KR102373933B1 - Diagnostic system for diagnosing semiconductor processing equipment and control method thereof - Google Patents

Abstract

The present invention relates to a diagnostic system for diagnosing a semiconductor processing apparatus capable of diagnosing the state of the semiconductor processing apparatus for processing a wafer, and a control method thereof. The diagnosis system is connected to a semiconductor processing apparatus that processes a wafer using plasma, and a communication unit that receives the sensor information from one or more sensors generating sensor information and diagnoses the semiconductor processing process using the sensor information and a control unit for generating diagnostic information.

Description

A diagnostic system for diagnosing a semiconductor processing apparatus and a control method thereof

The present invention relates to a diagnostic system for diagnosing a semiconductor processing apparatus capable of diagnosing the state of the semiconductor processing apparatus for processing a wafer, and a control method thereof.

In general, a semiconductor process consists of an etching process, a thin film deposition process, a cleaning process, a photo process, and the like, and a series of unit processes are sequentially or combined to manufacture a semiconductor device.

In general, the semiconductor process is performed in a vacuum chamber, and after the semiconductor substrate is seated in the chamber, plasma is generated on the seated substrate to form a thin film on the substrate or to perform etching.

During the process of the substrate, if an abnormality occurs in the components constituting the semiconductor processing device, for example, a mass flow controller (MFC), a radio frequency source (RF), and a bias power, the plasma The state is changed, which causes a problem in that deposition or etching properties are changed.

To solve this, a separate in-situ diagnostic tool is installed to monitor the plasma, but the cause of the failure is not provided. As another method, there is a method of diagnosing the cause of a failure by modeling the relationship between the electrical signal of the component and the in-situ relationship with a neural network. In this case, the installation of an expensive in-situ system is required, and there is a difficulty in generating a specific failure pattern in advance. Therefore, in case of a failure that is not known in advance, the cause of the failure cannot be provided. On the other hand, when the part sensor information collected from a plurality of plasma equipment is collected and processed at one processing place, it takes a considerable amount of time to collect and deliver a vast amount of sensor information, and failure diagnosis is difficult due to the large amount of sensor information. As the algorithm becomes complicated, the problem of lowering the accuracy of plasma fault diagnosis is caused.

SUMMARY OF THE INVENTION The present invention aims to solve the above and other problems.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a diagnostic system capable of immediately coping with a failure by increasing diagnostic accuracy of a semiconductor processing apparatus, and a control method thereof.

Another object of the present invention is to provide a diagnostic system capable of diagnosing the state of a semiconductor processing apparatus to increase production efficiency by detecting and immediately diagnosing the cause of failure of the semiconductor processing apparatus in real time and a control method thereof. .

One embodiment of the present invention for realizing the above problems relates to a diagnostic system for diagnosing a semiconductor processing apparatus and a control method thereof.

The diagnosis system may include: a communication unit connected to a semiconductor processing apparatus for processing a wafer using plasma and receiving the sensor information from one or more sensors generating sensor information; and a control unit generating diagnostic information for diagnosing a semiconductor processing process by using the sensor information.

According to an embodiment, the controller generates time-series data or a shaped image including a sensing time and a sensing value using the sensor information, and inputs the time-series data or the image to an artificial intelligence model to receive the diagnosis information. can create

According to an embodiment, the controller determines the semiconductor processing process as any one of a plurality of sections by using the sensor information, and displays the time series data or the image in different artificial intelligence models according to the determined section. can be entered.

According to an embodiment, the artificial intelligence model may be a convolutional neural network model that performs machine learning learning based on the sensor information.

According to an embodiment, the communication unit receives the sensor information from the one or more sensors as an analog signal, and converts the analog signal into a digital signal at a sampling rate of one megahertz (1 MHz) or more. and the control unit may generate the diagnostic information using the digital signal.

According to an embodiment, the communication unit receives the sensor information from the one or more sensors as an analog signal, and converts the analog signal into a first digital signal and a second digital signal having different sampling rates, , the first digital signal may be transmitted to the control unit, and the second digital signal may be transmitted to another device.

According to an embodiment, the sampling rate of the first digital signal may be greater than or equal to one megahertz (1MHz), and the sampling rate of the second digital signal may be less than or equal to one kilohertz (1KHz).

According to an embodiment, when there are a plurality of the one or more sensors, the control unit may simultaneously receive a plurality of sensor information through the communication unit and generate a plurality of diagnostic information corresponding to each of the plurality of sensor information. can

According to an embodiment, the control unit may generate a clock signal for time synchronization of a plurality of sensors, and transmit the clock signal to the plurality of sensors through the communication unit.

According to an embodiment, when a failure that does not meet a reference condition is detected in at least one of the plurality of sensor information, the controller determines a time region in which the failure is detected, and the time region for each of the plurality of sensor information Extracts data corresponding to , and the extracted data can be transmitted to another device.

In addition, the present invention provides a control method performed by a computing device including a communication unit and a control unit. The control method may include: receiving the sensor information from one or more sensors connected to a semiconductor processing apparatus for processing a wafer using plasma and generating sensor information; and generating diagnostic information for diagnosing a semiconductor processing process by using the sensor information.

According to an embodiment, the generating of the diagnostic information may include: generating time series data or a shaped image including a sensing time and a sensing value using the sensor information; and generating the diagnostic information by inputting the time series data or the image into an artificial intelligence model.

According to an embodiment, the generating of the diagnostic information may include: determining the semiconductor processing process as any one of a plurality of sections using the sensor information; and inputting the time series data or the image to different AI models according to the determined section.

According to an embodiment, the artificial intelligence model may be a convolutional neural network model that performs machine learning learning based on the sensor information.

According to an embodiment, the method further comprises receiving the sensor information as an analog signal, and converting the analog signal into a digital signal at a sampling rate of one megahertz (1 MHz) or more, wherein the diagnostic information is the It may be generated by a digital signal.

According to an embodiment, the sensor information is received as an analog signal, converting the analog signal into a first digital signal and a second digital signal having different sampling rates; and transmitting the first digital signal to the controller and transmitting the second digital signal to another device.

According to an embodiment, the sampling rate of the first digital signal may be greater than or equal to one megahertz (1MHz), and the sampling rate of the second digital signal may be less than or equal to one kilohertz (1KHz).

According to an embodiment, when there are a plurality of the one or more sensors, a plurality of pieces of sensor information may be simultaneously received through the communication unit, and a plurality of pieces of diagnostic information corresponding to each of the plurality of sensor information may be generated.

According to an embodiment, the control method includes generating a clock signal for time synchronization of a plurality of sensors; and transmitting the clock signal to a plurality of sensors through the communication unit.

According to an embodiment, the control method may include: when a failure that does not meet a reference condition is detected in at least one of a plurality of sensor information, determining a time region in which the failure is detected; extracting data corresponding to the time domain for each of a plurality of sensor information; and transmitting the extracted data to another device.

Effects of the diagnostic system and the control method thereof according to the present invention will be described as follows.

The present invention has the effect of improving equipment productivity, process yield, and process quality by diagnosing the failure of each component constituting the semiconductor device and finding the cause of the failure in real time using a computerized intelligence system.

In addition, the present invention has the effect of increasing the reliability of the fault diagnosis of the semiconductor device because the fault diagnosis can be additionally performed after the process is completed.

1 is a block diagram illustrating a semiconductor processing apparatus according to an embodiment of the present invention;
2 is a block diagram illustrating a functional configuration of a semiconductor processing system according to an embodiment of the present invention;
3 is a block diagram illustrating a diagnosis system according to an embodiment of the present invention;
4 is a flowchart illustrating a control method performed in the diagnostic system of FIG. 3 ;
5A to 5F are block diagrams illustrating various embodiments of a diagnosis system;

Hereinafter, the embodiments disclosed in the present specification will be described in detail with reference to the accompanying drawings, but the same or similar components are assigned the same reference numbers regardless of reference numerals, and redundant description thereof will be omitted. The suffixes "module" and "part" for components used in the following description are given or mixed in consideration of only the ease of writing the specification, and do not have distinct meanings or roles by themselves. In addition, in describing the embodiments disclosed in the present specification, if it is determined that detailed descriptions of related known technologies may obscure the gist of the embodiments disclosed in this specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical idea disclosed herein is not limited by the accompanying drawings, and all changes included in the spirit and scope of the present invention , should be understood to include equivalents or substitutes.

Terms including an ordinal number such as 1st, 2nd, etc. may be used to describe various elements, but the elements are not limited by the terms. The above terms are used only for the purpose of distinguishing one component from another.

When an element is referred to as being “connected” or “connected” to another element, it is understood that it may be directly connected or connected to the other element, but other elements may exist in between. it should be On the other hand, when it is said that a certain element is "directly connected" or "directly connected" to another element, it should be understood that the other element does not exist in the middle.

The singular expression includes the plural expression unless the context clearly dictates otherwise.

In the present application, terms such as “comprises” or “have” are intended to designate that a feature, number, step, operation, component, part, or combination thereof described in the specification exists, but one or more other features It should be understood that this does not preclude the existence or addition of numbers, steps, operations, components, parts, or combinations thereof.

1 : is a figure which shows the semiconductor processing apparatus 300 which concerns on embodiment of this invention.

In the drawing, the semiconductor processing apparatus 300 includes a processing chamber, a gas supply means for supplying a processing gas into the processing chamber, and a gas exhaust means for controlling the pressure in the processing chamber by exhausting the processing gas. In addition, a sample table for supporting a sample to be processed is provided in the processing chamber, and plasma generating means for generating plasma is provided in the processing chamber.

The plasma generating means includes electromagnetic wave supply means for transmitting and supplying electromagnetic waves into the processing chamber, and a solenoid coil for generating a magnetic field in the processing chamber. In addition, a high-frequency voltage is applied from a high-frequency power supply to the sample stage to direct a reactant generated by the generated plasma toward the sample side.

A sensor unit 303 may be installed in the semiconductor processing apparatus 300 . The sensor unit 303 includes, for example, a monitor detecting a gas flow rate supplied from a gas supply means, a detector detecting a current and voltage of a power supply path supplying power for plasma generation, and detecting a phase difference between the current and voltage It consists of a detector, a detector that detects a traveling wave and a reflected wave of high-frequency power supplied for plasma generation, and an impedance monitor.

The sensor unit 303 is provided with an analysis device for detecting and analyzing light emission from plasma generated by the plasma generating means in the processing chamber. The sensor unit 303 is preferably a detector that outputs a plurality of signals, such as a spectrometer that outputs a wavelength-resolved emission spectrum, but may be a detector that extracts light of a single wavelength, such as a monochromator. The emission spectrum of the spectrometer output is a signal indicating the light intensity for each wavelength. Further, the sensor unit 303 may be a general plasma state monitor such as a quadrupole mass spectrometer that outputs a mass spectrum of a substance in plasma.

In addition, in this embodiment, the control unit 305 is provided for receiving the output from the sensor unit 303 and controlling the operation of the device. The control unit 305 controls input and output power or input power to a plasma generating means including, for example, a magnetron for generating an electromagnetic wave or a magnetic field for generating plasma. Another means may be used to control the output of the generated plasma. For example, the sensor unit 303 may increase/decrease a specific reaction amount related to processing, reaction rate, or plasma intensity based on detection data that detects light of a predetermined wavelength generated while processing a sample using plasma. It is possible to control the operation of the device by detecting the reaction state such as plasma generation/stop, and starting/stop of the device.

The control unit 305 may receive an output from the information processing apparatus 200 installed separately or integrated with the semiconductor processing apparatus 300 to control the operation of the processing apparatus. The information processing device 200 is, for example, a CD-SEM (scanning electron microscope) for measuring the processed shape after etching, but may also be an optical processing shape measuring means called scatterometry using scattered light. .

2 is a block diagram illustrating a functional configuration of a semiconductor processing system according to an embodiment of the present invention.

The semiconductor processing apparatus 300 is an apparatus for processing a semiconductor using plasma. For example, the semiconductor processing apparatus 300 may be any of an etching apparatus, a PVD apparatus, a CVD apparatus, a plasma ashing apparatus, or a plasma cleaning apparatus. The plasma used by the semiconductor processing apparatus 300 may be, for example, high-frequency plasma, ECR plasma, capacitively coupled plasma, inductively coupled plasma, helicon wave plasma, or UHF or VHF plasma, and other mechanisms. Plasma generated by

The semiconductor processing apparatus 300 includes a processing unit 301 , a sensor unit 303 , and a control unit 305 .

The processing unit 301 includes a chamber in which processing of a semiconductor wafer or the like is performed. For example, the processing unit 301 generates plasma by ionizing the gas introduced into the chamber, and performs etching, film formation, ashing, or cleaning on the workpiece. The process performed by the processing unit 301 is not particularly limited as long as it is a process using plasma.

The sensor unit 303 measures information about the inside of the processing unit 301 . Specifically, the sensor unit 303 measures the plasma state or the like inside the processing unit 301 with sensors provided in the processing unit 301 , and passes the measured information to the information processing device 200 . give.

Sensors provided in the processing unit 301 are, for example, an optical emission spectrometer (OES), a mass spectrometer (Quadrupole Mass Spectrometry: QMS), an absorption spectrometer (Infrared Laser Absorption Spectroscopy: IRLAS), or an energy spectrum. An analyzer, etc. may be sufficient. With these sensors, the sensor unit 303 can always monitor the state of plasma inside the processing unit 301 . The measurement by the sensor unit 303 may be performed, for example, at a sampling rate of 0.1 second.

The control unit 305 controls process conditions in the processing unit 301 . Specifically, when receiving the correction of the process condition from the correction determining unit 211 of the information processing device 200 , the control unit 305 , based on the received correction, performs the process condition in the processing unit 301 . to control In addition, when receiving a stop instruction from the processing stop unit 213 of the information processing device 200 , the control unit 305 stops the processing in the processing unit 301 . That is, the control unit 305 may control the overall processing performed by the processing unit 301 , and may reflect the result of the damage prediction method performed by the information processing apparatus 200 to the processing unit 301 .

The information processing apparatus 200 includes at least one of an incident flux calculating unit 203 , a machining surface flux calculating unit 205 , a shape calculating unit 207 , a damage calculating unit 209 , a correction determining unit 211 , and a processing stop unit 213 . to provide

The incident flux calculating unit 203 calculates incident fluxes of ions and light generated by the plasma. Specifically, the incident flux calculating unit 203 calculates the gas density and ion energy inside the processing unit 301 based on the information about the plasma state measured by the sensor unit 303, and uses the plasma Calculate the incident flux of generated ions and light. Incidentally, when the calculation time is sufficiently shorter than the actual processing time, the incident flux calculating unit 203 may calculate the incident flux of ions and light generated by the plasma by plasma simulation.

The processing surface flux calculating unit 205 calculates the fluxes of ions and light by the aperture ratio of the wafer, the aperture ratio at the chip level, and the pattern-shaped apertures of the workpiece, from the input information such as the shape, layer structure, and film thickness of the workpiece. Calculate the impact of In addition, the processing surface flux calculating unit 205 calculates the flux of ions and light reaching the processing surface of the workpiece from the fluxes of ions and light generated in the plasma by using ray tracing.

The shape calculating unit 207 calculates a surface reaction caused by ions reaching the machining surface of the work piece, and calculates the shape evolution of the machined surface of the work piece. Specifically, the shape calculating unit 207 calculates a normal vector obtained by adding all vectors of fluxes of ions incident on the processing surface as a progress vector of a reaction on the processing surface. In addition, the shape calculating unit 207 calculates the reaction of the machined surface due to the incident of ions, and calculates the shape evolution of the machined surface of the work piece.

The damage calculation unit 209 calculates a damage distribution that the ions and light give to the workpiece based on the fluxes of ions and light incident on the processing surface. Specifically, the damage calculating unit 209 calculates the damage that the ions and the light give to the work piece by using different models in the area where both ions and light penetrate and the area where only light penetrates. In addition, the damage calculating unit 209 can calculate the damage caused by the ions and the light at the same time by using the direction of the advance vector of the surface reaction as the direction of the damage to the workpiece by the ions and the light.

The correction determining unit 211 determines correction for the process conditions when the shape and damage distribution of the workpiece calculated by the shape calculating unit 207 and the damage calculating unit 209 exceed desired prescribed values. Specifically, the correction determination unit 211 determines a process condition that satisfies the desired specified value when the shape of the workpiece and the damage distribution predicted based on the internal state of the processing unit 301 exceed desired specified values. to determine the correction for the current state of the process condition.

For example, in an etching process, when the dimensional variation value of a recess formed by etching is ±10% or more, or damage to a workpiece exceeds a desired specified value (for example, the number of defects 1011 pieces/cm2) by 50% or more In this case, the correction determination unit 211 changes the process conditions by ±50% in the order of the flow rate of the source gas, the gas pressure, the applied power, and the wafer temperature, and again calculates the shape and damage distribution of the workpiece as the incident flux. The calculation unit 203 , the machining surface flux calculation unit 205 , the shape calculation unit 207 , and the damage calculation unit 209 make calculations. By repeating this, the correction determination unit 211 finds a process condition in which the shape of the workpiece predicted by calculation and the damage distribution satisfy desired prescribed values, and judges correction to the current state process conditions.

On the other hand, when the correction determination unit 211 cannot find a process condition that satisfies the desired prescribed value, the correction determination unit 211 transmits a boundary state signal to the processing stop unit 213, and the processing unit 301 ) may be stopped.

In addition, when the calculation time is about the same as the actual processing time, the information processing device 200 prepares in advance a database in which perturbation computation is performed for various process conditions, for example. good night. In such a case, the correction determination unit 211 may search the database to find process conditions in which the shape of the workpiece and the damage distribution satisfy desired prescribed values.

The correction determined by the correction determination unit 211 is transmitted to the control unit 305 of the semiconductor processing apparatus 300 to be reflected in the process conditions in the processing unit 301 .

The processing stopper 213 stops processing by the semiconductor processing apparatus 300 when it is determined that it is difficult to form a desired workpiece. Specifically, when the correction determination unit 211 determines that it cannot find a process condition for forming a workpiece having a shape and damage distribution that satisfies a desired prescribed value, the processing stop unit 213 is configured to perform a semiconductor processing apparatus ( 300) to stop the processing. The machining stopper 213 may be, for example, an FDC/EES (Fault Detection and Classification/Equipment Engineering System) or the like. According to the processing stop part 213, when it is difficult to form a desired to-be-processed object, by stopping the semiconductor processing apparatus 300 early, it is possible to deal with an error early.

According to the semiconductor processing system according to the present embodiment, it is possible to efficiently form a semiconductor element having desired characteristics by predicting the damage distribution of the workpiece in the plasma process and feeding back the correction based on the prediction result to the process conditions. possible.

In addition, in the above, although this embodiment was demonstrated as a system which consists of the semiconductor processing apparatus 300 and the information processing apparatus 200, the technique concerning this indication is not limited to this illustration. For example, this embodiment may be a semiconductor processing apparatus in which the semiconductor processing apparatus 300 and the information processing apparatus 200 are integrated.

3 is a block diagram illustrating a failure diagnosis system 300 of a semiconductor processing apparatus according to an embodiment of the present invention.

The diagnostic system 300 of the semiconductor processing apparatus 100 according to the present invention is a semiconductor processing apparatus 100 using the communication unit 310 that collects sensor information for each component of the semiconductor processing apparatus 100 in real time and the sensor information. ) includes a control unit 330 for diagnosing the state.

The control unit 330 forms a monitoring model for each component from the collected sensor information, collects predicted sensor information, and outputs a failure belief value of the collected predicted sensor information, a component signal monitoring unit 332 and the component and a failure determination unit 334 which compares the failure belief value obtained from the signal monitoring unit 332 with a critical point to determine whether or not there is a failure.

The communication unit 310 collects sensor information of components driving the semiconductor device, for example, the plasma processing device 100 . The plasma processing apparatus 100 may generate plasma by applying upper and lower RF power and pressure to the reaction gas injected into the chamber, and the communication unit 310 may generate plasma and transmit information on the parts sensor while the substrate is processed in real time. can be collected as

As described above, when sensor information of each component constituting the semiconductor processing apparatus 100 is collected by the communication unit 310, the component signal monitoring unit 332 may form a monitoring model to monitor the sensor information collected in real time. . As such a monitoring model, a neural network or a fuzzy logic model, which is a computational intelligence model, may be used.

Of course, in addition to the computational intelligence model as a monitoring model, a statistical model or an Auto-Regressive model or Auto-Moving Average model, which is a statistical time series method, can be applied. It can be constructed as standard deviation.

The neural network used as a computational intelligence model divides the sensor information of parts into training data and test data, uses the training data to learn a part state pattern, and uses the learned model to determine the failure state of parts. It is a predictive system.

A neural network consists of an input layer, a hidden layer, and an output layer. Predict the output or find the most similar pattern group. The number of neurons in the input layer coincides with the number of all elements constituting the feature vector, and the number of neurons in the output layer consists of one. Of course, the number of neurons in the output layer is not limited thereto, and may consist of two or more.

When using the neural network as described above, a generalized delta rule may be used as a learning algorithm used for learning, and an optimized learning algorithm may be used by modifying the generalized delta rule.

Sensor information of the j-th part, that is, information at a future time (t+k) of Sj can be predicted using the present time (t) and past information (t-m) of the sensor information of the same part. The present and past information constitute the input part of the learning pattern, and the future time information constitute the output part of the learning pattern.

As described above, a neural network structure that outputs predictive sensor information of a magnetic component signal using the magnetic component sensor information as input data is called an auto-correlated model.

Accordingly, the component signal monitoring unit 332 may collect prediction information based on the sensor information collected in real time from the communication unit 310 .

In the above, the state of the part is monitored using the autocorrelation model, but the state of the part may be monitored using a neural network having a different structure. A cross-correlated model, which is a neural network structure that uses a signal other than the sensor information of magnetic parts as input data in the input layer of the neural network, calculates this, and outputs predictive sensor information of magnetic parts, can be used.

In addition, in addition to the auto-correlation model and the cross-correlation model, the magnetic component sensor information and the sensor information other than the magnetic component are input data in the input layer of the neural network, and the self-intersection is a neural network structure that calculates and outputs the predicted sensor information of the magnetic component. An Auto-Cross Correlated Model can also be used.

When the prediction signal is obtained, the component signal monitoring unit 332 may output a failure belief value from the obtained prediction signal. Here, the failure belief value may be obtained by accumulating component sensor information by applying the predicted sensor information to a control chart, and outputting the accumulated component sensor information as a belief function.

As described above, the failure determination unit 334 may perform failure determination according to whether the output value calculated by the failure belief function is greater than an arbitrarily set reference value. For example, if the output value calculated by the failure belief function is between 0 and 1, an arbitrary reference value may be set to 0.5, and if the output value is 0.5 or more, it may be considered that a failure has occurred. Of course, any reference value may be changed to a value other than 0.5.

As described above, the diagnostic system of the semiconductor processing apparatus according to the present invention can diagnose the failure of each component in real time, and has the effect of increasing the process yield by identifying the cause of the failure.

On the other hand, when the process is completed and the collection of all sensor information for each part is completed, it is applied as an input to the monitoring model of the parts signal part to output the predicted value, and the error between the output value and the actual value is calculated as RMSE (Root Mean Square Error) so that the fault can be determined. The method of diagnosing using the RMSE can be applied to monitoring the plasma by applying to the entire sensor pattern collected during the process.

FIG. 4 is a flowchart illustrating a control method performed in the diagnosis system of FIG. 3 .

The control unit 330 is connected to a semiconductor processing apparatus for processing a wafer using plasma through the communication unit 310 and receives the sensor information from one or more sensors that generate sensor information (S410).

The sensor refers to the sensor unit 303 described above in FIG. 1 , and may include a pressure sensor, a temperature sensor, a plasma ion sensor, a gas sensor, a light emission analyzer, an absorption spectrometer, a mass analyzer, and the like.

The sensor information may be expressed as time series data including a sensing time and a sensing value. For example, sensor information is transmitted in real time as an analog signal, and the communication unit and/or control unit may convert the analog signal into a digital signal to generate time series data including a sensing time and a sensing value.

delete

The controller 330 generates diagnostic information for diagnosing a semiconductor processing process by using the sensor information (S430).

The controller 330 may generate time series data or a shaped image using the sensor information received through the communication unit 310 , and input the generated image to the artificial intelligence model to generate diagnostic information. For example, the image may be a graph in which time is an x-axis variable and a sensing value is a y-axis variable.

The artificial intelligence model may be a convolutional neural network model that performs machine learning learning based on sensor information.

The diagnostic system may have different types of artificial intelligence models. In this case, the controller 330 may determine the semiconductor processing process as any one of a plurality of sections by using the sensor information, and input images into different AI models according to the determined sections.

For example, the artificial intelligence model for diagnosing the pressure sensor may include a first model corresponding to the first process and a second model corresponding to the second process. The controller may determine whether the ongoing process is the first process or the second process based on pressure sensor information provided by the pressure sensor. When the first process is in progress, diagnostic information may be generated using the first model, and when the second process is in progress, diagnostic information may be generated using the second model.

Even with the same sensor information, different diagnostic information may be generated depending on which AI model is used. Since different AI models are used for each process, more accurate diagnostic information can be generated and the time required for diagnosis can be minimized.

Meanwhile, the communication unit 310 may receive sensor information from one or more sensors as an analog signal.

The communication unit 310 converts the analog signal into a digital signal at a sampling rate of 1 megahertz (1 MHz) or higher, and the control unit 330 may generate the diagnostic information using the digital signal. Yield improvement is a task to be overcome due to ultra-miniaturization of the semiconductor process, because it is necessary to collect multi-sensor big data of the semiconductor processing apparatus 100 at high speed.

The communication unit 310 may convert the analog signal into a first digital signal and a second digital signal having different sampling rates. In this case, the first digital signal may be transmitted to the controller 330 , and the second digital signal may be transmitted to another device. The other device may be, for example, the information processing device 200 or the server described above in FIG. 2 .

The sampling rate of the first digital signal may be greater than or equal to one megahertz (1 MHz), and the sampling rate of the second digital signal may be less than or equal to one kilohertz (1KHz). A plurality of digital signals having different sampling rates according to system characteristics are simultaneously generated.

When there are a plurality of sensors provided in the semiconductor processing apparatus 100, the control unit 330 simultaneously receives a plurality of sensor information through the communication unit 310, and generates a plurality of diagnostic information corresponding to each of the plurality of sensor information. can

The controller 330 may generate a clock signal for time synchronization of several sensors, and transmit the clock signal to the plurality of sensors through the communication unit. The time of the sensed values generated by different sensors may be synchronized according to the clock signal.

When a failure that does not meet the reference condition is detected in at least one of the plurality of sensor information, the controller 330 determines a time region in which the failure is detected, and provides data corresponding to the time region for each of the plurality of sensor information. It can be extracted and the extracted data can be transmitted to another device. Because big data is classified according to process sections and analyzed using an artificial intelligence model, diagnosis accuracy can be improved.

5A to 5F are block diagrams for explaining various embodiments of a diagnosis system.

Referring to FIG. 5A , an edge controller may be connected to each of a plurality of sensors provided in the semiconductor processing apparatus 100 to perform the role of the above-described communication unit 310 . The edge controller converts the sensor information into a digital signal after real-time pre-processing and transmits it to the controller 330 . The controller 330 generates diagnostic information and transmits the generated diagnostic information to the information processing apparatus 200 .

The controller 330 may transmit the diagnostic information to the information processing apparatus 200 and the server simultaneously as shown in FIG. 5B or to the server as shown in FIG. 5C .

Referring to FIG. 5D , a signal splitter may be connected to each of the plurality of sensors provided in the semiconductor processing apparatus 100 to perform the role of the above-described communication unit 310 . 5E and 5F , the sensor and the diagnostic system may be directly connected or may be indirectly connected by a distributor.

The present invention described above can be implemented as computer-readable code (or application or software) on a medium in which a program is recorded. The above-described control method of the diagnostic system may be realized by a code stored in a memory or the like.

The computer-readable medium includes all types of recording devices in which data readable by a computer system is stored. Examples of computer-readable media include Hard Disk Drive (HDD), Solid State Disk (SSD), Silicon Disk Drive (SDD), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. There is also a carrier wave (eg, transmission over the Internet) that is implemented in the form of. In addition, the computer may include a processor or a processor. Accordingly, the above detailed description should not be construed as restrictive in all respects but as exemplary. The scope of the present invention should be determined by a reasonable interpretation of the appended claims, and all modifications within the equivalent scope of the present invention are included in the scope of the present invention.

Claims (9)

a communication unit configured to receive a plurality of sensor information from a plurality of sensors connected to a semiconductor processing apparatus for processing a wafer using plasma; and
A control unit for generating diagnostic information for diagnosing a semiconductor processing process by using the sensor information,
The control unit is
generating time-series data or a shaped image composed of a sensing time and a sensed value using the sensor information, and inputting the time-series data or the image to an artificial intelligence model to generate the diagnostic information;
receiving a plurality of sensor information simultaneously through the communication unit, and generating a plurality of diagnostic information corresponding to each of the plurality of sensor information;
When a failure that does not meet the reference condition is detected in at least one of the plurality of sensor information, determining a time region in which the failure is detected, and extracting data corresponding to the time region for each of the plurality of sensor information diagnostic system. According to claim 1,
The controller classifies the sensor information according to a plurality of sections of the semiconductor processing process, and uses the sensor information for each section as data for learning the AI model for each section. 3. The method of claim 2,
The AI model for each section is different from each other according to the section of the semiconductor processing process. According to claim 1,
The control unit extracts the data and then transmits the extracted data to another device. 5. The method of claim 4,
The other device predicts a damage distribution of the workpiece, and based on the prediction result, transmits a feedback for controlling a semiconductor processing process to the control unit. According to claim 1,
The communication unit,
receiving the sensor information as an analog signal from the plurality of sensors, and converting the analog signal into a digital signal at a sampling rate of one megahertz (1 MHz) or more,
The control unit generates the diagnosis information by using the digital signal. According to claim 1,
The communication unit,
receiving the sensor information from the plurality of sensors as an analog signal, converting the analog signal into a first digital signal and a second digital signal having different sampling rates, and transmitting the first digital signal to the control unit; , wherein the second digital signal is transmitted to another device. 8. The method of claim 7,
A sampling rate of the first digital signal is greater than or equal to one megahertz (1 MHz), and the sampling rate of the second digital signal is less than or equal to one kilohertz (KHz). A control method performed by a computing device including a communication unit and a control unit,
receiving a plurality of sensor information from a plurality of sensors connected to a semiconductor processing apparatus for processing a wafer using plasma through the communication unit; and
and generating, by the controller, diagnostic information for diagnosing a semiconductor processing process using the sensor information,
The generating of the diagnostic information may include generating time-series data or a shaped image including a sensed time and a sensed value using the sensor information, and inputting the time-series data or the image into an artificial intelligence model to generate the diagnostic information to do; and
When a failure that does not meet the reference condition is detected in at least one of a plurality of sensor information, determining a time region in which the failure is detected, and extracting data corresponding to the time region from each of the plurality of sensor information and
The communication unit simultaneously receives a plurality of sensor information, and the control unit generates a plurality of diagnostic information corresponding to each of the plurality of sensor information.

Images