Learning model generation method and learning model

Abstract

To provide a substrate processing device with which it is possible to improve the uniformity of processing among a plurality of substrates.SOLUTION: A substrate processing device 100 comprises a plurality of processing units 220, a transfer mechanism 210, and a control unit CNT. The plurality of processing units 220 respectively process a plurality of substrates Wx to be processed. The transfer mechanism 210 carries the substrate Wx to be processed into and out of the processing units 220. The control unit CNT controls the plurality of processing units 220 and the transfer mechanism 210. The control unit CNT controls one or more processing units 220 and the transfer mechanism 210 so that each of the substrates Wx to be processed is processed by one or more processing units 220 that are selected from among the plurality of processing units 220 on the basis of the processing results of a plurality of substrates Wp by each of the plurality of processing units 220.SELECTED DRAWING: Figure 8

Description

[Technical field]

[0001] The present invention relates to a substrate processing About The present invention relates to a learning model generating method and a learning model. [Background technology]

[0002] The substrate processing apparatus described in Patent Document 1 includes a carrier holding section that holds a carrier that houses a substrate, a plurality of processing units that process the substrate, a substrate transport unit that transports the substrate between the carrier and the processing units, and a computer. The computer classifies the plurality of processing units into a plurality of groups, and selects one of the plurality of groups when determining a processing unit that should process a substrate. The computer further selects one processing unit that belongs to the selected group, and controls the substrate transport unit to load the substrate into the selected processing unit. As a result, the variation in frequency of use of the plurality of processing units can be reduced. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2017-183545 A Summary of the Invention [Problem to be solved by the invention]

[0004] However, the substrate processing apparatus described in Patent Document 1 focuses only on the variation in the frequency of use of the processing units.

[0005] In response to this, the inventors of the present application focused on the possibility of variation in processing characteristics between multiple processing units, and conducted extensive research to improve the uniformity of processing between multiple substrates even when the processing characteristics vary.

[0006] The present invention One embodimentis the uniformity of processing across multiple substrates of improvement Contribute to Can Learning A method for generating a learning model and a method for providing a learning model do. [Means for solving the problem]

[0007] One aspect of the present invention In accordance with an embodiment The learning model generating method includes a step of acquiring learning data and a step of generating a learning model by performing machine learning on a plurality of the learning data. Each of the plurality of learning data includes processing result information indicating a processing result of a learning target substrate, and evaluation information indicating an evaluation of the processing result of the learning target substrate. The processing result information included in the learning data is one of two or more pieces of processing result information indicating processing results of two or more of the learning target substrates by two or more processing units having the same processing content. The learning model is a program that inputs, as input information, each of two or more pieces of processing result information indicating processing results of two or more substrates by the two or more processing units having the same processing content, and outputs, as output information, evaluation information indicating an evaluation of each of the processing results of the two or more substrates.

[0008] The present invention In one embodiment The learning model generating method includes a step of acquiring learning data, and a step of generating a learning model by performing machine learning on a plurality of the learning data. Each of the plurality of learning data includes two or more pieces of processing result information, each of which indicates a processing result of a learning target substrate by two or more processing units having different processing contents. The learning model is a program that inputs, as input information, each of two or more pieces of processing result information, each of which indicates a processing result of a substrate by the two or more processing units having different processing contents, clusters the two or more pieces of processing result information, and outputs, as output information, clustering information indicating a result of the clustering.

[0009] The present invention In one embodimentThe learning model is constructed by executing machine learning on a plurality of learning data, and causes the computer to function to evaluate the processing results of the substrate. Each of the plurality of learning data includes processing result information indicating the processing result of the learning target substrate, and evaluation information indicating an evaluation of the processing result of the learning target substrate. The processing result information included in the learning data is one of two or more pieces of processing result information indicating the processing results of the two or more learning target substrates by two or more processing units having the same processing content. The learning model causes the computer to function to input each of two or more pieces of input information and output each of two or more pieces of output information. The two or more pieces of input information are two or more pieces of processing result information indicating each of the processing results of the two or more substrates by the two or more processing units having the same processing content. The two or more pieces of output information are two or more pieces of evaluation information indicating the evaluation of the processing results of the two or more substrates.

[0010] The present invention In one embodiment The learning model is constructed by performing machine learning on a plurality of learning data, and causes the computer to function so as to cluster processing result information indicating the processing results of the substrate. Each of the plurality of learning data includes two or more pieces of processing result information, each indicating the processing results of the substrate to be learned by two or more processing units having different processing contents. The learning model causes the computer to function so as to input input information and output output information. The input information is two or more pieces of processing result information, each indicating the processing results of the substrate by the two or more processing units having different processing contents. The output information is clustering information indicating the result of clustering the two or more pieces of processing result information. Effect of the Invention

[0011] The present invention provides uniformity of processing across multiple substrates. of improvement Contribute to Can Learning A learning model generation method and a learning model can be provided. [Brief description of the drawings]

[0012] [Figure 1] 1 is a plan view showing a substrate processing apparatus according to a first embodiment of the present invention. [Diagram 2] 2 is a view showing a coating processing section, a developing processing section, and a cleaning / drying processing section of the substrate processing apparatus according to the first embodiment, as viewed from a direction D1 in FIG. [Diagram 3] 2 is a view showing a heat processing section and a cleaning / drying processing section of the substrate processing apparatus according to the first embodiment, as viewed from a direction D1 in FIG. [Figure 4] 2 is a view showing a coating processing section, a transport section, and a heat processing section of the substrate processing apparatus according to the first embodiment, as viewed from a direction D2 in FIG. [Diagram 5] 2 is a view of a transfer section of the substrate processing apparatus according to the first embodiment, as viewed from a direction D1 in FIG. [Figure 6] 5 is a flowchart showing processing of one substrate on an outward trip by the substrate processing apparatus according to the first embodiment. [Figure 7] 5 is a flowchart showing processing of one substrate on a return pass by the substrate processing apparatus according to the first embodiment. [Figure 8] 1 is a block diagram showing a control device in the substrate processing apparatus according to embodiment 1. FIG. [Figure 9] 4 is a flowchart showing a substrate processing method performed by the substrate processing apparatus according to the first embodiment. [Figure 10] 4 is a flowchart showing a "processing unit selection" process of the substrate processing method according to the first embodiment. [Figure 11] 5A to 5C are diagrams showing examples of processing result information, priority order information, selection result information, and inspection target information according to the first embodiment. [Figure 12] FIG. 11 is a block diagram showing a control device in a substrate processing apparatus according to a second embodiment of the present invention. [Figure 13] 10 is a flowchart showing a "processing unit selection" process of the substrate processing method according to the second embodiment. [Figure 14] 13 is a flowchart showing an example of a “selection of one or more processing units based on processing result information” process in the substrate processing method according to the second embodiment. [Figure 15]10 is a flowchart showing a learning method executed by the substrate processing apparatus according to the second embodiment. [Figure 16] 13A to 13C are diagrams showing examples of processing result information, priority order information, selection result information, and inspection target information according to the second embodiment. [Figure 17] FIG. 11 is a diagram showing a contact angle map according to the second embodiment. [Figure 18] FIG. 11 is a diagram showing another contact angle map according to the second embodiment. [Figure 19] FIG. 11 is a diagram showing yet another contact angle map according to the second embodiment. [Figure 20] FIG. 11 is a diagram illustrating a storage unit and a control unit according to a third embodiment of the present invention. [Figure 21] 13 is a flowchart showing a part of a "processing unit selection" process of the substrate processing method according to the third embodiment. [Figure 22] 13 is a flowchart showing another part of the “processing unit selection” process of the substrate processing method according to the third embodiment. [Diagram 23] 13 is a flowchart showing an example of a “selection of one or more processing units based on the processing result information set” process of the substrate processing method according to the third embodiment. [Figure 24] 13 is a flowchart showing an example of a “selection of one or more processing units based on a correlation” process in the substrate processing method according to the third embodiment. [Diagram 25] 13A to 13C are diagrams illustrating an example of a processing result information set, priority order information, selection result information, and inspection target information according to the third embodiment. [Figure 26] FIG. 13 is a diagram showing a contact angle map, a film thickness map, a temperature map, and a pattern line width map, which constitute a processing result information set according to the third embodiment. [Figure 27] FIG. 13 is a diagram showing a contact angle map, a film thickness map, a temperature map, and a pattern line width map, which constitute another processing result information set according to the third embodiment. [Figure 28] FIG. 13 is a diagram showing a contact angle map, a film thickness map, a temperature map, and a pattern line width map, which constitute yet another processing result information set according to the third embodiment. [Figure 29] FIG. 11 is a diagram showing a substrate processing system according to a fourth embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the drawings, the same or corresponding parts are denoted by the same reference characters and the description will not be repeated. In the embodiment of the present invention, the X-axis, Y-axis, and Z-axis are mutually orthogonal, the X-axis and Y-axis are parallel to the horizontal direction, and the Z-axis is parallel to the vertical direction.

[0014] (Embodiment 1) A substrate processing apparatus 100 according to a first embodiment of the present invention will be described with reference to Figs. 1 to 7. The substrate processing apparatus 100 processes a plurality of substrates W in parallel. The substrate W is, for example, a semiconductor wafer, a substrate for a liquid crystal display device, a substrate for a plasma display, a substrate for a field emission display (FED), a substrate for an optical disk, a substrate for a magnetic disk, a substrate for a magneto-optical disk, a substrate for a photomask, a ceramic substrate, or a substrate for a solar cell. The substrate W is, for example, substantially disk-shaped. In the following description of the first embodiment, the substrate W is a semiconductor substrate.

[0015] First, a substrate processing apparatus 100 will be described with reference to Fig. 1. Fig. 1 is a plan view showing the substrate processing apparatus 100 according to the first embodiment. As shown in Fig. 1, the substrate processing apparatus 100 includes an indexer block 11, a first processing block 12, a second processing block 13, a cleaning and drying processing block 14A, a load / unload block 14B, and a control device CM. The cleaning and drying processing block 14A and the load / unload block 14B form an interface block 14. An exposure device 15 is disposed adjacent to the load / unload block 14B. In the exposure device 15, an exposure process is performed on a substrate W by an immersion method.

[0016] The control device CM controls the indexer block 11, the first processing block 12, the second processing block 13, and the interface block 14. The control device CM is, for example, a computer.

[0017] The indexer block 11 includes a plurality of carrier placement parts 111 and a transport part 112. A carrier 113 that stores a plurality of substrates W in multiple stages is placed on each carrier placement part 111. In the first embodiment, a FOUP (front opening unified pod) is used as the carrier 113, but the present invention is not limited to this, and for example, a SMIF (Standard Mechanical Interface) pod or an OC (open cassette) that exposes the stored substrates W to the outside air may also be used.

[0018] The transport section 112 is provided with a transport mechanism 115. The transport mechanism 115 has a hand 116 for holding the substrate W. The transport mechanism 115 transports the substrate W while holding the substrate W with the hand 116. As shown in FIG. 5 described later, the transport section 112 is formed with an opening 117 for transferring the substrate W between the carrier 113 and the transport mechanism 115.

[0019] The first processing block 12 includes a coating processing section 121, a transport section 122, and a thermal processing section 123. The thermal processing section 123 includes a plurality of plates 5. A substrate W is placed on the plates 5. The coating processing section 121 and the thermal processing section 123 face each other with the transport section 122 in between. Between the transport section 122 and the indexer block 11, there are provided a substrate placement section PASS1 on which the substrate W is placed, and substrate placement sections PASS2 to PASS4 (see FIG. 5) described later. The transport section 122 is provided with a transport mechanism 127 for transporting the substrate W, and a transport mechanism 128 (see FIG. 5) described later.

[0020] The second processing block 13 includes a development processing section 131, a transport section 132, and a thermal processing section 133. The thermal processing section 133 includes a plurality of plates 5. A substrate W is placed on the plate 5. The development processing section 131 and the thermal processing section 133 are provided to face each other with the transport section 132 in between. A substrate placement section PASS5 on which the substrate W is placed and substrate placement sections PASS6 to PASS8 (see FIG. 5), which will be described later, are provided between the transport section 132 and the transport section 122. The transport section 132 is provided with a transport mechanism 137 for transporting the substrate W and a transport mechanism 138 (see FIG. 5), which will be described later. In the second processing block 13, a packing 145 is provided between the thermal processing section 133 and the interface block 14.

[0021] The cleaning and drying processing block 14A includes cleaning and drying processing sections 161 and 162 and a transport section 163. The cleaning and drying processing sections 161 and 162 face each other with the transport section 163 in between. The transport section 163 is provided with transport mechanisms 141 and 142.

[0022] A placement and buffer unit P-BF1 and a placement and buffer unit P-BF2 (see FIG. 5), which will be described later, are provided between the transport unit 163 and the transport unit 132. The placement and buffer units P-BF1 and P-BF2 are capable of accommodating a plurality of substrates W.

[0023] Furthermore, between the transport mechanisms 141, 142, substrate placement parts PASS11, PASS12 and a placement / cooling part P-CP (see FIG. 5) described below are provided adjacent to the load-in / load-out block 14B. The placement / cooling part P-CP has a function of cooling the substrate W (e.g., a cooling plate). In the placement / cooling part P-CP, the substrate W is cooled to a temperature suitable for exposure processing. Therefore, the placement / cooling part P-CP functions as a cooling processing unit. Hereinafter, the placement / cooling part P-CP may be referred to as a processing unit P-CP or a cooling processing unit P-CP.

[0024] The load / unload block 14B is provided with a transport mechanism 146. The transport mechanism 146 loads and unloads the substrate W to and from the exposure device 15. The transport mechanism 146 has hands H7, H8 for holding the substrate W. The exposure device 15 is provided with a substrate load section 15a for loading the substrate W and a substrate unload section 15b for unloading the substrate W.

[0025] Next, the coating processing unit 121 and the developing processing unit 131 in Fig. 1 will be described with reference to Fig. 2. Fig. 2 is a view of the coating processing unit 121, the developing processing unit 131, and the cleaning / drying processing unit 161 in Fig. 1 as viewed from a direction D1 in Fig. 1.

[0026] 2, the coating processing section 121 is provided with coating processing chambers 21, 22, 23, and 24 arranged in a hierarchical manner. The coating processing chamber 21 is provided with anti-reflective film coating processing units AR

[11] and AR

[12] . The coating processing chamber 23 is provided with anti-reflective film coating processing units AR

[21] and AR

[22] . The coating processing chamber 22 is provided with resist film coating processing units RT

[11] and RT

[12] . The coating processing chamber 24 is provided with resist film coating processing units RT

[21] and RT

[22] .

[0027] Hereinafter, the anti-reflective coating processing units AR

[11] , AR

[12] , AR

[21] , and AR

[22] may be collectively referred to as processing units AR or anti-reflective coating processing units AR. Also, the resist film coating processing units RT

[11] , RT

[12] , RT

[21] , and RT

[22] may be collectively referred to as processing units RT or resist film coating processing units RT.

[0028] Each of the processing units RT, AR includes a spin chuck 25 for holding the substrate W, and a cup 27 provided to cover the periphery of the spin chuck 25. The spin chuck 25 is rotated by a driving device (for example, an electric motor) not shown.

[0029] 1, the coating processing chamber 24 in which the processing unit RT is disposed is provided with a plurality of nozzles 28 for discharging processing liquid, and a nozzle transport mechanism 29. The nozzle transport mechanism 29 transports the nozzles 28.

[0030] In the coating processing chamber 24, one of the multiple nozzles 28 is moved above the substrate W by the nozzle transport mechanism 29. The processing liquid is discharged from the nozzle 28, thereby coating the substrate W with the processing liquid. When the processing liquid is supplied from the nozzle 28 to the substrate W, the spin chuck 25 is rotated by a drive device (not shown). As a result, the substrate W is rotated.

[0031] Similarly to the coating processing chamber 24, each of the coating processing chambers 23, 22, and 21 (FIG. 2) is provided with a plurality of nozzles 28 and a nozzle transport mechanism 29.

[0032] The nozzles 28 and nozzle transport mechanism 29 of the coating processing chambers 24 and 22 constitute a part of a processing unit RT. The nozzles 28 and nozzle transport mechanism 29 of the coating processing chambers 23 and 21 constitute a part of a processing unit AR.

[0033] In the first embodiment, in the coating processing chambers 24, 22 in which the processing units RT are arranged, a processing liquid for a resist film is supplied to the substrate W from a nozzle 28. Therefore, the nozzle 28 of the coating processing chambers 24, 22 constitutes a part of the processing unit RT.

[0034] In the coating processing chambers 23 and 21 in which the processing units AR are arranged, a processing liquid for an anti-reflection film is supplied to the substrate W from a nozzle 28. Therefore, the nozzle 28 of the coating processing chambers 23 and 21 constitutes a part of the processing unit AR.

[0035] As shown in FIG. 2, the development processing section 131 is provided with development processing chambers 31, 32, 33, and 34 arranged in a hierarchical manner. The development processing chamber 31 is provided with development processing units DV

[12] , DV

[14] , and DV

[16] . The development processing chamber 32 is provided with development processing units DV

[11] , DV

[13] , and DV

[15] . The development processing chamber 33 is provided with development processing units DV

[22] , DV

[24] , and DV

[26] . The development processing chamber 34 is provided with development processing units DV

[21] , DV

[23] , and DV

[25] .

[0036] Hereinafter, the development processing units DV

[11] to DV

[16] and DV

[21] to DV

[26] may be collectively referred to as the processing unit DV or the development processing unit DV.

[0037] Like the processing units RT and AR, the processing unit DV includes a spin chuck 35 and a cup 37. As shown in Fig. 1, the processing unit DV also includes two nozzles 38 that eject the developer, and a moving mechanism 39. The moving mechanism 39 moves the nozzles 38 in the X direction.

[0038] In the processing unit DV, first, one nozzle 38 moves in the X direction to supply the developing solution to each substrate W. Then, the other nozzle 38 moves in the X direction to supply the developing solution to each substrate W. When the developing solution is supplied to the substrate W from the nozzle 38, the spin chuck 35 is rotated by a driving device (not shown). As a result, the substrate W is rotated.

[0039] In the first embodiment, a developer is supplied to the substrate W in the processing unit DV, thereby removing the resist cover film on the substrate W and developing the substrate W. In the first embodiment, different developers are discharged from the two nozzles 38. Therefore, two types of developers can be supplied to each substrate W.

[0040] 2, a plurality of (four in this example) cleaning and drying processing units SD1 are provided in the cleaning and drying processing section 161. In the cleaning and drying processing units SD1, cleaning and drying processing is performed on the substrate W before the exposure processing.

[0041] Moreover, an air supply unit 41 is provided above the processing units AR, RT in the coating processing chambers 21-24 to supply clean air whose temperature and humidity have been adjusted into the coating processing chambers 21-24. Moreover, an air supply unit 47 is provided above the processing unit DV in the development processing chambers 31-34 to supply clean air whose temperature and humidity have been adjusted into the development processing chambers 31-34.

[0042] Further, in the coating processing chambers 21-24, below the processing units AR, RT, an exhaust unit 42 for exhausting the atmosphere in the cup 27 is provided. Further, in the development processing chambers 31-34, below the processing unit DV, an exhaust unit 48 for exhausting the atmosphere in the cup 37 is provided.

[0043] 1 and 2, a fluid box unit 50 is provided adjacent to the development processing unit 131 in the coating processing unit 121. Similarly, a fluid box unit 60 is provided adjacent to the cleaning and drying processing block 14A in the development processing unit 131. The fluid box units 50 and 60 house fluid-related devices such as conduits, joints, valves, flow meters, regulators, pumps, and temperature regulators for supplying chemicals to the processing units AR, RT, and DV and for draining and exhausting from the processing units AR, RT, and DV.

[0044] Next, the heat treatment units 123, 133 and the cleaning / drying treatment unit 162 in Fig. 1 will be described with reference to Fig. 3. Fig. 3 is a view of the heat treatment units 123, 133 and the cleaning / drying treatment unit 162 as viewed from a direction D1 in Fig. 1.

[0045] 3, the thermal processing section 123 has an upper thermal processing section 301 provided above and a lower thermal processing section 302 provided below. The lower thermal processing section 302 is provided with adhesion promotion processing units PAHP

[11] -PAHP

[14] , multiple thermal processing units PHP

[15] -PHP

[22] , and multiple cooling processing units CP

[11] -CP

[13] . The upper thermal processing section 301 is provided with adhesion promotion processing units PAHP

[51] -PAHP

[54] , multiple thermal processing units PHP

[55] -PHP

[62] , and multiple cooling processing units CP

[21] -CP

[23] .

[0046] The thermal processing section 133 has an upper thermal processing section 303 provided above and a lower thermal processing section 304 provided below. The lower thermal processing section 304 is provided with multiple thermal processing units PHP

[23] -PHP

[34] , multiple cooling processing units CP

[14] , CP

[15] , and an edge exposure processing unit EEW[1]. The upper thermal processing section 303 is provided with multiple thermal processing units PHP

[63] -PHP

[74] , multiple cooling processing units CP

[24] , CP

[25] , and an edge exposure processing unit EEW[2].

[0047] Hereinafter, the thermal processing units PHP

[15] -PHP

[34] and PHP

[55] -PHP

[74] may be collectively referred to as the processing unit PHP or the thermal processing unit PHP. The adhesion reinforcement processing units PAHP

[11] -PAHP

[14] and PAHP

[51] -PAHP

[54] may be collectively referred to as the processing unit PAHP or the adhesion reinforcement processing unit PAHP. The cooling processing units CP

[11] -CP

[15] and CP

[21] -CP

[25] may be collectively referred to as the processing unit CP or the cooling processing unit CP. The edge exposure processing units EEW[1] and EEW[2] may be collectively referred to as the processing unit EEW or the edge exposure processing unit EEW.

[0048] In the thermal processing unit PHP, heating processing and cooling processing are performed on the substrate W. In the thermal processing unit PHP, heating processing and cooling processing are performed continuously on the substrate W. For example, the thermal processing unit PHP includes two plates 5 (FIG. 1). The substrate W is placed on one of the plates 5, and the substrate W is subjected to heating processing. In addition, the substrate W is placed on the other plate 5, and the substrate W is subjected to cooling processing.

[0049] In the adhesion reinforcement processing unit PAHP, an adhesion reinforcement process is performed to improve the adhesion between the substrate W and the anti-reflection film. Specifically, in the adhesion reinforcement processing unit PAHP, an adhesion reinforcement agent such as HMDS (hexamethyldisilane) is applied to the substrate W, and the substrate W is subjected to a heat treatment. Applying the adhesion reinforcement agent to the substrate W corresponds to performing a hydrophobization process on the substrate W. In addition, after the substrate W is subjected to the heat treatment, a cooling process is performed on the substrate W. For example, the adhesion reinforcement processing unit PAHP includes two plates 5 (FIG. 1). The substrate W is placed on one of the plates 5, the adhesion reinforcement agent is applied to the substrate W, and the substrate W is subjected to a heat treatment. The substrate W is placed on the other plate 5, and the substrate W is subjected to a cooling process.

[0050] In the cooling processing unit CP, a cooling process is performed on the substrate W. For example, the cooling processing unit CP includes one or two plates (not shown). The substrate W is placed on the plates, and the cooling process is performed on the substrate W.

[0051] In the edge exposure processing unit EEW, exposure processing is performed on the peripheral portion of the substrate W. For example, the edge exposure processing unit EEW includes one plate 5 (FIG. 1). The substrate W is placed on the plate 5, and exposure processing is performed on the peripheral portion of the substrate W.

[0052] Hereinafter, the thermal processing unit PHP, adhesion reinforcement processing unit PAHP, cooling processing unit CP, edge exposure processing unit EEW, anti-reflective film coating processing unit AR, resist film coating processing unit RT, development processing unit DV, placement and cooling part P-CP, and substrate placement parts PASS6 and PASS8 which can function as cooling processing units (described below) may be collectively referred to as processing unit 220 (FIG. 5 described below).

[0053] 3, the cleaning and drying processing section 162 is provided with a plurality of (five in this example) cleaning and drying processing units SD2. In the cleaning and drying processing units SD2, cleaning and drying processing are performed on the substrate W after the exposure processing.

[0054] Next, the coating processing unit 121, the transport unit 122, and the heat processing unit 123 in Fig. 1 will be described with reference to Fig. 4 and Fig. 5. Fig. 4 is a view of the coating processing unit 121, the transport unit 122, and the heat processing unit 123 as viewed from a direction D2 in Fig. 1. Fig. 5 is a view of the transport units 122, 132, and 163 as viewed from a direction D1 in Fig. 1.

[0055] 4 and 5, the transfer section 122 has an upper transfer chamber 125 and a lower transfer chamber 126. The transfer section 132 has an upper transfer chamber 135 and a lower transfer chamber 136.

[0056] The upper transfer chamber 125 is provided with a transfer mechanism 127, and the lower transfer chamber 126 is provided with a transfer mechanism 128. Furthermore, the upper transfer chamber 135 is provided with a transfer mechanism 137, and the lower transfer chamber 136 is provided with a transfer mechanism 138.

[0057] 4, the coating treatment chambers 23, 24 and the upper heat processing section 301 face each other across the upper transfer chamber 125, and the coating treatment chambers 21, 22 and the lower heat processing section 302 face each other across the lower transfer chamber 126. Similarly, the development treatment chambers 33, 34 (FIG. 2) and the upper heat processing section 303 (FIG. 3) face each other across the upper transfer chamber 135 (FIG. 5), and the development treatment chambers 31, 32 (FIG. 2) and the lower heat processing section 304 (FIG. 3) face each other across the lower transfer chamber 136 (FIG. 5).

[0058] 5, substrate placement parts PASS1 and PASS2 are provided between the transport part 112 and the upper transport chamber 125, and substrate placement parts PASS3 and PASS4 are provided between the transport part 112 and the lower transport chamber 126. For example, a placement / buffer part may be provided between the transport part 112 and the upper transport chamber 125. The placement / buffer part is configured to allow the transport mechanisms 115 and 127 to load and unload the substrate W. For example, a placement / buffer part may be provided between the transport part 112 and the lower transport chamber 126. The placement / buffer part is configured to allow the transport mechanisms 115 and 128 to load and unload the substrate W.

[0059] Substrate placement parts PASS5 and PASS6 are provided between the upper transport chamber 125 and the upper transport chamber 135, and substrate placement parts PASS7 and PASS8 are provided between the lower transport chamber 126 and the lower transport chamber 136. For example, a plurality of substrate placement parts PASS6 (e.g., three substrate placement parts PASS6) may be provided vertically, or a plurality of substrate placement parts PASS8 (e.g., three substrate placement parts PASS8) may be provided vertically. For example, a placement / buffer section may be provided between the upper transport chamber 125 and the upper transport chamber 135. The placement / buffer section is configured so that the substrate W can be carried in and out by the transport mechanisms 127 and 137. For example, a placement / buffer section may be provided between the lower transport chamber 126 and the lower transport chamber 136. The placement / buffer section is configured so that the substrate W can be carried in and out by the transport mechanisms 128 and 138.

[0060] A placement / buffer unit P-BF1 is provided between the upper transfer chamber 135 and the transfer unit 163, and a placement / buffer unit P-BF2 is provided between the lower transfer chamber 136 and the transfer unit 163. In addition, a substrate placement unit PASS9 is provided between the upper transfer chamber 135 and the transfer unit 163. Furthermore, a substrate placement unit PASS10 is provided between the lower transfer chamber 136 and the transfer unit 163.

[0061] In the transfer section 163, substrate placement parts PASS11, PASS12 and a plurality of placement / cooling parts P-CP are provided adjacent to the load / unload block 14B.

[0062] The placement / buffer section P-BF1 and the substrate placement section PASS9 are configured to allow the transfer mechanism 137 and the transfer mechanisms 141, 142 (FIG. 1) to load and unload the substrate W. The placement / buffer section P-BF2 and the substrate placement section PASS10 are configured to allow the transfer mechanism 138 and the transfer mechanisms 141, 142 (FIG. 1) to load and unload the substrate W. The substrate placement sections PASS11, PASS12 and the placement / cooling section P-CP are configured to allow the transfer mechanism 141, 142 (FIG. 1) and the transfer mechanism 146 to load and unload the substrate W.

[0063] The transport mechanisms 141, 142 (FIG. 1) can transfer the substrate W to and from the placement / buffer sections P-BF1, P-BF2, the substrate placement sections PASS9 to PASS12, and the placement / cooling section P-CP. The transport mechanism 146 can transfer the substrate W to and from the substrate placement sections PASS11, PASS12, the placement / cooling section P-CP, the substrate carry-in section 15a (FIG. 1), and the substrate carry-out section 15b (FIG. 1).

[0064] 5, only one substrate platform PASS11 is provided, but multiple substrate platform PASS11 (e.g., two substrate platform PASS11) may be provided above and below. Also, multiple substrate platform PASS12 (e.g., two substrate platform PASS12) may be provided above and below. In this case, the substrate platform PASS11 and PASS12 may be used as buffer sections for temporarily placing the substrate W.

[0065] In the first embodiment, substrates W to be transported from the indexer block 11 to the first processing block 12 are placed on the substrate placement parts PASS1 and PASS3, and substrates W to be transported from the first processing block 12 to the indexer block 11 are placed on the substrate placement parts PASS2 and PASS4. The transport mechanism 115 can use a hand 116 to transfer the substrates W to and from the carrier 113 and the substrate placement parts PASS1 to PASS4.

[0066] Further, on the substrate rest parts PASS5, PASS6, PASS7, and PASS8, a substrate W to be transported from the first processing block 12 to the second processing block 13 is placed, or a substrate W to be transported from the second processing block 13 to the first processing block 12 is placed. In the first embodiment, each of the substrate rest parts PASS6 and PASS8 has a function of cooling the substrate W (e.g., a cooling plate). Thus, the substrate rest parts PASS6 and PASS8 function as cooling processing units. Hereinafter, the substrate rest part PASS6 may be referred to as a processing unit PASS6 or a cooling processing unit PASS6. Further, the substrate rest part PASS8 may be referred to as a processing unit PASS8 or a cooling processing unit PASS8.

[0067] The placement / buffer section P-BF1 receives a substrate W to be transported from the upper transport chamber 135 of the second processing block 13 to the cleaning / drying processing block 14A, and the substrate placement section PASS9 receives a substrate W to be transported from the cleaning / drying processing block 14A to the upper transport chamber 135. The placement / buffer section P-BF2 receives a substrate W to be transported from the lower transport chamber 136 of the second processing block 13 to the cleaning / drying processing block 14A, and the substrate placement section PASS10 receives a substrate W to be transported from the cleaning / drying processing block 14A to the lower transport chamber 136.

[0068] The placement and cooling section P-CP holds the substrate W to be transported from the cleaning and drying processing block 14A to the load-unloading block 14B, the substrate placement section PASS11 holds the substrate W to be transported from the cleaning and drying processing block 14A to the load-unloading block 14B, and the substrate placement section PASS12 holds the substrate W to be transported from the load-unloading block 14B to the cleaning and drying processing block 14A.

[0069] An air supply unit 43 is provided above the transport mechanism 127 in the upper transport chamber 125, and an air supply unit 43 is provided above the transport mechanism 128 in the lower transport chamber 126. An air supply unit 43 is provided above the transport mechanism 137 in the upper transport chamber 135, and an air supply unit 43 is provided above the transport mechanism 138 in the lower transport chamber 136. Air whose temperature and humidity have been adjusted is supplied to the air supply unit 43 by a temperature adjustment device (not shown).

[0070] In addition, an exhaust unit 44 for evacuating the upper transport chamber 125 is provided below the transport mechanism 127 within the upper transport chamber 125, and an exhaust unit 44 for evacuating the lower transport chamber 126 is provided below the transport mechanism 128 within the lower transport chamber 126.

[0071] Similarly, an exhaust unit 44 for evacuating the upper transport chamber 135 is provided below the transport mechanism 137 in the upper transport chamber 135, and an exhaust unit 44 for evacuating the lower transport chamber 136 is provided below the transport mechanism 138 in the lower transport chamber 136.

[0072] As a result, the atmosphere in the upper transfer chambers 125, 135 and the lower transfer chambers 126, 136 is maintained at an appropriate temperature, humidity and cleanliness.

[0073] An air supply unit 45 is provided at the upper part of the transport section 163 of the cleaning and drying processing block 14A. An air supply unit 46 is provided at the upper part of the load-in / load-out block 14B. Air whose temperature and humidity have been adjusted is supplied to the air supply units 45 and 46 by a temperature adjustment device (not shown). As a result, the atmospheres in the cleaning and drying processing block 14A and the load-in / load-out block 14B are maintained at appropriate temperature, humidity and cleanliness.

[0074] 5, the transfer mechanism 127 will be described. The transfer mechanism 127 includes long guide rails 311, 312. The guide rail 311 is fixed to the transfer section 112 side so as to extend in the vertical direction within the upper transfer chamber 125. The guide rail 312 is fixed to the upper transfer chamber 135 side so as to extend in the vertical direction within the upper transfer chamber 125.

[0075] A long guide rail 313 is provided between guide rail 311 and guide rail 312. Guide rail 313 is attached to guide rails 311, 312 so as to be vertically movable. A moving member 314 is attached to guide rail 313. Moving member 314 is provided so as to be movable in the longitudinal direction of guide rail 313.

[0076] A long rotating member 315 is rotatably provided on the upper surface of the moving member 314. Hands H1 and H2 for holding the substrate W are attached to the rotating member 315. The hands H1 and H2 are provided to be movable in the longitudinal direction of the rotating member 315.

[0077] With the above-mentioned configuration, the transport mechanism 127 can freely move in the X and Z directions in the upper transport chamber 125. In addition, the transport mechanism 127 can use the hands H1 and H2 to deliver and receive the substrate W to and from the coating processing chambers 24 and 23 (FIG. 2), the substrate placement parts PASS1, PASS2, PASS5, and PASS6 (FIG. 5), and the upper thermal processing part 301 (FIG. 3).

[0078] As shown in FIG. 5, the transport mechanisms 128, 137, and 138 have the same configuration as the transport mechanism 127.

[0079] Therefore, the transport mechanism 128 can freely move in the X and Z directions within the lower transport chamber 126. In addition, the transport mechanism 128 can use the hands H1 and H2 to deliver the substrate W to and from the coating processing chambers 22 and 21 (FIG. 2), the substrate placement parts PASS3, PASS4, PASS7, and PASS8 (FIG. 5), and the lower thermal processing part 302 (FIG. 3).

[0080] The transport mechanism 137 can move freely in the X and Z directions in the upper transport chamber 135. The transport mechanism 137 can use the hands H1 and H2 to deliver the substrate W to and from the development processing chambers 34 and 33 (FIG. 2), the substrate placement parts PASS5 and PASS6 (FIG. 5), the placement / buffer part P-BF1 (FIG. 5), the substrate placement part PASS9 (FIG. 5), and the upper thermal processing part 303 (FIG. 3).

[0081] The transport mechanism 138 can move freely in the X and Z directions within the lower transport chamber 136. The transport mechanism 138 can use hands H1 and H2 to deliver and receive substrates W to and from the development processing chambers 32 and 31 (FIG. 2), the substrate placement parts PASS7 and PASS8 (FIG. 5), the placement / buffer part P-BF2 (FIG. 5), the substrate placement part PASS10 (FIG. 5), and the lower thermal processing part 304 (FIG. 3).

[0082] Hereinafter, the transport mechanisms 115, 127, 128, 137, 138, 141, 142, and 146 may be collectively referred to as a transport mechanism 210.

[0083] Next, with reference to Figures 6 and 7, the processing of the substrate W transported from the carrier 113 of the substrate processing apparatus 100 to the exposure device 15, and the processing of the substrate W transported from the exposure device 15 to the carrier 113 of the substrate processing apparatus 100 will be described.

[0084] Fig. 6 is a flowchart showing the processing of one substrate W on the outward journey from the substrate processing apparatus 100 to the exposure apparatus 15. Fig. 7 is a flowchart showing the processing of one substrate W on the return journey from the exposure apparatus 15 to the substrate processing apparatus 100.

[0085] To perform the adhesion strengthening process in step S1 shown in FIG. 6, the transport mechanisms 115 and 128 in FIG. 5 unload the substrate W from the carrier 113 and load it into the processing unit PAHP (for example, processing unit PAHP

[11] ) in FIG.

[0086] Then, in step S1, the processing unit PAHP (for example, the processing unit PAHP

[11] ) performs an adhesion enhancing process on the substrate W. That is, the processing unit PAHP applies an adhesion enhancing agent to the substrate W and heats the substrate W. Then, the processing unit PAHP cools the substrate W.

[0087] 2 (e.g., processing unit AR

[11] ) performs an anti-reflection film formation process on the substrate W. That is, the processing unit AR applies a processing liquid for an anti-reflection film to the substrate W that has been subjected to the adhesion strengthening process, thereby forming an anti-reflection film on the substrate W.

[0088] 3 (for example, the processing unit PHP

[15] ) performs a first heating process on the substrate W. That is, the processing unit PHP heats the substrate W on which the anti-reflection film is formed.

[0089] Next, in step S4, the processing unit CP (for example, the processing unit CP

[11] ) performs a first cooling process on the substrate W. That is, the processing unit CP cools the substrate W after the first heating process.

[0090] Next, in step S5, the processing unit RT (e.g., the processing unit RT

[11] ) in Fig. 2 performs a resist film formation process on the substrate W. That is, the processing unit RT applies a processing liquid for a resist film to the substrate W after the first cooling process to form a resist film on the substrate W. The resist film is formed on the anti-reflection coating.

[0091] Next, in step S6, the processing unit PHP in Fig. 3 (for example, the processing unit PHP

[19] ) performs a second heating process on the substrate W. That is, the processing unit PHP heats the substrate W on which the resist film has been formed.

[0092] 5 performs a second cooling process on the substrate W. That is, the processing unit PASS8 cools the substrate W after the second heating process. Note that, in step S7, the processing unit CP (e.g., the processing unit CP

[13] ) may perform the second cooling process on the substrate W.

[0093] 3 (for example, the processing unit EEW[1]) performs an edge exposure process on the substrate W. That is, the processing unit EEW performs an exposure process on the peripheral portion of the substrate W after the second cooling process.

[0094] Next, in step S9, the processing unit P-CP in Fig. 5 performs a cooling process on the substrate W. That is, the processing unit P-CP cools the substrate W after the edge exposure process.

[0095] 1 performs an exposure process on the substrate W. That is, the exposure device 15 exposes the substrate W after the cooling process, that is, the substrate W on which the resist film has been formed.

[0096] 7, in step S11, the processing unit PHP in FIG. 3 (for example, the processing unit PHP

[31] ) performs a third heating process on the substrate W. That is, the processing unit PHP heats the substrate W after the exposure process.

[0097] Next, in step S12, the processing unit CP (for example, the processing unit CP

[14] ) performs a third cooling process on the substrate W. That is, the processing unit PHP cools the substrate W after the third heating process.

[0098] 2 (for example, the processing unit DV

[15] ) performs a development process on the substrate W. That is, the processing unit DV supplies a developer to the substrate W after the third cooling to develop the substrate W.

[0099] 3 (for example, the processing unit PHP

[27] ) performs a fourth heating process on the substrate W. That is, the processing unit PHP heats the substrate W after the development process.

[0100] Next, in step S15, the processing unit CP (for example, the processing unit CP

[15] ) performs a fourth cooling process on the substrate W. That is, the processing unit CP cools the substrate W after the fourth heating process.

[0101] Then, the transport mechanisms 138, 128, and 115 in FIG. 5 unload the substrate W from the processing unit CP in which the fourth cooling process was performed in step S15, and load the substrate W into the carrier 113.

[0102] The processing unit PHP that performs the first heating process in step S3, the processing unit PHP that performs the second heating process in step S6, the processing unit PHP that performs the third heating process in step S11, and the processing unit PHP that performs the fourth heating process in step S14 are different from each other.

[0103] In addition, the processing unit CP that performs the first cooling process at step S4, the processing unit PASS8 that performs the second cooling process at step S7, the processing unit CP that performs the third cooling process at step S12, and the processing unit CP that performs the fourth cooling process at step S15 are different from each other.

[0104] Here, among steps S1 to S15, the series of steps S1 to S9 and S11 to S15 excluding step S10 are referred to as a series of steps SQ. In the first embodiment, the substrate processing apparatus 100 performs a series of steps SQ in parallel on each of a plurality of substrates W. Each of steps S1 to S9 and S11 to S15 corresponds to an example of a "processing step".

[0105] Specifically, the series of steps SQ includes a plurality of steps S1 to S9 and S11 to S15 that are executed in sequence. That is, the series of steps SQ includes a plurality of steps S1 to S9 and S11 to S15 that have different processing contents.

[0106] Next, the substrate processing apparatus 100 will be described with reference to Figures 6 to 8. Figure 8 is a block diagram showing a control device CM and a plurality of processing units 220 controlled by the control device CM.

[0107] As shown in Fig. 8, the control device CM controls a plurality of processing units 220. Then, as shown in Figs. 6 to 8, the plurality of processing units 220 execute a plurality of series of processes SQ in parallel.

[0108] Specifically, the multiple processing units 220 include two or more processing units 220 that perform the same processing content on the substrate W between the multiple series of processes SQ. "Having the same processing content" indicates that the processing function and processing conditions are the same. For example, one of the processes S1 to S9 and S11 to S15 shown in Figs. 6 and 7 is described as "process SP". In this case, the process SP is performed in parallel by the two or more processing units 220. In this case, the processing content of the two or more processing units 220 that perform the same process SP in parallel is the same. For example, the multiple processing units 220 include two or more processing units 220 that perform two or more processes S1 included in the multiple series of processes SQ, respectively.

[0109] 8, the control device CM controls each transport mechanism 210. Then, each transport mechanism 210 carries the substrate W into and out of each processing unit 220. Then, each of the processing units 220 processes the substrates W.

[0110] The substrate processing apparatus 100 may include one or more measurement units 200. In this case, for example, the measurement units 200 are arranged in the transfer part 112, the transfer part 122 (upper transfer chamber 125, lower transfer chamber 126), the transfer part 132 (upper transfer chamber 135, lower transfer chamber 136), the transfer part 163, the carry-in / carry-out block 14B, the coating processing part 121 (coating processing chambers 21-24), the development processing part 131 (development processing chambers 31-34), the processing unit PAHP, the processing unit PHP, the processing unit CP, or the processing unit EEW.

[0111] Furthermore, the measurement unit or units 200 do not necessarily constitute a part of the substrate processing apparatus 100. In other words, the measurement unit or units 200 may be disposed outside the substrate processing apparatus 100.

[0112] The measuring unit 200 measures the state of the substrate W after processing by the processing unit 220. For example, the measuring unit 200 measures the state at a plurality of measurement points on the surface of the substrate W after processing by the processing unit 220. The measuring unit 200 transmits measurement data ME2 indicating the measurement results of the state of the substrate W to the control device CM. For example, the measurement data ME2 includes a plurality of measurement values ​​indicating the respective states of the plurality of measurement points on the surface of the substrate W. The measuring unit 200 includes, for example, a sensor that measures the state of the substrate W and a transmitter that transmits the measurement data ME2 to the control device CM. According to the first embodiment, the substrate processing apparatus 100 includes the measuring unit 200, and thus the measurement data ME2 can be easily acquired.

[0113] For example, the one or more measuring units 200 measure the state of the substrate W after processing by the processing unit 220 for each of one or more steps selected from steps S1 to S9 and S11 to S15 shown in FIGS.

[0114] For example, one of the multiple measurement units 200 measures the contact angle of the processing liquid on the surface of the substrate W and transmits measurement data ME2 indicating the contact angle to the control device CM. For example, the measurement unit 200 measures the contact angle of the processing liquid at multiple measurement points on the surface of the substrate W. The measurement unit 200 measures the contact angle by, for example, the θ / 2 method, the tangent method, or the sessile drop method. For example, the measurement unit 200 measures the contact angle of the processing liquid on the surface of the substrate W after the adhesion reinforcement process in step S1 of FIG. 6.

[0115] For example, one of the multiple measurement units 200 measures the film thickness of a film formed on the substrate W and transmits measurement data ME2 indicating the film thickness to the control device CM. For example, the measurement unit 200 measures the film thickness at multiple measurement points on the surface of the substrate W. The measurement unit 200 measures the film thickness by, for example, spectroscopic interferometry. For example, the measurement unit 200 measures the film thickness of the resist film on the substrate W after the resist film formation process in step S5 of FIG. 6.

[0116] For example, one of the multiple measuring units 200 measures the temperature during the heat treatment on the substrate W and transmits measurement data ME2 indicating the temperature to the control device CM. For example, the measuring unit 200 measures the temperature during the heat treatment at multiple measurement points on the substrate W. The measuring unit 200 measures the temperature using, for example, a substrate W with a built-in temperature sensor. For example, the measuring unit 200 measures the temperature of the substrate W during the third heating treatment (during the heating treatment after exposure) in step S11 of FIG. 7.

[0117] For example, one of the multiple measurement units 200 measures the line width (CD: Critical Dimension) of a pattern on the surface of the substrate W and transmits measurement data ME2 indicating the line width to the control device CM. For example, the measurement unit 200 measures the line width of the pattern at multiple measurement points on the surface of the substrate W. The "line width of the pattern" measured by the measurement unit 200 is, for example, the "line width of a resist pattern." The measurement unit 200 is, for example, a CD-SEM or a CD-AFM. For example, the measurement unit 200 measures the line width of the resist pattern on the surface of the substrate W after the development process in step S13 of FIG. 7.

[0118] For example, one of the plurality of measuring units 200 measures the number of particles on the surface of the substrate W and transmits measurement data ME2 indicating the number of particles to the control device CM. For example, the measuring unit 200 measures the number of particles at a plurality of measurement points on the surface of the substrate W.

[0119] For example, one of the multiple measurement units 200 measures the number of pattern collapses on the surface of the substrate W and transmits measurement data ME2 indicating the number of collapses to the control device CM. For example, the measurement unit 200 measures the number of pattern collapses at multiple measurement points on the surface of the substrate W.

[0120] For example, one of the multiple measuring units 200 measures the cut width of the edge of the substrate W and transmits measurement data ME2 indicating the cut width to the control device CM. For example, the measuring unit 200 measures the cut width of the edge of the substrate W at multiple measurement points along the circumferential direction.

[0121] The control device CM includes a control unit CNT, a memory unit MEM, an input unit IN, a display unit DP and a communication unit TR.

[0122] The control unit CNT controls the multiple processing units 220 and each transport mechanism 210. Specifically, the control unit CNT controls one or more processing units 220 and each transport mechanism 210 so that each of the target substrates Wx is processed by one or more processing units 220 selected from the multiple processing units 220 based on the processing results of the multiple substrates Wp by each of the multiple processing units 220. Therefore, according to the first embodiment, even if the processing characteristics of the multiple processing units 220 vary, a processing unit 220 with good processing results can be selected and used. As a result, the uniformity of processing among the multiple target substrates Wx can be improved.

[0123] "Processing results of the multiple substrates Wp by each of the multiple processing units 220" refers to past processing results of the substrate Wp prior to processing the target substrate Wx. The target substrate Wx is the substrate W that is to be processed by the series of processes SQ. Each of the processes S1 to S9 and S11 to S15 included in the series of processes SQ corresponds to an example of a "process for processing the target substrate."

[0124] In this specification, when there is no need to distinguish between a previously processed substrate Wp and a substrate Wx to be processed, the substrate Wp and the substrate Wx to be processed are simply referred to as "substrate W".

[0125] The control unit CNT includes a processor such as a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The memory unit MEM includes a storage device and stores data and computer programs. The memory unit MEM includes, for example, a main storage device such as a semiconductor memory, and an auxiliary storage device such as a semiconductor memory and a hard disk drive. The memory unit MEM may include a removable medium such as an optical disk. The memory unit MEM is, for example, a non-transitory computer-readable storage medium.

[0126] The processor of the control unit CNT executes a computer program stored in the storage device of the memory unit MEM to control the memory unit MEM, the input unit IN, the display unit DP, the communication unit TR, each of the measurement units 200, each of the transport mechanisms 210, and each of the processing units 220.

[0127] The communication unit TR is connected to a network and communicates with external devices. The network includes, for example, the Internet, a Local Area Network (LAN), a public telephone network, and a short-range wireless network. The communication unit TR is a communicator, for example a network interface controller.

[0128] The input unit IN is an input device for inputting various information to the control unit CNT. For example, the input unit IN is a keyboard and pointing device, or a touch panel.

[0129] The display unit DP displays an image and is, for example, a liquid crystal display or an organic electroluminescence display.

[0130] Specifically, the control unit CNT executes the control program ME1 to function as a pre-processing unit CN1, an acquisition unit CN2, a first selection unit CN3, an analysis unit CN4, an adjustment unit CN5, a second selection unit CN6, a substrate path determination unit CN7, and a substrate processing controller CN8. That is, the control unit CNT includes a pre-processing unit CN1, an acquisition unit CN2, a first selection unit CN3, an analysis unit CN4, an adjustment unit CN5, a second selection unit CN6, a substrate path determination unit CN7, and a substrate processing controller CN8. The first selection unit CN3 corresponds to an example of a "selection unit." The memory unit MEM stores the control program ME1.

[0131] The acquiring unit CN2 acquires the measurement data ME2 from each measuring unit 200. When the measuring unit 200 is disposed outside the substrate processing apparatus 100, the acquiring unit CN2 acquires the measurement data ME2 from the measuring unit 200 via the communication unit TR.

[0132] The memory unit MEM stores a plurality of measurement data ME2 acquired from the plurality of measuring units 200, respectively.

[0133] Pre-processing unit CN1 processes the measurement data ME2 to calculate processing result information ME3. The processing result information ME3 is generated for each substrate W and indicates the processing result of the substrate W by the processing unit 220. Pre-processing unit CN1 processes each measurement data ME2 for each processing unit 220 to calculate processing result information ME3. The memory unit MEM stores the processing result information ME3. Hereinafter, the processing by pre-processing unit CN1 may be referred to as "pre-processing".

[0134] For example, pre-processing unit CN1 performs statistical processing on multiple measurement values ​​indicating the respective states of multiple measurement points on the surface of substrate W, and calculates a numerical value indicating the state of substrate W for each substrate W. The "numerical value indicating the state of substrate W" obtained by statistical processing is processing result information ME3. In this case, for example, pre-processing unit CN1 calculates an average value of multiple measurement values ​​for substrate W. The "average value" is an example of a "numerical value indicating the state of substrate W" obtained by statistical processing, and is processing result information ME3.

[0135] Furthermore, for example, pre-processing unit CN1 calculates a numerical value indicating the variation of a plurality of measured values ​​for substrate W. The "numerical value indicating the variation" is an example of a "numerical value indicating the state of substrate W" obtained by statistical processing, and is processing result information ME3. The numerical value indicating the variation may be, for example, a difference value between the maximum and minimum values ​​of the plurality of measured values, a standard deviation σ of the plurality of measured values, a numerical value based on the standard deviation σ, a variance σ of the plurality of measured values, 2 , or variance σ 2 The value based on the standard deviation σ is, for example, a 3σ value.

[0136] Furthermore, for example, pre-processing unit CN1 normalizes a plurality of measurement values ​​for substrate W to make the plurality of measurement values ​​dimensionless. The "normalized plurality of measurement values" is an example of a "numerical value indicating the state of substrate W" obtained by statistical processing, and is processing result information ME3.

[0137] The preprocessing unit CN1 can calculate multiple types of processing result information ME3 from the same measurement data ME2. The multiple types of processing result information ME3 include, for example, an average value, a difference between a maximum value and a minimum value, a standard deviation σ, a value based on the standard deviation σ, a variance σ, etc. 2 , and variance σ 2 It is a number equal to or greater than two of the numbers based on the above.

[0138] Furthermore, the measurement data ME2 itself may be the processing result information ME3. For example, the processing result information ME3 is map data indicated by the measurement data ME2. The map data indicates a distribution of a plurality of measurement values ​​indicating the respective states of a plurality of measurement points on the surface of the substrate W, or a distribution of a plurality of measurement values ​​indicating the respective states of a plurality of measurement points on the substrate W.

[0139] For example, the processing result information ME3 indicates at least one of the contact angle θ of the processing liquid on the surface of the substrate W, a numerical value based on the contact angle θ, the film thickness t of the film formed on the substrate W, a numerical value based on the film thickness t, the temperature T after heat treatment on the surface of the substrate W, a numerical value based on the temperature T, the line width W1 of the pattern on the surface of the substrate W, a numerical value based on the line width W1, the number of particles NM1 on the surface of the substrate W, a numerical value based on the number of particles NM1, the number of collapses NM2 of the pattern on the surface of the substrate W, a numerical value based on the number of collapses NM2, and the cut width W2 of the edge of the substrate W. Therefore, according to the first embodiment, the processing result information ME3 can accurately indicate the state of the surface of the substrate W.

[0140] Specifically, the processing result information ME3 is, for example, contact angle map data, film thickness map data, temperature map data, pattern line width map data, particle map data, or collapse map data. The contact angle map data indicates the distribution of the contact angle θ with respect to the processing liquid on the surface of the substrate W. The film thickness map data indicates the distribution of the film thickness t of the substrate W. The temperature map data indicates the distribution of the temperature T of the substrate W. The pattern line width map data indicates the distribution of the line width W1 of the pattern on the surface of the substrate W. The particle map data indicates the distribution of the particle number NM1 on the surface of the substrate W. The collapse map data indicates the distribution of the number of pattern collapses NM2 on the surface of the substrate W.

[0141] The "numeric value based on contact angle θ" as the processing result information ME3 is, for example, a numerical value or a normalized value obtained by performing statistical processing on a plurality of contact angles θ at a plurality of measurement points on the surface of the substrate W. The "numeric value based on film thickness t" as the processing result information ME3 is, for example, a numerical value or a normalized value obtained by performing statistical processing on a plurality of film thicknesses t at a plurality of measurement points on the surface of the substrate W. The "numeric value based on temperature T" as the processing result information ME3 is, for example, a numerical value or a normalized value obtained by performing statistical processing on a plurality of temperatures T at a plurality of measurement points on the surface of the substrate W. The "numeric value based on line width W1" as the processing result information ME3 is, for example, a numerical value or a normalized value obtained by performing statistical processing on a plurality of line widths W1 at a plurality of measurement points on the surface of the substrate W. The "numeric value based on particle number NM1" as the processing result information ME3 is, for example, a numerical value or a normalized value obtained by performing statistical processing on the particle number NM1 at each measurement point on the surface of the substrate W. The "numeric value based on the number of collapses NM2" as the processing result information ME3 is, for example, a numerical value or a normalized value obtained by performing statistical processing on the number of collapses NM2 at each measurement point on the surface of the substrate W. In these cases, the "numeric value" is, for example, an average value or a numerical value indicating variation.

[0142] The processing result information ME3 may be stored in an external device of the substrate processing apparatus 100. In this case, the acquisition unit CN2 acquires the processing result information ME3 indicating the processing results of the substrate W by each of the multiple processing units 220 from the external device via the communication unit TR and the network, and stores the processing result information ME3 in the memory unit MEM. The external device is, for example, a server or a storage device.

[0143] Furthermore, the acquiring unit CN2 acquires the processing result information ME3 from the storage unit MEM, and causes the first selecting unit CN3 to select one or more processing units 220 to be used for processing each of the processing target substrates Wx.

[0144] The first selection unit CN3 has at least one analysis algorithm. In the first embodiment, the first selection unit CN3 has a plurality of analysis algorithms. The analysis algorithm is an algorithm for analyzing the processing result information ME3 when selecting one or more processing units 220 to be used for processing each of the processing target substrates Wx. The analysis algorithm is prepared, for example, for each type of processing result information ME3. For example, the analysis algorithm indicates the type of processing result information ME3 to be analyzed and a method of analyzing the processing result information ME3. The analysis method includes a selection condition for the processing unit 220. The analysis algorithm is incorporated, for example, in the control program ME1.

[0145] The first selection unit CN3 selects one or more processing units 220 to be used to process each of the target substrates Wx from among the multiple processing units 220 based on multiple pieces of processing result information ME3 indicating the processing results of the multiple substrates Wp by the multiple processing units 220, and generates selection result information ME4 indicating the selection result. Therefore, according to the first embodiment, a processing unit 220 with good processing results can be selected, and the quality of the target substrates Wx after processing can be improved. The control device CM can be considered as a "selection device."

[0146] For example, the selection result information ME4 indicates one or more processing units 220 to be used to process each of the processing target substrates Wx, among the plurality of processing units 220.

[0147] It is preferable that the first selection part CN3 selects two or more processing units 220 to be used for processing each of the processing target substrates Wx from among the multiple processing units 220, based on multiple pieces of processing result information ME3 indicating processing results of the multiple substrates Wp by the multiple processing units 220. This is because the throughput of the processing of the substrates W can be improved.

[0148] Moreover, the first selection unit CN3 sets the processing units 220 that were not selected from among the multiple processing units 220 as processing units 220 to be unused. In other words, the first selection unit CN3 determines the processing units 220 to be unused from among the multiple processing units 220 based on the multiple pieces of processing result information ME3 corresponding to the multiple processing units 220, respectively. In this case, the selection result information ME4 includes not only information indicating one or more processing units 220 to be used for processing each of the processing target substrates Wx, but also information indicating the processing units 220 to be unused. According to the first embodiment, by determining the processing units 220 to be unused, it is possible to suppress a decrease in uniformity of processing among the multiple substrates W caused by the processing units 220 having poor processing characteristics.

[0149] The first selection unit CN3 may select one or more processing units 220 to be used for processing each of the target substrates Wx from among the multiple processing units 220, with a priority order of use, based on multiple pieces of processing result information ME3 indicating respective processing results of the multiple substrates Wp by the multiple processing units 220. Therefore, according to the first embodiment, by increasing the frequency of use of the processing units 220 with higher priority, the quality after processing of the target substrates Wx can be further improved.

[0150] For example, the first selection part CN3 assigns a higher priority to a processing unit 220 when the post-processing quality of the substrate Wp indicated by the processing result information ME3 is higher.

[0151] For example, the first selection unit CN3 prioritizes use and selects all or some of the processing units 220 based on a plurality of pieces of processing result information ME3 corresponding to the plurality of processing units 220. In this case, the selection result information ME4 includes information indicating the use priority. Even in this case, the selection result information ME4 may include information indicating the processing units 220 that are not to be used.

[0152] For example, the first selection part CN3 compares the processing result information ME3 with a threshold value, and selects one or more processing units 220 to be used for processing each of the processing target substrates Wx based on the comparison result. The "threshold value" is, for example, a specification value required for the substrate processing apparatus 100.

[0153] For example, for processing result information ME3 in which the smaller the numerical value, the better the quality (e.g., a numerical value indicating variation), the first selection unit CN3 increases the priority of the processing unit 220 corresponding to the processing result information ME3 as the numerical value of the processing result information ME3 decreases. Also, for example, for processing result information ME3 in which the larger the numerical value, the better the quality, the first selection unit CN3 increases the priority of the processing unit 220 corresponding to the processing result information ME3 as the numerical value of the processing result information ME3 increases.

[0154] For example, the first selection unit CN3 judges whether or not a numerical value (e.g., an average value) indicated by the processing result information ME3 is within a predetermined range, and selects one or more processing units 220 to be used for processing each of the processing target substrates Wx based on the judgment result. The predetermined range includes a reference value. In this case, for example, when the numerical value indicated by the processing result information ME3 is within the predetermined range, the first selection unit CN3 selects the processing unit 220 corresponding to the processing result information ME3 as a processing unit to be used, and when the numerical value indicated by the processing result information ME3 is outside the predetermined range, the first selection unit CN3 sets the processing unit 220 corresponding to the processing result information ME3 as a processing unit not to be used. Also, for example, the closer the numerical value indicated by the processing result information ME3 is to the reference value, the higher the priority of the processing unit 220 corresponding to the processing result information ME3. The "predetermined range" is, for example, a range indicated by the specifications required for the substrate processing apparatus 100. The "reference value" is, for example, a median value of the predetermined range.

[0155] The first selection part CN3 generates selection result information ME4 for each analysis algorithm based on each processing result information ME3 prepared for each analysis algorithm. The storage unit MEM stores each selection result information ME4.

[0156] Based on the selection result information ME4, the first selection unit CN3 sets the target substrates Wx to be processed by the processing units 220 with a lower priority than a predetermined priority as the target substrates for quality inspection. Therefore, according to the first embodiment, the time required for the quality inspection can be shortened compared to the case where all the target substrates Wx are subjected to the quality inspection. In addition, by only subjecting the target substrates Wx to be processed by the processing units 220 with a relatively low priority to the quality inspection, the quality inspection can be performed efficiently.

[0157] The predetermined priority is a priority of use, and may be, for example, a priority one rank higher than the lowest priority, or a priority N ranks higher than the lowest priority, where N is an integer equal to or greater than 2.

[0158] Specifically, the first selection part CN3 determines, from the plurality of processing units 220, a processing unit 220 in which the processed substrate W will be subjected to quality inspection, based on the selection result information ME4.

[0159] For example, the first selection unit CN3 generates inspection target information ME5 indicating the processing units 220 with a lower priority than a predetermined priority based on the selection result information ME4. That is, the inspection target information ME5 indicates the processing units 220 in which the processed substrate W will be subjected to quality inspection.

[0160] The first selection unit CN3 may assign a priority to the quality inspection and determine, from the multiple processing units 220, a processing unit 220 that will be the target of the quality inspection of the processed substrate W. In this case, the first selection unit CN3 assigns a higher priority to the quality inspection as the priority of use becomes lower. In this case, for example, the inspection target information ME5 includes information indicating the priority of the quality inspection.

[0161] The multiple processing units 220 include a predetermined number of processing units 220. The predetermined number is a number equal to or greater than two. The first selection unit CN3 may select one or more processing units 220 to be used for processing each of the processing target substrates Wx from among the predetermined number of processing units 220, based on a predetermined number of pieces of processing result information ME3 indicating processing results of the predetermined number of substrates Wp by the predetermined number of processing units 220. In this case, the second selection unit CN6 selects whether to use one or more processing units 220 selected from the predetermined number of processing units 220, or to use all of the predetermined number of processing units 220.

[0162] That is, when it is selected to use one or more processing units 220 selected from the predetermined number of processing units 220, priority is given to the quality of the processed substrate W (processing quality). On the other hand, when it is selected to use all of the predetermined number of processing units 220, priority is given to the throughput (processing speed) of the processing of the substrate W. Therefore, according to the first embodiment, it is possible to select whether to prioritize the quality of the processed substrate W or the throughput of the processing of the substrate W, depending on the user's request. As a result, user convenience can be improved.

[0163] The memory unit MEM stores priority order information ME6 when priority is given to throughput of processing of substrates W. The priority order information ME6 indicates the priority of use for all of the predetermined number of processing units 220. The priority order information ME6 indicates, for example, consecutive numbers assigned to all of the predetermined number of processing units 220.

[0164] Next, a substrate processing method according to the first embodiment will be described with reference to Fig. 9 and Fig. 10. Fig. 9 is a flowchart showing the substrate processing method according to the first embodiment. The substrate processing method is performed by the substrate processing apparatus 100. Specifically, the control unit CNT shown in Fig. 8 executes the substrate processing method. As shown in Fig. 9, the substrate processing method includes steps S31 to S36.

[0165] In step S31, the control unit CNT selects one or more processing units 220 to be used for processing each of the processing target substrates Wx from among the predetermined number of processing units 220, and generates selection result information ME4 indicating the selection result. In the first embodiment, the predetermined number of processing units 220 are two or more processing units 220 having the same processing content.

[0166] Next, in step S32, the second selection part CN6 of the control unit CNT selects whether to perform processing with priority given to throughput on each substrate W, or to perform processing with priority given to the quality of the substrate W after processing.

[0167] If it is selected in step S32 that throughput-priority processing be performed, the process proceeds to step S33.

[0168] In step S33, the substrate path determination unit CN7 of the control unit CNT determines a plurality of transport paths for each of the substrates W so that all of the predetermined number of processing units 220 are used in parallel. In this case, the substrate path determination unit CN7 determines each transport path so that the processing unit 220 with a higher priority is used more frequently, for example, based on the priority of use indicated by the priority information ME6. Specifically, the substrate path determination unit CN7 determines each transport path so that not only the predetermined number of processing units 220 but also two or more other processing units 220 are used in parallel. Each transport path includes a plurality of processing units 220 that perform steps S1 to S9 and S11 to S15 shown in FIG. 6 and FIG. 7, respectively.

[0169] Next, in step S34, the substrate processing control unit CN8 of the control unit CNT controls each transport mechanism 210 and each processing unit 220 to transport and process each substrate W according to each transport path determined in step S33. As a result, all of the predetermined number of processing units 220 process substrates W in parallel, and two or more other processing units 220 process substrates W in parallel. When processing of the desired number of substrates W is completed, the processing ends.

[0170] On the other hand, if quality-priority processing is selected in step S32, the process proceeds to step S35.

[0171] In step S35, the substrate path determination unit CN7 determines a plurality of transport paths for each of the substrates W so that each processing unit 220 selected from a predetermined number of processing units 220 is used in parallel based on the selection result information ME4. In this case, the substrate path determination unit CN7 determines each transport path so that the processing unit 220 with a higher priority is used more frequently, for example, based on the priority of use indicated by the selection result information ME4. Specifically, the substrate path determination unit CN7 determines each transport path so that not only each processing unit 220 selected from the predetermined number of processing units 220, but also two or more other processing units 220 are used in parallel. Each transport path includes a plurality of processing units 220 that perform steps S1 to S9 and S11 to S15 shown in FIG. 6 and FIG. 7, respectively.

[0172] Next, in step S36, the substrate processing control unit CN8 controls each transport mechanism 210 and each processing unit 220 to transport and process each substrate W according to each transport path determined in step S35. As a result, each processing unit 220 selected from the predetermined number of processing units 220 processes the substrate W in parallel, and two or more other processing units 220 process the substrate W in parallel. When the processing of the desired number of substrates W is completed, the processing ends.

[0173] Fig. 10 is a flowchart showing the process of step S31 shown in Fig. 9. As shown in Fig. 10, the process of step S31 includes steps S51 to S61. The process of step S31 corresponds to an example of a "selection method".

[0174] In step S51, the preprocessing unit CN1 of the control unit CNT processes a plurality of pieces of measurement data ME2 to generate a plurality of pieces of processing result information ME3.

[0175] Next, in step S52, the acquisition unit CN2 acquires two or more pieces of processing result information ME3 indicating the processing results of two or more substrates Wp by two or more processing units 220 having the same processing contents.

[0176] Next, in step S53, the first selection part CN3 selects one or more processing units 220 to be used for processing each of the target substrates Wx from among the two or more processing units 220 (specifically, a predetermined number of processing units 220) based on two or more pieces of processing result information ME3 indicating processing results of two or more substrates Wp by two or more processing units 220 having the same processing contents. Therefore, according to the first embodiment, a processing unit 220 with good processing results can be selected from among two or more processing units 220 having the same processing contents, and the quality of the target substrates Wx after processing can be improved.

[0177] Specifically, the first selection unit CN3 analyzes the processing result information ME3 using an analysis algorithm, selects one or more processing units 220 to be used to process each of the target substrates Wx from among two or more processing units 220 having the same processing content, and generates selection result information ME4.

[0178] Next, in step S54, the first selection part CN3 determines whether or not the selection result information ME4 includes information indicating a processing unit 220 to be unused.

[0179] If the determination in step S54 is negative, the process proceeds to step S59.

[0180] On the other hand, if the determination in step S54 is affirmative, the process proceeds to step S55.

[0181] In step S55, the first selection part CN3 judges whether or not a request to adjust the unused processing unit 220 has been received.

[0182] If the determination in step S55 is negative, the process proceeds to step S59.

[0183] On the other hand, if the determination in step S55 is affirmative, the process proceeds to step S56.

[0184] In step S56, the analysis unit CN4 analyzes the processing result information ME3 indicating the processing result of the substrate Wp by the processing unit 220 that is not to be used.

[0185] Next, in step S57, the adjustment unit CN5 adjusts the unused processing unit 220 based on the analysis result of the processing result information ME3 by the analysis unit CN4, and returns the unused processing unit 220 to the processing unit 220 to be used. Therefore, according to the first embodiment, in the case where priority is given to quality, it is possible to increase the number of usable processing units 220. As a result, it is possible to improve the throughput of the processing of the substrates Wx to be processed while improving the quality after processing of the substrates Wx to be processed.

[0186] Next, in step S58, the first selection part CN3 updates the selection result information ME4 so that the processing unit 220 restored by the adjustment part CN5 is selected for processing the processing target substrate Wx.

[0187] Next, in step S59, the first selection part CN3 sets, as a target for quality inspection, the processing target substrate Wx to be processed by the processing unit 220 having a lower priority than a predetermined priority, based on the selection result information ME4.

[0188] Next, in step S60, the first selection unit CN3 notifies the selection result information ME4. For example, the first selection unit CN3 displays the selection result information ME4 on the display unit DP. For example, the first selection unit CN3 transmits the selection result information ME4 to an external device via the communication unit TR. The external device is, for example, a terminal device of an administrator of the substrate processing apparatus 100.

[0189] Next, in step S61, the first selection unit CN3 determines whether or not the processes based on all the analysis algorithms have been completed.

[0190] If a negative determination is made in step S61, the process proceeds to step S52. Then, in step S52, the acquisition unit CN2 acquires each processing result information ME3 corresponding to an analysis algorithm different from the previous analysis algorithm. Also, in step S53, the first selection unit CN3 analyzes the processing result information ME3 by an analysis algorithm different from the previous analysis algorithm, and selects one or more processing units 220 to be used for processing each of the processing target substrates Wx from two or more processing units 220 having the same processing contents.

[0191] On the other hand, if the determination in step S61 is affirmative, the process returns to the main routine in FIG. 9 and proceeds to step S32.

[0192] As described above with reference to Fig. 10, according to the first embodiment, the first selection unit CN3 analyzes each corresponding processing result information ME3 using a plurality of mutually different analysis algorithms, and selects, for each analysis algorithm, one or more processing units 220 to be used for processing each of the processing target substrates Wx from among two or more processing units 220 having the same processing content. Therefore, the first selection unit CN3 generates selection result information ME4 for each analysis algorithm. As a result, according to the first embodiment, it is possible to select the processing unit 220 to be used based on the selection result information ME4 desired by the user from among the plurality of selection result information ME4.

[0193] When a plurality of pieces of selection result information ME4 are stored in the storage unit MEM, the substrate path determination unit CN7 determines the transport path of each substrate W based on, for example, one piece of selection result information ME4 out of the plurality of pieces of selection result information ME4.

[0194] The "processing contents" in "two or more processing units 220 having the same processing contents" may be, for example, the processing contents in any of steps S1 to S9 and S11 to S15 constituting the series of steps SQ in FIGS.

[0195] Next, the selection result information ME4 will be described with a specific example with reference to Fig. 11. In addition, ten processing units PAHP are taken as an example of two or more processing units 220 having the same processing content. The processing units PAHP perform an adhesion reinforcement process (step S1 in Fig. 1) on a substrate W. For ease of explanation, identification information U1 to U10 is assigned to the ten processing units PAHP, respectively.

[0196] FIG. 11 is a diagram showing an example of the processing result information ME3, the priority order information ME6, the selection result information ME4, and the inspection target information ME5.

[0197] 11 shows, as a first example of the processing result information ME3, the average value of the contact angle (degrees) on the surface of the substrate W. Hereinafter, the average value of the contact angle may be referred to as the average contact angle. Furthermore, as a second example of the processing result information ME3, the difference value between the maximum and minimum values ​​of the contact angle (degrees) on the surface of the substrate W is shown. Furthermore, as a third example of the processing result information ME3, a 3σ value calculated from the distribution of the contact angle (degrees) on the surface of the substrate W is shown. In the first to third examples, the contact angle shows the contact angle with respect to the processing liquid for the anti-reflective coating.

[0198] Moreover, priority information A1 is shown as an example of priority information ME6 in the case where throughput is prioritized.

[0199] When throughput is prioritized, all 10 processing units U1 to U10 are used. The priority order information A1 when throughput is prioritized indicates the processing units U1 to U10 in descending order of priority for use. Therefore, for example, when 25 substrates W are to be processed, each of the processing units U1 to U5 processes three substrates W, and each of the processing units U6 to U10 processes two substrates W.

[0200] Moreover, selection results information A2 is shown as an example of selection results information ME4 in the case where priority is given to quality.

[0201] In the selection result information A2 in the case of quality priority, all ten processing units U1 to U10 are selected. In the selection result information A2, the priority of use is determined based on the 3σ value. Specifically, in the selection result information A2, the smaller the 3σ value, the higher the priority assigned to the processing units U1 to U10. In other words, the selection result information A2 shows the processing units U5, U6, U2, U3, U7, U4, U8, U1, U9, and U10 in order of priority of use.

[0202] Therefore, for example, when processing 25 substrates W, the processing units U5, U6, U2, U3, and U7, which have the smallest 3σ values, each process three substrates W, and the processing units U8, U1, U9, and U10, which have the largest 3σ values, each process two substrates W.

[0203] Moreover, selection results information A3 is shown as an example of selection results information ME4 in the case where priority is given to quality.

[0204] In the selection result information A3 in the case where quality is prioritized, the processing units U8, U5, U9, U1, U5, and U2 that processed substrates W with average contact angles within a predetermined range are selected. In the example of Fig. 11, the predetermined range is "59±0.5". "59" is a reference value.

[0205] In the selection result information A3, the priority order of use is determined based on the average contact angle. Specifically, in the selection result information A3, the closer the average contact angle is to "59", the higher the priority order is assigned to the processing units U8, U5, U9, U1, U5, and U2. In other words, the selection result information A3 indicates the processing units U8, U5, U9, U1, U5, and U2 in order of priority of use.

[0206] Therefore, for example, when processing 25 substrates W, the number of substrates W processed by processing unit U8 is 5, and the number of substrates W processed by processing units U5, U9, U1, U5, and U2 is 4, in order of average contact angle closest to the reference value "59".

[0207] On the other hand, in the selection result information A3, the processing units U10, U7, U3, and U4 that processed the substrates W with the average contact angles outside the predetermined range are not selected and are set as targets for non-use. In Fig. 11, the letters "NU" indicate that they are targets for non-use.

[0208] Furthermore, the inspection target information ME5 indicates that the substrate W processed by the processing unit U2 is set as the target of the quality inspection. That is, the inspection target information ME5 indicates the processing unit U2 in which the processed substrate W is to be the target of the quality inspection. In the example of Fig. 11, the inspection target information ME5 sets the processing unit U2 that processed the substrate W whose average contact angle is the most deviant from the reference value "59" as the processing unit U2 that processes the substrate W that is to be the target of the quality inspection in relation to the selection result information A3. That is, in the inspection target information ME5, the substrate W that is processed by the processing unit U2 that has the lowest priority of use among the processing units U8, U5, U9, U1, U5, and U2 selected as the target of use in the selection result information A3 is set as the target of the quality inspection.

[0209] (Embodiment 2) A substrate processing apparatus 100 according to a second embodiment of the present invention will be described with reference to Figs. 12 to 19. The second embodiment is mainly different from the first embodiment in that the substrate processing apparatus 100 according to the second embodiment performs machine learning. The configuration of the substrate processing apparatus 100 according to the second embodiment is similar to the configuration of the substrate processing apparatus 100 according to the first embodiment described with reference to Figs. 1 to 7. The following mainly describes the differences between the second embodiment and the first embodiment.

[0210] First, the substrate processing apparatus 100 according to the second embodiment will be described with reference to Fig. 12. Fig. 12 is a block diagram showing a control device CM of the substrate processing apparatus 100 according to the second embodiment. As shown in Fig. 12, the control unit CNT further includes a learning unit CN9. The memory unit MEM further stores a learning program ME7, a plurality of learning data ME8, and a learning model ME9.

[0211] The learning model ME9 is constructed by executing machine learning (e.g., supervised machine learning) on ​​a plurality of pieces of learning data ME8. The learning model ME9 is, for example, a trained model.

[0212] Each of the multiple learning data ME8 includes processing result information (hereinafter referred to as "processing result information F1") indicating the processing result of a learning target substrate (hereinafter referred to as "learning target substrate Wn") by the processing unit 220, and evaluation information (hereinafter referred to as "evaluation information TV1") indicating an evaluation of the processing result of the learning target substrate Wn indicated by the processing result information F1.

[0213] The multiple pieces of processing result information F1 included in the multiple pieces of learning data ME8 respectively indicate processing results by two or more processing units 220 having the same processing content. Moreover, the processing result information F1 is an explanatory variable. That is, the processing result information F1 is a feature. The configuration of the processing result information F1 is the same as the configuration of the processing result information ME3. The evaluation information TV1 is an objective variable. That is, the evaluation information TV1 is a label.

[0214] The evaluation information TV1 indicates, for example, the processing result of the learning target substrate Wn as either "normal" or "abnormal." The evaluation information TV1 may indicate, for example, the processing result of the learning target substrate Wn in a plurality of stages. The evaluation information TV1 may also be regarded as information indicating the evaluation of the processing result information F1.

[0215] The learning program ME7 is a program for executing a machine learning algorithm to find certain rules from a plurality of pieces of learning data ME8 and generate a learning model ME9 that expresses the found rules.

[0216] In embodiment 1, the machine learning algorithm is a supervised learning algorithm, for example, a decision tree, a nearest neighbor method, a naive Bayes classifier, a support vector machine, or a neural network. Therefore, the learning model ME9 includes a decision tree, a nearest neighbor method, a naive Bayes classifier, a support vector machine, or a neural network. In the machine learning that generates the learning model ME9, an error backpropagation method may be used.

[0217] For example, the neural network includes an input layer, one or more intermediate layers, and an output layer. Specifically, the neural network is a deep neural network (DNN), a recurrent neural network (RNN), or a convolutional neural network (CNN), and performs deep learning. For example, the deep neural network includes an input layer, multiple intermediate layers, and an output layer.

[0218] The learning unit CN9 performs machine learning on a plurality of pieces of learning data ME8 based on the learning program ME7. As a result, certain rules are found from the plurality of pieces of learning data ME8, and a learning model ME9 is generated.

[0219] Specifically, the learning unit CN9 finds certain rules between the processing result information F1 and the evaluation information TV1 included in the learning data ME8, and generates a learning model ME9.

[0220] More specifically, the learning unit CN9 calculates a plurality of learned parameters by machine learning a plurality of pieces of learning data ME8 based on the learning program ME7, and generates a learning model ME9 including one or more functions to which the plurality of learned parameters are applied. The learned parameters are parameters (coefficients) obtained based on the results of machine learning using the plurality of pieces of learning data ME8.

[0221] The learning model ME9 receives as input information processing result information ME3 indicating the processing result of the substrate Wp by the processing unit 220, and outputs as output information evaluation information (hereinafter referred to as "evaluation information OT1") indicating an evaluation of the processing result of the substrate Wp. In other words, the learning model ME9 evaluates the substrate Wp after processing by the processing unit 220. The configuration of the evaluation information OT1 is the same as the configuration of the evaluation information TV1.

[0222] The first selection unit CN3 inputs, as input information, each of two or more pieces of processing result information ME3 indicating the processing results of two or more substrates Wp, respectively, to the learning model ME9, and obtains, as output information, evaluation information OT1 indicating an evaluation of each of the processing results of the two or more substrates Wp, from the learning model ME9. In this case, the "processing results of two or more substrates Wp" refers to the processing results of two or more substrates Wp by two or more processing units 220 having the same processing content.

[0223] Then, the first selection unit CN3 selects one or more processing units 220 to be used for processing each of the processing target substrates Wx from among two or more processing units 220 having the same processing content, based on two or more pieces of evaluation information OT1 acquired from the learning model ME9. Therefore, according to the second embodiment, by using the learning model ME9 that has learned a plurality of learning data ME8 each including a plurality of pieces of processing result information F1, it is possible to quickly and accurately select the processing unit 220 to be used.

[0224] Next, a substrate processing method according to a second embodiment will be described with reference to Figures 13 and 14. The substrate processing method according to the second embodiment is similar to the substrate processing method according to the first embodiment described with reference to Figure 9. Below, differences between the substrate processing method according to the second embodiment and the substrate processing method of the first embodiment will be mainly described.

[0225] Fig. 13 is a flowchart showing step S31 of Fig. 9 in embodiment 2. As shown in Fig. 13, the process of step S31 according to embodiment 2 includes steps S81 to S94.

[0226] The processes in steps S81 to S91 are similar to the processes in steps S51 to S61, respectively, described with reference to Fig. 10. However, in step S83, the process shown in Fig. 14 may be performed.

[0227] Fig. 14 is a flowchart showing an example of the process of step S83 in Fig. 13. As shown in Fig. 14, the process of step S83 includes steps S111 to S114.

[0228] In step S111, the first selection unit CN3 inputs processing result information ME3 indicating the processing result of the substrate Wp as input information to the learning model ME9.

[0229] Next, in step S112, the first selection unit CN3 acquires, as output information, evaluation information OT1 indicating an evaluation of the processing result of the substrate Wp from the learning model ME9.

[0230] Next, in step S113, the first selection unit CN3 selects, based on the evaluation information OT1, one or more processing units 220 to be used to process each of the substrates Wx to be processed, from among two or more processing units 220 having the same processing content. For example, the first selection unit CN3 selects, as the processing unit to be used, the processing unit 220 that processed the substrate Wp for which the evaluation information OT1 indicates "normal", and sets, as the processing unit not to be used, the processing unit 220 that processed the substrate Wp for which the evaluation information OT1 indicates "abnormal".

[0231] Next, in step S114, the first selection unit CN3 determines whether or not the processing has been completed for all the processing result information ME3.

[0232] If a negative determination is made in step S114, the process proceeds to step S111.

[0233] On the other hand, if an affirmative decision is made in step S114, the first selection unit CN3 outputs selection result information ME4, and the process returns to the routine shown in FIG. 13 and proceeds to step S84.

[0234] 13, in step S92, the first selection unit CN3 selects one or more processing units 220 to be used for processing each of the plurality of processing target substrates Wx, based on the plurality of different selection result information ME4 obtained by the plurality of different analysis algorithms. For example, the first selection unit CN3 processes the plurality of different selection result information ME4 obtained by the plurality of different analysis algorithms, using an AND condition, to generate new selection result information ME4.

[0235] Next, in step S93, the first selection unit CN3 sets, as a target for quality inspection, the processing target substrate Wx to be processed by the processing unit 220 having a lower priority than the predetermined priority based on the selection result information ME4 generated in step S92. This is the same as in step S89.

[0236] Next, in step S94, the first selection unit CN3 notifies the selection result information ME4 generated in step S92. This is the same as in step S90. After completing step S94, the process returns to the main routine in Fig. 9 and proceeds to step S32.

[0237] The "processing contents" in "two or more processing units 220 having the same processing contents" may be, for example, the processing contents in any of steps S1 to S9 and S11 to S15 constituting the series of steps SQ in FIGS.

[0238] Next, the learning unit CN9 will be described with reference to Fig. 12 and Fig. 15. Fig. 15 is a flowchart showing a learning method executed by the learning unit CN9. As shown in Fig. 15, the learning method includes steps S131 to S134.

[0239] In step S131, the learning unit CN9 acquires learning data ME8.

[0240] Next, in step S132, the learning unit CN9 performs machine learning on the learning data ME8 based on the learning program ME7.

[0241] Next, in step S133, the learning unit CN9 determines whether or not a learning end condition is satisfied. The learning end condition is a condition that is determined in advance for ending the machine learning. For example, the learning end condition is that the number of iterations reaches a specified number.

[0242] If the determination in step S133 is negative, the process proceeds to step S131. As a result, the machine learning is repeated.

[0243] On the other hand, if the result of step S133 is affirmative, the process proceeds to step S134.

[0244] In step S134, the learning unit CN9 outputs a model (one or more functions) to which the latest parameters (coefficients), that is, the learned parameters (coefficients), are applied as a learning model ME9. Then, the memory unit MEM stores the learning model ME9.

[0245] As described above, the learning unit CN9 executes steps S131 to S134 to generate a learning model ME9. The learning method corresponds to one example of a "learning model generating method." The control device CM can be regarded as a "learning device."

[0246] In the repeated step S131, each of the multiple learning data ME8 includes processing result information F1 indicating the processing result of the learning target substrate Wn, and evaluation information TV1 indicating an evaluation of the processing result of the learning target substrate Wn. The processing result information F1 included in each learning data ME8 is one of two or more pieces of processing result information F1 indicating the processing results of two or more learning target substrates Wn by two or more processing units 220 having the same processing content. The learning model ME9 is a program that inputs, as input information, each of two or more pieces of processing result information ME3 indicating the processing results of two or more substrates Wp by two or more processing units 220 having the same processing content, and outputs, as output information, evaluation information OT1 indicating an evaluation of each of the processing results of the two or more substrates Wp.

[0247] In other words, the learning model ME9 causes the control device CM to function so as to evaluate the processing results of the substrate Wp. The control device CM is an example of a "computer." Specifically, the learning model ME9 causes the control device CM to function so as to input two or more pieces of input information and output two or more pieces of output information. The two or more pieces of input information are two or more pieces of processing result information ME3 each indicating the processing results of two or more substrates Wp by two or more processing units 220 having the same processing content. The two or more pieces of output information are two or more pieces of evaluation information OT1 each indicating the evaluation of the processing results of two or more substrates Wp.

[0248] Next, the selection result information ME4 will be described with specific examples with reference to Figures 16 to 19. As an example of two or more processing units 220 with the same processing content, ten processing units PAHP that perform the adhesion reinforcement processing (step S1 in Figure 6) are used, as in the case of Figure 11. For ease of explanation, identification information U1 to U10 are assigned to the ten processing units PAHP, respectively.

[0249] Fig. 16 is a diagram showing an example of the processing result information ME3, the priority order information ME6, the selection result information ME4, and the inspection target information ME5. Fig. 17 is a diagram showing contact angle map data M1 to M10 according to the second embodiment.

[0250] 16, in the second embodiment, contact angle map data M1 to M10 is shown as an example of the processing result information ME3. The contact angles represented by the contact angle map data M1 to M10 indicate the contact angles with respect to the processing liquid for an anti-reflection film.

[0251] 17 to 19, the contact angle map data M1 to M10 correspond to the processing units U1 to U10, respectively. For example, the contact angle map data M1 indicates the distribution of contact angles on the surface of the substrate W processed by the processing unit U1.

[0252] In the contact angle map data M1 to M10, the contact angle is smallest in region G1 and largest in region G5. The contact angle increases from region G1 toward region G5 in the order of region G1, region G2, region G3, region G4, and region G5. Region G1 is indicated by dense diagonal lines sloping to the upper right. Region G2 is indicated by sparse diagonal lines sloping to the lower right. Region G3 is indicated by sparse diagonal lines sloping to the upper right. Region G4 is indicated by crossed diagonal lines. Region G5 is indicated by dots. In Figs. 17 to 19, regions G1 to G5 are shown to make the drawings easier to understand, but in reality, the contact angle distribution is indicated by a gradation in the contact angle map data M1 to M10.

[0253] Returning to FIG. 16, selection results information B2 is shown as an example of selection results information ME4 in the case where priority is given to quality.

[0254] In the selection result information B2 in the case of quality priority, the processing units U2, U3, U4, U5, U6, and U7 are selected based on the evaluation information OT1 obtained by inputting the contact angle map data M1 to M10 into the learning model ME9. In Fig. 16, "○" indicates that the processing units U2, U3, U4, U5, U6, and U7 have been selected. Specifically, in the selection result information B2, the processing units U2, U3, U4, U5, U6, and U7 that processed the substrate Wp for which the evaluation information OT1 indicates "normal" are selected.

[0255] Therefore, for example, when 25 substrates W are to be processed, the processing unit U2 processes five substrates W, and each of the processing units U3, U4, U5, U6, and U7 processes four substrates W.

[0256] On the other hand, in the selection result information B2, the processing units U1, U8, U9, and U10 that processed the substrate Wp for which the evaluation information OT1 indicates "abnormal" are not selected and are set as targets for non-use. In Fig. 16, the letters "NU" indicate that they are targets for non-use.

[0257] Moreover, the priority order information B1 in the case where the throughput is prioritized is the same as the priority order information A1 in the case where the throughput is prioritized as shown in FIG.

[0258] Furthermore, the selection result information B3 in the case where the quality is prioritized is the same as the selection result information A3 in the case where the quality is prioritized shown in FIG.

[0259] Furthermore, selection result information B4 is shown as an example of selection result information ME4 in the case where priority is given to quality.

[0260] In the selection result information B4 in the case of quality priority, the result of processing the selection result information B2 and the selection result information B3 with the AND condition is shown for each of the processing units U1 to U10. For example, in the processing unit U1, when the AND condition is taken for the "unused target: NU" of the selection result information B2 and the "used target: priority order 4" of the selection result information B3, the result of the processing with the AND condition is "unused target: NU" in the selection result information B4. That is, when one or both of the selection result information B2 and the selection result information B3 indicate "unused target: NU", the result of the processing with the AND condition is "unused target: NU". On the other hand, when both the selection result information B2 and the selection result information B3 indicate "used target", that is, when both the selection result information B2 and the selection result information B3 indicate "selected", the result of the processing with the AND condition is "used target", that is, "selected" in the selection result information B4. In this case, the priority order of the selection result information B3 is adopted in the selection result information B4. This is because the selection result information B2 does not include a priority order.

[0261] In the selection result information B4, the processing units U6, U5, and U2 are selected. In the selection result information B4, the priority of use is determined based on the priority of the selection result information B3 that is the target of the processing based on the AND condition. Specifically, the selection result information B4 shows the processing units U6, U5, and U2 in order of highest priority of use according to the priority of the selection result information B3.

[0262] Therefore, for example, when 25 substrates W are to be processed, the processing unit U6 processes 9 substrates W, and the processing units U5 and U2 each process 8 substrates W.

[0263] On the other hand, in the selection result information B4, the processing units U1, U3, U4, and U7 to U10 whose processing results based on the AND condition indicate "unused" are not selected and are set as unusable. In Fig. 16, the characters "NU" indicate that they are unusable.

[0264] (Embodiment 3) A substrate processing apparatus 100 according to a third embodiment of the present invention will be described with reference to Fig. 12 and Figs. 20 to 28. The third embodiment is mainly different from the first and second embodiments in that one or more processing units 220 are selected based on processing result information ME3 from the processing units 220 performing different processing contents. The configuration of the control device CM of the substrate processing apparatus 100 according to the third embodiment is similar to the configuration of the control device CM of the substrate processing apparatus 100 according to the second embodiment described with reference to Fig. 12. Therefore, Fig. 12 will be referred to as appropriate in the description of the third embodiment. Below, the differences between the third embodiment and the first and second embodiments will be mainly described.

[0265] First, a substrate processing apparatus 100 according to a third embodiment will be described with reference to Fig. 6, Fig. 7, Fig. 12, and Fig. 20. Fig. 20 is a diagram showing a memory unit MEM and a control unit CNT of the substrate processing apparatus 100 according to the third embodiment. Fig. 20 mainly shows the portions of the substrate processing apparatus 100 according to the third embodiment that are different from the substrate processing apparatus 100 in Fig. 12.

[0266] As shown in FIGS. 6, 7, and 12, the multiple processing units 220 perform multiple series of processes SQ in parallel. Each of the multiple series of processes SQ includes multiple processes S1 to S9, S11 to S15 that are different in the processing content to be performed on the substrate W. The multiple processing units 220 include two or more processing units 220 that sequentially perform the multiple processes S1 to S9, S11 to S15 that are different in the processing content to be performed on the substrate W. "Different processing content" means that at least one of the processing function and the processing condition is different. In the third embodiment, similarly to the first embodiment, the multiple processing units 220 include two or more processing units 220 that perform the same processing content to be performed on the substrate W between the multiple series of processes SQ.

[0267] As shown in Fig. 20, in the substrate processing apparatus 100 according to the third embodiment, the storage unit MEM stores a plurality of processing result information sets ME30 instead of the plurality of processing result information ME3 in Fig. 12. The plurality of processing result information sets ME30 are obtained in a plurality of series of processes SQ, respectively. Each of the plurality of processing result information sets ME30 includes two or more pieces of processing result information ME3 indicating processing results of the substrate Wp by two or more processing units 220 having different processing contents. In other words, each of the plurality of processing result information sets ME30 includes two or more pieces of processing result information ME3 indicating processing results in two or more steps among steps S1 to S9 and S11 to S15 included in the series of processes SQ.

[0268] In the following description, a case where each of the multiple processing result information sets ME30 includes processing result information DA1, processing result information DA2, processing result information DA3, and processing result information DA4 is appropriately illustrated. Also, as targets for selection by the first selection unit CN3, processing units 220 from which the processing result information DA1 to DA3 can be obtained are appropriately illustrated.

[0269] The processing result information DA1 indicates the processing result of the substrate Wp by the processing unit 220 that performed the adhesion reinforcement processing in step S1 (FIG. 6). For example, the processing result information DA1 is contact angle map data after the adhesion reinforcement processing.

[0270] The processing result information DA2 indicates the processing result of the substrate Wp by the processing unit 220 that executed the resist film forming process in step S5 (FIG. 6). For example, the processing result information DA2 is resist film thickness map data after the resist film forming process.

[0271] The processing result information DA3 indicates the processing result of the substrate Wp by the processing unit 220 that executed the third heating process (post-exposure heating process) in step S11 (FIG. 7). For example, the processing result information DA3 is temperature map data after the third heating process.

[0272] The processing result information DA4 indicates the processing result of the substrate Wp by the processing unit 220 that executed the development processing in step S13 (FIG. 7). For example, the processing result information DA4 is resist pattern line width map data after the development processing.

[0273] 12 and 20, the first selection part CN3 selects one or more processing units 220 to be used for processing the target substrate Wx from among two or more processing units 220 having different processing contents based on the processing result information set ME30, and generates selection result information ME4 indicating the selection result. Therefore, according to the third embodiment, it is possible to select a processing unit 220 with good processing results, and to improve the quality of the target substrate Wx after processing.

[0274] For example, the selection result information ME4 indicates one or more processing units 220 to be used to process each of the processing target substrates Wx, among two or more processing units 220 having different processing contents.

[0275] In the following description, the two or more pieces of processing result information ME3 included in the processing result information set ME30 are normalized by the preprocessing unit CN1 (FIG. 12) and are dimensionless quantities.

[0276] In the substrate processing apparatus 100 according to the third embodiment, the storage unit MEM stores a first learning program 71, a second learning program 72, a plurality of third learning programs 73, a plurality of first learning data 81, a plurality of second learning data 82, a plurality of third learning data 83, a first learning model 91, a second learning model 92, and a plurality of third learning models 93, instead of the learning program ME7, the plurality of learning data ME8, and the learning model ME9 of FIG. 12. Each of the first learning program 71, the second learning program 72, and the third learning program 73 corresponds to an example of a "learning program". Each of the first learning data 81, the second learning data 82, and the third learning data 83 corresponds to an example of a "learning data". Each of the first learning model 91, the second learning model 92, and the third learning model 93 corresponds to an example of a "learning model".

[0277] In the substrate processing apparatus 100 according to the third embodiment, the control unit CNT includes a first learning unit 61, a second learning unit 62, and a plurality of third learning units 63, instead of the learning unit CN9 in Fig. 12. Each of the first learning unit 61, the second learning unit 62, and the third learning unit 63 corresponds to an example of a "learning unit".

[0278] The first learning model 91 is constructed by executing machine learning (e.g., supervised machine learning) on ​​a plurality of first learning data 81. The first learning model 91 is, for example, a trained model.

[0279] Each of the multiple first learning data 81 includes processing result information F1 indicating the processing result of the learning target substrate Wn by the processing unit 220, and evaluation information TV1 indicating an evaluation of the processing result of the learning target substrate Wn indicated by the processing result information F1. The multiple pieces of processing result information F1 included in each of the multiple first learning data 81 indicate processing results by two or more processing units 220 having the same processing content. Otherwise, the configuration of the first learning data 81 is similar to the configuration of the learning data ME8 in FIG.

[0280] In this case, for example, the processing result information F1 is processing result information indicating the processing result of the learning target substrate Wn obtained in a predetermined step (hereinafter, referred to as "predetermined step ST") among steps S1 to S9 and S11 to S15.

[0281] In the third embodiment, the predetermined process ST is process S13. Therefore, the configuration of the processing result information F1 is similar to the configuration of the processing result information DA4. The processing result information F1 is normalized by the pre-processing unit CN1 (FIG. 12) and is a dimensionless quantity. The predetermined process ST corresponds to an example of a "predetermined processing process".

[0282] For example, the predetermined process ST is not particularly limited as long as it is one of the processes S2 to S9 and S11 to S15 that are subsequent to the first process S1 among the processes S1 to S9 and S11 to S15 included in the series of processes SQ. For example, the predetermined process ST is a process subsequent to the exposure process of process S10.

[0283] The first learning program 71 is a program for executing a machine learning algorithm for finding certain rules from a plurality of first learning data 81 and generating a first learning model 91 that expresses the found rules. Otherwise, the configuration of the first learning program 71 is similar to the configuration of the learning program ME7 in FIG.

[0284] The first learning unit 61 performs machine learning on a plurality of first learning data 81 based on the first learning program 71 to generate a first learning model 91. Otherwise, the configuration of the first learning unit 61 is similar to the configuration of the learning unit CN9 in FIG.

[0285] The first learning model 91 receives processing result information DA4 indicating the processing result of the substrate Wp by the processing unit 220 as input information, and outputs evaluation information OT1 indicating an evaluation of the processing result of the substrate Wp as output information. The evaluation information OT1 indicates, for example, the processing result of the substrate Wp as either "normal" or "abnormal". In other words, the first learning model 91 evaluates the substrate W after processing by the processing unit 220. The evaluation information OT1 can also be regarded as information indicating the evaluation of the processing result information DA4. Otherwise, the configuration of the first learning model 91 is similar to the configuration of the learning model ME9 in FIG. 12.

[0286] The first selection unit CN3 inputs, as input information, each of two or more pieces of processing result information DA4 indicating the processing results of two or more substrates Wp to the first learning model 91, and obtains, as output information, evaluation information OT1 indicating an evaluation of the processing results of each of the two or more substrates Wp from the first learning model 91. In this case, the "processing results of two or more substrates Wp" indicates the processing results of two or more substrates Wp by two or more processing units 220 having the same processing content. The two or more pieces of processing result information DA4 are each included in two or more processing result information sets ME30.

[0287] That is, the first selection unit CN3 inputs the processing result information DA4 as input information to the first learning model 91 for each of the multiple processing result information sets ME30, and obtains the evaluation information OT1 from the first learning model 91 as output information.

[0288] The processing result information F1 included in the first learning data 81 and the processing result information DA4 input to the first learning model 91 may be information before normalization.

[0289] The second learning model 92 is constructed by performing machine learning (e.g., unsupervised machine learning) on ​​a plurality of second learning data 82. The second learning model 92 is, for example, a trained model.

[0290] Each of the multiple second learning data 82 includes a processing result information set (hereinafter, referred to as a "processing result information set FS"). The processing result information set FS includes two or more pieces of processing result information (hereinafter, referred to as "processing result information F2") each indicating the processing results of the learning target substrate Wn by two or more processing units 220 having different processing contents.

[0291] Moreover, the processing result information F2 is an explanatory variable. In other words, the processing result information F2 is a feature. The configuration of the processing result information set FS is the same as the configuration of the processing result information set ME30. In other words, the configurations of the two or more pieces of processing result information F2 included in the processing result information set FS are respectively the same as the configurations of the two or more pieces of processing result information ME3 included in the processing result information set ME30. The two or more pieces of processing result information F2 included in the processing result information set FS are normalized by the preprocessing unit CN1 (FIG. 12) and are dimensionless quantities.

[0292] In the third embodiment, the processing result information set FS includes four pieces of processing result information F2. Specifically, the processing result information set FS includes processing result information DB1 indicating the processing result of the learning target substrate Wn in step S1, processing result information DB2 indicating the processing result of the learning target substrate Wn in step S5, processing result information DB3 indicating the processing result of the learning target substrate Wn in step S11, and processing result information DB4 indicating the processing result of the learning target substrate Wn in step S13. The structure of the processing result information DB1 is similar to the structure of the processing result information DA1, and the structure of the processing result information DB2 is similar to the structure of the processing result information DA2. In addition, the structure of the processing result information DB3 is similar to the structure of the processing result information DA3, and the structure of the processing result information DB4 is similar to the structure of the processing result information DA4.

[0293] The second learning program 72 is a program for executing a machine learning algorithm to find certain rules from each processing result information F2 of multiple second learning data 82 and generate a second learning model 92 that represents the found rules.

[0294] In embodiment 3, the machine learning algorithm is unsupervised learning, such as k-means, k-medoids, hierarchical clustering, self-organizing maps, fuzzy c-means, Gaussian mixture models, or neural networks.

[0295] The second learning unit 62 performs machine learning on the multiple second learning data 82 based on the second learning program 72. As a result, certain rules are found from the processing result information F2 of each of the multiple second learning data 82, and a second learning model 92 is generated.

[0296] Specifically, the second learning unit 62 calculates multiple learned parameters by machine learning the multiple second learning data 82 based on the second learning program 72, and generates a second learning model 92 including one or more functions to which the multiple learned parameters are applied. The learned parameters are parameters (coefficients) acquired based on the results of machine learning using the multiple second learning data 82.

[0297] The second learning model 92 receives as input information each of two or more pieces of processing result information DA1-DA4 included in the processing result information set ME30, clusters the two or more pieces of processing result information DA1-DA4, and outputs as output information first clustering information indicating the result of the clustering. Clustering refers to finding similar or correlated information and grouping the similar or correlated information. Thus, by clustering, similar or correlated information is classified into one cluster. The first clustering information indicates each cluster into which each piece of processing result information DA1-DA4 is classified. The first clustering information corresponds to an example of "clustering information".

[0298] The first selection unit CN3 inputs each of the two or more pieces of processing result information DA1-DA4 included in the processing result information set ME30 as input information to the second learning model 92, and acquires first clustering information from the second learning model 92. In this case, the "two or more pieces of processing result information DA1-DA4" indicate the processing results of the substrate Wp by each of two or more processing units 220 having different processing contents.

[0299] Then, the first selection unit CN3 selects one or more processing units 220 to be used for processing the target substrate Wx from among two or more processing units 220 having different processing contents based on the first clustering information acquired from the second learning model 92. Therefore, according to the third embodiment, by utilizing the second learning model 92 that has learned the multiple second learning data 82 including the processing result information DB1 to DB4, the processing unit 220 to be used can be selected quickly and accurately.

[0300] For example, via the input unit IN, the user labels each cluster generated by the second learning model 92 as "normal" indicating that the processing result is normal or "abnormal" indicating that the processing result is abnormal.

[0301] Then, for example, the first selection unit CN3 selects the processing unit 220 from which processing result information DA1-DA4 classified into a cluster labeled "normal" is obtained as the processing unit 220 to be used for processing the processing target substrate Wx. On the other hand, for example, the first selection unit CN3 sets the processing unit 220 from which processing result information DA1-DA4 classified into a cluster labeled "abnormal" is obtained as the processing unit 220 to be unused.

[0302] Particularly in the third embodiment, the first selection unit CN3 judges, based on the first clustering information, whether or not the processing result information DA1-DA3 obtained in the processes S1, S5, S11, S13 executed before the predetermined process S13 among the multiple processes S1, S5, S11, S13 having different processing contents belongs to the same cluster as the processing result information DA4 obtained in the predetermined process S13, and selects, based on the judgment result, one or more processing units 220 to be used for processing the target substrate Wx from among the two or more processing units 220. In other words, the first selection unit CN3 selects the processing unit 220 to be used for processing the target substrate Wx based on the cluster to which the processing result information DA4 obtained in the predetermined process S13, which is the downstream process among the multiple processes S1, S5, S11, S13, belongs. Therefore, according to the third embodiment, it is possible to identify an upstream process that caused the processing result indicated by the processing result information DA4 among the multiple processes S1, S5, S11, and S13, and select a processing unit 220 to be used for processing the target substrate Wx. The predetermined process S13 corresponds to an example of a "predetermined processing process".

[0303] For example, in the first analysis algorithm AG1, the first selection unit CN3 selects, as the processing unit 220 to be used for processing the processing target substrate Wx, the processing unit 220 from which processing result information DA1 to DA3 has been obtained that belongs to a different cluster from the processing result information DA4 indicated as "abnormal" by the evaluation information OT1 acquired from the first learning model 91. In other words, the processing unit 220 from which processing result information uncorrelated or less correlated with the processing result information DA4 indicated as "abnormal" by the evaluation information OT1 is selected.

[0304] Furthermore, in the first analysis algorithm AG1, if the evaluation information OT1 obtained from the first learning model 91 indicates that the processing result information DA4 is “normal”, the first selection unit CN3 selects all processing units 220 from which the processing result information DA1 to DA3 has been obtained as the processing units 220 to be used for processing the substrate Wx to be processed.

[0305] Furthermore, in the first analysis algorithm AG1, the first selection unit CN3 sets, as a processing unit 220 to be unused, the processing unit 220 from which processing result information has been obtained that belongs to the same cluster as the processing result information DA4 indicated as "abnormal" by the evaluation information OT1 acquired from the first learning model 91, among the processing result information DA1 to DA3. In other words, the processing unit 220 from which processing result information has been obtained that is correlated with the processing result information DA4 indicated as "abnormal" by the evaluation information OT1, among the processing result information DA1 to DA3, is set as a processing unit to be unused.

[0306] Also, for example, to execute the second analysis algorithm AG2, the user labels at least one cluster among the multiple clusters generated by the second learning model 92 with "specificity" via the input unit IN, indicating that the processing result has specificity. Whether the processing result has specificity or not is determined experimentally and / or empirically.

[0307] The second analysis algorithm AG2 is the same as the first analysis algorithm AG1. However, even if the evaluation information OT1 acquired from the first learning model 91 indicates that the processing result information DA4 is “normal”, the processing unit 220 is set as an unused target under certain conditions.

[0308] That is, in the second analysis algorithm AG2, the first selection unit CN3 sets, among the processing result information DA1 to DA3, the processing result information that belongs to a cluster different from the processing result information DA4 indicated as "normal" by the evaluation information OT1 obtained from the first learning model 91 and that belongs to a cluster labeled with "uniqueness" to the processing unit 220 to be unused.

[0309] For example, the selection targets by the first selection unit CN3 are two or more processing units 220 that perform a process other than a predetermined process ST (e.g., process S13) among the multiple processing units 220 from which the multiple pieces of processing result information ME3 included in the processing result information set ME30 are obtained. However, the selection targets by the first selection unit CN3 may be all of the multiple processing units 220 from which the multiple pieces of processing result information ME3 included in the processing result information set ME30 are obtained. In this case, for example, when the processing result information DA4 belongs to a cluster of "normal", the first selection unit CN3 selects the processing unit 220 from which the processing result information DA4 is obtained as the processing unit 220 to be used for processing the processing target substrate Wx. On the other hand, for example, when the processing result information DA4 belongs to a cluster of "abnormal", the first selection unit CN3 sets the processing unit 220 from which the processing result information DA4 is obtained as the processing unit 220 to be unused.

[0310] Here, the first selection unit CN3 inputs each of the processing result information DA1 to DA4 as input information to the second learning model 92 for each of the multiple processing result information sets ME30, and acquires first clustering information from the second learning model 92. Then, the first selection unit CN3 selects one or more processing units 220 to be used to process the target substrate Wx for each of the multiple processing result information sets ME30 based on the first clustering information, and generates selection result information ME4. That is, the first selection unit CN3 selects one or more processing units 220 to be used to process the target substrate Wx for each of the multiple series of processes SQ based on the first clustering information, and generates selection result information ME4. That is, multiple pieces of selection result information ME4 are generated corresponding to the multiple series of processes SQ, respectively.

[0311] The first selection unit CN3 may assign priorities to the multiple pieces of selection result information ME4. In this case, the priorities are the priority order of use for a set of one or more processing units 220 selected by the selection result information ME4.

[0312] For example, in the first analysis algorithm AG1 and the second analysis algorithm AG2, when the evaluation information OT1 obtained from the first learning model 91 indicates that the processing result information DA4 is "abnormal" and at least one of the processing result information DA1 to DA3 belongs to the same cluster as the processing result information DA4, the first selection unit CN3 assigns the lowest priority to the selection result information ME4 based on the processing result information DA1 to DA4.

[0313] For example, in the first analysis algorithm AG1, when the evaluation information OT1 obtained from the first learning model 91 indicates that the processing result information DA4 is "normal" and at least one of the processing result information DA1 to DA3 belongs to the same cluster as the processing result information DA4, the first selection unit CN3 assigns a lower priority to the selection result information ME4 based on the processing result information DA1 to DA4 than to a case where all of the processing result information DA1 to DA3 belong to a different cluster from the processing result information DA4.

[0314] For example, in the second analysis algorithm AG2, when the evaluation information OT1 obtained from the first learning model 91 indicates that the processing result information DA4 is "normal" and at least one of the processing result information DA1 to DA3 belongs to a cluster labeled with "uniqueness", the first selection unit CN3 assigns a lower priority to the selection result information ME4 based on the processing result information DA1 to DA4 than to a case where none of the processing result information DA1 to DA3 belong to a cluster labeled with "uniqueness".

[0315] Also, for example, the first selection unit CN3 sets, as a target for quality inspection, the processing target substrate Wx to be processed by a set of one or more processing units 220 selected by the selection result information ME4 to which a priority lower than a predetermined priority has been assigned. In this case, the first selection unit CN3 may determine the set of processing units 220 by assigning a priority to the quality inspection.

[0316] Furthermore, in the third embodiment, in the first clustering information, when the processing result information (e.g., processing result information DA1) obtained in a process (e.g., process S1) executed in a step prior to a predetermined process S13 included in a series of processes SQ belongs to the same cluster as the processing result information DA4 obtained in the predetermined process S13, the first selection unit CN3 judges whether or not the processing result information (e.g., processing result information DA1) obtained in a process (e.g., process S1 of another series of processes SQ) having the same processing content as the process executed in the previous step (e.g., process S1) is correlated with the processing result information (e.g., processing result information DA1) obtained in the process executed in the previous step (e.g., process S1) based on the judgment result. Then, the first selection unit CN3 selects one or more processing units 220 to be used for processing the processing target substrate Wx from among two or more processing units 220 having the same processing content. Therefore, according to the third embodiment, it is possible to select a processing unit 220 that has produced good processing results from among two or more processing units 220 having the same processing contents, thereby improving the quality of the processed substrate Wx.

[0317] For example, when the evaluation information OT1 acquired from the first learning model 91 indicates that the processing result information DA4 is "abnormal" and the processing result information DA1 by the process S1 executed before the predetermined process S13 in the first clustering information belongs to the same cluster as the processing result information DA4 by the predetermined process S13, if the first selection unit CN3 determines that another processing result information DA1 by the process S1 having the same processing content as the process S1 executed before is correlated with the processing result information DA1 by the process S1 executed before, the first selection unit CN3 sets the processing unit 220 from which the other processing result information DA1 by the process S1 having the same processing content was obtained as the processing unit 220 to be unused. Therefore, according to the third embodiment, the processing unit 220 with poor processing results can be set as the processing unit to be unused from among two or more processing units 220 having the same processing content, and the quality of the processed substrate Wx can be improved as a whole in the substrate processing apparatus 100.

[0318] In this specification, setting a processing unit 220 to be unused (that is, determining a processing unit 220 to be unused) is synonymous with selecting a processing unit 220 other than the processing unit 220 to be unused.

[0319] In particular, in the third embodiment, the first selection unit CN3 utilizes the third learning model 93 to select one or more processing units 220 based on the correlation.

[0320] The memory unit MEM stores a plurality of third learning models 93 corresponding to the plurality of steps S1 to S9 and S11 to S15 having different processing contents in the series of steps SQ. In the third embodiment, as an example, the memory unit MEM stores a plurality of third learning models 93 corresponding to the plurality of steps S1, S5, and S11, respectively.

[0321] The following describes the third learning model 93, the multiple third learning data 83, and the third learning unit 63, which correspond to step S1. The configurations of the third learning model 93, the third learning data 83, and the third learning unit 63, which correspond to steps S5 and S11, are similar to the configurations of the third learning model 93, the third learning data 83, and the third learning unit 63, which correspond to step S1, and therefore descriptions thereof will be omitted.

[0322] The third learning model 93 is constructed by performing machine learning (e.g., unsupervised machine learning) on ​​a plurality of third learning data 83. The third learning model 93 is, for example, a trained model.

[0323] Each of the multiple third learning data 83 includes processing result information (hereinafter, referred to as "processing result information F3") indicating the processing result of the learning target substrate Wn by the processing unit 220 that executes the process S1. The multiple pieces of processing result information F3 are obtained in two or more processes S1 having the same processing content among the multiple series of processes SQ.

[0324] Moreover, the processing result information F3 is an explanatory variable. In other words, the processing result information F3 is a feature amount. The configuration of the processing result information F3 is the same as the configuration of the processing result information DA1 obtained in the step S1. 3 is normalized by the preprocessing unit CN1 (FIG. 12) and is a dimensionless quantity.

[0325] The third learning program 73 is a program for executing a machine learning algorithm for finding certain rules from a plurality of third learning data 83 and generating a third learning model 93 that expresses the found rules. The machine learning algorithm of the third learning program 73 is similar to the machine learning algorithm of the second learning program 72.

[0326] The third learning unit 63 performs machine learning on a plurality of third learning data 83 based on the third learning program 73. As a result, certain rules are found from the plurality of third learning data 83, and a third learning model 93 is generated.

[0327] Specifically, the third learning unit 63 calculates multiple learned parameters by machine learning multiple third learning data 83 based on the third learning program 73, and generates a third learning model 93 including one or more functions to which the multiple learned parameters are applied. The learned parameters are parameters (coefficients) acquired based on the results of machine learning using the multiple third learning data 83.

[0328] The third learning model 93 receives as input information two or more pieces of processing result information DA1 obtained in two or more steps S1 having the same processing content among a plurality of series of steps SQ, clusters the two or more pieces of processing result information DA1, and outputs as output information second clustering information indicating the clustering result. The second clustering information indicates each cluster into which each of the two or more pieces of processing result information DA1 is classified.

[0329] The first selection part CN3 selects a processing unit 220 to be used for processing the processing target substrate Wx from two or more processing units 220 having the same processing content among a plurality of series of processes SQ in the following manner.

[0330] That is, the first selection unit CN3 refers to the first clustering information acquired from the second learning model 92 for the processing result information set ME30, and identifies processing result information belonging to the same cluster as the processing result information DA4 obtained in the predetermined step S13 from among the processing result information DA1-DA3 obtained in steps S1, S5, and S11 executed prior to the predetermined step S13 in the series of steps SQ. For simplicity, the following illustrates an example in which the identified processing result information is the processing result information DA1 obtained in step S1.

[0331] Then, the first selection unit CN3 selects the third learning model 93 corresponding to the process S1 in which the identified processing result information DA1 was obtained.

[0332] Furthermore, the first selection unit CN3 inputs, as input information for the selected third learning model 93, the identified processing result information DA1 and processing result information DA1 obtained in another process S1 having the same processing content among the multiple series of processes SQ, and obtains second clustering information indicating the result of clustering for the multiple processing result information DA1.

[0333] In addition, the first selection unit CN3 refers to the second clustering information to determine whether or not there is other processing result information DA1 that belongs to the same cluster as the identified processing result information DA1.

[0334] Then, based on the judgment result, the first selection unit CN3 selects a processing unit 220 to be used to process the target substrate Wx from two or more processing units 220 that perform other processes S1 having the same processing content among the multiple series of processes SQ.

[0335] For example, if the evaluation information OT1 obtained from the first learning model 91 indicates that the processing result information DA4 is “abnormal,” the first selection unit CN3 sets as an unused target the processing unit 220 that executes other processes S1 from which other processing result information DA1 that belongs to the same cluster as the identified processing result information DA1 has been obtained.

[0336] Note that the processing result information F3 included in the third learning data 83 and the processing result information DA1 input to the third learning model 93 may be information before normalization.

[0337] As shown in FIG. 12, in the third embodiment, as in the second embodiment, the storage unit MEM stores priority information ME6 for giving priority to throughput. Here, a plurality of processing units 220 that execute the steps S1, S5, and S11 included in a series of steps SQ are regarded as one set. In this case, a plurality of sets exist corresponding to a plurality of series of steps SQ. The priority information ME6 in the third embodiment indicates the priority of use for the plurality of sets. The number of all processing units 220 that constitute the plurality of sets is a predetermined number. In other words, in the priority information ME6, all of the predetermined number of processing units 220 are selected. For example, the priority information ME6 indicates a consecutive number assigned to the plurality of sets.

[0338] Next, a substrate processing method according to embodiment 3 will be described with reference to Fig. 9 and Fig. 21 to Fig. 24. The substrate processing method according to embodiment 3 is similar to the substrate processing method according to embodiment 1 described with reference to Fig. 9. Below, differences between the substrate processing method according to embodiment 3 and the substrate processing method of embodiment 1 will be mainly described.

[0339] Figures 21 and 22 are flow charts showing step S31 of Figure 9 in embodiment 3. As shown in Figures 21 and 22, the process of step S31 according to embodiment 3 includes steps S151 to S166.

[0340] In step S151, the preprocessing unit CN1 in FIG. 12 processes a plurality of pieces of measurement data ME2 to generate a plurality of processing result information sets ME30.

[0341] Next, in step S152, the acquisition unit CN2 in FIG. 12 acquires a processing result information set ME30 by two or more processing units 220 having different processing contents in the series of steps SQ.

[0342] Next, in step S153, the first selection unit CN3 in FIG. 12 selects one or more processing units 220 to be used for processing each of the target substrates Wx from among two or more processing units 220 (specifically, a predetermined number of processing units 220) based on a processing result information set ME30 from two or more processing units 220 having different processing contents.

[0343] Specifically, the first selection unit CN3 analyzes the processing result information set ME30 using an analysis algorithm, selects one or more processing units 220 to be used for processing each of the target substrates Wx from among two or more processing units 220 having different processing contents, and generates selection result information ME4.

[0344] Fig. 23 is a flowchart showing an example of the process of step S153 in Fig. 21. As shown in Fig. 23, the process of step S153 includes steps S181 to S185.

[0345] In step S181, the first selection unit CN3 inputs, as input information to the first learning model 91, the processing result information DA4 obtained in a predetermined step S13 (FIG. 20) from among the processing result information DA1 to DA4 included in the processing result information set ME30.

[0346] Next, in step S182, the first selection unit CN3 acquires, as output information, from the first learning model 91, evaluation information OT1 indicating an evaluation of the processing result of the substrate Wp in the predetermined step S13.

[0347] Next, in step S183, the first selection unit CN3 inputs the processing result information DA1 to DA4 included in the processing result information set ME30 to the second learning model 92 as input information.

[0348] Next, in step S184, the first selection unit CN3 acquires, from the second learning model 92, first clustering information indicating the result of clustering the processing result information DA1 to DA4 as output information.

[0349] Next, in step S185, the first selection part CN3 selects one or more processing units 220 to be used for processing each of the processing target substrates Wx from among two or more processing units 220 having different processing contents in the series of processes SQ, that is, from among two or more processing units 220 that execute the processes S1, S5, and S11 (FIG. 20) included in the series of processes SQ, based on the evaluation information OT1 and the first clustering information, and generates selection result information ME4. Then, the process returns to the routine shown in FIG. 21 and proceeds to step S154.

[0350] Steps S154 to S158 shown in FIG. 21 are similar to steps S54 to S58 shown in FIG. 10, respectively, and therefore will not be described.

[0351] In step S159, the first selection unit CN3 determines whether or not the processes of steps S152 to S158 have been completed for all of the multiple processing result information sets ME30 corresponding to the multiple series of steps SQ, respectively.

[0352] If a negative determination is made in step S159, the process proceeds to step S152.

[0353] On the other hand, if an affirmative determination is made in step S159, the process proceeds to step S160.

[0354] In step S160, the first selection unit CN3 prioritizes the multiple pieces of selection result information ME4 generated corresponding to the multiple processing result information sets ME30, respectively.

[0355] Next, in step S161, the first selection unit CN3 determines whether or not the processes based on all the analysis algorithms have been completed.

[0356] If the result of the determination in step S161 is negative, the process proceeds to step S152. Then, in step S153 following step S152, the first selection part CN3 analyzes the processing result information set ME30 by an analysis algorithm different from the previous analysis algorithm, and selects a processing unit 220.

[0357] On the other hand, if an affirmative decision is made in step S161, the processing proceeds to step S162 in FIG.

[0358] As shown in FIG. 22, in step S162, the first selection unit CN3 selects one or more processing units 220 to be used to process the target substrate Wx from among the two or more processing units 220 based on the correlation between two or more pieces of processing result information ME3 from two or more processing units 220 having the same processing content.

[0359] Fig. 24 is a flowchart showing an example of the process of step S162 in Fig. 22. As shown in Fig. 24, the process of step S162 includes steps S201 to S204.

[0360] In step S201, the first selection unit CN3 refers to the first clustering information and identifies processing result information that belongs to the same cluster as the processing result information DA4 obtained in the predetermined step S13 (FIG. 20) from among the processing result information DA1-DA3 obtained in steps S1, S5, and S11 (FIG. 20) that precede the predetermined step S13 (FIG. 20). For simplicity, the following illustrates an example in which the identified processing result information is the processing result information DA1 obtained in step S1.

[0361] Next, in step S202, the first selection unit CN3 selects, from the multiple third learning models 93, the third learning model 93 that corresponds to the step S1 from which the processing result information DA1 identified in step S201 was obtained.

[0362] Next, in step S203, the first selection unit CN3 inputs, as input information to the third learning model 93 selected in step S202, multiple pieces of processing result information DA1 obtained in steps S1 having the same processing content among multiple series of steps SQ, to obtain second clustering information.

[0363] Next, in step S205, the first selection unit CN3 selects a processing unit 220 to be used for processing the target substrate Wx from two or more processing units 220 that execute the process S1 having the same processing content among the plurality of series of processes SQ based on the second clustering information. Then, the process returns to the routine shown in FIG. 22 and proceeds to step S163.

[0364] As shown in FIG. 22, in step S163, the first selection unit CN3 updates each piece of selection result information ME4 based on the result of step S162.

[0365] Next, in step S164, the first selection unit CN3 updates the priority assigned to each piece of selection result information ME4 based on the result of step S162.

[0366] Next, in step S165, the first selection part CN3 determines the processing unit 220 in which the processed substrate W will be subjected to quality inspection, based on each piece of selection result information ME4.

[0367] Next, in step S166, the first selection unit CN3 notifies each piece of selection result information ME4. This is similar to step S60 in Fig. 10. Then, the process returns to the main routine in Fig. 9 and proceeds to step S32.

[0368] In addition, the "two or more processing units 220 having different processing contents" is not particularly limited, and may be, for example, two or more processing units 220 each performing two or more steps among steps S1 to S9 and S11 to S15 constituting the series of steps SQ in Figures 6 and 7.

[0369] Next, a first learning method by the first learning unit 61, a second learning method by the second learning unit 62, and a third learning method by the third learning unit 63 will be described with reference to FIG.

[0370] The first learning method by the first learning unit 61 is similar to the learning method by the learning unit CN9 described with reference to Fig. 15. That is, in the description of Fig. 15, the learning method is replaced with the first learning method, the learning unit CN9 is replaced with the first learning unit 61, the learning data ME8 is replaced with the first learning data 81, the learning program ME7 is replaced with the first learning program 71, and the learning model ME9 is replaced with the first learning model 91.

[0371] The second learning method by the second learning unit 62 is similar to the learning method by the learning unit CN9 described with reference to Fig. 15. That is, in the description of Fig. 15, the learning method is replaced with the second learning method, the learning unit CN9 is replaced with the second learning unit 62, the learning data ME8 is replaced with the second learning data 82, the learning program ME7 is replaced with the second learning program 72, and the learning model ME9 is replaced with the second learning model 92.

[0372] In particular, each of the multiple second learning data 82 includes two or more pieces of processing result information F2, each indicating the processing result of a learning target substrate Wn by two or more processing units 220 having different processing contents. The second learning model 92 is a program that receives as input information each of two or more pieces of processing result information DA1-DA4, each indicating the processing result of a substrate Wp by two or more processing units 220 having different processing contents, clusters the two or more pieces of processing result information DA1-DA4, and outputs as output information first clustering information indicating the clustering result.

[0373] In other words, the second learning model 92 causes the control device CM to function so as to cluster the processing result information DA1-DA4 indicating the processing results of the substrate Wp. Specifically, the second learning model 92 causes the control device CM to function so as to input input information and output output information. The input information is two or more pieces of processing result information DA1-DA4 each indicating the processing results of the substrate Wp by two or more processing units 220 having different processing contents. The output information is first clustering information indicating the result of clustering the two or more pieces of processing result information DA1-DA4.

[0374] The third learning method by the third learning unit 63 is similar to the learning method by the learning unit CN9 described with reference to Fig. 15. That is, in the description of Fig. 15, the learning method is replaced with the third learning method, the learning unit CN9 is replaced with the third learning unit 63, the learning data ME8 is replaced with the third learning data 83, the learning program ME7 is replaced with the third learning program 73, and the learning model ME9 is replaced with the third learning model 93. Also, the third learning method by the third learning unit 63 is similar to the second learning method by the second learning unit 62 in that it performs clustering.

[0375] Next, the selection result information ME4 will be described with reference to Figs. 25 to 28, with specific examples. In addition, as examples of two or more processing units 220 whose processing contents are different in a series of steps SQ, the processing unit PAHP

[11] in Fig. 3, the processing unit RT

[11] in Fig. 2, the processing unit PHP

[31] in Fig. 3, and the processing unit DV

[11] in Fig. 2 are given. As examples of other two or more processing units 220 whose processing contents are different in another series of steps SQ, the processing unit PAHP

[12] , the processing unit RT

[12] , the processing unit PHP

[32] , and the processing unit DV

[12] are given. As examples of still other two or more processing units 220 whose processing contents are different in yet another series of steps SQ, the processing unit PAHP

[13] , the processing unit RT

[11] , the processing unit PHP

[33] , and the processing unit DV

[13] are given.

[0376] That is, examples of processing units to be selected by the first selection unit CN3 include PAHP

[11] , RT

[11] , PHP

[31] , PAHP

[12] , RT

[12] , PHP

[32] , PAHP

[13] , RT

[11] , and PHP

[33] .

[0377] Fig. 25 is a diagram showing an example of a processing result information set ME30, priority order information ME6, selection result information ME4, and examination target information ME5 according to embodiment 3. In Fig. 25, unused targets are indicated by "NU".

[0378] As shown in FIG. 25, in the third embodiment, processing result information sets ME30a, ME30b, and ME30c are shown as an example of the processing result information set ME30.

[0379] The processing result information set ME30a includes contact angle map data M10, film thickness map data M11, temperature map data M12, and pattern line width map data M13. The map data M10, M11, M12, and M13 indicate processing results of the substrate Wp after processing in the processing units PAHP

[11] , RT

[11] , PHP

[31] , and DV

[11] , respectively.

[0380] The processing result information set ME30b includes contact angle map data M20, film thickness map data M21, temperature map data M22, and pattern line width map data M23. The map data M20, M21, M22, and M23 indicate processing results of the substrate Wp after processing in the processing units PAHP

[12] , RT

[12] , PHP

[32] , and DV

[12] , respectively.

[0381] The processing result information set ME30c includes contact angle map data M30, film thickness map data M31, temperature map data M32, and pattern line width map data M33. The map data M30, M31, M32, and M33 indicate processing results of the substrate Wp after processing in the processing units PAHP

[13] , RT

[11] , PHP

[33] , and DV

[13] , respectively.

[0382] The contact angles represented by the contact angle map data M10, M20, and M30 indicate the processing results of the substrate Wp after the adhesion strengthening processing in step S1, and indicate the contact angles with respect to the processing liquid for the anti-reflection film.

[0383] The film thicknesses represented by the film thickness map data M11, M21, and M31 indicate the processing results of the substrate Wp after the resist film forming processing in step S5, and indicate the film thicknesses of the resist films.

[0384] The temperatures represented by the temperature map data M12, M22, and M32 indicate the processing results of the substrate Wp after the third heating process in step S11 (after the heating process after exposure), and indicate the temperature of the substrate Wp.

[0385] The line widths indicated by the pattern line width map data M13, M23, and M33 indicate the processing results of the substrate Wp after the development processing in step S13, and indicate the line widths of the resist patterns.

[0386] FIG. 26 is a diagram showing contact angle map data M10, film thickness map data M11, temperature map data M12, and pattern line width map data M13, which constitute the processing result information set ME30a of FIG.

[0387] FIG. 27 is a diagram showing contact angle map data M20, film thickness map data M21, temperature map data M22, and pattern line width map data M23, which constitute the processing result information set ME30b in FIG.

[0388] FIG. 28 is a diagram showing contact angle map data M30, film thickness map data M31, temperature map data M32, and pattern line width map data M33, which constitute the processing result information set ME30c in FIG.

[0389] For ease of understanding, FIGS. 26 to 28 show map data M10 to M13, M20 to M23, and M30 to M33 before normalization by preprocessing unit CN1.

[0390] In Figs. 26 to 28, the region G1 is indicated by dense slanting lines sloping upward to the right. The region G2 is indicated by sparse slanting lines sloping downward to the right. The region G3 is indicated by sparse slanting lines sloping upward to the right. The region G4 is indicated by crossed slanting lines. The region G5 is indicated by dots. The region G6 is indicated by dense slanting lines sloping downward to the right. In Figs. 26 to 28, the regions G1, G2, G3, G4, G5, and G6 are shown to make the drawings easier to understand. However, in reality, the contact angle distribution is indicated by gradation in the contact angle map data M10, M20, and M30. Similarly, in reality, the film thickness of the film thickness map data M11, M21, and M31, the temperature of the temperature map data M12, M22, and M32, and the line width of the pattern line width map data M13, M23, and M33 are also indicated by gradation.

[0391] In the contact angle map data M10, M20, and M30, the contact angle is smallest in the region G1 and largest in the region G6. The contact angle increases in the order of the regions G1, G2, G3, G4, G5, and G6 from the region G1 toward the region G6. Similarly, in the film thickness map data M11, M21, and M31, the film thickness increases in the order of the regions G1, G2, G3, G4, G5, and G6 from the region G1 toward the region G6. Similarly, in the temperature map data M12, M22, and M32, the temperature increases in the order of the regions G1, G2, G3, G4, G5, and G6 from the region G1 toward the region G6. Similarly, in the pattern line width map data M13, M23, and M33, the line width increases in the order of the regions G1, G2, G3, G4, G5, and G6 from the region G1 toward the region G6.

[0392] Returning to FIG. 25, priority information C1 is shown as an example of priority information ME6 in the case where throughput is prioritized.

[0393] In the case of throughput priority, all the processing units to be selected, PAHP

[11] , RT

[11] , PHP

[31] , PAHP

[12] , RT

[12] , PHP

[32] , PAHP

[13] , RT

[11] , PHP

[33] , are used. The priority information C1 indicates, in order of highest priority of use, the set of processing units PAHP

[11] , RT

[11] , PHP

[31] , the set of processing units PAHP

[12] , RT

[12] , PHP

[32] , and the set of processing units PAHP

[13] , RT

[11] , PHP

[33] .

[0394] Moreover, as an example of the selection results information ME4 in the case of quality priority, the selection results information C2 is shown. The selection results information C2 is obtained based on the above-mentioned first analysis algorithm AG1.

[0395] Regarding the selection result information C2, the pattern line width map data M13 of the processing result information set ME30a is indicated as “abnormal” by the evaluation information OT1 of the first learning model 91.

[0396] Also, as shown in FIG. 26, the temperature map data M12 is correlated with the pattern line width map data M13. In the example of FIG. 26, the temperature map data M12 is negatively correlated with the pattern line width map data M13. Therefore, the first clustering information by the second learning model 92 indicates that the temperature map data M12 belongs to the same cluster as the pattern line width map data M13. In this case, it can be estimated that the cause of the "abnormality" of the pattern line width map data M13 is the temperature map data M12. In other words, it can be estimated that the cause of the "abnormality" of the pattern line width map data M13 is the processing unit PHP

[31] from which the temperature map data M12 was obtained.

[0397] Therefore, as shown in FIG. 25, in the selection result information C2 for the processing result information set ME30a, the first selection unit CN3 sets the processing unit PHP

[31] from which the temperature map data M12 was obtained as an unused target, and selects the processing unit PAHP

[11] from which the contact angle map data M10 was obtained and the processing unit RT

[11] from which the film thickness map data M11 was obtained as the processing units for processing the substrate Wx to be processed.

[0398] Furthermore, with regard to the selection result information C2, the pattern line width map data M23 of the processing result information set ME30b is indicated as “normal” by the evaluation information OT1 of the first learning model 91.

[0399] 27, none of the contact angle map data M20, film thickness map data M21, and temperature map data M22 is correlated with the pattern line width map data M13. Therefore, the first clustering information by the second learning model 92 indicates that the contact angle map data M20, film thickness map data M21, and temperature map data M22 belong to a different cluster from the pattern line width map data M13.

[0400] Then, as shown in FIG. 25, for the processing result information set ME30b, in the selection result information C2, the first selection unit CN3 selects all of the processing units PAHP

[12] , RT

[12] , and PHP

[32] as processing units for processing the target substrate Wx.

[0401] Furthermore, with regard to the selection result information C2, the pattern line width map data M33 of the processing result information set ME30c is indicated as “normal” by the evaluation information OT1 of the first learning model 91.

[0402] 28, the film thickness map data M31 is correlated with the pattern line width map data M33. Therefore, the first clustering information by the second learning model 92 indicates that the film thickness map data M31 belongs to the same cluster as the pattern line width map data M33. In this case, it can be estimated that the film thickness distribution of the film thickness map data M31 has influenced the line width distribution of the pattern line width map data M33. In other words, it can be estimated that the processing unit RT

[11] that obtained the film thickness map data M31 has influenced the line width distribution of the pattern line width map data M33.

[0403] Then, as shown in FIG. 25, for the processing result information set ME30c, in the selection result information C2, the first selection unit CN3 selects all of the processing units PAHP

[13] , RT

[11] , and PHP

[33] as processing units for processing the target substrate Wx.

[0404] In addition, in the selection result information C2, the lowest priority is assigned to the set of processing units PAHP

[11] , RT

[11] selected based on the processing result information set ME30a according to the first analysis algorithm AG1. In addition, in the selection result information C2, the set of processing units PAHP

[13] , RT

[11] , PHP

[33] selected based on the processing result information set ME30c according to the first analysis algorithm AG1 is assigned a lower priority than the set of processing units PAHP

[12] , RT

[12] , PHP

[32] selected based on the processing result information set ME30b according to the first analysis algorithm AG1. This is because, in the processing result information set ME30c, the film thickness map data M31 belongs to the same cluster as the pattern line width map data M33.

[0405] Moreover, as an example of the selection results information ME4 in the case of quality priority, the selection results information C3 is shown. The selection results information C3 is obtained based on the above-mentioned second analysis algorithm AG2.

[0406] Regarding the selection result information C3, the pattern line width map data M13 of the processing result information set ME30a is indicated as "abnormal" by the evaluation information OT1 of the first learning model 91. In addition, the selection result information C3 obtained from the processing result information set ME30a is the same as the selection result information C2 obtained from the processing result information set ME30a.

[0407] Furthermore, with regard to the selection result information C3, the pattern line width map data M23 of the processing result information set ME30b is indicated as “normal” by the evaluation information OT1 of the first learning model 91.

[0408] 27, the contact angle map data M20 has singularity. Therefore, the first clustering information by the second learning model 92 indicates that the contact angle map data M20 belongs to a cluster labeled with "singularity." It can be estimated that the processing unit PAHP

[12] from which the contact angle map data M20 having singularity was obtained is likely to affect the line width of a resist pattern in the future.

[0409] Therefore, as shown in FIG. 25, in the selection result information C3 for the processing result information set ME30b, the first selection unit CN3 sets the processing unit PAHP

[12] from which the contact angle map data M20 was obtained as an unused target, and selects the processing unit RT

[12] from which the film thickness map data M21 was obtained and the processing unit PHP

[32] from which the temperature map data M22 was obtained as the processing units for processing the substrate Wx to be processed.

[0410] Furthermore, with regard to the selection result information C3, the pattern line width map data M33 of the processing result information set ME30c is indicated as “normal” by the evaluation information OT1 of the first learning model 91.

[0411] With regard to the processing result information set ME30c, in the selection result information C3, the first selection part CN3 selects all of the processing units PAHP

[13] , RT

[11] , and PHP

[33] as processing units for processing the processing target substrate Wx.

[0412] In addition, in the selection result information C3, the lowest priority is given to the pair of processing units PAHP

[11] and RT

[11] selected based on the processing result information set ME30a according to the second analysis algorithm AG2. In addition, in the selection result information C3, the pair of processing units RT

[12] and PHP

[32] selected based on the processing result information set ME30b according to the second analysis algorithm AG2 is given a lower priority than the pair of processing units PAHP

[13] , RT

[11] , and PHP

[33] selected based on the processing result information set ME30c. This is because the contact angle map data M20 of the processing result information set ME30b belongs to the cluster labeled "singularity".

[0413] For example, with respect to the processing result information set ME30b, the first selection unit CN3 may update the selection result information C2 by replacing at least one of the processing units PAHP

[12] , RT

[12] , and PHP

[32] selected in the selection result information C2 with another processing unit 220 having the same processing contents. In this case, the other processing unit 220 has a better processing result for the substrate Wp than the processing unit to be replaced.

[0414] (Embodiment 4) A substrate processing system SYS according to a fourth embodiment of the present invention will be described with reference to Fig. 29. The substrate processing system SYS according to the fourth embodiment is mainly different from the first embodiment in that the substrate processing system SYS according to the fourth embodiment has a selection device 400 for selecting a processing unit 220 outside the substrate processing apparatus 100A. The following mainly describes the differences between the fourth embodiment and the first embodiment.

[0415] Fig. 29 is a diagram showing a substrate processing system SYS according to a fourth embodiment of the present invention. As shown in Fig. 29, the substrate processing system SYS includes a substrate processing apparatus 100A and a selection device 400. The substrate processing apparatus 100A and the selection device 400 communicate with each other via a network NW. The network NW includes, for example, the Internet, a LAN (Local Area Network), a public telephone network, and a short-distance wireless network.

[0416] The configuration of the substrate processing apparatus 100A is similar to that of the substrate processing apparatus 100 according to the first embodiment described with reference to Figures 1 to 8. However, the substrate processing apparatus 100A does not include, for example, the acquisition unit CN2 and the first selection unit CN3 of Figure 8. Furthermore, the substrate processing apparatus 100A does not necessarily have to include the pre-processing unit CN1 of Figure 8. Furthermore, the substrate processing apparatus 100A does not necessarily have to store the measurement data ME2 and the processing result information ME3.

[0417] The selection device 400 selects one or more processing units 220 from the multiple processing units 220 that process multiple target substrates Wx, respectively. Specifically, the selection device 400 includes a control unit 401, a storage unit 402, an input unit 403, a display unit 404, and a communication unit 405. The configurations of the control unit 401, the storage unit 402, the input unit 403, the display unit 404, and the communication unit 405 are similar to the configurations of the control unit CNT, the storage unit MEM, the input unit IN, the display unit DP, and the communication unit TR in Fig. 8, respectively, and therefore descriptions thereof will be omitted as appropriate.

[0418] The control unit 401 includes the acquisition unit CN2 and the first selection unit CN3 of FIG.

[0419] The acquisition unit CN2 of the control unit 401 acquires a plurality of pieces of processing result information ME3 indicating respective processing results of a plurality of substrates Wp by a plurality of processing units 220. For example, the acquisition unit CN2 may acquire the processing result information ME3 from the substrate processing apparatus 100A via the communication unit 405 and the network NW, and store the information in the storage unit 402. For example, the acquisition unit CN2 may acquire measurement data ME2 from a measuring unit 200 outside the substrate processing apparatus 100A and / or a measuring unit 200 inside the substrate processing apparatus 100A via the communication unit 405 and the network NW, and store the measurement data ME2 in the storage unit 402 as processing result information ME3.

[0420] The control unit 401 may include a pre-processing unit CN1 of Fig. 8. In this case, for example, the acquiring unit CN2 acquires measurement data ME2 from a measuring unit 200 outside the substrate processing apparatus 100A and / or a measuring unit 200 inside the substrate processing apparatus 100A via the communication unit 405 and the network NW. Then, the pre-processing unit CN1 of the control unit 401 processes the measurement data ME2 to generate processing result information ME3.

[0421] The first selection section CN3 of the control unit 401 selects one or more processing units 220 to be used for processing each of the target substrates Wx from among the multiple processing units 220 based on the multiple pieces of processing result information ME3, and generates selection result information ME4. Then, the communication unit 405 transmits the selection result information ME4 to the substrate processing apparatus 100A via the network NW.

[0422] Therefore, the control unit CNT of the substrate processing apparatus 100A can control each processing unit 220 and each transport mechanism 210 so that the processing target substrate Wx is processed by the processing unit 220 selected by the selection result information ME4. As a result, according to the fourth embodiment, even if the processing characteristics of the multiple processing units 220 vary, the processing unit 220 that provides good processing results can be selected and used. Therefore, the uniformity of processing among the multiple processing target substrates Wx can be improved.

[0423] Moreover, the first selection unit CN3 generates inspection target information ME5 and transmits the inspection target information ME5 to the substrate processing apparatus 100A via the communication unit 405 and the network NW.

[0424] In addition, the acquisition unit CN2, the first selection unit CN3, and the pre-processing unit CN1 of the control unit 401 according to the embodiment 4 operate in the same manner as the acquisition unit CN2, the first selection unit CN3, and the pre-processing unit CN1 of the control unit CNT according to the embodiment 1. In addition, the embodiment 4 has the same effects as the embodiment 1.

[0425] (First Modification) A substrate processing system SYS according to a first modified example of the fourth embodiment of the present invention will be described with reference to Fig. 29. The first modified example is mainly different from the second embodiment in that the substrate processing system SYS according to the first modified example has a selection device 400 for selecting a processing unit 220 outside the substrate processing apparatus 100A. The following mainly describes the differences between the second and fourth embodiments of the first modified example.

[0426] The configuration of the substrate processing apparatus 100A according to the first modified example is similar to the configuration of the substrate processing apparatus 100 according to the second embodiment described with reference to Figs. 1 to 7 and 12. However, the substrate processing apparatus 100A does not include, for example, the acquisition unit CN2, the first selection unit CN3, and the learning unit CN9 of Fig. 12. The substrate processing apparatus 100A does not need to include the pre-processing unit CN1 of Fig. 12. The substrate processing apparatus 100A does not store the learning program ME7 and the learning data ME8 of Fig. 12. The substrate processing apparatus 100A does not need to store the measurement data ME2 and the processing result information ME3 of Fig. 12.

[0427] The control unit 401 according to the first modification includes an acquisition unit CN2, a first selection unit CN3, and a learning unit CN9 in Fig. 12. The control unit 401 may also include a pre-processing unit CN1 in Fig. 12. The storage unit 402 further stores a learning program ME7, learning data ME8, and learning model ME9 in Fig. 12. The storage unit 402 also stores the measurement data ME2 and processing result information ME3 in Fig. 12, which is the same as in the fourth embodiment.

[0428] A first selection section CN3 of the control unit 401 according to the first modification selects one or more processing units 220 to be used for processing each of the target substrates Wx from among the multiple processing units 220 based on the multiple pieces of processing result information ME3, and generates selection result information ME4. Then, the communication unit 405 transmits the selection result information ME4 to the substrate processing apparatus 100A via the network NW.

[0429] As a result, in the first modified example, similarly to the fourth embodiment, the uniformity of processing among a plurality of target substrates Wx can be improved in the substrate processing apparatus 100A.

[0430] Additionally, the acquisition unit CN2, the first selection unit CN3, the learning unit CN9, and the pre-processing unit CN1 of the control unit 401 according to the first modified example operate in the same manner as the acquisition unit CN2, the first selection unit CN3, the learning unit CN9, and the pre-processing unit CN1 of the control unit CNT according to the second embodiment, respectively. Additionally, the first modified example has the same effects as those of the second embodiment. The selection device 400 can also be regarded as a "learning device."

[0431] The communication unit 405 may transmit the learning model ME9 to the substrate processing apparatus 100A via the network NW. The learning model ME9 may be stored in a removable medium and provided to the substrate processing apparatus 100A from the removable medium.

[0432] (Second Modification) A substrate processing system SYS according to a second modified example of the fourth embodiment of the present invention will be described with reference to Fig. 29. The second modified example is mainly different from the third embodiment in that the substrate processing system SYS according to the second modified example has a selection device 400 for selecting a processing unit 220 outside the substrate processing apparatus 100A. The following mainly describes the differences between the third and fourth embodiments of the second modified example.

[0433] The configuration of the substrate processing apparatus 100A according to the second modification is the same as that of the substrate processing apparatus 100 according to the third embodiment described with reference to FIGS. 1 to 7, 12, and 20. However, the substrate processing apparatus 100A does not include, for example, the acquisition unit CN2, the first selection unit CN3, and the learning unit CN9 in FIG. 12, and does not include the first learning unit 61, the second learning unit 62, and the third learning unit 63 in FIG. 20. In addition, the substrate processing apparatus 100A may not include the pre-processing unit CN1 in FIG. 12. In addition, the substrate processing apparatus 100A does not store the learning program ME7 and the learning data ME8 in FIG. 12, and does not store the first learning program 71, the second learning program 72, the third learning program 73, the first learning data 81, the second learning data 82, and the third learning data 83 in FIG. 20. Furthermore, the substrate processing apparatus 100A does not necessarily have to store the measurement data ME2 and processing result information ME3 of FIG. 12, and does not necessarily have to store the processing result information set ME30 of FIG.

[0434] The control unit 401 according to the second modification includes the acquisition unit CN2 and the first selection unit CN3 of FIG. 12, and the first learning unit 61, the second learning unit 62, and the third learning unit 63 of FIG. 20. The control unit 401 may also include the pre-processing unit CN1 of FIG. 12. The storage unit 402 stores the first learning program 71, the second learning program 72, the third learning program 73, the first learning data 81, the second learning data 82, the third learning data 83, the first learning model 91, the second learning model 92, and the third learning model 93 of FIG. 20. The storage unit 402 also stores the measurement data ME2 of FIG. 12 and the processing result information set ME30 of FIG. 20, which is the same as in the fourth embodiment.

[0435] A first selection section CN3 of a control unit 401 according to the second modification selects one or more processing units 220 to be used for processing each of the target substrates Wx from among the multiple processing units 220 based on the multiple pieces of processing result information ME3, and generates selection result information ME4. Then, a communication unit 405 transmits the selection result information ME4 to the substrate processing apparatus 100A via the network NW.

[0436] As a result, in the second modified example, similarly to the fourth embodiment, the substrate processing apparatus 100A can improve the uniformity of processing among a plurality of substrates Wx to be processed.

[0437] In addition, the acquisition unit CN2, first selection unit CN3, first learning unit 61, second learning unit 62, third learning unit 63, and pre-processing unit CN1 of the control unit 401 in the second modified example operate in the same manner as the acquisition unit CN2, first selection unit CN3, first learning unit 61, second learning unit 62, third learning unit 63, and pre-processing unit CN1 of the control unit CNT in embodiment 3, respectively. In addition, the second modified example has the same effects as embodiment 3. The selection device 400 can also be considered as a "learning device".

[0438] Furthermore, the communication unit 405 may transmit the first learning model 91, the second learning model 92, and the third learning model 93 to the substrate processing apparatus 100A via the network NW. The first learning model 91, the second learning model 92, and the third learning model 93 may be stored in a removable medium, and the first learning model 91, the second learning model 92, and the third learning model 93 may be provided from the removable medium to the substrate processing apparatus 100A.

[0439] The embodiments of the present invention have been described above with reference to the drawings. However, the present invention is not limited to the above embodiments, and can be implemented in various aspects without departing from the gist of the present invention. In addition, the components disclosed in the above embodiments can be modified as appropriate. For example, a certain component among all the components shown in one embodiment may be added to a component of another embodiment, or some components among all the components shown in one embodiment may be deleted from the embodiment.

[0440] In addition, the drawings are mainly schematic illustrations of each component to facilitate understanding of the invention, and the thickness, length, number, spacing, etc. of each component shown in the drawings may differ from the actual ones due to the convenience of creating the drawings. In addition, the configurations of each component shown in the above embodiment are merely examples and are not particularly limited, and it goes without saying that various modifications are possible within a range that does not substantially deviate from the effects of the present invention. [Industrial Applicability]

[0441] The present invention relates to a substrate processing About The present invention relates to a learning model generation method and a learning model, and has industrial applicability. [Explanation of symbols]

[0442] 61 First Learning Section (Learning Section) 62 Second Learning Section (Learning Section) 63 Third Learning Section (Learning Section) 91 First Learning Model (Learning Model) 92 Second Learning Model (Learning Model) 93 Third Learning Model (Learning Model) 100, 100A Substrate Processing Equipment 200 measurement units 220 Processing Unit 210 Conveyor mechanism 400 Selection Device 401 Control Unit 402 Storage Unit AR, RT, DV, PAHP, PHP, CP, EEW Processing Unit CNT Control Unit CN1 pre-processing section CN2 Acquisition Department CN3 1st selection section (selection section) CN4 analysis section CN5 adjustment section CN6 Second selection section CN7 Board path determination section CN8 Substrate processing control unit CN9 Learning Department MEM Storage Unit ME9 Learning Model

Claims

1. Obtaining training data; generating a learning model by performing machine learning on the plurality of learning data; Including, Each of the plurality of learning data includes two or more pieces of processing result information each indicating a processing result of a learning target substrate by two or more processing units having different processing contents, A learning model generation method in which the learning model is a program that inputs as input information two or more pieces of processing result information each indicating the processing results of a substrate by the two or more processing units having different processing contents, clusters the two or more pieces of processing result information, and outputs as output information clustering information indicating the results of the clustering.

2. A learning model that is constructed by performing machine learning on a plurality of learning data and causes a computer to function so as to cluster processing result information indicating a processing result of a substrate, Each of the plurality of learning data includes two or more pieces of processing result information each indicating a processing result of a learning target substrate by two or more processing units having different processing contents, causing said computer to receive input information and output output information; the input information is two or more pieces of processing result information each indicating a processing result of the substrate by the two or more processing units having different processing contents, A learning model, wherein the output information is clustering information indicating a result of clustering the two or more pieces of processing result information.

Images