etched image mentioned here is not necessarily a single image, but can also be two or more images, and the distribution information of the lines in the three-dimensional space can be subsequently analyzed through the multiple etched images.

In one embodiment, the cross-sectional area of the line can be obtained by calculating the line width (for example, including the upper line width UW and the lower line width DW) and the line thickness SH, such as the trapezoidal area formula. Further, in an embodiment, the line volume can be obtained by calculating the line cross-sectional area and the line length SL. In another embodiment, in a more accurate calculation method, the line volume can be obtained by calculating the cross-sectional area of each sampled line and the length of each sampled line.

In one embodiment, when the image processing device 12 performs measurement, the pixel coordinates can be used as an index to record the value of each line on the etched image. With the calculation performance of the image processing device 12, at least a single pixel can be used as the reference point for recording the line size value. The smallest unit; in the case of limited computing performance of the image processing device 12, two or more pixel distances can also be used as the segment to record the line size value of the segment to which it belongs. These implementation methods are not within the scope of the present invention; In other embodiments, the route measurement information can also be recorded by using a self-defined coordinate record as an index, which is not limited in the present invention.

After the image processing device 12 completes all the measurements, the compensation device 20 compares the measured value with the line value of the standard sheet (such as the master sheet) pre-stored in the image database A1 to determine the difference between the etched image and the standard sheet. difference. Specifically, before the equipment goes on the production line, the required standard sheet can be sent to the equipment of the present invention to measure the entire board surface through the measuring device 10, so as to obtain the various circuits of the standard sheet in one step The measurement information is stored in the database as the expected value of each line coordinate according to the coordinates as the index. When the equipment enters the production line, the compensation device 20 subtracts the measured value of the captured etching image from the expected value of the corresponding position of the standard sheet to obtain a difference value indexed by the coordinates. In one embodiment, the standard chip can be a master chip or a good product chip, which is not limited in the present invention.

Furthermore, when the supplementary line measurement module A1 obtains the gap value between the coordinate measurement value of each line and the expected value, it can calculate the compensation value of the coordinate position of the exposure image according to the gap value. The compensation value specifically refers to the compensation value required to correct the line size to the expected value by exposing the line mask (such as a mask or a part without a mask) on the exposure image. The correction parameter for converting the gap value to the compensation value can be, for example, Machine parameters (such as LDI accuracy, incident angle, output power, etc.), mask registration errors, material properties and uniformity, environmental parameters (such as ambient temperature, humidity, etc.), or any other factors that may affect the final etching result Conditions, all or part of these variable parameters are substituted to generate compensation values. In another embodiment, in an ideal state where the environmental conditions are controllable, the difference between the measured value of the line coordinates and the expected value can be directly output as the compensation value of the line on the exposure film to obtain etching compensation information. No limitation is intended in the invention.

After obtaining the compensation value of the full version, the image generation device 30 corrects the line mask at the mapping coordinate position corresponding to the compensation value (for example, widening, reduce the width of the circuit mask), generate an exposure image when all circuit mask corrections are completed, and transmit the exposure image to the exposure device 40 at the front end of the production line to update the exposure film of the etching process.

Another embodiment of the present invention discloses a deep learning system, please refer to "Fig. 4" and "Fig. 5" together. It is a combined block of the deep learning system of the present invention and the compensation system for the etching process of the full board measurement A schematic diagram and a schematic diagram of the training process of the deep learning system are shown in the figure.

This embodiment discloses a deep learning system 200 configured to cooperate with the aforementioned compensation system 100 . In this embodiment, the deep learning system 200 includes a data storage device 60 and a processing device 70 connected to the data storage device 60 .

The data storage device 60 is connected to the compensation system 100, and the compensation system 100 receives etching images and exposure images to establish a sample database. Specifically, the compensation system 100 transmits the corresponding value of the etching image corresponding to the material number to the data storage device 60 for storage in the sample database each time the circuit mask is corrected. Wherein, the corresponding numerical value may include etching image, line measurement information (including coordinate position) of the full-page line size of the workpiece, full-page compensation value (including coordinate position) of the exposure film, and the updated exposure film.

The processing device 70 is connected to the data storage device 60 to access the sample database of the data storage device 60. The processing device 70 includes a deep neural network, which uses the etching images and exposure negatives stored in the sample database to train the deep neural network. network.

The deep neural network in the present invention can be, for example, but not limited to LeNet model, AlexNet model, GoogleNet model or VGG model (Visual Geometry Group), etc., which are not limited in the present invention. Specifically, the deep neural network can be trained to convert etching images into exposure negatives, use the etching images as the input of the deep neural network, and output corresponding exposure images based on the etching images; the present invention The deep neural network is used to input the etching image to the deep neural network for forward propagation to obtain the stimulus response, and calculate the difference between the stimulus response result and the exposure image after circuit correction, so as to obtain the response error and It performs backpropagation (Backpropagation) to update the weight, and iterates through the above two steps until the response of the deep neural network to the input reaches the target range.

In one embodiment, the training process of the deep learning system in the present invention is as follows, please refer to "Fig. 5" together: first, the workpiece W is sent to the exposure device 40, and the exposure device 40 generates a photomask based on the exposure image EP1 The AH performs an exposure process on the workpiece W, so as to transfer the mask pattern on the workpiece W. Next, the workpiece W on which the mask pattern is generated is sent to the developing device 80 and the etching device 90. The developing device 80 first develops the workpiece W to form an etching window, and then etches the workpiece W through the etching device 90; after the etching is completed The workpiece W is sent to the compensation system 100, and the etching image CP1 is obtained through the compensation system 100, and the entire board surface measurement is performed according to the etching image CP1 to obtain line measurement information. The compensation system 100 calculates etching compensation information based on the measurement results of the entire board, and generates an exposure image EP2 to be imaged according to the etching compensation information. For example, in one embodiment, the CAM file of the original exposure image/negative film can be adjusted and adjusted according to the etching compensation information to generate the exposure image EP2 to be imaged. Subsequently, the exposure image EP2 is sent to the exposure device 40 to update the exposure film/image of the exposure device 40 .

On the other hand, in one embodiment, the compensation system 100 can send the captured etching image CP1 and the exposure image EP1 corresponding to the etching image CP1 to the deep learning system 200 for training (special attention is required here for training The exposure image of the deep learning system 200 must be the exposure image ( EP1 ) used to generate the etching image CP1 .

Through the above training results, the trained deep neural network will acquire the ability to convert etching images into exposure images, so as to obtain an artificial intelligence model for workpiece etching measurement.

Another embodiment of the present invention discloses a circuit compensation method for the workpiece etching process. Please refer to FIG. 6 , which is a schematic flow chart of the circuit compensation method for the workpiece etching process of the present invention, as shown in the figure.

In another embodiment, the present invention discloses a line compensation method for a workpiece etching process, and the line compensation method includes the following steps.

First, photograph the workpiece undergoing the etching process to obtain an etching image (step S01); in one embodiment, the workpiece includes a printed circuit board, a flexible circuit board, a high-density interconnection technology circuit board, a touch panel or an integrated circuit chip and other similar substrates, these embodiments are not within the scope of the present invention.

Next, measure the workpiece according to the etching image to obtain line measurement information of the workpiece (step S02); in one embodiment, in an embodiment further comprising an image detection device, the image processing device performs workpiece measurement on the etching image At the same time, flaw detection can be performed on the etched image at the same time.

Next, calculate etching compensation information according to the measurement information of the line (step S03 ); in this embodiment, the etching compensation information includes the coordinate value of the line and the compensation value corresponding to the position of the coordinate value.

Next, generate an exposure image to be imaged according to the etching compensation information (step S04); the etching compensation information is used to modify the original exposure image (to generate an exposure film corresponding to the workpiece circuit), and correct the circuit at the corresponding mapped coordinate position according to the compensation value. An updated exposure image is generated.

Finally, the exposure image to be imaged is sent to the exposure device, so as to update the exposure film of the etching process (step S05 ). Exposure devices include, but are not limited to, digital direct imaging exposure machines, laser direct imaging devices (Laser Direct Imaging Device, LDI), inner layer exposure machines, outer layer exposure machines, solder mask exposure machines, parallel light exposure machines or non-parallel Light exposure machine.

To sum up, the present invention can carry out full-scale and comprehensive measurement, so as to quickly enter mass production and quickly reach the optimum level through the functions of fully automatic fast compensation value setting, fully automatic production intermediate monitoring, and real-time feedback adjustment. Optimum compensation value setting, monitoring in intelligent production, and the effect of improving and stabilizing quality.

The present invention has been described in detail above, but the above is only one of the preferred embodiments of the present invention, and should not limit the scope of the present invention with this, that is, all equivalents made according to the patent scope of the present invention Changes and modifications should still fall within the scope of the patent coverage of the present invention.

100: Compensation system 10: Measuring device 11: Image capture device 12: Image processing device 20: Compensation device 30: Image generation device 40: Exposure device 50: Image detection device W: Workpiece A1: Image database UW: line width DW: line lower width SH: line thickness SL: line length 200: Deep Learning Systems 60: data storage device 70: Processing device 80: Developing device 90: Etching device EP1: Exposure Image EP2: Update exposure image CP1: Etched Image AH: mask S01~S05: Steps

FIG. 1 is a schematic block diagram of a compensation system used in an etching process according to the present invention.

Fig. 2 is a schematic cross-sectional view of the circuit in the present invention.

Fig. 3 is an isometric schematic diagram of the circuit in the present invention.

Fig. 4 is a combined block diagram of the deep learning system of the present invention and the compensation system for etching process.

Fig. 5 is a schematic diagram of the training process of the deep learning system in the present invention.

FIG. 6 is a schematic flow chart of the method for compensating lines in the etching process of a workpiece according to the present invention.

100: Compensation system

10: Measuring device

11: Image capture device

12: Image processing device

20: Compensation device

30: Image generation device

40: Exposure device

50: Image detection device

W: Workpiece

A1: Image database

Claims (15)

A compensation method for an etching process, comprising: photographing a workpiece that has undergone an etching process to obtain an etching image; measuring the workpiece according to the etching image to obtain a line measurement information of the workpiece; calculating an etching compensation information according to the line measurement information; generating an exposure image to be imaged according to the etching compensation information; and The exposure image to be imaged is sent to an exposure device to update the exposure film of the etching process. The compensation method for an etching process as described in Claim 1 further includes: detecting the workpiece for defects according to the etching image, so as to obtain defect detection information of the workpiece. The compensation method for etching process as described in claim 1, wherein the exposure device includes a digital direct imaging exposure machine, a laser direct imaging device (Laser Direct Imaging Device, LDI), an inner layer exposure machine, an outer layer exposure machine, an anti- Welding exposure machine, parallel light exposure machine or non-parallel light exposure machine. The compensation method for etching process as claimed in claim 1, wherein the workpiece includes a printed circuit board (PCB), a flexible circuit board (FPC), a ceramic substrate, an integrated circuit wafer or an integrated circuit wafer. The compensation method for the etching process as described in claim 1, wherein the step of measuring the workpiece according to the etching image includes: measuring the entire surface of the workpiece according to the etching image, so as to obtain the line quantity of the workpiece test information. The compensation method for etching process as described in Claim 1, wherein the step of calculating the etching compensation information according to the line measurement information includes: calculating the Etch compensation information. The compensation method for etching process according to claim 1, wherein the line measurement information includes line width, line pitch, line direction, line thickness, line volume or line type. A compensation system for an etching process comprising: A measurement device, photographing a workpiece that has undergone an etching process to obtain an etching image, and analyzing the etching image to generate a line measurement information of the workpiece; A compensation device, coupled to the measurement device, calculates an etching compensation information based on the line measurement information; and An image generation device, coupled to the compensation device, generates an exposure image to be imaged according to the etching compensation information, and provides it to an exposure device to update the exposure film of the etching process. The compensation system for an etching process as described in claim 8 further includes an image detection device for detecting defects on the workpiece according to the etching image, so as to obtain defect detection information of the workpiece. The compensation system for etching process as described in claim item 8, wherein the exposure device includes a digital direct imaging exposure machine, a laser direct imaging device (Laser Direct Imaging Device, LDI), an inner layer exposure machine, an outer layer exposure machine, an anti- Welding exposure machine, parallel light exposure machine or non-parallel light exposure machine. The compensation system for etching process as claimed in claim 8, wherein the workpiece includes a printed circuit board (PCB), a flexible circuit board (FPC), a ceramic substrate, an integrated circuit wafer or an integrated circuit wafer. The compensation system for etching process as described in claim 8, wherein the measurement device measures the entire surface of the workpiece according to the etching image, so as to obtain the line measurement information of the workpiece. The compensation system for etching process as claimed in claim 8, wherein the compensation device calculates the etching compensation information according to the difference between the line measurement information and a preset information. The compensation system for etching process as claimed in claim 8, wherein the line measurement information includes line width, line pitch, line direction, line thickness, line volume or line type. A deep learning system, used to cooperate with the compensation system settings as described in claim items 8-14, comprising: a data storage device connected to the image generation device, the image generation device receives the etching image and the exposure image to establish a sample database; and a processing device connected to the data storage device to access the sample database, wherein the processing device includes a deep neural network for training the deep neural network using the etching image and the exposure image stored in the sample database network.

Images