Linear motor stage control device and control method

Abstract

To provide a control device and a control method for a linear motor stage, capable of suppressing vibration due to timing deviation of feedforward control and shortening a settling time at the time of stoppage.SOLUTION: A control device comprises a feedback control unit 6 which feedback-controls a stage 10 by PI control or PID control based on position information of a movable stage 10, and a feedforward control unit 7 which feedforward-controls the stage 10 by a preliminarily learned feedforward value. In feedforward control performed by the feedforward control unit 7, a proportional value and a differential value of feedback control are not included in a feedforward value, and an integral value of feedback control is used as a feedforward value.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a control device and a control method for a linear motor stage. [Background technology]

[0002] Linear motor stages are installed in industrial machinery such as machine tools, measuring instruments, and semiconductor manufacturing equipment and play an important role.

[0003] For example, step-and-repeat semiconductor exposure equipment (steppers) uses a linear motor stage for the stage on which the substrate is placed. The stepper reduces and projects the pattern formed on the photomask onto the shot area on the substrate, and when the pattern transfer to one shot area is complete, it moves the substrate to transfer the pattern to another shot area, repeating this operation. Therefore, it is necessary to stop the substrate at the required position early, and the stage is required to shorten the settling time when stopped.

[0004] A linear motor stage is controlled by feedback control based on the position information of the stage, and among them, PID control is the most common. In order to shorten the settling time when stopping, it is possible to increase the gain of the PID control to improve responsiveness, but there is a limit to how much the gain can be increased because oscillation can occur depending on the position and rigidity of the stage.

[0005] Also known is a control method that combines feedback control with feedforward control (for example, Patent Document 1). This makes it possible to reduce the influence of disturbances such as friction in the linear guide, resistance in the Cableveyor (registered trademark), and vibration of the frame, thereby shortening the settling time. However, as a result of the inventor's investigation, it was found that there are cases where vibrations occur that are thought to be due to timing deviations in the feedforward control, and that the feedforward control may in fact become a disturbance. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2005-322720 A Summary of the Invention [Problem to be solved by the invention]

[0007] The present invention has been made in consideration of the above-mentioned circumstances, and an object of the present invention is to provide a control device and control method for a linear motor stage that can suppress vibrations caused by timing deviations in feedforward control and shorten the settling time when stopping. [Means for solving the problem]

[0008] In a control method that combines feedback control with feedforward control, a feedforward control section can be designed using learning control. In learning control, for example, the control amount when the stage is actually moved under a specified condition and feedback processing is performed is learned in advance, and this is used as a feedforward value for feedforward control. Given the purpose of suppressing the effects of various disturbances by feedforward control, it is normally considered that the servo output, which is the total control amount, is used as the feedforward value as it is.

[0009] However, after extensive research, the inventors discovered that [1] such servo output contains components that suppress disturbances of various frequencies, from high to low, that accompany the movement of the stage, [2] among these, the high-frequency components are mainly derived from proportional and differential values, and the low-frequency components are mainly derived from integral values, and [3] the vibrations caused by the timing deviation of the feedforward control described above are caused by the high-frequency components. The present invention was made based on these findings, and the above-mentioned object can be achieved by the following configuration.

[0010] The linear motor stage control device according to the present invention includes a feedback control unit that performs feedback control of the stage by PI control or PID control based on position information of a movable stage, and a feedforward control unit that performs feedforward control of the stage by a feedforward value learned in advance, the feedforward control is a control in which the proportional value and differential value of the feedback control are not included in the feedforward value, and the integral value of the feedback control is used as the feedforward value. With this device, it is possible to reduce vibrations due to timing deviation of the feedforward control and shorten the settling time when stopping.

[0011] The control method of the linear motor stage according to the present invention feedback controls the stage by PI control or PID control based on position information of a movable stage, and feedforward controls the stage by a feedforward value learned in advance, in which the feedforward control does not include a proportional value and a differential value of the feedback control, but uses an integral value of the feedback control as the feedforward value. This method makes it possible to reduce vibrations caused by timing deviations in the feedforward control and shorten the settling time when the stage is stopped.

[0012] In the above-mentioned control device or control method, the feedback control may be, for example, a control that derives a proportional value and an integral value based on a position deviation between a command position and a detected position of the stage, and derives a differential value based on the detected position. [Brief description of the drawings]

[0013] [Figure 1] Schematic diagram showing an example of the configuration of a linear motor stage. [Diagram 2] A block diagram showing an example of a control system provided in the control device. [Figure 3A] Graph showing the motion waveform just before the stage stops (feedforward control: none) [Figure 3B]Graph showing the motion waveform just before the stage stops (feedforward control: enabled, feedforward value: servo output) [Figure 4A] Graph showing the motion waveform just before the stage stops (feedforward control: none) [Figure 4B] Graph showing the motion waveform just before the stage stops (feedforward control: enabled, feedforward value: integral value) [Diagram 5] A graph showing the transition data of the integral value measured at different stopping positions DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to the drawings.

[0015] 1 is a schematic diagram showing an example of the configuration of a linear motor stage to which the control device and control method of this embodiment are applied. The linear motor stage 1 has a stage 10 that can be moved in the X direction by a linear motor. When mounted on a stepper that is a semiconductor exposure device, the stage 10 can be moved while holding a substrate as a processing object. For example, the linear motor can be a coreless linear motor having a mover made of an electromagnet and a stator made of a permanent magnet. However, the linear motor is not limited to this, and any conventionally known linear motor can be used without any particular restrictions.

[0016] The stage 10 is supported by a linear guide 3 installed on a base plate 2. To ensure positioning accuracy, the base plate 2 is made of a material (such as stone, iron, or ceramics) that is highly rigid and not easily affected by changes in the surrounding temperature and humidity. The surface of the base plate 2 is processed to have high surface accuracy and flatness. Stoppers 31 are provided on both ends of the linear guide 3 to restrict the movement range of the stage 10.

[0017] The position detector 4 is configured to be able to detect the position of the stage 10 in the X direction. For example, a linear encoder can be used for the position detector 4. The linear encoder may be a magnetic linear encoder, but is preferably an optical linear encoder, which is relatively easy to miniaturize. The optical linear encoder may be either a reflective type that detects light pulses due to reflection / non-reflection of light according to the graduations of a linear scale installed on the stator side, or a transmissive type that detects light pulses due to transmission / blocking of light according to the presence or absence of slits on the linear scale. The position detector 4 can be configured using a laser interferometer or the like, instead of a linear encoder.

[0018] The position detector 4 detects the position of the stage 10 and generates a detected position signal PS2 indicating the detected position of the stage 10, and the position information as the detected position is supplied to a control device 5 of the linear motor stage. The control device 5 controls the linear motor stage 1 by a control system described later in order to move the stage 10 to a required position. The control device 5 can be provided by computer hardware and software.

[0019] Fig. 2 is a block diagram showing a control system provided in the control device 5. As shown in Fig. 2, the control device 5 includes a position command unit 51, a feedback control unit 6, and a feedforward control unit 7. The position command unit 51 generates a command position signal PS1 that gives a command position (target position) of the stage 10. The feedback control unit 6 feedback controls the stage 10 by PI control or PID control based on position information of the movable stage 10. The feedforward control unit 7 feedforward controls the stage 10 by a feedforward value learned in advance.

[0020] In this embodiment, the feedback control unit 6 has a calculator 61, a proportional calculator 62, an integrator 63, a calculator 64, a differentiator 65, and a calculator 66. The calculator 61 receives the command position signal PS1 output from the position command unit 51 and the detected position signal PS2 output from the position detection unit 4. The calculator 61 calculates the difference between the command position and the detected position of the stage 10, and a calculation signal CS1 giving the position deviation as the calculation result is input to the proportional calculator 62. The proportional calculator 62 multiplies the position deviation by a proportional gain, and a calculation signal CS2 giving the proportional value as the calculation result is input to the integrator 63 and the calculator 64.

[0021] The integrator 63 integrates the multiplication result of the proportional value and the integral gain, and a calculation signal CS3 that provides the integral value as the calculation result is input to the calculator 64. The calculator 64 adds the proportional value and the integral value, and a calculation signal CS4 that provides the calculation result is input to the calculator 66. The differentiator 65 differentiates the multiplication result of the detected position of the stage 10 and the differential gain, and a calculation signal CS5 that provides the differential value as the calculation result is input to the calculator 66. In addition, a control signal CS6 that provides a feedforward value is input to the calculator 66 by feedforward control, which will be described later.

[0022] The calculator 66 adds the differential value to the sum of the proportional value and the integral value, and the calculation result becomes a servo output (proportional value+integral value+differential value) which is an overall control amount for feedback controlling the stage 10. When a control signal CS6 is input to the calculator 66, the calculator 66 compensates the servo output by adding a feedforward value. A control signal CS7 which provides a servo output as a calculation result by the calculator 66 is input to a drive circuit (not shown) of the stage 10. The drive circuit supplies a required drive current to the coil of the mover so that a thrust corresponding to the control signal CS7 is applied to the stage 10.

[0023] The feedforward control unit 7 performs feedforward control of the stage 10 based on the command position signal PS1 output from the position command unit 51. The feedforward control will be described in detail later.

[0024] The monitoring units 81 to 84 monitor signals that provide servo output, proportional value, integral value, and differential value, respectively. These monitoring results are stored in a predetermined location (for example, a storage unit provided in the control device 5). Transition data such as servo output shown in a graph described later is based on these monitoring results.

[0025] To design the feedforward control section 7 using learning control, it is usually thought that the servo output obtained when the stage 10 is actually moved under predetermined conditions (for example, speed, acceleration, and stop position) and feedback processing is performed is learned in advance, and that the servo output is used as the feedforward value. However, as a result of the inventor's investigation, as will be described with reference to Figures 3A and 3B, it was found that there are cases where vibrations that are thought to be due to timing deviations in the feedforward control occur, and that the feedforward control can actually become a disturbance.

[0026] FIG. 3A is a graph showing the motion waveform of the stage 10, which has been moved under specified conditions, immediately before coming to a halt. However, the feedforward control unit 7 is not in operation, and feedforward control is not being performed. The horizontal axis represents time (sec), and the vertical axis represents position deviation (μm) and command acceleration (μm / s 2 ) and servo output. When a predetermined time (a time derived from the relationship between speed and distance) has elapsed since the position command (issuance of command position signal PS1), the command acceleration becomes zero. The time on the horizontal axis is set to zero at that point. The settling time Ts is the time it takes for the position deviation to fall within the desired settling width Ws (for example, ±100 nm).

[0027] Although the graphs referred to in this specification do not show numerical values ​​on the horizontal axis, the boundary values ​​(i.e., minimum and maximum values) are made to coincide with each other in Fig. 3A and Fig. 3B, Fig. 4A, and Fig. 4B described later. The same is true for the vertical axis. Therefore, by comparing these graphs, it is possible to relatively evaluate the position deviation, settling time Ts, etc.

[0028] FIG. 3B shows the operating waveform when the stage 10 is moved under the same conditions as FIG. 3A. However, feedforward control is being executed. The feedforward value uses the servo output learned in advance, specifically the transition data of the servo output after time zero (see FIG. 3A), and is controlled so that it is added at the timing of time zero. In FIG. 3B, although the settling time Ts is somewhat shorter than in FIG. 3A, the waveform of the position deviation is distorted and the amplitude is large. This is thought to be caused by vibration due to a timing shift in the feedforward control, and the occurrence of such vibration may cause disturbance to the feedforward control.

[0029] As a result of extensive research by the inventors into the above phenomenon, it was found that, as will be explained with reference to Figure 4A, [1] the servo output as a pre-learned feedforward value contains components that suppress disturbances of various frequencies, from high to low, that accompany the movement of stage 10, [2] among these, the high-frequency components are mainly derived from proportional values ​​and differential values, and the low-frequency components are mainly derived from integral values, and [3] the vibrations caused by timing deviations in feedforward control are caused by the high-frequency components.

[0030] FIG. 4A is a graph showing the operating waveforms of the stage 10, which has been moved under specified conditions, just before it comes to a halt. However, the feedforward control section 7 is not operated, and feedforward control is not being performed. A settling time Ts similar to that of FIG. 3B is obtained, but this is because the conditions are different from those of FIG. 3A and FIG. 3B. FIG. 4A also shows the integral value of the feedback control and the sum of the proportional value and the differential value. The servo output corresponds to the sum of these three. As in FIG. 3A, the command acceleration is 0 μm / s after time zero. 2 Since the transition is as shown in FIG. 4B, the illustration is omitted.

[0031] In Fig. 4A, the waveform of the position error shows the influence of a relatively high-frequency disturbance near time zero, followed by the influence of a relatively low-frequency disturbance. Therefore, the servo output for suppressing the influence of the disturbance contains various frequency components from high to low frequencies. If this servo output is used as the feedforward value as is, the high-frequency components are likely to cause a timing shift in the feedforward control, and it is thought that a slight shift in timing can cause the feedforward control to become a disturbance.

[0032] Also, from Figure 4A, it can be seen that the proportional value and the differential value mainly contribute to suppressing the effects of high-frequency disturbances, and the integral value mainly contributes to suppressing the effects of low-frequency disturbances. Therefore, it can be seen that the high-frequency components contained in the servo output are mainly derived from the proportional value and the differential value, and the low-frequency components are mainly derived from the integral value. This is also considered to be the same when the feedback control is PI control, except that the differential value is not used.

[0033] Based on this knowledge, in this embodiment, the feedforward control is a control in which the proportional value and differential value of the feedback control are not included in the feedforward value, and the integral value of the feedback control is used as the feedforward value. Therefore, the feedforward value used in this embodiment includes the integral value of the feedback control, but does not include the proportional value or differential value. In the case of FIG. 3B, the servo output, which is the monitoring result of the monitoring unit 81, can be used as the feedforward value, but in this embodiment, the integral value, which is the monitoring result of the monitoring unit 83, can be used as the feedforward value.

[0034] The feedforward control unit 7 outputs a control signal CS6 that provides a previously learned feedforward value based on the command position signal PS1 output from the position command unit 51. As described above, an integral value stored in a predetermined location by prior learning is used as the feedforward value. The calculator 66 adds a feedforward value (integral value) according to the input control signal CS6 to compensate the servo output. As will be described later, in this embodiment, control is performed so that the feedforward value is added at the timing of time zero.

[0035] FIG. 4B shows the operating waveform when the stage 10 is moved under the same conditions as FIG. 4A. However, feedforward control is executed. A previously learned integral value, specifically, transition data of the integral value after time zero (see FIG. 4A) is used as the feedforward value, and control is performed so that the integral value is added at the timing of time zero. In this embodiment, such timing is adopted to perform feedforward control just before stopping, but this is not limited to this, and the timing for adding the integral value as the feedforward value is arbitrary. For example, it is also possible to use a point back from time zero as a reference and add the integral value data thereafter at the reference timing.

[0036] Looking at the waveform of the position error in Figure 4B, the effects of low-frequency disturbances are effectively suppressed, and as a result, fluctuations in the position error are contained relatively quickly. As a result, the settling time Ts is shortened, decreasing to less than half that of Figure 4A. This is believed to be because the feedforward value does not include proportional or differential values, and vibrations caused by timing deviations in feedforward control due to high-frequency components are suppressed. By shortening the settling time Ts when the stage 10 stops, the stepper can start exposure earlier, which helps improve the takt time.

[0037] In FIG. 4B, the integral value of the feedback control remains almost constant due to the feedforward processing of the integral value. In this way, the integral does almost no work in the feedback control, so the integral gain can be lowered. If the integral gain is excessively increased, it acts not only on low-frequency disturbances but also on high-frequency disturbances, making it easier to cause oscillation, so lowering the integral gain is a measure to prevent oscillation. High-frequency disturbances can be countered by increasing the proportional gain or differential gain, so the gain adjustment method is clear and easy to understand. From the viewpoint of enhancing the disturbance suppression effect by the feedforward control, it is preferable to increase the integral gain and make the integral value larger when learning the integral value that becomes the feedforward value.

[0038] When learning the integral value that will be the feedforward value, the conditions when the linear motor stage 1 is actually used are used for the conditions such as speed, acceleration, and stop position. If there is concern about fluctuations in the integral value depending on the stop position of the stage 10, it is possible to measure the integral values ​​at multiple stop positions and use their average value as the feedforward value. Figure 4B shows the result of using such an average value, specifically, the average value data shown in Figure 5 is adopted as the feedforward value. However, as will be described later, it is not necessary to use the average value.

[0039] FIG. 5 is a graph showing transition data of the integral value measured by changing the stop position for the operation waveform of FIG. 4A, and the scale of the vertical axis is enlarged compared to FIG. 4A. Ten types of stop positions were used, varying at a specified pitch (for example, 15 mm pitch) from position A to position J. In FIG. 5, there is almost no variation in the integral value due to differences in the stop position, and each shows a similar transition. Therefore, if the speed and acceleration conditions are the same, it is considered that a certain integral value (for example, the integral value at position E) can be used as a feedforward value regardless of the position at which the motor is stopped.

[0040] In the above-mentioned Fig. 3B, an example was shown in which the vibration and amplitude of the position deviation increased when feedforward control was performed, but this was the result of using the servo output as the pre-learned feedforward value. Therefore, even under the operating conditions of Fig. 3B, if the integral value of the feedback control is used as the pre-learned feedforward value, the settling time can be expected to be shortened as in the case of Fig. 4B.

[0041] In the present embodiment, an example has been shown in which the feedback control executed by the feedback control unit 6 derives a proportional value and an integral value based on the position deviation between the command position and the detected position of the stage 10, and derives a differential value based on the detected position. However, the feedback control may be executed by other PI control or PID control, for example, PID control that derives a differential value based on the position deviation. In that case, the configuration may be such that the calculation signal CS1 is input to the differentiator 65, and a signal that gives the calculation result obtained by differentiating the result of multiplying the position deviation and the differential gain is input to the calculator 66.

[0042] As described above, in the control method of the linear motor stage in this embodiment, the stage 10 is feedback-controlled by PI control or PID control based on position information of the movable stage 10, and the stage 10 is feedforward-controlled by a feedforward value learned in advance. In addition, the feedforward control does not include the proportional value and differential value of the feedback control in the feedforward value, but uses the integral value of the feedback control as the feedforward value.

[0043] The linear motor stage to which the control device and control method according to the present invention are applied is not limited to a one-axis linear motor stage, and may be, for example, one or both of a two-axis linear motor stage. The two-axis linear motor stage may be an XY stage equipped with two-axis linear motor stages installed in directions perpendicular to each other, or a gantry stage in which two-axis linear motor stages installed parallel to each other are connected by a connection table.

[0044] The control device and control method according to the present invention have the above-mentioned advantageous effects and can shorten the settling time when the stage stops, and are therefore particularly useful for a linear motor stage that moves a substrate in a semiconductor exposure apparatus. The semiconductor exposure apparatus is not limited to the above-mentioned stepper, and may be a step-and-scan type semiconductor exposure apparatus (scanner).

[0045] The present invention is not limited to the above-described embodiment, and various improvements and modifications can be made without departing from the spirit and scope of the present invention. [Explanation of symbols]

[0046] 1 Linear motor stage 5. Control device 6 Feedback control section 7 Feedforward control section 10 Stages

Claims

1. a feedback control unit that feedback controls the stage by PI control or PID control based on position information of a movable stage; a feedforward control unit that controls the stage by a feedforward value; A control device for a linear motor stage, which does not use a proportional value and a differential value of the feedback control that have been learned in advance as the feedforward value, but uses an integral value of the feedback control that has been learned in advance.

2. 2. The control device for a linear motor stage according to claim 1, wherein the feedback control derives a proportional value and an integral value based on a position deviation between a command position and a detected position of the stage, and derives a differential value based on the detected position.

3. feedback-controlling the stage by PI control or PID control based on position information of a movable stage, and feedforward-controlling the stage by a feedforward value; A method for controlling a linear motor stage, comprising the steps of: not using a proportional value and a differential value of the feedback control that have been learned in advance, but using an integral value of the feedback control that has been learned in advance, as the feedforward value.

4. 4. The control method for a linear motor stage according to claim 3, wherein the feedback control derives a proportional value and an integral value based on a position deviation between a command position and a detected position of the stage, and derives a differential value based on the detected position.

Images