TW202101563A - Control system for controlling polishing device for polishing substrate, and polishing method - Google Patents

Abstract

Provided are a control system having an improved yield, for controlling a polishing device, and a polishing method. A defect data receiving portion 112 of a control system 102 receives defect data relating to a defect of a substrate. A processing data receiving portion 114 receives processing data relating to the content of processing of the substrate by a polishing device 1000. A defect occurrence cause identifying portion 116 identifies a cause of occurrence of the defect on the basis of the defect data and the processing data. A defect correcting measure portion 118 determines a defect correcting measure for correcting the defect, with regard to the identified occurrence cause, and instructs at least one of a polishing unit 300, a washing unit 400, and an EFEM 200 to carry out the defect correcting measure.

Description

Control system and polishing method for controlling polishing device for polishing substrate

The present invention relates to a control system and a polishing method for controlling a polishing device for polishing a substrate.

The manufacturing of semiconductor devices is to form many kinds of materials in a film on a substrate, such as a silicon wafer, to form a multilayer structure. In order to form this layered structure, a technique to flatten the surface of the wafer is important. One of the methods for flattening the surface of a wafer is a polishing device (also known as a chemical mechanical polishing device) that performs chemical mechanical polishing (CMP).

A chemical mechanical polishing (CMP) device usually has: a polishing table with a polishing pad installed; an annular turntable that holds the wafer above; and a nozzle that supplies polishing liquid to the polishing pad. The polishing liquid is supplied to the polishing pad from the nozzle, and the wafer is pressed against the polishing pad by the upper ring turntable, and the upper ring turntable and the polishing table are moved relative to each other to polish the wafer and flatten its surface. The CMP device has a cleaning section that cleans the polished wafer and further dries it.

In this type of polishing device, it is required to increase the yield rate of processing substrates. Since the polishing device has various treatment parts for polishing and cleaning, if defects occur in each treatment part, the pass rate of the entire polishing device will decrease. The pass rate of the entire polishing device not only affects the processing parts such as the polishing part and the cleaning part, but also affects the wafer transfer mechanism (transport part). Therefore, the pass rate of the entire polishing device depends on various processing steps and conveying steps.

In a semiconductor factory, when a defect is found on a wafer being manufactured, it is necessary to trace the process in which the defect was found and the process upstream of the process to find out the cause. In the investigation of the cause, stopping the device that may cause defects for too long will delay the production process and reduce the production volume. [Prior Technical Literature] [Patent Literature]

[Patent Document 1] Japanese Patent Application Publication No. 2000-269108

(The problem to be solved by the invention)

One aspect of the present invention is to solve this problem, and its purpose is to provide a control system and a polishing method for controlling a polishing device with improved yield. (Means to solve the problem)

In order to solve the above-mentioned problems, the first form adopts a configuration of a control system, which is used to control a polishing device for polishing a substrate, and is characterized in that the aforementioned polishing device has: a polishing portion configured to polish the substrate; a cleaning portion , Which is configured to clean the substrate after polishing; and a transport section, which is configured to transport the substrate between the polishing section and the cleaning section; the control system has: a defect data receiving section, which is It is constituted by receiving the defect data about the defect of the aforementioned substrate; the processing data receiving unit is constituted by receiving the processing data regarding the processing content of the aforementioned substrate in the aforementioned polishing device; the defect occurrence cause determination unit is based on The aforementioned defect data and the aforementioned processing data are constituted by the method to determine the cause of the aforementioned defect; and the defect correction measure department, which determines the defect correction measure to correct the aforementioned defect based on the confirmed cause of the aforementioned defect, and provides an assessment of the aforementioned grinding part It is constructed with at least one of the aforementioned cleaning unit and the aforementioned conveying unit instructing the aforementioned defect correction measures.

This embodiment has: a defect cause determination unit constructed by determining the cause of the defect based on the defect data and processing data; and determining the defect correction measures to correct the defect based on the determined cause of occurrence, and the grinding At least one of the cleaning department, the cleaning department, and the conveying department constitutes a defect correction measure department that indicates defect correction measures; therefore, a control system that improves the pass rate for the polishing device can be provided.

Here, the so-called defect data refers to data showing information about substrate defects. Defect data such as the type of defect (the presence or absence of dust on the wafer (hereinafter also referred to as "particles"); the presence or absence of wafer damage (hereinafter also referred to as "scratches"); poor polishing (hereinafter also referred to as "scratches") "Outline abnormality"), etc. (type of flaw); and state of flaw (number of dust damage, size, shape, color, position, distribution state, depth of flaw, amount of over or under grinding, etc.) 1 piece.

The so-called processing data refers to data showing what kind of processing the substrate receives in the polishing device. The processing data is, for example, batch information for identifying the substrates in units of individual or plural pieces; information about the processing of the substrates received in each process inside and outside the polishing device; alarm information received by the substrates in each process; At least one of the processed consumables information (type of consumables such as polishing pads, usage time, etc.). In addition, the defective data receiving section and the processing data receiving section can be combined into one data receiving section. That is, one embodiment of the present invention can also receive defect data and processing data by one data receiving unit.

Defect correction measures can be, for example, to quickly stop the constituent units of the polishing device (grinding section, cleaning section, conveying section, etc.) that are the cause of the defect and cut it away from the production line; or change the configuration of the polishing device to solve the cause of the defect. Unit operation type (adjustment of grinding volume, adjustment of cleaning water volume, etc.).

The form 2 adopts the configuration of the control system described in form 1, and is characterized in that it has a measure result data processing section, which implements the defect correction measures by at least one of the polishing section, the cleaning section, and the transport section After receiving the measure result data on the results of the aforementioned defect correction measures, the method composition of the aforementioned defect correction measures can be changed based on the aforementioned measure result data.

Form 3 adopts the configuration of the control system described in Form 1 or Form 2, and is characterized by having a defect knowledge database, which is constructed by accumulating at least one of the aforementioned defect data and the aforementioned processing data. For example, batch information of defective wafers, type of defect, cause of defect occurrence, and defect correction measures can also be correlated and accumulated into a database. Thereby, the operation method of the grinding device and the accuracy of determining the cause of the defect can be improved.

The form 4 is a configuration using a polishing device, and is characterized by having the control system described in any one of the forms 1 to 3, the polishing section, the cleaning section, and the conveying section.

Mode 5 is a configuration using a polishing device system, and is characterized by having the control system described in any one of Modes 1 to 3, and a plurality of the aforementioned polishing devices.

Form 6 is a configuration using a polishing method, using a polishing device having a polishing section for polishing the substrate, a cleaning section for cleaning the substrate after polishing, and a transporting section for transporting the substrate between the polishing section and the cleaning section, To grind the aforementioned substrate, it is characterized by: receiving defect data on the defect of the aforementioned substrate, receiving processing data on the processing content of the aforementioned substrate in the aforementioned polishing device, and determining the cause of the aforementioned defect based on the aforementioned defect data and the aforementioned processing data. The determined cause of occurrence determines the defect correction measures used to correct the defect, and at least one of the polishing part, the cleaning part, and the conveying part indicates the defect correction measure.

Form 7 is a configuration using the polishing method described in Form 6, characterized in that at least one of the polishing part, the cleaning part, and the conveying part has performed the defect correction measures, and then receives the result of the defect correction measures. According to the above-mentioned measure result data, change the aforementioned defect correction measures.

The form 8 adopts the polishing method described in form 6 or form 7, and is characterized in that at least one of the aforementioned defect data and the aforementioned processing data is accumulated in the defect knowledge database.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. In addition, in the various embodiments below, the same or equivalent members are given the same reference numerals, and repeated descriptions are omitted. In addition, as long as the features shown in the various embodiments do not contradict each other, they can also be applied to other embodiments.

The embodiment of the present invention may be the case of only one polishing device or the case of a plurality of polishing devices. In addition, a control system for controlling the polishing device for polishing the substrate can be configured in the polishing device as a part of the polishing device. In addition, the control system can also be used as an external device of the polishing device, and is arranged outside the polishing device. Furthermore, one control system can control a plurality of polishing devices. Hereinafter, the outline of the polishing device will be described first. Then, a polishing device system that controls a plurality of polishing devices with one control system will be described. ＜Grinding device＞

Figure 1 is a top view of the polishing device. As shown in FIG. 1, the polishing apparatus 1000 includes an EFEM (Equipment Front End Module) 200, a polishing unit 300, and a cleaning unit 400. In addition, the polishing apparatus 1000 further includes a control unit 500 for controlling various operations of the EFEM 200, the polishing unit 300, and the cleaning unit 400.

The polishing unit 300 is a polishing part configured to polish a substrate. The cleaning unit 400 is a cleaning section configured to clean a polished substrate. The EFEM200 is a conveying section configured to convey the substrates in a carrier containing a plurality of substrates and convey them to the polishing apparatus 1000 for processing, and to convey the processed substrates from the polishing apparatus 1000 to the carrier. Hereinafter, the EFEM 200, the polishing unit 300, and the cleaning unit 400 will be described. ＜EFEM＞

EFEM200 is a unit used to deliver the substrate before processing such as polishing and cleaning to the polishing unit 300, and to receive the substrate after processing such as polishing and cleaning from the cleaning unit 400. The EFEM200 is equipped with a front loading unit 220 of plural units (4 units in this embodiment). The front loading part 220 is respectively mounted with cassettes 222 for storing substrates.

The EFEM200 is equipped with a rail 230 installed inside the housing 100; and a plurality of (two in this embodiment) transport robot 240 arranged on the rail 230. The transfer robot 240 takes out the substrate before processing such as polishing and cleaning from the cassette 222 and sends it to the polishing unit 300. In addition, the transport robot 240 receives the substrate after processing such as polishing and cleaning from the cleaning unit 400 and returns it to the cassette 222. ＜Grinding unit＞

The polishing unit 300 is a unit for polishing a substrate. The polishing unit 300 includes a first polishing unit 300A, a second polishing unit 300B, a third polishing unit 300C, and a fourth polishing unit 300D. The first polishing unit 300A, the second polishing unit 300B, the third polishing unit 300C, and the fourth polishing unit 300D have the same configuration. Therefore, only the first polishing unit 300A will be described below.

The first polishing unit 300A includes a polishing table 320A and an upper ring turntable 330A. The polishing table 320A is rotationally driven by a driving source not shown. A polishing pad 310A is attached to the polishing table 320A. The upper ring turntable 330A holds the substrate and is pressed against the polishing pad 310A. The upper ring wheel 330A is rotationally driven by a driving source not shown. The substrate is held on the upper ring turntable 330A and is polished by pressing on the polishing pad 310A.

Next, the conveying mechanism for conveying the substrate is explained. The transport mechanism includes an elevator 370, a first linear conveyor 372, a swing conveyor 374, a second linear conveyor 376, and a temporary table 378.

The elevator 370 receives the substrate from the transfer robot 240. The first linear conveyor 372 conveys the substrate received from the elevator 370 between the first transfer position TP1, the second transfer position TP2, the third transfer position TP3, and the fourth transfer position TP4. The first polishing unit 300A and the second polishing unit 300B receive the substrate from the first linear conveyor 372 for polishing. The first polishing unit 300A and the second polishing unit 300B send the polished substrates to the first linear conveyor 372.

The swing conveyor 374 transfers substrates between the first linear conveyor 372 and the second linear conveyor 376. The second linear conveyor 376 transfers the substrate received from the swing conveyor 374 between the fifth transfer position TP5, the sixth transfer position TP6, and the seventh transfer position TP7. The third polishing unit 300C and the fourth polishing unit 300D receive the substrate from the second linear conveyor 376 for polishing. The third polishing unit 300C and the fourth polishing unit 300D send the polished substrate to the second linear conveyor 376. The substrate after being polished by the polishing unit 300 is placed on the temporary table 378 by the swing conveyor 374. ＜Cleaning unit＞

The cleaning unit 400 is a unit for cleaning and drying the substrate after the polishing process. The cleaning unit 400 includes a first cleaning room 410, a first transfer room 420, a second cleaning room 430, a second transfer room 440, and a drying room 450.

The substrate placed on the temporary stage 378 is transferred to the first cleaning chamber 410 or the second cleaning chamber 430 via the first transfer chamber 420. The substrate is cleaned in the first cleaning chamber 410 or the second cleaning chamber 430. The substrate cleaned in the first cleaning chamber 410 or the second cleaning chamber 430 is transported to the drying chamber 450 via the second transport chamber 440. The substrate is dried in the drying chamber 450. After the drying process, the substrate is taken out from the drying chamber 450 by the transport robot 240 and returned to the cassette 222. <Detailed structure of the first grinding unit>

Next, the first polishing unit 300A will be described in detail. FIG. 2 is a perspective view of the first polishing unit 300A. The first polishing unit 300A includes a polishing liquid supply nozzle 340A for supplying polishing liquid or dressing liquid to the polishing pad 310A. The polishing liquid is, for example, a slurry. The dressing liquid is pure water, for example. In addition, the first polishing unit 300A further includes a dresser 350A for conditioning the polishing pad 310A. In addition, the first polishing unit 300A further includes a sprayer 360A for spraying liquid or a mixed fluid of liquid and gas toward the polishing pad 310A. The liquid is pure water, for example. The gas is, for example, nitrogen.

The upper ring turntable 330A is supported by the upper ring turntable shaft 332A. The upper ring turntable 330A can be rotated around the axis of the upper ring turntable shaft 332A by a driving part (not shown) as shown by arrow AB. The polishing table 320A is supported by the table shaft 322A. The polishing table 320A can be rotated around the axis of the table shaft 322A by a driving part not shown in the figure as shown by the arrow AC.

The substrate WF is held on the surface of the upper ring turntable 330A opposite to the polishing pad 310A by vacuum suction. During polishing, the polishing liquid is supplied to the polishing surface of the polishing pad 310A from the polishing liquid supply nozzle 340A. In addition, during grinding, the grinding table 320A and the upper ring turntable 330A are rotationally driven. The substrate WF is pressed against the polishing surface of the polishing pad 310A by the upper ring turntable 330A for polishing. ＜Detailed structure of cleaning unit＞

Next, the cleaning unit 400 will be described in detail. 3A is a top view of the cleaning unit 400. 3B is a side view of the cleaning unit 400.

In the first transfer chamber 420, a support shaft 424 extending in the vertical direction and a first transfer robot 422 supported by the support shaft 424 are arranged. The first transfer robot 422 can move in the vertical direction along the support shaft 424 by a driving mechanism such as a motor. The first transfer robot 422 receives the substrate WF placed on the temporary stage 378. The first transfer robot 422 transfers the substrate WF received from the temporary placement table 378 to the first cleaning chamber 410 or the second cleaning chamber 430.

In the first cleaning chamber 410, the upper primary cleaning module 412A and the lower primary cleaning module 412B are arranged along the vertical direction. Similarly, in the second cleaning chamber 430, the upper secondary cleaning module 432A and the lower secondary cleaning module 432B are arranged along the vertical direction. The upper primary cleaning module 412A, the lower primary cleaning module 412B, the upper secondary cleaning module 432A, and the lower secondary cleaning module 432B are cleaning machines that use cleaning fluid to clean substrates. Between the upper secondary cleaning module 432A and the lower secondary cleaning module 432B, a temporary placement table 434 for the substrate is provided.

The first transport robot 422 can transport between the temporary table 378, the upper primary cleaning module 412A, the lower primary cleaning module 412B, the temporary table 434, the upper secondary cleaning module 432A, and the lower secondary cleaning module 432B Substrate WF. The first transfer robot 422 has an upper arm and a lower arm. Since the slurry is attached to the substrate before cleaning, the first transfer robot 422 uses the lower arm when transporting the substrate before cleaning. In addition, the first transfer robot 422 uses the upper arm when transferring the cleaned substrate.

In the second transfer chamber 440, a support shaft 444 extending in the vertical direction and a second transfer robot 442 movably supported by the support shaft 444 are arranged. The second transfer robot 442 can be moved in the vertical direction along the support shaft 444 by a driving mechanism such as a motor. The second transfer robot 442 receives the substrate WF from the second cleaning chamber 430 and transfers it to the drying chamber 450.

In the drying chamber 450, the upper drying module 452A and the lower drying module 452B are arranged along the vertical direction. The upper drying module 452A includes a drying unit 456A for drying the substrate WF, and a fan filter unit 454A for supplying clean air to the drying unit 456A. The lower drying module 452B includes a drying unit 456B for drying the substrate WF, and a fan filter unit 454B for supplying clean air to the drying unit 456B.

The second transfer robot 442 can transfer the substrate WF between the upper secondary cleaning module 432A, the lower secondary cleaning module 432B, the temporary stage 434, the upper drying module 452A, and the lower drying module 452B. Since the second transfer robot 442 transfers the cleaned substrate, it only has one arm.

After the substrate WF is dried by the upper drying module 452A or the lower drying module 452B, it is taken out from the drying chamber 450 by the upper arm of the transport robot 240 shown in FIG. 1. The transfer robot 240 returns the substrate WF taken out from the drying chamber 450 to the cassette 222.

Next, referring to FIG. 4, a polishing device system in which a plurality of polishing devices 1000 are controlled by one control system 102 will be described. The polishing device system 101 includes a control system 102 and a plurality of polishing devices 1000. Multiple polishing devices 1000 are placed on multiple manufacturing lines 104-1, 104-2, ‧‧‧, 104-N. Each manufacturing line 104-1, 104-2, ‧‧‧, 104-N is equipped with a plurality of semiconductor manufacturing equipment exposure equipment, dry etching equipment, CVD equipment, polishing equipment 1000, etc. FIG. 4 only shows the polishing device 1000. Other semiconductor manufacturing equipment is omitted. The plural polishing devices 1000 are referred to as a polishing device group 108. The control system 102 is a pass rate improvement system that controls the polishing device 1000, improves the pass rate of the polishing device 1000 when polishing the substrate, and improves the pass rate.

In addition, in the embodiment of FIG. 4, one control system 102 controls a plurality of polishing devices 1000. In other embodiments, the control system 102 may be configured for each polishing device 1000. Furthermore, the control system 102 may also be arranged in a computer outside the factory where the polishing device group 108 and the inspection device group 110 are arranged. The computer installed outside the factory can also be the cloud. When a computer installed outside the factory is used, the computer is connected to the polishing device group 108 and the inspection device group 110 through communication means such as the Internet or a dedicated line.

The plurality of polishing devices 1000 are not limited to those having the same structure and function, and may be those having different structures and functions. For example, a polishing device 1000 suitable for polishing metal films and a polishing device 1000 suitable for polishing insulating films can also be configured. The polishing device 1000 is marked with a serial number such as "N-2", which shows the serial number of the manufacturing line 104 and the serial number in the manufacturing line 104.

Each manufacturing line 104-1, 104-2, ‧‧‧, 104-N is equipped with a plurality of inspection devices 106 for inspecting substrates. The inspection device 106 inspects the substrate for defects and the like against the designated reference. Defects include dust, foreign matter, damage, circuit breakage, excessive or insufficient film thickness, etc. The inspection device 106 is an SEM (scanning electron microscope) type inspection device, an optical type inspection device, a film thickness inspection device, or the like. The plural inspection devices 106 are not limited to those having the same structure and function, and may be those having different structures and functions. The inspection device 106 is marked with a number showing the serial number of the manufacturing line 104 and the serial number such as "N-2" in the manufacturing line 104.

The inspection device 106 and the polishing device 1000 can correspond to various types. For example, one inspection device 106 corresponds to one polishing device 1000, and the substrate can be inspected before and after polishing or during polishing. In some cases, one inspection device 106 corresponds to a plurality of polishing devices 1000 to inspect the substrates processed by the plurality of polishing devices 1000. Or, a plurality of inspection devices 106 may correspond to one polishing device 1000, and the plurality of inspection devices 106 may inspect the substrate processed by one polishing device 1000.

In addition, the inspection device 106 is sometimes used in addition to the polishing process by the polishing device 1000. In addition, the polishing device 1000 may also have a film thickness sensor to measure the film thickness. When the film thickness is abnormal, the defect is transmitted to the defect data receiving unit 112 via the signal line 122.

Figure 4 shows the same number of grinding devices 1000 and the same number of inspection devices 106 on each manufacturing line 104-1, 104-2, ‧‧‧, 104-N, that is, N grinding devices 1000 and M inspections装置106。 Device 106. However, the number of polishing devices 1000 and the number of inspection devices 106 in each manufacturing line 104 may also be different in each manufacturing line 104. The plural inspection devices 106 are referred to as an inspection device group 110.

The control system 102 has: a defect data receiving unit 112 configured to receive defect data about a substrate defect; a processing data receiving unit 114 configured to receive processing data about the processing content of the substrate in the polishing device; and based on the defect Data and processing data, the defect cause determination unit 116 constituted by the method to determine the cause of the defect; and determine the defect correction measures for correcting the defect based on the determined cause of occurrence, for at least one of the grinding department, cleaning department, and conveying department The defect correction measure section 118 constituted by the method of indicating defect correction measures.

The control system 102 further has a database update unit 142 (measurement result data processing unit), which receives the measure result data about the result of the defect correction measure after at least one of the polishing department, the cleaning department, and the conveying department executes defect correction measures , The defect correction measures can be changed based on the measure result data.

The defect data receiving unit 112, the processing data receiving unit 114, the defect occurrence cause determination unit 116, the defect correction measure unit 118, and the database update unit 142 accept instructions from the control unit 146 to perform respective processing as described later. In addition, the defect data receiving unit 112, the processing data receiving unit 114, the defect occurrence cause determination unit 116, the defect correction measure unit 118, and the database update unit 142 may also be configured not to accept instructions from the control unit 146, but to exchange signals with each other. , And the sequence of actions described later.

The above-mentioned parts of the control system 102 are controlled by the control part 146, and are controlled by repeatedly performing the following procedures. That is, the control unit 146 a) sends an instruction to the defect data receiving unit 112 via the signal line 148, receives defect data about the defect from the inspection device group 110, and collects defect information. The defect data receiving unit 112 receives all inspection data from the inspection device group 110, and when the inspection data contains information indicating defects, the inspection data is treated as the defect data. In addition, the defect data receiving unit 112 may receive only the defect data from the inspection device group 110. When the defect data receiving unit 112 receives the defect data, the content is transmitted to the control unit 146 via the signal line 148. Here, the inspection data refers to data obtained by inspecting the substrate by the inspection device group 110. b) Secondly, send an instruction to the processing data receiving unit 114 via the signal line 148, and request the processing data receiving unit 114 to receive processing data on the processing content of the substrate in the polishing device, and collect production information. At this time, the processing data receiving unit 114 may also receive processing data about the processing content of the substrate only for the polishing device 1000 that has the defect, or may receive processing data about the processing content of the substrate for all the polishing devices 1000. c) Next, when the defect data receiving unit 112 receives the defect data from the inspection device group 110, it sends an instruction to the defect occurrence cause determination unit 116 via the signal line 148 to determine the cause of the defect occurrence. When the defect data receiving unit 112 does not receive the defect data from the inspection device group 110, it returns to step a). d) Secondly, send an instruction to the defect correction measure section 118 via the signal line 148, determine the defect correction measure to correct the defect based on the confirmed cause, and correct at least one of the grinding section, cleaning section, and conveying section. Measures. At least one of the polishing section, cleaning section, and conveying section shall perform defect correction measures in accordance with the instructions. e) Next, an instruction is sent to the defect data receiving unit 112 via the signal line 148 to request inspection data from the inspection device group 110. At the same time, an instruction is sent to the processing data receiving unit 114 via the signal line 148 to request the processing data receiving unit 114 to receive processing data about the processing content of the substrate in the polishing device, and to collect production information. Then, an instruction is sent to the database update unit 142 via the signal line 148 to request to receive the measure result data (received inspection data and processing data) regarding the result of the defect correction measure, and change the defect correction measure based on the measure result data. In this way, corrective measures are implemented and the effect measurement after implementation is performed. Then return to step a), repeat steps a) to e).

In step e), the database update unit 142 does not change the defect in the defect knowledge database 144 when the obtained measure result data is consistent with the previous measure result data accumulated in the defect knowledge database 144 within the specified range. Corrective measures. The database update unit 142 changes the defect correction measures in the defect knowledge database 144 when the obtained measure result data is inconsistent with the previous measure result data within the specified range and has been improved. The database update unit 142 sends an alarm to the control unit 146 when the obtained measure result data is inconsistent with the previous measure result data within the specified range and worsens. The control unit 146 further sends the alarm to the control unit 500 or the host computer 126.

The relationship between the control system 102 and the control unit 500 in the polishing device 1000 is as follows. When the polishing device 1000 includes a control system 102, the control system 102 can be used as a part of the control unit 500. When the polishing device 1000 does not include the control system 102, for example, when a plurality of polishing devices 1000 are operated as shown in FIG. 4, the control system 102 can control each part of the polishing device 1000 via the control unit 500 to perform defect correction measures.

The actions of the embodiments of the present invention can also be performed using the following software and/or systems. For example, each polishing device 1000 may also have: a main controller (control unit 500) that controls each polishing device 1000; and a separate control unit (EFEM 200, polishing unit 300, cleaning unit 400, etc.) of each polishing device 1000. Multiple unit controllers. The main controller and the unit controller respectively have a CPU, a memory, a recording medium, and software stored in the recording medium to operate each unit.

The control system 102 has a CPU, a memory, a recording medium, and software recorded on the recording medium. The control system 102 monitors and controls the defects of the entire polishing device 1000, and therefore performs signal transmission and reception, information recording, and calculation. The software can be updated. In addition, the software may not be updated.

Next, each element of the control system 102 will be described. The defect data receiving unit 112 receives defect data from the inspection device group 110, the polishing device group 108, and the host computer 126 via the signal line 120, the signal line 122, and the signal line 128 (the batch information of the substrate, the defect, the processing conditions, etc.). Information). And from the inspection device group 110 through the signal line 120, the batch information and defect data (dust, damage, etc.) of the substrate are transmitted. And from the polishing device group 108 via the signal line 122, data on the batch information and defects of the substrate (data on the identification number of the polishing device 1000, processing conditions, film thickness outside the regulations, etc.) are transmitted.

The number of the signal lines 120 and the signal lines 122 can also be set to correspond to the number of the inspection device 106 and the polishing device 1000. In addition, the signal lines 120 and the signal lines 122 transmit data in a time-division multiplexing manner. The number of the signal lines 120 and the signal lines 122 can also be less than the total number of the inspection device 106 and the polishing device 1000.

From the host computer 126, through the signal line 128, the batch information about the substrate and the information about the defect (dust, damage, etc.) (dust, damage, etc., and whether it can be polished by grinding) are generated in the manufacturing process before the manufacturing process shown in FIG. The device 1000 corrects the defect) and sends it to the defect data receiving unit 112.

The defect data receiving unit 112 sends the received defect data to the defect database 124 via the signal line 130. The defect database 124 accumulates the defect data received from the defect data receiving unit 112 in each manufacturing line 104.

The defect data received by the defect data receiving unit 112 are as follows. The defect data receiving unit 112 receives batch information of the defective substrate from the inspection device group 110. When the defect information is fine particles (foreign matter), it includes data such as the number, size, and color of the fine particles; when it is a scratch (scratch), it includes scratches. The number, size, depth, location (distance from the center of the substrate) of scars (scars), etc. The defect data receiving unit 112 receives the lot number information of the defective substrate from the polishing device group 108, and the film thickness distribution or the maximum and minimum values of the film thickness as the defect information.

The defect data receiving unit 112 also receives the same data from the host computer 126 as the defect information of another process. That is, to receive batch information of defective substrates. When the defect information is particles (foreign matter), it includes data such as the number, size, and color of the particles; when it is a scratch (scar), it includes the number, size, and size of the scratch (scar). Depth, location (distance from the center of the substrate), etc.; and film thickness distribution or maximum and minimum film thickness.

The processing data receiving unit 114 receives processing data (batch information of the substrate, information of the polishing device 1000 that generated the defect (process data, alarm history, etc.) from the polishing device group 108 via the signal line 132 on the processing content of the substrate in the polishing device 1000 The history information of the consumables in the polishing device 1000, etc.)). The processing data receiving unit 114 sends the received processing data to the polishing database 136 via the signal line 134. The polishing database 136 accumulates the processing data received from the processing data receiving unit 114 in each manufacturing line 104.

The processing data received by the processing data receiving unit 114 is as follows. a) The information about the processing batch is the batch information of the substrate, and the contents of each processing plan (recipe) of conveying/polishing/cleaning. (Here, the so-called processing plan refers to the processing of the substrate, and the description of each part of the polishing device 1000 The assigned commands, procedures, settings, and parameters are in the form of lists, for example.) b) The process data is the conveying route, which can determine how to deal with the records of the plan allocated to the grinding/cleaning; c) together with all the time Record the occurrence/release of the alarm of each grinding device 1000, the on (NO)/disconnect (OFF) of the contact, or the record of all process data in the device at a specified time interval; d) To ensure that each grinding device The quality of 1000, the rise and fall time of various data in substrate processing, the data to judge the change of the processing result from the standard deviation of the process data; e) the use time of consumables, etc.

The defect database 124 and the polishing database 136 transmit the defect data and the processing data to the database update unit 142 via the signal line 138 and the signal line 140, respectively. The database update unit 142 accumulates the received defect data and processing data in the defect knowledge database 144.

The defect knowledge database 144 will be the defect data detected by the inspection device group 110 during the polishing process, and the defect data detected outside the polishing process, and the polishing device 1000 can deal with the range of defects and by polishing The processing data detected by the device group 108 is accumulated. The defects detected outside the polishing process are the defect data and processing data sent from the host computer 126 to the defect data receiving unit 112 via the signal line 128.

The defect knowledge database 144 has defect data and processing data; and data showing the corresponding relationship with the cause of the defect and the defect correction measure. As a result of the inspection, the correspondence between the newly added defect data and processing data, the occurrence cause of the defect and the defect correction measures are updated by the database update unit 142 as follows, and this is illustrated by FIGS. 5 to 10.

In order to obtain the corresponding relationship, the database update unit 142 may use any method such as decision tree, neural network, simulation, etc., to determine the cause of the defect and the defect correction measure. Figure 5 shows an example of using neural networks. Figures 6 to 10 show examples of using simulation. These are sample models, and the embodiment of the present invention is not limited to these.

Figure 5 shows a block diagram of the neural network 150 used to update the database. The neural network 150 is implemented as a computer program that executes AI (artificial intelligence). In addition, a part of the neural network 150 can also be implemented as hardware. The neural network 150 is an automatic learner, specifically, a machine learning, a deep learner as a machine learning. The neural network 150 inputs learning data (defect data and processing data), and the neural network 150 outputs output data (defect occurrence causes and defect correction measures).

The neural network 150 after learning by the neural network 150 is used in the actual polishing process. In the actual polishing process, the neural network 150 inputs the learning data made from the data obtained in the actual polishing process, and outputs the aforementioned output data (the cause of the defect and the defect correction measure).

FIG. 5 is a diagram showing a configuration example of the neural network 150. As shown in this figure, the neural network 150 has an input layer including neurons (input nodes) x1, x2, x3, ‧‧‧, xl that input input data (defect data and processing data). The neural network 150 has m neurons y11, y21, y31, ‧‧‧yml (hidden nodes) and p neurons y12, y22, y32, ‧‧‧, yp2 (hidden nodes) connecting input nodes and output nodes ) In the middle layer (hidden layer). Furthermore, the neural network 150 has an output layer containing n neurons z1, z2, z3, ‧‧‧ and zn (output nodes) that include the cause of the output defect and the defect correction measures. In addition, in this figure, the intermediate layer only shows two layers, but one layer or three or more intermediate layers may be provided.

The neural network 150 can also learn fault prediction according to mechanical learning (called deep learning or deep learning) of a multi-layer (4 or more) neural network (deep neural network), for example. This picture is a 4-layer neural network (deep neural network).

The neural network 150 learns the normal/abnormal grinding of the grinding unit. The neural network 150 learns the defect data and processing by the so-called "learning with teacher" according to the data set created based on the combination of the defect data obtained by the inspection device group 110 and the polishing device group 108 and the normal data and the processing data The data and the relevance of the reason for the defect and the corrective measures. Here, the so-called "learning with teacher" refers to the model of inferring the results from the input by assigning a large number of data sets of some inputs and results (labels) to the learning device, learning some features in these data sets, and obtaining inductive results from the inputs. That is, its connectedness.

In addition, the neural network can also only accumulate abnormal states, that is, only accumulate state variables when the grinding unit is operating normally, and learn the normal/abnormality of grinding through the so-called "learning without teacher". For example, when the abnormal frequency of the grinding unit is extremely low, the "learning without teacher" method is effective. Here, the so-called "learning without a teacher" means that only a large amount of input data is given to the neural network 150 to learn how the input data is distributed. Even if the corresponding teacher output data is not given, it still learns to compress, classify, and trim the input data. method. The similar characteristics of these data sets can be clustered (Clustering) and so on. Using this result, by setting some benchmarks and configuring the most suitable output, the normal/abnormal judgment can be realized.

In this figure, neurons x1, x2, x3, ‧‧‧ and xl input input data about particles, scratches, contour abnormalities, and machine abnormalities. This picture is for each input data about particles, scratches, contour abnormalities, and machine abnormalities. However, this is an example. You can also enter more than two input data about particles, scratches, contour abnormalities, and machine abnormalities. In addition, input data other than those related to particles, scratches, contour abnormalities, and machine abnormalities can also be input.

Neurons z1, z2, z3, ‧‧‧ and zn respectively output output data about the plan (that is, parameters, etc.), component replacement measures, and unit stop. This figure outputs one output data about the plan, component replacement measures, and unit stop, but this is an example, and more than two output data about the plan, component replacement measures, and unit stop can also be output. In addition, output data other than the output data about the plan, component replacement measures, and unit stop can also be output.

In addition, when new defect data and processing data are added, the number of rough defect types for input nodes and the number of nodes for correction measures for output nodes may not be increased. However, at this time, it is also possible to add or change the number of input nodes and/or output nodes from the host computer 126 and the control server (same row as the host computer 126) on the polishing device 1000.

Next, use FIGS. 6 to 9 to illustrate another method of using the above-mentioned neural network 150 and simulation to determine the cause of the defect and corrective measures. Figures 6 to 9 show block diagrams of the neural network and simulation model used to update the database. The procedures shown in FIGS. 6-9 are used to improve the defect correction measures and improve the accuracy of the neural network 150. The initial form of the neural network 150 and the simulation model g152 can also be prepared in advance, and can also be designated and/or updated from the host computer 126.

FIG. 6 is a diagram showing the simulation result of the polishing process according to the correction measures z1~zn obtained by the neural network 150 before the polishing after the correction is implemented. The simulation result is obtained by inputting the correction measures z1~zn into the simulation model g152.

Fig. 6 refers to the middle layer containing m neurons y11, y21, y31, ‧‧‧yml and p neurons y12, y22, y32, ‧‧‧ and yp2 of Fig. 5 as the correction measure generation model f154. In addition, the output layer containing n neurons z1, z2, z3, ‧‧‧ and zn in Fig. 5 is called correction measure 156. The simulation model g152 inputs corrective measures 156, and outputs the inspection data (or defect data) and processing data after grinding as the simulation result 158. The simulation model g152 can use multivariate input and multivariate output functions or neural networks.

The output as the simulation result 158 is the aforementioned defect data. In other words, it displays information about substrate defects. However, because the simulation result 158 is obtained as a result of corrective measures, the simulation result 158 is sometimes a normal value. The simulation results 158 in this figure output one simulation result 158 about particles, scratches, and contour abnormalities.

In addition, this figure outputs one simulation result 158 about particles, scratches, and contour abnormalities. However, this is an example, and two or more simulation results 158 about fine particles, scratches, and contour abnormalities may be output. In addition, simulation results 158 other than the simulation results 158 for particles, scratches, and contour abnormalities can also be output. The simulation results 158 about fine particles, scratches, and contour abnormalities are called predicted value C, predicted value D, and predicted value E, respectively. The predicted value C, the predicted value D, and the predicted value E all meet the specifications at this stage, and have values that meet the qualification criteria.

After the predicted value C, the predicted value D, and the predicted value E are obtained by simulation, polishing is actually performed. The simulation model g152 is improved by the actual measurement values obtained by actual polishing. The improved simulation model g152 is used as the simulation model g'1521. This is shown in Figure 7. When obtaining processing results C', processing results D', and processing results E'that are different from the predicted value C, predicted value D, and predicted value E, use the corrective measures x156 in Figure 6 to obtain processing results C', processing results D', The way of processing the result E'improves the simulation model g152. The method of obtaining the improved simulation model g'1521 can adopt the least square method to change the coefficient of the function, or to change the neural network by back propagation.

After obtaining the improved simulation model g'1521, by using the method shown in Figure 8, the simulation model g'1521 can be used to obtain qualified reference values C, D, E (that is, predicted value C, predicted value D, predicted value Value E) The way to improve and correct measures to generate model f154. This is illustrated by Figure 8.

First, calculate the corrective measures x'1561 that can obtain the qualified reference values C, D, and E using the improved simulation model g'1521. Calculate the corrective measures x'1561 using the simulation model g'1521 to obtain the qualified reference values C, D, E, and solve the equation when simulating the model g'1521 system function. The method to find the corrective measures x'1561 when the simulation model g'1521 is a neural network is to use virtual input data to repeat the learner from the simulation model g'1521 before obtaining the qualified reference values C, D, E. The learning of the neural network is to correct the weight and bias of the neurons of the neural network by the error back propagation method.

After obtaining the corrective measure x'1561, the corrective measure generation model f154 is improved by the method of obtaining the corrective measure x'1561, and the improved corrective measure generation model f'1541 is obtained. Use the corrective measure x'1561 to find the corrective measure generation model f'1541 that can obtain the corrective measure x'1561. Before the corrective measure x'1561 can be obtained, the corrective measure generation model f154 uses virtual input data and repeats As a learner of AI. The learning of the neural network is to correct the weight and bias of the neurons of the neural network by the error back propagation method. Therefore, a corrective measure generation model f'1541 that can generate corrective measure x'1561 is obtained.

Fig. 6 to Fig. 8 are an example of the method of generating the model f154 for the improvement and correction measures. Other methods can also be used to improve (learn). For example, in the method shown in Figure 8, instead of the qualified reference values C, D, and E, the ideal values C", D", E" and the simulation model g'1521 can be used to learn the corrective measure generation model f154, and Obtain corrective measures to generate model f” 1542. It is shown in Figure 9. The corrective measure x”1562 can be obtained by the corrective measure generation model f”1542.

In Figures 6-9, the neural network 150 and simulation are used to determine the cause of the defect and the corrective measures for the defect. In Figs. 6-9, it is also possible to replace the neural network 150 and use the simulation model 150' to find the corrective measures z1~zn. That is, the simulation model 150' and the simulation model g152 can also be used. The simulation result 158 is obtained by inputting the correction measures z1~zn obtained by the simulation model 150' into the simulation model g152.

Returning to FIG. 4, the defect occurrence cause determination unit 116 enters the defect knowledge database 144, determines the cause of the defect occurrence with respect to the defect data and processing data showing the newly added defect, and outputs it to the defect correction measure unit 118. The data output by the defect occurrence cause determination unit 116 is the lot information of the defective substrate, the identification information of the unit (also referred to as the "chamber") in the polishing device 1000 that generated the defect, and the polishing device 1000 that generated the defect. The reason for the defect (the part that generated the defect and/or the processing content), etc. Factors causing defects include machine failure and/or machine malfunctions and/or machine setting errors and/or machine handling improperly.

The defect correction measure unit 118 enters the defect knowledge database 144 to determine the defect correction measure for correcting the defect based on the input cause of the defect. The defect knowledge database 144 accumulates defect correction measures corresponding to the cause of the defect. Then, the defect correction measure part 118 indicates defect correction measures to at least one of the polishing unit 300, the cleaning unit 400, and the EFEM 200. The flaws to be corrected in this embodiment are flaws generated by the polishing device 1000 and flaws that the polishing device 1000 can handle.

The data output by the defect correction measure unit 118 is the identification information of the polishing device 1000, the unit identification information of the polishing device 1000, and the specific correction measures. Corrective measures include: a) the unit is stopped; b) the processing plan is changed (adjustment of various parameters, the conveying route change, etc.); c) the setting constant (parameter) of the grinding device is changed (in addition to the processing plan, there are also information about the grinding device Action parameters) and so on.

The specific correction measures corresponding to the cause of occurrence of the defect output by the defect correction measure unit 118 are as follows. 1. When there is a defect about fine particles In the inspection after cleaning by the cleaning unit 400 in the polishing apparatus 1000, when it is basically detected that the number of particles is increasing, corrective measures are taken. When the number of fine particles increases and the number of fine particles exceeds the reference value, the cleaning module in the cleaning unit 400 is stopped, and the conveying route that used the cleaning module cannot be used. Explain about this.

In normal cleaning, the cleaning module on the upper side and the cleaning module on the lower side are used for parallel cleaning and transportation in FIG. 3B. Then, the four grinding units in the grinding device 1000 use the two first grinding units 300A and the second grinding unit 300B on the right side of FIG. 1 in pairs, and the two third grinding units 300C and the fourth grinding unit 300D on the left side. The two polishing units on the right are combined with the cleaning module on the upper side, for example. At this time, the 2 grinding units on the left are combined with the cleaning module on the lower side.

The typical transportation route is as follows. Upper route: first grinding unit 300A→second grinding unit 300B→upper primary cleaning module 412A→upper secondary cleaning module 432A→upper drying module 452A. Lower route: third grinding unit 300C→fourth grinding unit 300D→lower primary cleaning module 412B→lower secondary cleaning module 432B→lower drying module 452B. And use the transportation plan to set which of these transportation routes to use? Among these representative transport routes, a cleaning module determined to be defective is stopped, and the transport route using the cleaning module cannot be used. An example is as follows. For example, when it is determined that there is an abnormality in the upper primary cleaning module 412A, a corrective measure of stopping the upper primary cleaning module 412A and using the lower route is adopted.

In addition, in these representative transport routes, when it is determined that one of the polishing units is defective, the polishing unit is stopped, and the transport route using the polishing unit cannot be used. The example is as follows. For example, when it is judged that there is an abnormality in the first polishing unit 300A, the first polishing unit 300A and the second polishing unit 300B are stopped, and the third polishing unit 300C and the fourth polishing unit 300D are used for polishing.

When it is detected that the number of particles is increasing, the following corrective measures shall be implemented. That is, when the amount of fine particles is found to increase in the inspection after cleaning of the cleaning unit 400, it is estimated that the cause of the defect is incomplete cleaning. (1) Change the cleaning plan to increase the cleaning time and/or the flow rate of the agent. (2) One or more correction measures that correct the deviation of the drug supply point (not dripping at the designated position on the wafer) are judged as defect correction measures. When taking (2) as a corrective measure, the maintenance person in charge can correct the medicine supply point, notify the incorrect medicine supply position alarm, stop the cleaning module, and instruct to use the transport route that does not pass through the module.

When it is determined by the inspection of the inspection device group 110 that the particle distribution map (data showing the distribution state of the particles) is generated due to the shape of the roller member, it is estimated that the roller member is the cause of the defect. Then, (1) replace the roller component; (2) adjust the self-cleaning condition of the roller component (adjust the cleaning time of the roller component and the load of the roller component pressing the self-cleaning plate); (3) replace the self-cleaning plate of the roller component ( Because it may cause anti-contamination from the self-cleaning plate to the roller member), one or more of them are judged as defect correction measures.

Here, the self-cleaning of the roller member refers to the cleaning process of the roller member itself. The self-cleaning of the roller member uses a self-cleaning plate to replace the wafer, and is performed by cleaning the roller member in the cleaning unit 400. The self-cleaning of the pen member described later is the cleaning process of the pen member itself, and the self-cleaning of the pen member is also performed in the same way.

When taking (1) and (3) as corrective measures, notify the replacement component alarm, stop the cleaning module, and instruct to use the transport route that does not pass through the module. When taking (2) as a corrective measure, instruct to change the cleaning time or pressing load and other self-cleaning parameters and execute self-cleaning. Here, the so-called roller member is a part used to clean the substrate WF. The roller component is installed on the rotating shaft in the cleaning module. The roller component is a sponge arranged on the cylinder (rotating shaft). In addition, the inspection devices included in the inspection device group 110 include a wafer surface inspection device, a defect inspection device, and the like as described above. These belong to the field of semiconductor manufacturing and are generally called inspection devices.

When it is determined by the inspection of the inspection device group 110 that the distribution map of the fine particles is caused by the shape of the pen member, the pen member is estimated to be the cause of the defect. Then, (1) replace the pen component; (2) adjust the self-cleaning condition of the pen component (adjust the cleaning time and pressing load of the pen component); (3) replace the self-cleaning plate of the pen component (because it may cause self-cleaning One or more of the anti-pollution of the board to the pen member) is judged as defect correction measures.

When taking (1) and (3) as corrective measures, notify the replacement component alarm, stop the cleaning module, and instruct to use the transport route that does not pass through the module. When taking (2) as a corrective measure, instruct to change the cleaning time or pressing load and other self-cleaning parameters and execute self-cleaning. Here, the so-called pen member is a part used to clean the substrate WF. The pen member is installed on the rotating shaft in the cleaning module. The pen member is a sponge arranged at the front end of the rotating shaft. In addition, the cleaning module equipped with the roller member and the cleaning module equipped with the pen member are usually separate cleaning modules.

After inspection by the inspection device group 110, although there is a tendency for particles to increase, but the cause cannot be determined, the following measures to determine the cause are taken as defect correction measures. That is, for all the polishing units 300 and the cleaning units 400 that process wafers WF in which particles tend to increase, each unit is transported so that only this unit is processed. After all the processed wafers are re-inspected by the inspection device group 110, the unit that becomes the source is determined. And instruct to perform the cutting test for this reason.

When the unit for which the defect occurs is determined through the inspection of the inspection device group 110, it is instructed to change the transportation plan, change to a route that does not pass through the unit, and stop the defect correction measure. 2. In case of flaws related to scratches

Basically, the unit with scratches is stopped, and the conveying route using the unit cannot be used. And take corrective measures to eliminate the cause.

After inspection by the inspection device, when the scratches are on the long shape of the wafer surface, the coagulation on the polishing table (the coagulation of the diamond and the slurry) on the polishing table is the cause of the defect, so the polishing unit 300 is cleaned The program is interrupted, the pad is cleaned (by the sprayer cleaning and dressing), and the correction measures for the defects of the coagulation on the washing table are indicated.

In the polishing unit 300, when the scratches occurred after replacing the slurry batch, it is possible that the coagulum solidified in the slurry supply piping was supplied to the device (that is, on the polishing table), and the coagulum was affected by the scratches. . Therefore, a long Dummy Dispense (that is, slurry is supplied, but polishing is not performed) is performed, and the inside of the pipe is thoroughly cleaned. At the same time, implement pad cleaning (cleaning and dressing by sprayer), and instruct to flush the coagulum on the grinding table to correct the defects.

When the grinding device 1000 continues to produce scratches after the idling state (the state where no grinding is performed) for a long time, it should be that the solidified material solidified by the slurry attached to the slurry supply nozzle or the upper annular turntable falls on the polishing table, because the Scratches occur under the influence of coagulum. At this time, perform long slurry nozzle cleaning and virtual deployment to thoroughly clean the inside of the piping. At the same time, implement pad cleaning (cleaning and dressing by sprayer), and instruct to flush the coagulum on the grinding table to correct the defects.

When the scratches occurred after changing the polishing plan, the reason should be that the water polishing (polishing using pure water only) performed before the end of the polishing, the pressure of pressing the wafer on the polishing pad was too strong, and the adhesion was not thoroughly washed. Wafer slurry. At this time, it is instructed to check the pressure of pressing the wafer against the polishing pad in the water polishing step of the polishing scheme, and to reduce the pressure to the range of the reference value allowed by the host computer 126 to correct defects.

If the cause cannot be determined, the following measures to determine the cause shall be taken as the defect correction measures. That is, regarding all the polishing units 300 that process defective wafers, each polishing unit 300 is transported so that only the polishing unit 300 processes. Then, all the used wafers are re-inspected by the inspection device group 110 to determine the polishing unit 300 that is the source of generation. And instruct to perform the cutting test for this reason. 3. When there are defects with insufficient grinding amount

When an abnormality (that is, a defect) is found through the inspection of the film thickness measuring device mounted in the polishing device 1000, the polishing status information on the wafer is not notified of the completion of the processing, so the rework processing (ie , Grinding and subsequent cleaning treatment). This is a heavy-duty process that does not exchange information with the host computer 126, that is, does not accept the control of the host computer 126, and the polishing device 1000 judges separately and implements the heavy processing. In addition, changing the polishing plan along with the heavy work process is based on the heavy work plan received from the host computer 126 in advance.

In the polishing process where the output from the end point detector that detects the end point of the polishing is used to determine the time for the end of the polishing, when the amount of polishing is insufficient, the over process time is extended as a defect correction measure. Here, the term “exceeding processing time” refers to the time required for the extra polishing offset after the actual detection of the characteristic point (that is, the polishing end point). In addition, the opposite is true. That is, when the amount of polishing is too large, shorten the polishing time as a defect correction measure.

After the rework process, if the inspection result of the film thickness measuring device is abnormal, the control unit 146 stops the polishing unit 300 in use. The control unit 146 prohibits the use of the conveying route using the polishing unit 300.

When the abnormality is determined by the inspection of the film thickness measuring device outside the polishing device 1000, the host computer 126 requests the polishing device 1000 to reprocess as a new process (plan). The grinding device 1000 performs processing according to this scheme. 4. When there are defects with abnormal film thickness

The coping method at this time is the same as the insufficient grinding amount shown in item 3. That is, the heavy-duty processing (that is, re-grinding and subsequent cleaning processing) and other processing are performed in the same manner as in item 3. 5. Although a defect is found but no corrective measures are needed

In the inspection after cleaning with the cleaning unit 400, although the defect is caused by fine particles, if it is caused by the user of the polishing device 1000 changing the batch of the polishing agent or slurry, and the adhesion in the piping outside the polishing device 1000 , No need to change the cleaning plan. However, the host computer 126 is sent information indicating that the user's change may be the source after the cleaning in the piping is performed (or before the cleaning is performed). The reason for sending to the host computer 126 is to stop the production as soon as possible. In this way, polishing devices made by manufacturers other than the polishing device 1000 are also included, which can increase the yield rate of the entire factory.

That is, the reason is that when the batches of chemicals used in the factory are switched, even the batches processed by other brands of grinding devices may still have the same flaws. Because the coagulum of the slurry stuck in the pipe outside the polishing device 1000, the newly used slurry and the medicine itself may be inappropriate, it is necessary to send information to the host computer 126 indicating that the user's change may be the source of the occurrence. And the whole factory will handle it.

After the corrective measures are taken, the control unit 146 measures the effects of the corrective measures. That is, the control unit 146 sends an instruction to the defect data receiving unit 112 to receive the measure result data of the polishing device 1000 (the inspection data received from the inspection device group 110 and the processing data received from the polishing device 1000).

Then, an instruction is sent to the database update unit 142 via the signal line 148 to receive the measure result data received regarding the result of the defect correction measure, and change the defect correction measure based on the measure result data. The database update unit 142 enters the defect knowledge database 144, and updates the defect knowledge database 144 when necessary, as explained in step e).

The foregoing description has been given of examples of embodiments of the present invention, but the above-mentioned embodiments of the present invention are intended to facilitate the understanding of the present inventors and do not limit the present inventors. The present invention can be changed and improved without departing from the scope of the gist thereof, and the equivalents are naturally included in the present invention. In addition, as long as at least a part of the above-mentioned problems can be solved or at least part of the effect can be achieved, the various elements described in the scope of patent application and the specification can be combined or omitted arbitrarily.

100: frame 101: Grinding device system 102: Control System 104-1~104-N: Manufacturing line 106: check device 108: Grinding device group 110: Inspection device group 112: Defective data receiving department 114: Processing data receiving department 116: Defects cause determination department 118: Defect Correction Measures Department 120, 122, 128, 130, 132, 134, 138, 140, 148: signal line 124: Defect Database 126: main computer 136: Grinding Database 142: Database Update Department 144: Defect Knowledge Database 146: Control Department 150: Neural Network g152: Simulation model f154: Corrective measures generation model 156: Corrective Measures 158: Simulation results 200: EFEM 220: Front loading section 222: Box 230: track 240: transport robot 300: Grinding unit 300A: The first grinding unit 300B: The second grinding unit 300C: The third grinding unit 300D: Fourth grinding unit 310A: polishing pad 320A: Grinding table 322A: Workbench shaft 330A: upper ring turntable 332A: Upper circular turntable shaft 340A: Slurry supply nozzle 350A: Dresser 360A: sprayer 370: Lift 372: The first linear conveyor 374: Swing conveyor 376: The second linear conveyor 378: Temporary Set 400: Cleaning unit 410: First cleaning room 412A: Primary cleaning module on the upper side 412B: Primary cleaning module on the lower side 420: first transfer room 422: The first transport robot 424: Support shaft 430: second cleaning room 432A: Upper secondary cleaning module 432B: Lower side secondary cleaning module 434: Temporary Set 440: Second transfer room 442: The second transport robot 444: Support shaft 450: drying room 452A: Upper side drying module 452B: Lower side drying module 454A, 454B: fan filter unit 456A, 456B: Drying unit 500: control unit 1000: Grinding device TP1: The first transfer position TP2: second transfer position TP3: Third transfer position TP4: Fourth transfer position TP5: Fifth transfer position TP6: The sixth transfer position TP7: seventh transfer position WF: substrate

Figure 1 is a top view of the polishing device. Figure 2 is a perspective view of the first grinding unit. Figure 3A is a top view of the cleaning unit. Figure 3B is a side view of the cleaning unit. Figure 4 is a block diagram of the polishing device system. Figure 5 shows a block diagram of the neural network used to update the database. Figure 6 is a block diagram showing the neural network and simulation model used to update the database. Figure 7 is a block diagram showing the neural network and simulation model used to update the database. Figure 8 is a block diagram showing the neural network and simulation model used to update the database. Figure 9 is a block diagram showing the neural network and simulation model used to update the database.

101: Grinding device system

102: Control System

104-1~104-N: Manufacturing line

106: check device

108: Grinding device group

110: Inspection device group

112: Defective data receiving department

114: Processing data receiving department

116: Defects cause determination department

118: Defect Correction Measures Department

120, 122, 128, 130, 132, 134, 138, 140, 148: signal line

124: Defect Database

126: main computer

136: Grinding Database

142: Database Update Department

144: Defect Knowledge Database

146: Control Department

1000: Grinding device

Claims (8)

A control system is used to control a polishing device for polishing a substrate, It is characterized in that: the aforementioned grinding device has: Grinding part, which is constructed by grinding the substrate; The cleaning part is constructed by cleaning the aforementioned substrate after polishing; and A conveying part, which is configured to convey the substrate between the polishing part and the cleaning part; The aforementioned control system has: Defect data receiving section, which is constituted by receiving the defect data about the defects of the aforementioned substrate; The processing data receiving section is constructed to receive processing data on the processing content of the substrate in the polishing device; The defect cause determination unit is constituted by the method of determining the cause of the aforementioned defect based on the aforementioned defect data and the aforementioned processing data; and Defect Correction Measures Department, which determines the defect correction measures to correct the aforementioned defects based on the identified cause of occurrence, and instructs at least one of the aforementioned polishing part, the aforementioned cleaning part, and the aforementioned conveying part to correct the aforementioned defects Mode composition. The control system according to claim 1, wherein there is a measure result data processing unit, which receives the correction of the defect after at least one of the polishing unit, the cleaning unit, and the conveying unit has performed the defect correction measure The measure result data of the result of the measure may be based on the aforementioned measure result data to change the method of the aforementioned defect correction measures. The control system according to claim 1 or 2, which has a defect knowledge database, which is constructed by accumulating at least one of the aforementioned defect data and the aforementioned processing data. A polishing device characterized by having: the control system according to any one of claims 1 to 3, the polishing part, the cleaning part, and the conveying part. A polishing device system, characterized by having: the control system described in any one of claims 1 to 3, and a plurality of the aforementioned polishing devices. A polishing method using a polishing device having a polishing section for polishing a substrate, a cleaning section for cleaning the substrate after polishing, and a transporting section for transporting the substrate between the polishing section and the cleaning section, to polish the substrate, Its characteristics are: Receive defect information about the aforementioned defects of the substrate, Receive processing data on the processing content of the aforementioned substrate in the aforementioned polishing device, Determine the cause of the aforementioned defects based on the aforementioned defect data and the aforementioned processing data, Determine the defect correction measures for correcting the aforementioned defects based on the confirmed cause of occurrence, and indicate the aforementioned defect correction measures to at least one of the aforementioned polishing part, the aforementioned cleaning part, and the aforementioned conveying part. The polishing method according to claim 6, wherein after at least one of the polishing part, the cleaning part, and the conveying part has performed the defect correction measure, it receives the measure result data about the result of the defect correction measure and bases it on The results of the aforementioned measures have changed the aforementioned defect correction measures. The polishing method according to claim 6 or 7, wherein at least one of the aforementioned defect data and the aforementioned processing data is accumulated in a defect knowledge database.

Images