JP2000275370A - Method for updating compensation parameter of stage and active vibration isolation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To achieve a method for updating compensation parameters without interfering with productivity while securing position accuracy or the like in a required scene, by applying a pseudo irregular signal, identifying the physical parameters of a stage according to the response data and closed loop identification and updating the compensation parameters. SOLUTION: When an M-series signal generator 14 is started while a stage 7 is being fed speedily in a speed control mode, an M-series signal Ms is applied to the prestage of a power amplifier 10. The signal Ms and an output signal Mr of a speed controller 9 where the signal has been applied are taken into a system identification operation part 15, the characteristics of a rotary direct drive mechanism by a ball screw/nut and the fluctuation are captured rather than the closed loop characteristics of speed control, and each compensation parameter of an optimum speed controller 9 and a position controller 13 is calculated. Then, by using the speed compensation parameter-setting means 18 and the position compensation parameter-setting means 19, each compensation parameter of the speed controller 9 and the speed controller 13 is updated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the measurement of characteristics of a motion mechanism, particularly to the measurement of characteristics of a motion mechanism such as an XY stage or an active vibration isolator in an exposure apparatus used for manufacturing semiconductor devices and the like, and based on the results. The present invention relates to a method for adjusting and updating a compensation parameter.

[0002]

2. Description of the Related Art An electron microscope using an electron beam or a semiconductor exposure apparatus represented by a stepper or a scanner includes:
Various servo devices such as an XY stage, a fine movement stage mounted on the stage, and an active anti-vibration device that insulates them from floor vibration are operating.

FIG. 6 shows the structure of an active vibration isolator and an XY stage mounted thereon. In FIG. 1, reference numeral 1 denotes a vibration isolation table, and 2 (2-1 to 2-3) denote mechanical units of an active vibration isolation device that supports the entire main body structure including the vibration isolation table 1 and the XY stage 3. 4 shows an active mount. The active vibration isolator comprises at least three or more active mounts 2 supporting the main body structure, insulates the vibration of the floor on which the exposure apparatus is installed from propagating to the main body structure, and
It has a function of suppressing the swing of the main body structure due to the reaction force caused by the rapid acceleration and deceleration of the Y stage 3. The former function is called vibration isolation, and the latter function is called vibration suppression.

In order to secure and maintain the performance of the exposure apparatus, a servo device must be implemented in which a feedback for maximizing the performance of the above-described motion mechanism is implemented. In addition, it is necessary to quickly capture a change in the characteristics of the motion mechanism over time and reflect the change in the adjustment of the compensation parameter in the servo device, thereby maintaining the performance of the device at a constant level. Therefore, conventionally, identification that quantitatively captures a nominal value of a physical parameter of a mechanism and a variation from the nominal value has been performed, and a parameter adjustment of a servo device has been performed based on the result. That is, this is a technique called by the term auto tuning.

However, the servo apparatus in the above-described exposure apparatus has the following problems that must be solved. Therefore, identification has not been performed and parameter adjustment based on the results has not been performed. (1) The application of the excitation signal is a disturbance to the servo device and acts as a disturbance that disturbs the steady / transient characteristics. However, in order to ensure identification accuracy, the input signal must include many frequency components. That is, the excitation signal needs to be rich so as to excite the vibration of the object to be measured. Naturally, for a closed-loop system, a richer excitation signal is perceived as a larger disturbance, and thus cannot be applied to a servo device in an exposure apparatus that achieves a position accuracy on the order of nanometers. (2) A sine wave signal is generally used as the excitation signal for identification. However, as long as this signal is used, there is a problem that it takes time to perform measurement with high analysis accuracy over a necessary frequency band. Therefore, in an exposure apparatus that emphasizes throughput (productivity), measurement for identification that hinders production cannot be performed.

(3) In general, it is desirable to identify a mechanical mechanism in a servo device in an open loop state from the viewpoint of accuracy. However, actually, it is not easy to measure the open-loop characteristics of the servo device in the actual operation state. For example, if the positioning servo of the XY stage in operation is removed for open loop frequency response measurement, the operation for initialization must be performed again after the measurement is completed. Also, this is remarkable in the case of an XY stage having a hydrostatic bearing as a guide, but due to its low friction characteristics, the XY stage gradually moves due to the stimulation of the excitation signal for measurement, and this is specified during the measurement period. Because you can't stay in that place. Taking the active vibration isolator as an example, cutting the servo system to measure the mechanical characteristics gives a large disturbance to the measurement values of various measuring instruments mounted on the main body structure, Above all, there is a danger that the equilibrium position of the main body structure will be uncertain and mechanical collision will occur. In other words, it is concluded that the servo system of the active anti-vibration apparatus cannot be removed for the purpose of measuring the characteristics. That is, although the ultimate purpose is to improve the performance of the exposure apparatus, the characteristic measurement (identification) by cutting the loop of the servo apparatus has not been permitted especially in an actual apparatus. That is, the identification has to be performed under a closed loop state where the servo device is actually operating. However, there is no appropriate method for identifying the servo device in the actual operation state of the exposure apparatus.

[0007]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and achieves a positional accuracy on the order of nanometers and sufficiently utilizes the specialty of an exposure apparatus which emphasizes productivity. It is an object of the present invention to provide a closed-loop identification method with consideration given thereto and a method for updating a compensation parameter based on the identification result.

The problems leading to the present invention are summarized as follows. It is desirable that the XY stage in the exposure apparatus in the actual operation state be adjusted periodically to exhibit the best performance at all times. Conventionally, there has been a technique for identifying a servo device and performing adjustment based on the result. However, identification and parameter adjustment based on servo equipment such as an XY stage or an active anti-vibration device in an exposure system that ensures positioning accuracy on the order of nanometers and must not impede productivity are performed. Had not been. The reason is that the application of the identification signal disturbs the position accuracy, and the measurement for the identification takes time to lower the productivity. Also, using the physical parameters obtained by the identification,
This is because there was no algorithm to improve or maintain the positioning characteristics of the Y stage. Anyone would think that if the physical parameters representing the mechanical characteristics of the XY stage could be reflected in the setting of the compensation parameters for positioning, it would be possible to achieve preferable positioning. He was absent.

The same applies to the active anti-vibration apparatus. It is of course possible to identify the active anti-vibration apparatus offline, that is, independently of the operation as the exposure apparatus. There was no known means of identification in the presence state. And there was no guide as to how to use the obtained identification results to readjust the compensation parameters of the active anti-vibration device.

In the present invention, taking into account the specialty of the XY stage or the active vibration isolator in the exposure apparatus, it is possible to prevent the productivity while observing not only the positional accuracy and the like in the necessary scene. It provides a closed loop identification method and a compensation parameter adjustment method.

[0011]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems. Specifically, in the stage parameter updating method according to the present invention, a stage control device for performing positioning by performing step driving or a stage control device for performing scanning driving, and applies a pseudo-random signal for closed-loop identification. Then, the physical parameters of the stage are identified by closed-loop identification from the pseudo-random signal and the data of the response thereto, and the compensation parameters in the stage position control device are updated using the physical parameters.

Here, the application period of the pseudo irregular signal is a constant speed section of the stage, an arbitrary period except for the final positioning, or between the scan drive and the next scan drive. The compensation parameter to be updated is a parameter of the speed controller and the position controller in the case of a control system having a triple loop of current, speed, and position. If the position loop has a PID compensator, these parameters are used. It is.

In addition, each time the closed loop identification is performed, the compensation parameter is updated when it is detected that the control limit is exceeded by continuous or periodic identification. The closed-loop identification by applying the pseudo-random signal can be performed every time the stage is positioned, every time a predetermined number of steps and repeats or steps and scans are performed, or every time a predetermined time elapses. Then, an M-sequence signal can be suitably used as the pseudorandom signal to be applied for identification.

Next, a pseudo-random signal for identifying a closed loop is applied to an active vibration isolator in the exposure apparatus, and the closed loop identification is performed based on the signal and response data of at least the number of degrees of freedom of motion. The physical parameters of the active anti-vibration apparatus are identified by using the parameter, and the compensation parameters in the active anti-vibration apparatus are updated using the physical parameters.

Here, the application of the pseudo-random signal is uncorrelated for the number of degrees of freedom of motion controlled by the active anti-vibration device, and is simultaneously performed. The application period of the pseudo-irregular signals that are uncorrelated with each other in the number of degrees of freedom of motion is the period between the sequential positioning of all the shots of the semiconductor wafer performed by the stage mounted on the active vibration isolator and the next sequential positioning. During the period from the scan performed by the stage mounted on the active vibration isolation device to the next scan,
This is a transfer period for transferring a semiconductor wafer or a transfer period for exchanging a reticle. Then, an M-sequence signal can be suitably used as the pseudo irregular signal.

[0016]

Embodiments of the present invention will be described below with reference to the drawings. [Embodiment 1] A semiconductor exposure apparatus (stepper or scanner, hereinafter referred to as a stepper) according to a first embodiment of the present invention.
FIGS. 1 and 2 are diagrams illustrating a method of updating a compensation parameter of a stage in FIG. Before a detailed description based on the figure, a basic idea will be described.

First, attention is paid to the fact that the positioning of the XY stage in the stepper is PTP (point-to-point) control. That is, it is important for the performance of the stepper that the XY stage is settled at the target position, and the behavior during the movement does not matter. Therefore,
A vibration signal for closed loop identification is applied during step driving to perform closed loop identification, and the compensation parameter is updated based on the result.

More specifically, in the case of an XY stage control device in which the control system of the XY stage has switching means from the speed control mode to the position control mode, only in the state of the position control mode without mode switching means. Let us clarify the timing of applying the excitation signal to both the step driving over a long distance and the XY stage control device for positioning to the final target position.

(1) In the case of an XY stage control device having mode switching means When the XY stage is switched from the speed control mode to the position control mode to perform the step-and-repeat operation, the speed control mode is performed at a constant speed. A vibration signal for identification is applied to the section. Since the XY stage is in a steady state with a constant speed, the response when the excitation signal is applied corresponds to the excitation signal on a one-to-one basis, so that a good SN can be measured.

FIG. 3 shows the actual speed profile of the XY stage over time. Usually, a trapezoidal speed profile is obtained, but a vibration signal for identification is applied to a section in which the XY stage is moved at the maximum speed. Here, the excitation signal for identification is a pseudo irregular signal. Among them, an M-sequence signal is preferably used. In recent years, the positioning time of the XY stage has been shortened. Therefore, the constant speed section in FIG. 3 is extremely short, but the application time of the M-sequence signal for identification is also extremely short.

(2) XY without mode switching means
In the case of the stage control device When all the steps and repeat operations are executed in the position control mode without switching the control mode as described above, at least a period excluding the final positioning period is used for identification. Here, a pseudo-irregular signal is used as the excitation signal for identification. Then, among the pseudo-irregular signals, the M-sequence signal is preferably used.

FIG. 4 shows a profile of a position target (upper stage) and an actual position (lower stage) when positioning the XY stage by step driving in the XY stage controller in the position control mode. Here, the excitation signal for identification is applied to the moving section excluding the period immediately after the start of the positioning and the period of the final target value positioning. However, the excitation signal for identification can be applied at least during a period excluding the period of final target value positioning.

Now, when applying a vibration signal for identification,
The XY stage control device is not in an adjustment state different from the original operation state. In the stepper, the XY stage control device is in an adjusted state as it should be, and the excitation signal is applied thereto to obtain the physical parameters of the XY stage itself. So-called closed loop identification is performed.

Next, based on the identification result, the compensation parameters are updated to optimize the positioning of the XY stage. Regarding the update time, there are several patterns as described below. (A) For each step drive, application of an excitation signal, closed loop identification based on the result, and a compensation parameter to be updated based on the identification result are calculated, and the compensation parameter is updated immediately before final positioning. That is, the compensation parameter is updated for each positioning shot of the semiconductor wafer.

(B) After the step-and-repeat of the entire semiconductor wafer is completed, the compensation parameter is updated at the same step location that is followed when step-and-repeat is performed. That is, the application of the excitation signal, the subsequent process of identifying the closed loop, and the process of updating the compensation parameter are alternately performed for each step and repeat of the entire semiconductor wafer. The same applies to the case of the step-and-scan operation. (C) Close-loop identification is performed by applying a pseudo-random signal continuously or periodically during step-and-repeat or step-and-scan, and the identification result is continuously monitored and there is a fluctuation that deviates from the initial characteristic. Update the compensation parameter only when

In addition to the above, the timing and method of updating the compensation parameters can be selected according to the usage of the XY stage and the operation results. Specifically, as the algorithm for updating the compensation parameter of the XY stage control device, for example, the following two can be suitably used. Both Algorithm 1 and Algorithm 2 are described in Japanese Patent Application No. 10-231128 filed by the present applicant. Here, the control mode of the XY stage controller is position, and a PID compensator is used for compensation.

(1) Algorithm 1 The compensator of the XY stage controller is F_x + F_i / S + F_v 
PID compensator, stage mass, viscous friction
Coefficient, spring constant, and power amplifier to drive stage
The first-order lag time constant of₁ , B₁ , K₁ , T_d When
Every K _p Is the position gain, K_d The power amplifier
Inn, K_t Is the thrust constant and σ is the responsiveness specified by the designer.
Prescribed numerical value, α₀ , Α₁ , Α_Two , Α_Three The desired response
In the coefficient of the reference model as
The parameters are determined based on the following equation.

[0028]

(Equation 1)

(2) Algorithm 2 The compensator of the position control system of the XY stage is F _x + F _i / s +
F _v in the form of a PID compensator, the mass of the stage, the viscous friction coefficient, the spring constant and respectively a first-order delay time constant of the power amplifier that drives the positioning stage m _1, b _1,,
k ₁ and T _d , K _p is a position gain, K _d is a gain of a power amplifier, _Δt is a thrust constant, σ is a time scale for defining responsiveness, α ₀ , α ₁ , α ₂ , α ₃ , Α ₄ as the desired response coefficient of the reference model, and a cubic equation for σ

[0030]

(Equation 2) , The positive minimum real root σ is selected from the solutions, and F _i , F _x , and F _v are determined based on the equations (1a) to (1c) using the minimum real root σ.

Regarding the application of the above algorithm, several measures for shortening the positioning time are conceivable. (1) The PID parameter based on Algorithm 1 is σ
The smaller the is specified, the higher the gain and the higher the convergence. On the other hand, power amplifiers and actuators tend to be saturated. Therefore, σ can be switched according to the magnitude of the position deviation signal during step driving. Of course,
In the XY stage where the step-and-repeat is performed, the smaller the value of σ, the closer to the final target position.

Further, the optimum σ may be specified based on the known moving distance information before actually driving the XY stage. That is, when the moving distance is long, σ is relatively large, and when the moving distance is short, σ is specified to be small within a range that does not cause saturation. (2) Since the PID parameter based on the algorithm 2 has a high gain, it tends to cause saturation of the power amplifier and the actuator. Therefore, the present invention is applied only at the time of minute positioning with a small position deviation signal.

By the above, the entire flow from the closed loop identification performed at the time of positioning the XY stage to the subsequent update of the compensation parameter can be understood. Hereinafter, the configuration of the XY stage control device having the functions of closed-loop identification and parameter updating will be described more specifically.

First, FIG. 1 shows that a motor (abbreviated as M in the figure) 4 rotates a stage 7 using a ball screw 5 and a nut 6.
1 shows a stage control device for linearly moving a stage. In the figure, a motor 4 is energized by energizing a power amplifier 10 including a current controller and a current feedback thereto.
Is driven. The power amplifier 10 is driven by the output signal of the speed controller 9. The input to the speed controller 9 is a signal obtained by comparing the output signal of the amplifier 17 or the position controller 13 with the speed signal of the tacho generator (abbreviated as TG in the figure) 8. The laser interferometer 11 as a position detecting means detects the position of the stage 7 by measuring a movable mirror 12 mounted on the stage 7. This negative feedback signal is compared with a speed or position command value (target value). When compared with the target speed value, the state changeover switch 16 is connected to the side of V (speed control mode), the speed target signal passes through the amplifier 17, and this output signal is compared with the output of the tacho generator 8. Then, a speed deviation signal is obtained to drive the speed controller 9. On the other hand, when comparing with the target position value, the state changeover switch 16 is connected to the P (position control mode) side, and the position deviation obtained by comparing the target position value with the output of the laser interferometer 11 is obtained. The signal is guided to the position controller 13. Subsequently, the position controller 1
3 and the output of the tacho generator 8 are compared.
Further, in the stage control device of FIG. 1, while the state changeover switch 16 is on the V side and the stage 7 is moving at high speed in the speed control mode, the M-sequence signal generator 1
When 4 is started, the M-sequence signal Ms, which is the output, is applied to the front stage of the power amplifier 10. Ms
And the output signal Mr of the speed controller 9 in which the signal has made a round, is taken into the system identification calculation unit 15, and first, the characteristics or the characteristic fluctuation of the rotation / linear motion mechanism using the ball screw / nut are grasped from the closed loop characteristics of the speed control. Then, optimal compensation parameters of the speed controller 9 and the position controller 13 are calculated based on the characteristics or characteristic fluctuations of the rotary / linear motion mechanism. Then, in order to reflect the result to the actual compensation, the respective compensation parameters of the speed controller 9 and the position controller 13 are updated using the speed compensation parameter setting means 18 and the position compensation parameter setting means 19.

Next, FIG. 2 is a block diagram of a stage positioning control system of a type in which the stage is directly positioned and driven without passing through a rotary / linear motion mechanism as shown in FIG. Specifically, it is a block diagram of a position control system for a stage using a non-contact static pressure guide and a linear motor as a representative of an electromagnetic actuator. This figure is disclosed by the present applicant in Japanese Patent Application Laid-Open No. 5-250041.

Now, reference numeral 40 surrounded by a broken line in the figure represents a dynamic model of a stage mounted on the vibration isolator. Here, m ₁ is the mass of the stage, b ₁ is the viscous friction coefficient of the stage, k ₁ is the spring constant of the stage, m ₂ is the mass of the vibration isolation table, b ₂ is the viscous friction coefficient of the vibration isolation table, and k ₂ is The spring constant of the vibration isolation table, f is a driving force, and f _ext is a disturbance acting on the vibration isolation table.

Here, a laser interferometer as a position detecting means is installed on the vibration isolation table, and therefore the feedback signal is x ₁ -x ₂ . This signal is the target position signal x. Is compared with a position deviation signal, and the position gain K _p
, And then to the PID compensator 41. This transfer function is represented by the following equation as shown.

[0038]

(Equation 3) Here, _Fx is a proportional gain, _Fi is an integral gain, and _Fv is a differential gain. The output of the PID compensator 41 drives the power amplifier 42. Power amplifier 4
The 2, are also subjected feedback acceleration feedback gain A detects the acceleration s ² x ₂ of the anti-vibration table at a primary delay time constant T _a. The role of A is to make the relative distance between the stage and the vibration isolation table constant by causing the stage to follow the vibration of the vibration isolation table. The optimum value A _(opt) of A for this is given by the following equation.

[0039]

(Equation 4)

As described with reference to FIG. 1, the output Ms of the M-sequence signal generator 14 for closed-loop identification is added to the previous stage of the power amplifier 42.
s and the signal Mr that makes a round of the closed loop system are taken into the system identification calculation unit 15. Here, closed-loop identification is performed to capture the characteristics of the stage mounted on the vibration isolation table. Then, based on these characteristics, the desired algorithm of the PID compensator 41 is calculated by applying the algorithm 1 or the algorithm 2 described above. Then, the PID parameter setting means 43 updates each gain of the PID compensator 41. Here, the application timing of the output signal Ms of the M-sequence signal generator 14 is selected at least during a period other than the final positioning settling period, and for that purpose, a command is issued from the system identification calculation unit 15 to the M-sequence signal generator 14. . In the case of FIG. 4, the M-sequence signal is applied during the moving section, excluding the period immediately after the start of positioning.

In realizing this, the position gain K _p ,
A portion 44 enclosed by a broken line including the PID compensator 41, the M-sequence signal generator 14, the system identification operation unit 15, and the PID parameter setting unit 43 is a controller unit that can be realized by hardware or software.

Finally, the effects brought about by applying an M-sequence signal, which is a kind of pseudorandom signal, for identification during step-and-repeat or step-and-scan of the XY stage will be summarized. . (A) Since the M-sequence signal is used as the excitation signal,
The application time is overwhelmingly shorter than that of the excitation using the sine wave, so that the excitation for identification can be completed during the movement of the stage. For this reason, it is possible to capture the characteristic variation of the stage for each stage movement position, that is, for each shot of the semiconductor wafer. (B) The adjustment for the optimum positioning can be executed in real time or in a learning manner based on the identification result for each moving place. (C) As a result, the positioning characteristics of the XY stage can always be maintained at the highest state, which has the effect of contributing to the maintenance of the productivity of the stepper.

(D) It is known that when the guide mechanism of the XY stage slides or rolls, a problem occurs due to friction of these guides. Typically, the oscillation is called stick-slip. As a control method for suppressing this phenomenon, superposition of a dither signal is known. The application of the dither signal linearizes non-linear friction characteristics (for example, Coulomb friction), and acts to prevent stick-slip from occurring. Here, the original purpose of the application of the M-sequence signal lies in identification using a step-and-repeat period irrelevant to the positioning of the PTP, but it can also have a role as a dither signal for compensating for nonlinear friction.

(E) When applying the M-sequence signal, the frequency band to be measured can be selected. It is not necessary to cover all frequency ranges necessary and sufficient to express the characteristics of the XY stage. In general, characteristic fluctuations in a high frequency range pose a problem. At this time, it is sufficient to excite the M-sequence signal so that only the characteristics in the high frequency range can be captured. In this case, the application time of the signal is extremely short. This is because the characteristics in the low frequency range of the mechanism created through the normal design and manufacturing processes are stable. If the variation in the low frequency range is large, this is because there is a fundamental problem in the design scene or the quality control in the manufacturing process is insufficient. If there is no defect in the quality control of the design and manufacturing processes, variation in high-frequency dynamics generally becomes a problem. This variation may destabilize the closed loop system and has a subtle effect on the performance of the closed loop servo system. Therefore, an M-sequence signal that can measure only the high-frequency characteristics of the control target may be input. In this case, the application time (identification cycle) of the M-sequence signal becomes extremely short, and does not cause disturbance that impairs the positioning index (positioning time and positioning accuracy).

[Second Embodiment] In the first embodiment, noting that the XY stage for performing step and repeat is PTP control, an M-sequence signal which is a kind of a pseudo irregular signal between positioning and the next positioning. Then, the physical parameters of the XY stage mechanism were captured by closed-loop identification, and the appropriate compensation parameters were updated based on the identification results to reflect the final positioning characteristics.

The same concept can be applied to an active vibration isolator in a stepper. For this device, vibration for closed-loop identification is performed using a function that does not significantly disturb vibration isolation and vibration suppression, which are functions of the active vibration isolation device. Then, the compensation parameter is updated.

First, the mounting state of the actuator and the sensor will be described with reference to the active vibration isolator shown in FIG. 6 again. AC- is, for example, an acceleration sensor as vibration detecting means, PO- is a position sensor as position measuring means, and SV- is a working fluid to an air spring (not shown) for supporting a device main body including a large-weight anti-vibration table. This is a servo valve that controls the supply and exhaust of air. The symbol attached after the hyphen "-" in the symbols representing these sensors and actuators indicates the bearing and the site on the vibration isolation table. The azimuth will be clear by referring to the coordinate system shown in FIG.

Next, the active mounts 2-1, 2-2, 2-
In the active vibration isolator that controls 3
The system is shown in FIG. Here, the position sensors PO-Z1, -Z
2, -Z3, -X1, -Y2, -Y3 are output from the anti-vibration table 1.
Output [z] of the position target value output unit 20 for specifying the equilibrium position of
_Ten, Z₂₀, Z₃₀, X_Ten, Y₂₀, Y₃₀] And position compared
The deviation signal [e_z1, E_z2, E_z3, E_x1, E_y2, E_y3]
You. Subsequently, the motion mode extraction calculation means 21 relating to the position
Modes such as translation and rotation of the vibration isolation table 1
The deviation signal [e_x , E_y , E_z , Eθ_x , Eθ_y , Eθ
_z ] Is obtained. This motion mode deviation signal gives a steady-state deviation of zero
It is led to a PI compensator 22 for guaranteeing. here
Where P is proportional and I is integral operation. Acceleration sensor
AC-Z1, -Z2, -Z3, -X1, -Y2, -Y
The output of 3 is a motion mode extraction calculating means 2 relating to acceleration.
It is led to 3. And the translation and rotation of the vibration isolation table 1
Motion mode acceleration signal [a_x , A_y , A_z , Aθ
_x , Aθ_y , Aθ_z ] Is calculated. And the vibration isolation table 1
Gain compensator 2 for adjusting damping for each motion mode
4 The output of the gain compensator 24 is described earlier.
Of the motion mode by adding the output of the
Drive signal [d_x , D_y , D_z , Dθ_x , Dθ _y , D
θ_z ]. The drive signal for this motion mode is motion
Actuator for each axis guided to the mode distribution calculation means 25
The drive signal [d_z1, D_z2, D _z3, D_x1, D
_y1, D_y2] Is generated. By these drive signals,
-Bo valve SV-Z1, -Z2, -Z3, -X1, -Y
2. Voltage-current converter for controlling -Y3 valve opening and closing
26 (abbreviated as VI converter in the figure).

According to the control structure of the active vibration isolator as described above, the active mount 2-1 supporting the vibration isolator 1 and
Damping is given to 2-2 and 2-3, and the vibration isolation table 1 is positioned at an equilibrium position without a steady deviation.

Note that LA-X, LA-Y and LA-θ are X
A laser interferometer as a position detecting means for positioning the Y stage 3 is attached to the exposure apparatus main body 34, and detects rotations in the X direction, the Y direction, and the XY plane, respectively. Of course, the active anti-vibration device also includes the
In FIG. 5, SV, PO, A
C is depicted as being attached to 34.

Now, in addition to the above-described configuration, the output of the multi-channel, uncorrelated, M-sequence signal generator 27 is added to the previous stage of the motion mode distribution calculating means 25. The active mounts 2-1 2-2, 2-
The anti-vibration table 1 supported by 3 is excited in a six-degree-of-freedom motion mode such as translation or rotation by M-sequence signals that are uncorrelated with each other. Then, the excitation signal and the output of the gain compensator 24 in which the excitation signal has passed through are fetched into the data storage unit 28. Appropriate filtering is performed in the pre-filtering 29 on the time-series data of each of the captured signals, and the signal is guided to the closed-loop identification unit 30. Here, the characteristics of the control target are identified from the input / output data representing the characteristics of the closed loop. That is, the active mount 2-
The mechanical characteristics of the vibration isolation table 1 supported by 1, 2-2, and 2-3 are clarified. Subsequently, the update parameter calculator 3
In step 1, based on the calculation result of the closed loop identification unit 30, a compensation parameter of a control system for maintaining a constant vibration isolation and vibration suppression characteristic, which are performance indexes of the active vibration isolation device, is calculated. For example, based on the calculation result of the closed loop identification unit 30, a change in the spring constant or the viscous friction coefficient of the active mounts 2-1 2-2, 2-3 can be captured and the update amount of the compensation parameter can be calculated. . Finally, the parameter change setting section 32 includes an update parameter calculation section 31
The parameter adjustment is executed by accessing a specific adjustable portion in the control loop based on the calculation result of. In the case shown, readjustment has been performed on the gain compensator 24 indicated by A and the PI compensator 22 indicated by B.

In FIG. 5, the response data as a closed loop system by exciting the M-sequence signal is
4 output is selected. That is, here, the closed loop characteristic of the acceleration minor loop is measured.
Of course, the closed-loop characteristics of the feedback based on the output of the position sensor PO are measured for the application location of the output of the multi-channel, uncorrelated, M-sequence signal generator 27 and the location of the response data obtained by the excitation, if necessary. May be set as follows.

The adjustment location is not limited to A and B. For example, the position target value output unit 20 may be adjusted again. In short, it is only necessary to adjust the closed loop characteristics of the active vibration isolation device.

Although not shown in FIG. 5, the active anti-vibration apparatus may include a feed forward. For example, in addition to the air spring actuator used in FIG. 5, an active anti-vibration apparatus that includes an electromagnetic motor typified by a linear motor or the like and is hybridized is also in practical use. In such a case, in order to suppress the vibration of the vibration isolation table 1 caused by the driving reaction force of the XY stage, the drive signal of the XY stage is appropriately shaped and applied to a closed loop system configured by using an electromagnetic motor as an actuator. A feed-forward path is provided. As can be easily understood, the transfer characteristic of the feedforward path has a great effect in suppressing the vibration of the vibration isolation table. Therefore,
It is not impeded that the fluctuations in the mechanical characteristics of the vibration isolation table due to the closed-loop identification are captured and the results are used to update the transfer characteristics of the feedforward path.

That is, the second embodiment includes that the result of the closed loop identification is used not only for updating the parameters in the closed loop system but also for updating the parameters of the open loop such as a feed forward path.

By the above description, the operation of the control system for the active anti-vibration apparatus shown in FIG. 5 and the apparatus configuration and concept for performing the closed loop identification and updating the parameters based on the result can be understood. .

Here, reference will be made to the timing of activating the parameter updating device 33 surrounded by the broken line in FIG.
The excitation signal for identification is an M-sequence signal, but the superposition of this signal is only a disturbance for the active vibration isolator. For example, when trying to converge the XY stage 3 to the target position, the active vibration isolation device operates to suppress the vibration of the vibration isolation table 1 due to the step driving of the XY stage 3. In such a case, if an M-sequence signal is applied to the active anti-vibration apparatus for identification, the convergence of the positioning of the XY stage is adversely affected, and thus the original operation of the stepper is inhibited. Therefore, the application of the M-sequence signal for closed-loop identification is performed during a period in which the basic function of the stepper is not impaired. Such periods include the following.

(1) The period between the step and repeat for all shots on the semiconductor wafer surface and the next one. (2) The period for shifting from the scan performed for the shot on the semiconductor wafer surface to the scan for the next shot ( 3) Transport period for transferring semiconductor wafers (4) Transport period for reticle exchange In other words, it adversely affects the positioning of the XY stage 3, the measurement of measurement equipment mounted on the exposure apparatus main body, the exposure of the semiconductor wafer, and the like. If the anti-vibration table 1 supported by the active anti-vibration device is in a steady state, these periods can be used for closed-loop identification. for that reason,
Starting and stopping of the parameter updating device 33 is controlled by a control signal (not shown), and the parameter updating device 33 can be started during any of the above periods and stopped during any other period.

The technical idea of the present invention is embodied in the first embodiment for the XY stage and in the second embodiment for the active vibration isolator. Of course, the motion mechanism in the stepper is not limited to the XY stage and the active vibration isolator, and the application of the idea of the present invention is not impeded to other servo devices including those of the stepper. Needless to say.

[0060]

The effects of the present invention are as follows. (1) In a state where a servo device in an exposure apparatus is actually operating, closed-loop identification of the device can be performed without disturbing accuracy or productivity. (2) Based on the closed loop identification, it is possible to capture the characteristic of the mechanical mechanism or its fluctuation in the servo device in the actual operation state. (3) The compensation parameter can be updated based on the characteristic or characteristic fluctuation of the mechanical mechanism. (4) Therefore, since the compensation parameter for compensating for the characteristic change of the mechanical mechanism is constantly updated, a constant servo performance can be exhibited. (5) Therefore, there is an effect that it greatly contributes to the productivity of the exposure apparatus.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a method of updating a compensation parameter of a stage according to a first embodiment of the present invention.

FIG. 2 is a block diagram of a stage positioning control system.

FIG. 3 is a diagram showing a speed profile of an XY stage with respect to time.

FIG. 4 is a diagram showing a profile of a position target and an actual position in a position control mode.

FIG. 5 is a diagram illustrating a configuration of a control system for the active vibration isolation device.

FIG. 6 shows an active vibration isolator and XY mounted thereon.
FIG. 3 is a diagram illustrating a structure of a stage.

[Explanation of symbols]

1: anti-vibration table, 2: active mount, 3: XY stage,
4: Motor, 5: Ball screw, 6: Nut, 7: Stage, 8: Tach generator, 9: Speed controller, 10: Power amplifier, 11: Laser interferometer, 12: Moving mirror, 13:
Position controller, 14: M-sequence signal generator, 15: System identification operation unit, 16: State changeover switch, 17: Amplifier, 18: Speed compensation parameter setting means, 19: Position compensation parameter setting means, 20: Position target value Output section, 2
1: Exercise mode extraction calculation means for position, 22: PI
Compensator, 23: Motion mode extraction calculation means related to acceleration, 24: Gain compensator, 25: Motion mode distribution calculation means, 26: Voltage-current converter, 27: Multi-channel, uncorrelated, M-sequence signal generator, 28: Data storage unit, 29: pre-filtering, 30: closed loop identification unit, 31: update parameter calculation unit, 32: parameter update setting unit, 3
3: Parameter updating device, 34: Exposure device body, 40:
Dynamic model, 41: PID compensator, 42: power amplifier,
43: PID parameter setting means, 44: controller unit.

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) // B23Q 17/12 G05D 3/12 305V 5H303 G05D 3/12 305 H01L 21/68 K H01L 21/68 21 / 30 503A 503F F term (reference) 2F078 CA01 CA08 CB05 CB13 CC11 3C048 BB12 DD06 EE10 3J048 AB07 AD02 EA13 5F031 HA55 5F046 AA23 BA04 BA05 CC01 CC02 CC03 DA06 DA07 DD06 5H303 AA06 BB01 KK01 KK01 DD01 KK23 KK24

Claims (9)

[Claims] 1. A pseudorandom signal for identifying a closed loop is applied to a control closed loop of a stage for performing positioning by step drive or a stage for performing scan drive, and the pseudorandom signal and data of a response to the signal are applied. A method of updating a compensation parameter of a stage, comprising identifying a physical parameter of the stage by closed-loop identification from, and updating a compensation parameter in a control device of the stage using the physical parameter. 2. The method according to claim 1, wherein the application of the pseudo-irregular signal is performed in a constant speed section of the stage, an arbitrary period except for final positioning,
2. The method according to claim 1, wherein the step is performed during any period between the scan drive and the next scan drive. 3. In the case of a control system having a triple loop of current, speed, and position, the compensation parameter is a parameter of a speed controller and a position controller.
3. The stage compensation parameter updating method according to claim 1, wherein the PID parameter is used when a compensator is provided. 4. The stage compensation parameter updating method according to claim 1, wherein the pseudorandom signal applied for the closed loop identification is an M-sequence signal. 5. The updating of the compensation parameter is performed each time the closed loop identification is performed, every time a step-and-repeat or a step-and-scan in the process of the closed-loop identification is performed, The method according to any one of claims 1 to 4, wherein the method is performed when a control limit is exceeded by continuous or periodic identification. 6. The closed loop identification by applying the pseudo-random signal is performed every time the stage is positioned or scanned, every time a predetermined number of steps and repeats or steps and scans are performed, or every time a predetermined time elapses. The method according to any one of claims 1 to 4, wherein the compensation parameter of the stage is updated. 7. A pseudorandom signal for identifying a closed loop is applied to a closed loop for control of an active vibration isolator on which a stage for performing positioning by step drive or a stage for performing scan drive is mounted. And identifying the physical parameters of the active vibration isolation device by closed-loop identification from the response data of at least the number of degrees of freedom of motion, and updating the compensation parameters in the active vibration isolation device using the physical parameters. Of updating the compensation parameter of the active anti-vibration apparatus to be used. 8. The active anti-vibration device according to claim 7, wherein the application of the pseudo-random signal is performed simultaneously and uncorrelated with the number of degrees of freedom of motion controlled by the active anti-vibration device. A method for updating a compensation parameter of a vibration device. 9. The pseudo-irregular signal is applied to the active anti-vibration apparatus during a period of sequential positioning of all shots of a semiconductor wafer performed by a stage mounted on the active anti-vibration apparatus and a next sequential positioning. 9. The method according to claim 7, wherein the voltage is applied during a period from a scan performed by the placed stage to a next scan, a transfer period for transferring a semiconductor wafer, or a transfer period for replacing a reticle. 10. Update method of active vibration isolator.