Information processor, information processing method, and program

Abstract

To provide a technique advantageous for suppressing abrasion of a substrate and a holding part for holding the substrate.SOLUTION: A control part for controlling a holding part for holding an object is configured to control supply of gas, on the basis of a control parameter provided by an information processor, when holding of the object by the holding part is canceled. The information processor includes: an output part for outputting the control parameter to be provided to the control part, according to an estimation model; and a machine learning part for performing machine learning of the estimation model, on the basis of a result of separation between the object and a holding surface holding the object, which has been executed while controlling the supply of the gas on the basis of the control parameter by the control part in the holding part.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing method, and a program. [Background technology]

[0002] Substrate processing apparatuses used in the manufacturing process of semiconductor devices, liquid crystal display devices, etc. may include a substrate chuck (substrate holding unit) that holds a substrate. The smoothness with which the substrate chuck holds and releases the substrate is related not only to productivity but also to the wear resistance of the substrate and the substrate chuck.

[0003] Patent Document 1 describes supplying gas to a gap between a base surface of a support table and the substrate through gas flow openings formed in the support table when the substrate is unloaded from the support table. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Special Publication No. 20199-505841 Summary of the Invention [Problem to be solved by the invention]

[0005] Even if the substrate chuck chucks a substrate and then releases the chuck, an adhesive force still acts between the substrate and the substrate chuck. If the amount of gas supplied when separating the substrate from the substrate chuck is excessive relative to the adhesive force, the substrate will lift up from the substrate chuck and slide sideways. On the other hand, if the amount of gas supplied when separating the substrate from the substrate chuck is insufficient relative to the adhesive force, the substrate will not separate from the substrate chuck, causing wear of the substrate chuck. Since the adhesive force depends on the characteristics of the substrate and the substrate chuck, it is preferable that the supply of gas for separating the substrate from the substrate chuck is appropriately set each time depending on the adhesive force.

[0006] Furthermore, for example, if the substrate stage is moved at high speed when separating the substrate from the substrate chuck, the substrate and the substrate chuck may slide against each other, which may cause the substrate chuck to wear out. This sliding is also caused by the adhesive force between the substrate and the substrate chuck. In order to reduce the influence of the adhesive force when separating the substrate from the substrate chuck, it is preferable to appropriately set the drive of the substrate stage each time.

[0007] The present invention provides an advantageous technique for suppressing wear of a substrate and a holder that holds the substrate. [Means for solving the problem]

[0008] According to one aspect of the present invention, there is provided an information processing device that provides control parameters of a holding unit that holds an object to a control unit that controls the holding unit, wherein the holding unit has a gas flow path arranged so that gas from a gas supply source is supplied between the object and the holding unit through an opening formed on a surface facing the object, and the control unit is configured to control the supply of gas through the gas flow path and the opening based on the control parameters provided from the information processing device when the holding unit releases the object, and the information processing device has an output unit that outputs the control parameters to be provided to the control unit in accordance with an inference model, and a machine learning unit that performs machine learning of the inference model based on the result of separation of the object and the holding surface that holds the object, performed in the holding unit while the supply of gas is controlled by the control unit based on the control parameters. Effect of the Invention

[0009] According to the present invention, it is possible to provide an advantageous technique for suppressing wear on a substrate and a holder that holds the substrate. [Brief description of the drawings]

[0010] [Figure 1] FIG. 1 is a schematic diagram showing the configuration of an exposure apparatus. [Diagram 2]FIG. 2 is a diagram showing the configuration of a substrate stage. [Diagram 3] FIG. 4 is a diagram showing the functional configuration of a control unit. [Figure 4] FIG. [Diagram 5] FIG. 4 is a diagram for explaining control of gas supply. [Figure 6] 4A to 4C are diagrams illustrating a substrate transfer operation. [Figure 7] FIG. [Figure 8] FIG. 4 is a diagram showing a drive profile of a fine movement stage. [Figure 9] FIG. 13 is a graph showing the time progression of the Z deviation of the fine movement stage. [Figure 10] FIG. 4 is a diagram showing a change in pressure over time during gas supply. [Figure 11] FIG. 1 shows the configuration of a neural network. [Figure 12] Flowchart of the learning phase. [Figure 13] Flowchart of the estimation phase. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] Hereinafter, the embodiments will be described in detail with reference to the attached drawings. Note that the following embodiments do not limit the invention according to the claims. Although the embodiments describe a number of features, not all of these features are essential to the invention, and the features may be combined in any manner. Furthermore, in the attached drawings, the same reference numbers are used for the same or similar configurations, and duplicated descriptions are omitted.

[0012] First Embodiment The present disclosure relates to an information processing apparatus that provides a control parameter of a holding unit to a control unit that controls a holding unit that holds an object. The target object may be a flat surface or an object with a gentle curvature, such as a substrate, an original, or an optical element in a substrate processing apparatus that processes a substrate. In the present disclosure, the substrate processing apparatus may be, for example, a lithography apparatus that forms a pattern on a substrate, such as an exposure apparatus or an imprint apparatus. Alternatively, the substrate processing apparatus may be an inspection apparatus, such as an overlay inspection apparatus, a CD inspection apparatus, a defect inspection apparatus, or an electrical characteristic inspection apparatus. Alternatively, the substrate processing apparatus may be a processing apparatus, such as an etching apparatus or a film formation apparatus. In the following, the substrate processing apparatus in the present disclosure will be described as an exposure apparatus, which is an example of a lithography apparatus. As described in detail below, the substrate processing apparatus may include an original holding unit that holds an original as an object, and a substrate holding unit that holds a substrate as an object.

[0013] FIG. 1 is a schematic diagram of an exposure apparatus 200 in an embodiment. In this specification and the drawings, directions are shown in an XYZ coordinate system with a horizontal plane as the XY plane. In general, a substrate 2 is placed on a substrate chuck 3 of a substrate stage 1 so that its surface is parallel to the horizontal plane (XY plane). Therefore, in the following, directions that are orthogonal to each other in a plane along the surface of the substrate 2 are referred to as the X-axis and Y-axis, and a direction perpendicular to the X-axis and Y-axis is referred to as the Z-axis. In the following, directions parallel to the X-axis, Y-axis, and Z-axis in the XYZ coordinate system are referred to as the X-direction, Y-direction, and Z-direction, respectively, and the rotation direction around the X-axis, the rotation direction around the Y-axis, and the rotation direction around the Z-axis are referred to as the θX direction, the θY direction, and the θZ direction, respectively.

[0014] In the exposure apparatus 200, exposure light guided from a light source (not shown) through an illumination optical system 202 is irradiated onto an original 203. The original is also called a reticle or a mask. The exposure light that has passed through the original 203 passes through a projection optical system 206 and is irradiated onto a substrate 2. Examples of materials that can be used for the substrate include glass, ceramics, metals, semiconductors, and resins. If necessary, a member made of a material different from that of the substrate may be provided on the surface of the substrate. The substrate is, for example, a silicon substrate, a compound semiconductor substrate, or quartz glass.

[0015] The original stage 204 is equipped with an original chuck (not shown) which is an original holding section. The original stage 204 holds the original 203 by the original chuck. The original stage 204 can move linearly in the X, Y, and Z directions, and can rotate in the θX, θY, and θZ directions, and each of these axes can be controlled independently.

[0016] A laser interferometer (not shown) is provided to measure with high precision the position of the original stage 204. The position of the original stage 204 can be identified from the amount of change in the measurement value obtained by the laser interferometer.

[0017] The original transport system 214 may include an original hand 215, an original pre-alignment stage 216, an original transport robot 217, and a storage unit 218. The original transport robot 217 is an articulated robot. The original transport robot 217 has a hand that holds the original 203, and can move this hand to any position in the XYZ space. The storage unit 218 stores the original 203. The original 203 can be placed in the storage unit 218 from outside the apparatus. The original 203 placed in the storage unit 218 is placed on the original pre-alignment stage 216 by the original transport robot 217, which is an articulated robot.

[0018] The position of the original 203 on the original pre-alignment stage 216 is measured by observing the marks on the original 203 with a microscope (not shown) on the original pre-alignment stage 216. After the measurement, the original hand 215 holds the original 203 on the original pre-alignment stage 216 and drives it in the +Y direction along a guide (not shown) to a position where the original 203 is transferred to the original stage 204. After this drive is completed, the original hand 215 drives in the -Z direction to transfer the original 203 to the original stage 204. The original stage 204 holds the original 203 by vacuum suction. The deviations of the original 203 in the X direction, Y direction, and θZ direction relative to the original pre-alignment stage 216 measured by the original pre-alignment stage 216 can be corrected by changing the position where the original stage 204 receives the original 203.

[0019] The substrate transport system 210 may include a supply hand 211, a collection hand 212, a storage unit 213, a pre-alignment unit 219, a transport robot 220, and a temporary placement table 221. The substrate 2 is stored in the storage unit 213. The transport robot 220 is an articulated robot. The transport robot 220 has a hand for holding the substrate 2, and can move this hand to any position in the XYZ space. The transport robot 220 takes the substrate 2 from the storage unit 213 and places the substrate 2 on the pre-alignment unit 219. In the pre-alignment unit 219, a measurement light is irradiated onto the outer periphery of the substrate 2, and the reflected light is measured by a sensor, thereby measuring the outer position of the substrate 2 relative to the pre-alignment unit 219.

[0020] The supply hand 211 is a transport mechanism that transports the substrate 2 from the pre-alignment unit 219 to the substrate stage 1, specifically, onto the pin members 9 configured in the coarse movement stage 13. The supply hand 211 can hold the substrate 2 by vacuum suction. The recovery hand 212 is a transport mechanism that recovers the substrate 2 from the substrate stage 1 and transports it to the temporary placement table 221. The recovery hand 212 can hold the substrate 2 by vacuum suction.

[0021] The temporary placement table 221 is a table for temporarily placing the substrate 2 that has been retrieved from the substrate stage 1 by the retrieval hand 212. Since the transfer robot 220 cannot directly pick up the substrate from the retrieval hand 212, the transfer robot 220 transfers the substrate 2 via the temporary placement table 221.

[0022] The substrate stage 1 may include a coarse movement stage 13 and a fine movement stage 8 mounted on the coarse movement stage 13. A substrate chuck 3 is mounted on the fine movement stage 8. The substrate chuck 3 is a substrate holding device that holds the substrate 2. The substrate chuck 3 may be a vacuum chuck that adsorbs the lower surface of the substrate 2 to the holding surface by vacuum suction. Alternatively, the substrate chuck 3 may be an electrostatic chuck that adsorbs the lower surface of the substrate 2 to the holding surface by electrostatic action. In the following, the substrate chuck 3 will be described as a vacuum chuck. The fine movement stage 8 is configured to be linearly movable in the X, Y, and Z directions, and to be rotatable in the θX, θY, and θZ directions. The substrate stage 1 includes a driving mechanism (not shown) that drives the fine movement stage 8 along each of these axes, and each of these axes can be controlled independently.

[0023] A laser interferometer (not shown) is provided to measure with high precision the position of the fine movement stage 8. The position of the fine movement stage 8 can be identified from the amount of displacement of the measurement value obtained by the laser interferometer.

[0024] The pin members 9 are used when receiving the substrate 2 from the supply hand 211 and when transferring the substrate 2 to the collection hand 212. The pin members 9 can hold the substrate 2 by suction. In order to stably support the substrate 2, the pin members 9 may include at least three pin members arranged at or near the apexes of a triangle as large as possible within the holding surface. The pin members 9 pass through holes formed in the fine movement stage 8 and the substrate chuck 3, and are arranged so that the longitudinal axis of the pin members is parallel to the direction perpendicular to the holding surface of the substrate chuck 3 (holding unit) that holds the substrate (i.e., the Z direction). The lower end of the pin member 9 is fixed to the coarse movement stage 13, and the upper end is configured to be able to protrude and retract from the holding surface of the substrate chuck 3. The pin members 9 protrude above the holding surface to serve as a support member that supports the substrate 2. The pin members 9 can protrude and retract from the holding surface of the substrate chuck 3 by at least one of the following: the extension and contraction of the pin members 9, the vertical movement of the coarse movement stage 13, and the vertical movement of the fine movement stage 8.

[0025] The coarse movement stage 13 can move linearly in the X and Y directions and rotate in the θZ direction. For example, a capacitance sensor is used to measure the distance between the coarse movement stage 13 and the fine movement stage 8. Based on the result of this measurement, an actuator such as a linear motor is used to control the position of the coarse movement stage 13 so that it follows the position of the fine movement stage 8. The pin member 9 is fixed to the coarse movement stage 13 and therefore moves together with the coarse movement stage 13.

[0026] The off-axis scope 209 can measure alignment marks on the substrate 2 .

[0027] The control unit 100 controls the operation of the exposure apparatus 200. The control unit 100 can include the functions of a driver and a controller that control sensors, actuators, and the like in the exposure apparatus 200.

[0028] The information processing device 700 is communicably connected to the control unit 100 that controls the exposure apparatus 200. The information processing device 700 can provide the control unit 100 that controls the exposure apparatus 200 (i.e., the holding unit that holds the object) with control parameters of the holding unit. The information processing device 700 can include a processing unit 701, a display unit 705, and an input unit 706. The processing unit 701 can be configured by a computer including a processor such as a CPU (Central Processing Unit), a memory, and a GPU (Graphical Processing Unit). The processing unit 701 can include an acquisition unit 702, a calculation unit 703, a supply unit 704, and a storage unit 707. The acquisition unit 702 acquires information from the exposure apparatus 200. The calculation unit 703 calculates the control parameters based on the information acquired by the acquisition unit 702. The control parameters can include parameters related to gas supply and parameters related to drive control of the fine movement stage. The supply unit 704 supplies (transmits, provides) control parameters to the control unit 100 of the exposure apparatus 200. The storage unit 707 stores information acquired from the exposure apparatus 200. The acquisition unit 702 and the supply unit 704 may be understood as units that constitute a communication unit that communicates with the control unit 100 that controls the exposure apparatus 200 to send and receive information.

[0029] The recovery sequence of the substrate 2 in this embodiment will be described below. The recovery sequence of the substrate 2 may include an operation of transferring the substrate 2 from the substrate chuck 3 to the pin members 9 after the exposure is completed. This operation may be performed by driving the substrate chuck 3 and the pin members 9 relatively in the Z direction. Here, an adhesive force acts between the substrate 2 and the substrate chuck 3. In this specification, the adhesive force refers to a force required to apply a force to the substrate 2 and the substrate chuck 3 in a direction in which they move away from each other to separate them. That is, the adhesive force is related to the ease of separation of the substrate 2 from the substrate chuck 3. Even if the substrate 2 is unloaded, that is, even if the substrate chuck 3 releases the substrate 2 from its holding, the adhesive force may remain depending on the characteristics of the holding surface of the substrate chuck 3, the characteristics of the substrate 2, the characteristics of the exposure process, the influence of foreign matter mixed between the substrate 2 and the substrate chuck 3, and the like. In addition, even if the vacuum suction is released when the substrate is unloaded, the pressure does not immediately become atmospheric pressure, and the pressure that does not escape may become the adhesive force. Furthermore, even when the substrate chuck 3 is an electrostatic chuck, an adhesive force may remain due to the electrostatic chuck or the substrate being charged when the power is off. If the adhesive force is strong, the substrate 2 will be pulled away from the substrate chuck 3 against the adhesive force, which may damage the substrate 2, the substrate chuck 3, or both.

[0030] In this embodiment, therefore, when the substrate 2 is released from the substrate chuck 3, gas is supplied between the substrate 2 and the substrate chuck 3 to facilitate separation of the two. However, if the amount of gas supplied when separating the substrate 2 from the substrate chuck 3 is excessive relative to the adhesive force, the substrate 2 will float up from the substrate chuck 3 and slide sideways. On the other hand, if the amount of gas supplied when separating the substrate 2 from the substrate chuck 3 is insufficient relative to the adhesive force, the substrate 2 will not separate from the substrate chuck 3, causing wear of the substrate chuck 3. Since the adhesive force is due to the characteristics of the substrate 2, the characteristics of the substrate chuck 3, the characteristics of the exposure process, and the like, the supply of gas for separating the substrate 2 from the substrate chuck 3 needs to be appropriately set each time according to the adhesive force.

[0031] Furthermore, for example, if the fine movement stage 8 is moved at high speed when separating the substrate 2 from the substrate chuck 3, the substrate 2 and the substrate chuck 3 may slide against each other, which may cause wear on the substrate chuck 3. This sliding is also caused by the adhesive force between the substrate 2 and the substrate chuck 3. In order to reduce the influence of the adhesive force and separate the substrate 2 from the substrate chuck 3, the drive of the fine movement stage 8 must also be appropriately set each time.

[0032] The following describes the holding and release of the substrate 2 by the substrate chuck 3, and the control of the supply of gas between the substrate 2 and the substrate chuck 3.

[0033] 2 is a schematic diagram showing the configuration of the substrate stage 1. The substrate chuck 3 is configured to hold the substrate 2 by reducing the pressure in the space between the substrate 2 and the holding surface of the substrate chuck 3 and vacuum-adhering the substrate 2 to the holding surface. The substrate chuck 3 is held on the fine movement stage 8 by vacuum adsorption. The fine movement stage 8 is mounted on the coarse movement stage 13.

[0034] A gas flow path 10 is disposed within the coarse movement stage 13. The gas flow path 10 is disposed so that gas is sucked from a vacuum pump 5 (vacuum source) serving as negative pressure generating means through an opening formed on a surface of the substrate chuck 3 facing the substrate 2. The gas flow path 10 is also disposed so that gas is supplied from a positive pressure tank 6 serving as positive pressure generating means (gas supply source) through the opening. In addition, a pressure sensor 20, regulators 41, 61, 62, 63, and solenoid valves 31, 51, 52, 53 are provided on the gas flow path 10 within the coarse movement stage 13.

[0035] A plurality of openings 7 (see FIG. 4) are formed on the surface of the substrate chuck 3 facing the substrate 2. The gas flow path 10 branches and connects to each opening via the fine movement stage 8. The gas flow path 10 is also connected to a vacuum pump 5 via an electromagnetic valve 31 and a regulator 41, and is connected to a positive pressure tank 6 via electromagnetic valves 51, 52, 53 and regulators 61, 62, 63. In general, the vacuum pump 5 and the positive pressure tank 6 are configured as factory equipment, and therefore are installed outside the substrate stage 1.

[0036] The pressure inside the gas flow path 10 is measured by a pressure sensor 20. In one example, the pressure sensor 20 is arranged so as to measure the pressure inside the gas flow path 10 at a position closer to the vacuum pump than the branch of the gas flow path 10 and closer to the substrate chuck 3 than the solenoid valves 31, 51, 52, and 53. This is to enable the pressure sensor 20 to accurately measure the pressure in the space between the substrate 2 and the substrate chuck 3.

[0037] The regulator 41 and the solenoid valve 31 are disposed between the vacuum pump 5 and the substrate chuck 3. The regulator 41 is configured to adjust the pressure in the gas flow path 10 to a desired vacuum pressure. The solenoid valve 31 is disposed on the substrate chuck 3 side of the regulator 41 in order to supply the pressure adjusted by the regulator 41 to the substrate chuck 3. When the solenoid valve 31 is turned ON, the vacuum pump 5 adjusts the pressure in the space between the substrate 2 and the substrate chuck 3. When the solenoid valve 31 is turned OFF, the pressure adjustment in the space between the substrate 2 and the substrate chuck 3 is no longer performed.

[0038] The regulators 61, 62, 63 and the solenoid valves 51, 52, 53 are disposed between the positive pressure tank 6 and the substrate chuck 3. The regulators 61, 62, 63 are configured to adjust the supply pressure from the positive pressure tank 6 in the gas flow path 10. The ability to adjust the supply pressure means that the supply flow rate can also be adjusted. The solenoid valves 51, 52, 53 are disposed on the gas flow path 10 on the substrate chuck 3 side with respect to the regulators 61, 62, 63. When the solenoid valves 51, 52, 53 are turned ON, the positive pressure tank 6 adjusts the pressure in the space between the substrate 2 and the substrate chuck 3. When the solenoid valves 51, 52, 53 are turned OFF, the pressure in the space between the substrate 2 and the substrate chuck 3 is no longer adjusted. Note that, although FIG. 2 shows three sets of regulators and solenoid valves between the positive pressure tank 6 and the substrate chuck 3, the number of sets is not limited to a specific number. There may be one set, or there may be four or more sets. Also, instead of the combination of the solenoid valve and the regulator, a servo valve whose opening and closing degree can be freely set may be configured.

[0039] When the substrate 2 is vacuum-sucked by the substrate chuck 3, the solenoid valve 31 is turned ON, and the solenoid valves 51, 52, and 53 are turned OFF. This allows the pressure in the space between the substrate 2 and the substrate chuck 3 to be reduced by the vacuum pump 5. When the substrate 2 is to be released from the substrate chuck 3, the solenoid valve 31 is turned OFF, and at least one of the solenoid valves 51, 52, and 53 is turned ON. This allows the space between the substrate 2 and the substrate chuck 3 to be pressurized with gas from the positive pressure tank 6.

[0040] By changing the settings of the regulators 61, 62, and 63, respectively, it is possible to change the flow rate and pressure of the gas filling the space between the substrate 2 and the substrate chuck 3 when any one of the solenoid valves 51, 52, and 53 is turned on. In one example, the regulators 61, 62, and 63 are set so that the flow rate of the gas supplied increases in the order of the regulators 61, 62, and 63. When releasing the substrate 2 from the substrate chuck 3, the control unit 100 controls the supply of gas through the gas flow path 10 and the opening 7 based on the control parameters provided by the information processing device 700. The control parameters include information on the ease of separation of the substrate 2 from the holding surface. For example, the control unit 100 determines which of the solenoid valves 51, 52, and 53 to turn on and the gas supply time based on the information on the ease of separation of the substrate 2 from the holding surface.

[0041] The control unit 100 may be configured, for example, by a general-purpose or dedicated processor with a program built into a memory. Alternatively, the control unit 100 may be configured by a PLD (abbreviation of Programmable Logic Device) such as an FPGA (abbreviation of Field Programmable Gate Array) or an ASIC (abbreviation of Application Specific Integrated Circuit). Alternatively, the control unit 100 may be configured by a combination of all or part of a processor, a PLD, and an ASIC. The control unit 100 may be configured inside the exposure apparatus 200 or outside the exposure apparatus 200. When the control unit 100 is configured outside the exposure apparatus 200, for example, the function of the control unit 100 is installed in a server device or the like communicably connected to the exposure apparatus 200. The function of the control unit 100 may be realized by an information processing device 700.

[0042] 3 shows the functional configuration of the control unit 100. The control unit 100 can include an instruction unit 110, a storage unit 120, and a processing unit 130. The instruction unit 110 instructs the amount of control for each driving unit including the fine movement stage 8. The storage unit 120 stores various data and programs related to exposure control. The processing unit 130 executes various processes related to exposure control. For example, the processing unit 130 calculates the drive target positions of the substrate stage 1 and the original stage 204. The processing unit 130 also executes an exposure sequence and a substrate supply and recovery sequence according to the control program stored in the storage unit 120.

[0043] The function of the processing unit 130 will be described later. Note that, although an example in which the control unit 100 includes the storage unit 120 and the processing unit 130 is shown in FIG.

[0044] Fig. 4 is a plan view of the substrate chuck 3 as viewed from the +Z direction. An annular seal portion 60 is provided at a position along the outer periphery of the substrate chuck 3. However, the presence or absence, number, and arrangement of the seal portion 60 are not limited to the example shown in Fig. 4. For example, the substrate chuck 3 may not be provided with a seal portion. Alternatively, a plurality of seal portions may be provided.

[0045] The surface of the substrate chuck 3 is provided with a plurality of protrusions (not shown), and the upper surfaces of these protrusions can serve as a holding surface for holding the substrate 2. These protrusions are called chuck pins. The substrate chuck 3 has openings 7 connected to a gas flow path 10. Although twelve openings 7 are shown in FIG. 4, the arrangement and number of the openings 7 are not limited to the example shown in FIG. 4.

[0046] The operation of vacuum-adhering the substrate 2 to the substrate chuck 3 is performed by exhausting the gas in the space between the substrate 2 and the substrate chuck 3 through the opening 7. The provision of a seal portion 60 between the substrate 2 and the substrate chuck 3 prevents the inflow of air from the surroundings, making vacuum adsorption possible.

[0047] The operation of releasing the substrate 2 from the substrate chuck 3 is performed by supplying gas between the substrate 2 and the substrate chuck 3 through the opening 7. If the amount of gas supplied (pressure and flow rate) or the gas supply time is excessive relative to the adhesive force, the substrate 2 will lift up from the substrate chuck 3 and slide sideways. On the other hand, if the amount of gas supplied or the gas supply time when separating the substrate 2 from the substrate chuck 3 is insufficient relative to the adhesive force, the substrate 2 will not separate from the substrate chuck 3, causing wear of the substrate chuck when the substrate 2 is picked up. A detailed explanation of the adhesive force and a method for optimizing the amount of gas supplied relative to the adhesive force will be given later.

[0048] A method for transferring the substrate 2 from the substrate chuck 3 to the pin members 9 will be described with reference to Fig. 6. Fig. 6 shows a schematic cross-sectional view of the substrate stage 1. The fine movement stage 8 moves in the Z direction along the longitudinal direction of the pin members 9. The fine movement stage 8 can move in the Z direction while tilting in accordance with the Z positional relationship of the three pin members 9. The pin members 9 are placed on a coarse movement stage 13 that can be driven in the X and Y directions. The Z direction position of the pin members 9 can be measured using, for example, an interferometer or a capacitance sensor (not shown).

[0049] The pin member 9 is connected to the vacuum pump 5 via a gas flow path 70. A regulator 73 and an electromagnetic valve 72 are provided between the vacuum pump 5 and the pin member 9. The regulator 73 is configured to adjust the vacuum pressure to a desired level. The electromagnetic valve 72 is disposed closer to the pin member 9 than the regulator 73. This is to supply the pressure adjusted by the regulator 73 to the substrate chuck 3. When the electromagnetic valve 72 is turned on, the vacuum pump 5 adjusts the pressure in the space between the substrate 2 and the substrate chuck 3, and when the electromagnetic valve 72 is turned off, the pressure adjustment in the space between the substrate 2 and the substrate chuck 3 is no longer performed. In addition, a pressure sensor 71 is disposed to monitor the pressure between the pin member 9 and the substrate 2. The pressure sensor 71 needs to be disposed closer to the pin member 9 than the electromagnetic valve 72. In FIG. 8, the pressure sensor 71, the electromagnetic valve 72, and the regulator 73 are disposed outside the coarse movement stage 13, but they may be disposed inside the coarse movement stage.

[0050] In Fig. 6(a), the fine movement stage 8 is positioned on the +Z direction side, and the substrate chuck 3 vacuum-adsorbs the substrate 2. In Fig. 6(b), the fine movement stage 8 moves in the -Z direction, causing the pin members 9 to come into contact with the substrate 2, and the fine movement stage 8 stops moving. The control unit 100 then releases the substrate chuck 3 from its suction hold on the substrate 2. At this time, the control unit 100 controls the supply of gas via the gas flow path 10 and the opening 7, based on information on the ease of separation of the substrate 2 from the holding surface (for example, the adhesive force between the substrate 2 and the substrate chuck 3).

[0051] Thereafter, the solenoid valve 72 is turned ON, so that the pin members 9 adsorb and hold the substrate 2. Then, as shown in Fig. 6(c), while the pin members 9 adsorb and hold the substrate 2, the fine movement stage 8 further moves in the -Z direction. In this manner, the substrate 2 is transferred from the substrate chuck 3 to the pin members 9.

[0052] 6, the fine movement stage 8 moves to transfer the substrate 2 from the substrate chuck 3 to the pin members 9, but it is only necessary that the relative positions of the substrate chuck 3 and the pin members 9 can be changed. For example, the entirety or the tip of the pin members 9 may move in the +Z direction. Changing the relative positions of the substrate chuck 3 and the pin members 9 is synonymous with moving the substrate chuck 3 and the pin members 9 relative to each other.

[0053] 7 is a plan view of the substrate chuck 3 as viewed from the +Z direction. A plurality of protrusions (chuck pins) (not shown) are provided on the surface of the substrate chuck 3, and the upper surfaces of these plurality of protrusions can serve as a holding surface for holding the substrate 2. An annular seal portion 60 is provided on the outer periphery of the substrate chuck 3 to prevent air from flowing out of the substrate chuck 3 when the substrate 2 is vacuum-sucked. The substrate chuck 3 has an opening 7 connected to a gas flow path 10. The substrate chuck 3 vacuum-sucks the substrate 2 through the opening 7. The space between the substrate 2 and the chuck pins is in a vacuum state, and the substrate 2 can be sucked and held with a uniform force.

[0054] The substrate chuck 3 and the fine movement stage 8 are formed with holes 14 for passing the pin members 9 in the Z direction, and the fine movement stage 8 can be driven in the Z direction without interfering with the pin members 9. Fig. 7 shows the holes 14 through which the pin members 9 pass. Fig. 7 shows a configuration in which three pin members 9 are installed, but the number of pin members 9 is not limited as long as the substrate 2 can be stably held depending on the size, material, and mass of the substrate 2.

[0055] In the example shown in FIG. 6, the fine movement stage 8 moves to cause the pin members 9 to adsorb and hold the substrate 2, but it is sufficient as long as the relative position between the substrate chuck 3 and the pin members 9 can be changed. For example, the entire pin members 9 or the tip portions thereof may move in the +Z direction. Changing the relative position between the substrate chuck 3 and the pin members 9 is synonymous with moving the substrate chuck 3 and the pin members 9 relatively. Alternatively, a configuration may be adopted in which the pin members 9 adsorb and hold the substrate 2 by extending the pin members 9 rather than by moving the fine movement stage 8. In this way, the substrate stage 1 may include a drive unit that drives the fine movement stage 8 and / or drives or deforms the pin members 9 to perform the operation of separating the substrate 2 from the substrate chuck 3.

[0056] Fig. 8 shows a drive profile of the fine movement stage 8. Fig. 8 shows the relationship between time and the Z direction position of the fine movement stage 8 in the operation of transferring the substrate 2 from the substrate chuck 3 to the pin members 9. Note that the amount of change in the Z direction position of the fine movement stage 8 is synonymous with the amount of change in the Z direction position of the substrate chuck 3.

[0057] In the first section, the instruction unit 110 instructs the fine movement stage 8 to move in the -Z direction at high speed using the first position stored in the storage unit 120 as the target position. When the fine movement stage 8 is driven in the -Z direction, the first position is set to be the same as the Z direction position of the fine movement stage 8 when the substrate 2 and the pin member 9 come into contact with each other, or a position on the +Z direction side of that. Here, an example is shown in which the fine movement stage 8 moves in the -Z direction as shown in FIG. 6. However, when a configuration is adopted in which the pin member 9 moves in the +Z direction to transfer the substrate, the -Z direction and the +Z direction can be interchanged for application, since only the direction of movement is different. In the figures and explanations described below, when the pin member 9 is moved in the +Z direction, the -Z direction and the +Z direction can be interchanged for application. In addition, in a configuration in which both the fine movement stage 8 and the pin member 9 are driven, the movement of the fine movement stage 8 in the Z direction may be replaced with an explanation of the relative movement of the fine movement stage 8 and the pin member 9.

[0058] In the second section, the control unit 100 moves the fine movement stage 8 at a low speed in the -Z direction and monitors the drive current value in driving the fine movement stage 8. By monitoring the drive current value when moving the fine movement stage 8 in the -Z direction, it is possible to detect the drive current value that changes when the substrate 2 and the pin members 9 come into contact with each other, and to detect that the substrate 2 and the pin members 9 have come into contact with each other. The reason for moving the fine movement stage 8 at a low speed in the second section is that if the substrate 2 comes into contact with the pin members 9 at high speed, the substrate 2 and the pin members 9 may be damaged. When the control unit 100 detects that the pin members 9 and the substrate 2 have come into contact with each other by monitoring the drive current value, it stops the movement of the fine movement stage 8 in the -Z direction and ends the second section. The control unit 100 may update the value of the first position stored in the storage unit 120 with the value of the Z direction position of the fine movement stage 8 when the contact between the pin members 9 and the substrate 2 is detected in the second section. By updating the first position in this way, the time required for the second section can be reduced.

[0059] Furthermore, if the coordinates of the fine movement stage 8 when the substrate 2 and pin members 9 come into contact are the same as the first position, the substrate 2 and pin members 9 come into contact when moved to the first position. In this case, contact between the substrate 2 and pin members 9 is detected by the drive current value at the same time as the start of the second section, so the fine movement stage 8 is not moved at low speed in the second section, but transitions to the third section. Note that if it is known that the coordinates of the fine movement stage 8 when the substrate 2 and pin members 9 come into contact are the same as the first position, the second section may be omitted, and a transition to the third section may be made after the first section.

[0060] In the third section, the control unit 100 stops suction of the gas between the substrate 2 and the substrate chuck 3, and in parallel with this, supplies gas between the substrate 2 and the substrate chuck 3. At this time, by supplying gas according to the adhesive force, the phenomenon that the substrate 2 floats from the substrate chuck 3 and slides sideways is prevented. In addition, in the fourth section described later, the substrate 2 is forcibly separated from the substrate chuck 3 while the adhesive force remains, and thus the wear of the substrate chuck 3 caused by the sliding between the substrate 2 and the substrate chuck 3 can be suppressed. In the third section, the pin members 9 also start suctioning the gas between the substrate 2 and the pin members 9. In the third section, the control unit 100 estimates the adhesive force based on the characteristics of the substrate 2, the characteristics of the substrate chuck 3, the characteristics of the exposure process, etc., optimizes the gas supply parameters, and can supply the gas. The estimation of the adhesive force and the supply of the gas will be described later.

[0061] In the fourth section, the control unit 100 moves the fine movement stage 8 at a low speed in the -Z direction to separate the substrate 2 and the substrate chuck 3 by a small distance. If the fine movement stage 8 is moved at a high speed in the -Z direction to forcibly separate the substrate 2 and the substrate chuck 3 while there is an adhesive force, the substrate 2 will slide sideways on the substrate chuck 3, causing wear on the substrate chuck 3. Wear on the substrate chuck 3 will affect the subsequent suction and holding of the substrate 2. Therefore, in order to reduce wear on the substrate chuck 3, gas is supplied between the substrate 2 and the substrate chuck 3 in the third section according to the adhesive force.

[0062] In the fifth section, the fine movement stage 8 is moved at high speed in the −Z direction to a final position in the Z direction.

[0063] If the adhesive force between the substrate 2 and the substrate chuck 3 can be reduced by supplying gas in the third section, the moving speed of the fine movement stage 8 in the fourth section can be increased. For example, it is also possible to eliminate the fifth section described below and drive the fine movement stage 8 at high speed in the -Z direction to the final position in the Z direction.

[0064] A method for estimating adhesive force will be described. One of the indicators showing the ease of separation of the substrate is the adhesive force between the substrate 2 and the (holding surface of) the substrate chuck 3. In other words, information on the ease of separation of the substrate may include information on the adhesive force. The adhesive force is the resultant force of the suction pressure acting between the substrate 2 and the substrate chuck 3 and the adhesive force between the substrate 2 and each chuck pin of the substrate chuck 3. The adhesive force may be estimated based on the characteristics of the substrate 2, the characteristics of the substrate chuck 3, the characteristics of the exposure process, etc. In one example, the adhesive force may be estimated based on at least one of the characteristics of the substrate 2 and the characteristics of the substrate chuck 3.

[0065] The characteristics of the substrate 2 are based on at least one of, for example, the amount of warping of the substrate 2, the roughness of the surface (back surface) of the substrate 2 that contacts the holding surface of the substrate chuck 3, the type of film formed on the substrate 2, the material of the substrate 2, and the thickness of the substrate 2. For example, if the substrate 2 is warped, the adhesion between the substrate 2 and the substrate chuck 3 is reduced. This is because the warping of the substrate 2 causes stress in the direction in which the substrate 2 moves away from the substrate chuck 3, decreasing the degree of adhesion. On the other hand, if the substrate 2 is not warped, the degree of adhesion is high and the adhesion is strong. The roughness of the back surface of the substrate 2 also affects the adhesion. For example, if the back surface of the substrate 2 is not rough, the substrate 2 and the substrate chuck 3 are firmly attached to each other. This is a state in which the adhesion is strong. On the other hand, if the back surface of the substrate 2 is rough, the substrate 2 and the substrate chuck 3 are not in close contact with each other, resulting in a weak adhesion. In addition, if a film is formed on the back surface of the substrate, the type of film may affect the adhesion. Furthermore, the material and thickness of the substrate 2 also affect the weight of the substrate 2. The heavier the substrate, the stronger the adhesion between the substrate 2 and the substrate chuck 3. This is a state in which the adhesive force is strong.

[0066] The characteristics of the substrate chuck 3 are based on at least one of the uniformity of the chuck pin height (flatness of the holding surface), the surface coating of the chuck pin (surface roughness of the holding surface), the thickness and material of the chuck pin, etc. The more uniform the chuck pin height, the stronger the adhesion between the substrate 2 and the substrate chuck 3, resulting in a stronger adhesive force. Also, the thicker the chuck pin, the larger the contact area with the substrate, resulting in a stronger adhesive force. If the chuck pin is made of a material with fine particles, for example, the substrate 2 and the substrate chuck 3 will adhere firmly to each other, resulting in a stronger adhesive force.

[0067] The characteristics of the exposure process may include the exposure time, the exposure amount, and the presence or absence of θ correction drive. For example, if the exposure time is long, the time during which the substrate 2 and the substrate chuck 4 are vacuum-adsorbed is long, and the adhesion between the substrate 2 and the substrate chuck 4 becomes strong. If the exposure amount is large, the substrate 2 thermally expands. During exposure, the substrate 2 is vacuum-adsorbed on the substrate chuck 3, so the back surface of the substrate 2 does not expand, although internal stress acts on it. However, when the vacuum adsorption is released, the internal stress is released and the substrate 2 expands. When the substrate 2 shifts slightly laterally on the substrate chuck 3, the adhesive force between the substrate 2 and the substrate chuck 3 weakens. Also, in the exposure sequence, when the alignment mark is observed with an off-axis scope, the substrate 2 may be shifted in the θZ direction on the substrate chuck 3. In order to rotate the substrate 2 in the θZ direction on the substrate stage 1, an operation called θ correction drive as follows may be performed. An example will be described in which the substrate 2 is rotated by 200 μrad. The control unit 100 rotates the fine movement stage 8 by 200 μrad from the 0 μrad position. Next, the control unit 100 transfers the substrate 2 from the substrate chuck 3 to the pin member 9. In this state, the control unit 100 rotates the fine movement stage 8 by -200 μrad and returns it to the 0 μrad position. The substrate 2 on the pin member 9 is not affected by this rotation. In this state, the control unit 100 transfers the substrate 2 to the substrate chuck 3 again. Then, the substrate 2 is rotated by 200 μrad relative to the substrate chuck 3. This method makes it possible to correct the deviation in the θZ direction. Once, the substrate 2 is released from the substrate chuck 3 and vacuum-adsorbed again, which is equivalent to a change in the exposure time. In this way, the adhesion force also changes depending on the exposure process.

[0068] In this way, if the respective characteristics of the substrate 2, the substrate chuck 3, and the exposure process are known, it is possible to estimate the adhesive force acting between the substrate 2 and the substrate chuck 3. This adhesive force can be estimated using a simulator (not shown) that models the substrate 2 and the substrate chuck 3.

[0069] Next, a simulation of the change in adhesive force when the amount of gas supplied and the time of gas supply are changed will be described. The greater the amount of gas supplied, the weaker the adhesive force between the substrate 2 and the substrate chuck 3. Also, even if the amount of gas supplied is the same, the longer the supply time, the weaker the adhesive force between the substrate 2 and the substrate chuck 3. Parameters of the amount of gas supplied and the time of gas supply are determined so that the adhesive force approaches zero after gas supply. These are set as optimal parameters for gas supply. In this way, information on the ease of separation of the substrate 2 from the holding surface of the substrate chuck 3 can be obtained in advance.

[0070] Control of gas supply will be described with reference to Fig. 5(a). Fig. 5(a) shows an example of the transition of the gas supply flow rate over time when the substrate 2 is released from the substrate chuck 3. Fig. 5(a) shows that the gas supply flow rate is constant when each solenoid valve is turned ON. Since the gas supply flow rates are set to increase in the order of regulators 61, 62, and 63, the gas supply flow rates supplied to the substrate 2 and substrate chuck 3 increase in the order of solenoid valves 51, 52, and 53 being turned ON.

[0071] FIG. 5(a) is a diagram assuming a case where the adhesive force between the substrate 2 and the substrate chuck 3 is large. The gas supply amount and gas supply time are determined based on the results of a simulation performed in advance using the characteristics of the substrate 2 and the characteristics of the substrate chuck 3. The determined gas supply amount and gas supply time are stored in the memory unit 120. In FIG. 5(a), the control unit 100 (instruction unit 110) turns on the solenoid valve 51 in accordance with the gas supply amount stored in the memory unit 120. The length of time for which the solenoid valve 51 is turned on also follows the gas supply time stored in the memory unit 120.

[0072] FIG. 5(b) illustrates an example of a change in the adhesive force between the substrate 2 and the substrate chuck 3 while gas is being supplied under the gas supply control shown in FIG. 5(a). During gas supply, the adhesive force gradually weakens. If the adhesive force is weakened by gas supply to the optimum point shown in FIG. 5(b), the substrate 2 will not slip on the substrate chuck 3 and the substrate chuck 3 will not wear out. The point at which the adhesive force is 0 is not set as the optimum point because a small amount of adhesive force is left to prevent the substrate 2 from slipping.

[0073] By changing the solenoid valve that is turned ON, it is possible to control the amount of gas supplied between the substrate 2 and the substrate chuck 3. Also, it is possible to control the solenoid valve to be turned OFF and stop the gas supply after a predetermined time has elapsed since the start of the gas supply. This makes it possible to create a state in which the substrate 2 does not slide sideways on the substrate chuck 3 and the substrate chuck 3 is not worn.

[0074] A method for measuring the adhesive force from the Z deviation will be described below. As described above, the fine movement stage 8 can be driven in the Z direction (first direction perpendicular to the holding surface) by a driving mechanism (driving unit) (not shown). The position of the fine movement stage 8 in the Z direction at that time can be measured by a laser interferometer (measuring unit) (not shown).

[0075] The Z deviation refers to the control deviation in the Z direction by the driving unit. Specifically, the Z deviation refers to the difference between the target position of the substrate chuck 3 in the Z direction and the current position of the substrate chuck 3 in the Z direction measured by a laser interferometer (measurement unit). This is equivalent to the difference between the target position of the fine movement stage 8 in the Z direction indicated by the indication unit 110 and the position of the fine movement stage 8 in the Z direction measured by a laser interferometer (not shown). Also, this is equivalent to the difference between the target position of the pin members 9 in the Z direction and the position of the pin members 9 in the Z direction measured by a laser interferometer (not shown).

[0076] The adhesive force can be obtained from the Z deviation in the fourth section (FIG. 8). FIG. 9 is a diagram showing the time transition of the Z deviation. In FIG. 9, D1 indicates the start time of the fourth section (the end time of the third section), and D2 indicates the end time of the fourth section (the start time of the fifth section). In this embodiment, the control unit 100 controls the fine movement stage 8 based on the Z deviation (first condition) acquired in advance. As described above, when the fine movement stage 8 is moved in the -Z direction at a low speed in the fourth section, the substrate 2 and the substrate chuck 3 may be in a state where they are difficult to separate due to the adhesive force. In that case, the fine movement stage 8 cannot move to the target position instructed by the instruction unit 110 (the substrate chuck 3 or the pin member 9 cannot be moved to the target position), and the target position and the actual position in the Z direction are different, and a Z deviation occurs. The processing unit 130 calculates the Z deviation, which is the difference between the target position in the Z direction of the fine movement stage 8, the substrate chuck 3, or the pin member 9, and the actual position in the Z direction measured by a sensor not shown. The Z deviation is shown, for example, by waveform 17 in Fig. 9. Note that waveform 18 shows the reference Z deviation. Reference waveform 18 is a reference waveform that shows the Z deviation when fine movement stage 8 is moved without holding substrate 2.

[0077] In this embodiment, the absolute value of the Z deviation when the acquired Z deviation is the maximum or minimum value is set as the Z deviation representative value. Also, the representative value of the Z deviation that serves as a reference when the fine movement stage 8 is moved without the substrate 2 being placed on the substrate chuck 3 is set as the Z deviation reference value. The maximum or minimum value of the Z deviation is the maximum or minimum value of the Z deviation representative values ​​from when the substrate 2 and the pin member 9 come into contact with each other to when the entire surface of the substrate 2 and the substrate chuck 3 are separated. The processing unit 130 obtains the absolute value of the Z deviation when the Z deviation is maximum or minimum in the acquired Z deviation waveform. This absolute value is set as the Z deviation representative value. The adhesion force is obtained based on the difference between the Z deviation representative value and the Z deviation reference value.

[0078] In this embodiment, parameters (gas supply parameters) related to the supply of gas between the substrate 2 and the substrate chuck 3 are inferred using an inference model (learning model). The calculation unit 703 and the supply unit 704 can function as an output unit that outputs control parameters to be provided to the control unit 100 that controls the exposure apparatus 200 according to the inference model. The information processing device 700 can also include a machine learning unit that generates an inference model by machine learning. The calculation unit 703 can function as the machine learning unit. The machine learning unit performs machine learning of the inference model based on the result of separation of the substrate 2 and the holding surface of the substrate chuck 3, which is performed while the supply of gas is controlled by the control unit 100 based on the control parameters, in the substrate chuck 3. As an inference method, existing methods such as a neural network, a Bayesian network, a support vector machine, a linear regression, and a logistic regression can be used. An example of determining gas supply parameters by machine learning using a neural network as an inference model (learning model) will be described below. By using a neural network, it is possible to determine optimal parameters from the causal relationship between input and output without using a complex calculation formula. Furthermore, by learning using the results of gas supply performed in an exposure apparatus, it is possible to provide optimal parameters for substrate unloading with higher accuracy than simulation.

[0079] FIG. 11 shows an example of the configuration of a neural network 91 created as an inference model. The neural network 91 includes an input layer 910, one or more intermediate layers 911, and an output layer 912. Each value of the input data 915 in the input layer 910 is expressed as A 1 , A 2 , ,A N In response to this input, output data 917 of the output layer 912 is determined via one or more intermediate layers 911. In the example of FIG. 11, there are two intermediate layers 911, and each value in the intermediate data 916 is expressed as M 11 , M 12 , ,M 1M , and M 21 , M 22 , ,M 2PIn addition, each value of the output data 917 in the output layer 912 is represented as O 1 , O 2 , , O O The calculation of each layer of the neural network 91 is performed as follows.

[0080] The calculation method in the process of converting input data into intermediate layer data is as follows. First, a linear transformation is performed on the input data. Here, if the intermediate data is U, the intermediate data U is expressed as follows:

number

number

number

[0081] This intermediate data U is subjected to nonlinear transformation using an activation function to obtain intermediate layer data M. For example, when ReLU is used as the activation function, intermediate layer data M can be obtained by the following formula.

number

[0082] In this way, by performing linear transformation and nonlinear transformation, each data of the neural network can be calculated and found. Here, an example of calculating intermediate data 916 from input data 915 is shown, but calculation from intermediate data 916 to intermediate data 916 can be performed in a similar manner. In calculation from the intermediate layer 911 to the output layer 912, a tanh function can be used as an activation function. Finally, the output layer 912 has an expected value normalized between -1 and 1.

[0083] The calculation of this neural network can be performed by the calculation unit 703 of the information processing device 700. The parameters of the neural network 91 need to be adjusted in advance by some method. As an adjustment method, for example, a network parameter adjustment method using reinforcement learning can be used. The network parameter adjustment method here is called "learning". The information processing method executed by the information processing device 700 will be described below.

[0084] In the learning phase (learning process), an inference model that outputs control parameters for the substrate chuck 3 to be provided to the control unit 100 is learned. In the learning phase, supervised learning is used. Based on the data acquired by the exposure apparatus 200, the parameters of the neural network are adjusted so that the output obtained for a certain input is the same as the output calculated using the neural network 91 and the output obtained by the exposure apparatus.

[0085] The calculation unit 703 performs learning of the neural network 91 according to the flowchart of Fig. 12. In S1701, the calculation unit 703 changes the neural network parameters to their initial values. Specifically, the calculation unit 703 sets the vector W (weighting matrix) and the bias vector B to 0.

[0086] In S1702, the acquisition unit 702 of the information processing device 700 acquires data from the exposure apparatus 200 when the substrate 2 is transferred from the substrate chuck 3 to the pin members 9. The data acquired from the exposure apparatus 200 is referred to as learning data. The learning data is input data A'1 , A' 2 , ,A' N And the training data O' 1 , O' 2 , , , O' O The training data is used for training the neural network 91. The training data is stored in the storage unit 707. In the calculations of the neural network 91, it is desirable that the values ​​of the input layer and the output layer are normalized between −1 and +1. Therefore, in one example, the training data can be a value divided by the maximum value that the device can take.

[0087] In one example, the input data for the training data may include at least one of the following: A' 1 : Estimated adhesion force, A' 2 : Amount of gas supplied in the third section (Figure 8), A' 3 : gas supply time in the third section, A' 4 : Driving speed of the fine movement stage 8 in the Z direction in the fourth section

[0088] In addition, the teacher data of the learning data may include the following. O' 1 : the amount of change in the measurement value of the pressure sensor 20 (i.e., the pressure in the space between the substrate 2 and the substrate chuck 3) in the third and fourth sections, O' 2 : Difference between the typical Z deviation value and the standard Z deviation value in the fourth section

[0089] The input data and the training data are regarded as one set of data. Note that the number of sets of data to be acquired may be one, but multiple sets can also be acquired. The accuracy of learning is improved by acquiring multiple sets.

[0090] Further explanation on input data and output data will be provided.

[0091] An adhesive force acts between the substrate 2 and the substrate chuck 3. The adhesive force can be estimated using at least one of the characteristics of the substrate 2, the characteristics of the substrate chuck 3, and the characteristics of the exposure process. This estimated value, which is information about the adhesive force, is represented as A' 1 is given as:

[0092] A' 2 and A' 3 A' is data on the supply of gas in the third section used to weaken the adhesive force between the substrate 2 and the substrate chuck 3. 2 is data related to the amount of gas supplied. The data related to the amount of gas supplied can be, for example, a gas flow rate or a gas pressure value. Since the pressure and flow rate of a gas are proportional to each other, either can be data related to the amount of gas supplied. A' 3 is data related to gas supply time. A' 4 is data relating to the speed at which the substrate 2 and the substrate chuck 3 are separated.

[0093] The training data O' 1 and O' 2 shows information regarding the result of separation of the substrate 2 from the substrate chuck 3. 1 can be an indicator of whether the amount of gas supplied between the substrate 2 and the substrate chuck 3 is excessive. If the gas supply is excessive, the increased pressure drops suddenly because the substrate 2 is away from the substrate chuck 3, as shown in FIG. 1 The pressure A', which indicates the adhesive force, becomes close to atmospheric pressure. 1 For A', 2 The supply indicated by A' is too low, or 3 If the supply time indicated by A' is too short, this pressure drop will not occur. 1 , A' 2 , A' 3 And O' 1 There is a correlation between.

[0094] O' 2is information on the control deviation of the driving unit and is an index of the adhesive force remaining after the gas supply. The larger the residual adhesive force, the more the substrate chuck 3 may wear out. Data A' regarding the gas supply 2 and A' 3 is small compared to the adhesive force, the residual adhesive force will be large. In this case, A' 4 If the driving speed indicated by is fast, O' 2 The value of A' becomes large. 4 If the driving speed indicated by is slow, O' 2 The value of A' becomes smaller. 1 ,A' 2 , A' 3 , A' 4 And O' 2 There is a correlation between.

[0095] A' 4 The driving speed of the fine movement stage 8 in the Z direction in the fourth section is set as A' because the driving speed has the greatest effect on the wear of the substrate chuck 3 when there is residual adhesion between the substrate 2 and the substrate chuck 3 after the gas supply. Although the degree of influence is small, A' 5 As subsequent input data, other parameters relating to driving, such as acceleration, jerk, etc. may be set.

[0096] O' 1 As the pressure sensor 71, the amount of change in the measurement value (i.e., the pressure between the substrate 2 and the pin members 9) may be used as the pressure sensor 71. When there is an excessive supply of gas, the excess supply can be detected by the substrate 2 being separated from the pin members 9 and the pressure sensor 71 indicating a value close to atmospheric pressure.

[0097] As output data, O' 1 and O' 2 However, the accuracy of the control parameters when the substrate is unloaded decreases, but the calculation time and learning time of the neural network can be shortened.

[0098] In S1702 described above, the information processing device 700 acquires the learning data from the exposure device 200. However, the learning data to be acquired may be stored in advance in the memory unit 707, and the learning data may be acquired from the memory unit 707.

[0099] In S1703, the calculation unit 703 executes reward calculation based on the learning data acquired in S1702. For example, the calculation unit 703 randomly selects one set of data from the data acquired in S1702. The calculation unit 703 inputs the learning data acquired in S1702 to the input data 915 of the neural network 91. Specifically, A' 1 A 1 To, A' 2 A 2 To, A' 3 A 3 To, A' 4 A 4 The calculation unit 703 performs the calculation of the neural network and outputs the output data O 1 and O 2 The reward is calculated by the output O 1 , O 2 and O' of the training data acquired from the exposure apparatus 200. 1 , O' 2 The smaller the difference, the higher the reward.

[0100] In S1704, the calculation unit 703 changes the neural network 91 (i.e., updates the parameters of the neural network 91) so as to increase the reward. Specifically, the vector W (weighting matrix) and the bias vector B are updated.

[0101] By repeating steps S1703 and S1704, the neural network 91 is fed with input data A 1 ~A 4 The calculated O 1 and O 2 The result is the training data O' obtained from the exposure apparatus 200. 1 , O' 2can be approached.

[0102] In S1705, the calculation unit 703 determines whether the number of pieces of learning data used in the learning has reached the designated number. The designated number may be input via the input unit 706 at the start of the learning, or may be set in advance in the processing unit 701. In one example, the designated number is the same as the number of pieces of learning data acquired from the exposure apparatus 200. In another example, the designated number is a number greater than the number of pieces of learning data acquired from the exposure apparatus 200. In yet another example, the designated number is a number less than the number of pieces of learning data acquired from the exposure apparatus 200. If the designated number is set to a number greater than the number of pieces of learning data acquired from the exposure apparatus 200, in S1703, the learning is repeated using the same learning data. In general, the more learning data there is, the higher the accuracy of the learning result becomes. In S1705, if the number of pieces of learning data used in the learning has not yet reached the designated number, the process returns to S1703 and the learning is repeated. In S1705, when the number of pieces of learning data used in learning reaches the designated number, the calculation unit 703 stores the parameters of the neural network 91 in the storage unit 707, and the learning phase ends.

[0103] Also, in S1702, evaluation data may be obtained separately from the learning data, and after the number of pieces of learning data used for learning reaches a designated number in S1705, an evaluation step may be performed. In the evaluation step, the calculation unit 703 judges whether the learning is good or bad. In the evaluation step, the calculation unit 703 calculates a reward from the output obtained by inputting the evaluation data into the neural network 91 and the evaluation output data. If the obtained reward is higher than a predetermined value, it is judged that the learning is successful. If the obtained reward is equal to or lower than a predetermined value, it is judged that the learning should be continued.

[0104] Next, the estimation (inference) phase using the trained neural network will be described. In the estimation phase, the information processing device 700 calculates control parameters for unloading the substrate using the trained neural network based on input data from the exposure apparatus 200, and provides the calculated parameters to the exposure apparatus 200.

[0105] The estimation phase is outlined below. The information processing device 700 receives input data A 1 is obtained from the exposure tool 200. The calculation unit 703 2 , A 3 , A 4 is set to a tentative value, calculations are performed on the neural network 91, and the output data of the neural network 91, O 1 , O 2 The calculation unit 703 obtains the output data O 1 and O 2 A, which gives the most favorable result. 2 , A 3 , A 4 Explore combinations of.

[0106] The flow in the estimation phase will be described with reference to Fig. 13. In the estimation phase, the control parameters for unloading the substrate are optimized using the neural network trained in the learning phase.

[0107] In S1801, the calculation unit 703 initializes the control parameters. 2 , A 3 , A 4 is initialized to 0.

[0108] In S1802, the acquisition unit 702 acquires A of the neural network 91. 1 The parameters related to the adhesion force to be put into are obtained from the exposure tool 200. 1 The information may be information that has been learned in the past, or may be information that has not been learned yet.

[0109] In S1803, the calculation unit 703 updates the control parameters. 2 , A 3 , A 4 Change at least one of the following:

[0110] In S1804, the calculation unit 703 1 ~A 4 Based on the input data, the neural network 91 is calculated, and O 1 and O 2 Calculate.

[0111] By repeating S1803 and S1804, A 2 , A 3 , A 4 By changing the parameter value, the output data 1 and O 2 At this time, the output data O 1 and O 2 The parameter values ​​are searched for such that O approaches 0. 1 indicates the amount of decrease in the measurement value of the pressure sensor 20 (i.e., the pressure in the space between the substrate 2 and the substrate chuck 3) in the third and fourth sections. The amount of decrease being 0 indicates that the substrate 2 and the substrate chuck 3 are not sliding sideways. 2 is the difference between the Z deviation representative value and the Z deviation reference value in the fourth section. The closer this is to 0, the smaller the adhesion force is, and the smaller the wear of the substrate chuck 3 is.

[0112] As mentioned above, O 1 and O 2By obtaining this, it is possible to prevent the substrate 2 from sliding sideways relative to the substrate chuck 3, and also to prevent the substrate 2 from separating from the substrate chuck 3, which would cause wear on the substrate chuck 3. This has the effect of reducing the adhesion force to the optimum point shown in FIG. 5(b). By using a neural network, it is possible to provide optimal parameters for unloading the substrate, even for unknown adhesion forces that have not been calculated by the simulator. The provided parameters are obtained by learning actual data, so the parameters are highly reliable.

[0113] As a result of the search, O 1 and O 2 A is equivalent to 2 , A 3 , A 4 If multiple parameter value sets are found, a set of parameter values ​​that maximizes the productivity of exposure apparatus 200 may be selected from among them. 3 and A 4 contributes to productivity. Therefore, A with short supply time 3 Or select A with a faster drive speed. 4 Alternatively, A may be selected. 3 and A 4 It is also possible to combine both times and select the combination that results in the shortest substrate unloading time.

[0114] In S1805, the calculation unit 703 2 , A 3 , A 4 It is determined whether the parameter search for has been completed. If the parameter search has not been completed, the process returns to S1803 and the search is repeated. If the parameter search has been completed, the process proceeds to S1806. In S1806, the supply unit 704 supplies the determined parameter A to the exposure apparatus 200. 2 , A 3 , A 4 is supplied as a control parameter.

[0115] <Second embodiment> In the above-described first embodiment, an estimated value of the adhesive force between the substrate 2 and the substrate chuck 3 is provided as an input to the neural network 91. In the second embodiment, instead of the estimated value of the adhesive force, the characteristics of the substrate 2, the characteristics of the substrate chuck 3, and the characteristics of the exposure process, which are factors of the adhesive force, are provided as input. In the second embodiment, the input data for the learning data may include at least any of the following: A 1 : Warpage of substrate 2, A 2 : Roughness of the back surface of the substrate 2 (the surface facing the holding surface of the substrate chuck), A 3 : thickness of substrate 2, A 4 : the area or diameter of the tip surface of the chuck pin (holding member) of the substrate chuck 3 that comes into contact with the substrate, A 5 : Material of the substrate chuck 3, A 6 : exposure time, A 7 :Exposure A 8 : Amount of gas supplied in the third section, A 9 : gas supply time in the third section, A 10 : Driving speed of the fine movement stage 8 in the Z direction in the fourth section

[0116] In addition, the teacher data of the learning data may include the following. O 1 : the amount of change in the measurement value of the pressure sensor 20 (i.e., the pressure in the space between the substrate 2 and the substrate chuck 3) in the third and fourth sections, O 2 : Difference between the typical Z deviation value and the standard Z deviation value in the fourth section

[0117] The above input data and training data are considered as one set of data.

[0118] As in the first embodiment, the neural network 91 is trained in the learning phase, and the control parameters during substrate unloading are estimated using the trained neural network 91 in the estimation phase.

[0119] By inputting the characteristics of the substrate 2, the characteristics of the substrate chuck 3, and the characteristics of the exposure process, which are factors in the adhesion force, into the neural network 91 instead of the estimated value of the adhesion force, the adhesion force estimation process can be included in the neural network. This provides the effect of further optimizing the control parameters when the substrate is unloaded. Also, even for combinations with different characteristics that have not been processed in the exposure apparatus 200, the effect of further optimizing the control parameters when the substrate is unloaded can be provided.

[0120] In the first and second embodiments, the gas flow path for supplying gas is described as one system, but the gas flow path may be two systems. By providing two gas flow paths, the supply of gas can be controlled more precisely.

[0121] Furthermore, the processing unit 701 may cause the display unit 705 to display a list of input parameters and output results, and allow the user to select parameters to be supplied to the exposure apparatus 200 via the input unit 706. At this time, the processing unit 701 may cause the display unit 705 to highlight recommended settings.

[0122] The information processing device 700 may also be communicatively connected to a plurality of exposure apparatuses, acquire learning data from each of the plurality of exposure apparatuses, and supply each of the plurality of exposure apparatuses with optimal parameters for substrate unloading. By collecting data from a plurality of exposure apparatuses, learning can be performed based on a large amount of data. The characteristics of the substrate chuck differ from apparatus to apparatus, and information on the substrate chuck under various conditions can be included in the learning data. The substrate chuck may be replaced after it wears out. Even if a new substrate chuck is mounted on the exposure apparatus, optimal control parameters for substrate unloading can be provided to the exposure apparatus because there is learning data from another exposure apparatus.

[0123] In a mass production factory for semiconductor devices, a specific number of substrates (e.g., 25 substrates) that have undergone the same manufacturing process are generally managed as one lot. Therefore, an exposure processing job is input on a lot-by-lot basis. The information processing device 700 may supply optimal parameters for substrate unloading on a lot-by-lot basis. This is because it is expected that the characteristics of the substrate 2, the characteristics of the substrate chuck 3, and the characteristics of the exposure process, which are used to estimate the adhesion force, are substantially the same within a lot.

[0124] The information processing device 700 may be connected to a larger capacity external storage unit (not shown) to store the learning data acquired from the exposure apparatus 200. This allows more information to be stored in the external storage unit, and the information processing device 700 can perform learning using more information. This improves the performance of the generated (updated) neural network 91.

[0125] <Embodiment of the article manufacturing method> The article manufacturing method according to the embodiment of the present invention is suitable for manufacturing articles such as microdevices such as semiconductor devices and elements having a fine structure. The article manufacturing method according to the present embodiment may include a forming step of forming a pattern of an original on a substrate using the above-mentioned substrate processing apparatus, and a processing step of processing the substrate on which the pattern has been formed in the forming step. Furthermore, such an article manufacturing method may include other well-known steps (oxidation, film formation, deposition, doping, planarization, etching, resist stripping, dicing, bonding, packaging, etc.). The article manufacturing method according to the present embodiment is advantageous in at least one of the performance, quality, productivity, and production cost of the article compared to conventional methods.

[0126] The disclosure of the present specification includes at least the following techniques. (Item 1) An information processing device that provides a control parameter of a holding unit that holds an object to a control unit that controls the holding unit, the holding unit has a gas flow path arranged so that gas from a gas supply source is supplied between the object and the holding unit through an opening formed in a surface facing the object, the control unit is configured to control the supply of gas through the gas flow path and the opening based on the control parameter provided by the information processing device when releasing the holding unit from the holding unit; The information processing device includes: an output unit that outputs control parameters to be provided to the control unit in accordance with an inference model; a machine learning unit that performs machine learning of the inference model based on a result of separation of the object and a holding surface that holds the object, the separation being performed in the holding unit while the supply of gas is controlled by the control unit based on the control parameters; and 13. An information processing device comprising: (Item 2) The information processing device described in item 1, characterized in that the machine learning unit performs machine learning using information regarding the adhesive force of the object on the holding surface and information regarding the supply of gas by the control unit as input data and information regarding the result of the separation as training data. (Item 3) 3. The information processing device according to item 2, wherein the information relating to the supply of gas includes information on the amount and time of supply of gas. (Item 4) 2. The information processing device according to item 1, wherein the machine learning unit performs machine learning using the characteristics of the object and the characteristics of the holding unit as input data and information on the results of the separation as training data. (Item 5) 5. The information processing device according to item 4, wherein the characteristics of the object include information on at least one of the amount of warping of the object, the roughness of the surface of the object opposite the holding surface, and the thickness of the object. (Item 6) the holding portion has a support member that is configured to be retractable from the holding surface and supports the object by protruding above the holding surface, 5. The information processing device according to item 4, characterized in that the characteristics of the holding portion include at least one of information on the area or diameter of the tip surface of the support member that contacts the object and information on the material of the holding portion. (Item 7) the holding part is configured to hold the substrate as the object in an exposure apparatus that exposes a substrate, the input data further includes characteristics of an exposure process in the exposure apparatus; 7. The information processing device according to any one of items 4 to 6, (Item 8) 8. The information processing apparatus according to item 7, wherein the characteristics of the exposure process include information on at least one of an exposure time and an exposure amount for the substrate. (Item 9) the holding portion has a support member that is configured to be retractable from the holding surface and supports the object by protruding above the holding surface, 3. The information processing device according to item 2, wherein the information relating to the result of the separation as the teaching data includes information on the amount of change in pressure between the object and the support member. (Item 10) 3. The information processing device according to item 2, characterized in that the input data further includes a drive speed of a drive unit that performs the separation operation, and the information regarding the result of the separation as the teacher data includes information on a control deviation of the drive unit. (Item 11) 11. The information processing device according to any one of items 1 to 10, characterized in that the inference model is created by a neural network. (Item 12) A program for causing a computer to function as each part of the information processing device according to any one of items 1 to 11. (Item 13) 1. An information processing method executed by an information processing device that provides a control parameter of a holding unit that holds an object to a control unit that controls the holding unit, the holding unit has a gas flow path arranged so that gas from a gas supply source is supplied between the object and the holding unit through an opening formed in a surface facing the object, the control unit is configured to control the supply of gas through the gas flow path and the opening based on the control parameter provided by the information processing device when releasing the holding unit from the holding unit; The information processing method includes: A learning step of learning an inference model that outputs a control parameter to be provided to the control unit, The learning step includes: and performing machine learning using information on the adhesive force of the object to the holding surface of the holding unit that holds the object and information on the supply of gas by the control unit as input data, and information on the result of separation of the object from the holding surface executed in the holding unit while the supply of gas is controlled by the control unit based on the control parameters as training data. 23. An information processing method comprising:

[0127] The invention is not limited to the above-described embodiments, and various modifications and variations are possible without departing from the spirit and scope of the invention. Accordingly, the following claims are appended to apprise the public of the scope of the invention. [Explanation of symbols]

[0128] 1: substrate stage, 2: substrate, 3: substrate chuck, 5: vacuum pump, 6: positive pressure tank, 8: fine movement stage, 10: gas flow path, 13: coarse movement stage, 100: control unit, 700: information processing device, 701: processing unit

Claims

1. An information processing device that provides a control parameter of a holding unit that holds an object to a control unit that controls the holding unit, the holding unit has a gas flow path arranged so that gas from a gas supply source is supplied between the object and the holding unit through an opening formed in a surface facing the object, the control unit is configured to control the supply of gas through the gas flow path and the opening based on the control parameter provided by the information processing device when releasing the holding unit from the holding unit; The information processing device includes: an output unit that outputs control parameters to be provided to the control unit in accordance with an inference model; a machine learning unit that performs machine learning of the inference model based on a result of separation of the object and a holding surface that holds the object, the separation being performed in the holding unit while the supply of gas is controlled by the control unit based on the control parameters; and 13. An information processing device comprising:

2. The information processing device according to claim 1, characterized in that the machine learning unit performs machine learning using information regarding the adhesive force of the object relative to the holding surface and information regarding the supply of gas by the control unit as input data, and information regarding the results of the separation as training data.

3. 3. The information processing apparatus according to claim 2, wherein the information on the supply of gas includes information on a supply amount and a supply time of the gas.

4. The information processing apparatus according to claim 1 , wherein the machine learning unit performs machine learning using the characteristics of the object and the characteristics of the holder as input data and information on the results of the separation as training data.

5. 5. The information processing apparatus according to claim 4, wherein the characteristics of the object include information on at least one of an amount of warping of the object, a roughness of a surface of the object facing the holding surface, and a thickness of the object.

6. the holding portion has a support member that is configured to be retractable from the holding surface and supports the object by protruding above the holding surface, 5 . The information processing device according to claim 4 , wherein the characteristics of the holding portion include at least one of information on an area or diameter of a tip surface of the support member that contacts the object, and information on a material of the holding portion.

7. the holding part is configured to hold the substrate as the object in an exposure apparatus that exposes a substrate, the input data further includes characteristics of an exposure process in the exposure apparatus; 5. The information processing apparatus according to claim 4.

8. 8. The information processing apparatus according to claim 7, wherein the characteristics of the exposure process include information on at least one of an exposure time and an exposure amount for the substrate.

9. the holding portion has a support member that is configured to be retractable from the holding surface and supports the object by protruding above the holding surface, The information processing apparatus according to claim 2 , wherein the information relating to the result of the separation as the teaching data includes information on an amount of change in pressure between the object and the support member.

10. 3. The information processing device according to claim 2, wherein the input data further includes a drive speed of a drive unit that performs the separation operation, and the information regarding the result of the separation as the teaching data includes information on a control deviation of the drive unit.

11. 2. The information processing apparatus according to claim 1, wherein the inference model is created by a neural network.

12. A program for causing a computer to function as each unit of the information processing device according to any one of claims 1 to 11.

13. 1. An information processing method executed by an information processing device that provides a control parameter of a holding unit that holds an object to a control unit that controls the holding unit, the holding unit has a gas flow path arranged so that gas from a gas supply source is supplied between the object and the holding unit through an opening formed in a surface facing the object, the control unit is configured to control the supply of gas through the gas flow path and the opening based on the control parameter provided by the information processing device when releasing the holding unit from the holding unit; The information processing method includes: A learning step of learning an inference model that outputs a control parameter to be provided to the control unit, The learning step includes: and performing machine learning using information on the adhesive force of the object to the holding surface of the holding unit that holds the object and information on the supply of gas by the control unit as input data, and information on the result of separation of the object from the holding surface executed in the holding unit while the supply of gas is controlled by the control unit based on the control parameters as training data.

23. An information processing method comprising:

Images