Semiconductor process control system based on AI

Abstract

The present invention relates to an AI-based semiconductor process control system that prevents corrosion of equipment / parts and wafer defects caused by fumes and ensures the process is carried out under optimal conditions, thereby maximizing wafer quality, increasing the efficiency and convenience of monitoring and control of the semiconductor process, and enabling rapid prevention of equipment failures. This is achieved by installing additional cleaning equipment to remove residual hazardous substances from a wafer that has undergone an etching process in the semiconductor process facility, while simultaneously having an integrated control server collect raw data and sensing values ​​from PLCs and IoT sensors of the semiconductor process facility, monitoring them, analyzing the collected data using pre-trained AI models, and outputting equipment failures, quality predictions, and optimal parameter values.

Description

Technology Field

[0001] The present invention relates to an AI-based semiconductor process control system, and more specifically, to an AI-based semiconductor process control system that prevents corrosion of equipment / parts and wafer defects caused by fumes and ensures the process is carried out under optimal conditions, thereby maximizing wafer quality, increasing the efficiency and convenience of monitoring and control of the semiconductor process, and enabling rapid prevention of equipment failures. This is achieved by installing additional cleaning equipment to remove residual hazardous substances from a wafer that has undergone an etching process in the semiconductor process facility, while simultaneously having an integrated control server collect raw data and sensing values ​​from PLCs and IoT sensors of the semiconductor process facility, monitoring them, analyzing the collected data using pre-trained AI models, and outputting equipment failure, quality prediction, and optimal parameter values. Background Technology

[0002] Typically, a semiconductor is a complex and highly precise device completed by forming and processing various types of films on a semiconductor substrate called a wafer, and then connecting electronic and electrical components with wiring.

[0003] Such semiconductor manufacturing is carried out through very complex and diverse processes, but basically, it involves forming and processing conductors, insulators, and semiconductor films on a semiconductor wafer to create electronic and electrical components and print circuits.

[0004] Generally, semiconductor manufacturing processes consist of deposition, etching, cleaning, and drying processes.

[0005] In particular, the etching process is carried out with process gas injected into the chamber. At this time, a chlorine (Cl)-based gas is used as the process gas to etch the thin film formed on the wafer.

[0006] When etching processes are repeatedly performed using such chlorine-based gases as process gases, residual hazardous substances remain. These residual hazardous substances are deposited inside the chamber and act as contaminants on wafers during subsequent processes. Furthermore, they react with atmospheric gases to generate fumes, which can corrode equipment / components and cause wafer defects. Additionally, the lifespan of equipment / components is shortened, leading to increased operating costs.

[0007] Meanwhile, with the recent expansion of communication infrastructure and the advancement of sensing technology, facility management technologies based on Ubiquitous Sensor Networks (USN) are rapidly being adopted for the purpose of monitoring various on-site facilities and preventing failures.

[0008] In other words, it is necessary to integrate ICT (Information and Communication Technology) into semiconductor equipment to control the semiconductor process and improve wafer quality and yield.

[0009] In particular, there is a great need for technology that utilizes pre-trained artificial intelligence (AI) algorithms for semiconductor equipment to analyze collected data, detect equipment malfunctions, predict quality, and identify optimal parameter values, and to provide application services using the detected data; however, research on optimizing the operation of semiconductor equipment processes is still very insufficient. The problem to be solved

[0010] The present invention aims to solve these problems, and the objective of the present invention is to provide an AI-based semiconductor process control system that can improve the efficiency, accuracy, and convenience of semiconductor process control by having an integrated control server analyze collected raw data and sensing values ​​to generate process data, detect equipment abnormalities, and predict quality, as well as enable a rapid response to equipment abnormalities.

[0011] In addition, another objective of the present invention is to provide an AI-based semiconductor process control system capable of maximizing wafer yield and quality compared to the same conditions, since the semiconductor process equipment enables the process to operate under optimal conditions.

[0012] In addition, another problem to be solved by the present invention is to provide an AI-based semiconductor process control system that can further increase the reliability of the semiconductor process by enabling accurate and objective recognition of the semiconductor process status through the integrated control server generating a report every second cycle (T2) and providing it to managers.

[0013] In addition, another problem to be solved by the present invention is to provide an AI-based semiconductor process control system that effectively prevents corrosion of equipment / components caused by fumes by installing additional cleaning equipment to remove residual harmful substances from a wafer that has undergone an etching process in the semiconductor process facility, and simultaneously introducing steam into the wafer that has undergone the etching process by the cleaning equipment, thereby improving wafer quality and extending the lifespan of equipment / components. means of solving the problem

[0014] The means of the present invention for solving the above problem comprises: a PLC (Program Logic Controller) connected to semiconductor process equipment that performs a semiconductor process and generates raw data of the semiconductor process equipment; IoT sensors installed in the semiconductor process equipment and measuring preset parameters; an integrated control server that collects raw data received from the PLC and sensing values ​​received from the IoT sensors, matches them to generate process data, analyzes the generated process data to detect equipment abnormalities, and generates incident data when equipment abnormalities are detected; and a manager terminal that is a terminal possessed by a manager of the semiconductor process equipment and receives incident data from the integrated control server.

[0015] In addition, in the present invention, the integrated control server preferably includes: a DB server in which an anomaly detection algorithm, which is an AI algorithm that analyzes process data and outputs whether there is an anomaly in each component of the semiconductor process equipment (5), and a quality prediction algorithm, which is an AI algorithm that analyzes process data and outputs a wafer quality prediction; a process data generation unit that collects raw data transmitted from the PLCs and sensing values ​​transmitted from the IoT sensors and then generates process data; an equipment anomaly detection unit that uses the anomaly detection algorithm to analyze the process data generated by the process data generation unit and detect whether there is an equipment anomaly; a quality prediction unit that uses the quality prediction algorithm to analyze the process data generated by the process data generation unit and predict the quality; an incident-data generation unit that, when an equipment anomaly is detected by the equipment anomaly detection unit, generates incident-data including identification information of the semiconductor process equipment where the equipment anomaly was detected, equipment identification information, and anomaly details; and a control unit that transmits the incident-data generated by the incident-data generation unit to the administrator terminal.

[0016] In addition, in the present invention, the integrated control server further includes a learning unit that is executed at a preset period (T), and the learning unit performs the learning of the anomaly detection algorithm by generating learning data capable of learning the correlation between process data and the anomaly detection result of each component, and then utilizing the generated learning data to derive an extraction model which is a set of parameter values ​​regarding the correlation between process data and the anomaly detection result, and performs the learning of the quality prediction algorithm by generating learning data capable of learning the correlation between process data and wafer quality, and then utilizing the generated learning data to derive an extraction model which is a set of parameter values ​​regarding the correlation between process data and wafer quality.

[0017] In addition, in the present invention, the semiconductor process equipment preferably includes: a transfer chamber including a transfer unit; a load lock chamber that performs the function of loading or unloading a wafer after maintaining a predetermined pressure when a wafer in an unprocessed state is received from the outside; an orient chamber that aligns the orientation of the received wafers when the wafers are transferred and received by the transfer unit, including slot units having a plurality of slots; an etching chamber that performs an etching process when a wafer is received by the transfer unit; and a vapor chamber that removes residual harmful substances by introducing vapor to remove residual harmful substances from the wafer when a wafer is received by the transfer unit.

[0018] In addition, in the present invention, the operation process (S1) of the semiconductor process equipment comprises: a step 10 (S10) of stacking wafers in a storage space of a cassette; a step 20 (S20) in which the transfer unit transfers the cassette containing the wafers to the load lock chamber, and the load lock chamber maintains a preset pressure; a step 30 (S30) in which the transfer unit transfers the cassette of the load lock chamber to the orient chamber; a step 40 (S40) in which the orient chamber aligns the orientation of the wafers; a step 50 (S50) in which the transfer unit transfers the cassette of the orient chamber to the etching chamber; a step 60 (S60) in which the etching chamber performs an etching process of the wafers contained therein; a step 70 (S70) in which the transfer unit transfers the cassette to the vapor chamber; a step 80 (S80) in which vapor is introduced to remove residual harmful substances remaining inside the vapor chamber; and a step 90 (S90) in which the transfer unit transfers the cassette to the vapor chamber. It is preferable to include step 100 (S100) of removing the cassette from the load lock chamber.

[0019] In addition, the semiconductor process equipment in the present invention further comprises cleaning equipment including the vapor chamber, a DIW supply unit, a nitrogen supply unit, a flow meter, a steam generation unit, a mass flow controller, a mass flow meter, and a pump; the DIW supply unit, the flow meter, the steam generation unit, the mass flow meter, and the vapor chamber are connected by a first flow path (C1); the nitrogen supply unit, the mass flow controller, and the steam generation unit are connected by a second flow path (C2); a third flow path (C3) is connected to the second flow path (C2) between the nitrogen supply unit and the mass flow controller and the first flow path (C1) between the mass flow meter and the vapor chamber; the vapor chamber and the pump are connected by a fourth flow path (C4); the vapor chamber is connected to the mass flow meter and the first flow path (C1) on one side so that steam is introduced, and is connected to the pump through the fourth flow path (C4); the DIW supply unit supplies DI water to the first flow path (C1); The nitrogen supply unit supplies nitrogen to the second flow path (C2), and the flow meter is connected to the DIW supply unit and the first flow path (C1) to measure the flow rate of DIW supplied through the first flow path (C1). The steam generation unit is connected to the first flow path (C1), and the mass flow rate controller controls the flow rate of nitrogen supplied from the nitrogen supply unit through the second flow path (C2). The steam generation unit is connected to the second flow path (C2) to supply an appropriate flow rate of nitrogen to the steam generation unit through the second flow path (C2). The steam generation unit is connected to the flow meter and the first flow path (C1), and simultaneously with the inflow of DI water, it is connected to the mass flow rate controller and the second flow path (C2) to allow nitrogen (N2) to flow in and generate steam while maintaining a preset set-pressure. The steam generation unit is connected to the mass flow meter and the first flow path (C1) to the mass flow meter through the first flow path (C1). Steam of a set pressure is supplied, and is connected to the steam generating unit and the first flow path (C1).It is preferable to measure the flow rate of steam introduced through the first flow path (C1) and to include a mass flow meter connected to the steam chamber and the first flow path (C1).

[0020] In addition, in the present invention, the cleaning equipment preferably further includes a first valve (EV#1) installed in a first flow path (C1) between the flow meter and the steam generating unit, a second valve (EV#2) installed in a first flow path (C1) between the mass flow meter and the steam chamber, a third valve (EV#3) installed in a third flow path (C3), and a fourth valve (EV#4) installed in a fourth flow path (C4) between the pump and the steam chamber. Effects of the invention

[0021] According to the present invention having the above-mentioned problem and solution means, an integrated control server analyzes collected raw data and sensing values ​​to generate process data, simultaneously detect equipment abnormalities and predict quality, thereby increasing the control efficiency, accuracy, and convenience of the semiconductor process, as well as enabling a rapid response to equipment abnormalities.

[0022] In addition, according to the present invention, since the semiconductor process equipment enables a process under optimal conditions, the yield and quality of the wafer can be maximized compared to the same conditions.

[0023] In addition, according to the present invention, the integrated control server generates a report every second cycle (T2) that is pre-set and provides it to managers, thereby enabling accurate and objective recognition of the semiconductor process status, which further increases the reliability of the semiconductor process.

[0024] In addition, according to the present invention, by installing a cleaning device to remove residual harmful substances from a wafer that has undergone an etching process in addition to the semiconductor process equipment, and by introducing steam into the wafer that has undergone the etching process through the cleaning device, corrosion of equipment / parts caused by fumes can be effectively prevented, as well as wafer quality is improved and the lifespan of equipment / parts is extended. Brief explanation of the drawing

[0025] FIG. 1 is a configuration diagram showing an AI-based semiconductor process control system, which is an embodiment of the present invention. Figure 2 is a conceptual diagram representing Figure 1. Figure 3 is an example diagram showing an equipment malfunction interface provided by the control service application of Figure 1. Figure 4 is a block diagram showing the semiconductor process equipment of Figure 1. Figure 5 is a flowchart showing the operation process of the semiconductor process equipment of Figure 4. Figure 6 is a conceptual diagram showing cleaning equipment including the vapor chamber of Figure 4. Figure 7 is a conceptual diagram illustrating the optimal recipe applied to the cleaning equipment of Figure 6. Figure 8 is a block diagram showing the integrated control server of Figure 1. Figure 9 is a conceptual diagram to explain Figure 8. Specific details for implementing the invention

[0026] Hereinafter, an embodiment of the present invention will be described with reference to the attached drawings.

[0027] FIG. 1 is a configuration diagram showing an AI-based semiconductor process control system, which is an embodiment of the present invention, and FIG. 2 is a conceptual diagram showing FIG. 1.

[0028] An AI-based semiconductor process control system (1), which is an embodiment of the present invention, is designed to prevent corrosion of equipment / parts and wafer defects caused by fumes, as well as to maximize wafer quality and enable rapid prevention of equipment malfunctions by installing additional cleaning equipment to remove residual harmful substances from a wafer that has undergone an etching process in the semiconductor process facility, while simultaneously having an integrated control server collect raw data and sensing values ​​from the PLC and IoT sensors of the semiconductor process facility, monitor them, analyze the collected data using pre-trained AI models, and output equipment malfunctions, quality predictions, and optimal parameter values.

[0029] In addition, the AI-based semiconductor process control system (1) of the present invention, as illustrated in FIGS. 1 and 2, comprises semiconductor process equipment (5) for manufacturing semiconductors, PLCs (Program Logic Controllers) (7) for managing and controlling the operation of each semiconductor process equipment (5), IoT sensors (9) installed in each semiconductor process equipment (5) for measuring the value of a preset parameter, an integrated control server (3) that stores and monitors raw data received from the PLCs (7) and sensing values ​​received from the IoT sensors (9), and simultaneously analyzes collected data using pre-trained AI models to output equipment malfunction, quality prediction, and optimal parameter values, manager terminals (11) that are terminals held by the manager of each semiconductor process equipment (5) and have a control service application installed that provides semiconductor equipment control services in conjunction with the integrated control server (3), and a communication network (10) that provides a data transfer path between the integrated control server (3), the PLCs (7), the IoT sensors (9), and the manager terminals (11).

[0030] In this invention, for convenience of explanation, the manager terminal (11) is described as receiving monitoring information, abnormal information, quality prediction information, etc. from the integrated control server (3), but the manager terminal (11) may be configured to be connected to the PLCs (7) via a wired cable.

[0031] The communication network (10) supports data communication between the integrated control server (3), PLCs (7), IoT sensors (9), and administrator terminals (11), and in detail, can be composed of a wide area network (WAN), an optical communication network, a mobile radio communication network, 3G / 4G / 5G, etc.

[0032] The administrator terminal (11) is a digital device that the administrator possesses or carries, and in detail, may consist of a desktop PC, a laptop, a smartphone, a tablet PC, etc.

[0033] In addition, a control service application (12) is installed on the administrator terminal (11).

[0034] The control service application (12) is an application, software, or application installed on the administrator terminal (11), and provides a control service for semiconductor process equipment assigned to the administrator in conjunction with the integrated control server (3).

[0035] In addition, the control service application (12) requests monitoring information from the integrated control server (3) in response to a request from the administrator, and at the same time displays the monitoring information received from the integrated control server (3) on the monitor of the administrator terminal (11) through the GUI (Graphic User Interface).

[0036] Figure 3 is an example diagram showing an equipment malfunction interface provided by the control service application of Figure 1.

[0037] In addition, the control service application (12) requests equipment fault information from the integrated control server (3) in response to a request from the administrator, and at the same time displays the equipment fault interface (900) of FIG. 3, which is a GUI that displays the equipment fault information received from the integrated control server (3), on the monitor of the administrator terminal (11).

[0038] In addition, the control service application (12) requests quality prediction information from the integrated control server (3) at the request of the manager, and at the same time displays the quality prediction information received from the integrated control server (3) on the monitor of the manager terminal (11) through the GUI (Graphic User Interface).

[0039] In addition, the control service application (12) provides GUIs with various functions, such as receiving command requests from an administrator or displaying response data in response to the request.

[0040] Additionally, when the control service application (12) receives control data from an administrator to control the operation of a specific piece of equipment / part, it transmits the input control data to an integrated control server (3), and the integrated control server (3) transmits the control data to the PLC (7) of the semiconductor process equipment (5) so that the operation of the semiconductor process equipment (5) is controlled according to the administrator's request.

[0041] Figure 4 is a block diagram showing the semiconductor process equipment of Figure 1.

[0042] As shown in FIG. 4, the semiconductor process equipment (5) consists of a transfer chamber (50) including a transfer unit (501), a plurality of etching chambers (55), a wafer-oriented chamber (53), a load lock chamber (51), and a vapor chamber (57).

[0043] At this time, the etching chambers (55), orient chamber (53), load lock chamber (51), and vapor chamber (57) are sequentially connected to the outside based on the central transfer chamber (50), but are installed so that adjacent chambers are interlocked or sealed by opening and closing the door, and are maintained in a vacuum atmosphere of constant pressure.

[0044] When the load lock chamber (51) receives an unprocessed wafer from the outside, it maintains a predetermined pressure and then performs the function of loading or unloading the wafer.

[0045] At this time, when the predetermined pressure of the load lock chamber (51) is maintained, the transfer chamber (50) transfers the wafer of the load lock chamber (51) to the orient chamber (53) using the transfer unit (501).

[0046] The orient chamber (53) includes slot units equipped with multiple slots, and simultaneously accommodates multiple wafers and aligns and aligns the wafers.

[0047] At this time, when the alignment of the orient chamber (53) is completed, the transfer chamber (50) transfers the wafer to the etching chamber (55) using the transfer unit (501).

[0048] When the etching chamber (55) receives a wafer, it performs etching processing. At this time, when the etching processing of the etching chamber (55) is completed, the transfer chamber (50) transfers the wafer to the vapor chamber (57) using the transfer unit (501).

[0049] The vapor chamber (57) is an accessory of the cleaning equipment for removing residual harmful substances from wafers that have undergone an etching process, and introduces vapor into the chamber to remove residual harmful substances.

[0050] At this time, since chlorine (Cl) is mainly used as the process gas for the etching process, the vapor chamber (57) can effectively remove residual harmful substances by introducing steam (H2O) that can react with chlorine (Cl) to chemically react into hydrogen chloride (HCl) and then discharging it to the outside.

[0051] Additionally, when the removal of residual harmful substances from the vapor chamber (57) is completed, the transfer chamber (50) transfers the wafer back to the load lock chamber (51) using the transfer unit (501).

[0052] That is, the semiconductor process equipment of the present invention is additionally equipped with a cleaning device including a vapor chamber (57) so that residual harmful substances on the wafer after the etching process are effectively removed, thereby preventing corrosion of equipment / parts and wafer defects caused by fumes, as well as improving wafer yield and extending the lifespan of equipment / parts, and thus reducing operating costs.

[0053] In addition, when the cleaning equipment of the present invention receives the optimal parameter value generated through repeated experiments from the integrated control server (3), it proceeds with the process according to the received optimal parameter value, thereby performing the etching process under optimal conditions and maximizing wafer quality.

[0054] Figure 5 is a flowchart showing the operation process of the semiconductor process equipment of Figure 4.

[0055] The operation process (S1) of the semiconductor process equipment of FIG. 5 consists of a cassette mounting step (S10), a load lock chamber transfer step (S20), an orient chamber transfer step (S30), a wafer orientation alignment step (S40), an etching chamber transfer step (S50), an etching process step (S60), a vapor chamber transfer step (S70), a cleaning process step (S80), a load lock chamber transfer step (S90), and a cassette removal step (S100).

[0056] The cassette mounting step (10) is a process step of stacking wafers in the storage space of the cassette.

[0057] The load lock chamber transfer step (S20) is a process step in which the transfer unit (501) of the transfer chamber (50) transfers a cassette containing a wafer to the load lock chamber (51), and the load lock chamber (51) maintains a preset pressure.

[0058] The orient chamber transfer step (S30) is a process step in which the transfer unit (501) of the transfer chamber (50) transfers the cassette of the load lock chamber (51) to the orient chamber (53).

[0059] The wafer orientation alignment step (S40) is a process step in which the orientation chamber (53) aligns the orientation of the wafer.

[0060] The etching chamber transfer step (S50) is a process step in which the transfer unit (501) of the transfer chamber (50) transfers the cassette of the orient chamber (53) to the etching chamber (55).

[0061] The etching process step (S60) is a step of performing an etching process on a wafer in which an etching chamber (55) is housed.

[0062] The steam chamber transfer step (S70) is a process step in which the transfer unit (501) of the transfer chamber (50) transfers the cassette to the steam chamber (57).

[0063] The cleaning process step (S80) is a process step for removing residual harmful substances remaining inside the vapor chamber (57).

[0064] The load lock chamber transfer step (S90) is a process step in which the transfer unit (501) of the transfer chamber (50) transfers the cassette to the steam chamber (57).

[0065] The cassette removal step (S100) is a process step for removing the cassette from the load lock chamber (57).

[0066] Figure 6 is a conceptual diagram showing cleaning equipment including the vapor chamber of Figure 4.

[0067] As shown in FIG. 6, the cleaning equipment (570) consists of the steam chamber (57) of FIG. 4 described above, a DIW supply unit (571), a nitrogen supply unit (572), a flow meter (573), a steam generation unit (574), a mass flow controller (575), a mass flow meter (576), a pump (577), and a heater line (578).

[0068] At this time, the DIW supply unit (571), flow meter (573), steam generation unit (574), mass flow meter (576) and steam chamber (57) are connected to the first flow path (C1), the nitrogen supply unit (572), mass flow controller (575) and steam generation unit (574) are connected to the second flow path (C2), the second flow path (C2) between the nitrogen supply unit (572) and the mass flow controller (575) and the first flow path (C1) between the mass flow meter (576) and steam chamber (57) are connected to the third flow path (C3), and the steam chamber (57) and pump (577) are connected to the fourth flow path (C4).

[0069] As previously described in FIG. 4, when wafers that have undergone an etching process in the etching chamber (55) are placed in the vapor chamber (57), vapor is introduced to remove residual harmful substances.

[0070] Additionally, the steam chamber (57) is connected to a first flow path (C1) on one side, which has a mass flow meter (576) and a heater line (578) installed therein, through which steam is introduced, and is connected to a pump (577) through a fourth flow path (C4).

[0071] The DIW supply unit (571) supplies DI water to the first flow path (C1), and the nitrogen supply unit (572) supplies nitrogen to the second flow path (C2).

[0072] At this time, nitrogen is supplied to maintain a vacuum state.

[0073] A liquid flow meter (LFM) (573) is connected to a DIW supply unit (571) and a first flow path (C1) to measure the flow rate of DIW supplied through the first flow path (C1).

[0074] Additionally, the flow meter (573) is connected to the steam generating unit (574) and the first flow path (C1), and the first valve (EV#1) is installed in the first flow path (C1) between them.

[0075] The mass flow controller (575) controls the flow rate of nitrogen supplied from the nitrogen supply unit (572) through the second flow path (C2).

[0076] Additionally, the mass flow controller (575) is connected to the steam generating unit (574) and the second flow path (C2), and supplies an appropriate flow rate of nitrogen to the steam generating unit (574) through the second flow path (C2).

[0077] The steam generating unit (574) is connected to the flow meter (573) and the first flow path (C1), so that DI water is introduced, and simultaneously connected to the mass flow controller (575) and the second flow path (C2), so that nitrogen (N2) is introduced, and steam is generated while maintaining a preset set-pressure.

[0078] Additionally, the steam generating unit (574) is connected to the mass flow meter (576) and the first flow path (C1), and supplies steam at a set pressure to the mass flow meter (576) through the first flow path (C1).

[0079] The mass flow meter (576) is connected to the steam generating unit (574) and the first flow path (C1) to measure the flow rate of steam introduced through the first flow path (C1).

[0080] In addition, the mass flow meter (576) is connected to the steam chamber (57) and the first flow path (C1). At this time, a second valve (EV#2) is installed in the first flow path (C1) connected between the mass flow meter (576) and the steam chamber (57), and a third flow path (C3) is connected to the first flow path (C1) between the second valve (EV#2) and the steam chamber (57).

[0081] In addition, a heater line (578) is installed in the first flow path (C1) connected between the mass flow meter (576) and the steam chamber (57) to heat the moving steam.

[0082] Meanwhile, a third flow path (C3) is connected to the second flow path (C2) between the nitrogen supply unit (572) and the mass flow controller (575), and to the first flow path (C1) between the mass flow meter (576) and the steam chamber (57), and a third valve (EV#3) is installed in the third flow path (C3).

[0083] The pump (577) is connected to the steam chamber (57) and the fourth flow path (C4) to draw in the chemically reacted gas inside the steam chamber (57) and discharge it to the outside.

[0084] At this time, the fourth valve (EV#4) is installed in the fourth Euro (C4).

[0085] This cleaning device (570) can maximize the precision and accuracy of removing residual harmful substances by carrying out the cleaning process according to a preset optimal recipe, thereby effectively preventing corrosion of equipment / parts and wafer defects caused by fumes.

[0086] In this context, the optimal recipe refers to data where the optimal values ​​of each parameter are matched according to the process timeline.

[0087] Figure 7 is a conceptual diagram illustrating the optimal recipe applied to the cleaning equipment of Figure 6.

[0088] The optimal recipe applied to the cleaning equipment (570) of FIG. 6 consists of Step 1, Step 2, Step 3, Step 4, Step 5, Step 6, and Step 7, as shown in FIG. 7.

[0089] Step 1 closes the first and third valves (EV#1) and (EV#3), but opens the second and fourth valves (EV#2) and (EV#4), thereby supplying nitrogen from the nitrogen supply unit (72) so that nitrogen flows into the vapor chamber (57) for N seconds.

[0090] At this time, the maximum fixed time of Step 1 is set to 10 seconds, the throttle valve is opened to the maximum, and nitrogen is supplied at a flow rate of 2000 sccm.

[0091] Step 2 opens all first, second, and fourth valves (EV#1), (EV#2), and (EV#4) to the steam chamber (57) to supply nitrogen and DIW simultaneously, thereby ensuring time for the gas inflow into the steam chamber (57) to stabilize.

[0092] At this time, the maximum process time of Step 2 is set to 20 seconds, the throttle valve is opened to the maximum, nitrogen is supplied at a flow rate of 2000 sccm, and DIW is supplied at a mass of 0.5g.

[0093] Step 3 is a step in which the vapor chamber (57) performs a cleaning process. At this time, Step 3 opens all of the first, second, and fourth valves (EV#1), (EV#2), and (EV#4) to supply nitrogen and DIW simultaneously.

[0094] At this time, the maximum process time of Step 3 is set to 60 seconds, the throttle valve is used to maintain a pressure of 3 Torr using a 10 Torr pressure gauge, nitrogen is supplied at a flow rate of 2000 sccm, and DIW is supplied at a mass of 0.5 g.

[0095] Step 4 is a step of stopping the cleaning process of the vapor chamber (57) by stopping the DIW supply, closing the first and third valves (EV#1) and (EV#3), opening the second and fourth valves (EV#2) and (EV#4) to supply nitrogen, and stopping the supply of DIW.

[0096] At this time, the maximum process time of Step 4 is set to 20 seconds, the throttle valve is opened to the maximum, and nitrogen is supplied at a flow rate of 2000 sccm.

[0097] Step 5 is a step of purging the inside of the vapor chamber (57) by stopping the DIW supply and maintaining the nitrogen supply, closing the first and third valves (EV#1) and (EV#3), and opening the second and fourth valves (EV#2) and (EV#4).

[0098] At this time, the maximum process time of Step 5 is set to 30 seconds, the throttle valve is opened to the maximum, and nitrogen is supplied at a flow rate of 2000 sccm.

[0099] Step 6 is a step to stop the supply of nitrogen, by closing the first and third valves (EV#1) and (EV#3), and opening the second and fourth valves (EV#2) and (EV#4), so that the supply of both nitrogen and DIW is stopped.

[0100] At this time, the maximum process time of Step 6 is set to 10 seconds, and the throttle valve is opened to the maximum.

[0101] Step 7 is a step of relieving residual gas inside the steam chamber (57), by closing the first and third valves (EV#1) and (EV#3) and opening the second and fourth valves (EV#2) and (EV#4) so ​​that residual gas in the steam chamber (57) is relieved by a pump without supplying nitrogen and DIW.7

[0102] At this time, the maximum process time for Step 7 is set to 120 seconds, and the throttle valve is opened to the maximum.

[0103] As described above, the optimal recipe consisting of steps 1, 2, 3, 4, 5, 6, and 7 allows the cleaning process of the vapor chamber (57) to proceed under optimal conditions, thereby precisely and quickly removing fumes, improving wafer quality, and at the same time preventing corrosion of equipment / parts, so that the lifespan can be extended.

[0104] FIG. 8 is a block diagram showing the integrated control server of FIG. 1, and FIG. 9 is a conceptual diagram to explain FIG. 8.

[0105] As shown in FIGS. 8 and 9, the integrated control server (3) is composed of a control unit (30), a DB server (31), a data transmission and reception unit (32), a learning unit (33), a data collection / storage unit (34), a process data generation unit (35), a monitoring unit (36), an optimal parameter reference-value detection unit (37), an equipment anomaly detection unit (38), a quality prediction unit (39), an incident-data generation unit (40), a report generation unit (41), and a platform operation unit (42).

[0106] The control unit (30) is the OS (Operating System) of the integrated control server (3) and manages and controls the operation of the control targets (31), (32), (33), (34), (35), (36), (37), (38), (39), (40), (41), (42).

[0107] In addition, when the control unit (30) receives raw data from the PLCs (7) through the data transmission / reception unit (32), it inputs the received raw data into the data collection / storage unit (34).

[0108] Additionally, when the control unit (30) receives sensing values ​​from IoT sensors (9) through the data transmission / reception unit (32), it inputs the received sensing values ​​into the data collection / storage unit (34).

[0109] Additionally, the control unit (30) executes the sudden-data generation unit (40) when an equipment abnormality is detected by the equipment abnormality detection unit (38).

[0110] Additionally, when sudden data is generated in the sudden data generation unit (40), the control unit (30) controls the data transmission and reception unit (32) so that the sudden data is transmitted to the manager terminal (11) of the semiconductor process equipment (5).

[0111] In addition, the control unit (30) executes the report generation unit (41) every preset second cycle (T2).

[0112] In addition, the control unit (30) executes the learning unit (33) at every preset period (T).

[0113] The DB server (31) stores communication identification information and location information of each PLC, as well as information on the connected semiconductor process equipment (5).

[0114] In addition, the DB server (31) stores communication identification information and location information of each IoT sensor (9).

[0115] In addition, the DB server (31) stores an anomaly detection algorithm and a quality prediction algorithm learned by the learning unit (33).

[0116] At this time, the anomaly detection algorithm is an AI algorithm that analyzes process data and outputs whether each component of the semiconductor process equipment (5) is abnormal, and the quality prediction algorithm is an AI algorithm that analyzes process data and outputs a prediction of the wafer's quality.

[0117] In addition, the report generated by the report generation unit (40) is stored in the DB server (31).

[0118] In addition, the optimal recipe information is stored in the DB server (31).

[0119] The data transmission and reception unit (32) transmits and receives data with the PLCs (7), IoT sensors (9), and the administrator terminal (11).

[0120] The learning unit (33) is executed at a preset period (T) under the control of the control unit (30) and performs learning of an anomaly detection algorithm and a quality prediction algorithm.

[0121] At this time, the anomaly detection algorithm generates training data capable of learning the correlation between process data and the anomaly detection results of each part, and performs training by utilizing the generated training data to derive an extraction model, which is a set of parameter values ​​regarding the correlation between the process data and the anomaly detection results.

[0122] In addition, the quality prediction algorithm generates training data capable of learning the correlation between process data and wafer quality, and performs training by utilizing the generated training data to derive an extraction model, which is a set of parameter values ​​regarding the correlation between process data and wafer quality.

[0123] The data collection / storage unit (34) collects raw data transmitted from PLCs (7) and sensing values ​​transmitted from IoT sensors (9) through the data transmission / reception unit (32), and stores them in the DB server (31).

[0124] The process data generation unit (35) analyzes and utilizes raw data and sensing values ​​collected from the data collection / storage unit (34) to generate process data for the semiconductor process equipment (5).

[0125] At this time, process data refers to data in which the process state, the operating state of each part, and the sensing value of each IoT sensor are matched, and is stored in the DB server (31) according to the control of the control unit (30).

[0126] The monitoring unit (36) monitors the process data generated by the process data generation unit (35).

[0127] The optimal parameter reference-value detection unit (37) detects the optimal parameter reference-value of each part or sensor by referencing and utilizing the process data and optimal recipe information generated by the process data generation unit (35).

[0128] At this time, the control unit (30) transmits the optimal parameter reference value detected by the optimal parameter reference value detection unit (37) to the corresponding PLC (7).

[0129] The equipment anomaly detection unit (38) detects whether there is an equipment anomaly by analyzing process data generated by the process data generation unit (35) using an AI-based anomaly detection algorithm learned by the learning unit (33).

[0130] At this time, when the equipment abnormality is detected by the equipment abnormality detection unit (38), the control unit (30) executes the sudden-data generation unit (40).

[0131] The quality prediction unit (39) predicts quality by analyzing process data generated by the process data generation unit (35) using an AI-based quality prediction algorithm learned by the learning unit (33).

[0132] When an equipment abnormality is detected by the equipment abnormality detection unit (38), the sudden-data generation unit (40) generates sudden-data including identification information of the semiconductor process equipment where the equipment abnormality was detected, equipment identification information, and details of the abnormality.

[0133] At this time, the sudden data generated by the sudden data generation unit (40) is transmitted to the administrator terminal (11) through the data transmission and reception unit (32) under the control of the control unit (30).

[0134] The report generation unit (41) is executed every second cycle (T2) set according to the control of the control unit (30) and generates a report including process details, process data, equipment abnormality results, quality prediction results, etc. of each semiconductor process equipment (5).

[0135] At this time, the report generated by the report generation unit (41) is transmitted to the administrator terminal (11) through the data transmission and reception unit (32) under the control of the control unit (30).

[0136] The platform operation unit (42) manages firmware updates, UI updates, etc. of the process management platform (13) and controls the operation of the process management platform (13).

[0137] As described above, the AI-based semiconductor process control system (1), which is an embodiment of the present invention, can increase the efficiency, accuracy, and convenience of controlling the semiconductor process by having an integrated control server analyze collected raw data and sensing values ​​to generate process data, detect equipment abnormalities, and predict quality.

[0138] In addition, the AI-based semiconductor process control system (1) of the present invention enables the semiconductor process equipment to perform the process under optimal conditions, thereby maximizing the yield and quality of the wafer compared to the same conditions.

[0139] In addition, the AI-based semiconductor process control system (1) of the present invention enables accurate and objective recognition of the semiconductor process status by having the integrated control server generate a report every second cycle (T2) that is pre-set and provide it to managers, thereby further increasing the reliability of the semiconductor process.

[0140] In addition, the AI-based semiconductor process control system (1) of the present invention can effectively prevent corrosion of equipment / parts caused by fumes by installing additional cleaning equipment to remove residual harmful substances from a wafer that has undergone an etching process in the semiconductor process equipment, and at the same time, introducing steam into the wafer that has undergone the etching process by the cleaning equipment. Furthermore, the wafer quality is improved and the lifespan of the equipment / parts is extended. Explanation of the symbols

[0141] 1: AI-based semiconductor process control system 3: Integrated control server 5: Semiconductor process equipment 7: PLC 9: IoT Sensor 10: Network 11: Administrator Terminal 12: Control Service Application 13: Process Management Platform 30: Control Unit 31: DB Server 32: Data Transmitter / Receiver 33: Learning Unit 34: Data Collection / Storage Unit 35: Process Data Generation Unit 36: Monitoring unit 37: Optimal parameter reference-value detection unit 38: Equipment Anomaly Detection Unit 39: Quality Prediction Unit 40: Unexpected-Data Generation Unit 41: Report Generation Unit 42: Platform Operations Department 50: Transfer Chamber 51: Load lock chamber 53: Wafer orient chamber 55: Etching chamber 57: Vapor chamber 501: Transfer Unit 570: Cleaning Equipment 571:DIW Supply Unit 572:Nitrogen Supply Unit 573: Flow meter 574: Steam generator 575: Mass flow controller 576: Mass flow meter 577: Pump 578: Heater line

Claims

Claim 1 In an AI-based semiconductor process control system, semiconductor process equipment for performing a semiconductor process; an integrated control server connected to the semiconductor process equipment; The system includes an administrator terminal that receives incident data from the integrated control server, wherein the integrated control server comprises: a DB server in which an anomaly detection algorithm, which is an AI algorithm that analyzes process data and outputs whether each component of the semiconductor process equipment is abnormal, and a quality prediction algorithm, which is an AI algorithm that analyzes process data and outputs a wafer quality prediction, are stored; a process data generation unit that generates process data by matching raw data received from a PLC (Program Logic Controller) of the semiconductor process equipment and sensing values ​​received from IoT sensors installed in the semiconductor process equipment, after collecting the raw data and sensing values; an equipment anomaly detection unit that detects whether there is an equipment anomaly by analyzing the process data generated by the process data generation unit using the anomaly detection algorithm; a quality prediction unit that predicts the wafer quality by analyzing the process data generated by the process data generation unit using the quality prediction algorithm; and an incident data generation unit that generates incident data including identification information of the semiconductor process equipment where the equipment anomaly was detected, equipment identification information, and details of the anomaly when the equipment anomaly is detected by the equipment anomaly detection unit. The semiconductor process equipment includes a control unit that transmits the incident-data generated by the incident-data generation unit to the administrator terminal, and comprises: a transfer chamber including a transfer unit; a load lock chamber that receives a wafer in an unprocessed state from the outside, maintains a predetermined pressure, and performs the function of loading or unloading the wafer; an orient chamber that includes slot units equipped with a plurality of slots and aligns the orientation of the received wafers when the wafers are transferred and received by the transfer unit; and an etching chamber that receives the wafer by the transfer unit and performs etching processing.An AI-based semiconductor process control system characterized by including a vapor chamber that receives a wafer via the above-mentioned transfer unit and introduces vapor to remove residual harmful substances from the wafer. Claim 2 delete Claim 3 The AI-based semiconductor process control system according to claim 1, wherein the integrated control server further includes a learning unit executed at each preset period (T), and the learning unit performs the learning of the anomaly detection algorithm by generating learning data capable of learning the correlation between process data and anomaly detection results for each component of the semiconductor process equipment, and then using the generated learning data to derive an extraction model which is a set of parameter values ​​regarding the correlation between process data and anomaly detection results, and performs the learning of the quality prediction algorithm by generating learning data capable of learning the correlation between process data and wafer quality, and then using the generated learning data to derive an extraction model which is a set of parameter values ​​regarding the correlation between process data and wafer quality. Claim 4 delete Claim 5 The AI-based semiconductor process control system according to claim 1, wherein the semiconductor process equipment is configured to stack wafers in a storage space of a cassette, and the transfer unit transfers the cassette containing the wafers to the load lock chamber, and the load lock chamber maintains a preset pressure, and the transfer unit transfers the cassette of the load lock chamber to the orient chamber, and the orient chamber aligns the orientation of the wafers, and the transfer unit transfers the cassette of the orient chamber to the etching chamber, and the etching chamber performs an etching process of the wafers contained therein, and the transfer unit transfers the cassette to the vapor chamber, and the vapor chamber introduces vapor to remove residual harmful substances remaining inside, and the transfer unit transfers the cassette from the vapor chamber to the load lock chamber, and removes the cassette from the load lock chamber. Claim 6 In claim 5, the semiconductor process equipment further comprises cleaning equipment including the vapor chamber, a DIW supply unit, a nitrogen supply unit, a flow meter, a vapor generation unit, a mass flow controller, a mass flow meter, and a pump; the DIW supply unit, the flow meter, the vapor generation unit, the mass flow meter, and the vapor chamber are connected by a first flow path (C1); the nitrogen supply unit, the mass flow controller, and the vapor generation unit are connected by a second flow path (C2); a third flow path (C3) is connected to the second flow path (C2) between the nitrogen supply unit and the mass flow controller and the first flow path (C1) between the mass flow meter and the vapor chamber; the vapor chamber and the pump are connected by a fourth flow path (C4); the vapor chamber is connected to the mass flow meter and the first flow path (C1) on one side so that vapor is introduced, and is connected to the pump through the fourth flow path (C4); and the DIW supply unit is connected to the first flow path (C1). DI water is supplied, the nitrogen supply unit supplies nitrogen to the second flow path (C2), the flow meter is connected to the DIW supply unit and the first flow path (C1) to measure the flow rate of DI water supplied through the first flow path (C1), the steam generation unit is connected to the first flow path (C1), and the mass flow rate controller controls the flow rate of nitrogen supplied from the nitrogen supply unit through the second flow path (C2), the steam generation unit is connected to the second flow path (C2) to supply nitrogen to the steam generation unit through the second flow path (C2), the steam generation unit is connected to the flow meter and the first flow path (C1) so that DI water flows in and simultaneously nitrogen flows in, and generates steam while maintaining a preset set-pressure, and is connected to the mass flow meter and the first flow path (C1) so that through the first flow path (C1) to the mass flow meter Steam of a set pressure is supplied, and the mass flow meter is connected to the steam generating unit and the first flow path (C1).An AI-based semiconductor process control system characterized by measuring the flow rate of steam introduced through the first flow path (C1) and connecting the steam chamber to the first flow path (C1). Claim 7 delete

Images