Substrate processing apparatus and information processing system

Abstract

To provide a substrate processing device for estimating whether a substrate after substrate processing is a defective article without inspecting the substrate by a defect inspection device, and an information processing system.SOLUTION: A substrate processing device includes: at least one sensor for detecting a physical amount of an object during substrate polishing and / or washing and / or drying; a conversion part for converting a sensor value during polishing and / or washing and / or drying to be detected by the sensor into a feature amount in each processing step with respect to a learned machine learning model; and an inference part for outputting at least one prediction value among the number of defects of a substrate of an object, the sizes of the defect positions of the defects by inputting object data including the feature amount into the model. The machine learning model regards a sensor value detected in a production line of the object or the same kind of a production line as input data including the feature amount converted in each processing step, and performs learning by using a learning data set obtained by regarding at least one of the number of defects of the substrate, the sizes of the defects, and the positions of the defects as output data.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to a substrate processing apparatus and an information processing system.

Background Art

[0002] The cleaning performance of a substrate processing apparatus (for example, a polishing apparatus) is evaluated by measuring, with a dedicated defect inspection apparatus, a substrate (specifically, a wafer) discharged from the apparatus after processes such as polishing, cleaning, and drying (see, for example, Patent Document 1). Since defect inspection requires cost (mainly time), it is difficult to perform a full inspection at the manufacturing site after substrate processing (for example, polishing, cleaning, drying), and sampling inspection is carried out.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, there is also a possibility that defective products may occur due to insufficient cleaning or the like in the substrates (specifically, wafers) that have not been inspected.

[0005] The present invention has been made in view of the above problems, and an object thereof is to provide a substrate processing apparatus and an information processing system that can estimate whether a substrate after substrate processing is a defective product without inspecting it with a defect inspection apparatus.

Means for Solving the Problems

[0006] A substrate processing apparatus according to an aspect of the present invention includes at least one sensor that detects a target physical quantity during polishing and / or cleaning and / or drying of a substrate, and a conversion unit that converts sensor values during polishing and / or cleaning and / or drying detected by the sensor into feature quantities for each processing step with respect to a learned machine learning model, and an inference unit that outputs at least one predicted value among the number of defects, the size of the defects, and the position of the defects in the target substrate by inputting target data including the feature quantities into the learned machine learning model, wherein the learned machine learning model is input data including feature quantities obtained by converting sensor values during polishing and / or cleaning detected by a sensor in a target production line or a production line of the same type as the target production line for each processing step, and is learned using a learning data set having at least one of the number of defects, the size of the defects, and the position of the defects in the substrate as output data.

[0007] According to this configuration, since at least one predicted value among the number of defects, the size of the defects, and the position of the defects in the substrate after substrate processing can be obtained without inspecting with a defect inspection apparatus, it is possible to estimate whether the substrate after substrate processing is a defective product without inspecting with a defect inspection apparatus.

[0008] Further, in the above substrate processing apparatus, the input data during learning of the machine learning model further includes the residence time counted for each unit included in the substrate processing apparatus and staying in the unit, and each unit included in the substrate processing apparatus further includes a unit residence time counting unit that counts the residence time staying in the unit, and the target data input to the learned machine learning model may further include the residence time counted for each unit by the unit residence time counting unit and staying in the unit.

[0009] In addition, in the above substrate processing apparatus, the input data during the learning of the machine learning model further includes a second feature quantity obtained by converting the position of a member used for polishing or cleaning. The conversion unit converts the position of the member used for polishing or cleaning into the second feature quantity, and the target data input to the learned machine learning model may further include the second feature quantity for each member converted by the conversion unit.

[0010] In addition, in the above substrate processing apparatus, the input data during the learning of the machine learning model further includes recipe information including command values for units included in the substrate processing apparatus, and the target data input to the learned machine learning model may further include recipe information including command values for units included in the substrate processing apparatus.

[0011] In addition, in the above substrate processing apparatus, a regression analysis unit that outputs a correlation parameter representing a correlation between each of a plurality of sensor values and any one of the number of defects, the size of the defects, and the position of the defects in the substrate according to a predetermined regression analysis algorithm, a reception unit that receives at least one sensor that outputs a sensor value serving as a basis for a feature quantity included in the input data of the machine learning model, and a learning unit that retrains the machine learning model with the feature quantity obtained by converting the sensor value of the received sensor may be provided. The inference unit may output the predicted value using the machine learning model retrained by the learning unit.

[0012] An information processing system according to another aspect of the present invention includes a conversion unit that converts sensor values detected by sensors included in a substrate processing apparatus during polishing and / or cleaning and / or drying of a substrate into feature amounts for each processing step for a learned machine learning model, and an inference unit that outputs at least one predicted value of the number of defects, the size of the defects, and the position of the defects in the substrate by inputting target data including the feature amounts. The learned machine learning model is learned using an input data set including feature amounts obtained by converting sensor values detected by sensors during polishing and / or cleaning in a target production line or a production line of the same type as the target production line for each processing step, and output data including at least one of the number of defects, the size of the defects, and the position of the defects in the substrate.

Effects of the Invention

[0013] According to one aspect of the present invention, since at least one predicted value of the number of defects, the size of the defects, and the position of the defects in the substrate after substrate processing can be obtained without inspecting with a defect inspection apparatus, it is possible to estimate whether the substrate after substrate processing is a defective product without inspecting with a defect inspection apparatus.

Brief Description of the Drawings

[0014]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6A

Figure 6B

Figure 7

Figure 8A

Figure 8B

Figure 9A

Figure 9B

Figure 10A

Figure 10B

Figure 11A

Figure 11B

Figure 12A

Figure 12B

Mode for Carrying Out the Invention

[0015] Hereinafter, each embodiment will be described with reference to the drawings. However, a more detailed description than necessary may be omitted. For example, detailed descriptions of well-known matters and duplicate descriptions of substantially the same configurations may be omitted. This is to avoid making the following description unnecessarily redundant and to facilitate the understanding of those skilled in the art. In the present embodiment, as an example of a substrate processing apparatus, a polishing apparatus for chemically and mechanically polishing (CMP: Chemical Mechanical Polishing) the surface of a substrate W flat will be described.

[0016] FIG. 1 is a diagram showing an example of a schematic configuration of a substrate processing apparatus according to the present embodiment. As shown in FIG. 1, the substrate processing apparatus 1 is a polishing apparatus that chemically and mechanically polishes (CMP: Chemical Mechanical Polishing) the surface of a substrate W such as a silicon wafer (hereinafter also simply referred to as a wafer) flat.

[0017] As shown in FIG. 1, the substrate processing apparatus 1 includes, for example, a rectangular box-shaped housing 2. The housing 2 is formed in a substantially rectangular shape in plan view. The housing 2 includes a substrate transfer path 3 extending in the longitudinal direction at its center. A load / unload unit 10 is disposed at one end in the longitudinal direction of the substrate transfer path 3. A polishing unit 20 is disposed on one side in the width direction of the substrate transfer path 3, that is, the direction orthogonal to the longitudinal direction in plan view, and a cleaning unit 30 is disposed on the other side. A substrate transfer unit 40 for transferring the substrate W is provided in the substrate transfer path 3. Further, the substrate processing apparatus 1 includes a control unit 50 that comprehensively controls the operations of the load / unload unit 10, the polishing unit 20, the cleaning unit 30, and the substrate transfer unit 40.

[0018] The load / unload unit 10 includes a front load unit 11 for accommodating the substrate W. A plurality of front load units 11 are provided on one side surface in the longitudinal direction of the housing 2. The plurality of front load units 11 are arranged in the width direction of the housing 2. The front load unit 11 is mounted with, for example, an open cassette, a SMIF (Standard Manufacturing Interface) pod, or a FOUP (Front Opening Unified Pod). Both SMIF and FOUP are sealed containers in which a cassette of the substrate W is stored inside and covered with a partition wall, and can maintain an environment independent of the external space.

[0019] In addition, the load / unload unit 10 includes two transfer robots 12 for loading and unloading the substrate W from and to the front load unit 11, and a traveling mechanism 13 for moving each transfer robot 12 along the arrangement of the front load unit 11. Each transfer robot 12 is provided with two hands vertically, and these two hands are used separately before and after the processing of the substrate W. For example, when each transfer robot 12 returns the substrate W to the front load unit 11, the upper hand is used, and when taking out the substrate W before processing from the front load unit 11, the lower hand is used.

[0020] The polishing unit 20 includes a plurality of polishing devices 21 (21A, 21B, 21C, 21D) for polishing the substrate W. The plurality of polishing devices 21 are arranged in the longitudinal direction of the substrate transfer path 3. The polishing device 21 includes a polishing table 23 that rotates a polishing pad 22 having a polishing surface, a top ring 24 for holding the substrate W and pressing the substrate W against the polishing pad 22 on the polishing table 23 for polishing, a polishing liquid supply nozzle 25 for supplying a polishing liquid, a dressing liquid, etc. to the polishing pad 22, a dresser 26 for dressing the polishing surface of the polishing pad 22, and an atomizer 27 for spraying a mixed fluid of a liquid such as pure water and a gas such as nitrogen gas or a liquid such as pure water in a mist form onto the polishing surface.

[0021] While supplying the polishing liquid from the polishing liquid supply nozzle 25 onto the polishing pad 22, the polishing device 21 presses the substrate W against the polishing pad 22 by the top ring 24, and moves the top ring 24 and the polishing table 23 relative to each other to polish the substrate W and flatten the surface of the substrate W. In addition, the top ring 24 is provided with a plurality of pressure increase / decrease areas (for example, air bags) arranged concentrically. The top ring 24 adjusts the pressing condition of the substrate W against the polishing pad 22 by adjusting the pressure in these plurality of pressure increase / decrease areas.

[0022] The dresser 26 has hard particles such as diamond particles or ceramic particles fixed to the rotating part at the tip that contacts the polishing pad 22, and by rotating and swinging the rotating part, the entire polishing surface of the polishing pad 22 is uniformly dressed to form a flat polishing surface.

[0023] The atomizer 27 performs purification of the polishing surface and dressing work of the polishing surface by the dresser 26, that is, regeneration of the polishing surface, by washing away polishing debris, abrasive grains, etc. remaining on the polishing surface of the polishing pad 22 with a high-pressure fluid.

[0024] The cleaning unit 30 includes a plurality of cleaning devices 31 (31A, 31B) for cleaning the substrate W and a substrate drying device 32 for drying the cleaned substrate W. The plurality of cleaning devices 31 and the substrate drying device 32 are arranged in the longitudinal direction of the substrate conveyance path 3. A first conveyance chamber 33 is provided between the cleaning device 31A and the cleaning device 31B. In the first conveyance chamber 33, a conveyance robot 35 for conveying the substrate W between the substrate conveyance unit 40, the cleaning device 31A, and the cleaning device 31B is provided.

[0025] The conveyance robot 35 has two hands vertically, and these two hands are used separately before and after cleaning of the substrate W in the cleaning device 31A. For example, when the conveyance robot 35 takes out the substrate W before cleaning from a temporary placement table (also referred to as a wafer station) 47 to be described later and conveys it to the cleaning device 31A, the lower hand is used, and when taking out the substrate W from the cleaning device 31A after cleaning and conveying it to the cleaning device 31, the upper hand is used.

[0026] Also, a second conveyance chamber 34 is provided between the cleaning device 31B and the substrate drying device 32. In the second conveyance chamber 34, a conveyance robot 36 for conveying the substrate W between the cleaning device 31B and the substrate drying device 32 is provided.

[0027] The cleaning device 31 includes a roll sponge (hereinafter also referred to as a roll) type cleaning module, and uses this cleaning module to clean the substrate W. Note that the cleaning devices 31A and 31B may be of the same type or different types of cleaning modules. Further, instead of the roll sponge type cleaning module, the cleaning devices 31A and 31B may include, for example, a pen sponge (hereinafter also referred to as a pen) type cleaning module or a two-fluid jet type cleaning module. Here, in the case of the two-fluid jet type cleaning module, the two fluids are, for example, a mixture of nitrogen (N2) and pure water.

[0028] The substrate drying device 32 includes a drying module that performs, for example, nitrogen gas (N2) drying or rotagoni drying using isopropyl alcohol (IPA: Iso-Propyl Alcohol). After the rotagoni drying is performed on the substrate W, the shutter 1a provided in the partition wall between the substrate drying device 32 and the load / unload unit 10 is opened, and the substrate W is unloaded from the substrate drying device 32 by the transfer robot 12.

[0029] The substrate transfer unit 40 includes a lifter 41, a first linear transporter 42, a second linear transporter 43, and a swing transporter 44. In the substrate transfer path 3, a first transfer position TP1, a second transfer position TP2, a third transfer position TP3, a fourth transfer position TP4, a fifth transfer position TP5, a sixth transfer position TP6, and a seventh transfer position TP7 are set in order from the load / unload unit 10 side.

[0030] The lifter 41 is a mechanism that vertically transfers the substrate W at the first transfer position TP1. The lifter 41 receives the substrate W from the transfer robot 12 of the load / unload unit 10 at the first transfer position TP1. Then, the lifter 41 delivers the substrate W received from the transfer robot 12 to the first linear transporter 42. A shutter 1b is provided in the partition wall between the first transfer position TP1 and the load / unload unit 10, and the shutter 1b is opened during the transfer of the substrate W, and the substrate W is delivered from the transfer robot 12 to the lifter 41.

[0031] The first linear transporter 42 is a mechanism for transporting the substrate W between two of the first transport position TP1, the second transport position TP2, the third transport position TP3, and the fourth transport position TP4. The first linear transporter 42 includes a plurality of transport hands 45 (45A, 45B, 45C, 45D) and a linear guide mechanism 46 that horizontally moves each transport hand 45 at a plurality of heights.

[0032] The transport hand 45A moves between the first transport position TP1 and the fourth transport position TP4 by the linear guide mechanism 46. The transport hand 45A is a pass hand for receiving the substrate W from the lifter 41 and delivering the substrate W to the second linear transporter 43.

[0033] The transport hand 45B moves between the first transport position TP1 and the second transport position TP2 by the linear guide mechanism 46. The transport hand 45B receives the substrate W from the lifter 41 at the first transport position TP1 and delivers the substrate W to the polishing device 21A at the second transport position TP2. The transport hand 45B is provided with an elevating drive unit that rises when delivering the substrate W to the top ring 24 of the polishing device 21A and descends after delivering the substrate W to the top ring 24. Note that the transport hands 45C and 45D are also provided with similar elevating drive units.

[0034] The transport hand 45C moves between the first transport position TP1 and the third transport position TP3 by the linear guide mechanism 46. The transport hand 45C receives the substrate W from the lifter 41 at the first transport position TP1 and delivers the substrate W to the polishing device 21B at the third transport position TP3. Further, the transport hand 45C also functions as an access hand for receiving the substrate W from the top ring 24 of the polishing device 21A at the second transport position TP2 and delivering the substrate W to the polishing device 21B at the third transport position TP3.

[0035] The transfer hand 45D moves between the second transfer position TP2 and the fourth transfer position TP4 by means of a linear guide mechanism 46. The transfer hand 45D functions as an access hand for receiving the substrate W from the top ring 24 of the polishing device 21A or the polishing device 21B at the second transfer position TP2 or the third transfer position TP3 and delivering the substrate W to the swing transporter 44 at the fourth transfer position TP4.

[0036] The swing transporter 44 has a hand that can move between the fourth transfer position TP4 and the fifth transfer position TP5, and delivers the substrate W from the first linear transporter 42 to the second linear transporter 43. Further, the swing transporter 44 delivers the substrate W polished in the polishing unit 20 to the cleaning unit 30. A temporary placement table 47 for the substrate W is provided on the side of the swing transporter 44. The swing transporter 44 turns the substrate W received at the fourth transfer position TP4 or the fifth transfer position TP5 upside down and places it on the temporary placement table 47. The substrate W placed on the temporary placement table 47 is transported to the first transfer chamber 33 by the transfer robot 35 of the cleaning unit 30.

[0037] The second linear transporter 43 is a mechanism for transporting the substrate W between two of the fifth transfer position TP5, the sixth transfer position TP6, and the seventh transfer position TP7. The second linear transporter 43 includes a plurality of transfer hands 48 (48A, 48B, 48C) and a linear guide mechanism 49 that moves each transfer hand 45 horizontally at a plurality of heights. The transfer hand 48A moves between the fifth transfer position TP5 and the sixth transfer position TP6 by means of the linear guide mechanism 49. The transfer hand 45A functions as an access hand for receiving the substrate W from the swing transporter 44 and delivering the substrate W to the polishing device 21C.

[0038] The transfer hand 48B moves between the sixth transfer position TP6 and the seventh transfer position TP7. The transfer hand 48B functions as an access hand for receiving the substrate W from the polishing apparatus 21C and delivering the substrate W to the polishing apparatus 21D. The transfer hand 48C moves between the seventh transfer position TP7 and the fifth transfer position TP5. The transfer hand 48C functions as an access hand for receiving the substrate W from the top ring 24 of the polishing apparatus 21C or the polishing apparatus 21D at the sixth transfer position TP6 or the seventh transfer position TP7 and delivering the substrate W to the swing transporter 44 at the fifth transfer position TP5. Although the description is omitted, the operation of the transfer hand 48 when delivering the substrate W is the same as the operation of the first linear transporter 42 described above.

[0039] The substrate processing apparatus 1 includes at least one sensor (not shown) that detects a target physical quantity during polishing and / or cleaning and / or drying of a substrate (e.g., a wafer). The target physical quantity is, for example, as follows when polishing. · Rotational speed and / or torque of the polishing table 23 · Rotational speed and / or torque of the top ring 24 · Airbag pressure of the top ring 24 · Rotational speed and / or load of the dresser 26 · Flow rate of slurry / water · Flow rate of the atomizer 27 · Flow rate of nitrogen (N2) in the atomizer 27

[0040] The target physical quantity is, for example, as follows when cleaning. · Rotational speed and / or torque and / or load of the roll · Rotational speed and / or torque and / or load of the pen · Flow rate of chemical solution / water · Rotational speed of the wafer · Flow rate of nitrogen (N2)

[0041] The target physical quantity is, for example, as follows when drying. · Flow rate of nitrogen (N2) · Flow rate of isopropyl alcohol (IPA)

[0042] As an example of the above sensor, the substrate processing apparatus 1 may include a sensor for detecting the table rotation speed and / or torque, a sensor for detecting the top ring rotation speed and / or torque, a sensor for detecting the top ring airbag pressure, a sensor for detecting the dresser rotation speed and / or load, a sensor for detecting the slurry / water flow rate, and a sensor for detecting the slurry / water flow rate. The substrate processing apparatus 1 may also include a sensor for detecting the roll rotation speed and / or torque and / or load of the cleaning apparatus 31, a sensor for detecting the pen rotation speed and / or torque and / or load of the cleaning apparatus 31, a sensor for detecting the chemical solution / water flow rate of the cleaning apparatus 31, and a sensor for detecting the wafer rotation speed of the cleaning apparatus 31. The substrate processing apparatus 1 may also include a sensor for detecting the flow rate of nitrogen (N2) and a sensor for detecting the flow rate of isopropyl alcohol (IPA).

[0043] Hereinafter, each member constituting the load / unload unit 10, the polishing unit 20, the cleaning unit 30, and the substrate transfer unit 40 is referred to as a unit.

[0044] FIG. 2 is a block diagram showing an example of the schematic configuration of the control unit according to the present embodiment. As shown in FIG. 2, the control unit 50 includes a unit control unit 51, a processor 6, and a storage unit 7. The unit control unit 51 comprehensively controls the operations of the units of the load / unload unit 10, the polishing unit 20, the cleaning unit 30, and the substrate transfer unit 40.

[0045] The learning model 71 that has been trained is stored in the memory unit 7. The trained machine learning model 71 uses, as input data, feature quantities obtained by converting sensor values during polishing and / or cleaning detected by sensors at each processing step in a target production line or a production line of the same type as the target production line, and is trained using a learning dataset that outputs, as output data, at least one of the number of defects, the size of defects, and the positions of defects in the substrate. In the present embodiment, as an example, the output data of the learning dataset will be described as the number of defects in the substrate. Here, the feature quantities are the average, maximum, minimum, sum, median, standard deviation, variance, kurtosis, or skewness of the time-series values of the sensors, or the average, maximum, minimum, sum, median, standard deviation, variance, kurtosis, or skewness of the time-series data of the differential values of the sensor values.

[0046] The processor 6 functions as a unit residence time counter 61, a conversion unit 62, a learning unit 63, an inference unit 64, a regression analysis unit 65, and a reception unit 66 by reading and executing a predetermined program from the memory unit 7.

[0047] The conversion unit 62 converts the sensor values during polishing and / or cleaning and / or drying detected by the sensors into feature quantities at each processing step for the trained machine learning model.

[0048] The learning unit 63 trains the machine learning model 71 using, as input data, feature quantities obtained by converting sensor values during polishing and / or cleaning detected by sensors at each processing step in a target production line or a production line of the same type as the target production line, and using a learning dataset that outputs, as output data, at least one of the number of defects, the size of defects, and the positions of defects in the substrate.

[0049] The inference unit 64 outputs a predicted value of at least one of the number of defects, the size of defects, and the positions of defects in the target substrate by inputting the target data including the feature quantities into the trained machine learning model.

[0050] The regression analysis unit 65 outputs a correlation parameter (e.g., a correlation coefficient) representing the correlation between each of a plurality of sensor values and any one of the number of defects, the size of the defects, and the position of the defects in the substrate according to a predetermined regression analysis algorithm. Here, this regression analysis algorithm may be LASSO (least absolute shrinkage and selection operator) regression, Ridge regression, Support Vector Regression (SVR), random forest regression (RFR), or Light GBM. The regression analysis unit 65 may display this correlation parameter (e.g., a correlation coefficient) on the display device 8. Thereby, the operator or user of the substrate processing apparatus 1 can confirm the correlation parameter (e.g., a correlation coefficient).

[0051] The reception unit 66 receives, for example, at least one sensor that outputs a sensor value that is the basis of a feature amount included in the input data of the machine learning model from an operator or user who has confirmed the output correlation parameter (e.g., a correlation coefficient). At this time, the operator or user inputs or selects, for example, a sensor that outputs a sensor value that is the basis of a highly correlated feature amount. Then, the learning unit 63 re-learns the previous machine learning model 71 with the feature amount obtained by converting the sensor value of the received sensor. The inference unit 64 outputs the prediction value using the machine learning model re-learned by the learning unit 63. According to this configuration, by the reception unit 66 receiving, for example, a sensor that outputs a sensor value that is the basis of a highly correlated feature amount from an operator or user, the prediction accuracy after re-learning of the machine learning model 71 can be improved.

[0052] In this embodiment, as an example, the input data during the learning of the machine learning model further includes the residence time counted for each unit included in the substrate processing apparatus and staying in the unit. On this premise, the unit residence time counting unit 61 counts the residence time staying in the unit for each unit included in the substrate processing apparatus 1. In this embodiment, as an example, the target data input to the learned machine learning model further includes the residence time staying in the unit counted for each unit by the unit residence time counting unit.

[0053] In this embodiment, as an example, the input data during the learning of the machine learning model further includes a second feature quantity obtained by converting the position of a member used for polishing or cleaning. On this premise, the conversion unit 62 converts the position of the member used for polishing or cleaning into the second feature quantity. In this case, the target data input to the learned machine learning model further includes the second feature quantity for each member converted by the conversion unit 62.

[0054] In this embodiment, as an example, the input data during the learning of the machine learning model further includes recipe information including command values for units included in the substrate processing apparatus. This command value is a set value of the above-described target physical quantity (for example, the rotation speed of the polishing table 23, etc.), and each set value of the target physical quantity is one value for each processing step. On this premise, the target data input to the learned machine learning model further includes recipe information including command values for units included in the substrate processing apparatus.

[0055] Next, the residence time of the substrate in each unit will be described with reference to FIG. 3. FIG. 3 is a graph showing an example of the residence time in each unit according to the present embodiment. In the graph of FIG. 3, the vertical axis represents the processing step number, and the horizontal axis represents time. FIG. 3 shows the residence time of each unit in the following process. That is, the transfer robot 12 delivers the substrate to the lifter 41, and then the transfer hand 45A of the first linear transporter 42 takes out the substrate from the lifter 41, and the transfer hand 45A delivers the substrate W to the polishing device 21A at the second transfer position TP2. After polishing by the polishing device 21A, the transfer hand 45D of the first linear transporter 42 receives the substrate from the polishing device 21A and delivers the substrate W to the swing transporter 44 at the fourth transfer position TP4. The swing transporter 44 turns the received substrate upside down and places it on the temporary placement table 47. Then, the lower hand of the transfer robot 35 receives the substrate W and transports it to the cleaning device 31A. After cleaning by the cleaning device 31A, the upper hand of the transfer robot 35 takes out the substrate W from the cleaning device 31A and transports it to the cleaning device 31B. After cleaning by the cleaning device 31B, the hand of the transfer robot 36 takes out the substrate W from the cleaning device 31B and transports it to the substrate drying device 32. After drying by the substrate drying device 32, the transfer robot 12 takes out the substrate from the substrate drying device 32.

[0056] In FIG. 3, TRBDs is the residence time of the substrate in the hand of the transfer robot 12 before the polishing process, TLFT is the residence time of the substrate in the lifter 41, TLTP1 is the residence time of the substrate in the transfer hand 45A of the first linear transporter 42, and TPoliA is the residence time of the substrate in the polishing apparatus 21A. The residence time TPoliA of the substrate in this polishing apparatus 21A includes not only the time of the polishing process but also the standby time. This standby time may also be used as input data for the learning and inference of the machine learning model. TLTP3 is the residence time of the substrate in the transfer hand 45D of the first linear transporter 42, TSTP is the residence time of the substrate in the swing transporter 44, TWS1 is the residence time of the substrate on the temporary placement table, TRB1L is the residence time of the substrate in the lower hand of the transfer robot 35, TCL1A is the residence time of the substrate in the cleaning apparatus 31A, TRB1LU is the residence time of the substrate in the upper hand of the transfer robot 35, TCL3A is the residence time of the substrate in the cleaning apparatus 31B, TRB3 is the residence time of the substrate in the hand of the transfer robot 36, TCL4A is the residence time of the substrate in the substrate drying apparatus 32, and TRBDe is the residence time of the substrate in the hand of the transfer robot 12 after drying.

[0057] As shown in FIG. 3, the polishing process has a plurality of processing steps, the first cleaning process also has a plurality of processing steps, the second cleaning process also has a plurality of processing steps, and the drying process also has a plurality of processing steps. Processing step numbers are assigned to these respective processing steps so that the processing steps can be identified. Here, assuming that these processing steps range from the 1st to the Nth (that is, the processing step numbers are 1 to N, and N is a natural number), the following description will be given.

[0058] Next, with reference to FIGS. 4 and 5, an example of the data structure of the feature amounts included in the input data input to the machine learning model will be described. Here, in the substrate processing apparatus 1, it is assumed that there are M sensors from sensor D1 to sensor DM (M is a natural number). FIG. 4 is an example of the data structure of the feature amounts included in the input data input to the machine learning model. As shown in FIG. 4, the feature amount of the value of sensor D1 is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively. Similarly, the feature amount of the value of sensor D2 is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively, and the feature amount of the value of sensor DM is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively. Thus, the feature amount of the value of sensor Di (i is an integer from 1 to M) included in the input data input to the machine learning model is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively.

[0059] Here, each member included in the above-described substrate processing apparatus 1 will be described as being referred to as members U1 to UL (L is a natural number). For example, the processor 6 stores the positions of members 1 to L included in the substrate processing apparatus 1 in the storage unit 7 in time series. The positions of these members U1 to UL may be calculated by a predetermined conversion formula from a command signal in which the unit control unit 51 commands operations on the members U1 to UL, may be determined from a known correspondence between the command signal and the position, or may be positions detected by a sensor. As shown in FIG. 4, the feature amount of the position of member U1 is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively. Similarly, the feature amount of the position of member U2 is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively. Similarly, the feature amount of the position of member UL is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively. Thus, the feature amount of the position of member Uj (j is an integer from 1 to L) included in the input data input to the machine learning model is represented by an array (or vector) having the feature amounts of processing steps 1 to N as elements respectively.

[0060] FIG. 5 is an example of the data structure of the residence time for each unit included in the input data input to the machine learning model. As shown in FIG. 5, the residence time for each unit included in the input data input to the machine learning model is represented by an array (or vector) having, as elements, the residence times of the substrates for each of the above units.

[0061] Next, an example of the learning process and the inference process of the machine learning model 71 will be described with reference to FIGS. 6A and 6B. FIG. 6A is a schematic diagram for explaining an example of the learning process of the machine learning model. FIG. 6B is a schematic diagram for explaining an example of the inference process of the machine learning model. As shown in FIG. 6A, in the learning process, the machine learning model 71 learns using a learning data set in which one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit, which were described above with reference to FIG. 4, are input data and the number of defects on the substrate is output data. As shown in FIG. 6B, in the inference process, when one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit of the target substrate are input to the machine learning model 71, the number of defects on the target substrate is output. Here, the feature amounts in this inference process are of the same type as the feature amounts in the learning process.

[0062] Note that one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit are used as the input data for machine learning, but any one of them may be used, or any combination of two of them may be used.

[0063] FIG. 7 is an example of a graph comparing the measured value of the number of defects and the predicted value of the number of defects. In FIG. 7, the vertical axis represents the predicted value of the number of defects, the horizontal axis represents the measured value of the number of defects, and for each of the polishing apparatuses 21A, 21B, 21C, and 21D, the plots are plotted with different shades of the plots. The closer the plot group is distributed to the broken line L1, the higher the prediction accuracy of the number of defects. As shown in FIG. 7, for any of the polishing apparatuses, the plot group is distributed close to the broken line L1, indicating that the prediction accuracy of the number of defects is high for any of the polishing apparatuses.

[0064] As described above, the substrate processing apparatus 1 according to the present embodiment includes at least one sensor that detects a target physical quantity during polishing and / or cleaning and / or drying of a substrate, a conversion unit 62 that converts sensor values during polishing and / or cleaning and / or drying detected by the sensor into feature quantities for each processing step with respect to a learned machine learning model, and an inference unit 64 that outputs a predicted value of the number of defects in the substrate by inputting target data including the feature quantities into the learned machine learning model 71. This learned machine learning model 71 is learned using a learning data set in which input data includes feature quantities obtained by converting sensor values during polishing and / or cleaning detected by a sensor in a target production line or a production line of the same type as the target production line for each processing step, and the output data is the number of defects in the substrate.

[0065] According to this configuration, since at least one predicted value of the number of defects, the size of the defects, and the position of the defects in the substrate after substrate processing can be obtained without inspecting with a defect inspection apparatus, it is possible to estimate whether the substrate after substrate processing is a defective product without inspecting with a defect inspection apparatus.

[0066] <Modification Example 1 of Machine Learning Model> Subsequently, a modification example 1 of the learning process and the inference process of the machine learning model 71 will be described with reference to FIGS. 8A and 8B. FIG. 8A is a schematic diagram for explaining a modification example 1 of the learning process of the machine learning model. FIG. 8B is a schematic diagram for explaining a modification example 1 of the inference process of the machine learning model. As shown in FIG. 8A, in the learning process, the machine learning model 71 is learned using a learning data set in which one or more arrays of feature quantities for each processing step, recipe information, and an array of residence times for each unit, which were described above with reference to FIG. 4, are input data, and the number of defects in the substrate and the size of the defects in the substrate are output data. As shown in FIG. 8B, in the inference process, when one or more arrays of feature quantities for each processing step, recipe information, and an array of residence times for each unit of the target substrate are input into the machine learning model 71, the number of defects in the target substrate and the size of the defects in the target substrate are output. The feature quantities in this inference process are of the same type as the feature quantities in the learning process.

[0067] <Modification Example 2 of Machine Learning Model> Subsequently, a second modification example of the learning process and the inference process of the machine learning model 71 will be described with reference to FIGS. 9A and 9B. FIG. 9A is a schematic diagram for explaining the second modification example of the learning process of the machine learning model. FIG. 8B is a schematic diagram for explaining the second modification example of the inference process of the machine learning model. As shown in FIG. 9A, in the learning process, the machine learning model 71 learns using a learning data set in which one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit, which were described above with reference to FIG. 4, are input data, and the number of defects on the substrate and the positions of the defects on the substrate are output data. As shown in FIG. 9B, in the inference process, when one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit of the target substrate are input to the machine learning model 71, the number of defects on the target substrate and the positions of the defects on the target substrate are output. Here, the feature amounts in this inference process are of the same type as the feature amounts in the learning process.

[0068] <Modification Example 3 of Machine Learning Model> Subsequently, a third modification example of the learning process and the inference process of the machine learning model 71 will be described with reference to FIGS. 10A and 10B. FIG. 10A is a schematic diagram for explaining the third modification example of the learning process of the machine learning model. FIG. 10B is a schematic diagram for explaining the third modification example of the inference process of the machine learning model. As shown in FIG. 10A, in the learning process, the machine learning model 71 learns using a learning data set in which one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit, which were described above with reference to FIG. 4, are input data, and the number of defects on the substrate, the sizes of the defects on the substrate, and the positions of the defects on the substrate are output data. As shown in FIG. 9B, in the inference process, when one or more arrays of feature amounts for each processing step, recipe information, and an array of residence times for each unit of the target substrate are input to the machine learning model 71, the number of defects in the target substrate, the sizes of the defects in the target substrate, and the positions of the defects in the target substrate are output. Here, the feature amounts in this inference process are of the same type as the feature amounts in the learning process.

[0069] Note that the inference unit 64 may output not only the combinations of output data in the first to third modification examples, but also one, or any two of the number of defects, the size of defects, and the positions of defects in the target substrate. In this way, the inference unit 64 may input the target data including the feature amount into a learned machine learning model to output at least one predicted value of the number of defects, the size of defects, and the positions of defects in the target substrate.

[0070] In this embodiment, the substrate processing apparatus 1 includes the processor 6 and the storage unit 7, but the present invention is not limited thereto. <First Modification Example of the Present Embodiment> FIG. 11A is a schematic configuration diagram according to the first modification example of the present embodiment. As shown in FIG. 11A, an information processing system S1 that is communicably connected to the substrate processing apparatus 1 is provided. The information processing system S1 has a storage unit 7 in which a processor 6 and a machine learning model 71 are stored. By reading and executing a program from the storage unit 7, the processor 6 may function as the unit residence time counting unit 61, the conversion unit 62, the learning unit 63, the inference unit 64, the regression analysis unit 65, and the reception unit 66.

[0071] <Second Modification Example of the Present Embodiment> FIG. 11B is a schematic configuration diagram according to the second modification example of the present embodiment. As shown in FIG. 11B, an information processing system S1 that is communicably connected to the substrate processing apparatus 1 via a communication circuit network CN is provided. The information processing system S1 has a storage unit 7 in which a processor 6 and a machine learning model 71 are stored. By reading and executing a predetermined program from the storage unit 7, the processor 6 may function as the unit residence time counting unit 61, the conversion unit 62, the learning unit 63, the inference unit 64, the regression analysis unit 65, and the reception unit 66.

[0072] <Third Modification Example of the Present Embodiment> FIG. 12A is a schematic configuration diagram according to Modification Example 3 of the present embodiment. As shown in FIG. 12A, a server 70 is provided which is connected to be able to exchange information with the substrate processing apparatus 1 via a communication circuit network CN. The substrate processing apparatus 1 has a processor 6 and a storage unit 7a in which a predetermined program is stored. The server 70 may have a storage unit 7b in which a machine learning model 71 is stored. In this case, the processor 6 of the substrate processing apparatus 1 reads out and executes a predetermined program from the storage unit 7a, thereby functioning as a unit residence time counting unit 61, a conversion unit 62, a learning unit 63, an inference unit 64, a regression analysis unit 65, and a reception unit 66. The inference unit 64 may output at least one predicted value among the number of defects, the size of the defects, and the position of the defects in the target substrate by inputting the target data including the feature amount into the learned machine learning model 71 of the server 70.

[0073] <Modification Example 4 of the Present Embodiment> FIG. 12A is a schematic configuration diagram according to Modification Example 3 of the present embodiment. As shown in FIG. 12A, a configuration including the substrate processing apparatus 1 and an information processing system S3 may be provided. The information processing system S3 may include an information processing apparatus 60 that is connected to the substrate processing apparatus 1 so as to be able to exchange information, and a server 70 that is connected to the information processing apparatus 60 so as to be able to exchange information via a communication circuit network CN. In this case, the information processing apparatus 60 has a processor 6 and a storage unit 7a in which a predetermined program is stored. The server 70 may have a storage unit 7b in which a machine learning model 71 is stored. Here, the processor 6 of the information processing apparatus 60 reads out and executes a predetermined program from the storage unit 7a, thereby functioning as a unit residence time counting unit 61, a conversion unit 62, a learning unit 63, an inference unit 64, a regression analysis unit 65, and a reception unit 66. The inference unit 64 may output at least one predicted value among the number of defects, the size of the defects, and the position of the defects in the target substrate by inputting the target data including the feature amount into the learned machine learning model 71 of the server 70.

[0074] In this way, the information processing systems S1, S2, and S3 convert, for a learned machine learning model, sensor values during polishing and / or cleaning and / or drying of a substrate detected by sensors provided in the substrate processing apparatus 1 into feature amounts for each processing step, and by inputting target data including the feature amounts, an inference unit 64 that outputs at least one predicted value among the number of defects, the size of defects, and the positions of defects in the substrate may be provided. Here, this learned machine learning model is used as input data including feature amounts obtained by converting sensor values during polishing and / or cleaning detected by sensors in a target production line or a production line of the same type as the target production line for each processing step, and is learned using a learning data set that outputs at least one of the number of defects, the size of defects, and the positions of defects in the substrate as output data.

[0075] According to this configuration, since at least one predicted value among the number of defects, the size of defects, and the positions of defects in the substrate after substrate processing can be obtained without inspecting with a defect inspection apparatus, it is possible to estimate whether the substrate after substrate processing is a defective product without inspecting with a defect inspection apparatus.

[0076] Note that at least some functions of the information processing systems S1, S2, and S3 or the processor 6 described in the above embodiments may be configured by hardware or may be configured by software. In the case of configuring with software, a program that realizes at least some functions of the information processing systems S1, S2, and S3 or the processor 6 may be stored in a computer-readable recording medium and read and executed by a computer. The recording medium is not limited to removable ones such as magnetic disks and optical disks, and may be a fixed-type recording medium such as a hard disk device or a memory.

[0077] Also, a program that realizes at least some functions of the information processing systems S1, S2, S3 or the processor 6 may be distributed via a communication line (including wireless communication) such as the Internet. Further, the program may be distributed in an encrypted, modulated, or compressed state via a wired or wireless line such as the Internet, or stored in a recording medium.

[0078] Furthermore, the information processing systems S1, S2, S3 may be made to function by one or a plurality of information devices. When using a plurality of information devices, one of them may be used as a computer, and the functions of at least one means of the information processing systems S1, S2, S3 may be realized by the computer executing a predetermined program.

[0079] As described above, the present invention is not limited to the above-described embodiments as they are, and at the implementation stage, the components can be modified and embodied without departing from the gist thereof. Also, various inventions can be formed by appropriately combining a plurality of components disclosed in the above embodiments. For example, some components may be deleted from all the components shown in the embodiments. Further, components from different embodiments may be appropriately combined.

Explanation of Reference Numerals

[0080] 1 Substrate processing apparatus 50 Control unit 51 Unit control unit 6 Processor 61 Unit residence time counting unit 62 Conversion unit 63 Learning unit 64 Inference unit 65 Regression analysis unit 66 Reception unit 7 Storage unit 71 Machine learning model 8 Display device S1, S2, S3 Information processing system

Claims

1. At least one sensor for detecting a target physical quantity during polishing and / or cleaning and / or drying of a substrate, a conversion unit that converts sensor values during polishing and / or cleaning and / or drying detected by the sensor into feature quantities for each processing step for a learned machine learning model, an inference unit that outputs at least one predicted value of the number of defects, the size of the defects, and the position of the defects in the target substrate by inputting target data including the feature quantities into a learned machine learning model, comprising: the learned machine learning model is learned using a learning data set in which input data includes feature quantities obtained by converting sensor values during polishing and / or cleaning detected by a sensor in a target production line or a production line of the same type as the target production line for each processing step, and output data includes at least one of the number of defects, the size of the defects, and the position of the defects in the substrate, the input data during learning of the machine learning model further includes the residence time counted for each unit included in the substrate processing apparatus and staying in the unit, each unit included in the substrate processing apparatus further comprises a unit residence time counting unit that counts the residence time staying in the unit, a substrate processing apparatus, wherein the target data input to the learned machine learning model further includes the residence time counted for each unit by the unit residence time counting unit and staying in the unit.

2. At least one sensor for detecting a target physical quantity during polishing and / or cleaning and / or drying of a substrate, a conversion unit that converts sensor values during polishing and / or cleaning and / or drying detected by the sensor into feature quantities for each processing step for a learned machine learning model, an inference unit that outputs at least one predicted value of the number of defects, the size of the defects, and the position of the defects in the target substrate by inputting target data including the feature quantities into a learned machine learning model, a regression analysis unit that outputs a correlation parameter representing a correlation between each of a plurality of sensor values and any one of the number of defects, the size of the defects, and the position of the defects in the substrate according to a predetermined regression analysis algorithm, a reception unit that receives at least one sensor that outputs a sensor value serving as a basis for a feature quantity included in the input data of the machine learning model. A learning unit that retrains the machine learning model using the feature quantity obtained by converting the sensor value of the received sensor; comprising: The trained machine learning model uses a training dataset in which input data includes feature quantities obtained by converting sensor values during polishing and / or cleaning detected by a sensor in a target production line or a production line of the same type as the target production line for each processing step, and output data includes at least one of the number of defects, the size of the defects, and the positions of the defects in the substrate. The inference unit is a substrate processing apparatus that outputs the predicted value using the machine learning model retrained by the learning unit.

3. In the input data during the training of the machine learning model, the residence time counted for each unit included in the substrate processing apparatus and staying in the unit is further included. The substrate processing apparatus further includes a unit residence time counting unit that counts the residence time staying in each unit for each unit included in the substrate processing apparatus. In the target data input to the trained machine learning model, the residence time counted for each unit by the unit residence time counting unit and staying in the unit is further included. The substrate processing apparatus according to claim 2.

4. In the input data during the training of the machine learning model, a second feature quantity obtained by converting the position of a member used for polishing or cleaning is further included. The conversion unit converts the position of the member used for polishing or cleaning into a second feature quantity. In the target data input to the trained machine learning model, the second feature quantity for each member converted by the conversion unit is further included. The substrate processing apparatus according to any one of claims 1 to 3.

5. In the input data during the training of the machine learning model, recipe information including command values for units included in the substrate processing apparatus is further included. In the target data input to the trained machine learning model, recipe information including command values for units included in the substrate processing apparatus is further included. The substrate processing apparatus according to any one of claims 1 to 4.

Images