JP7536626B2 - CONTROL DEVICE, ADJUSTMENT METHOD, LITHOGRAPHIC APPARATUS, AND METHOD FOR MANUFACTURING ARTICLE - Patent application - Google Patents

Description

The present invention relates to a control device, an adjustment method, a lithography apparatus, and a method for manufacturing an article.

In the photolithography process used to manufacture devices such as semiconductor devices and flat panel displays (FPDs), exposure apparatuses are used to transfer a mask pattern onto a substrate. For example, exposure apparatuses are required to perform highly accurate positional and synchronous control of the mask stage that holds the mask and the substrate stage that holds the substrate in order to align the mask and substrate.

The demands for precision required for position control and synchronization control of stages and the like as described above are becoming stricter as devices become more precise, and conventional feedback control alone may not be able to achieve the required precision. Therefore, efforts are being made to configure neural network controllers in parallel in addition to conventional controllers (Patent Document 1). Also, a method has been devised that switches between neural network controllers depending on the state of the controlled object and performs compensation tailored to the controlled object (Patent Document 2).

Special Publication No. 7-503563 Japanese Patent Application Publication No. 7-277286

However, while accuracy can be improved by configuring multiple neural network controllers, the control calculation time increases. In addition, the parameters of neural network controllers are adjusted by machine learning, and it takes a lot of time to train the parameters of multiple neural networks. Furthermore, if there is a change in the state of the controlled object or a change in the external disturbance environment, the predetermined neural network parameters will no longer be optimal, and a lot of time will be required to readjust the parameters.

The present invention aims to provide a control device that uses a neural network and is advantageous for adjusting appropriate control characteristics in a short period of time.

In order to achieve the above object, a control device according to one aspect of the present invention is a control device that generates a control signal for controlling a controlled object, and is characterized by comprising: a first compensator that generates a first signal based on a control deviation of the controlled object; a corrector that corrects the control deviation using one of a plurality of adjustment units that generate a correction signal by correcting the control deviation according to an arithmetic expression having an adjustable coefficient; a second compensator that generates a second signal by a neural network based on the correction signal; and a calculator that generates the control signal based on the first signal and the second signal.

The present invention provides a control device that uses a neural network and is advantageous for adjusting appropriate control characteristics in a short period of time.

FIG. 1 is a diagram illustrating an example of a system configuration in a first embodiment. FIG. 1 is a diagram illustrating an example of a system configuration in a first embodiment. FIG. 2 is a diagram illustrating an example of the configuration of a controller in the system of the first embodiment. FIG. 23 is a diagram illustrating a configuration example of a controller in a sixth embodiment. FIG. 23 is a diagram illustrating a configuration example of a controller in a seventh embodiment. FIG. 23 is a diagram illustrating an example of the configuration of a controller in the eighth embodiment. FIG. 1 is a diagram illustrating an example of a system configuration in a first embodiment. 5 is a flowchart showing an operation example when the system of the first embodiment is applied to a production device. 11A and 11B are diagrams illustrating an example of measurement results of disturbance suppression characteristics. FIG. 13 is a diagram illustrating an example of the configuration of a stage control device according to the second embodiment. FIG. 13 is a diagram illustrating an example of the configuration of a control board in the system of the second embodiment. 13 is a flowchart showing adjustment of a corrector. FIG. 11 is a diagram illustrating an example of a position control deviation. FIG. 11 is a diagram illustrating an example of a result of frequency analysis. FIG. 1 is a diagram illustrating an example of the configuration of an exposure apparatus.

Below, a preferred embodiment of the present invention will be described in detail with reference to the attached drawings. Note that in each drawing, the same reference numbers are used for the same components, and duplicated explanations will be omitted.

First Embodiment
1 shows the configuration of a system SS in this embodiment. The system SS is applied to, for example, a manufacturing apparatus for manufacturing an article. The manufacturing apparatus includes, for example, a processing apparatus for processing an article or a member constituting a part of the article. The processing apparatus can be, for example, any of a lithography apparatus that transfers a pattern to a material or member, a film forming apparatus that forms a film on a material or member, an apparatus that etches a material or member, and a heating apparatus that heats a material or member.

The system SS includes, for example, a sequence unit 101, a control device 100, and a controlled object 103. The control device 100 includes a controller 102. The control device 100 or the controller 102 generates a control signal MV for controlling the controlled object 103. When the system SS is applied to a production system, a production sequence is provided to the sequence unit 101. The production sequence specifies a procedure for production. The sequence unit 101 generates a target value R for controlling the controlled object 103 based on the production sequence, and provides the target value R to the control device 100 or the controller 102.

The control device 100 or the controller 102 performs feedback control of the controlled object 103. Specifically, the control device 100 controls the controlled object 103 so that the control amount CV of the controlled object 103 follows the target value R based on a control deviation, which is the difference between the target value R provided by the sequence unit 101 and the control amount CV provided by the controlled object 103. The controlled object 103 may have a sensor that detects the control amount CV, and the control amount CV detected by the sensor may be provided to the controller 102. The target value R, the control signal MV, and the control amount CV may be time-series data whose values change over time.

As illustrated in FIG. 2, the system SS may incorporate a learning unit 201. The learning unit 201 may be configured as a part of the control device 100, or may be configured as an external device of the control device 100. When the learning unit 201 is configured as an external device of the control device 100, the learning unit 201 may be disconnected from the control device 100 after learning is completed. The learning unit 201 is configured to send a learning sequence prepared in advance to the sequence unit 101. The sequence unit 101 generates a target value R according to the learning sequence and provides it to the controller 102.

The controller 102 generates a control signal MV based on a control deviation, which is the difference between a target value R generated and provided from the sequence unit 101 according to a learning sequence and a control amount CV provided from the controlled object 103. Here, the controller 102 has a neural network and generates a control signal MV using the neural network. The control signal MV generated by the controller 102 is provided to the controlled object 103, and the controlled object 103 operates according to this control signal MV. The control amount CV as a result of this operation is provided to the controller 102. The controller 102 provides the learning unit 201 with an operation history indicating the history of the operation of the controller 102 based on the target value R. The learning unit 201 determines parameter values of the neural network of the controller 102 based on the operation history, and sets the parameter values in the neural network. The parameter values are determined by machine learning, such as reinforcement learning, for example.

Figure 3 is a diagram showing one example of the configuration of the controller 102. The controller 102 includes a first compensator 301 that generates a first signal S1 based on the control deviation E, and a corrector 303 that generates a correction signal CS by calculating the control deviation E according to an arithmetic expression with an adjustable coefficient. The controller 102 also includes a second compensator 302 that generates a second signal S2 by a neural network based on the correction signal CS, and a calculator 306 that generates a control signal MV based on the first signal S1 and the second signal S2.

The corrector 303 has a plurality of adjustment units including a first adjustment unit 303a and a second adjustment unit 303b, and can select an adjustment unit to be used (a connected adjustment unit) according to a control state. The control signal MV is the sum of the first signal S1 and the second signal S2, and the calculator 306 can be configured with an adder. The control signal MV is a signal obtained by correcting the first signal S1 based on the second signal S2. The controller 102 includes a subtractor 305 that generates a control deviation E that is a difference between the target value R and the control amount CV. The control amount CV is acquired by being measured by a sensor (not shown) or the like that is provided in the controlled object 103. In addition, the difference between the control amount and the target value R, which is a result of controlling the controlled object 103 based on the control signal MV, is smaller than the difference between the control amount and the target value R, which is a result of controlling the controlled object 103 based on the first signal S1.

The controller 102 further includes an operation history recording unit 304. The learning unit 201 in FIG. 2 is configured to perform learning to determine parameter values of the neural network of the second compensator 302 in FIG. 3. For the learning by the learning unit 201, the operation history recording unit 304 records the operation history required for the learning by the learning unit 201 and provides the recorded operation history to the learning unit 201. The operation history is, for example, the correction signal CS which is input data to the second compensator 302 and the second signal S2 which is output data of the second compensator 302, but may be the control deviation and the second signal S2 which is output data of the second compensator 302, or other data. The first adjustment unit 303a and the second adjustment unit 303b can perform learning with any parameter as an initial value.

In the following Examples 1 to 5, configuration examples of the corrector 303 are described. In Examples 1 to 5, examples of arithmetic expressions used by the corrector 303 to generate the correction signal CS based on the control deviation E are shown. The arithmetic expression can be, for example, a monomial or a polynomial.

Example 1
In the first embodiment, the first adjustment unit 303a and the second adjustment unit 303b have control characteristics expressed by the following formula (1): Here, the input (E) to the corrector 303 is x, the output (CS) of the corrector 303 is y, and an arbitrary coefficient (constant) is Kp.

Example 2
In the second embodiment, the first adjustment unit 303a and the second adjustment unit 303b have control characteristics expressed by the following formula (2). Here, the input (E) to the corrector 303 is x, the output (CS) of the corrector 303 is y, the time is t, and an arbitrary coefficient (constant) is Ki. Note that integration may be performed multiple times. The integration may be a definite integral over a certain time interval, or an indefinite integral.

Example 3
In the third embodiment, the first adjustment unit 303a and the second adjustment unit 303b have control characteristics expressed by the following formula (3). Here, the input (E) to the corrector 303 is x, the output (CS) of the corrector 303 is y, the time is t, and an arbitrary coefficient (constant) is Kd. Differentiation may be performed multiple times.

Example 4
In the fourth embodiment, the first adjustment unit 303a and the second adjustment unit 303b have control characteristics expressed by the following formula (4). Here, the input (E) to the corrector 303 is x, the output (CS) of the corrector 303 is y, and arbitrary coefficients (constants) are Kp, Ki, and Kd. Note that integration and differentiation may be performed multiple times.

Example 5
In the fifth embodiment, the first adjustment unit 303a and the second adjustment unit 303b have control characteristics expressed by the following arithmetic expression (5): Here, the input (E) to the corrector 303 is x, the output (CS) of the corrector 303 is y, the integral order of the multiple integral is n, the differential order is m, an arbitrary coefficient (constant) is Kp, an arbitrary coefficient (constant) for n-fold integral is Ki_n, and an arbitrary constant for m-th order differential is Kd_m.

Examples 1 to 5 can be understood as examples in which the equation used by the corrector 303 to generate the correction signal CS includes at least one of a term proportional to the control deviation E, an integral term, and a differential term.

The coefficients (constants) Kp, Ki, Kd, Ki_n, and Kd_m of the arithmetic expressions given in Examples 1 to 5 are examples of adjustable parameters of the corrector 303. The first adjustment unit 303a and the second adjustment unit 303b determine optimal parameters using any of Examples 1 to 5 in accordance with changes in the control state that are expected in advance. The control state is, for example, switching of synchronous control, switching of controllers, switching of operation patterns, and changes in the environment and disturbances such as temperature, noise, and floor vibration. By selecting the first adjustment unit 303a and the second adjustment unit 303b according to the control state, optimal control characteristics can be obtained. The adjustment time required by configuring multiple correctors is shorter than the adjustment time required by configuring multiple neural networks, so it is advantageous in terms of time reduction.

Furthermore, if the state of the controlled object 103 or the disturbance environment changes while the system SS is operating, the change can be accommodated by adjusting the values (parameter values) of the arithmetic equations (coefficients) exemplified as Examples 1 to 5. The time required to adjust the values of the arithmetic equations (coefficients) of the corrector 303 is shorter than the time required to re-learn the neural network. Therefore, control accuracy can be maintained without reducing the productivity of the system SS. In other words, by introducing the corrector 303, it is possible to improve tolerance to changes in the state of the controlled object 103 and changes in the disturbance environment.

Example 6
In the sixth to eighth embodiments, the relationship between the change in the control state and the switching of the adjustment unit used in the corrector 303 will be described.

Figure 4 is a diagram showing an example of the configuration of the controller 102 in the sixth embodiment. In the sixth embodiment, as shown in Figure 4, it is possible to switch between controlling each of a plurality of control objects including the control object 103a and the control object 103b individually or synchronously. The control state in the sixth embodiment is a state determined by whether the plurality of control objects are controlled individually or synchronously, and the appropriate adjustment unit is switched depending on whether the plurality of control objects are synchronously controlled or not. In the sixth embodiment, the compensator 303 is switched depending on the state of the synchronous control switching unit 402, and it is possible to select whether to use the first adjustment unit 303a or the second adjustment unit 303b. In the sixth embodiment, the axis on which the control object 103a is controlled is the master axis, and the axis on which the control object 103b is controlled is the slave axis, and the slave axis follows the master axis, which is called a master-slave method of synchronous control.

The controller 102 acquires the control quantities CVa and CVb of the control objects 103a and 103b, respectively, measured by a sensor (not shown) provided in the control object 103, and calculates the difference between the control quantities CVa and CVb and the target values Ra and Rb, respectively, as the control deviations Ea and Eb.

The control deviation Ea is input to the compensator 301a. The input to the compensator 301b and the corrector 303 provided in the front stage of the neural network 302 configured in parallel with the compensator 301b can be switched depending on whether or not the controlled object 103a and the controlled object 103b are synchronously controlled. The input to the corrector 303 can select the control deviation Eb switched by a synchronous control switching unit 402 that switches the synchronous control between the controlled object 103a and the controlled object 103b, or the synchronous deviation Ec which is the difference between the control deviation Eb and the control deviation Ea. The output of the corrector 303 can select whether to use the first adjustment unit 303a or the second adjustment unit 303b depending on the state of the synchronous control switching unit 402.

As a specific example of the sixth embodiment, when applied to an exposure apparatus, different adjustment units may be selected when the plate stage and the mask stage are synchronized and when other operations are performed. At this time, the first adjustment unit 303a and the second adjustment unit 303b have their parameters optimized when the plate stage and the mask stage are synchronized and when other operations are performed. The outputs of the first adjustment unit 303a and the second adjustment unit 303b selected according to the state of the synchronization control switching unit 402 are input to the neural network 302 (second compensator). The output of the compensator 301a is set as a control signal MVa. The output of the compensator 301b and the output of the neural network 302 are added together to obtain a control signal MVb. The controller 102 outputs the control signals MVa and MVb to the control targets 103a and 103b, respectively.

The first adjustment unit 303a and the second adjustment unit 303b determine optimal parameters using any of the first to fifth embodiments according to the state of the synchronization control switching unit 402. By selecting the first adjustment unit 303a and the second adjustment unit 303b according to the state of the synchronization control switching unit 402, optimal control characteristics can be obtained.

The increase in adjustment time due to the compensator 303 being configured to select the optimal adjustment unit from multiple adjustment units is shorter than the increase in adjustment time due to the configuration of multiple neural networks. Furthermore, if the state of the controlled object 103 or the disturbance environment changes during operation using any of the first to fifth embodiments, the change can be accommodated by adjusting the parameters of the first to fifth embodiments. The time required to adjust the first adjustment unit 303a and the second adjustment unit 303b is shorter than the time required to re-learn the neural network. In the sixth embodiment, even when multiple compensations are made according to the state of the controlled object or the disturbance environment, the increase in calculation time and learning time can be suppressed, and the appropriate control characteristics can be adjusted in a short time even if the state of the controlled object or the disturbance environment changes.

(Example 7)
5 is a diagram showing a configuration example of the controller 102 in the seventh embodiment. In the seventh embodiment, the compensator 301 is switched depending on the state and operation of the controlled object 103, and it is possible to select whether to use the compensator 301a or the compensator 301b. The control state in the seventh embodiment is a state determined by which of a plurality of compensators is to be used, and the adjustment unit is switched depending on whether to use the compensator 301a or the compensator 301b. In the seventh embodiment, the corrector 303 is switched depending on the state of the compensator 301, and it is possible to select whether to use the first adjustment unit 303a or the second adjustment unit 303b.

As a specific example of Example 7, when applied to an exposure apparatus, for example, it may be applied when gain switching occurs, such as using compensator 301a during the exposure operation of the plate stage and compensator 301b during the plate transport operation. That is, an adjustment unit to be used for correcting the control deviation E may be selected from multiple adjustment units based on whether or not gain switching of the control target 103 has occurred. At this time, the first adjustment unit 303a and the second adjustment unit 303b determine optimal parameters for compensator 301a and compensator 301b using any of Examples 1 to 5. By selecting the first adjustment unit 303a and the second adjustment unit 303b according to the state of compensator 301, optimal control characteristics can be obtained.

The increase in adjustment time due to the compensator 303 being configured to select the optimal adjustment unit from multiple adjustment units is shorter than the increase in adjustment time due to the configuration of multiple neural networks. Furthermore, if the state of the controlled object 103 or the disturbance environment changes during operation using any of the first to fifth embodiments, the change can be accommodated by adjusting the parameters of the first to fifth embodiments. The time required to adjust the first adjustment unit 303a and the second adjustment unit 303b is shorter than the time required to re-learn the neural network. In the seventh embodiment, even when multiple compensations are made according to the state of the controlled object or the disturbance environment, the increase in calculation time and learning time can be suppressed, and the appropriate control characteristics can be adjusted in a short time even if the state of the controlled object or the disturbance environment changes.

(Example 8)
6 is a diagram showing a configuration example of the controller 102 in Example 8. The control state in Example 8 is a state determined by whether or not the operation pattern 403 of the controlled object is changed. A specific example of the operation pattern 403 will be described later. In Example 8, the adjustment unit is configured to be switched depending on the state of the operation pattern 403, and it is possible to select whether to use the first adjustment unit 303a or the second adjustment unit 303b.

As a specific example of Example 8 , when applied to a stage device used in an exposure apparatus or the like, it may be applied by switching between an acceleration section during driving of the stage and other operation patterns. At this time, the first adjustment unit 303a and the second adjustment unit 303b determine optimal parameters using any of Examples 1 to 5 according to the state of the operation pattern 403 of the control target 103. By selecting the first adjustment unit 303a and the second adjustment unit 303b according to the state of the operation pattern 403, optimal control characteristics can be obtained.

The increase in adjustment time due to the compensator 303 being configured to select the optimal adjustment unit from multiple adjustment units is shorter than the increase in adjustment time due to the configuration of multiple neural networks. Furthermore, if the state of the controlled object 103 or the disturbance environment changes during operation using any of the first to fifth embodiments, the change can be accommodated by adjusting the parameters of the first to fifth embodiments. The time required to adjust the first adjustment unit 303a and the second adjustment unit 303b is shorter than the time required to re-learn the neural network. In the eighth embodiment, even when multiple compensations are made according to the state of the controlled object or the disturbance environment, the increase in calculation time and learning time can be suppressed, and the appropriate control characteristics can be adjusted in a short time even if the state of the controlled object or the disturbance environment changes.

As illustrated in FIG. 7, the control device 100 may include a setting unit 202 that selects whether to use the first adjustment unit 303a or the second adjustment unit. The setting unit 202 may also have a role of setting the parameter value of the corrector 303.

The setting unit 202 may execute an adjustment process for switching the adjustment unit or adjusting the parameter value, and may determine and set the adjustment unit switching or parameter value by this adjustment process, or may set the adjustment unit switching or parameter value based on an instruction from a user. In the former case, the setting unit 202 may send a confirmation sequence for confirming the operation of the controller 102 to the sequence unit 101, and may cause the sequence unit 101 to generate a target value R based on this confirmation sequence. The setting unit 202 may then obtain an operation history (e.g., control deviation) from the controller 102 that operates based on the target value R, and may determine the necessity of switching the corrector 303 and the parameter value based on the operation history. The setting unit 202 having such a function may be understood as an adjustment unit that switches the corrector 303 and adjusts the parameter value.

The setting unit 202 may acquire an operation history (e.g., a control deviation) from the controller 102 during production when the sequence unit 101 generates the target value R based on the production sequence, and may determine whether or not to adjust the parameter value of the corrector 303 based on the operation history. Alternatively, a judgment unit may be provided separately from the setting unit 202, which judges whether or not to adjust the parameter value of the corrector 303 by the setting unit 202 during production when the sequence unit 101 generates the target value R based on the production sequence.

Next, an example of production performed by the system of this embodiment will be described. Figure 8 shows an example of the operation of the system SS of this embodiment when it is applied to a production device. In step S501, the sequence unit 101 generates a target value R based on a given production sequence and provides it to the control device 100 or the controller 102. The control device 100 or the controller 102 controls the controlled object 103 based on the target value R.

In step S502, the setting unit 202 acquires the operation history (e.g., control deviation) of the controller 102 in step S501.

In step S503, the setting unit 202 may determine whether to perform adjustment of the corrector 303, such as switching the adjustment unit or adjusting (or readjusting) the parameter value, based on the operation history acquired in step S502. For example, when the operation history satisfies a predetermined condition, the setting unit 202 can determine to perform adjustment (or readjustment) of the switching parameter value of the adjustment unit. The predetermined condition is a condition under which production should be stopped, and for example, it is determined that adjustment of the corrector 303 is necessary when the control deviation acquired as the operation history exceeds a specified value. Then, if adjustment of the corrector 303 by the setting unit 202 is to be performed, the process proceeds to step S504, and if not, the process proceeds to step S505.

In step S504, the setting unit 202 performs adjustment of the corrector 303. This adjustment is performed while the parameter values of the second compensator 302 are maintained in their previous states, and this adjustment resets, for example, the parameter values (coefficients) of the corrector 303.

In step S505, the sequence unit 101 determines whether or not to end production according to the production sequence, and if not, returns to step S501, and if so, ends production. According to the above process, even if production needs to be stopped, the parameter value of the corrector 303 can be quickly adjusted, and production can be resumed while minimizing interruptions to production.

In step S504, the setting unit 202 may send the confirmation sequence to the sequence unit 101, cause the sequence unit 101 to execute the confirmation sequence, and acquire the operation history (e.g., control deviation) in the confirmation sequence from the controller 102. The setting unit 202 may then perform a frequency analysis of the operation history, determine the frequency to be improved based on the result, and determine the parameter value of the corrector 303 so that the control deviation at that frequency falls within a specified value. A more specific example of step S504 will be described in the second embodiment.

Figure 9 shows an example of the measurement results of the disturbance suppression characteristic. The disturbance suppression characteristic is the result of measuring the frequency response when the control deviation when a sine wave is input as the control signal MV in Figure 2 is output. In Figure 9, the horizontal axis represents frequency, and the vertical axis represents the gain of the disturbance suppression characteristic. Since the disturbance suppression characteristic represents the frequency response of the control deviation E when a disturbance is added to the control signal MV, a large gain indicates a low effect of suppressing the disturbance. On the other hand, a small gain indicates a high effect of suppressing the disturbance. In Figure 9, the dashed line represents the disturbance suppression characteristic before adjustment, and the solid line represents the disturbance suppression characteristic after adjustment.

When the frequency indicated by the dashed dotted line in FIG. 9 is defined as the frequency at which the disturbance suppression characteristic should be improved and step S504 is executed, it is possible to obtain a disturbance suppression characteristic as indicated by the solid line, for example. It can be seen that the gain of the disturbance suppression characteristic becomes smaller at the frequency at which the disturbance suppression characteristic should be improved. In Examples 1 to 8, when adjusting the parameters of the corrector 303 provided in the front stage of the neural network, the parameter adjustment may be performed using the disturbance suppression characteristic shown in FIG. 9 as an index.

Second Embodiment
In this embodiment, an example will be described in which the control system SS described in the first embodiment is applied to a stage control device 800. Matters not mentioned in this embodiment follow the first embodiment. Fig. 10 is a diagram showing a hardware configuration when the control system SS shown in Fig. 1 is applied to a stage control device 800.

The stage control device 800 is configured to control the stage 804 while holding an object such as a substrate on the stage 804 in order to control the position of the object. The stage control device 800 includes a control board 801, a current driver 802, a motor 803, a stage 804, and a sensor 805. The control board 801 corresponds to the control device 100 or the controller 102 in the system SS of the first embodiment. The current driver 802, the motor 803, the stage 804, and the sensor 805 correspond to the control target 103 in the system SS of the first embodiment. However, the current driver 802 may be incorporated in the control board 801. Although not shown in FIG. 10, the stage control device 800 may include a sequence unit 101, a learning unit 201, and a setting unit 202.

The control board 801 can be supplied with a position target value as a target value from the sequence unit 101. The control board 801 can generate a current command as a control signal based on the position target value supplied from the sequence unit 101 and position information supplied from the sensor 805, and supply it to the current driver 802. The control board 801 can also supply an operation history to the sequence unit 101.

The current driver 802 can supply a current according to a current command to the motor 803. The motor 803 can be an actuator that converts the current supplied from the current driver 802 into thrust and drives the stage 804 with the thrust. The stage 804 can hold an object such as a plate or a mask. The sensor 805 can detect the position of the stage 804 and supply the position information obtained thereby to the control board 801.

In FIG. 11, a configuration example of the control board 801 is shown as a block diagram. The control board 801 may include a first compensator 301 that generates a first signal S1 based on a position control deviation E of a stage 804 as a control target, and a corrector 303 that generates a correction signal CS by correcting the control deviation E according to an arithmetic expression with an adjustable coefficient. The control board 801 may also include a second compensator 302 that generates a second signal S2 by a neural network based on the correction signal CS, and a calculator 306 that generates a current command as a control signal based on the first signal S1 and the second signal S2. The control board 801 may also include a subtractor 305 that generates a control deviation E, which is the difference between the position target value PR and the position information.

The stage control device 100 of the second embodiment may also include a learning unit 201, as in the first embodiment described in FIG. 7. The learning unit 201 may be configured to perform learning to determine parameter values of the neural network of the second compensator 302. For learning by the learning unit 201, the operation history recording unit 304 may record the operation history required for learning by the learning unit 201 and provide the recorded operation history to the learning unit 201. The operation history may be, for example, a correction signal CS that is input data to the second compensator 302 and a second signal S2 that is output data from the second compensator 302, but may also be other data.

The stage control device 100 of the second embodiment can include a setting unit 202. The setting unit 202 executes an adjustment process for adjusting the parameter values of the corrector 303, and may determine and set the parameter values of the corrector 303 through this adjustment process, or may set the parameter values of the corrector 303 based on an instruction from a user.

8, an exemplary operation of the stage control device 800 of the second embodiment when applied to a production device will be described. In step S501, the sequence unit 101 generates a position target value PR based on a given production sequence and provides it to the stage control device 800. The stage control device 800 controls the position of the stage 804 based on the position target value PR.

In step S502, the setting unit 202 acquires the operation history (e.g., control deviation) of the control board 801 in step S501.

In step S503, the setting unit 202 may determine whether to adjust the corrector 303, such as by switching the adjustment unit or adjusting (or re-adjusting) parameter values, based on the operation history acquired in step S502. The setting unit 202 may determine to adjust the corrector 303, for example, when the operation history satisfies a predetermined condition. The predetermined condition is a condition under which production should be stopped, and for example, it is determined that adjustment of the corrector 303 is necessary when the maximum value of the position control deviation during constant speed driving of the stage 804 exceeds a predetermined specified value. Then, if adjustment of the corrector 303 is to be performed by the setting unit 202, the process proceeds to step S504, and if not, the process proceeds to step S505.

In step S504 , the setting unit 202 may adjust the corrector 303. In step S505, the sequence unit 101 determines whether or not to end the production according to the production sequence, and if not, returns to step S501, and if to end the production, ends the production.

Figure 12 shows a specific example of the process of adjusting the parameter values (or readjusting the parameter values) during the adjustment of the corrector 303 in step S504. In step S601, the setting unit 202 can send a confirmation sequence for confirming the operation of the stage control device 800 to the sequence unit 101, and can cause the sequence unit 101 to generate a position target value PR based on this confirmation sequence. In step S602, the setting unit 202 can obtain a position control deviation E as an operation history from the controller 102, which operates based on the position target value PR.

Now, referring to Figure 13, we will explain the change in position control deviation E before and after parameter adjustment. Figure 13 is a diagram illustrating an example of the position control deviation before and after parameter adjustment. In Figure 13, the horizontal axis represents time, and the vertical axis represents the position control deviation E. Here, the dotted curve represents the position control deviation E before adjusting the parameter value of the compensator 303, and indicates that the position control accuracy has deteriorated. By adjusting the parameter value, it is possible to reduce the fluctuation in the position control deviation E.

In step S603, the setting unit 202 may perform a frequency analysis of the position control error E acquired in step S602. Here, the results of the frequency analysis before and after parameter adjustment will be described with reference to FIG. 14. FIG. 14 is a diagram illustrating the results of the frequency analysis before and after parameter adjustment. In FIG. 14, the horizontal axis is frequency, and the vertical axis is power spectrum. The dotted line indicates the frequency showing the maximum spectrum before adjustment. In step S604, the setting unit 202 may determine, for example, the frequency showing the maximum spectrum in the power spectrum as the frequency to be improved.

Steps S605 to S610 are a specific example of an adjustment process for adjusting the parameter values of the corrector 303. Here, an example is described in which the steepest descent method is adopted as a method for adjusting the parameter values, but other methods may be used. In step S605, the setting unit 202 initializes n to 1. For example, when the arithmetic expression of the corrector 303 is composed of three terms, a first-order integral term, a proportional term, and a first-order differential term, the parameters whose parameter values should be adjusted are Ki, Kp, and Kd. The parameter value pn in the n-th adjustment is shown in the following formula (6).

In step S606, the setting unit 202 can set an arbitrary initial value for the parameter value p1 in the first adjustment of the parameter value pn. In the nth adjustment, the parameter value pn shown in equation (8) described below can be set.

The objective function J(pn) for adjusting the parameter value pn may be, for example, the gain of the disturbance suppression characteristic at the frequency determined in step S604. In step S607, the setting unit 202 may measure a gradient vector grad J(pn) of the objective function J(pn). The gradient vector grad J(pn) may be given by the following equation (7). The gradient vector grad J(pn) may be measured by minutely changing the elements Ki-n, Kp-n, and Kd-n constituting the parameter value pn.

In step S608, the setting unit 202 may determine whether the value of each element of the gradient vector grad J(pn) is equal to or less than a specified value as a convergence determination of the steepest descent method. If the value of each element of the gradient vector grad J(pn) is equal to or less than the specified value, the setting unit 202 may end the adjustment of the parameter value of the corrector 303. On the other hand, if the value of each element of the gradient vector grad J(pn) exceeds the specified value, in step S609, the setting unit 202 may calculate a parameter value pn+1. Here, the parameter value pn+1 may be calculated according to the following formula (8) using, for example, an arbitrary constant α greater than 0. In step S610, the setting unit 202 adds 1 to the value of n and returns to step S606.

In step S611, the setting unit 202 can send a confirmation sequence for confirming the operation of the stage control device 800 to the sequence unit 101, and can cause the sequence unit 101 to generate a position target value PR based on this confirmation sequence. In step S612, the setting unit 202 can obtain a position control deviation E as an operation history from the controller 102, which operates based on the position target value PR.

In step S613, the setting unit 202 determines whether the position control deviation E acquired in step S612 is equal to or less than a specified value. If the position control deviation E exceeds the specified value, the setting unit 202 returns to step S601 and performs the adjustment again. If the position control deviation E is equal to or less than the specified value, the setting unit 202 can end the adjustment.

According to this embodiment, when the state or disturbance of the controlled object including the stage 804 changes, the change can be accommodated by adjusting the parameter value of the corrector 303. For example, in the example of FIG. 13, the position control deviation indicated by the dotted line is reduced to the position control deviation indicated by the solid line, improving the control accuracy.

In the example of equation (6), the number of parameters of the corrector 303 is only three, which is far fewer than the number of parameters of a typical neural network. For example, when using a deep neural network, if the number of dimensions of the input layer is five, the number of dimensions of the hidden layer is two stages of 32, and the number of dimensions of the output layer is eight, the number of parameters will be 1,545. It is possible to complete the adjustment in a shorter time by adjusting the parameter values of the corrector 303 than by determining the values of these 1,545 parameters by re-learning. Therefore, it is possible to maintain the control precision without reducing the productivity of the stage control device 800.

Third Embodiment
In this embodiment, an example will be described in which the control system SS described in the first embodiment is applied to an exposure apparatus EXP. Matters not mentioned in this embodiment follow the first embodiment. Figure 15 shows a schematic configuration example of the exposure apparatus EXP of this embodiment. The exposure apparatus EXP can be configured as a scanning exposure apparatus.

The exposure apparatus EXP may include, for example, an illumination light source 1000, an illumination optical system 1001, a mask stage 1003, a projection optical system 1004, and a plate stage 1006. The illumination light source 1000 may include, but is not limited to, a mercury lamp, an excimer laser light source, or an EUV light source. The exposure light 1010 from the illumination light source 1000 is shaped by the illumination optical system 1001 into the shape of the irradiation area of the projection optical system 1004 with uniform illuminance. In one example, the exposure light 1010 may be shaped into a rectangle that is long in the X direction, which is an axis perpendicular to the plane defined by the Y axis and the Z axis. Depending on the type of the projection optical system 1004, the exposure light 1010 may be shaped into an arc shape. The shaped exposure light 1010 is irradiated onto the pattern of the mask (original) 1002, and the exposure light 1010 that passes through the pattern of the mask 1002 forms an image of the pattern of the mask 1002 on the surface of the plate 1005 (substrate) via the projection optical system 1004.

The mask 1002 is held by the mask stage 1003 by vacuum suction or the like. The plate 1005 is held by the chuck 1007 of the plate stage 1006 by vacuum suction or the like. The positions of the mask stage 1003 and the plate stage 1006 can be controlled by a multi-axis position control device equipped with a position sensor 1030 such as a laser interferometer or a laser scale, a drive system 1031 such as a linear motor, and a controller 1032. The position measurement value output from the position sensor 1030 can be provided to the controller 1032. The controller 1032 generates a control signal based on a position control deviation, which is the difference between the position target value and the position measurement value, and provides it to the drive system 1031 to drive the mask stage 1003 and the plate stage 1006. The pattern of the mask 1002 is transferred to the plate 1005 (the photosensitive material thereon) by scanning and exposing the plate 1005 while synchronously driving the mask stage 1003 and the plate stage 1006 in the Y direction.

The second embodiment will be described when applied to the control of the plate stage 1006. In FIG. 11, the control board 801 corresponds to the controller 1032, the current driver 802 and the motor 803 correspond to the drive system 1031, the stage 804 corresponds to the plate stage 1006, and the sensor 805 corresponds to the position sensor 1030. By applying a controller having a neural network to the control of the plate stage 1006, the position control deviation of the plate stage 1006 can be reduced. This can improve the overlay accuracy and the like. The parameter values of the neural network can be determined by a predetermined learning sequence. However, when the state of the control object changes from the time of learning or the disturbance environment changes, the control accuracy of the plate stage 1006 decreases. Even in such a case, by adjusting the parameter values of the corrector, the adjustment can be completed in a shorter time than by re-learning the neural network. As a result, the control accuracy can be maintained without reducing the productivity of the exposure apparatus.

The second embodiment will be described when applied to the control of the mask stage 1003. In FIG. 11, the control board 801 corresponds to the controller 1032, the current driver 802 and the motor 803 correspond to the drive system 1031, the stage 804 corresponds to the mask stage 1003, and the sensor 805 corresponds to the position sensor 1030.

Even when the second embodiment is applied to the control of the mask stage 1003, it is possible to reduce the position control deviation of the mask stage 1003. This makes it possible to improve the overlay accuracy and the like. The parameter values of the neural network can be determined by a predetermined learning sequence. However, when the state of the controlled object changes from the time of learning or the disturbance environment changes, the control accuracy of the mask stage 1003 decreases. Even in such cases, by adjusting the parameter values of the corrector, the adjustment can be completed in a shorter time than by re-learning the neural network. As a result, it is possible to maintain the control accuracy without reducing the productivity of the exposure apparatus.

The second embodiment can be applied not only to the control of a stage in an exposure apparatus, but also to the control of a stage in other lithography apparatuses such as an imprint apparatus and an electron beam drawing apparatus. In addition, the first or second embodiment can also be applied to the control of a movable part in a transport mechanism that transports an object, such as a hand that holds the object.

<Embodiments of a method for manufacturing an article>
The method for manufacturing an article according to an embodiment of the present invention is suitable for manufacturing, for example, a flat panel display (FPD). The method for manufacturing an article according to this embodiment includes a step of forming a latent image pattern on a photosensitive agent applied on a substrate using the above-mentioned exposure apparatus (a step of exposing the substrate), and a step of developing the substrate on which the latent image pattern has been formed in this step. Furthermore, this manufacturing method includes other well-known steps (oxidation, film formation, deposition, doping, planarization, etching, resist stripping, dicing, bonding, packaging, etc.). The method for manufacturing an article according to this embodiment is advantageous in at least one of the performance, quality, productivity, and production cost of the article compared to conventional methods.

The above describes preferred embodiments of the present invention, but it goes without saying that the present invention is not limited to these embodiments, and various modifications and variations are possible within the scope of the gist of the invention.

REFERENCE SIGNS LIST 100 Control device 103 Control target 301 First compensator 302 Second compensator 303 Corrector 303a First adjustment unit 303b Second adjustment unit 306 Calculator S1 First signal S2 Second signal

Claims (23)

A control device that generates a control signal for controlling a controlled object,
a first compensator that generates a first signal based on a control deviation of the controlled object;
a corrector that corrects the control deviation using one of a plurality of adjustment units that generate a correction signal by correcting the control deviation according to an arithmetic expression with an adjustable coefficient; and
a second compensator that generates a second signal by a neural network based on the correction signal;
a computing unit that generates the control signal based on the first signal and the second signal;
A control device comprising: The control device according to claim 1, characterized in that in the corrector, an adjustment unit to be used for correcting the control deviation is selected from the plurality of adjustment units based on the control state of the controlled object. The control device according to claim 1 or 2, characterized in that an adjustment unit to be used for correcting the control deviation is selected from the plurality of adjustment units based on whether or not the control target and another control target other than the control target are controlled in synchronization with each other. The control device according to claim 1 or 2, characterized in that an adjustment unit to be used for correcting the control deviation is selected from the plurality of adjustment units based on whether or not a switch in the gain of the controlled object has occurred. The control device according to claim 1 or 2, characterized in that an adjustment unit to be used for correcting the control deviation is selected from the plurality of adjustment units based on the state of the first compensator. The control device according to claim 1 or 2, characterized in that an adjustment unit to be used for correcting the control deviation is selected from the plurality of adjustment units based on whether or not the operation pattern of the controlled object has changed. The control device according to any one of claims 1 to 6, characterized in that the calculation formula includes a term proportional to the control deviation. The control device according to any one of claims 1 to 7, characterized in that the calculation formula includes a term that performs integration on the control deviation. The control device according to any one of claims 1 to 8, characterized in that the calculation formula includes a term for differentiating the control deviation. The control device according to any one of claims 1 to 6, characterized in that the arithmetic expression includes at least one of a term proportional to the control deviation, a term for integration, and a term for differentiation. A setting unit that adjusts the corrector is further provided,
11. The control device according to claim 1, wherein the setting unit selects an adjustment unit to be used for correcting the control deviation from the plurality of adjustment units. The control device according to claim 11, characterized in that the setting unit sets the arithmetic expression. The control device according to claim 11 or 12, characterized in that the setting unit resets the coefficients of the calculation formula when the control deviation satisfies a predetermined condition. The control device according to claim 13, characterized in that the predetermined condition includes the control deviation exceeding a specified value. The control device according to any one of claims 1 to 14, further comprising a learning unit that determines parameter values of the neural network by machine learning. the control signal is a signal obtained by correcting the first signal based on the second signal,
The control device according to any one of claims 1 to 15, characterized in that a difference between a control amount resulting from controlling the controlled object based on the control signal and a target value for controlling the controlled object is smaller than a difference between a control amount resulting from controlling the controlled object based on the first signal and a target value for controlling the controlled object. A control device that generates a control signal for controlling a controlled object,
a first compensator that generates a first signal based on a control deviation of the controlled object;
a corrector that corrects the control deviation using one of a plurality of adjustment units that generate a correction signal by correcting the control deviation according to an arithmetic expression;
a second compensator that generates a second signal by a neural network based on the correction signal;
a computing unit that generates the control signal based on the first signal and the second signal;
A control device comprising: The control device according to claim 17, characterized in that the corrector selects an adjustment unit to be used for correcting the control deviation from the plurality of adjustment units based on the control state of the controlled object. The control device according to claim 17 or 18, characterized in that the arithmetic expression includes at least one of a term proportional to the control deviation, an integral term, and a differential term. A stage control device that controls a stage that holds an object in order to control a position of the object,
A stage control device comprising the control device according to any one of claims 1 to 19. A lithography apparatus for transferring a pattern of an original onto a substrate, comprising:
A lithographic apparatus comprising a controller according to any one of claims 1 to 19, configured to control the position of the substrate or the master. A transfer step of transferring a pattern of a master onto a substrate using the lithography apparatus according to claim 21 ;
A processing step of processing the substrate that has been subjected to the transfer step,
A method for manufacturing an article, comprising obtaining an article from the substrate that has been subjected to the processing step. An adjustment method for adjusting a control device including: a first compensator that generates a first signal based on a control deviation of a controlled object; a corrector that corrects the control deviation using one of a plurality of adjustment units that generate a correction signal by correcting the control deviation; a second compensator that generates a second signal by a neural network based on the correction signal; and a calculator that generates a control signal for controlling the controlled object based on the first signal and the second signal,
an adjustment step of selecting, in the corrector, an adjustment unit to be used for correcting the control deviation from the plurality of adjustment units based on a control state of the controlled object;
The adjustment method according to claim 1,

Images